{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.915816640039327,
  "eval_steps": 500,
  "global_step": 280000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00017556488000140453,
      "grad_norm": 5.521957336675931,
      "learning_rate": 3.160112359550562e-07,
      "loss": 1.0812,
      "step": 10
    },
    {
      "epoch": 0.00035112976000280905,
      "grad_norm": 5.708572602993713,
      "learning_rate": 6.671348314606742e-07,
      "loss": 1.0907,
      "step": 20
    },
    {
      "epoch": 0.0005266946400042135,
      "grad_norm": 5.375281817925661,
      "learning_rate": 1.0182584269662922e-06,
      "loss": 1.0868,
      "step": 30
    },
    {
      "epoch": 0.0007022595200056181,
      "grad_norm": 5.052791845417721,
      "learning_rate": 1.3693820224719102e-06,
      "loss": 1.0727,
      "step": 40
    },
    {
      "epoch": 0.0008778244000070225,
      "grad_norm": 3.8611207439866524,
      "learning_rate": 1.7205056179775281e-06,
      "loss": 1.0479,
      "step": 50
    },
    {
      "epoch": 0.001053389280008427,
      "grad_norm": 3.6465595676164804,
      "learning_rate": 2.071629213483146e-06,
      "loss": 1.0199,
      "step": 60
    },
    {
      "epoch": 0.0012289541600098315,
      "grad_norm": 1.3902749235806497,
      "learning_rate": 2.4227528089887643e-06,
      "loss": 0.9591,
      "step": 70
    },
    {
      "epoch": 0.0014045190400112362,
      "grad_norm": 1.0776370002359033,
      "learning_rate": 2.773876404494382e-06,
      "loss": 0.9458,
      "step": 80
    },
    {
      "epoch": 0.0015800839200126406,
      "grad_norm": 0.5087540307713369,
      "learning_rate": 3.125e-06,
      "loss": 0.9419,
      "step": 90
    },
    {
      "epoch": 0.001755648800014045,
      "grad_norm": 0.34249186022454586,
      "learning_rate": 3.4761235955056183e-06,
      "loss": 0.9196,
      "step": 100
    },
    {
      "epoch": 0.0019312136800154497,
      "grad_norm": 0.3453262311951503,
      "learning_rate": 3.827247191011236e-06,
      "loss": 0.9236,
      "step": 110
    },
    {
      "epoch": 0.002106778560016854,
      "grad_norm": 0.2403328842081639,
      "learning_rate": 4.178370786516854e-06,
      "loss": 0.9158,
      "step": 120
    },
    {
      "epoch": 0.002282343440018259,
      "grad_norm": 0.1820471162175142,
      "learning_rate": 4.529494382022472e-06,
      "loss": 0.9108,
      "step": 130
    },
    {
      "epoch": 0.002457908320019663,
      "grad_norm": 0.19566371419518408,
      "learning_rate": 4.88061797752809e-06,
      "loss": 0.9078,
      "step": 140
    },
    {
      "epoch": 0.0026334732000210677,
      "grad_norm": 0.14067728487599765,
      "learning_rate": 5.231741573033709e-06,
      "loss": 0.9071,
      "step": 150
    },
    {
      "epoch": 0.0028090380800224724,
      "grad_norm": 0.15601909764113486,
      "learning_rate": 5.582865168539326e-06,
      "loss": 0.9036,
      "step": 160
    },
    {
      "epoch": 0.0029846029600238766,
      "grad_norm": 0.13106316674985408,
      "learning_rate": 5.933988764044944e-06,
      "loss": 0.8971,
      "step": 170
    },
    {
      "epoch": 0.0031601678400252813,
      "grad_norm": 0.23159646369429554,
      "learning_rate": 6.285112359550562e-06,
      "loss": 0.9001,
      "step": 180
    },
    {
      "epoch": 0.003335732720026686,
      "grad_norm": 0.139445412751372,
      "learning_rate": 6.6362359550561805e-06,
      "loss": 0.9069,
      "step": 190
    },
    {
      "epoch": 0.00351129760002809,
      "grad_norm": 0.12232303227649992,
      "learning_rate": 6.987359550561798e-06,
      "loss": 0.8973,
      "step": 200
    },
    {
      "epoch": 0.003686862480029495,
      "grad_norm": 0.09774383208297045,
      "learning_rate": 7.338483146067417e-06,
      "loss": 0.8984,
      "step": 210
    },
    {
      "epoch": 0.0038624273600308995,
      "grad_norm": 0.09872372482967814,
      "learning_rate": 7.689606741573034e-06,
      "loss": 0.9006,
      "step": 220
    },
    {
      "epoch": 0.004037992240032304,
      "grad_norm": 0.09927394772845226,
      "learning_rate": 8.040730337078651e-06,
      "loss": 0.8925,
      "step": 230
    },
    {
      "epoch": 0.004213557120033708,
      "grad_norm": 0.10205270462284845,
      "learning_rate": 8.391853932584271e-06,
      "loss": 0.897,
      "step": 240
    },
    {
      "epoch": 0.004389122000035113,
      "grad_norm": 0.08598854537659635,
      "learning_rate": 8.742977528089889e-06,
      "loss": 0.8952,
      "step": 250
    },
    {
      "epoch": 0.004564686880036518,
      "grad_norm": 0.14052247056430883,
      "learning_rate": 9.094101123595505e-06,
      "loss": 0.903,
      "step": 260
    },
    {
      "epoch": 0.004740251760037922,
      "grad_norm": 0.11633339538747048,
      "learning_rate": 9.445224719101124e-06,
      "loss": 0.8993,
      "step": 270
    },
    {
      "epoch": 0.004915816640039326,
      "grad_norm": 0.15667199732434572,
      "learning_rate": 9.796348314606742e-06,
      "loss": 0.8926,
      "step": 280
    },
    {
      "epoch": 0.005091381520040731,
      "grad_norm": 0.1580863352124487,
      "learning_rate": 1.0147471910112361e-05,
      "loss": 0.8906,
      "step": 290
    },
    {
      "epoch": 0.0052669464000421355,
      "grad_norm": 0.11700961833744775,
      "learning_rate": 1.0498595505617977e-05,
      "loss": 0.8911,
      "step": 300
    },
    {
      "epoch": 0.00544251128004354,
      "grad_norm": 0.1002158140214453,
      "learning_rate": 1.0849719101123595e-05,
      "loss": 0.8964,
      "step": 310
    },
    {
      "epoch": 0.005618076160044945,
      "grad_norm": 0.11450525112897186,
      "learning_rate": 1.1200842696629214e-05,
      "loss": 0.8914,
      "step": 320
    },
    {
      "epoch": 0.0057936410400463495,
      "grad_norm": 0.10355642114335632,
      "learning_rate": 1.1551966292134832e-05,
      "loss": 0.8935,
      "step": 330
    },
    {
      "epoch": 0.005969205920047753,
      "grad_norm": 0.13003862574172384,
      "learning_rate": 1.190308988764045e-05,
      "loss": 0.8923,
      "step": 340
    },
    {
      "epoch": 0.006144770800049158,
      "grad_norm": 0.11942176095930303,
      "learning_rate": 1.2254213483146068e-05,
      "loss": 0.8976,
      "step": 350
    },
    {
      "epoch": 0.006320335680050563,
      "grad_norm": 0.174729880743143,
      "learning_rate": 1.2605337078651685e-05,
      "loss": 0.8954,
      "step": 360
    },
    {
      "epoch": 0.006495900560051967,
      "grad_norm": 0.17454279494445282,
      "learning_rate": 1.2956460674157303e-05,
      "loss": 0.8842,
      "step": 370
    },
    {
      "epoch": 0.006671465440053372,
      "grad_norm": 0.11093942310863414,
      "learning_rate": 1.3307584269662923e-05,
      "loss": 0.8938,
      "step": 380
    },
    {
      "epoch": 0.0068470303200547766,
      "grad_norm": 0.12459459333977055,
      "learning_rate": 1.365870786516854e-05,
      "loss": 0.9032,
      "step": 390
    },
    {
      "epoch": 0.00702259520005618,
      "grad_norm": 0.10478445850007473,
      "learning_rate": 1.4009831460674158e-05,
      "loss": 0.8899,
      "step": 400
    },
    {
      "epoch": 0.007198160080057585,
      "grad_norm": 0.21758319521997785,
      "learning_rate": 1.4360955056179776e-05,
      "loss": 0.8911,
      "step": 410
    },
    {
      "epoch": 0.00737372496005899,
      "grad_norm": 0.3799682257333132,
      "learning_rate": 1.4712078651685394e-05,
      "loss": 0.8952,
      "step": 420
    },
    {
      "epoch": 0.007549289840060394,
      "grad_norm": 0.16613723696367122,
      "learning_rate": 1.5063202247191011e-05,
      "loss": 0.8957,
      "step": 430
    },
    {
      "epoch": 0.007724854720061799,
      "grad_norm": 0.10950583373419281,
      "learning_rate": 1.541432584269663e-05,
      "loss": 0.8922,
      "step": 440
    },
    {
      "epoch": 0.007900419600063203,
      "grad_norm": 0.11813836136178629,
      "learning_rate": 1.5765449438202247e-05,
      "loss": 0.8824,
      "step": 450
    },
    {
      "epoch": 0.008075984480064607,
      "grad_norm": 0.13696044256853676,
      "learning_rate": 1.6116573033707866e-05,
      "loss": 0.8916,
      "step": 460
    },
    {
      "epoch": 0.008251549360066012,
      "grad_norm": 0.2135809828531423,
      "learning_rate": 1.6467696629213482e-05,
      "loss": 0.8898,
      "step": 470
    },
    {
      "epoch": 0.008427114240067417,
      "grad_norm": 0.12168929213785253,
      "learning_rate": 1.68188202247191e-05,
      "loss": 0.893,
      "step": 480
    },
    {
      "epoch": 0.008602679120068821,
      "grad_norm": 0.1845567646344545,
      "learning_rate": 1.716994382022472e-05,
      "loss": 0.891,
      "step": 490
    },
    {
      "epoch": 0.008778244000070226,
      "grad_norm": 0.15251799572779476,
      "learning_rate": 1.7521067415730337e-05,
      "loss": 0.8886,
      "step": 500
    },
    {
      "epoch": 0.00895380888007163,
      "grad_norm": 0.2092028472029786,
      "learning_rate": 1.7872191011235957e-05,
      "loss": 0.8878,
      "step": 510
    },
    {
      "epoch": 0.009129373760073035,
      "grad_norm": 0.11402792829918096,
      "learning_rate": 1.8223314606741573e-05,
      "loss": 0.8819,
      "step": 520
    },
    {
      "epoch": 0.00930493864007444,
      "grad_norm": 0.11373947249157612,
      "learning_rate": 1.8574438202247192e-05,
      "loss": 0.8886,
      "step": 530
    },
    {
      "epoch": 0.009480503520075845,
      "grad_norm": 0.16039738696762704,
      "learning_rate": 1.892556179775281e-05,
      "loss": 0.8836,
      "step": 540
    },
    {
      "epoch": 0.00965606840007725,
      "grad_norm": 0.09881854664635709,
      "learning_rate": 1.9276685393258428e-05,
      "loss": 0.8869,
      "step": 550
    },
    {
      "epoch": 0.009831633280078652,
      "grad_norm": 0.18258558704333444,
      "learning_rate": 1.9627808988764047e-05,
      "loss": 0.8813,
      "step": 560
    },
    {
      "epoch": 0.010007198160080057,
      "grad_norm": 0.1383098131580563,
      "learning_rate": 1.9978932584269663e-05,
      "loss": 0.8838,
      "step": 570
    },
    {
      "epoch": 0.010182763040081462,
      "grad_norm": 0.23244818037178885,
      "learning_rate": 2.0330056179775282e-05,
      "loss": 0.8871,
      "step": 580
    },
    {
      "epoch": 0.010358327920082866,
      "grad_norm": 0.17421140066252747,
      "learning_rate": 2.0681179775280902e-05,
      "loss": 0.8878,
      "step": 590
    },
    {
      "epoch": 0.010533892800084271,
      "grad_norm": 0.12457583224780955,
      "learning_rate": 2.1032303370786518e-05,
      "loss": 0.887,
      "step": 600
    },
    {
      "epoch": 0.010709457680085676,
      "grad_norm": 0.11724554014808711,
      "learning_rate": 2.1383426966292137e-05,
      "loss": 0.8806,
      "step": 610
    },
    {
      "epoch": 0.01088502256008708,
      "grad_norm": 0.1244203189846869,
      "learning_rate": 2.1734550561797753e-05,
      "loss": 0.8882,
      "step": 620
    },
    {
      "epoch": 0.011060587440088485,
      "grad_norm": 0.14846248260345,
      "learning_rate": 2.208567415730337e-05,
      "loss": 0.8895,
      "step": 630
    },
    {
      "epoch": 0.01123615232008989,
      "grad_norm": 0.3889962019808758,
      "learning_rate": 2.2436797752808992e-05,
      "loss": 0.8867,
      "step": 640
    },
    {
      "epoch": 0.011411717200091294,
      "grad_norm": 0.13123791801487897,
      "learning_rate": 2.2787921348314608e-05,
      "loss": 0.8865,
      "step": 650
    },
    {
      "epoch": 0.011587282080092699,
      "grad_norm": 0.11073845463814745,
      "learning_rate": 2.3139044943820228e-05,
      "loss": 0.8806,
      "step": 660
    },
    {
      "epoch": 0.011762846960094104,
      "grad_norm": 0.1578732363955825,
      "learning_rate": 2.3490168539325844e-05,
      "loss": 0.8887,
      "step": 670
    },
    {
      "epoch": 0.011938411840095507,
      "grad_norm": 0.10143291619102777,
      "learning_rate": 2.384129213483146e-05,
      "loss": 0.8805,
      "step": 680
    },
    {
      "epoch": 0.012113976720096911,
      "grad_norm": 0.08763109379315187,
      "learning_rate": 2.419241573033708e-05,
      "loss": 0.8851,
      "step": 690
    },
    {
      "epoch": 0.012289541600098316,
      "grad_norm": 0.08784034286918553,
      "learning_rate": 2.45435393258427e-05,
      "loss": 0.8779,
      "step": 700
    },
    {
      "epoch": 0.01246510648009972,
      "grad_norm": 0.09181187838551577,
      "learning_rate": 2.4894662921348315e-05,
      "loss": 0.887,
      "step": 710
    },
    {
      "epoch": 0.012640671360101125,
      "grad_norm": 0.19649377203734933,
      "learning_rate": 2.5245786516853938e-05,
      "loss": 0.8834,
      "step": 720
    },
    {
      "epoch": 0.01281623624010253,
      "grad_norm": 0.12061678019396534,
      "learning_rate": 2.559691011235955e-05,
      "loss": 0.8929,
      "step": 730
    },
    {
      "epoch": 0.012991801120103934,
      "grad_norm": 0.10118288204090287,
      "learning_rate": 2.5948033707865173e-05,
      "loss": 0.8786,
      "step": 740
    },
    {
      "epoch": 0.01316736600010534,
      "grad_norm": 0.0951775531904648,
      "learning_rate": 2.6299157303370786e-05,
      "loss": 0.8823,
      "step": 750
    },
    {
      "epoch": 0.013342930880106744,
      "grad_norm": 0.14484406391415428,
      "learning_rate": 2.6650280898876405e-05,
      "loss": 0.8932,
      "step": 760
    },
    {
      "epoch": 0.013518495760108148,
      "grad_norm": 0.09667707906081065,
      "learning_rate": 2.700140449438202e-05,
      "loss": 0.8853,
      "step": 770
    },
    {
      "epoch": 0.013694060640109553,
      "grad_norm": 0.09430730124868125,
      "learning_rate": 2.735252808988764e-05,
      "loss": 0.8846,
      "step": 780
    },
    {
      "epoch": 0.013869625520110958,
      "grad_norm": 0.19445492363204217,
      "learning_rate": 2.770365168539326e-05,
      "loss": 0.8856,
      "step": 790
    },
    {
      "epoch": 0.01404519040011236,
      "grad_norm": 0.14603069834354043,
      "learning_rate": 2.8054775280898876e-05,
      "loss": 0.8811,
      "step": 800
    },
    {
      "epoch": 0.014220755280113765,
      "grad_norm": 0.16915073790782253,
      "learning_rate": 2.8405898876404495e-05,
      "loss": 0.8896,
      "step": 810
    },
    {
      "epoch": 0.01439632016011517,
      "grad_norm": 0.19749962008297817,
      "learning_rate": 2.875702247191011e-05,
      "loss": 0.8755,
      "step": 820
    },
    {
      "epoch": 0.014571885040116575,
      "grad_norm": 0.20199712665010328,
      "learning_rate": 2.910814606741573e-05,
      "loss": 0.8858,
      "step": 830
    },
    {
      "epoch": 0.01474744992011798,
      "grad_norm": 0.19151917433720766,
      "learning_rate": 2.945926966292135e-05,
      "loss": 0.8867,
      "step": 840
    },
    {
      "epoch": 0.014923014800119384,
      "grad_norm": 0.15104805977276609,
      "learning_rate": 2.9810393258426966e-05,
      "loss": 0.889,
      "step": 850
    },
    {
      "epoch": 0.015098579680120789,
      "grad_norm": 0.08698921718514607,
      "learning_rate": 3.0161516853932586e-05,
      "loss": 0.8833,
      "step": 860
    },
    {
      "epoch": 0.015274144560122193,
      "grad_norm": 0.21719231893707253,
      "learning_rate": 3.0512640449438202e-05,
      "loss": 0.8772,
      "step": 870
    },
    {
      "epoch": 0.015449709440123598,
      "grad_norm": 0.15292399935888182,
      "learning_rate": 3.086376404494382e-05,
      "loss": 0.8864,
      "step": 880
    },
    {
      "epoch": 0.015625274320125,
      "grad_norm": 0.12868821622479323,
      "learning_rate": 3.121488764044944e-05,
      "loss": 0.8808,
      "step": 890
    },
    {
      "epoch": 0.015800839200126406,
      "grad_norm": 0.1114463273839004,
      "learning_rate": 3.1566011235955053e-05,
      "loss": 0.8834,
      "step": 900
    },
    {
      "epoch": 0.01597640408012781,
      "grad_norm": 0.14646938220861777,
      "learning_rate": 3.191713483146068e-05,
      "loss": 0.876,
      "step": 910
    },
    {
      "epoch": 0.016151968960129215,
      "grad_norm": 0.08950874145362948,
      "learning_rate": 3.226825842696629e-05,
      "loss": 0.8834,
      "step": 920
    },
    {
      "epoch": 0.01632753384013062,
      "grad_norm": 0.09747202190192505,
      "learning_rate": 3.261938202247191e-05,
      "loss": 0.8788,
      "step": 930
    },
    {
      "epoch": 0.016503098720132024,
      "grad_norm": 0.08582650256353923,
      "learning_rate": 3.297050561797753e-05,
      "loss": 0.881,
      "step": 940
    },
    {
      "epoch": 0.01667866360013343,
      "grad_norm": 0.13023059248395036,
      "learning_rate": 3.3321629213483144e-05,
      "loss": 0.8792,
      "step": 950
    },
    {
      "epoch": 0.016854228480134834,
      "grad_norm": 0.10247776876484127,
      "learning_rate": 3.367275280898877e-05,
      "loss": 0.8817,
      "step": 960
    },
    {
      "epoch": 0.017029793360136238,
      "grad_norm": 0.08775525370839382,
      "learning_rate": 3.402387640449438e-05,
      "loss": 0.8846,
      "step": 970
    },
    {
      "epoch": 0.017205358240137643,
      "grad_norm": 0.11476314763996193,
      "learning_rate": 3.4375e-05,
      "loss": 0.8725,
      "step": 980
    },
    {
      "epoch": 0.017380923120139048,
      "grad_norm": 0.09282239496851163,
      "learning_rate": 3.472612359550562e-05,
      "loss": 0.8805,
      "step": 990
    },
    {
      "epoch": 0.017556488000140452,
      "grad_norm": 0.10890505592962989,
      "learning_rate": 3.5077247191011234e-05,
      "loss": 0.8842,
      "step": 1000
    },
    {
      "epoch": 0.017732052880141857,
      "grad_norm": 0.09240520159701895,
      "learning_rate": 3.5428370786516854e-05,
      "loss": 0.8832,
      "step": 1010
    },
    {
      "epoch": 0.01790761776014326,
      "grad_norm": 0.10706185216074135,
      "learning_rate": 3.577949438202247e-05,
      "loss": 0.8821,
      "step": 1020
    },
    {
      "epoch": 0.018083182640144666,
      "grad_norm": 0.15557394691058934,
      "learning_rate": 3.613061797752809e-05,
      "loss": 0.8815,
      "step": 1030
    },
    {
      "epoch": 0.01825874752014607,
      "grad_norm": 0.3040345859600493,
      "learning_rate": 3.648174157303371e-05,
      "loss": 0.8858,
      "step": 1040
    },
    {
      "epoch": 0.018434312400147475,
      "grad_norm": 0.12407916724674464,
      "learning_rate": 3.6832865168539325e-05,
      "loss": 0.8771,
      "step": 1050
    },
    {
      "epoch": 0.01860987728014888,
      "grad_norm": 0.11502050992909092,
      "learning_rate": 3.7183988764044944e-05,
      "loss": 0.8803,
      "step": 1060
    },
    {
      "epoch": 0.018785442160150285,
      "grad_norm": 0.14017187555612576,
      "learning_rate": 3.7535112359550563e-05,
      "loss": 0.8769,
      "step": 1070
    },
    {
      "epoch": 0.01896100704015169,
      "grad_norm": 0.09143403988859881,
      "learning_rate": 3.788623595505618e-05,
      "loss": 0.8871,
      "step": 1080
    },
    {
      "epoch": 0.019136571920153094,
      "grad_norm": 0.11545895191184127,
      "learning_rate": 3.82373595505618e-05,
      "loss": 0.8815,
      "step": 1090
    },
    {
      "epoch": 0.0193121368001545,
      "grad_norm": 0.13316062255351235,
      "learning_rate": 3.8588483146067415e-05,
      "loss": 0.8738,
      "step": 1100
    },
    {
      "epoch": 0.0194877016801559,
      "grad_norm": 0.08756710132417034,
      "learning_rate": 3.8939606741573034e-05,
      "loss": 0.8799,
      "step": 1110
    },
    {
      "epoch": 0.019663266560157305,
      "grad_norm": 0.10678008159217663,
      "learning_rate": 3.9290730337078654e-05,
      "loss": 0.8767,
      "step": 1120
    },
    {
      "epoch": 0.01983883144015871,
      "grad_norm": 0.11334060600352037,
      "learning_rate": 3.964185393258427e-05,
      "loss": 0.882,
      "step": 1130
    },
    {
      "epoch": 0.020014396320160114,
      "grad_norm": 0.1160410339823135,
      "learning_rate": 3.999297752808989e-05,
      "loss": 0.8816,
      "step": 1140
    },
    {
      "epoch": 0.02018996120016152,
      "grad_norm": 0.12459030880076682,
      "learning_rate": 4.0344101123595505e-05,
      "loss": 0.8795,
      "step": 1150
    },
    {
      "epoch": 0.020365526080162923,
      "grad_norm": 0.12559299403358756,
      "learning_rate": 4.0695224719101125e-05,
      "loss": 0.8782,
      "step": 1160
    },
    {
      "epoch": 0.020541090960164328,
      "grad_norm": 0.10023943972764061,
      "learning_rate": 4.1046348314606744e-05,
      "loss": 0.8791,
      "step": 1170
    },
    {
      "epoch": 0.020716655840165733,
      "grad_norm": 0.10316844757941478,
      "learning_rate": 4.1397471910112364e-05,
      "loss": 0.8823,
      "step": 1180
    },
    {
      "epoch": 0.020892220720167137,
      "grad_norm": 0.14926091459568772,
      "learning_rate": 4.174859550561798e-05,
      "loss": 0.8907,
      "step": 1190
    },
    {
      "epoch": 0.021067785600168542,
      "grad_norm": 0.08984367645265927,
      "learning_rate": 4.2099719101123596e-05,
      "loss": 0.881,
      "step": 1200
    },
    {
      "epoch": 0.021243350480169947,
      "grad_norm": 0.17623413875591554,
      "learning_rate": 4.2450842696629215e-05,
      "loss": 0.8739,
      "step": 1210
    },
    {
      "epoch": 0.02141891536017135,
      "grad_norm": 0.16791972225000076,
      "learning_rate": 4.2801966292134835e-05,
      "loss": 0.8794,
      "step": 1220
    },
    {
      "epoch": 0.021594480240172756,
      "grad_norm": 0.10879541248589036,
      "learning_rate": 4.3153089887640454e-05,
      "loss": 0.8802,
      "step": 1230
    },
    {
      "epoch": 0.02177004512017416,
      "grad_norm": 0.11338165637195254,
      "learning_rate": 4.350421348314607e-05,
      "loss": 0.8805,
      "step": 1240
    },
    {
      "epoch": 0.021945610000175565,
      "grad_norm": 0.12179767113807875,
      "learning_rate": 4.3855337078651686e-05,
      "loss": 0.8762,
      "step": 1250
    },
    {
      "epoch": 0.02212117488017697,
      "grad_norm": 0.09004762223463969,
      "learning_rate": 4.4206460674157305e-05,
      "loss": 0.8807,
      "step": 1260
    },
    {
      "epoch": 0.022296739760178375,
      "grad_norm": 0.11161869071576214,
      "learning_rate": 4.4557584269662925e-05,
      "loss": 0.8747,
      "step": 1270
    },
    {
      "epoch": 0.02247230464017978,
      "grad_norm": 0.1320558927752511,
      "learning_rate": 4.4908707865168544e-05,
      "loss": 0.8751,
      "step": 1280
    },
    {
      "epoch": 0.022647869520181184,
      "grad_norm": 0.10185680177260303,
      "learning_rate": 4.525983146067416e-05,
      "loss": 0.8717,
      "step": 1290
    },
    {
      "epoch": 0.02282343440018259,
      "grad_norm": 0.14466921189363552,
      "learning_rate": 4.5610955056179776e-05,
      "loss": 0.8801,
      "step": 1300
    },
    {
      "epoch": 0.022998999280183993,
      "grad_norm": 0.1622565537783581,
      "learning_rate": 4.5962078651685396e-05,
      "loss": 0.883,
      "step": 1310
    },
    {
      "epoch": 0.023174564160185398,
      "grad_norm": 0.12906234720991022,
      "learning_rate": 4.6313202247191015e-05,
      "loss": 0.884,
      "step": 1320
    },
    {
      "epoch": 0.023350129040186803,
      "grad_norm": 0.0823508745676725,
      "learning_rate": 4.6664325842696635e-05,
      "loss": 0.8718,
      "step": 1330
    },
    {
      "epoch": 0.023525693920188207,
      "grad_norm": 0.06900689900788833,
      "learning_rate": 4.701544943820225e-05,
      "loss": 0.8714,
      "step": 1340
    },
    {
      "epoch": 0.02370125880018961,
      "grad_norm": 0.10270879313362849,
      "learning_rate": 4.736657303370787e-05,
      "loss": 0.8774,
      "step": 1350
    },
    {
      "epoch": 0.023876823680191013,
      "grad_norm": 0.10273776368395167,
      "learning_rate": 4.7717696629213486e-05,
      "loss": 0.8826,
      "step": 1360
    },
    {
      "epoch": 0.024052388560192418,
      "grad_norm": 0.09596988299799603,
      "learning_rate": 4.80688202247191e-05,
      "loss": 0.8798,
      "step": 1370
    },
    {
      "epoch": 0.024227953440193822,
      "grad_norm": 0.11847816310379901,
      "learning_rate": 4.8419943820224725e-05,
      "loss": 0.8729,
      "step": 1380
    },
    {
      "epoch": 0.024403518320195227,
      "grad_norm": 0.10770913945601095,
      "learning_rate": 4.877106741573034e-05,
      "loss": 0.8725,
      "step": 1390
    },
    {
      "epoch": 0.02457908320019663,
      "grad_norm": 0.09760188151512993,
      "learning_rate": 4.912219101123596e-05,
      "loss": 0.8782,
      "step": 1400
    },
    {
      "epoch": 0.024754648080198036,
      "grad_norm": 0.12070470238690403,
      "learning_rate": 4.947331460674158e-05,
      "loss": 0.8765,
      "step": 1410
    },
    {
      "epoch": 0.02493021296019944,
      "grad_norm": 0.08369159988123433,
      "learning_rate": 4.982443820224719e-05,
      "loss": 0.873,
      "step": 1420
    },
    {
      "epoch": 0.025105777840200846,
      "grad_norm": 0.11018167891867899,
      "learning_rate": 5.0175561797752815e-05,
      "loss": 0.8772,
      "step": 1430
    },
    {
      "epoch": 0.02528134272020225,
      "grad_norm": 0.11718188814986187,
      "learning_rate": 5.0526685393258435e-05,
      "loss": 0.8794,
      "step": 1440
    },
    {
      "epoch": 0.025456907600203655,
      "grad_norm": 0.12314955573510664,
      "learning_rate": 5.087780898876404e-05,
      "loss": 0.8754,
      "step": 1450
    },
    {
      "epoch": 0.02563247248020506,
      "grad_norm": 0.12931857851416387,
      "learning_rate": 5.122893258426966e-05,
      "loss": 0.8775,
      "step": 1460
    },
    {
      "epoch": 0.025808037360206464,
      "grad_norm": 0.07576528257708266,
      "learning_rate": 5.158005617977528e-05,
      "loss": 0.8762,
      "step": 1470
    },
    {
      "epoch": 0.02598360224020787,
      "grad_norm": 0.10621841321010858,
      "learning_rate": 5.1931179775280906e-05,
      "loss": 0.8832,
      "step": 1480
    },
    {
      "epoch": 0.026159167120209274,
      "grad_norm": 0.11818241117123471,
      "learning_rate": 5.2282303370786525e-05,
      "loss": 0.8801,
      "step": 1490
    },
    {
      "epoch": 0.02633473200021068,
      "grad_norm": 0.09873292829166555,
      "learning_rate": 5.263342696629213e-05,
      "loss": 0.8743,
      "step": 1500
    },
    {
      "epoch": 0.026510296880212083,
      "grad_norm": 0.1140487973108858,
      "learning_rate": 5.298455056179775e-05,
      "loss": 0.878,
      "step": 1510
    },
    {
      "epoch": 0.026685861760213488,
      "grad_norm": 0.08472200669063173,
      "learning_rate": 5.333567415730337e-05,
      "loss": 0.8773,
      "step": 1520
    },
    {
      "epoch": 0.026861426640214892,
      "grad_norm": 0.108810534675764,
      "learning_rate": 5.368679775280899e-05,
      "loss": 0.8721,
      "step": 1530
    },
    {
      "epoch": 0.027036991520216297,
      "grad_norm": 0.09625556890133662,
      "learning_rate": 5.4037921348314616e-05,
      "loss": 0.8738,
      "step": 1540
    },
    {
      "epoch": 0.0272125564002177,
      "grad_norm": 0.07747487756307622,
      "learning_rate": 5.438904494382022e-05,
      "loss": 0.8717,
      "step": 1550
    },
    {
      "epoch": 0.027388121280219106,
      "grad_norm": 0.10360135602517416,
      "learning_rate": 5.474016853932584e-05,
      "loss": 0.8707,
      "step": 1560
    },
    {
      "epoch": 0.02756368616022051,
      "grad_norm": 0.09995946461117297,
      "learning_rate": 5.509129213483146e-05,
      "loss": 0.8715,
      "step": 1570
    },
    {
      "epoch": 0.027739251040221916,
      "grad_norm": 0.1213715210921303,
      "learning_rate": 5.544241573033708e-05,
      "loss": 0.874,
      "step": 1580
    },
    {
      "epoch": 0.02791481592022332,
      "grad_norm": 0.13437138004106108,
      "learning_rate": 5.5793539325842706e-05,
      "loss": 0.8792,
      "step": 1590
    },
    {
      "epoch": 0.02809038080022472,
      "grad_norm": 0.11119987780625132,
      "learning_rate": 5.614466292134831e-05,
      "loss": 0.8759,
      "step": 1600
    },
    {
      "epoch": 0.028265945680226126,
      "grad_norm": 0.1465789182089292,
      "learning_rate": 5.649578651685393e-05,
      "loss": 0.8739,
      "step": 1610
    },
    {
      "epoch": 0.02844151056022753,
      "grad_norm": 0.12862822684017766,
      "learning_rate": 5.684691011235955e-05,
      "loss": 0.8765,
      "step": 1620
    },
    {
      "epoch": 0.028617075440228935,
      "grad_norm": 0.13542549581629815,
      "learning_rate": 5.719803370786517e-05,
      "loss": 0.8761,
      "step": 1630
    },
    {
      "epoch": 0.02879264032023034,
      "grad_norm": 0.11928026764126487,
      "learning_rate": 5.7549157303370796e-05,
      "loss": 0.8784,
      "step": 1640
    },
    {
      "epoch": 0.028968205200231745,
      "grad_norm": 0.0763259314761483,
      "learning_rate": 5.79002808988764e-05,
      "loss": 0.881,
      "step": 1650
    },
    {
      "epoch": 0.02914377008023315,
      "grad_norm": 0.07782812695027325,
      "learning_rate": 5.825140449438202e-05,
      "loss": 0.8739,
      "step": 1660
    },
    {
      "epoch": 0.029319334960234554,
      "grad_norm": 0.13227242239160308,
      "learning_rate": 5.860252808988764e-05,
      "loss": 0.8755,
      "step": 1670
    },
    {
      "epoch": 0.02949489984023596,
      "grad_norm": 0.09658700802290138,
      "learning_rate": 5.895365168539326e-05,
      "loss": 0.8747,
      "step": 1680
    },
    {
      "epoch": 0.029670464720237363,
      "grad_norm": 0.12127904263227096,
      "learning_rate": 5.930477528089888e-05,
      "loss": 0.8669,
      "step": 1690
    },
    {
      "epoch": 0.029846029600238768,
      "grad_norm": 0.15261930954246689,
      "learning_rate": 5.965589887640449e-05,
      "loss": 0.8819,
      "step": 1700
    },
    {
      "epoch": 0.030021594480240173,
      "grad_norm": 0.09311605073275837,
      "learning_rate": 6.000702247191011e-05,
      "loss": 0.8753,
      "step": 1710
    },
    {
      "epoch": 0.030197159360241577,
      "grad_norm": 0.1603394526918929,
      "learning_rate": 6.035814606741573e-05,
      "loss": 0.8677,
      "step": 1720
    },
    {
      "epoch": 0.030372724240242982,
      "grad_norm": 0.10052514144643267,
      "learning_rate": 6.070926966292135e-05,
      "loss": 0.8769,
      "step": 1730
    },
    {
      "epoch": 0.030548289120244387,
      "grad_norm": 0.1114893564591888,
      "learning_rate": 6.106039325842698e-05,
      "loss": 0.8754,
      "step": 1740
    },
    {
      "epoch": 0.03072385400024579,
      "grad_norm": 0.07851976586212815,
      "learning_rate": 6.141151685393258e-05,
      "loss": 0.8779,
      "step": 1750
    },
    {
      "epoch": 0.030899418880247196,
      "grad_norm": 0.09385890866527777,
      "learning_rate": 6.17626404494382e-05,
      "loss": 0.8724,
      "step": 1760
    },
    {
      "epoch": 0.0310749837602486,
      "grad_norm": 0.07885972168006505,
      "learning_rate": 6.211376404494383e-05,
      "loss": 0.8804,
      "step": 1770
    },
    {
      "epoch": 0.03125054864025,
      "grad_norm": 0.07208905134588864,
      "learning_rate": 6.246488764044944e-05,
      "loss": 0.8699,
      "step": 1780
    },
    {
      "epoch": 0.03142611352025141,
      "grad_norm": 0.08292397779923719,
      "learning_rate": 6.281601123595507e-05,
      "loss": 0.8778,
      "step": 1790
    },
    {
      "epoch": 0.03160167840025281,
      "grad_norm": 0.08427477325386275,
      "learning_rate": 6.316713483146067e-05,
      "loss": 0.8805,
      "step": 1800
    },
    {
      "epoch": 0.03177724328025422,
      "grad_norm": 0.12272744268665463,
      "learning_rate": 6.351825842696629e-05,
      "loss": 0.8688,
      "step": 1810
    },
    {
      "epoch": 0.03195280816025562,
      "grad_norm": 0.07212266832330874,
      "learning_rate": 6.386938202247192e-05,
      "loss": 0.8792,
      "step": 1820
    },
    {
      "epoch": 0.03212837304025703,
      "grad_norm": 0.09181690769470355,
      "learning_rate": 6.422050561797753e-05,
      "loss": 0.8725,
      "step": 1830
    },
    {
      "epoch": 0.03230393792025843,
      "grad_norm": 0.08977185572727775,
      "learning_rate": 6.457162921348316e-05,
      "loss": 0.8692,
      "step": 1840
    },
    {
      "epoch": 0.03247950280025984,
      "grad_norm": 0.0930865006899593,
      "learning_rate": 6.492275280898876e-05,
      "loss": 0.8722,
      "step": 1850
    },
    {
      "epoch": 0.03265506768026124,
      "grad_norm": 0.0960602121051624,
      "learning_rate": 6.527387640449438e-05,
      "loss": 0.8731,
      "step": 1860
    },
    {
      "epoch": 0.03283063256026265,
      "grad_norm": 0.08424816990217762,
      "learning_rate": 6.562500000000001e-05,
      "loss": 0.8775,
      "step": 1870
    },
    {
      "epoch": 0.03300619744026405,
      "grad_norm": 0.1428367714092643,
      "learning_rate": 6.597612359550562e-05,
      "loss": 0.8723,
      "step": 1880
    },
    {
      "epoch": 0.03318176232026546,
      "grad_norm": 0.09339724033251974,
      "learning_rate": 6.632724719101125e-05,
      "loss": 0.8816,
      "step": 1890
    },
    {
      "epoch": 0.03335732720026686,
      "grad_norm": 0.1149360897725946,
      "learning_rate": 6.667837078651685e-05,
      "loss": 0.8671,
      "step": 1900
    },
    {
      "epoch": 0.033532892080268266,
      "grad_norm": 0.07463221208225414,
      "learning_rate": 6.702949438202247e-05,
      "loss": 0.8744,
      "step": 1910
    },
    {
      "epoch": 0.03370845696026967,
      "grad_norm": 0.09317303782517729,
      "learning_rate": 6.73806179775281e-05,
      "loss": 0.8688,
      "step": 1920
    },
    {
      "epoch": 0.033884021840271075,
      "grad_norm": 0.0827646211262487,
      "learning_rate": 6.773174157303371e-05,
      "loss": 0.867,
      "step": 1930
    },
    {
      "epoch": 0.034059586720272476,
      "grad_norm": 0.11607234004724722,
      "learning_rate": 6.808286516853934e-05,
      "loss": 0.8789,
      "step": 1940
    },
    {
      "epoch": 0.034235151600273885,
      "grad_norm": 0.08829585851860171,
      "learning_rate": 6.843398876404494e-05,
      "loss": 0.8668,
      "step": 1950
    },
    {
      "epoch": 0.034410716480275286,
      "grad_norm": 0.08630338550914773,
      "learning_rate": 6.878511235955056e-05,
      "loss": 0.877,
      "step": 1960
    },
    {
      "epoch": 0.03458628136027669,
      "grad_norm": 0.1163233532454095,
      "learning_rate": 6.913623595505618e-05,
      "loss": 0.8714,
      "step": 1970
    },
    {
      "epoch": 0.034761846240278095,
      "grad_norm": 0.12675596910350276,
      "learning_rate": 6.94873595505618e-05,
      "loss": 0.8774,
      "step": 1980
    },
    {
      "epoch": 0.034937411120279496,
      "grad_norm": 0.09444714022240416,
      "learning_rate": 6.983848314606743e-05,
      "loss": 0.8744,
      "step": 1990
    },
    {
      "epoch": 0.035112976000280904,
      "grad_norm": 0.08960441180721453,
      "learning_rate": 7.018960674157303e-05,
      "loss": 0.8808,
      "step": 2000
    },
    {
      "epoch": 0.035288540880282306,
      "grad_norm": 0.0998993091001679,
      "learning_rate": 7.054073033707865e-05,
      "loss": 0.8781,
      "step": 2010
    },
    {
      "epoch": 0.035464105760283714,
      "grad_norm": 0.11235014703105918,
      "learning_rate": 7.089185393258427e-05,
      "loss": 0.8783,
      "step": 2020
    },
    {
      "epoch": 0.035639670640285115,
      "grad_norm": 0.08286826522891387,
      "learning_rate": 7.12429775280899e-05,
      "loss": 0.8769,
      "step": 2030
    },
    {
      "epoch": 0.03581523552028652,
      "grad_norm": 0.10936917717514412,
      "learning_rate": 7.159410112359552e-05,
      "loss": 0.8748,
      "step": 2040
    },
    {
      "epoch": 0.035990800400287924,
      "grad_norm": 0.12021495932110862,
      "learning_rate": 7.194522471910112e-05,
      "loss": 0.8688,
      "step": 2050
    },
    {
      "epoch": 0.03616636528028933,
      "grad_norm": 0.0866737134060902,
      "learning_rate": 7.229634831460674e-05,
      "loss": 0.8705,
      "step": 2060
    },
    {
      "epoch": 0.036341930160290734,
      "grad_norm": 0.09024807003894653,
      "learning_rate": 7.264747191011236e-05,
      "loss": 0.8786,
      "step": 2070
    },
    {
      "epoch": 0.03651749504029214,
      "grad_norm": 0.11950063547577987,
      "learning_rate": 7.299859550561798e-05,
      "loss": 0.8701,
      "step": 2080
    },
    {
      "epoch": 0.03669305992029354,
      "grad_norm": 0.1302159495426378,
      "learning_rate": 7.334971910112361e-05,
      "loss": 0.8702,
      "step": 2090
    },
    {
      "epoch": 0.03686862480029495,
      "grad_norm": 0.1106051898767845,
      "learning_rate": 7.370084269662921e-05,
      "loss": 0.8731,
      "step": 2100
    },
    {
      "epoch": 0.03704418968029635,
      "grad_norm": 0.12559592446553916,
      "learning_rate": 7.405196629213484e-05,
      "loss": 0.875,
      "step": 2110
    },
    {
      "epoch": 0.03721975456029776,
      "grad_norm": 0.0744267921977051,
      "learning_rate": 7.440308988764045e-05,
      "loss": 0.8666,
      "step": 2120
    },
    {
      "epoch": 0.03739531944029916,
      "grad_norm": 0.08962407126469685,
      "learning_rate": 7.475421348314607e-05,
      "loss": 0.8716,
      "step": 2130
    },
    {
      "epoch": 0.03757088432030057,
      "grad_norm": 0.09835657623995289,
      "learning_rate": 7.51053370786517e-05,
      "loss": 0.8701,
      "step": 2140
    },
    {
      "epoch": 0.03774644920030197,
      "grad_norm": 0.07510032325589777,
      "learning_rate": 7.54564606741573e-05,
      "loss": 0.8651,
      "step": 2150
    },
    {
      "epoch": 0.03792201408030338,
      "grad_norm": 0.1279588603520965,
      "learning_rate": 7.580758426966293e-05,
      "loss": 0.8747,
      "step": 2160
    },
    {
      "epoch": 0.03809757896030478,
      "grad_norm": 0.0836110324331831,
      "learning_rate": 7.615870786516854e-05,
      "loss": 0.8704,
      "step": 2170
    },
    {
      "epoch": 0.03827314384030619,
      "grad_norm": 0.07018143541902473,
      "learning_rate": 7.650983146067416e-05,
      "loss": 0.8827,
      "step": 2180
    },
    {
      "epoch": 0.03844870872030759,
      "grad_norm": 0.08837340260940743,
      "learning_rate": 7.686095505617979e-05,
      "loss": 0.8718,
      "step": 2190
    },
    {
      "epoch": 0.038624273600309,
      "grad_norm": 0.08585956793197065,
      "learning_rate": 7.721207865168539e-05,
      "loss": 0.875,
      "step": 2200
    },
    {
      "epoch": 0.0387998384803104,
      "grad_norm": 0.09076681816308635,
      "learning_rate": 7.756320224719102e-05,
      "loss": 0.8745,
      "step": 2210
    },
    {
      "epoch": 0.0389754033603118,
      "grad_norm": 0.0900400659383941,
      "learning_rate": 7.791432584269663e-05,
      "loss": 0.8734,
      "step": 2220
    },
    {
      "epoch": 0.03915096824031321,
      "grad_norm": 0.11490939261904487,
      "learning_rate": 7.826544943820225e-05,
      "loss": 0.8773,
      "step": 2230
    },
    {
      "epoch": 0.03932653312031461,
      "grad_norm": 0.09547733989058134,
      "learning_rate": 7.861657303370788e-05,
      "loss": 0.871,
      "step": 2240
    },
    {
      "epoch": 0.03950209800031602,
      "grad_norm": 0.09366012973645699,
      "learning_rate": 7.896769662921348e-05,
      "loss": 0.8789,
      "step": 2250
    },
    {
      "epoch": 0.03967766288031742,
      "grad_norm": 0.13061623207691203,
      "learning_rate": 7.93188202247191e-05,
      "loss": 0.8701,
      "step": 2260
    },
    {
      "epoch": 0.03985322776031883,
      "grad_norm": 0.09396449884425516,
      "learning_rate": 7.966994382022472e-05,
      "loss": 0.8674,
      "step": 2270
    },
    {
      "epoch": 0.04002879264032023,
      "grad_norm": 0.081202074367791,
      "learning_rate": 8.002106741573035e-05,
      "loss": 0.8708,
      "step": 2280
    },
    {
      "epoch": 0.040204357520321636,
      "grad_norm": 0.08152083969532947,
      "learning_rate": 8.037219101123596e-05,
      "loss": 0.8751,
      "step": 2290
    },
    {
      "epoch": 0.04037992240032304,
      "grad_norm": 0.10445825729166426,
      "learning_rate": 8.072331460674157e-05,
      "loss": 0.8746,
      "step": 2300
    },
    {
      "epoch": 0.040555487280324445,
      "grad_norm": 0.13780276532848207,
      "learning_rate": 8.10744382022472e-05,
      "loss": 0.8714,
      "step": 2310
    },
    {
      "epoch": 0.04073105216032585,
      "grad_norm": 0.07835817898931924,
      "learning_rate": 8.142556179775281e-05,
      "loss": 0.8736,
      "step": 2320
    },
    {
      "epoch": 0.040906617040327255,
      "grad_norm": 0.07331873240648742,
      "learning_rate": 8.177668539325844e-05,
      "loss": 0.8731,
      "step": 2330
    },
    {
      "epoch": 0.041082181920328656,
      "grad_norm": 0.11721648321005346,
      "learning_rate": 8.212780898876405e-05,
      "loss": 0.8733,
      "step": 2340
    },
    {
      "epoch": 0.041257746800330064,
      "grad_norm": 0.1719497813435053,
      "learning_rate": 8.247893258426966e-05,
      "loss": 0.8791,
      "step": 2350
    },
    {
      "epoch": 0.041433311680331465,
      "grad_norm": 0.08148721441455263,
      "learning_rate": 8.283005617977529e-05,
      "loss": 0.8723,
      "step": 2360
    },
    {
      "epoch": 0.04160887656033287,
      "grad_norm": 0.10026605180058305,
      "learning_rate": 8.31811797752809e-05,
      "loss": 0.8748,
      "step": 2370
    },
    {
      "epoch": 0.041784441440334275,
      "grad_norm": 0.11754014286672843,
      "learning_rate": 8.353230337078653e-05,
      "loss": 0.8782,
      "step": 2380
    },
    {
      "epoch": 0.04196000632033568,
      "grad_norm": 0.10731994918529643,
      "learning_rate": 8.388342696629214e-05,
      "loss": 0.8802,
      "step": 2390
    },
    {
      "epoch": 0.042135571200337084,
      "grad_norm": 0.10859563034401212,
      "learning_rate": 8.423455056179775e-05,
      "loss": 0.8689,
      "step": 2400
    },
    {
      "epoch": 0.04231113608033849,
      "grad_norm": 0.09597930546324353,
      "learning_rate": 8.458567415730338e-05,
      "loss": 0.8696,
      "step": 2410
    },
    {
      "epoch": 0.04248670096033989,
      "grad_norm": 0.10547186388674545,
      "learning_rate": 8.493679775280899e-05,
      "loss": 0.8787,
      "step": 2420
    },
    {
      "epoch": 0.0426622658403413,
      "grad_norm": 0.09348780419089932,
      "learning_rate": 8.528792134831462e-05,
      "loss": 0.8692,
      "step": 2430
    },
    {
      "epoch": 0.0428378307203427,
      "grad_norm": 0.08611926721115874,
      "learning_rate": 8.563904494382023e-05,
      "loss": 0.8722,
      "step": 2440
    },
    {
      "epoch": 0.043013395600344104,
      "grad_norm": 0.11653422929769607,
      "learning_rate": 8.599016853932584e-05,
      "loss": 0.8776,
      "step": 2450
    },
    {
      "epoch": 0.04318896048034551,
      "grad_norm": 0.1111063126732818,
      "learning_rate": 8.634129213483147e-05,
      "loss": 0.8705,
      "step": 2460
    },
    {
      "epoch": 0.04336452536034691,
      "grad_norm": 0.14743627127639225,
      "learning_rate": 8.669241573033708e-05,
      "loss": 0.8715,
      "step": 2470
    },
    {
      "epoch": 0.04354009024034832,
      "grad_norm": 0.1007333490386571,
      "learning_rate": 8.70435393258427e-05,
      "loss": 0.8733,
      "step": 2480
    },
    {
      "epoch": 0.04371565512034972,
      "grad_norm": 0.07123765213968385,
      "learning_rate": 8.739466292134832e-05,
      "loss": 0.8701,
      "step": 2490
    },
    {
      "epoch": 0.04389122000035113,
      "grad_norm": 0.09039947304214216,
      "learning_rate": 8.774578651685393e-05,
      "loss": 0.8852,
      "step": 2500
    },
    {
      "epoch": 0.04406678488035253,
      "grad_norm": 0.0877478104769651,
      "learning_rate": 8.809691011235956e-05,
      "loss": 0.8696,
      "step": 2510
    },
    {
      "epoch": 0.04424234976035394,
      "grad_norm": 0.07930286645977104,
      "learning_rate": 8.844803370786517e-05,
      "loss": 0.8661,
      "step": 2520
    },
    {
      "epoch": 0.04441791464035534,
      "grad_norm": 0.15239307871870406,
      "learning_rate": 8.87991573033708e-05,
      "loss": 0.8696,
      "step": 2530
    },
    {
      "epoch": 0.04459347952035675,
      "grad_norm": 0.0661294853551783,
      "learning_rate": 8.915028089887641e-05,
      "loss": 0.8713,
      "step": 2540
    },
    {
      "epoch": 0.04476904440035815,
      "grad_norm": 0.07691426284051446,
      "learning_rate": 8.950140449438202e-05,
      "loss": 0.871,
      "step": 2550
    },
    {
      "epoch": 0.04494460928035956,
      "grad_norm": 0.09285817049029198,
      "learning_rate": 8.985252808988765e-05,
      "loss": 0.8618,
      "step": 2560
    },
    {
      "epoch": 0.04512017416036096,
      "grad_norm": 0.1445573685383954,
      "learning_rate": 9.020365168539326e-05,
      "loss": 0.8722,
      "step": 2570
    },
    {
      "epoch": 0.04529573904036237,
      "grad_norm": 0.09214466523692007,
      "learning_rate": 9.055477528089889e-05,
      "loss": 0.8641,
      "step": 2580
    },
    {
      "epoch": 0.04547130392036377,
      "grad_norm": 0.08881843255206193,
      "learning_rate": 9.09058988764045e-05,
      "loss": 0.8712,
      "step": 2590
    },
    {
      "epoch": 0.04564686880036518,
      "grad_norm": 0.09426639476722427,
      "learning_rate": 9.125702247191011e-05,
      "loss": 0.8687,
      "step": 2600
    },
    {
      "epoch": 0.04582243368036658,
      "grad_norm": 0.08790925243557784,
      "learning_rate": 9.160814606741574e-05,
      "loss": 0.8703,
      "step": 2610
    },
    {
      "epoch": 0.045997998560367986,
      "grad_norm": 0.1344042728083833,
      "learning_rate": 9.195926966292135e-05,
      "loss": 0.8747,
      "step": 2620
    },
    {
      "epoch": 0.04617356344036939,
      "grad_norm": 0.0819882931403838,
      "learning_rate": 9.231039325842698e-05,
      "loss": 0.869,
      "step": 2630
    },
    {
      "epoch": 0.046349128320370796,
      "grad_norm": 0.07695996431276211,
      "learning_rate": 9.266151685393259e-05,
      "loss": 0.8763,
      "step": 2640
    },
    {
      "epoch": 0.0465246932003722,
      "grad_norm": 0.12103296580832704,
      "learning_rate": 9.30126404494382e-05,
      "loss": 0.8759,
      "step": 2650
    },
    {
      "epoch": 0.046700258080373605,
      "grad_norm": 0.10053373488238627,
      "learning_rate": 9.336376404494383e-05,
      "loss": 0.8628,
      "step": 2660
    },
    {
      "epoch": 0.046875822960375006,
      "grad_norm": 0.09363301253517695,
      "learning_rate": 9.371488764044944e-05,
      "loss": 0.8813,
      "step": 2670
    },
    {
      "epoch": 0.047051387840376414,
      "grad_norm": 0.10884568097127366,
      "learning_rate": 9.406601123595507e-05,
      "loss": 0.8715,
      "step": 2680
    },
    {
      "epoch": 0.047226952720377816,
      "grad_norm": 0.09105824767723958,
      "learning_rate": 9.441713483146067e-05,
      "loss": 0.868,
      "step": 2690
    },
    {
      "epoch": 0.04740251760037922,
      "grad_norm": 0.07610597950464966,
      "learning_rate": 9.47682584269663e-05,
      "loss": 0.873,
      "step": 2700
    },
    {
      "epoch": 0.047578082480380625,
      "grad_norm": 0.10464935004067749,
      "learning_rate": 9.511938202247192e-05,
      "loss": 0.8716,
      "step": 2710
    },
    {
      "epoch": 0.047753647360382026,
      "grad_norm": 0.06946899076922435,
      "learning_rate": 9.547050561797753e-05,
      "loss": 0.8684,
      "step": 2720
    },
    {
      "epoch": 0.047929212240383434,
      "grad_norm": 0.08061928733635612,
      "learning_rate": 9.582162921348316e-05,
      "loss": 0.8782,
      "step": 2730
    },
    {
      "epoch": 0.048104777120384835,
      "grad_norm": 0.11139900903899783,
      "learning_rate": 9.617275280898876e-05,
      "loss": 0.8867,
      "step": 2740
    },
    {
      "epoch": 0.048280342000386244,
      "grad_norm": 0.10383002385475375,
      "learning_rate": 9.652387640449438e-05,
      "loss": 0.8675,
      "step": 2750
    },
    {
      "epoch": 0.048455906880387645,
      "grad_norm": 0.0742451160080383,
      "learning_rate": 9.687500000000001e-05,
      "loss": 0.8719,
      "step": 2760
    },
    {
      "epoch": 0.04863147176038905,
      "grad_norm": 0.08339560204858726,
      "learning_rate": 9.722612359550562e-05,
      "loss": 0.874,
      "step": 2770
    },
    {
      "epoch": 0.048807036640390454,
      "grad_norm": 0.0886617007520003,
      "learning_rate": 9.757724719101125e-05,
      "loss": 0.8758,
      "step": 2780
    },
    {
      "epoch": 0.04898260152039186,
      "grad_norm": 0.07014192220796461,
      "learning_rate": 9.792837078651685e-05,
      "loss": 0.8703,
      "step": 2790
    },
    {
      "epoch": 0.04915816640039326,
      "grad_norm": 0.10642527077501172,
      "learning_rate": 9.827949438202247e-05,
      "loss": 0.8716,
      "step": 2800
    },
    {
      "epoch": 0.04933373128039467,
      "grad_norm": 0.12543116445892447,
      "learning_rate": 9.86306179775281e-05,
      "loss": 0.8704,
      "step": 2810
    },
    {
      "epoch": 0.04950929616039607,
      "grad_norm": 0.09696623122489874,
      "learning_rate": 9.898174157303371e-05,
      "loss": 0.8686,
      "step": 2820
    },
    {
      "epoch": 0.04968486104039748,
      "grad_norm": 0.08364880955280966,
      "learning_rate": 9.933286516853934e-05,
      "loss": 0.8677,
      "step": 2830
    },
    {
      "epoch": 0.04986042592039888,
      "grad_norm": 0.07627651929749238,
      "learning_rate": 9.968398876404494e-05,
      "loss": 0.8654,
      "step": 2840
    },
    {
      "epoch": 0.05003599080040029,
      "grad_norm": 0.07662343368914502,
      "learning_rate": 0.00010003511235955056,
      "loss": 0.8706,
      "step": 2850
    },
    {
      "epoch": 0.05021155568040169,
      "grad_norm": 0.09355597457773152,
      "learning_rate": 0.00010038623595505619,
      "loss": 0.8746,
      "step": 2860
    },
    {
      "epoch": 0.0503871205604031,
      "grad_norm": 0.14354171802669177,
      "learning_rate": 0.0001007373595505618,
      "loss": 0.8686,
      "step": 2870
    },
    {
      "epoch": 0.0505626854404045,
      "grad_norm": 0.08260488112384638,
      "learning_rate": 0.00010108848314606743,
      "loss": 0.865,
      "step": 2880
    },
    {
      "epoch": 0.05073825032040591,
      "grad_norm": 0.12336890236691185,
      "learning_rate": 0.00010143960674157304,
      "loss": 0.8682,
      "step": 2890
    },
    {
      "epoch": 0.05091381520040731,
      "grad_norm": 0.09619293620987679,
      "learning_rate": 0.00010179073033707867,
      "loss": 0.8657,
      "step": 2900
    },
    {
      "epoch": 0.05108938008040872,
      "grad_norm": 0.09299525913420573,
      "learning_rate": 0.00010214185393258427,
      "loss": 0.8702,
      "step": 2910
    },
    {
      "epoch": 0.05126494496041012,
      "grad_norm": 0.10287830534293611,
      "learning_rate": 0.00010249297752808988,
      "loss": 0.868,
      "step": 2920
    },
    {
      "epoch": 0.05144050984041153,
      "grad_norm": 0.06290306939053397,
      "learning_rate": 0.0001028441011235955,
      "loss": 0.8528,
      "step": 2930
    },
    {
      "epoch": 0.05161607472041293,
      "grad_norm": 0.08896019860214284,
      "learning_rate": 0.00010319522471910112,
      "loss": 0.8734,
      "step": 2940
    },
    {
      "epoch": 0.05179163960041433,
      "grad_norm": 0.08591212361928517,
      "learning_rate": 0.00010354634831460675,
      "loss": 0.8659,
      "step": 2950
    },
    {
      "epoch": 0.05196720448041574,
      "grad_norm": 0.10448251442617819,
      "learning_rate": 0.00010389747191011237,
      "loss": 0.8653,
      "step": 2960
    },
    {
      "epoch": 0.05214276936041714,
      "grad_norm": 0.0808799666528056,
      "learning_rate": 0.00010424859550561798,
      "loss": 0.8733,
      "step": 2970
    },
    {
      "epoch": 0.05231833424041855,
      "grad_norm": 0.0940345095307922,
      "learning_rate": 0.00010459971910112361,
      "loss": 0.8648,
      "step": 2980
    },
    {
      "epoch": 0.05249389912041995,
      "grad_norm": 0.07052264219731792,
      "learning_rate": 0.00010495084269662922,
      "loss": 0.8725,
      "step": 2990
    },
    {
      "epoch": 0.05266946400042136,
      "grad_norm": 0.07885099466722859,
      "learning_rate": 0.00010530196629213485,
      "loss": 0.881,
      "step": 3000
    },
    {
      "epoch": 0.05284502888042276,
      "grad_norm": 0.0754634601812287,
      "learning_rate": 0.00010565308988764045,
      "loss": 0.873,
      "step": 3010
    },
    {
      "epoch": 0.053020593760424166,
      "grad_norm": 0.06814422978053401,
      "learning_rate": 0.00010600421348314606,
      "loss": 0.8665,
      "step": 3020
    },
    {
      "epoch": 0.05319615864042557,
      "grad_norm": 0.07786495136330572,
      "learning_rate": 0.00010635533707865169,
      "loss": 0.8673,
      "step": 3030
    },
    {
      "epoch": 0.053371723520426975,
      "grad_norm": 0.10207021620693676,
      "learning_rate": 0.0001067064606741573,
      "loss": 0.8735,
      "step": 3040
    },
    {
      "epoch": 0.053547288400428376,
      "grad_norm": 0.08564142990231689,
      "learning_rate": 0.00010705758426966293,
      "loss": 0.8689,
      "step": 3050
    },
    {
      "epoch": 0.053722853280429785,
      "grad_norm": 0.08114109637591382,
      "learning_rate": 0.00010740870786516854,
      "loss": 0.8679,
      "step": 3060
    },
    {
      "epoch": 0.053898418160431186,
      "grad_norm": 0.09546023065439535,
      "learning_rate": 0.00010775983146067417,
      "loss": 0.8693,
      "step": 3070
    },
    {
      "epoch": 0.054073983040432594,
      "grad_norm": 0.10860912735636992,
      "learning_rate": 0.00010811095505617979,
      "loss": 0.8712,
      "step": 3080
    },
    {
      "epoch": 0.054249547920433995,
      "grad_norm": 0.08801223964664094,
      "learning_rate": 0.0001084620786516854,
      "loss": 0.8682,
      "step": 3090
    },
    {
      "epoch": 0.0544251128004354,
      "grad_norm": 0.12172961077070571,
      "learning_rate": 0.00010881320224719103,
      "loss": 0.8699,
      "step": 3100
    },
    {
      "epoch": 0.054600677680436804,
      "grad_norm": 0.08241615340468754,
      "learning_rate": 0.00010916432584269663,
      "loss": 0.8674,
      "step": 3110
    },
    {
      "epoch": 0.05477624256043821,
      "grad_norm": 0.0865794758965367,
      "learning_rate": 0.00010951544943820224,
      "loss": 0.8822,
      "step": 3120
    },
    {
      "epoch": 0.054951807440439614,
      "grad_norm": 0.11719867689285764,
      "learning_rate": 0.00010986657303370787,
      "loss": 0.8656,
      "step": 3130
    },
    {
      "epoch": 0.05512737232044102,
      "grad_norm": 0.0766996241800066,
      "learning_rate": 0.00011021769662921348,
      "loss": 0.8705,
      "step": 3140
    },
    {
      "epoch": 0.05530293720044242,
      "grad_norm": 0.083148551011847,
      "learning_rate": 0.00011056882022471911,
      "loss": 0.8699,
      "step": 3150
    },
    {
      "epoch": 0.05547850208044383,
      "grad_norm": 0.09156534426029721,
      "learning_rate": 0.00011091994382022472,
      "loss": 0.8627,
      "step": 3160
    },
    {
      "epoch": 0.05565406696044523,
      "grad_norm": 0.07732744747798344,
      "learning_rate": 0.00011127106741573035,
      "loss": 0.8655,
      "step": 3170
    },
    {
      "epoch": 0.05582963184044664,
      "grad_norm": 0.07871542809557351,
      "learning_rate": 0.00011162219101123597,
      "loss": 0.8646,
      "step": 3180
    },
    {
      "epoch": 0.05600519672044804,
      "grad_norm": 0.06860547571216925,
      "learning_rate": 0.00011197331460674158,
      "loss": 0.8725,
      "step": 3190
    },
    {
      "epoch": 0.05618076160044944,
      "grad_norm": 0.09898952493015431,
      "learning_rate": 0.00011232443820224721,
      "loss": 0.8697,
      "step": 3200
    },
    {
      "epoch": 0.05635632648045085,
      "grad_norm": 0.10503791319987942,
      "learning_rate": 0.00011267556179775281,
      "loss": 0.8728,
      "step": 3210
    },
    {
      "epoch": 0.05653189136045225,
      "grad_norm": 0.113397681516741,
      "learning_rate": 0.00011302668539325842,
      "loss": 0.8692,
      "step": 3220
    },
    {
      "epoch": 0.05670745624045366,
      "grad_norm": 0.14823671882268247,
      "learning_rate": 0.00011337780898876405,
      "loss": 0.871,
      "step": 3230
    },
    {
      "epoch": 0.05688302112045506,
      "grad_norm": 0.08112289479811689,
      "learning_rate": 0.00011372893258426966,
      "loss": 0.8683,
      "step": 3240
    },
    {
      "epoch": 0.05705858600045647,
      "grad_norm": 0.06459786081889814,
      "learning_rate": 0.00011408005617977529,
      "loss": 0.8673,
      "step": 3250
    },
    {
      "epoch": 0.05723415088045787,
      "grad_norm": 0.09265222017632906,
      "learning_rate": 0.0001144311797752809,
      "loss": 0.869,
      "step": 3260
    },
    {
      "epoch": 0.05740971576045928,
      "grad_norm": 0.07766313734105677,
      "learning_rate": 0.00011478230337078653,
      "loss": 0.8756,
      "step": 3270
    },
    {
      "epoch": 0.05758528064046068,
      "grad_norm": 0.061711919032600045,
      "learning_rate": 0.00011513342696629215,
      "loss": 0.8662,
      "step": 3280
    },
    {
      "epoch": 0.05776084552046209,
      "grad_norm": 0.08438941281492143,
      "learning_rate": 0.00011548455056179777,
      "loss": 0.8728,
      "step": 3290
    },
    {
      "epoch": 0.05793641040046349,
      "grad_norm": 0.08242077296391899,
      "learning_rate": 0.00011583567415730339,
      "loss": 0.8586,
      "step": 3300
    },
    {
      "epoch": 0.0581119752804649,
      "grad_norm": 0.0924121867789103,
      "learning_rate": 0.00011618679775280899,
      "loss": 0.8731,
      "step": 3310
    },
    {
      "epoch": 0.0582875401604663,
      "grad_norm": 0.12274655377056946,
      "learning_rate": 0.0001165379213483146,
      "loss": 0.8644,
      "step": 3320
    },
    {
      "epoch": 0.05846310504046771,
      "grad_norm": 0.11344404884514878,
      "learning_rate": 0.00011688904494382023,
      "loss": 0.8673,
      "step": 3330
    },
    {
      "epoch": 0.05863866992046911,
      "grad_norm": 0.10983789543228578,
      "learning_rate": 0.00011724016853932584,
      "loss": 0.8707,
      "step": 3340
    },
    {
      "epoch": 0.058814234800470516,
      "grad_norm": 0.08721443834475422,
      "learning_rate": 0.00011759129213483147,
      "loss": 0.8695,
      "step": 3350
    },
    {
      "epoch": 0.05898979968047192,
      "grad_norm": 0.07856964717018013,
      "learning_rate": 0.00011794241573033708,
      "loss": 0.8685,
      "step": 3360
    },
    {
      "epoch": 0.059165364560473326,
      "grad_norm": 0.08079654160183043,
      "learning_rate": 0.00011829353932584271,
      "loss": 0.8669,
      "step": 3370
    },
    {
      "epoch": 0.05934092944047473,
      "grad_norm": 0.07066320977870835,
      "learning_rate": 0.00011864466292134832,
      "loss": 0.8656,
      "step": 3380
    },
    {
      "epoch": 0.059516494320476135,
      "grad_norm": 0.08974000064477358,
      "learning_rate": 0.00011899578651685395,
      "loss": 0.868,
      "step": 3390
    },
    {
      "epoch": 0.059692059200477536,
      "grad_norm": 0.09442930657707849,
      "learning_rate": 0.00011934691011235957,
      "loss": 0.8713,
      "step": 3400
    },
    {
      "epoch": 0.059867624080478944,
      "grad_norm": 0.09146779738026516,
      "learning_rate": 0.00011969803370786517,
      "loss": 0.8618,
      "step": 3410
    },
    {
      "epoch": 0.060043188960480345,
      "grad_norm": 0.06633853866889391,
      "learning_rate": 0.00012004915730337078,
      "loss": 0.8717,
      "step": 3420
    },
    {
      "epoch": 0.06021875384048175,
      "grad_norm": 0.06690691662235126,
      "learning_rate": 0.00012040028089887641,
      "loss": 0.8681,
      "step": 3430
    },
    {
      "epoch": 0.060394318720483155,
      "grad_norm": 0.06881263548470298,
      "learning_rate": 0.00012075140449438202,
      "loss": 0.8668,
      "step": 3440
    },
    {
      "epoch": 0.060569883600484556,
      "grad_norm": 0.07719744730115442,
      "learning_rate": 0.00012110252808988765,
      "loss": 0.8733,
      "step": 3450
    },
    {
      "epoch": 0.060745448480485964,
      "grad_norm": 0.07070604496656566,
      "learning_rate": 0.00012145365168539326,
      "loss": 0.8605,
      "step": 3460
    },
    {
      "epoch": 0.060921013360487365,
      "grad_norm": 0.09341392473875189,
      "learning_rate": 0.00012180477528089889,
      "loss": 0.8677,
      "step": 3470
    },
    {
      "epoch": 0.06109657824048877,
      "grad_norm": 0.0918776267489599,
      "learning_rate": 0.0001221558988764045,
      "loss": 0.8632,
      "step": 3480
    },
    {
      "epoch": 0.061272143120490175,
      "grad_norm": 0.1591291175325041,
      "learning_rate": 0.00012250702247191013,
      "loss": 0.8677,
      "step": 3490
    },
    {
      "epoch": 0.06144770800049158,
      "grad_norm": 0.08929748142315258,
      "learning_rate": 0.00012285814606741575,
      "loss": 0.8738,
      "step": 3500
    },
    {
      "epoch": 0.061623272880492984,
      "grad_norm": 0.09921196662287508,
      "learning_rate": 0.00012320926966292135,
      "loss": 0.8639,
      "step": 3510
    },
    {
      "epoch": 0.06179883776049439,
      "grad_norm": 0.06902446483157539,
      "learning_rate": 0.00012356039325842698,
      "loss": 0.868,
      "step": 3520
    },
    {
      "epoch": 0.06197440264049579,
      "grad_norm": 0.12024732140770085,
      "learning_rate": 0.00012391151685393258,
      "loss": 0.8741,
      "step": 3530
    },
    {
      "epoch": 0.0621499675204972,
      "grad_norm": 0.07097550208531664,
      "learning_rate": 0.0001242626404494382,
      "loss": 0.865,
      "step": 3540
    },
    {
      "epoch": 0.0623255324004986,
      "grad_norm": 0.08653807971129462,
      "learning_rate": 0.00012461376404494383,
      "loss": 0.8739,
      "step": 3550
    },
    {
      "epoch": 0.0625010972805,
      "grad_norm": 0.09075586971741764,
      "learning_rate": 0.00012496488764044946,
      "loss": 0.871,
      "step": 3560
    },
    {
      "epoch": 0.06267666216050141,
      "grad_norm": 0.08939137018393155,
      "learning_rate": 0.00012531601123595506,
      "loss": 0.8649,
      "step": 3570
    },
    {
      "epoch": 0.06285222704050282,
      "grad_norm": 0.1073606370412846,
      "learning_rate": 0.00012566713483146068,
      "loss": 0.8763,
      "step": 3580
    },
    {
      "epoch": 0.06302779192050423,
      "grad_norm": 0.09772752366937706,
      "learning_rate": 0.0001260182584269663,
      "loss": 0.8694,
      "step": 3590
    },
    {
      "epoch": 0.06320335680050562,
      "grad_norm": 0.10046149176689853,
      "learning_rate": 0.00012636938202247193,
      "loss": 0.8666,
      "step": 3600
    },
    {
      "epoch": 0.06337892168050703,
      "grad_norm": 0.09555825182241347,
      "learning_rate": 0.00012672050561797753,
      "loss": 0.8776,
      "step": 3610
    },
    {
      "epoch": 0.06355448656050844,
      "grad_norm": 0.07682773877827448,
      "learning_rate": 0.00012707162921348313,
      "loss": 0.8643,
      "step": 3620
    },
    {
      "epoch": 0.06373005144050985,
      "grad_norm": 0.08374957848003826,
      "learning_rate": 0.00012742275280898876,
      "loss": 0.8704,
      "step": 3630
    },
    {
      "epoch": 0.06390561632051124,
      "grad_norm": 0.08844890922689236,
      "learning_rate": 0.00012777387640449438,
      "loss": 0.8726,
      "step": 3640
    },
    {
      "epoch": 0.06408118120051265,
      "grad_norm": 0.09399630003450928,
      "learning_rate": 0.000128125,
      "loss": 0.8667,
      "step": 3650
    },
    {
      "epoch": 0.06425674608051406,
      "grad_norm": 0.07835107143620082,
      "learning_rate": 0.00012847612359550564,
      "loss": 0.8655,
      "step": 3660
    },
    {
      "epoch": 0.06443231096051547,
      "grad_norm": 0.08024847587044613,
      "learning_rate": 0.00012882724719101124,
      "loss": 0.8677,
      "step": 3670
    },
    {
      "epoch": 0.06460787584051686,
      "grad_norm": 0.07468031044444032,
      "learning_rate": 0.00012917837078651686,
      "loss": 0.8638,
      "step": 3680
    },
    {
      "epoch": 0.06478344072051827,
      "grad_norm": 0.08474352502966452,
      "learning_rate": 0.0001295294943820225,
      "loss": 0.8719,
      "step": 3690
    },
    {
      "epoch": 0.06495900560051968,
      "grad_norm": 0.0899589308903606,
      "learning_rate": 0.0001298806179775281,
      "loss": 0.8687,
      "step": 3700
    },
    {
      "epoch": 0.06513457048052107,
      "grad_norm": 0.0860965337425579,
      "learning_rate": 0.00013023174157303371,
      "loss": 0.8704,
      "step": 3710
    },
    {
      "epoch": 0.06531013536052248,
      "grad_norm": 0.08497707086525352,
      "learning_rate": 0.0001305828651685393,
      "loss": 0.8651,
      "step": 3720
    },
    {
      "epoch": 0.06548570024052389,
      "grad_norm": 0.0634556600068835,
      "learning_rate": 0.00013093398876404494,
      "loss": 0.8692,
      "step": 3730
    },
    {
      "epoch": 0.0656612651205253,
      "grad_norm": 0.08224388864604819,
      "learning_rate": 0.00013128511235955057,
      "loss": 0.8638,
      "step": 3740
    },
    {
      "epoch": 0.06583683000052669,
      "grad_norm": 0.07870191128854613,
      "learning_rate": 0.0001316362359550562,
      "loss": 0.8616,
      "step": 3750
    },
    {
      "epoch": 0.0660123948805281,
      "grad_norm": 0.08332245595149168,
      "learning_rate": 0.00013198735955056182,
      "loss": 0.8661,
      "step": 3760
    },
    {
      "epoch": 0.0661879597605295,
      "grad_norm": 0.08866736905289187,
      "learning_rate": 0.00013233848314606742,
      "loss": 0.8708,
      "step": 3770
    },
    {
      "epoch": 0.06636352464053091,
      "grad_norm": 0.07721877461615047,
      "learning_rate": 0.00013268960674157304,
      "loss": 0.874,
      "step": 3780
    },
    {
      "epoch": 0.06653908952053231,
      "grad_norm": 0.09724981795207698,
      "learning_rate": 0.00013304073033707867,
      "loss": 0.8749,
      "step": 3790
    },
    {
      "epoch": 0.06671465440053372,
      "grad_norm": 0.11893054087661478,
      "learning_rate": 0.00013339185393258427,
      "loss": 0.8755,
      "step": 3800
    },
    {
      "epoch": 0.06689021928053512,
      "grad_norm": 0.10137589262985941,
      "learning_rate": 0.0001337429775280899,
      "loss": 0.8697,
      "step": 3810
    },
    {
      "epoch": 0.06706578416053653,
      "grad_norm": 0.07216692843322385,
      "learning_rate": 0.0001340941011235955,
      "loss": 0.8695,
      "step": 3820
    },
    {
      "epoch": 0.06724134904053793,
      "grad_norm": 0.06748229359881665,
      "learning_rate": 0.00013444522471910112,
      "loss": 0.8731,
      "step": 3830
    },
    {
      "epoch": 0.06741691392053933,
      "grad_norm": 0.08631753256432514,
      "learning_rate": 0.00013479634831460675,
      "loss": 0.8666,
      "step": 3840
    },
    {
      "epoch": 0.06759247880054074,
      "grad_norm": 0.09485659253323388,
      "learning_rate": 0.00013514747191011237,
      "loss": 0.8609,
      "step": 3850
    },
    {
      "epoch": 0.06776804368054215,
      "grad_norm": 0.10023775477120957,
      "learning_rate": 0.000135498595505618,
      "loss": 0.8717,
      "step": 3860
    },
    {
      "epoch": 0.06794360856054354,
      "grad_norm": 0.0867118224133698,
      "learning_rate": 0.0001358497191011236,
      "loss": 0.8698,
      "step": 3870
    },
    {
      "epoch": 0.06811917344054495,
      "grad_norm": 0.1105233685981463,
      "learning_rate": 0.00013620084269662922,
      "loss": 0.8669,
      "step": 3880
    },
    {
      "epoch": 0.06829473832054636,
      "grad_norm": 0.0949806796508284,
      "learning_rate": 0.00013655196629213485,
      "loss": 0.8722,
      "step": 3890
    },
    {
      "epoch": 0.06847030320054777,
      "grad_norm": 0.08375678745285936,
      "learning_rate": 0.00013690308988764045,
      "loss": 0.8739,
      "step": 3900
    },
    {
      "epoch": 0.06864586808054916,
      "grad_norm": 0.0644081927936635,
      "learning_rate": 0.00013725421348314608,
      "loss": 0.8661,
      "step": 3910
    },
    {
      "epoch": 0.06882143296055057,
      "grad_norm": 0.09912616189707003,
      "learning_rate": 0.00013760533707865167,
      "loss": 0.8676,
      "step": 3920
    },
    {
      "epoch": 0.06899699784055198,
      "grad_norm": 0.12052094066945607,
      "learning_rate": 0.0001379564606741573,
      "loss": 0.8619,
      "step": 3930
    },
    {
      "epoch": 0.06917256272055337,
      "grad_norm": 0.09153774261297007,
      "learning_rate": 0.00013830758426966293,
      "loss": 0.8754,
      "step": 3940
    },
    {
      "epoch": 0.06934812760055478,
      "grad_norm": 0.0969384167605854,
      "learning_rate": 0.00013865870786516855,
      "loss": 0.8683,
      "step": 3950
    },
    {
      "epoch": 0.06952369248055619,
      "grad_norm": 0.08974666853226083,
      "learning_rate": 0.00013900983146067418,
      "loss": 0.8665,
      "step": 3960
    },
    {
      "epoch": 0.0696992573605576,
      "grad_norm": 0.09511105897158184,
      "learning_rate": 0.00013936095505617978,
      "loss": 0.8666,
      "step": 3970
    },
    {
      "epoch": 0.06987482224055899,
      "grad_norm": 0.10151448001847216,
      "learning_rate": 0.0001397120786516854,
      "loss": 0.8717,
      "step": 3980
    },
    {
      "epoch": 0.0700503871205604,
      "grad_norm": 0.08104035612694072,
      "learning_rate": 0.00014006320224719103,
      "loss": 0.8739,
      "step": 3990
    },
    {
      "epoch": 0.07022595200056181,
      "grad_norm": 0.07699135641787441,
      "learning_rate": 0.00014041432584269663,
      "loss": 0.8744,
      "step": 4000
    },
    {
      "epoch": 0.07040151688056322,
      "grad_norm": 0.07000581689066553,
      "learning_rate": 0.00014076544943820226,
      "loss": 0.86,
      "step": 4010
    },
    {
      "epoch": 0.07057708176056461,
      "grad_norm": 0.08855740751215302,
      "learning_rate": 0.00014111657303370786,
      "loss": 0.871,
      "step": 4020
    },
    {
      "epoch": 0.07075264664056602,
      "grad_norm": 0.08166610452330605,
      "learning_rate": 0.00014146769662921348,
      "loss": 0.8685,
      "step": 4030
    },
    {
      "epoch": 0.07092821152056743,
      "grad_norm": 0.11790191119551124,
      "learning_rate": 0.0001418188202247191,
      "loss": 0.8677,
      "step": 4040
    },
    {
      "epoch": 0.07110377640056884,
      "grad_norm": 0.13799654867093156,
      "learning_rate": 0.00014216994382022473,
      "loss": 0.8582,
      "step": 4050
    },
    {
      "epoch": 0.07127934128057023,
      "grad_norm": 0.07327678821618046,
      "learning_rate": 0.00014252106741573036,
      "loss": 0.8675,
      "step": 4060
    },
    {
      "epoch": 0.07145490616057164,
      "grad_norm": 0.07863410269343978,
      "learning_rate": 0.00014287219101123596,
      "loss": 0.869,
      "step": 4070
    },
    {
      "epoch": 0.07163047104057305,
      "grad_norm": 0.07542836260788668,
      "learning_rate": 0.00014322331460674159,
      "loss": 0.8659,
      "step": 4080
    },
    {
      "epoch": 0.07180603592057445,
      "grad_norm": 0.11071694874391441,
      "learning_rate": 0.0001435744382022472,
      "loss": 0.8711,
      "step": 4090
    },
    {
      "epoch": 0.07198160080057585,
      "grad_norm": 0.07668577818265196,
      "learning_rate": 0.0001439255617977528,
      "loss": 0.8641,
      "step": 4100
    },
    {
      "epoch": 0.07215716568057726,
      "grad_norm": 0.06880767046653036,
      "learning_rate": 0.00014427668539325844,
      "loss": 0.8684,
      "step": 4110
    },
    {
      "epoch": 0.07233273056057866,
      "grad_norm": 0.09301324305077331,
      "learning_rate": 0.00014462780898876404,
      "loss": 0.8779,
      "step": 4120
    },
    {
      "epoch": 0.07250829544058007,
      "grad_norm": 0.0926534904459085,
      "learning_rate": 0.00014497893258426966,
      "loss": 0.8666,
      "step": 4130
    },
    {
      "epoch": 0.07268386032058147,
      "grad_norm": 0.10438820316693159,
      "learning_rate": 0.0001453300561797753,
      "loss": 0.8676,
      "step": 4140
    },
    {
      "epoch": 0.07285942520058288,
      "grad_norm": 0.10819014876339289,
      "learning_rate": 0.00014568117977528091,
      "loss": 0.8747,
      "step": 4150
    },
    {
      "epoch": 0.07303499008058428,
      "grad_norm": 0.07580563707929784,
      "learning_rate": 0.00014603230337078654,
      "loss": 0.8723,
      "step": 4160
    },
    {
      "epoch": 0.07321055496058568,
      "grad_norm": 0.07846311969047703,
      "learning_rate": 0.00014638342696629214,
      "loss": 0.8662,
      "step": 4170
    },
    {
      "epoch": 0.07338611984058709,
      "grad_norm": 0.07983271774368964,
      "learning_rate": 0.00014673455056179777,
      "loss": 0.863,
      "step": 4180
    },
    {
      "epoch": 0.0735616847205885,
      "grad_norm": 0.0825969791220882,
      "learning_rate": 0.0001470856741573034,
      "loss": 0.8658,
      "step": 4190
    },
    {
      "epoch": 0.0737372496005899,
      "grad_norm": 0.07234958039630493,
      "learning_rate": 0.000147436797752809,
      "loss": 0.8682,
      "step": 4200
    },
    {
      "epoch": 0.0739128144805913,
      "grad_norm": 0.08174310148214484,
      "learning_rate": 0.00014778792134831462,
      "loss": 0.8691,
      "step": 4210
    },
    {
      "epoch": 0.0740883793605927,
      "grad_norm": 0.07186689397781615,
      "learning_rate": 0.00014813904494382022,
      "loss": 0.8676,
      "step": 4220
    },
    {
      "epoch": 0.07426394424059411,
      "grad_norm": 0.08461427893764355,
      "learning_rate": 0.00014849016853932584,
      "loss": 0.8649,
      "step": 4230
    },
    {
      "epoch": 0.07443950912059552,
      "grad_norm": 0.06589368503481277,
      "learning_rate": 0.00014884129213483147,
      "loss": 0.8726,
      "step": 4240
    },
    {
      "epoch": 0.07461507400059691,
      "grad_norm": 0.06367472494772826,
      "learning_rate": 0.0001491924157303371,
      "loss": 0.8709,
      "step": 4250
    },
    {
      "epoch": 0.07479063888059832,
      "grad_norm": 0.09853738913632547,
      "learning_rate": 0.0001495435393258427,
      "loss": 0.8581,
      "step": 4260
    },
    {
      "epoch": 0.07496620376059973,
      "grad_norm": 0.10839327616134614,
      "learning_rate": 0.00014989466292134832,
      "loss": 0.8673,
      "step": 4270
    },
    {
      "epoch": 0.07514176864060114,
      "grad_norm": 0.08450106202556797,
      "learning_rate": 0.00015024578651685395,
      "loss": 0.8717,
      "step": 4280
    },
    {
      "epoch": 0.07531733352060253,
      "grad_norm": 0.09038532471554327,
      "learning_rate": 0.00015059691011235957,
      "loss": 0.8656,
      "step": 4290
    },
    {
      "epoch": 0.07549289840060394,
      "grad_norm": 0.08629338423113897,
      "learning_rate": 0.00015094803370786517,
      "loss": 0.8682,
      "step": 4300
    },
    {
      "epoch": 0.07566846328060535,
      "grad_norm": 0.0750463548101676,
      "learning_rate": 0.0001512991573033708,
      "loss": 0.8627,
      "step": 4310
    },
    {
      "epoch": 0.07584402816060676,
      "grad_norm": 0.07337608217570649,
      "learning_rate": 0.0001516502808988764,
      "loss": 0.8785,
      "step": 4320
    },
    {
      "epoch": 0.07601959304060815,
      "grad_norm": 0.08405988308033217,
      "learning_rate": 0.00015200140449438202,
      "loss": 0.867,
      "step": 4330
    },
    {
      "epoch": 0.07619515792060956,
      "grad_norm": 0.0701783258009735,
      "learning_rate": 0.00015235252808988765,
      "loss": 0.8662,
      "step": 4340
    },
    {
      "epoch": 0.07637072280061097,
      "grad_norm": 0.08586204400052112,
      "learning_rate": 0.00015270365168539328,
      "loss": 0.883,
      "step": 4350
    },
    {
      "epoch": 0.07654628768061238,
      "grad_norm": 0.0769610592561428,
      "learning_rate": 0.00015305477528089888,
      "loss": 0.8637,
      "step": 4360
    },
    {
      "epoch": 0.07672185256061377,
      "grad_norm": 0.07998400035283786,
      "learning_rate": 0.0001534058988764045,
      "loss": 0.8614,
      "step": 4370
    },
    {
      "epoch": 0.07689741744061518,
      "grad_norm": 0.10840681872396715,
      "learning_rate": 0.00015375702247191013,
      "loss": 0.8676,
      "step": 4380
    },
    {
      "epoch": 0.07707298232061659,
      "grad_norm": 0.09646002316237622,
      "learning_rate": 0.00015410814606741575,
      "loss": 0.8688,
      "step": 4390
    },
    {
      "epoch": 0.077248547200618,
      "grad_norm": 0.07038237154544184,
      "learning_rate": 0.00015445926966292135,
      "loss": 0.8589,
      "step": 4400
    },
    {
      "epoch": 0.07742411208061939,
      "grad_norm": 0.0854281808452838,
      "learning_rate": 0.00015481039325842698,
      "loss": 0.8661,
      "step": 4410
    },
    {
      "epoch": 0.0775996769606208,
      "grad_norm": 0.07270387940775015,
      "learning_rate": 0.00015516151685393258,
      "loss": 0.8695,
      "step": 4420
    },
    {
      "epoch": 0.0777752418406222,
      "grad_norm": 0.07818967782742531,
      "learning_rate": 0.0001555126404494382,
      "loss": 0.8672,
      "step": 4430
    },
    {
      "epoch": 0.0779508067206236,
      "grad_norm": 0.10054700292155842,
      "learning_rate": 0.00015586376404494383,
      "loss": 0.868,
      "step": 4440
    },
    {
      "epoch": 0.07812637160062501,
      "grad_norm": 0.07337802633079017,
      "learning_rate": 0.00015621488764044946,
      "loss": 0.8685,
      "step": 4450
    },
    {
      "epoch": 0.07830193648062642,
      "grad_norm": 0.07292774031953334,
      "learning_rate": 0.00015656601123595506,
      "loss": 0.8703,
      "step": 4460
    },
    {
      "epoch": 0.07847750136062782,
      "grad_norm": 0.09110409688313115,
      "learning_rate": 0.00015691713483146068,
      "loss": 0.8637,
      "step": 4470
    },
    {
      "epoch": 0.07865306624062922,
      "grad_norm": 0.10994224724783667,
      "learning_rate": 0.0001572682584269663,
      "loss": 0.8642,
      "step": 4480
    },
    {
      "epoch": 0.07882863112063063,
      "grad_norm": 0.07514051930797577,
      "learning_rate": 0.00015761938202247193,
      "loss": 0.876,
      "step": 4490
    },
    {
      "epoch": 0.07900419600063203,
      "grad_norm": 0.08473126096815146,
      "learning_rate": 0.00015797050561797753,
      "loss": 0.8694,
      "step": 4500
    },
    {
      "epoch": 0.07917976088063344,
      "grad_norm": 0.05889963636699825,
      "learning_rate": 0.00015832162921348316,
      "loss": 0.8693,
      "step": 4510
    },
    {
      "epoch": 0.07935532576063484,
      "grad_norm": 0.09644216724875465,
      "learning_rate": 0.00015867275280898876,
      "loss": 0.868,
      "step": 4520
    },
    {
      "epoch": 0.07953089064063625,
      "grad_norm": 0.08330160201996087,
      "learning_rate": 0.00015902387640449439,
      "loss": 0.8721,
      "step": 4530
    },
    {
      "epoch": 0.07970645552063765,
      "grad_norm": 0.0771144573061966,
      "learning_rate": 0.000159375,
      "loss": 0.8703,
      "step": 4540
    },
    {
      "epoch": 0.07988202040063906,
      "grad_norm": 0.07320722269239563,
      "learning_rate": 0.00015972612359550564,
      "loss": 0.864,
      "step": 4550
    },
    {
      "epoch": 0.08005758528064046,
      "grad_norm": 0.07410639814814954,
      "learning_rate": 0.00016007724719101124,
      "loss": 0.8689,
      "step": 4560
    },
    {
      "epoch": 0.08023315016064186,
      "grad_norm": 0.08436512009270412,
      "learning_rate": 0.00016042837078651686,
      "loss": 0.8646,
      "step": 4570
    },
    {
      "epoch": 0.08040871504064327,
      "grad_norm": 0.08810186765455941,
      "learning_rate": 0.0001607794943820225,
      "loss": 0.8697,
      "step": 4580
    },
    {
      "epoch": 0.08058427992064468,
      "grad_norm": 0.10799389452800459,
      "learning_rate": 0.0001611306179775281,
      "loss": 0.8657,
      "step": 4590
    },
    {
      "epoch": 0.08075984480064607,
      "grad_norm": 0.07023168870357091,
      "learning_rate": 0.00016148174157303371,
      "loss": 0.8694,
      "step": 4600
    },
    {
      "epoch": 0.08093540968064748,
      "grad_norm": 0.1087335669987877,
      "learning_rate": 0.00016183286516853931,
      "loss": 0.8647,
      "step": 4610
    },
    {
      "epoch": 0.08111097456064889,
      "grad_norm": 0.07648697691679036,
      "learning_rate": 0.00016218398876404494,
      "loss": 0.8732,
      "step": 4620
    },
    {
      "epoch": 0.0812865394406503,
      "grad_norm": 0.08907095196687016,
      "learning_rate": 0.00016253511235955057,
      "loss": 0.8664,
      "step": 4630
    },
    {
      "epoch": 0.0814621043206517,
      "grad_norm": 0.08090337098455115,
      "learning_rate": 0.0001628862359550562,
      "loss": 0.8735,
      "step": 4640
    },
    {
      "epoch": 0.0816376692006531,
      "grad_norm": 0.07864039379613462,
      "learning_rate": 0.00016323735955056182,
      "loss": 0.8729,
      "step": 4650
    },
    {
      "epoch": 0.08181323408065451,
      "grad_norm": 0.11340875558156789,
      "learning_rate": 0.00016358848314606742,
      "loss": 0.8699,
      "step": 4660
    },
    {
      "epoch": 0.0819887989606559,
      "grad_norm": 0.07735852976232371,
      "learning_rate": 0.00016393960674157304,
      "loss": 0.8602,
      "step": 4670
    },
    {
      "epoch": 0.08216436384065731,
      "grad_norm": 0.08806413794618977,
      "learning_rate": 0.00016429073033707867,
      "loss": 0.8678,
      "step": 4680
    },
    {
      "epoch": 0.08233992872065872,
      "grad_norm": 0.09118430042747845,
      "learning_rate": 0.00016464185393258427,
      "loss": 0.862,
      "step": 4690
    },
    {
      "epoch": 0.08251549360066013,
      "grad_norm": 0.07886876913844895,
      "learning_rate": 0.0001649929775280899,
      "loss": 0.8578,
      "step": 4700
    },
    {
      "epoch": 0.08269105848066152,
      "grad_norm": 0.09063438315459828,
      "learning_rate": 0.0001653441011235955,
      "loss": 0.8628,
      "step": 4710
    },
    {
      "epoch": 0.08286662336066293,
      "grad_norm": 0.10755783807824873,
      "learning_rate": 0.00016569522471910112,
      "loss": 0.8722,
      "step": 4720
    },
    {
      "epoch": 0.08304218824066434,
      "grad_norm": 0.06498041183357131,
      "learning_rate": 0.00016604634831460675,
      "loss": 0.8632,
      "step": 4730
    },
    {
      "epoch": 0.08321775312066575,
      "grad_norm": 0.07473440843261468,
      "learning_rate": 0.00016639747191011237,
      "loss": 0.865,
      "step": 4740
    },
    {
      "epoch": 0.08339331800066714,
      "grad_norm": 0.148929785661174,
      "learning_rate": 0.000166748595505618,
      "loss": 0.8677,
      "step": 4750
    },
    {
      "epoch": 0.08356888288066855,
      "grad_norm": 0.06992532896317065,
      "learning_rate": 0.0001670997191011236,
      "loss": 0.8676,
      "step": 4760
    },
    {
      "epoch": 0.08374444776066996,
      "grad_norm": 0.08049927042262585,
      "learning_rate": 0.00016745084269662922,
      "loss": 0.8628,
      "step": 4770
    },
    {
      "epoch": 0.08392001264067137,
      "grad_norm": 0.07725685529048437,
      "learning_rate": 0.00016780196629213485,
      "loss": 0.8651,
      "step": 4780
    },
    {
      "epoch": 0.08409557752067276,
      "grad_norm": 0.08866309175750381,
      "learning_rate": 0.00016815308988764045,
      "loss": 0.8649,
      "step": 4790
    },
    {
      "epoch": 0.08427114240067417,
      "grad_norm": 0.07268164801108132,
      "learning_rate": 0.00016850421348314608,
      "loss": 0.8619,
      "step": 4800
    },
    {
      "epoch": 0.08444670728067558,
      "grad_norm": 0.08583037056922549,
      "learning_rate": 0.00016885533707865168,
      "loss": 0.8691,
      "step": 4810
    },
    {
      "epoch": 0.08462227216067698,
      "grad_norm": 0.09705894012417274,
      "learning_rate": 0.0001692064606741573,
      "loss": 0.8608,
      "step": 4820
    },
    {
      "epoch": 0.08479783704067838,
      "grad_norm": 0.09740745345970428,
      "learning_rate": 0.00016955758426966293,
      "loss": 0.8686,
      "step": 4830
    },
    {
      "epoch": 0.08497340192067979,
      "grad_norm": 0.06227523777486907,
      "learning_rate": 0.00016990870786516855,
      "loss": 0.8724,
      "step": 4840
    },
    {
      "epoch": 0.0851489668006812,
      "grad_norm": 0.07892537529720367,
      "learning_rate": 0.00017025983146067418,
      "loss": 0.8685,
      "step": 4850
    },
    {
      "epoch": 0.0853245316806826,
      "grad_norm": 0.07015226131016203,
      "learning_rate": 0.00017061095505617978,
      "loss": 0.8661,
      "step": 4860
    },
    {
      "epoch": 0.085500096560684,
      "grad_norm": 0.10831127654925424,
      "learning_rate": 0.0001709620786516854,
      "loss": 0.8781,
      "step": 4870
    },
    {
      "epoch": 0.0856756614406854,
      "grad_norm": 0.07290902209263958,
      "learning_rate": 0.00017131320224719103,
      "loss": 0.8664,
      "step": 4880
    },
    {
      "epoch": 0.08585122632068681,
      "grad_norm": 0.060254225502320215,
      "learning_rate": 0.00017166432584269663,
      "loss": 0.8595,
      "step": 4890
    },
    {
      "epoch": 0.08602679120068821,
      "grad_norm": 0.09287350217820713,
      "learning_rate": 0.00017201544943820226,
      "loss": 0.8713,
      "step": 4900
    },
    {
      "epoch": 0.08620235608068962,
      "grad_norm": 0.08081451244307865,
      "learning_rate": 0.00017236657303370786,
      "loss": 0.8618,
      "step": 4910
    },
    {
      "epoch": 0.08637792096069102,
      "grad_norm": 0.08625108801626413,
      "learning_rate": 0.00017271769662921348,
      "loss": 0.8703,
      "step": 4920
    },
    {
      "epoch": 0.08655348584069243,
      "grad_norm": 0.06292383949617689,
      "learning_rate": 0.0001730688202247191,
      "loss": 0.8683,
      "step": 4930
    },
    {
      "epoch": 0.08672905072069383,
      "grad_norm": 0.06638756545664112,
      "learning_rate": 0.00017341994382022473,
      "loss": 0.8645,
      "step": 4940
    },
    {
      "epoch": 0.08690461560069523,
      "grad_norm": 0.09739006317211192,
      "learning_rate": 0.00017377106741573036,
      "loss": 0.8711,
      "step": 4950
    },
    {
      "epoch": 0.08708018048069664,
      "grad_norm": 0.06535366262039126,
      "learning_rate": 0.00017412219101123596,
      "loss": 0.8644,
      "step": 4960
    },
    {
      "epoch": 0.08725574536069805,
      "grad_norm": 0.07121829780754685,
      "learning_rate": 0.00017447331460674159,
      "loss": 0.8683,
      "step": 4970
    },
    {
      "epoch": 0.08743131024069944,
      "grad_norm": 0.06474340894680934,
      "learning_rate": 0.0001748244382022472,
      "loss": 0.8634,
      "step": 4980
    },
    {
      "epoch": 0.08760687512070085,
      "grad_norm": 0.12692213524882914,
      "learning_rate": 0.0001751755617977528,
      "loss": 0.8623,
      "step": 4990
    },
    {
      "epoch": 0.08778244000070226,
      "grad_norm": 0.0950664795038832,
      "learning_rate": 0.00017552668539325844,
      "loss": 0.8699,
      "step": 5000
    },
    {
      "epoch": 0.08795800488070367,
      "grad_norm": 0.07260680382355068,
      "learning_rate": 0.00017587780898876404,
      "loss": 0.8698,
      "step": 5010
    },
    {
      "epoch": 0.08813356976070506,
      "grad_norm": 0.08006083335946487,
      "learning_rate": 0.00017622893258426966,
      "loss": 0.8662,
      "step": 5020
    },
    {
      "epoch": 0.08830913464070647,
      "grad_norm": 0.08955359871919073,
      "learning_rate": 0.0001765800561797753,
      "loss": 0.8582,
      "step": 5030
    },
    {
      "epoch": 0.08848469952070788,
      "grad_norm": 0.0707756510968693,
      "learning_rate": 0.00017693117977528092,
      "loss": 0.866,
      "step": 5040
    },
    {
      "epoch": 0.08866026440070929,
      "grad_norm": 0.07998471007407254,
      "learning_rate": 0.00017728230337078654,
      "loss": 0.8742,
      "step": 5050
    },
    {
      "epoch": 0.08883582928071068,
      "grad_norm": 0.08220301787851143,
      "learning_rate": 0.00017763342696629214,
      "loss": 0.8744,
      "step": 5060
    },
    {
      "epoch": 0.08901139416071209,
      "grad_norm": 0.11196912685369766,
      "learning_rate": 0.00017798455056179777,
      "loss": 0.867,
      "step": 5070
    },
    {
      "epoch": 0.0891869590407135,
      "grad_norm": 0.08960575960300038,
      "learning_rate": 0.0001783356741573034,
      "loss": 0.8706,
      "step": 5080
    },
    {
      "epoch": 0.0893625239207149,
      "grad_norm": 0.11055748637690853,
      "learning_rate": 0.000178686797752809,
      "loss": 0.8614,
      "step": 5090
    },
    {
      "epoch": 0.0895380888007163,
      "grad_norm": 0.07733886541459249,
      "learning_rate": 0.00017903792134831462,
      "loss": 0.8661,
      "step": 5100
    },
    {
      "epoch": 0.08971365368071771,
      "grad_norm": 0.07175029446649613,
      "learning_rate": 0.00017938904494382022,
      "loss": 0.8653,
      "step": 5110
    },
    {
      "epoch": 0.08988921856071912,
      "grad_norm": 0.07954521033221447,
      "learning_rate": 0.00017974016853932584,
      "loss": 0.8718,
      "step": 5120
    },
    {
      "epoch": 0.09006478344072052,
      "grad_norm": 0.0877358714596494,
      "learning_rate": 0.00018009129213483147,
      "loss": 0.8679,
      "step": 5130
    },
    {
      "epoch": 0.09024034832072192,
      "grad_norm": 0.09314894763089218,
      "learning_rate": 0.0001804424157303371,
      "loss": 0.8704,
      "step": 5140
    },
    {
      "epoch": 0.09041591320072333,
      "grad_norm": 0.10118294864548553,
      "learning_rate": 0.00018079353932584272,
      "loss": 0.8641,
      "step": 5150
    },
    {
      "epoch": 0.09059147808072474,
      "grad_norm": 0.10062214394123596,
      "learning_rate": 0.00018114466292134832,
      "loss": 0.8666,
      "step": 5160
    },
    {
      "epoch": 0.09076704296072613,
      "grad_norm": 0.08645117471089833,
      "learning_rate": 0.00018149578651685395,
      "loss": 0.8728,
      "step": 5170
    },
    {
      "epoch": 0.09094260784072754,
      "grad_norm": 0.08680755326743796,
      "learning_rate": 0.00018184691011235957,
      "loss": 0.8747,
      "step": 5180
    },
    {
      "epoch": 0.09111817272072895,
      "grad_norm": 0.0746449117078722,
      "learning_rate": 0.00018219803370786517,
      "loss": 0.8714,
      "step": 5190
    },
    {
      "epoch": 0.09129373760073035,
      "grad_norm": 0.08648526637792477,
      "learning_rate": 0.0001825491573033708,
      "loss": 0.8622,
      "step": 5200
    },
    {
      "epoch": 0.09146930248073175,
      "grad_norm": 0.13431745028293973,
      "learning_rate": 0.0001829002808988764,
      "loss": 0.8594,
      "step": 5210
    },
    {
      "epoch": 0.09164486736073316,
      "grad_norm": 0.06981457627834535,
      "learning_rate": 0.00018325140449438202,
      "loss": 0.8681,
      "step": 5220
    },
    {
      "epoch": 0.09182043224073456,
      "grad_norm": 0.1130522452706824,
      "learning_rate": 0.00018360252808988765,
      "loss": 0.8688,
      "step": 5230
    },
    {
      "epoch": 0.09199599712073597,
      "grad_norm": 0.10309214926270541,
      "learning_rate": 0.00018395365168539328,
      "loss": 0.8657,
      "step": 5240
    },
    {
      "epoch": 0.09217156200073737,
      "grad_norm": 0.09141805361926296,
      "learning_rate": 0.00018430477528089888,
      "loss": 0.8599,
      "step": 5250
    },
    {
      "epoch": 0.09234712688073878,
      "grad_norm": 0.10325173008054479,
      "learning_rate": 0.0001846558988764045,
      "loss": 0.8594,
      "step": 5260
    },
    {
      "epoch": 0.09252269176074018,
      "grad_norm": 0.08491881491829187,
      "learning_rate": 0.00018500702247191013,
      "loss": 0.8688,
      "step": 5270
    },
    {
      "epoch": 0.09269825664074159,
      "grad_norm": 0.08887562234822251,
      "learning_rate": 0.00018535814606741575,
      "loss": 0.8583,
      "step": 5280
    },
    {
      "epoch": 0.09287382152074299,
      "grad_norm": 0.08426215291625319,
      "learning_rate": 0.00018570926966292135,
      "loss": 0.8646,
      "step": 5290
    },
    {
      "epoch": 0.0930493864007444,
      "grad_norm": 0.0827792330446448,
      "learning_rate": 0.00018606039325842698,
      "loss": 0.864,
      "step": 5300
    },
    {
      "epoch": 0.0932249512807458,
      "grad_norm": 0.07766589011550949,
      "learning_rate": 0.00018641151685393258,
      "loss": 0.8668,
      "step": 5310
    },
    {
      "epoch": 0.09340051616074721,
      "grad_norm": 0.09520977824458415,
      "learning_rate": 0.0001867626404494382,
      "loss": 0.871,
      "step": 5320
    },
    {
      "epoch": 0.0935760810407486,
      "grad_norm": 0.06359435360186315,
      "learning_rate": 0.00018711376404494383,
      "loss": 0.8651,
      "step": 5330
    },
    {
      "epoch": 0.09375164592075001,
      "grad_norm": 0.06989474927666056,
      "learning_rate": 0.00018746488764044946,
      "loss": 0.869,
      "step": 5340
    },
    {
      "epoch": 0.09392721080075142,
      "grad_norm": 0.07663947538179294,
      "learning_rate": 0.00018781601123595506,
      "loss": 0.8668,
      "step": 5350
    },
    {
      "epoch": 0.09410277568075283,
      "grad_norm": 0.08971002887767697,
      "learning_rate": 0.00018816713483146068,
      "loss": 0.8625,
      "step": 5360
    },
    {
      "epoch": 0.09427834056075422,
      "grad_norm": 0.10046597281574413,
      "learning_rate": 0.0001885182584269663,
      "loss": 0.8603,
      "step": 5370
    },
    {
      "epoch": 0.09445390544075563,
      "grad_norm": 0.0826615082943464,
      "learning_rate": 0.00018886938202247194,
      "loss": 0.8662,
      "step": 5380
    },
    {
      "epoch": 0.09462947032075704,
      "grad_norm": 0.12038149841092534,
      "learning_rate": 0.00018922050561797753,
      "loss": 0.8704,
      "step": 5390
    },
    {
      "epoch": 0.09480503520075843,
      "grad_norm": 0.07572646972459902,
      "learning_rate": 0.00018957162921348316,
      "loss": 0.8683,
      "step": 5400
    },
    {
      "epoch": 0.09498060008075984,
      "grad_norm": 0.08678404926154863,
      "learning_rate": 0.00018992275280898876,
      "loss": 0.867,
      "step": 5410
    },
    {
      "epoch": 0.09515616496076125,
      "grad_norm": 0.0770515566215959,
      "learning_rate": 0.00019027387640449439,
      "loss": 0.8656,
      "step": 5420
    },
    {
      "epoch": 0.09533172984076266,
      "grad_norm": 0.07366276219931929,
      "learning_rate": 0.000190625,
      "loss": 0.8649,
      "step": 5430
    },
    {
      "epoch": 0.09550729472076405,
      "grad_norm": 0.09576967283134216,
      "learning_rate": 0.00019097612359550564,
      "loss": 0.8704,
      "step": 5440
    },
    {
      "epoch": 0.09568285960076546,
      "grad_norm": 0.10867746900587937,
      "learning_rate": 0.00019132724719101124,
      "loss": 0.8694,
      "step": 5450
    },
    {
      "epoch": 0.09585842448076687,
      "grad_norm": 0.09173761799700574,
      "learning_rate": 0.00019167837078651686,
      "loss": 0.8678,
      "step": 5460
    },
    {
      "epoch": 0.09603398936076828,
      "grad_norm": 0.08497574132058541,
      "learning_rate": 0.0001920294943820225,
      "loss": 0.8632,
      "step": 5470
    },
    {
      "epoch": 0.09620955424076967,
      "grad_norm": 0.10063514854313432,
      "learning_rate": 0.0001923806179775281,
      "loss": 0.8569,
      "step": 5480
    },
    {
      "epoch": 0.09638511912077108,
      "grad_norm": 0.07634828188980673,
      "learning_rate": 0.00019273174157303372,
      "loss": 0.8705,
      "step": 5490
    },
    {
      "epoch": 0.09656068400077249,
      "grad_norm": 0.08062230045213784,
      "learning_rate": 0.00019308286516853934,
      "loss": 0.8655,
      "step": 5500
    },
    {
      "epoch": 0.0967362488807739,
      "grad_norm": 0.07785635604700651,
      "learning_rate": 0.00019343398876404494,
      "loss": 0.869,
      "step": 5510
    },
    {
      "epoch": 0.09691181376077529,
      "grad_norm": 0.08388067460689601,
      "learning_rate": 0.00019378511235955057,
      "loss": 0.8711,
      "step": 5520
    },
    {
      "epoch": 0.0970873786407767,
      "grad_norm": 0.08094237436290566,
      "learning_rate": 0.0001941362359550562,
      "loss": 0.858,
      "step": 5530
    },
    {
      "epoch": 0.0972629435207781,
      "grad_norm": 0.07257938176280468,
      "learning_rate": 0.00019448735955056182,
      "loss": 0.8656,
      "step": 5540
    },
    {
      "epoch": 0.09743850840077951,
      "grad_norm": 0.0660216945109028,
      "learning_rate": 0.00019483848314606742,
      "loss": 0.874,
      "step": 5550
    },
    {
      "epoch": 0.09761407328078091,
      "grad_norm": 0.07336033031504356,
      "learning_rate": 0.00019518960674157304,
      "loss": 0.8693,
      "step": 5560
    },
    {
      "epoch": 0.09778963816078232,
      "grad_norm": 0.11475596532129542,
      "learning_rate": 0.00019554073033707867,
      "loss": 0.866,
      "step": 5570
    },
    {
      "epoch": 0.09796520304078372,
      "grad_norm": 0.06676005376026309,
      "learning_rate": 0.00019589185393258427,
      "loss": 0.8668,
      "step": 5580
    },
    {
      "epoch": 0.09814076792078513,
      "grad_norm": 0.11670832249241093,
      "learning_rate": 0.0001962429775280899,
      "loss": 0.8646,
      "step": 5590
    },
    {
      "epoch": 0.09831633280078653,
      "grad_norm": 0.0946479465102158,
      "learning_rate": 0.00019659410112359552,
      "loss": 0.8607,
      "step": 5600
    },
    {
      "epoch": 0.09849189768078793,
      "grad_norm": 0.0903312358020182,
      "learning_rate": 0.00019694522471910112,
      "loss": 0.8725,
      "step": 5610
    },
    {
      "epoch": 0.09866746256078934,
      "grad_norm": 0.0892184536704075,
      "learning_rate": 0.00019729634831460675,
      "loss": 0.865,
      "step": 5620
    },
    {
      "epoch": 0.09884302744079075,
      "grad_norm": 0.10496426530372611,
      "learning_rate": 0.00019764747191011237,
      "loss": 0.8677,
      "step": 5630
    },
    {
      "epoch": 0.09901859232079215,
      "grad_norm": 0.10170756382432672,
      "learning_rate": 0.000197998595505618,
      "loss": 0.8638,
      "step": 5640
    },
    {
      "epoch": 0.09919415720079355,
      "grad_norm": 0.07844715045945563,
      "learning_rate": 0.0001983497191011236,
      "loss": 0.8673,
      "step": 5650
    },
    {
      "epoch": 0.09936972208079496,
      "grad_norm": 0.08308166854934049,
      "learning_rate": 0.00019870084269662923,
      "loss": 0.8594,
      "step": 5660
    },
    {
      "epoch": 0.09954528696079636,
      "grad_norm": 0.07078826247147489,
      "learning_rate": 0.00019905196629213485,
      "loss": 0.8689,
      "step": 5670
    },
    {
      "epoch": 0.09972085184079776,
      "grad_norm": 0.07122424794099147,
      "learning_rate": 0.00019940308988764045,
      "loss": 0.8686,
      "step": 5680
    },
    {
      "epoch": 0.09989641672079917,
      "grad_norm": 0.0662598705125894,
      "learning_rate": 0.00019975421348314608,
      "loss": 0.8585,
      "step": 5690
    },
    {
      "epoch": 0.10007198160080058,
      "grad_norm": 0.08115052049442587,
      "learning_rate": 0.00019999999994355433,
      "loss": 0.8687,
      "step": 5700
    },
    {
      "epoch": 0.10024754648080197,
      "grad_norm": 0.06120919579659045,
      "learning_rate": 0.00019999999894007513,
      "loss": 0.8656,
      "step": 5710
    },
    {
      "epoch": 0.10042311136080338,
      "grad_norm": 0.07271076099270422,
      "learning_rate": 0.000199999996682247,
      "loss": 0.8655,
      "step": 5720
    },
    {
      "epoch": 0.10059867624080479,
      "grad_norm": 0.07008063612434355,
      "learning_rate": 0.00019999999317006991,
      "loss": 0.8674,
      "step": 5730
    },
    {
      "epoch": 0.1007742411208062,
      "grad_norm": 0.08207146951118063,
      "learning_rate": 0.00019999998840354398,
      "loss": 0.8715,
      "step": 5740
    },
    {
      "epoch": 0.1009498060008076,
      "grad_norm": 0.11793559162189993,
      "learning_rate": 0.00019999998238266918,
      "loss": 0.861,
      "step": 5750
    },
    {
      "epoch": 0.101125370880809,
      "grad_norm": 0.09656625220519839,
      "learning_rate": 0.00019999997510744563,
      "loss": 0.872,
      "step": 5760
    },
    {
      "epoch": 0.10130093576081041,
      "grad_norm": 0.08344523597212511,
      "learning_rate": 0.00019999996657787348,
      "loss": 0.8655,
      "step": 5770
    },
    {
      "epoch": 0.10147650064081182,
      "grad_norm": 0.087657217897348,
      "learning_rate": 0.00019999995679395274,
      "loss": 0.8588,
      "step": 5780
    },
    {
      "epoch": 0.10165206552081321,
      "grad_norm": 0.08669015655096904,
      "learning_rate": 0.00019999994575568356,
      "loss": 0.8706,
      "step": 5790
    },
    {
      "epoch": 0.10182763040081462,
      "grad_norm": 0.08540805795078366,
      "learning_rate": 0.00019999993346306612,
      "loss": 0.8548,
      "step": 5800
    },
    {
      "epoch": 0.10200319528081603,
      "grad_norm": 0.0855003350024982,
      "learning_rate": 0.0001999999199161005,
      "loss": 0.8714,
      "step": 5810
    },
    {
      "epoch": 0.10217876016081744,
      "grad_norm": 0.14239010780553799,
      "learning_rate": 0.000199999905114787,
      "loss": 0.868,
      "step": 5820
    },
    {
      "epoch": 0.10235432504081883,
      "grad_norm": 0.07822582490916924,
      "learning_rate": 0.00019999988905912566,
      "loss": 0.8644,
      "step": 5830
    },
    {
      "epoch": 0.10252988992082024,
      "grad_norm": 0.05840941753870967,
      "learning_rate": 0.00019999987174911682,
      "loss": 0.872,
      "step": 5840
    },
    {
      "epoch": 0.10270545480082165,
      "grad_norm": 0.0777612893715757,
      "learning_rate": 0.00019999985318476057,
      "loss": 0.8699,
      "step": 5850
    },
    {
      "epoch": 0.10288101968082305,
      "grad_norm": 0.07908561980487473,
      "learning_rate": 0.0001999998333660572,
      "loss": 0.8641,
      "step": 5860
    },
    {
      "epoch": 0.10305658456082445,
      "grad_norm": 0.09908934997538962,
      "learning_rate": 0.00019999981229300702,
      "loss": 0.8616,
      "step": 5870
    },
    {
      "epoch": 0.10323214944082586,
      "grad_norm": 0.09482973389235208,
      "learning_rate": 0.00019999978996561023,
      "loss": 0.8682,
      "step": 5880
    },
    {
      "epoch": 0.10340771432082727,
      "grad_norm": 0.07115887895206233,
      "learning_rate": 0.00019999976638386711,
      "loss": 0.8692,
      "step": 5890
    },
    {
      "epoch": 0.10358327920082866,
      "grad_norm": 0.06499311559529795,
      "learning_rate": 0.000199999741547778,
      "loss": 0.8792,
      "step": 5900
    },
    {
      "epoch": 0.10375884408083007,
      "grad_norm": 0.09612461730681014,
      "learning_rate": 0.00019999971545734318,
      "loss": 0.8658,
      "step": 5910
    },
    {
      "epoch": 0.10393440896083148,
      "grad_norm": 0.08110422904202048,
      "learning_rate": 0.000199999688112563,
      "loss": 0.8707,
      "step": 5920
    },
    {
      "epoch": 0.10410997384083288,
      "grad_norm": 0.10369390784473606,
      "learning_rate": 0.00019999965951343778,
      "loss": 0.8699,
      "step": 5930
    },
    {
      "epoch": 0.10428553872083428,
      "grad_norm": 0.09699533552971036,
      "learning_rate": 0.00019999962965996788,
      "loss": 0.8646,
      "step": 5940
    },
    {
      "epoch": 0.10446110360083569,
      "grad_norm": 0.07308862974129239,
      "learning_rate": 0.00019999959855215374,
      "loss": 0.8655,
      "step": 5950
    },
    {
      "epoch": 0.1046366684808371,
      "grad_norm": 0.05968577837626002,
      "learning_rate": 0.00019999956618999568,
      "loss": 0.8656,
      "step": 5960
    },
    {
      "epoch": 0.1048122333608385,
      "grad_norm": 0.0719894546627652,
      "learning_rate": 0.0001999995325734942,
      "loss": 0.8764,
      "step": 5970
    },
    {
      "epoch": 0.1049877982408399,
      "grad_norm": 0.08471584877753886,
      "learning_rate": 0.00019999949770264962,
      "loss": 0.8685,
      "step": 5980
    },
    {
      "epoch": 0.1051633631208413,
      "grad_norm": 0.07196709755639648,
      "learning_rate": 0.00019999946157746244,
      "loss": 0.8646,
      "step": 5990
    },
    {
      "epoch": 0.10533892800084271,
      "grad_norm": 0.08281928995007487,
      "learning_rate": 0.00019999942419793312,
      "loss": 0.8659,
      "step": 6000
    },
    {
      "epoch": 0.10551449288084412,
      "grad_norm": 0.08451846739330868,
      "learning_rate": 0.00019999938556406212,
      "loss": 0.8714,
      "step": 6010
    },
    {
      "epoch": 0.10569005776084552,
      "grad_norm": 0.08684304361475265,
      "learning_rate": 0.00019999934567584994,
      "loss": 0.8692,
      "step": 6020
    },
    {
      "epoch": 0.10586562264084692,
      "grad_norm": 0.10077702584925932,
      "learning_rate": 0.00019999930453329704,
      "loss": 0.8639,
      "step": 6030
    },
    {
      "epoch": 0.10604118752084833,
      "grad_norm": 0.07492763032279787,
      "learning_rate": 0.00019999926213640402,
      "loss": 0.8659,
      "step": 6040
    },
    {
      "epoch": 0.10621675240084974,
      "grad_norm": 0.09287087464818368,
      "learning_rate": 0.00019999921848517136,
      "loss": 0.8661,
      "step": 6050
    },
    {
      "epoch": 0.10639231728085113,
      "grad_norm": 0.080949995790025,
      "learning_rate": 0.00019999917357959964,
      "loss": 0.8668,
      "step": 6060
    },
    {
      "epoch": 0.10656788216085254,
      "grad_norm": 0.05809611936487027,
      "learning_rate": 0.0001999991274196894,
      "loss": 0.8687,
      "step": 6070
    },
    {
      "epoch": 0.10674344704085395,
      "grad_norm": 0.0746887531457463,
      "learning_rate": 0.00019999908000544126,
      "loss": 0.8582,
      "step": 6080
    },
    {
      "epoch": 0.10691901192085536,
      "grad_norm": 0.09318559064731355,
      "learning_rate": 0.0001999990313368558,
      "loss": 0.8723,
      "step": 6090
    },
    {
      "epoch": 0.10709457680085675,
      "grad_norm": 0.08538566387847724,
      "learning_rate": 0.0001999989814139336,
      "loss": 0.8649,
      "step": 6100
    },
    {
      "epoch": 0.10727014168085816,
      "grad_norm": 0.09452315098754602,
      "learning_rate": 0.00019999893023667535,
      "loss": 0.8688,
      "step": 6110
    },
    {
      "epoch": 0.10744570656085957,
      "grad_norm": 0.07950987561053775,
      "learning_rate": 0.00019999887780508172,
      "loss": 0.871,
      "step": 6120
    },
    {
      "epoch": 0.10762127144086096,
      "grad_norm": 0.07096817035639305,
      "learning_rate": 0.00019999882411915332,
      "loss": 0.8665,
      "step": 6130
    },
    {
      "epoch": 0.10779683632086237,
      "grad_norm": 0.09326225773926203,
      "learning_rate": 0.00019999876917889083,
      "loss": 0.8674,
      "step": 6140
    },
    {
      "epoch": 0.10797240120086378,
      "grad_norm": 0.06126560978801633,
      "learning_rate": 0.00019999871298429495,
      "loss": 0.8643,
      "step": 6150
    },
    {
      "epoch": 0.10814796608086519,
      "grad_norm": 0.0797053726819108,
      "learning_rate": 0.00019999865553536642,
      "loss": 0.8654,
      "step": 6160
    },
    {
      "epoch": 0.10832353096086658,
      "grad_norm": 0.08125984243510861,
      "learning_rate": 0.00019999859683210596,
      "loss": 0.858,
      "step": 6170
    },
    {
      "epoch": 0.10849909584086799,
      "grad_norm": 0.08431219112265906,
      "learning_rate": 0.0001999985368745143,
      "loss": 0.8698,
      "step": 6180
    },
    {
      "epoch": 0.1086746607208694,
      "grad_norm": 0.06438939115349626,
      "learning_rate": 0.0001999984756625922,
      "loss": 0.8683,
      "step": 6190
    },
    {
      "epoch": 0.1088502256008708,
      "grad_norm": 0.07449894398507802,
      "learning_rate": 0.00019999841319634046,
      "loss": 0.8705,
      "step": 6200
    },
    {
      "epoch": 0.1090257904808722,
      "grad_norm": 0.08540272979984538,
      "learning_rate": 0.0001999983494757598,
      "loss": 0.8647,
      "step": 6210
    },
    {
      "epoch": 0.10920135536087361,
      "grad_norm": 0.07495797825829166,
      "learning_rate": 0.00019999828450085113,
      "loss": 0.868,
      "step": 6220
    },
    {
      "epoch": 0.10937692024087502,
      "grad_norm": 0.10085252850674534,
      "learning_rate": 0.0001999982182716152,
      "loss": 0.8687,
      "step": 6230
    },
    {
      "epoch": 0.10955248512087642,
      "grad_norm": 0.10049466153529564,
      "learning_rate": 0.0001999981507880529,
      "loss": 0.8674,
      "step": 6240
    },
    {
      "epoch": 0.10972805000087782,
      "grad_norm": 0.08928114376815789,
      "learning_rate": 0.000199998082050165,
      "loss": 0.8701,
      "step": 6250
    },
    {
      "epoch": 0.10990361488087923,
      "grad_norm": 0.07446344602630224,
      "learning_rate": 0.0001999980120579525,
      "loss": 0.8691,
      "step": 6260
    },
    {
      "epoch": 0.11007917976088064,
      "grad_norm": 0.06506682042836333,
      "learning_rate": 0.00019999794081141615,
      "loss": 0.8633,
      "step": 6270
    },
    {
      "epoch": 0.11025474464088204,
      "grad_norm": 0.07356071238125399,
      "learning_rate": 0.0001999978683105569,
      "loss": 0.8709,
      "step": 6280
    },
    {
      "epoch": 0.11043030952088344,
      "grad_norm": 0.08262323934648089,
      "learning_rate": 0.00019999779455537575,
      "loss": 0.8636,
      "step": 6290
    },
    {
      "epoch": 0.11060587440088485,
      "grad_norm": 0.10293518138400064,
      "learning_rate": 0.00019999771954587356,
      "loss": 0.869,
      "step": 6300
    },
    {
      "epoch": 0.11078143928088625,
      "grad_norm": 0.09351369426369147,
      "learning_rate": 0.0001999976432820513,
      "loss": 0.8665,
      "step": 6310
    },
    {
      "epoch": 0.11095700416088766,
      "grad_norm": 0.06767427898168131,
      "learning_rate": 0.00019999756576390986,
      "loss": 0.8683,
      "step": 6320
    },
    {
      "epoch": 0.11113256904088906,
      "grad_norm": 0.08935931793196067,
      "learning_rate": 0.00019999748699145036,
      "loss": 0.8712,
      "step": 6330
    },
    {
      "epoch": 0.11130813392089046,
      "grad_norm": 0.10245129977516058,
      "learning_rate": 0.0001999974069646737,
      "loss": 0.873,
      "step": 6340
    },
    {
      "epoch": 0.11148369880089187,
      "grad_norm": 0.06443295472017836,
      "learning_rate": 0.00019999732568358092,
      "loss": 0.8672,
      "step": 6350
    },
    {
      "epoch": 0.11165926368089328,
      "grad_norm": 0.12901080808613738,
      "learning_rate": 0.00019999724314817308,
      "loss": 0.8722,
      "step": 6360
    },
    {
      "epoch": 0.11183482856089468,
      "grad_norm": 0.08829695413936106,
      "learning_rate": 0.00019999715935845115,
      "loss": 0.862,
      "step": 6370
    },
    {
      "epoch": 0.11201039344089608,
      "grad_norm": 0.12886658976115806,
      "learning_rate": 0.00019999707431441626,
      "loss": 0.8644,
      "step": 6380
    },
    {
      "epoch": 0.11218595832089749,
      "grad_norm": 0.08992976476210034,
      "learning_rate": 0.00019999698801606946,
      "loss": 0.8676,
      "step": 6390
    },
    {
      "epoch": 0.11236152320089889,
      "grad_norm": 0.06288656344847394,
      "learning_rate": 0.00019999690046341184,
      "loss": 0.8665,
      "step": 6400
    },
    {
      "epoch": 0.1125370880809003,
      "grad_norm": 0.10361696689855823,
      "learning_rate": 0.00019999681165644453,
      "loss": 0.8634,
      "step": 6410
    },
    {
      "epoch": 0.1127126529609017,
      "grad_norm": 0.08416559105822072,
      "learning_rate": 0.00019999672159516866,
      "loss": 0.8666,
      "step": 6420
    },
    {
      "epoch": 0.11288821784090311,
      "grad_norm": 0.08755783557908572,
      "learning_rate": 0.00019999663027958536,
      "loss": 0.8644,
      "step": 6430
    },
    {
      "epoch": 0.1130637827209045,
      "grad_norm": 0.07017255913653044,
      "learning_rate": 0.00019999653770969573,
      "loss": 0.8648,
      "step": 6440
    },
    {
      "epoch": 0.11323934760090591,
      "grad_norm": 0.07583121919386751,
      "learning_rate": 0.00019999644388550103,
      "loss": 0.8631,
      "step": 6450
    },
    {
      "epoch": 0.11341491248090732,
      "grad_norm": 0.07455572372353134,
      "learning_rate": 0.00019999634880700243,
      "loss": 0.8709,
      "step": 6460
    },
    {
      "epoch": 0.11359047736090873,
      "grad_norm": 0.06354505214521729,
      "learning_rate": 0.00019999625247420105,
      "loss": 0.8624,
      "step": 6470
    },
    {
      "epoch": 0.11376604224091012,
      "grad_norm": 0.09167279147972723,
      "learning_rate": 0.00019999615488709827,
      "loss": 0.8628,
      "step": 6480
    },
    {
      "epoch": 0.11394160712091153,
      "grad_norm": 0.07530994927582323,
      "learning_rate": 0.00019999605604569519,
      "loss": 0.868,
      "step": 6490
    },
    {
      "epoch": 0.11411717200091294,
      "grad_norm": 0.10898181248596628,
      "learning_rate": 0.00019999595594999309,
      "loss": 0.8621,
      "step": 6500
    },
    {
      "epoch": 0.11429273688091435,
      "grad_norm": 0.06594956241740206,
      "learning_rate": 0.00019999585459999327,
      "loss": 0.8649,
      "step": 6510
    },
    {
      "epoch": 0.11446830176091574,
      "grad_norm": 0.08614647134189739,
      "learning_rate": 0.00019999575199569703,
      "loss": 0.8597,
      "step": 6520
    },
    {
      "epoch": 0.11464386664091715,
      "grad_norm": 0.09465561323595492,
      "learning_rate": 0.00019999564813710558,
      "loss": 0.871,
      "step": 6530
    },
    {
      "epoch": 0.11481943152091856,
      "grad_norm": 0.07353382388727248,
      "learning_rate": 0.00019999554302422033,
      "loss": 0.8583,
      "step": 6540
    },
    {
      "epoch": 0.11499499640091997,
      "grad_norm": 0.07259023633604288,
      "learning_rate": 0.00019999543665704262,
      "loss": 0.8648,
      "step": 6550
    },
    {
      "epoch": 0.11517056128092136,
      "grad_norm": 0.08061027167670781,
      "learning_rate": 0.00019999532903557367,
      "loss": 0.8746,
      "step": 6560
    },
    {
      "epoch": 0.11534612616092277,
      "grad_norm": 0.07897083524581373,
      "learning_rate": 0.00019999522015981498,
      "loss": 0.862,
      "step": 6570
    },
    {
      "epoch": 0.11552169104092418,
      "grad_norm": 0.09370825365000673,
      "learning_rate": 0.00019999511002976786,
      "loss": 0.8572,
      "step": 6580
    },
    {
      "epoch": 0.11569725592092558,
      "grad_norm": 0.11887697661952216,
      "learning_rate": 0.00019999499864543376,
      "loss": 0.8621,
      "step": 6590
    },
    {
      "epoch": 0.11587282080092698,
      "grad_norm": 0.09492930258042202,
      "learning_rate": 0.00019999488600681403,
      "loss": 0.8703,
      "step": 6600
    },
    {
      "epoch": 0.11604838568092839,
      "grad_norm": 0.07337217730463012,
      "learning_rate": 0.0001999947721139101,
      "loss": 0.8555,
      "step": 6610
    },
    {
      "epoch": 0.1162239505609298,
      "grad_norm": 0.08607413425578857,
      "learning_rate": 0.00019999465696672345,
      "loss": 0.8722,
      "step": 6620
    },
    {
      "epoch": 0.11639951544093119,
      "grad_norm": 0.07886298836159036,
      "learning_rate": 0.00019999454056525553,
      "loss": 0.8643,
      "step": 6630
    },
    {
      "epoch": 0.1165750803209326,
      "grad_norm": 0.11743638197999838,
      "learning_rate": 0.00019999442290950784,
      "loss": 0.8701,
      "step": 6640
    },
    {
      "epoch": 0.116750645200934,
      "grad_norm": 0.07461184649095323,
      "learning_rate": 0.00019999430399948178,
      "loss": 0.8628,
      "step": 6650
    },
    {
      "epoch": 0.11692621008093541,
      "grad_norm": 0.07431165118958825,
      "learning_rate": 0.00019999418383517893,
      "loss": 0.8672,
      "step": 6660
    },
    {
      "epoch": 0.11710177496093681,
      "grad_norm": 0.09049083204237948,
      "learning_rate": 0.0001999940624166008,
      "loss": 0.8604,
      "step": 6670
    },
    {
      "epoch": 0.11727733984093822,
      "grad_norm": 0.08053893181504287,
      "learning_rate": 0.00019999393974374897,
      "loss": 0.8622,
      "step": 6680
    },
    {
      "epoch": 0.11745290472093962,
      "grad_norm": 0.06790978019891217,
      "learning_rate": 0.0001999938158166249,
      "loss": 0.866,
      "step": 6690
    },
    {
      "epoch": 0.11762846960094103,
      "grad_norm": 0.06941697690309509,
      "learning_rate": 0.00019999369063523025,
      "loss": 0.8688,
      "step": 6700
    },
    {
      "epoch": 0.11780403448094243,
      "grad_norm": 0.08640293820886544,
      "learning_rate": 0.00019999356419956655,
      "loss": 0.8663,
      "step": 6710
    },
    {
      "epoch": 0.11797959936094383,
      "grad_norm": 0.08173441778188999,
      "learning_rate": 0.0001999934365096354,
      "loss": 0.8747,
      "step": 6720
    },
    {
      "epoch": 0.11815516424094524,
      "grad_norm": 0.08740826126255855,
      "learning_rate": 0.00019999330756543846,
      "loss": 0.8609,
      "step": 6730
    },
    {
      "epoch": 0.11833072912094665,
      "grad_norm": 0.06218179823133969,
      "learning_rate": 0.00019999317736697737,
      "loss": 0.8676,
      "step": 6740
    },
    {
      "epoch": 0.11850629400094805,
      "grad_norm": 0.08341363142816993,
      "learning_rate": 0.0001999930459142537,
      "loss": 0.8663,
      "step": 6750
    },
    {
      "epoch": 0.11868185888094945,
      "grad_norm": 0.08311774202815386,
      "learning_rate": 0.00019999291320726918,
      "loss": 0.8609,
      "step": 6760
    },
    {
      "epoch": 0.11885742376095086,
      "grad_norm": 0.09567146188622888,
      "learning_rate": 0.00019999277924602547,
      "loss": 0.8746,
      "step": 6770
    },
    {
      "epoch": 0.11903298864095227,
      "grad_norm": 0.10158800856109962,
      "learning_rate": 0.0001999926440305243,
      "loss": 0.8654,
      "step": 6780
    },
    {
      "epoch": 0.11920855352095366,
      "grad_norm": 0.07712153256744408,
      "learning_rate": 0.00019999250756076736,
      "loss": 0.864,
      "step": 6790
    },
    {
      "epoch": 0.11938411840095507,
      "grad_norm": 0.060324355022765515,
      "learning_rate": 0.00019999236983675635,
      "loss": 0.8637,
      "step": 6800
    },
    {
      "epoch": 0.11955968328095648,
      "grad_norm": 0.08411255978235048,
      "learning_rate": 0.00019999223085849307,
      "loss": 0.8536,
      "step": 6810
    },
    {
      "epoch": 0.11973524816095789,
      "grad_norm": 0.08102618790704574,
      "learning_rate": 0.0001999920906259792,
      "loss": 0.8685,
      "step": 6820
    },
    {
      "epoch": 0.11991081304095928,
      "grad_norm": 0.08160311119269288,
      "learning_rate": 0.0001999919491392166,
      "loss": 0.8595,
      "step": 6830
    },
    {
      "epoch": 0.12008637792096069,
      "grad_norm": 0.12122808782140802,
      "learning_rate": 0.00019999180639820708,
      "loss": 0.8644,
      "step": 6840
    },
    {
      "epoch": 0.1202619428009621,
      "grad_norm": 0.07563399350941598,
      "learning_rate": 0.00019999166240295234,
      "loss": 0.8635,
      "step": 6850
    },
    {
      "epoch": 0.1204375076809635,
      "grad_norm": 0.08006821286090464,
      "learning_rate": 0.0001999915171534543,
      "loss": 0.8587,
      "step": 6860
    },
    {
      "epoch": 0.1206130725609649,
      "grad_norm": 0.08339756380077658,
      "learning_rate": 0.00019999137064971475,
      "loss": 0.8651,
      "step": 6870
    },
    {
      "epoch": 0.12078863744096631,
      "grad_norm": 0.08399774951682784,
      "learning_rate": 0.00019999122289173556,
      "loss": 0.86,
      "step": 6880
    },
    {
      "epoch": 0.12096420232096772,
      "grad_norm": 0.08516551542864127,
      "learning_rate": 0.00019999107387951864,
      "loss": 0.859,
      "step": 6890
    },
    {
      "epoch": 0.12113976720096911,
      "grad_norm": 0.07829214773855364,
      "learning_rate": 0.0001999909236130658,
      "loss": 0.8607,
      "step": 6900
    },
    {
      "epoch": 0.12131533208097052,
      "grad_norm": 0.07150225604036,
      "learning_rate": 0.000199990772092379,
      "loss": 0.8662,
      "step": 6910
    },
    {
      "epoch": 0.12149089696097193,
      "grad_norm": 0.07477752417774537,
      "learning_rate": 0.00019999061931746012,
      "loss": 0.8646,
      "step": 6920
    },
    {
      "epoch": 0.12166646184097334,
      "grad_norm": 0.11078704328903252,
      "learning_rate": 0.00019999046528831118,
      "loss": 0.8629,
      "step": 6930
    },
    {
      "epoch": 0.12184202672097473,
      "grad_norm": 0.12418489595002523,
      "learning_rate": 0.000199990310004934,
      "loss": 0.8559,
      "step": 6940
    },
    {
      "epoch": 0.12201759160097614,
      "grad_norm": 0.06998456316081583,
      "learning_rate": 0.0001999901534673307,
      "loss": 0.8638,
      "step": 6950
    },
    {
      "epoch": 0.12219315648097755,
      "grad_norm": 0.06729024179566495,
      "learning_rate": 0.00019998999567550312,
      "loss": 0.8656,
      "step": 6960
    },
    {
      "epoch": 0.12236872136097895,
      "grad_norm": 0.07826644010008393,
      "learning_rate": 0.00019998983662945338,
      "loss": 0.8691,
      "step": 6970
    },
    {
      "epoch": 0.12254428624098035,
      "grad_norm": 0.09028200218501865,
      "learning_rate": 0.00019998967632918338,
      "loss": 0.8646,
      "step": 6980
    },
    {
      "epoch": 0.12271985112098176,
      "grad_norm": 0.06819629080257669,
      "learning_rate": 0.00019998951477469528,
      "loss": 0.8699,
      "step": 6990
    },
    {
      "epoch": 0.12289541600098317,
      "grad_norm": 0.09800816698358492,
      "learning_rate": 0.000199989351965991,
      "loss": 0.8551,
      "step": 7000
    },
    {
      "epoch": 0.12307098088098457,
      "grad_norm": 0.07584536567529526,
      "learning_rate": 0.00019998918790307266,
      "loss": 0.8674,
      "step": 7010
    },
    {
      "epoch": 0.12324654576098597,
      "grad_norm": 0.07651475306048983,
      "learning_rate": 0.00019998902258594234,
      "loss": 0.8651,
      "step": 7020
    },
    {
      "epoch": 0.12342211064098738,
      "grad_norm": 0.07451610141984215,
      "learning_rate": 0.00019998885601460215,
      "loss": 0.8679,
      "step": 7030
    },
    {
      "epoch": 0.12359767552098878,
      "grad_norm": 0.06829257767750715,
      "learning_rate": 0.00019998868818905417,
      "loss": 0.8603,
      "step": 7040
    },
    {
      "epoch": 0.12377324040099019,
      "grad_norm": 0.11661286758607511,
      "learning_rate": 0.00019998851910930056,
      "loss": 0.858,
      "step": 7050
    },
    {
      "epoch": 0.12394880528099159,
      "grad_norm": 0.07533027381527405,
      "learning_rate": 0.00019998834877534341,
      "loss": 0.8664,
      "step": 7060
    },
    {
      "epoch": 0.124124370160993,
      "grad_norm": 0.08518685375615481,
      "learning_rate": 0.00019998817718718497,
      "loss": 0.8674,
      "step": 7070
    },
    {
      "epoch": 0.1242999350409944,
      "grad_norm": 0.11718953267785284,
      "learning_rate": 0.0001999880043448273,
      "loss": 0.8597,
      "step": 7080
    },
    {
      "epoch": 0.12447549992099581,
      "grad_norm": 0.06041968996360231,
      "learning_rate": 0.00019998783024827266,
      "loss": 0.8591,
      "step": 7090
    },
    {
      "epoch": 0.1246510648009972,
      "grad_norm": 0.08895705776728655,
      "learning_rate": 0.0001999876548975232,
      "loss": 0.8688,
      "step": 7100
    },
    {
      "epoch": 0.12482662968099861,
      "grad_norm": 0.09209232934596556,
      "learning_rate": 0.0001999874782925812,
      "loss": 0.8631,
      "step": 7110
    },
    {
      "epoch": 0.125002194561,
      "grad_norm": 0.073680188021475,
      "learning_rate": 0.00019998730043344895,
      "loss": 0.8606,
      "step": 7120
    },
    {
      "epoch": 0.12517775944100143,
      "grad_norm": 0.06970140115454933,
      "learning_rate": 0.00019998712132012858,
      "loss": 0.8626,
      "step": 7130
    },
    {
      "epoch": 0.12535332432100282,
      "grad_norm": 0.07195509838980164,
      "learning_rate": 0.00019998694095262238,
      "loss": 0.866,
      "step": 7140
    },
    {
      "epoch": 0.12552888920100422,
      "grad_norm": 0.09359695167686342,
      "learning_rate": 0.00019998675933093272,
      "loss": 0.8622,
      "step": 7150
    },
    {
      "epoch": 0.12570445408100564,
      "grad_norm": 0.0945085929979948,
      "learning_rate": 0.00019998657645506183,
      "loss": 0.8602,
      "step": 7160
    },
    {
      "epoch": 0.12588001896100703,
      "grad_norm": 0.10430049127986885,
      "learning_rate": 0.00019998639232501203,
      "loss": 0.8669,
      "step": 7170
    },
    {
      "epoch": 0.12605558384100846,
      "grad_norm": 0.06948410625736152,
      "learning_rate": 0.0001999862069407857,
      "loss": 0.8607,
      "step": 7180
    },
    {
      "epoch": 0.12623114872100985,
      "grad_norm": 0.06202101545318056,
      "learning_rate": 0.00019998602030238513,
      "loss": 0.8685,
      "step": 7190
    },
    {
      "epoch": 0.12640671360101124,
      "grad_norm": 0.08321540080321779,
      "learning_rate": 0.00019998583240981273,
      "loss": 0.8628,
      "step": 7200
    },
    {
      "epoch": 0.12658227848101267,
      "grad_norm": 0.12479142934717083,
      "learning_rate": 0.00019998564326307087,
      "loss": 0.8648,
      "step": 7210
    },
    {
      "epoch": 0.12675784336101406,
      "grad_norm": 0.0678888725635754,
      "learning_rate": 0.0001999854528621619,
      "loss": 0.8617,
      "step": 7220
    },
    {
      "epoch": 0.12693340824101546,
      "grad_norm": 0.06172599705488479,
      "learning_rate": 0.0001999852612070883,
      "loss": 0.866,
      "step": 7230
    },
    {
      "epoch": 0.12710897312101688,
      "grad_norm": 0.07005113820190416,
      "learning_rate": 0.0001999850682978525,
      "loss": 0.8659,
      "step": 7240
    },
    {
      "epoch": 0.12728453800101827,
      "grad_norm": 0.07979955380921058,
      "learning_rate": 0.00019998487413445685,
      "loss": 0.8588,
      "step": 7250
    },
    {
      "epoch": 0.1274601028810197,
      "grad_norm": 0.06202932181058022,
      "learning_rate": 0.0001999846787169039,
      "loss": 0.8717,
      "step": 7260
    },
    {
      "epoch": 0.1276356677610211,
      "grad_norm": 0.07786650945228414,
      "learning_rate": 0.00019998448204519613,
      "loss": 0.8548,
      "step": 7270
    },
    {
      "epoch": 0.12781123264102248,
      "grad_norm": 0.09704058046716232,
      "learning_rate": 0.00019998428411933597,
      "loss": 0.8667,
      "step": 7280
    },
    {
      "epoch": 0.1279867975210239,
      "grad_norm": 0.13141578953770494,
      "learning_rate": 0.00019998408493932595,
      "loss": 0.8643,
      "step": 7290
    },
    {
      "epoch": 0.1281623624010253,
      "grad_norm": 0.09946917261695024,
      "learning_rate": 0.00019998388450516864,
      "loss": 0.86,
      "step": 7300
    },
    {
      "epoch": 0.1283379272810267,
      "grad_norm": 0.06468876121153408,
      "learning_rate": 0.00019998368281686648,
      "loss": 0.8663,
      "step": 7310
    },
    {
      "epoch": 0.12851349216102811,
      "grad_norm": 0.08299448610774098,
      "learning_rate": 0.00019998347987442217,
      "loss": 0.87,
      "step": 7320
    },
    {
      "epoch": 0.1286890570410295,
      "grad_norm": 0.06991833842567641,
      "learning_rate": 0.00019998327567783816,
      "loss": 0.8663,
      "step": 7330
    },
    {
      "epoch": 0.12886462192103093,
      "grad_norm": 0.06438973276071469,
      "learning_rate": 0.00019998307022711708,
      "loss": 0.8647,
      "step": 7340
    },
    {
      "epoch": 0.12904018680103232,
      "grad_norm": 0.07961236381273255,
      "learning_rate": 0.0001999828635222615,
      "loss": 0.864,
      "step": 7350
    },
    {
      "epoch": 0.12921575168103372,
      "grad_norm": 0.07903442970479649,
      "learning_rate": 0.0001999826555632741,
      "loss": 0.8683,
      "step": 7360
    },
    {
      "epoch": 0.12939131656103514,
      "grad_norm": 0.09440078614171286,
      "learning_rate": 0.00019998244635015752,
      "loss": 0.8723,
      "step": 7370
    },
    {
      "epoch": 0.12956688144103654,
      "grad_norm": 0.07896254632174128,
      "learning_rate": 0.00019998223588291433,
      "loss": 0.8684,
      "step": 7380
    },
    {
      "epoch": 0.12974244632103793,
      "grad_norm": 0.10629569772396467,
      "learning_rate": 0.00019998202416154724,
      "loss": 0.8642,
      "step": 7390
    },
    {
      "epoch": 0.12991801120103935,
      "grad_norm": 0.08337568127340755,
      "learning_rate": 0.00019998181118605893,
      "loss": 0.8572,
      "step": 7400
    },
    {
      "epoch": 0.13009357608104075,
      "grad_norm": 0.12933104531436018,
      "learning_rate": 0.00019998159695645213,
      "loss": 0.8587,
      "step": 7410
    },
    {
      "epoch": 0.13026914096104214,
      "grad_norm": 0.0624210331651339,
      "learning_rate": 0.00019998138147272953,
      "loss": 0.8695,
      "step": 7420
    },
    {
      "epoch": 0.13044470584104356,
      "grad_norm": 0.06198528454553228,
      "learning_rate": 0.00019998116473489384,
      "loss": 0.8699,
      "step": 7430
    },
    {
      "epoch": 0.13062027072104496,
      "grad_norm": 0.08986008441312672,
      "learning_rate": 0.00019998094674294778,
      "loss": 0.8563,
      "step": 7440
    },
    {
      "epoch": 0.13079583560104638,
      "grad_norm": 0.09576007207439642,
      "learning_rate": 0.0001999807274968942,
      "loss": 0.8641,
      "step": 7450
    },
    {
      "epoch": 0.13097140048104777,
      "grad_norm": 0.07811431741574616,
      "learning_rate": 0.00019998050699673585,
      "loss": 0.8722,
      "step": 7460
    },
    {
      "epoch": 0.13114696536104917,
      "grad_norm": 0.07436090882982085,
      "learning_rate": 0.0001999802852424755,
      "loss": 0.8633,
      "step": 7470
    },
    {
      "epoch": 0.1313225302410506,
      "grad_norm": 0.06900096948287676,
      "learning_rate": 0.00019998006223411594,
      "loss": 0.8618,
      "step": 7480
    },
    {
      "epoch": 0.13149809512105198,
      "grad_norm": 0.06991603081005608,
      "learning_rate": 0.00019997983797166003,
      "loss": 0.8622,
      "step": 7490
    },
    {
      "epoch": 0.13167366000105338,
      "grad_norm": 0.08336308943038788,
      "learning_rate": 0.0001999796124551106,
      "loss": 0.8614,
      "step": 7500
    },
    {
      "epoch": 0.1318492248810548,
      "grad_norm": 0.05857836997335782,
      "learning_rate": 0.00019997938568447053,
      "loss": 0.8696,
      "step": 7510
    },
    {
      "epoch": 0.1320247897610562,
      "grad_norm": 0.061331575939083324,
      "learning_rate": 0.00019997915765974263,
      "loss": 0.8627,
      "step": 7520
    },
    {
      "epoch": 0.13220035464105762,
      "grad_norm": 0.12452096382272163,
      "learning_rate": 0.0001999789283809299,
      "loss": 0.8601,
      "step": 7530
    },
    {
      "epoch": 0.132375919521059,
      "grad_norm": 0.07378478604646697,
      "learning_rate": 0.00019997869784803515,
      "loss": 0.8678,
      "step": 7540
    },
    {
      "epoch": 0.1325514844010604,
      "grad_norm": 0.08915917243276575,
      "learning_rate": 0.00019997846606106134,
      "loss": 0.8671,
      "step": 7550
    },
    {
      "epoch": 0.13272704928106183,
      "grad_norm": 0.06964537694834418,
      "learning_rate": 0.00019997823302001134,
      "loss": 0.8645,
      "step": 7560
    },
    {
      "epoch": 0.13290261416106322,
      "grad_norm": 0.06887584061236975,
      "learning_rate": 0.00019997799872488822,
      "loss": 0.8681,
      "step": 7570
    },
    {
      "epoch": 0.13307817904106461,
      "grad_norm": 0.07338452277336925,
      "learning_rate": 0.00019997776317569487,
      "loss": 0.8669,
      "step": 7580
    },
    {
      "epoch": 0.13325374392106604,
      "grad_norm": 0.07527237758228512,
      "learning_rate": 0.00019997752637243427,
      "loss": 0.8663,
      "step": 7590
    },
    {
      "epoch": 0.13342930880106743,
      "grad_norm": 0.07122887723854143,
      "learning_rate": 0.0001999772883151095,
      "loss": 0.8628,
      "step": 7600
    },
    {
      "epoch": 0.13360487368106883,
      "grad_norm": 0.08050908182130337,
      "learning_rate": 0.00019997704900372347,
      "loss": 0.8676,
      "step": 7610
    },
    {
      "epoch": 0.13378043856107025,
      "grad_norm": 0.06773373887509727,
      "learning_rate": 0.00019997680843827925,
      "loss": 0.8644,
      "step": 7620
    },
    {
      "epoch": 0.13395600344107164,
      "grad_norm": 0.07056489320852093,
      "learning_rate": 0.00019997656661877993,
      "loss": 0.8664,
      "step": 7630
    },
    {
      "epoch": 0.13413156832107306,
      "grad_norm": 0.08452788695089015,
      "learning_rate": 0.00019997632354522852,
      "loss": 0.8685,
      "step": 7640
    },
    {
      "epoch": 0.13430713320107446,
      "grad_norm": 0.07556254172026002,
      "learning_rate": 0.00019997607921762813,
      "loss": 0.8666,
      "step": 7650
    },
    {
      "epoch": 0.13448269808107585,
      "grad_norm": 0.0736956694792162,
      "learning_rate": 0.00019997583363598186,
      "loss": 0.8741,
      "step": 7660
    },
    {
      "epoch": 0.13465826296107727,
      "grad_norm": 0.09490239388328225,
      "learning_rate": 0.0001999755868002928,
      "loss": 0.8636,
      "step": 7670
    },
    {
      "epoch": 0.13483382784107867,
      "grad_norm": 0.07655251831927491,
      "learning_rate": 0.00019997533871056407,
      "loss": 0.8697,
      "step": 7680
    },
    {
      "epoch": 0.13500939272108006,
      "grad_norm": 0.10343269470152557,
      "learning_rate": 0.00019997508936679884,
      "loss": 0.8612,
      "step": 7690
    },
    {
      "epoch": 0.13518495760108148,
      "grad_norm": 0.06448586819635803,
      "learning_rate": 0.00019997483876900028,
      "loss": 0.8672,
      "step": 7700
    },
    {
      "epoch": 0.13536052248108288,
      "grad_norm": 0.07929609900453134,
      "learning_rate": 0.0001999745869171715,
      "loss": 0.8705,
      "step": 7710
    },
    {
      "epoch": 0.1355360873610843,
      "grad_norm": 0.07542779050514448,
      "learning_rate": 0.00019997433381131578,
      "loss": 0.862,
      "step": 7720
    },
    {
      "epoch": 0.1357116522410857,
      "grad_norm": 0.07390768465646513,
      "learning_rate": 0.00019997407945143627,
      "loss": 0.8683,
      "step": 7730
    },
    {
      "epoch": 0.1358872171210871,
      "grad_norm": 0.06234469323883626,
      "learning_rate": 0.0001999738238375362,
      "loss": 0.8698,
      "step": 7740
    },
    {
      "epoch": 0.1360627820010885,
      "grad_norm": 0.068467463857245,
      "learning_rate": 0.00019997356696961881,
      "loss": 0.8559,
      "step": 7750
    },
    {
      "epoch": 0.1362383468810899,
      "grad_norm": 0.08492980228703197,
      "learning_rate": 0.00019997330884768738,
      "loss": 0.8625,
      "step": 7760
    },
    {
      "epoch": 0.1364139117610913,
      "grad_norm": 0.06207522791229603,
      "learning_rate": 0.00019997304947174513,
      "loss": 0.8551,
      "step": 7770
    },
    {
      "epoch": 0.13658947664109272,
      "grad_norm": 0.0725808457815244,
      "learning_rate": 0.00019997278884179538,
      "loss": 0.861,
      "step": 7780
    },
    {
      "epoch": 0.13676504152109412,
      "grad_norm": 0.08033819820833364,
      "learning_rate": 0.00019997252695784143,
      "loss": 0.8689,
      "step": 7790
    },
    {
      "epoch": 0.13694060640109554,
      "grad_norm": 0.08982211061950059,
      "learning_rate": 0.00019997226381988662,
      "loss": 0.871,
      "step": 7800
    },
    {
      "epoch": 0.13711617128109693,
      "grad_norm": 0.07131737353291,
      "learning_rate": 0.0001999719994279342,
      "loss": 0.8598,
      "step": 7810
    },
    {
      "epoch": 0.13729173616109833,
      "grad_norm": 0.08004395486306835,
      "learning_rate": 0.00019997173378198766,
      "loss": 0.8571,
      "step": 7820
    },
    {
      "epoch": 0.13746730104109975,
      "grad_norm": 0.07305514861101732,
      "learning_rate": 0.0001999714668820502,
      "loss": 0.858,
      "step": 7830
    },
    {
      "epoch": 0.13764286592110114,
      "grad_norm": 0.05685932586683221,
      "learning_rate": 0.00019997119872812536,
      "loss": 0.8657,
      "step": 7840
    },
    {
      "epoch": 0.13781843080110254,
      "grad_norm": 0.06504601064166171,
      "learning_rate": 0.00019997092932021643,
      "loss": 0.8637,
      "step": 7850
    },
    {
      "epoch": 0.13799399568110396,
      "grad_norm": 0.06000046499294888,
      "learning_rate": 0.00019997065865832685,
      "loss": 0.8707,
      "step": 7860
    },
    {
      "epoch": 0.13816956056110535,
      "grad_norm": 0.07310407555002096,
      "learning_rate": 0.0001999703867424601,
      "loss": 0.8655,
      "step": 7870
    },
    {
      "epoch": 0.13834512544110675,
      "grad_norm": 0.07279701460333302,
      "learning_rate": 0.00019997011357261954,
      "loss": 0.8607,
      "step": 7880
    },
    {
      "epoch": 0.13852069032110817,
      "grad_norm": 0.05683728055548074,
      "learning_rate": 0.00019996983914880866,
      "loss": 0.8676,
      "step": 7890
    },
    {
      "epoch": 0.13869625520110956,
      "grad_norm": 0.0694588047563135,
      "learning_rate": 0.000199969563471031,
      "loss": 0.8532,
      "step": 7900
    },
    {
      "epoch": 0.13887182008111099,
      "grad_norm": 0.06397035234155944,
      "learning_rate": 0.00019996928653928997,
      "loss": 0.8636,
      "step": 7910
    },
    {
      "epoch": 0.13904738496111238,
      "grad_norm": 0.0593671427715556,
      "learning_rate": 0.00019996900835358915,
      "loss": 0.8663,
      "step": 7920
    },
    {
      "epoch": 0.13922294984111377,
      "grad_norm": 0.06991832355467917,
      "learning_rate": 0.000199968728913932,
      "loss": 0.8663,
      "step": 7930
    },
    {
      "epoch": 0.1393985147211152,
      "grad_norm": 0.08800082443723872,
      "learning_rate": 0.00019996844822032207,
      "loss": 0.8659,
      "step": 7940
    },
    {
      "epoch": 0.1395740796011166,
      "grad_norm": 0.06615879372452065,
      "learning_rate": 0.00019996816627276294,
      "loss": 0.8728,
      "step": 7950
    },
    {
      "epoch": 0.13974964448111798,
      "grad_norm": 0.074350658551748,
      "learning_rate": 0.00019996788307125819,
      "loss": 0.8609,
      "step": 7960
    },
    {
      "epoch": 0.1399252093611194,
      "grad_norm": 0.09286392327233237,
      "learning_rate": 0.00019996759861581138,
      "loss": 0.8643,
      "step": 7970
    },
    {
      "epoch": 0.1401007742411208,
      "grad_norm": 0.08870220132813415,
      "learning_rate": 0.00019996731290642614,
      "loss": 0.8627,
      "step": 7980
    },
    {
      "epoch": 0.14027633912112222,
      "grad_norm": 0.08546161863448751,
      "learning_rate": 0.00019996702594310607,
      "loss": 0.8751,
      "step": 7990
    },
    {
      "epoch": 0.14045190400112362,
      "grad_norm": 0.08198455410242135,
      "learning_rate": 0.00019996673772585484,
      "loss": 0.8625,
      "step": 8000
    },
    {
      "epoch": 0.140627468881125,
      "grad_norm": 0.10064043024920288,
      "learning_rate": 0.00019996644825467604,
      "loss": 0.8693,
      "step": 8010
    },
    {
      "epoch": 0.14080303376112643,
      "grad_norm": 0.06109620034871559,
      "learning_rate": 0.0001999661575295734,
      "loss": 0.8574,
      "step": 8020
    },
    {
      "epoch": 0.14097859864112783,
      "grad_norm": 0.0799266894829027,
      "learning_rate": 0.00019996586555055056,
      "loss": 0.8604,
      "step": 8030
    },
    {
      "epoch": 0.14115416352112922,
      "grad_norm": 0.068758539594058,
      "learning_rate": 0.00019996557231761127,
      "loss": 0.8668,
      "step": 8040
    },
    {
      "epoch": 0.14132972840113064,
      "grad_norm": 0.08429222535924202,
      "learning_rate": 0.00019996527783075919,
      "loss": 0.8634,
      "step": 8050
    },
    {
      "epoch": 0.14150529328113204,
      "grad_norm": 0.06745855639969878,
      "learning_rate": 0.00019996498208999804,
      "loss": 0.8711,
      "step": 8060
    },
    {
      "epoch": 0.14168085816113346,
      "grad_norm": 0.06647319369616542,
      "learning_rate": 0.00019996468509533163,
      "loss": 0.8694,
      "step": 8070
    },
    {
      "epoch": 0.14185642304113485,
      "grad_norm": 0.07391600089092751,
      "learning_rate": 0.0001999643868467637,
      "loss": 0.8646,
      "step": 8080
    },
    {
      "epoch": 0.14203198792113625,
      "grad_norm": 0.061424303117080026,
      "learning_rate": 0.000199964087344298,
      "loss": 0.8533,
      "step": 8090
    },
    {
      "epoch": 0.14220755280113767,
      "grad_norm": 0.07127557212068945,
      "learning_rate": 0.0001999637865879384,
      "loss": 0.8651,
      "step": 8100
    },
    {
      "epoch": 0.14238311768113907,
      "grad_norm": 0.07793301121588155,
      "learning_rate": 0.0001999634845776886,
      "loss": 0.864,
      "step": 8110
    },
    {
      "epoch": 0.14255868256114046,
      "grad_norm": 0.07168635736295148,
      "learning_rate": 0.00019996318131355252,
      "loss": 0.8574,
      "step": 8120
    },
    {
      "epoch": 0.14273424744114188,
      "grad_norm": 0.08339932699061672,
      "learning_rate": 0.00019996287679553394,
      "loss": 0.8597,
      "step": 8130
    },
    {
      "epoch": 0.14290981232114328,
      "grad_norm": 0.08332609478405832,
      "learning_rate": 0.00019996257102363675,
      "loss": 0.8637,
      "step": 8140
    },
    {
      "epoch": 0.14308537720114467,
      "grad_norm": 0.07660408820691532,
      "learning_rate": 0.00019996226399786484,
      "loss": 0.8623,
      "step": 8150
    },
    {
      "epoch": 0.1432609420811461,
      "grad_norm": 0.05308178861678294,
      "learning_rate": 0.00019996195571822206,
      "loss": 0.8749,
      "step": 8160
    },
    {
      "epoch": 0.1434365069611475,
      "grad_norm": 0.10094193393812588,
      "learning_rate": 0.00019996164618471234,
      "loss": 0.8578,
      "step": 8170
    },
    {
      "epoch": 0.1436120718411489,
      "grad_norm": 0.07224715398841358,
      "learning_rate": 0.00019996133539733959,
      "loss": 0.8718,
      "step": 8180
    },
    {
      "epoch": 0.1437876367211503,
      "grad_norm": 0.06172443073945614,
      "learning_rate": 0.00019996102335610778,
      "loss": 0.8681,
      "step": 8190
    },
    {
      "epoch": 0.1439632016011517,
      "grad_norm": 0.07628329806305256,
      "learning_rate": 0.0001999607100610208,
      "loss": 0.8692,
      "step": 8200
    },
    {
      "epoch": 0.14413876648115312,
      "grad_norm": 0.07257495093918971,
      "learning_rate": 0.0001999603955120827,
      "loss": 0.8706,
      "step": 8210
    },
    {
      "epoch": 0.1443143313611545,
      "grad_norm": 0.08652328706797828,
      "learning_rate": 0.0001999600797092974,
      "loss": 0.8612,
      "step": 8220
    },
    {
      "epoch": 0.1444898962411559,
      "grad_norm": 0.05985572657647732,
      "learning_rate": 0.0001999597626526689,
      "loss": 0.8724,
      "step": 8230
    },
    {
      "epoch": 0.14466546112115733,
      "grad_norm": 0.06907438585620056,
      "learning_rate": 0.00019995944434220126,
      "loss": 0.8658,
      "step": 8240
    },
    {
      "epoch": 0.14484102600115872,
      "grad_norm": 0.08038741236179113,
      "learning_rate": 0.0001999591247778985,
      "loss": 0.863,
      "step": 8250
    },
    {
      "epoch": 0.14501659088116015,
      "grad_norm": 0.09823924023835057,
      "learning_rate": 0.00019995880395976467,
      "loss": 0.8656,
      "step": 8260
    },
    {
      "epoch": 0.14519215576116154,
      "grad_norm": 0.063127811496834,
      "learning_rate": 0.00019995848188780383,
      "loss": 0.866,
      "step": 8270
    },
    {
      "epoch": 0.14536772064116293,
      "grad_norm": 0.09736354394589922,
      "learning_rate": 0.00019995815856202007,
      "loss": 0.8669,
      "step": 8280
    },
    {
      "epoch": 0.14554328552116436,
      "grad_norm": 0.09841097723260638,
      "learning_rate": 0.00019995783398241745,
      "loss": 0.8672,
      "step": 8290
    },
    {
      "epoch": 0.14571885040116575,
      "grad_norm": 0.07283386116763972,
      "learning_rate": 0.00019995750814900013,
      "loss": 0.8663,
      "step": 8300
    },
    {
      "epoch": 0.14589441528116714,
      "grad_norm": 0.0775444840781556,
      "learning_rate": 0.00019995718106177217,
      "loss": 0.8675,
      "step": 8310
    },
    {
      "epoch": 0.14606998016116857,
      "grad_norm": 0.08668953506115076,
      "learning_rate": 0.00019995685272073782,
      "loss": 0.859,
      "step": 8320
    },
    {
      "epoch": 0.14624554504116996,
      "grad_norm": 0.06396927920741455,
      "learning_rate": 0.00019995652312590115,
      "loss": 0.8659,
      "step": 8330
    },
    {
      "epoch": 0.14642110992117136,
      "grad_norm": 0.1030289024513569,
      "learning_rate": 0.0001999561922772664,
      "loss": 0.8741,
      "step": 8340
    },
    {
      "epoch": 0.14659667480117278,
      "grad_norm": 0.05465013085212943,
      "learning_rate": 0.00019995586017483767,
      "loss": 0.8564,
      "step": 8350
    },
    {
      "epoch": 0.14677223968117417,
      "grad_norm": 0.06074197273688347,
      "learning_rate": 0.00019995552681861925,
      "loss": 0.8682,
      "step": 8360
    },
    {
      "epoch": 0.1469478045611756,
      "grad_norm": 0.07312150928547612,
      "learning_rate": 0.00019995519220861535,
      "loss": 0.8619,
      "step": 8370
    },
    {
      "epoch": 0.147123369441177,
      "grad_norm": 0.07386732434699313,
      "learning_rate": 0.0001999548563448302,
      "loss": 0.8686,
      "step": 8380
    },
    {
      "epoch": 0.14729893432117838,
      "grad_norm": 0.07361871239458914,
      "learning_rate": 0.00019995451922726808,
      "loss": 0.8562,
      "step": 8390
    },
    {
      "epoch": 0.1474744992011798,
      "grad_norm": 0.06424309210528296,
      "learning_rate": 0.0001999541808559332,
      "loss": 0.8624,
      "step": 8400
    },
    {
      "epoch": 0.1476500640811812,
      "grad_norm": 0.0870024512994654,
      "learning_rate": 0.00019995384123082987,
      "loss": 0.8653,
      "step": 8410
    },
    {
      "epoch": 0.1478256289611826,
      "grad_norm": 0.09087337621831493,
      "learning_rate": 0.00019995350035196245,
      "loss": 0.8713,
      "step": 8420
    },
    {
      "epoch": 0.14800119384118401,
      "grad_norm": 0.07883004117975567,
      "learning_rate": 0.0001999531582193352,
      "loss": 0.8727,
      "step": 8430
    },
    {
      "epoch": 0.1481767587211854,
      "grad_norm": 0.09168322270945552,
      "learning_rate": 0.00019995281483295248,
      "loss": 0.8593,
      "step": 8440
    },
    {
      "epoch": 0.14835232360118683,
      "grad_norm": 0.0732946541279441,
      "learning_rate": 0.0001999524701928186,
      "loss": 0.862,
      "step": 8450
    },
    {
      "epoch": 0.14852788848118822,
      "grad_norm": 0.0550971982818378,
      "learning_rate": 0.000199952124298938,
      "loss": 0.8595,
      "step": 8460
    },
    {
      "epoch": 0.14870345336118962,
      "grad_norm": 0.06850952990859187,
      "learning_rate": 0.00019995177715131498,
      "loss": 0.8646,
      "step": 8470
    },
    {
      "epoch": 0.14887901824119104,
      "grad_norm": 0.07494472979392426,
      "learning_rate": 0.000199951428749954,
      "loss": 0.8605,
      "step": 8480
    },
    {
      "epoch": 0.14905458312119244,
      "grad_norm": 0.07434027217336076,
      "learning_rate": 0.00019995107909485948,
      "loss": 0.8644,
      "step": 8490
    },
    {
      "epoch": 0.14923014800119383,
      "grad_norm": 0.06681746802354996,
      "learning_rate": 0.00019995072818603578,
      "loss": 0.8643,
      "step": 8500
    },
    {
      "epoch": 0.14940571288119525,
      "grad_norm": 0.06917371879590063,
      "learning_rate": 0.00019995037602348742,
      "loss": 0.8642,
      "step": 8510
    },
    {
      "epoch": 0.14958127776119665,
      "grad_norm": 0.06394503379057809,
      "learning_rate": 0.0001999500226072188,
      "loss": 0.8573,
      "step": 8520
    },
    {
      "epoch": 0.14975684264119807,
      "grad_norm": 0.11756490259832941,
      "learning_rate": 0.00019994966793723446,
      "loss": 0.8662,
      "step": 8530
    },
    {
      "epoch": 0.14993240752119946,
      "grad_norm": 0.06736807555176572,
      "learning_rate": 0.00019994931201353883,
      "loss": 0.863,
      "step": 8540
    },
    {
      "epoch": 0.15010797240120086,
      "grad_norm": 0.11001681610355439,
      "learning_rate": 0.0001999489548361365,
      "loss": 0.8583,
      "step": 8550
    },
    {
      "epoch": 0.15028353728120228,
      "grad_norm": 0.08136238789073882,
      "learning_rate": 0.00019994859640503195,
      "loss": 0.8613,
      "step": 8560
    },
    {
      "epoch": 0.15045910216120367,
      "grad_norm": 0.0663069571743909,
      "learning_rate": 0.00019994823672022966,
      "loss": 0.8548,
      "step": 8570
    },
    {
      "epoch": 0.15063466704120507,
      "grad_norm": 0.07560663833352216,
      "learning_rate": 0.0001999478757817343,
      "loss": 0.8635,
      "step": 8580
    },
    {
      "epoch": 0.1508102319212065,
      "grad_norm": 0.06629747536370145,
      "learning_rate": 0.00019994751358955038,
      "loss": 0.8564,
      "step": 8590
    },
    {
      "epoch": 0.15098579680120788,
      "grad_norm": 0.061888001302653865,
      "learning_rate": 0.0001999471501436825,
      "loss": 0.8674,
      "step": 8600
    },
    {
      "epoch": 0.15116136168120928,
      "grad_norm": 0.09693073004753053,
      "learning_rate": 0.00019994678544413524,
      "loss": 0.8661,
      "step": 8610
    },
    {
      "epoch": 0.1513369265612107,
      "grad_norm": 0.06495360513609198,
      "learning_rate": 0.00019994641949091325,
      "loss": 0.8679,
      "step": 8620
    },
    {
      "epoch": 0.1515124914412121,
      "grad_norm": 0.08274464856364237,
      "learning_rate": 0.0001999460522840212,
      "loss": 0.8548,
      "step": 8630
    },
    {
      "epoch": 0.15168805632121352,
      "grad_norm": 0.08473108045655109,
      "learning_rate": 0.00019994568382346366,
      "loss": 0.8665,
      "step": 8640
    },
    {
      "epoch": 0.1518636212012149,
      "grad_norm": 0.05413280797841906,
      "learning_rate": 0.00019994531410924538,
      "loss": 0.8677,
      "step": 8650
    },
    {
      "epoch": 0.1520391860812163,
      "grad_norm": 0.09989397755513212,
      "learning_rate": 0.00019994494314137099,
      "loss": 0.8703,
      "step": 8660
    },
    {
      "epoch": 0.15221475096121773,
      "grad_norm": 0.06426253484216714,
      "learning_rate": 0.0001999445709198452,
      "loss": 0.8681,
      "step": 8670
    },
    {
      "epoch": 0.15239031584121912,
      "grad_norm": 0.08099422659628848,
      "learning_rate": 0.00019994419744467275,
      "loss": 0.8562,
      "step": 8680
    },
    {
      "epoch": 0.15256588072122051,
      "grad_norm": 0.08350950293638333,
      "learning_rate": 0.00019994382271585834,
      "loss": 0.8669,
      "step": 8690
    },
    {
      "epoch": 0.15274144560122194,
      "grad_norm": 0.08598099726378816,
      "learning_rate": 0.00019994344673340673,
      "loss": 0.859,
      "step": 8700
    },
    {
      "epoch": 0.15291701048122333,
      "grad_norm": 0.09080714439127639,
      "learning_rate": 0.0001999430694973227,
      "loss": 0.857,
      "step": 8710
    },
    {
      "epoch": 0.15309257536122475,
      "grad_norm": 0.09750691173403307,
      "learning_rate": 0.00019994269100761104,
      "loss": 0.8621,
      "step": 8720
    },
    {
      "epoch": 0.15326814024122615,
      "grad_norm": 0.09314856329411601,
      "learning_rate": 0.00019994231126427647,
      "loss": 0.8628,
      "step": 8730
    },
    {
      "epoch": 0.15344370512122754,
      "grad_norm": 0.0717703883288334,
      "learning_rate": 0.0001999419302673239,
      "loss": 0.8663,
      "step": 8740
    },
    {
      "epoch": 0.15361927000122896,
      "grad_norm": 0.07606708593332649,
      "learning_rate": 0.0001999415480167581,
      "loss": 0.8658,
      "step": 8750
    },
    {
      "epoch": 0.15379483488123036,
      "grad_norm": 0.0736511239698316,
      "learning_rate": 0.00019994116451258392,
      "loss": 0.8617,
      "step": 8760
    },
    {
      "epoch": 0.15397039976123175,
      "grad_norm": 0.06877581202991036,
      "learning_rate": 0.00019994077975480623,
      "loss": 0.8655,
      "step": 8770
    },
    {
      "epoch": 0.15414596464123317,
      "grad_norm": 0.08320934391222186,
      "learning_rate": 0.00019994039374342988,
      "loss": 0.8679,
      "step": 8780
    },
    {
      "epoch": 0.15432152952123457,
      "grad_norm": 0.09036133764983303,
      "learning_rate": 0.0001999400064784598,
      "loss": 0.8589,
      "step": 8790
    },
    {
      "epoch": 0.154497094401236,
      "grad_norm": 0.07083316708295039,
      "learning_rate": 0.00019993961795990086,
      "loss": 0.8633,
      "step": 8800
    },
    {
      "epoch": 0.15467265928123738,
      "grad_norm": 0.08465051474396952,
      "learning_rate": 0.00019993922818775802,
      "loss": 0.868,
      "step": 8810
    },
    {
      "epoch": 0.15484822416123878,
      "grad_norm": 0.08756670025858677,
      "learning_rate": 0.00019993883716203615,
      "loss": 0.8688,
      "step": 8820
    },
    {
      "epoch": 0.1550237890412402,
      "grad_norm": 0.05243013108793366,
      "learning_rate": 0.00019993844488274033,
      "loss": 0.862,
      "step": 8830
    },
    {
      "epoch": 0.1551993539212416,
      "grad_norm": 0.0725426233713026,
      "learning_rate": 0.00019993805134987537,
      "loss": 0.8609,
      "step": 8840
    },
    {
      "epoch": 0.155374918801243,
      "grad_norm": 0.07174115797642591,
      "learning_rate": 0.00019993765656344637,
      "loss": 0.865,
      "step": 8850
    },
    {
      "epoch": 0.1555504836812444,
      "grad_norm": 0.07427652010108483,
      "learning_rate": 0.00019993726052345832,
      "loss": 0.8645,
      "step": 8860
    },
    {
      "epoch": 0.1557260485612458,
      "grad_norm": 0.0661783330510821,
      "learning_rate": 0.00019993686322991618,
      "loss": 0.86,
      "step": 8870
    },
    {
      "epoch": 0.1559016134412472,
      "grad_norm": 0.06873827381797169,
      "learning_rate": 0.00019993646468282505,
      "loss": 0.8723,
      "step": 8880
    },
    {
      "epoch": 0.15607717832124862,
      "grad_norm": 0.08253006624171969,
      "learning_rate": 0.00019993606488218996,
      "loss": 0.8651,
      "step": 8890
    },
    {
      "epoch": 0.15625274320125002,
      "grad_norm": 0.06705395453343976,
      "learning_rate": 0.00019993566382801593,
      "loss": 0.8673,
      "step": 8900
    },
    {
      "epoch": 0.15642830808125144,
      "grad_norm": 0.0717721721666629,
      "learning_rate": 0.00019993526152030814,
      "loss": 0.8605,
      "step": 8910
    },
    {
      "epoch": 0.15660387296125283,
      "grad_norm": 0.09128937045023384,
      "learning_rate": 0.00019993485795907157,
      "loss": 0.8603,
      "step": 8920
    },
    {
      "epoch": 0.15677943784125423,
      "grad_norm": 0.06689444128726653,
      "learning_rate": 0.00019993445314431138,
      "loss": 0.86,
      "step": 8930
    },
    {
      "epoch": 0.15695500272125565,
      "grad_norm": 0.10546463248930352,
      "learning_rate": 0.00019993404707603274,
      "loss": 0.8608,
      "step": 8940
    },
    {
      "epoch": 0.15713056760125704,
      "grad_norm": 0.10985246939155664,
      "learning_rate": 0.00019993363975424078,
      "loss": 0.8559,
      "step": 8950
    },
    {
      "epoch": 0.15730613248125844,
      "grad_norm": 0.06452299563979243,
      "learning_rate": 0.0001999332311789406,
      "loss": 0.8574,
      "step": 8960
    },
    {
      "epoch": 0.15748169736125986,
      "grad_norm": 0.08962700170012455,
      "learning_rate": 0.00019993282135013745,
      "loss": 0.8516,
      "step": 8970
    },
    {
      "epoch": 0.15765726224126125,
      "grad_norm": 0.0640773885341476,
      "learning_rate": 0.0001999324102678365,
      "loss": 0.8554,
      "step": 8980
    },
    {
      "epoch": 0.15783282712126268,
      "grad_norm": 0.06504273381677973,
      "learning_rate": 0.00019993199793204294,
      "loss": 0.8572,
      "step": 8990
    },
    {
      "epoch": 0.15800839200126407,
      "grad_norm": 0.06773147516132705,
      "learning_rate": 0.00019993158434276198,
      "loss": 0.8636,
      "step": 9000
    },
    {
      "epoch": 0.15818395688126546,
      "grad_norm": 0.08432560892962908,
      "learning_rate": 0.00019993116949999892,
      "loss": 0.8695,
      "step": 9010
    },
    {
      "epoch": 0.15835952176126689,
      "grad_norm": 0.08843251950234295,
      "learning_rate": 0.00019993075340375894,
      "loss": 0.8667,
      "step": 9020
    },
    {
      "epoch": 0.15853508664126828,
      "grad_norm": 0.06882795747876028,
      "learning_rate": 0.00019993033605404737,
      "loss": 0.8695,
      "step": 9030
    },
    {
      "epoch": 0.15871065152126967,
      "grad_norm": 0.07453910741175059,
      "learning_rate": 0.00019992991745086954,
      "loss": 0.8645,
      "step": 9040
    },
    {
      "epoch": 0.1588862164012711,
      "grad_norm": 0.09526012524279467,
      "learning_rate": 0.00019992949759423066,
      "loss": 0.8593,
      "step": 9050
    },
    {
      "epoch": 0.1590617812812725,
      "grad_norm": 0.059080892730582174,
      "learning_rate": 0.00019992907648413603,
      "loss": 0.8635,
      "step": 9060
    },
    {
      "epoch": 0.15923734616127388,
      "grad_norm": 0.06289406373253201,
      "learning_rate": 0.00019992865412059112,
      "loss": 0.8654,
      "step": 9070
    },
    {
      "epoch": 0.1594129110412753,
      "grad_norm": 0.06531544699727798,
      "learning_rate": 0.00019992823050360117,
      "loss": 0.8632,
      "step": 9080
    },
    {
      "epoch": 0.1595884759212767,
      "grad_norm": 0.06129311560827095,
      "learning_rate": 0.00019992780563317158,
      "loss": 0.8595,
      "step": 9090
    },
    {
      "epoch": 0.15976404080127812,
      "grad_norm": 0.06118040644326025,
      "learning_rate": 0.00019992737950930774,
      "loss": 0.8611,
      "step": 9100
    },
    {
      "epoch": 0.15993960568127952,
      "grad_norm": 0.07307102016211205,
      "learning_rate": 0.00019992695213201503,
      "loss": 0.8735,
      "step": 9110
    },
    {
      "epoch": 0.1601151705612809,
      "grad_norm": 0.07474999947349535,
      "learning_rate": 0.0001999265235012989,
      "loss": 0.8592,
      "step": 9120
    },
    {
      "epoch": 0.16029073544128233,
      "grad_norm": 0.07473614119369282,
      "learning_rate": 0.00019992609361716473,
      "loss": 0.8637,
      "step": 9130
    },
    {
      "epoch": 0.16046630032128373,
      "grad_norm": 0.0694456417698654,
      "learning_rate": 0.000199925662479618,
      "loss": 0.8621,
      "step": 9140
    },
    {
      "epoch": 0.16064186520128512,
      "grad_norm": 0.07357894530195655,
      "learning_rate": 0.0001999252300886642,
      "loss": 0.8648,
      "step": 9150
    },
    {
      "epoch": 0.16081743008128654,
      "grad_norm": 0.0974038344717866,
      "learning_rate": 0.00019992479644430873,
      "loss": 0.8685,
      "step": 9160
    },
    {
      "epoch": 0.16099299496128794,
      "grad_norm": 0.07166206789135984,
      "learning_rate": 0.00019992436154655713,
      "loss": 0.8611,
      "step": 9170
    },
    {
      "epoch": 0.16116855984128936,
      "grad_norm": 0.06541982085557667,
      "learning_rate": 0.00019992392539541496,
      "loss": 0.8604,
      "step": 9180
    },
    {
      "epoch": 0.16134412472129075,
      "grad_norm": 0.09425575396858925,
      "learning_rate": 0.00019992348799088768,
      "loss": 0.8664,
      "step": 9190
    },
    {
      "epoch": 0.16151968960129215,
      "grad_norm": 0.07713841512934352,
      "learning_rate": 0.0001999230493329808,
      "loss": 0.8594,
      "step": 9200
    },
    {
      "epoch": 0.16169525448129357,
      "grad_norm": 0.08892005262494483,
      "learning_rate": 0.00019992260942169998,
      "loss": 0.8615,
      "step": 9210
    },
    {
      "epoch": 0.16187081936129497,
      "grad_norm": 0.08648344203153639,
      "learning_rate": 0.0001999221682570507,
      "loss": 0.8659,
      "step": 9220
    },
    {
      "epoch": 0.16204638424129636,
      "grad_norm": 0.07508612605281205,
      "learning_rate": 0.00019992172583903863,
      "loss": 0.8588,
      "step": 9230
    },
    {
      "epoch": 0.16222194912129778,
      "grad_norm": 0.08515763120203633,
      "learning_rate": 0.00019992128216766932,
      "loss": 0.8728,
      "step": 9240
    },
    {
      "epoch": 0.16239751400129918,
      "grad_norm": 0.07636141117605677,
      "learning_rate": 0.00019992083724294844,
      "loss": 0.8594,
      "step": 9250
    },
    {
      "epoch": 0.1625730788813006,
      "grad_norm": 0.07370722395278455,
      "learning_rate": 0.00019992039106488154,
      "loss": 0.8577,
      "step": 9260
    },
    {
      "epoch": 0.162748643761302,
      "grad_norm": 0.06145866029898632,
      "learning_rate": 0.00019991994363347432,
      "loss": 0.8626,
      "step": 9270
    },
    {
      "epoch": 0.1629242086413034,
      "grad_norm": 0.11141764531674557,
      "learning_rate": 0.0001999194949487325,
      "loss": 0.8624,
      "step": 9280
    },
    {
      "epoch": 0.1630997735213048,
      "grad_norm": 0.0584717758167465,
      "learning_rate": 0.00019991904501066168,
      "loss": 0.8691,
      "step": 9290
    },
    {
      "epoch": 0.1632753384013062,
      "grad_norm": 0.07412945180177175,
      "learning_rate": 0.00019991859381926765,
      "loss": 0.8715,
      "step": 9300
    },
    {
      "epoch": 0.1634509032813076,
      "grad_norm": 0.08891954981625651,
      "learning_rate": 0.00019991814137455604,
      "loss": 0.8645,
      "step": 9310
    },
    {
      "epoch": 0.16362646816130902,
      "grad_norm": 0.07560406663715527,
      "learning_rate": 0.00019991768767653264,
      "loss": 0.8587,
      "step": 9320
    },
    {
      "epoch": 0.1638020330413104,
      "grad_norm": 0.06974263696337847,
      "learning_rate": 0.0001999172327252032,
      "loss": 0.8718,
      "step": 9330
    },
    {
      "epoch": 0.1639775979213118,
      "grad_norm": 0.0930068009059249,
      "learning_rate": 0.00019991677652057344,
      "loss": 0.8676,
      "step": 9340
    },
    {
      "epoch": 0.16415316280131323,
      "grad_norm": 0.07334166879139871,
      "learning_rate": 0.00019991631906264917,
      "loss": 0.861,
      "step": 9350
    },
    {
      "epoch": 0.16432872768131462,
      "grad_norm": 0.08040603429290669,
      "learning_rate": 0.00019991586035143614,
      "loss": 0.8639,
      "step": 9360
    },
    {
      "epoch": 0.16450429256131605,
      "grad_norm": 0.08785954461313158,
      "learning_rate": 0.00019991540038694026,
      "loss": 0.8661,
      "step": 9370
    },
    {
      "epoch": 0.16467985744131744,
      "grad_norm": 0.07297271655161192,
      "learning_rate": 0.00019991493916916725,
      "loss": 0.8619,
      "step": 9380
    },
    {
      "epoch": 0.16485542232131883,
      "grad_norm": 0.0872625162816735,
      "learning_rate": 0.00019991447669812302,
      "loss": 0.8611,
      "step": 9390
    },
    {
      "epoch": 0.16503098720132026,
      "grad_norm": 0.058527402906654924,
      "learning_rate": 0.0001999140129738134,
      "loss": 0.854,
      "step": 9400
    },
    {
      "epoch": 0.16520655208132165,
      "grad_norm": 0.07645091298869681,
      "learning_rate": 0.00019991354799624435,
      "loss": 0.8596,
      "step": 9410
    },
    {
      "epoch": 0.16538211696132304,
      "grad_norm": 0.07515946241328171,
      "learning_rate": 0.00019991308176542166,
      "loss": 0.8648,
      "step": 9420
    },
    {
      "epoch": 0.16555768184132447,
      "grad_norm": 0.06370908578843607,
      "learning_rate": 0.00019991261428135122,
      "loss": 0.8592,
      "step": 9430
    },
    {
      "epoch": 0.16573324672132586,
      "grad_norm": 0.06769439804592255,
      "learning_rate": 0.00019991214554403904,
      "loss": 0.8591,
      "step": 9440
    },
    {
      "epoch": 0.16590881160132728,
      "grad_norm": 0.07474513832820746,
      "learning_rate": 0.000199911675553491,
      "loss": 0.8647,
      "step": 9450
    },
    {
      "epoch": 0.16608437648132868,
      "grad_norm": 0.08605403399755807,
      "learning_rate": 0.00019991120430971307,
      "loss": 0.8621,
      "step": 9460
    },
    {
      "epoch": 0.16625994136133007,
      "grad_norm": 0.09090060972933628,
      "learning_rate": 0.00019991073181271126,
      "loss": 0.8589,
      "step": 9470
    },
    {
      "epoch": 0.1664355062413315,
      "grad_norm": 0.06711289597312475,
      "learning_rate": 0.00019991025806249152,
      "loss": 0.8638,
      "step": 9480
    },
    {
      "epoch": 0.1666110711213329,
      "grad_norm": 0.08418296573504533,
      "learning_rate": 0.0001999097830590598,
      "loss": 0.8639,
      "step": 9490
    },
    {
      "epoch": 0.16678663600133428,
      "grad_norm": 0.09523003911502409,
      "learning_rate": 0.0001999093068024222,
      "loss": 0.8553,
      "step": 9500
    },
    {
      "epoch": 0.1669622008813357,
      "grad_norm": 0.0903098734399247,
      "learning_rate": 0.00019990882929258477,
      "loss": 0.857,
      "step": 9510
    },
    {
      "epoch": 0.1671377657613371,
      "grad_norm": 0.08511634410237613,
      "learning_rate": 0.0001999083505295535,
      "loss": 0.8599,
      "step": 9520
    },
    {
      "epoch": 0.16731333064133852,
      "grad_norm": 0.059944215657867214,
      "learning_rate": 0.0001999078705133344,
      "loss": 0.871,
      "step": 9530
    },
    {
      "epoch": 0.16748889552133991,
      "grad_norm": 0.09795039278638061,
      "learning_rate": 0.0001999073892439337,
      "loss": 0.8641,
      "step": 9540
    },
    {
      "epoch": 0.1676644604013413,
      "grad_norm": 0.08748207627566633,
      "learning_rate": 0.00019990690672135741,
      "loss": 0.8662,
      "step": 9550
    },
    {
      "epoch": 0.16784002528134273,
      "grad_norm": 0.08585452623106965,
      "learning_rate": 0.00019990642294561166,
      "loss": 0.8586,
      "step": 9560
    },
    {
      "epoch": 0.16801559016134412,
      "grad_norm": 0.06310994016824001,
      "learning_rate": 0.00019990593791670253,
      "loss": 0.8657,
      "step": 9570
    },
    {
      "epoch": 0.16819115504134552,
      "grad_norm": 0.07297600059616083,
      "learning_rate": 0.00019990545163463627,
      "loss": 0.8592,
      "step": 9580
    },
    {
      "epoch": 0.16836671992134694,
      "grad_norm": 0.08719214700742134,
      "learning_rate": 0.00019990496409941897,
      "loss": 0.8635,
      "step": 9590
    },
    {
      "epoch": 0.16854228480134834,
      "grad_norm": 0.07656986990784974,
      "learning_rate": 0.0001999044753110568,
      "loss": 0.8645,
      "step": 9600
    },
    {
      "epoch": 0.16871784968134973,
      "grad_norm": 0.055809223301214635,
      "learning_rate": 0.000199903985269556,
      "loss": 0.8581,
      "step": 9610
    },
    {
      "epoch": 0.16889341456135115,
      "grad_norm": 0.07736214787856818,
      "learning_rate": 0.0001999034939749227,
      "loss": 0.876,
      "step": 9620
    },
    {
      "epoch": 0.16906897944135255,
      "grad_norm": 0.07789077802051922,
      "learning_rate": 0.00019990300142716324,
      "loss": 0.8598,
      "step": 9630
    },
    {
      "epoch": 0.16924454432135397,
      "grad_norm": 0.05940681679442575,
      "learning_rate": 0.00019990250762628377,
      "loss": 0.8636,
      "step": 9640
    },
    {
      "epoch": 0.16942010920135536,
      "grad_norm": 0.08157368668557072,
      "learning_rate": 0.00019990201257229055,
      "loss": 0.8625,
      "step": 9650
    },
    {
      "epoch": 0.16959567408135676,
      "grad_norm": 0.06871332137266127,
      "learning_rate": 0.00019990151626518988,
      "loss": 0.8591,
      "step": 9660
    },
    {
      "epoch": 0.16977123896135818,
      "grad_norm": 0.09212902300392771,
      "learning_rate": 0.00019990101870498805,
      "loss": 0.8602,
      "step": 9670
    },
    {
      "epoch": 0.16994680384135957,
      "grad_norm": 0.07925498092327785,
      "learning_rate": 0.00019990051989169142,
      "loss": 0.855,
      "step": 9680
    },
    {
      "epoch": 0.17012236872136097,
      "grad_norm": 0.061002231256336664,
      "learning_rate": 0.00019990001982530617,
      "loss": 0.8723,
      "step": 9690
    },
    {
      "epoch": 0.1702979336013624,
      "grad_norm": 0.08544839049453211,
      "learning_rate": 0.00019989951850583873,
      "loss": 0.8612,
      "step": 9700
    },
    {
      "epoch": 0.17047349848136378,
      "grad_norm": 0.10128284114301066,
      "learning_rate": 0.00019989901593329547,
      "loss": 0.8617,
      "step": 9710
    },
    {
      "epoch": 0.1706490633613652,
      "grad_norm": 0.06371088723549555,
      "learning_rate": 0.00019989851210768268,
      "loss": 0.8614,
      "step": 9720
    },
    {
      "epoch": 0.1708246282413666,
      "grad_norm": 0.07405317542913095,
      "learning_rate": 0.00019989800702900683,
      "loss": 0.8643,
      "step": 9730
    },
    {
      "epoch": 0.171000193121368,
      "grad_norm": 0.06520291534634826,
      "learning_rate": 0.00019989750069727426,
      "loss": 0.8616,
      "step": 9740
    },
    {
      "epoch": 0.17117575800136942,
      "grad_norm": 0.07684439854613317,
      "learning_rate": 0.0001998969931124914,
      "loss": 0.8551,
      "step": 9750
    },
    {
      "epoch": 0.1713513228813708,
      "grad_norm": 0.0658949070761131,
      "learning_rate": 0.0001998964842746647,
      "loss": 0.8677,
      "step": 9760
    },
    {
      "epoch": 0.1715268877613722,
      "grad_norm": 0.07815243047259925,
      "learning_rate": 0.00019989597418380056,
      "loss": 0.8619,
      "step": 9770
    },
    {
      "epoch": 0.17170245264137363,
      "grad_norm": 0.08974931153381292,
      "learning_rate": 0.0001998954628399055,
      "loss": 0.8682,
      "step": 9780
    },
    {
      "epoch": 0.17187801752137502,
      "grad_norm": 0.06341055623095249,
      "learning_rate": 0.00019989495024298594,
      "loss": 0.8573,
      "step": 9790
    },
    {
      "epoch": 0.17205358240137641,
      "grad_norm": 0.07561132162507479,
      "learning_rate": 0.00019989443639304845,
      "loss": 0.8578,
      "step": 9800
    },
    {
      "epoch": 0.17222914728137784,
      "grad_norm": 0.07141542897370559,
      "learning_rate": 0.0001998939212900995,
      "loss": 0.8614,
      "step": 9810
    },
    {
      "epoch": 0.17240471216137923,
      "grad_norm": 0.06336151179522205,
      "learning_rate": 0.0001998934049341456,
      "loss": 0.868,
      "step": 9820
    },
    {
      "epoch": 0.17258027704138065,
      "grad_norm": 0.06267936064003009,
      "learning_rate": 0.0001998928873251933,
      "loss": 0.8639,
      "step": 9830
    },
    {
      "epoch": 0.17275584192138205,
      "grad_norm": 0.08223352671706154,
      "learning_rate": 0.00019989236846324916,
      "loss": 0.8519,
      "step": 9840
    },
    {
      "epoch": 0.17293140680138344,
      "grad_norm": 0.08007809677054974,
      "learning_rate": 0.00019989184834831974,
      "loss": 0.8639,
      "step": 9850
    },
    {
      "epoch": 0.17310697168138486,
      "grad_norm": 0.08067924754928234,
      "learning_rate": 0.00019989132698041166,
      "loss": 0.8632,
      "step": 9860
    },
    {
      "epoch": 0.17328253656138626,
      "grad_norm": 0.0872832718969682,
      "learning_rate": 0.00019989080435953155,
      "loss": 0.8626,
      "step": 9870
    },
    {
      "epoch": 0.17345810144138765,
      "grad_norm": 0.09930029191913392,
      "learning_rate": 0.00019989028048568596,
      "loss": 0.8635,
      "step": 9880
    },
    {
      "epoch": 0.17363366632138907,
      "grad_norm": 0.06457660700255355,
      "learning_rate": 0.0001998897553588816,
      "loss": 0.8678,
      "step": 9890
    },
    {
      "epoch": 0.17380923120139047,
      "grad_norm": 0.07163690067302121,
      "learning_rate": 0.00019988922897912502,
      "loss": 0.8584,
      "step": 9900
    },
    {
      "epoch": 0.1739847960813919,
      "grad_norm": 0.07333869228771073,
      "learning_rate": 0.00019988870134642303,
      "loss": 0.8584,
      "step": 9910
    },
    {
      "epoch": 0.17416036096139328,
      "grad_norm": 0.09920749126245873,
      "learning_rate": 0.0001998881724607822,
      "loss": 0.852,
      "step": 9920
    },
    {
      "epoch": 0.17433592584139468,
      "grad_norm": 0.0644264652166726,
      "learning_rate": 0.0001998876423222093,
      "loss": 0.872,
      "step": 9930
    },
    {
      "epoch": 0.1745114907213961,
      "grad_norm": 0.07553336032351063,
      "learning_rate": 0.000199887110930711,
      "loss": 0.8566,
      "step": 9940
    },
    {
      "epoch": 0.1746870556013975,
      "grad_norm": 0.08527588400704994,
      "learning_rate": 0.00019988657828629407,
      "loss": 0.8575,
      "step": 9950
    },
    {
      "epoch": 0.1748626204813989,
      "grad_norm": 0.08141966025606887,
      "learning_rate": 0.00019988604438896524,
      "loss": 0.8646,
      "step": 9960
    },
    {
      "epoch": 0.1750381853614003,
      "grad_norm": 0.07133836051483286,
      "learning_rate": 0.00019988550923873125,
      "loss": 0.8675,
      "step": 9970
    },
    {
      "epoch": 0.1752137502414017,
      "grad_norm": 0.09810695717967664,
      "learning_rate": 0.00019988497283559892,
      "loss": 0.8725,
      "step": 9980
    },
    {
      "epoch": 0.17538931512140313,
      "grad_norm": 0.11828264784501429,
      "learning_rate": 0.00019988443517957505,
      "loss": 0.864,
      "step": 9990
    },
    {
      "epoch": 0.17556488000140452,
      "grad_norm": 0.0981174809533362,
      "learning_rate": 0.0001998838962706664,
      "loss": 0.8655,
      "step": 10000
    },
    {
      "epoch": 0.17574044488140592,
      "grad_norm": 0.07510618966943149,
      "learning_rate": 0.00019988335610887986,
      "loss": 0.8572,
      "step": 10010
    },
    {
      "epoch": 0.17591600976140734,
      "grad_norm": 0.0794045636059406,
      "learning_rate": 0.00019988281469422226,
      "loss": 0.8545,
      "step": 10020
    },
    {
      "epoch": 0.17609157464140873,
      "grad_norm": 0.07908719577811062,
      "learning_rate": 0.00019988227202670042,
      "loss": 0.8601,
      "step": 10030
    },
    {
      "epoch": 0.17626713952141013,
      "grad_norm": 0.07650173189796122,
      "learning_rate": 0.00019988172810632125,
      "loss": 0.8686,
      "step": 10040
    },
    {
      "epoch": 0.17644270440141155,
      "grad_norm": 0.09709808068179498,
      "learning_rate": 0.00019988118293309164,
      "loss": 0.8622,
      "step": 10050
    },
    {
      "epoch": 0.17661826928141294,
      "grad_norm": 0.0652213694594962,
      "learning_rate": 0.00019988063650701847,
      "loss": 0.8646,
      "step": 10060
    },
    {
      "epoch": 0.17679383416141434,
      "grad_norm": 0.08625824341243918,
      "learning_rate": 0.00019988008882810873,
      "loss": 0.8657,
      "step": 10070
    },
    {
      "epoch": 0.17696939904141576,
      "grad_norm": 0.06513734209982208,
      "learning_rate": 0.00019987953989636929,
      "loss": 0.865,
      "step": 10080
    },
    {
      "epoch": 0.17714496392141715,
      "grad_norm": 0.06501170654325465,
      "learning_rate": 0.00019987898971180712,
      "loss": 0.8679,
      "step": 10090
    },
    {
      "epoch": 0.17732052880141858,
      "grad_norm": 0.08797938874798843,
      "learning_rate": 0.0001998784382744292,
      "loss": 0.8558,
      "step": 10100
    },
    {
      "epoch": 0.17749609368141997,
      "grad_norm": 0.0810016334857986,
      "learning_rate": 0.00019987788558424255,
      "loss": 0.867,
      "step": 10110
    },
    {
      "epoch": 0.17767165856142136,
      "grad_norm": 0.06992741931488768,
      "learning_rate": 0.0001998773316412541,
      "loss": 0.8624,
      "step": 10120
    },
    {
      "epoch": 0.17784722344142279,
      "grad_norm": 0.07819834816203552,
      "learning_rate": 0.00019987677644547092,
      "loss": 0.8552,
      "step": 10130
    },
    {
      "epoch": 0.17802278832142418,
      "grad_norm": 0.07305566569865295,
      "learning_rate": 0.00019987621999690002,
      "loss": 0.8684,
      "step": 10140
    },
    {
      "epoch": 0.17819835320142557,
      "grad_norm": 0.06063968947363287,
      "learning_rate": 0.00019987566229554847,
      "loss": 0.8566,
      "step": 10150
    },
    {
      "epoch": 0.178373918081427,
      "grad_norm": 0.06765427977699871,
      "learning_rate": 0.00019987510334142336,
      "loss": 0.8683,
      "step": 10160
    },
    {
      "epoch": 0.1785494829614284,
      "grad_norm": 0.09505738070814497,
      "learning_rate": 0.00019987454313453168,
      "loss": 0.8611,
      "step": 10170
    },
    {
      "epoch": 0.1787250478414298,
      "grad_norm": 0.10091325639644706,
      "learning_rate": 0.00019987398167488067,
      "loss": 0.8593,
      "step": 10180
    },
    {
      "epoch": 0.1789006127214312,
      "grad_norm": 0.05876109115479496,
      "learning_rate": 0.0001998734189624773,
      "loss": 0.8661,
      "step": 10190
    },
    {
      "epoch": 0.1790761776014326,
      "grad_norm": 0.07164693642008405,
      "learning_rate": 0.0001998728549973288,
      "loss": 0.8632,
      "step": 10200
    },
    {
      "epoch": 0.17925174248143402,
      "grad_norm": 0.10223541228131937,
      "learning_rate": 0.00019987228977944225,
      "loss": 0.8614,
      "step": 10210
    },
    {
      "epoch": 0.17942730736143542,
      "grad_norm": 0.10673932630499713,
      "learning_rate": 0.00019987172330882487,
      "loss": 0.8652,
      "step": 10220
    },
    {
      "epoch": 0.1796028722414368,
      "grad_norm": 0.08566428889908309,
      "learning_rate": 0.00019987115558548377,
      "loss": 0.8603,
      "step": 10230
    },
    {
      "epoch": 0.17977843712143823,
      "grad_norm": 0.07638623183843066,
      "learning_rate": 0.0001998705866094262,
      "loss": 0.8703,
      "step": 10240
    },
    {
      "epoch": 0.17995400200143963,
      "grad_norm": 0.06143653191409074,
      "learning_rate": 0.00019987001638065936,
      "loss": 0.8636,
      "step": 10250
    },
    {
      "epoch": 0.18012956688144105,
      "grad_norm": 0.09797499449695742,
      "learning_rate": 0.00019986944489919044,
      "loss": 0.8601,
      "step": 10260
    },
    {
      "epoch": 0.18030513176144244,
      "grad_norm": 0.09177709469051426,
      "learning_rate": 0.00019986887216502671,
      "loss": 0.8714,
      "step": 10270
    },
    {
      "epoch": 0.18048069664144384,
      "grad_norm": 0.07439648074158474,
      "learning_rate": 0.00019986829817817542,
      "loss": 0.8658,
      "step": 10280
    },
    {
      "epoch": 0.18065626152144526,
      "grad_norm": 0.08654006082653591,
      "learning_rate": 0.00019986772293864387,
      "loss": 0.8618,
      "step": 10290
    },
    {
      "epoch": 0.18083182640144665,
      "grad_norm": 0.07237361456581927,
      "learning_rate": 0.00019986714644643931,
      "loss": 0.8627,
      "step": 10300
    },
    {
      "epoch": 0.18100739128144805,
      "grad_norm": 0.09993362603667856,
      "learning_rate": 0.00019986656870156906,
      "loss": 0.8651,
      "step": 10310
    },
    {
      "epoch": 0.18118295616144947,
      "grad_norm": 0.07481642834526049,
      "learning_rate": 0.0001998659897040404,
      "loss": 0.8583,
      "step": 10320
    },
    {
      "epoch": 0.18135852104145087,
      "grad_norm": 0.1152709737564011,
      "learning_rate": 0.00019986540945386076,
      "loss": 0.8576,
      "step": 10330
    },
    {
      "epoch": 0.18153408592145226,
      "grad_norm": 0.069318894860848,
      "learning_rate": 0.0001998648279510374,
      "loss": 0.8584,
      "step": 10340
    },
    {
      "epoch": 0.18170965080145368,
      "grad_norm": 0.09216538914341131,
      "learning_rate": 0.00019986424519557775,
      "loss": 0.864,
      "step": 10350
    },
    {
      "epoch": 0.18188521568145508,
      "grad_norm": 0.07286481899983718,
      "learning_rate": 0.00019986366118748914,
      "loss": 0.8645,
      "step": 10360
    },
    {
      "epoch": 0.1820607805614565,
      "grad_norm": 0.0633612849678101,
      "learning_rate": 0.000199863075926779,
      "loss": 0.8594,
      "step": 10370
    },
    {
      "epoch": 0.1822363454414579,
      "grad_norm": 0.0727252022021232,
      "learning_rate": 0.00019986248941345473,
      "loss": 0.8567,
      "step": 10380
    },
    {
      "epoch": 0.1824119103214593,
      "grad_norm": 0.06255341485020453,
      "learning_rate": 0.00019986190164752378,
      "loss": 0.8664,
      "step": 10390
    },
    {
      "epoch": 0.1825874752014607,
      "grad_norm": 0.07910156387375265,
      "learning_rate": 0.0001998613126289936,
      "loss": 0.8636,
      "step": 10400
    },
    {
      "epoch": 0.1827630400814621,
      "grad_norm": 0.06288713765741527,
      "learning_rate": 0.00019986072235787163,
      "loss": 0.8662,
      "step": 10410
    },
    {
      "epoch": 0.1829386049614635,
      "grad_norm": 0.06922476971570464,
      "learning_rate": 0.00019986013083416536,
      "loss": 0.8605,
      "step": 10420
    },
    {
      "epoch": 0.18311416984146492,
      "grad_norm": 0.09526828707415765,
      "learning_rate": 0.00019985953805788228,
      "loss": 0.863,
      "step": 10430
    },
    {
      "epoch": 0.1832897347214663,
      "grad_norm": 0.07063473366972216,
      "learning_rate": 0.00019985894402902994,
      "loss": 0.8691,
      "step": 10440
    },
    {
      "epoch": 0.18346529960146774,
      "grad_norm": 0.08174686913646441,
      "learning_rate": 0.00019985834874761582,
      "loss": 0.8659,
      "step": 10450
    },
    {
      "epoch": 0.18364086448146913,
      "grad_norm": 0.08622751640065462,
      "learning_rate": 0.00019985775221364744,
      "loss": 0.8589,
      "step": 10460
    },
    {
      "epoch": 0.18381642936147052,
      "grad_norm": 0.0783923089370782,
      "learning_rate": 0.00019985715442713244,
      "loss": 0.8604,
      "step": 10470
    },
    {
      "epoch": 0.18399199424147195,
      "grad_norm": 0.11180507524826966,
      "learning_rate": 0.00019985655538807832,
      "loss": 0.8557,
      "step": 10480
    },
    {
      "epoch": 0.18416755912147334,
      "grad_norm": 0.09258147896363879,
      "learning_rate": 0.0001998559550964927,
      "loss": 0.8625,
      "step": 10490
    },
    {
      "epoch": 0.18434312400147473,
      "grad_norm": 0.07355802101648803,
      "learning_rate": 0.00019985535355238317,
      "loss": 0.8665,
      "step": 10500
    },
    {
      "epoch": 0.18451868888147616,
      "grad_norm": 0.059473917770376444,
      "learning_rate": 0.00019985475075575744,
      "loss": 0.8626,
      "step": 10510
    },
    {
      "epoch": 0.18469425376147755,
      "grad_norm": 0.05486967289995833,
      "learning_rate": 0.000199854146706623,
      "loss": 0.8585,
      "step": 10520
    },
    {
      "epoch": 0.18486981864147897,
      "grad_norm": 0.07351159326943973,
      "learning_rate": 0.00019985354140498757,
      "loss": 0.8651,
      "step": 10530
    },
    {
      "epoch": 0.18504538352148037,
      "grad_norm": 0.06073791995183849,
      "learning_rate": 0.00019985293485085887,
      "loss": 0.8592,
      "step": 10540
    },
    {
      "epoch": 0.18522094840148176,
      "grad_norm": 0.06580712461096797,
      "learning_rate": 0.0001998523270442445,
      "loss": 0.8713,
      "step": 10550
    },
    {
      "epoch": 0.18539651328148318,
      "grad_norm": 0.062035742732636096,
      "learning_rate": 0.0001998517179851522,
      "loss": 0.8671,
      "step": 10560
    },
    {
      "epoch": 0.18557207816148458,
      "grad_norm": 0.08481498201839642,
      "learning_rate": 0.00019985110767358974,
      "loss": 0.8699,
      "step": 10570
    },
    {
      "epoch": 0.18574764304148597,
      "grad_norm": 0.1009857302010947,
      "learning_rate": 0.00019985049610956476,
      "loss": 0.8594,
      "step": 10580
    },
    {
      "epoch": 0.1859232079214874,
      "grad_norm": 0.09539985912911572,
      "learning_rate": 0.00019984988329308507,
      "loss": 0.8649,
      "step": 10590
    },
    {
      "epoch": 0.1860987728014888,
      "grad_norm": 0.08590163307990636,
      "learning_rate": 0.00019984926922415842,
      "loss": 0.8637,
      "step": 10600
    },
    {
      "epoch": 0.18627433768149018,
      "grad_norm": 0.08194726497148588,
      "learning_rate": 0.00019984865390279254,
      "loss": 0.8636,
      "step": 10610
    },
    {
      "epoch": 0.1864499025614916,
      "grad_norm": 0.057813856647796716,
      "learning_rate": 0.0001998480373289953,
      "loss": 0.8622,
      "step": 10620
    },
    {
      "epoch": 0.186625467441493,
      "grad_norm": 0.12474027693734238,
      "learning_rate": 0.0001998474195027745,
      "loss": 0.8646,
      "step": 10630
    },
    {
      "epoch": 0.18680103232149442,
      "grad_norm": 0.05767733550029884,
      "learning_rate": 0.00019984680042413794,
      "loss": 0.8722,
      "step": 10640
    },
    {
      "epoch": 0.18697659720149581,
      "grad_norm": 0.06862991104147567,
      "learning_rate": 0.00019984618009309346,
      "loss": 0.8629,
      "step": 10650
    },
    {
      "epoch": 0.1871521620814972,
      "grad_norm": 0.07427064090856203,
      "learning_rate": 0.0001998455585096489,
      "loss": 0.8591,
      "step": 10660
    },
    {
      "epoch": 0.18732772696149863,
      "grad_norm": 0.06860451823827075,
      "learning_rate": 0.00019984493567381224,
      "loss": 0.8636,
      "step": 10670
    },
    {
      "epoch": 0.18750329184150002,
      "grad_norm": 0.06345813381783684,
      "learning_rate": 0.00019984431158559128,
      "loss": 0.8653,
      "step": 10680
    },
    {
      "epoch": 0.18767885672150142,
      "grad_norm": 0.09711219588180658,
      "learning_rate": 0.0001998436862449939,
      "loss": 0.8649,
      "step": 10690
    },
    {
      "epoch": 0.18785442160150284,
      "grad_norm": 0.08837972293904665,
      "learning_rate": 0.00019984305965202817,
      "loss": 0.8652,
      "step": 10700
    },
    {
      "epoch": 0.18802998648150424,
      "grad_norm": 0.08238708874363787,
      "learning_rate": 0.00019984243180670185,
      "loss": 0.859,
      "step": 10710
    },
    {
      "epoch": 0.18820555136150566,
      "grad_norm": 0.060340071533053825,
      "learning_rate": 0.000199841802709023,
      "loss": 0.8622,
      "step": 10720
    },
    {
      "epoch": 0.18838111624150705,
      "grad_norm": 0.10345411122114005,
      "learning_rate": 0.00019984117235899958,
      "loss": 0.863,
      "step": 10730
    },
    {
      "epoch": 0.18855668112150845,
      "grad_norm": 0.06709925173298303,
      "learning_rate": 0.00019984054075663954,
      "loss": 0.8605,
      "step": 10740
    },
    {
      "epoch": 0.18873224600150987,
      "grad_norm": 0.08781933699617814,
      "learning_rate": 0.0001998399079019509,
      "loss": 0.8652,
      "step": 10750
    },
    {
      "epoch": 0.18890781088151126,
      "grad_norm": 0.054989626672634837,
      "learning_rate": 0.0001998392737949417,
      "loss": 0.8578,
      "step": 10760
    },
    {
      "epoch": 0.18908337576151266,
      "grad_norm": 0.09930997781741459,
      "learning_rate": 0.00019983863843561997,
      "loss": 0.8637,
      "step": 10770
    },
    {
      "epoch": 0.18925894064151408,
      "grad_norm": 0.08787036559542731,
      "learning_rate": 0.00019983800182399374,
      "loss": 0.8632,
      "step": 10780
    },
    {
      "epoch": 0.18943450552151547,
      "grad_norm": 0.06013754318342283,
      "learning_rate": 0.00019983736396007108,
      "loss": 0.8543,
      "step": 10790
    },
    {
      "epoch": 0.18961007040151687,
      "grad_norm": 0.07577025932806009,
      "learning_rate": 0.0001998367248438601,
      "loss": 0.869,
      "step": 10800
    },
    {
      "epoch": 0.1897856352815183,
      "grad_norm": 0.0612473124830451,
      "learning_rate": 0.00019983608447536884,
      "loss": 0.8581,
      "step": 10810
    },
    {
      "epoch": 0.18996120016151968,
      "grad_norm": 0.0930281877903287,
      "learning_rate": 0.00019983544285460546,
      "loss": 0.8613,
      "step": 10820
    },
    {
      "epoch": 0.1901367650415211,
      "grad_norm": 0.06382876186842239,
      "learning_rate": 0.00019983479998157805,
      "loss": 0.8593,
      "step": 10830
    },
    {
      "epoch": 0.1903123299215225,
      "grad_norm": 0.06713671265324865,
      "learning_rate": 0.0001998341558562948,
      "loss": 0.8589,
      "step": 10840
    },
    {
      "epoch": 0.1904878948015239,
      "grad_norm": 0.07192818678303677,
      "learning_rate": 0.0001998335104787639,
      "loss": 0.8607,
      "step": 10850
    },
    {
      "epoch": 0.19066345968152532,
      "grad_norm": 0.07335446635119781,
      "learning_rate": 0.00019983286384899344,
      "loss": 0.8652,
      "step": 10860
    },
    {
      "epoch": 0.1908390245615267,
      "grad_norm": 0.08007270518931298,
      "learning_rate": 0.00019983221596699165,
      "loss": 0.8538,
      "step": 10870
    },
    {
      "epoch": 0.1910145894415281,
      "grad_norm": 0.07802218035351148,
      "learning_rate": 0.00019983156683276676,
      "loss": 0.8593,
      "step": 10880
    },
    {
      "epoch": 0.19119015432152953,
      "grad_norm": 0.07005898170223804,
      "learning_rate": 0.000199830916446327,
      "loss": 0.8618,
      "step": 10890
    },
    {
      "epoch": 0.19136571920153092,
      "grad_norm": 0.08864190491014051,
      "learning_rate": 0.00019983026480768052,
      "loss": 0.8652,
      "step": 10900
    },
    {
      "epoch": 0.19154128408153234,
      "grad_norm": 0.07526426022242097,
      "learning_rate": 0.0001998296119168357,
      "loss": 0.8656,
      "step": 10910
    },
    {
      "epoch": 0.19171684896153374,
      "grad_norm": 0.06433782607527952,
      "learning_rate": 0.00019982895777380074,
      "loss": 0.8617,
      "step": 10920
    },
    {
      "epoch": 0.19189241384153513,
      "grad_norm": 0.0634146377423622,
      "learning_rate": 0.0001998283023785839,
      "loss": 0.8595,
      "step": 10930
    },
    {
      "epoch": 0.19206797872153655,
      "grad_norm": 0.08879654579806198,
      "learning_rate": 0.0001998276457311936,
      "loss": 0.862,
      "step": 10940
    },
    {
      "epoch": 0.19224354360153795,
      "grad_norm": 0.08027176823222054,
      "learning_rate": 0.00019982698783163806,
      "loss": 0.8583,
      "step": 10950
    },
    {
      "epoch": 0.19241910848153934,
      "grad_norm": 0.09072746953058013,
      "learning_rate": 0.00019982632867992562,
      "loss": 0.8674,
      "step": 10960
    },
    {
      "epoch": 0.19259467336154076,
      "grad_norm": 0.06439002486022939,
      "learning_rate": 0.00019982566827606469,
      "loss": 0.8637,
      "step": 10970
    },
    {
      "epoch": 0.19277023824154216,
      "grad_norm": 0.10195226438734444,
      "learning_rate": 0.0001998250066200636,
      "loss": 0.8583,
      "step": 10980
    },
    {
      "epoch": 0.19294580312154358,
      "grad_norm": 0.09107915214181189,
      "learning_rate": 0.00019982434371193072,
      "loss": 0.861,
      "step": 10990
    },
    {
      "epoch": 0.19312136800154497,
      "grad_norm": 0.07710625417041224,
      "learning_rate": 0.0001998236795516745,
      "loss": 0.86,
      "step": 11000
    },
    {
      "epoch": 0.19329693288154637,
      "grad_norm": 0.05664192454959979,
      "learning_rate": 0.0001998230141393033,
      "loss": 0.8598,
      "step": 11010
    },
    {
      "epoch": 0.1934724977615478,
      "grad_norm": 0.06328225386522277,
      "learning_rate": 0.00019982234747482552,
      "loss": 0.8667,
      "step": 11020
    },
    {
      "epoch": 0.19364806264154918,
      "grad_norm": 0.08066251156125577,
      "learning_rate": 0.00019982167955824974,
      "loss": 0.8648,
      "step": 11030
    },
    {
      "epoch": 0.19382362752155058,
      "grad_norm": 0.08480385743337594,
      "learning_rate": 0.00019982101038958428,
      "loss": 0.8645,
      "step": 11040
    },
    {
      "epoch": 0.193999192401552,
      "grad_norm": 0.06109001401566652,
      "learning_rate": 0.0001998203399688377,
      "loss": 0.8693,
      "step": 11050
    },
    {
      "epoch": 0.1941747572815534,
      "grad_norm": 0.06516817049012194,
      "learning_rate": 0.0001998196682960185,
      "loss": 0.8638,
      "step": 11060
    },
    {
      "epoch": 0.1943503221615548,
      "grad_norm": 0.07468404178305846,
      "learning_rate": 0.00019981899537113514,
      "loss": 0.8605,
      "step": 11070
    },
    {
      "epoch": 0.1945258870415562,
      "grad_norm": 0.06518877176803846,
      "learning_rate": 0.00019981832119419618,
      "loss": 0.8597,
      "step": 11080
    },
    {
      "epoch": 0.1947014519215576,
      "grad_norm": 0.09716435378391924,
      "learning_rate": 0.00019981764576521013,
      "loss": 0.8594,
      "step": 11090
    },
    {
      "epoch": 0.19487701680155903,
      "grad_norm": 0.07029988431118245,
      "learning_rate": 0.00019981696908418558,
      "loss": 0.859,
      "step": 11100
    },
    {
      "epoch": 0.19505258168156042,
      "grad_norm": 0.07597425786466074,
      "learning_rate": 0.0001998162911511311,
      "loss": 0.8634,
      "step": 11110
    },
    {
      "epoch": 0.19522814656156182,
      "grad_norm": 0.07103170200190047,
      "learning_rate": 0.00019981561196605527,
      "loss": 0.8649,
      "step": 11120
    },
    {
      "epoch": 0.19540371144156324,
      "grad_norm": 0.08409596394265781,
      "learning_rate": 0.0001998149315289667,
      "loss": 0.8614,
      "step": 11130
    },
    {
      "epoch": 0.19557927632156463,
      "grad_norm": 0.06598690166878926,
      "learning_rate": 0.00019981424983987403,
      "loss": 0.8584,
      "step": 11140
    },
    {
      "epoch": 0.19575484120156603,
      "grad_norm": 0.08066397899992475,
      "learning_rate": 0.00019981356689878583,
      "loss": 0.8602,
      "step": 11150
    },
    {
      "epoch": 0.19593040608156745,
      "grad_norm": 0.10522217700109793,
      "learning_rate": 0.00019981288270571083,
      "loss": 0.8638,
      "step": 11160
    },
    {
      "epoch": 0.19610597096156884,
      "grad_norm": 0.08929543182791812,
      "learning_rate": 0.00019981219726065768,
      "loss": 0.8578,
      "step": 11170
    },
    {
      "epoch": 0.19628153584157026,
      "grad_norm": 0.1062191647129973,
      "learning_rate": 0.00019981151056363503,
      "loss": 0.863,
      "step": 11180
    },
    {
      "epoch": 0.19645710072157166,
      "grad_norm": 0.07284430449164271,
      "learning_rate": 0.0001998108226146516,
      "loss": 0.8579,
      "step": 11190
    },
    {
      "epoch": 0.19663266560157305,
      "grad_norm": 0.0699500337169853,
      "learning_rate": 0.00019981013341371614,
      "loss": 0.8567,
      "step": 11200
    },
    {
      "epoch": 0.19680823048157448,
      "grad_norm": 0.11056444393694914,
      "learning_rate": 0.00019980944296083732,
      "loss": 0.8611,
      "step": 11210
    },
    {
      "epoch": 0.19698379536157587,
      "grad_norm": 0.06689444397647458,
      "learning_rate": 0.00019980875125602394,
      "loss": 0.8717,
      "step": 11220
    },
    {
      "epoch": 0.19715936024157726,
      "grad_norm": 0.0757317716098211,
      "learning_rate": 0.00019980805829928472,
      "loss": 0.8622,
      "step": 11230
    },
    {
      "epoch": 0.19733492512157869,
      "grad_norm": 0.11467404944023815,
      "learning_rate": 0.00019980736409062853,
      "loss": 0.8565,
      "step": 11240
    },
    {
      "epoch": 0.19751049000158008,
      "grad_norm": 0.06896553570119207,
      "learning_rate": 0.00019980666863006404,
      "loss": 0.8584,
      "step": 11250
    },
    {
      "epoch": 0.1976860548815815,
      "grad_norm": 0.07411036410909033,
      "learning_rate": 0.00019980597191760011,
      "loss": 0.8538,
      "step": 11260
    },
    {
      "epoch": 0.1978616197615829,
      "grad_norm": 0.05414463987679983,
      "learning_rate": 0.00019980527395324562,
      "loss": 0.8625,
      "step": 11270
    },
    {
      "epoch": 0.1980371846415843,
      "grad_norm": 0.06654404691204416,
      "learning_rate": 0.00019980457473700938,
      "loss": 0.8607,
      "step": 11280
    },
    {
      "epoch": 0.1982127495215857,
      "grad_norm": 0.05940047775418309,
      "learning_rate": 0.00019980387426890022,
      "loss": 0.8604,
      "step": 11290
    },
    {
      "epoch": 0.1983883144015871,
      "grad_norm": 0.07291847057122354,
      "learning_rate": 0.00019980317254892705,
      "loss": 0.8572,
      "step": 11300
    },
    {
      "epoch": 0.1985638792815885,
      "grad_norm": 0.0502920404382447,
      "learning_rate": 0.0001998024695770987,
      "loss": 0.8646,
      "step": 11310
    },
    {
      "epoch": 0.19873944416158992,
      "grad_norm": 0.10904238733366504,
      "learning_rate": 0.00019980176535342418,
      "loss": 0.8555,
      "step": 11320
    },
    {
      "epoch": 0.19891500904159132,
      "grad_norm": 0.06915380819181416,
      "learning_rate": 0.00019980105987791235,
      "loss": 0.8616,
      "step": 11330
    },
    {
      "epoch": 0.1990905739215927,
      "grad_norm": 0.08383934681825642,
      "learning_rate": 0.00019980035315057216,
      "loss": 0.8651,
      "step": 11340
    },
    {
      "epoch": 0.19926613880159413,
      "grad_norm": 0.07747787055189378,
      "learning_rate": 0.00019979964517141254,
      "loss": 0.8533,
      "step": 11350
    },
    {
      "epoch": 0.19944170368159553,
      "grad_norm": 0.11025951747688771,
      "learning_rate": 0.0001997989359404425,
      "loss": 0.8568,
      "step": 11360
    },
    {
      "epoch": 0.19961726856159695,
      "grad_norm": 0.06865460749006481,
      "learning_rate": 0.00019979822545767098,
      "loss": 0.858,
      "step": 11370
    },
    {
      "epoch": 0.19979283344159834,
      "grad_norm": 0.07287576534441402,
      "learning_rate": 0.00019979751372310703,
      "loss": 0.8623,
      "step": 11380
    },
    {
      "epoch": 0.19996839832159974,
      "grad_norm": 0.06905723856559356,
      "learning_rate": 0.00019979680073675966,
      "loss": 0.8626,
      "step": 11390
    },
    {
      "epoch": 0.20014396320160116,
      "grad_norm": 0.06974754431925725,
      "learning_rate": 0.00019979608649863787,
      "loss": 0.8663,
      "step": 11400
    },
    {
      "epoch": 0.20031952808160255,
      "grad_norm": 0.07174432648102101,
      "learning_rate": 0.00019979537100875072,
      "loss": 0.8545,
      "step": 11410
    },
    {
      "epoch": 0.20049509296160395,
      "grad_norm": 0.05877411251113119,
      "learning_rate": 0.0001997946542671073,
      "loss": 0.8635,
      "step": 11420
    },
    {
      "epoch": 0.20067065784160537,
      "grad_norm": 0.06443476471552287,
      "learning_rate": 0.00019979393627371668,
      "loss": 0.8691,
      "step": 11430
    },
    {
      "epoch": 0.20084622272160677,
      "grad_norm": 0.06711153342882789,
      "learning_rate": 0.00019979321702858792,
      "loss": 0.8627,
      "step": 11440
    },
    {
      "epoch": 0.2010217876016082,
      "grad_norm": 0.05604570745637266,
      "learning_rate": 0.00019979249653173022,
      "loss": 0.8558,
      "step": 11450
    },
    {
      "epoch": 0.20119735248160958,
      "grad_norm": 0.07434165518192266,
      "learning_rate": 0.00019979177478315262,
      "loss": 0.8607,
      "step": 11460
    },
    {
      "epoch": 0.20137291736161098,
      "grad_norm": 0.07289731992796257,
      "learning_rate": 0.00019979105178286432,
      "loss": 0.8569,
      "step": 11470
    },
    {
      "epoch": 0.2015484822416124,
      "grad_norm": 0.07410009019828331,
      "learning_rate": 0.00019979032753087445,
      "loss": 0.8661,
      "step": 11480
    },
    {
      "epoch": 0.2017240471216138,
      "grad_norm": 0.08828807351892376,
      "learning_rate": 0.0001997896020271922,
      "loss": 0.8635,
      "step": 11490
    },
    {
      "epoch": 0.2018996120016152,
      "grad_norm": 0.06020463855990833,
      "learning_rate": 0.00019978887527182676,
      "loss": 0.8607,
      "step": 11500
    },
    {
      "epoch": 0.2020751768816166,
      "grad_norm": 0.07332687669787262,
      "learning_rate": 0.0001997881472647873,
      "loss": 0.8598,
      "step": 11510
    },
    {
      "epoch": 0.202250741761618,
      "grad_norm": 0.06753671941631845,
      "learning_rate": 0.00019978741800608313,
      "loss": 0.8505,
      "step": 11520
    },
    {
      "epoch": 0.2024263066416194,
      "grad_norm": 0.08485753273520577,
      "learning_rate": 0.0001997866874957234,
      "loss": 0.8627,
      "step": 11530
    },
    {
      "epoch": 0.20260187152162082,
      "grad_norm": 0.11056422178464878,
      "learning_rate": 0.00019978595573371747,
      "loss": 0.8624,
      "step": 11540
    },
    {
      "epoch": 0.2027774364016222,
      "grad_norm": 0.08050322495035325,
      "learning_rate": 0.00019978522272007452,
      "loss": 0.8675,
      "step": 11550
    },
    {
      "epoch": 0.20295300128162364,
      "grad_norm": 0.08792617465616541,
      "learning_rate": 0.00019978448845480386,
      "loss": 0.8575,
      "step": 11560
    },
    {
      "epoch": 0.20312856616162503,
      "grad_norm": 0.07745796590646534,
      "learning_rate": 0.0001997837529379148,
      "loss": 0.8565,
      "step": 11570
    },
    {
      "epoch": 0.20330413104162642,
      "grad_norm": 0.06831318790288808,
      "learning_rate": 0.00019978301616941667,
      "loss": 0.8637,
      "step": 11580
    },
    {
      "epoch": 0.20347969592162785,
      "grad_norm": 0.10158133947793363,
      "learning_rate": 0.00019978227814931878,
      "loss": 0.8535,
      "step": 11590
    },
    {
      "epoch": 0.20365526080162924,
      "grad_norm": 0.06764835212078524,
      "learning_rate": 0.00019978153887763048,
      "loss": 0.8616,
      "step": 11600
    },
    {
      "epoch": 0.20383082568163063,
      "grad_norm": 0.07481663328383992,
      "learning_rate": 0.00019978079835436118,
      "loss": 0.8646,
      "step": 11610
    },
    {
      "epoch": 0.20400639056163206,
      "grad_norm": 0.06183354158690388,
      "learning_rate": 0.0001997800565795202,
      "loss": 0.8645,
      "step": 11620
    },
    {
      "epoch": 0.20418195544163345,
      "grad_norm": 0.05399871093881159,
      "learning_rate": 0.00019977931355311698,
      "loss": 0.859,
      "step": 11630
    },
    {
      "epoch": 0.20435752032163487,
      "grad_norm": 0.08299789254881017,
      "learning_rate": 0.00019977856927516093,
      "loss": 0.8619,
      "step": 11640
    },
    {
      "epoch": 0.20453308520163627,
      "grad_norm": 0.07270330571629706,
      "learning_rate": 0.0001997778237456615,
      "loss": 0.8637,
      "step": 11650
    },
    {
      "epoch": 0.20470865008163766,
      "grad_norm": 0.06890626820832899,
      "learning_rate": 0.00019977707696462808,
      "loss": 0.8658,
      "step": 11660
    },
    {
      "epoch": 0.20488421496163908,
      "grad_norm": 0.09422152704997061,
      "learning_rate": 0.0001997763289320702,
      "loss": 0.8569,
      "step": 11670
    },
    {
      "epoch": 0.20505977984164048,
      "grad_norm": 0.0717091126915168,
      "learning_rate": 0.0001997755796479973,
      "loss": 0.8598,
      "step": 11680
    },
    {
      "epoch": 0.20523534472164187,
      "grad_norm": 0.06599919251117033,
      "learning_rate": 0.00019977482911241883,
      "loss": 0.8595,
      "step": 11690
    },
    {
      "epoch": 0.2054109096016433,
      "grad_norm": 0.05899318554996764,
      "learning_rate": 0.0001997740773253444,
      "loss": 0.8602,
      "step": 11700
    },
    {
      "epoch": 0.2055864744816447,
      "grad_norm": 0.05902171827836532,
      "learning_rate": 0.00019977332428678345,
      "loss": 0.8633,
      "step": 11710
    },
    {
      "epoch": 0.2057620393616461,
      "grad_norm": 0.0569672865231465,
      "learning_rate": 0.00019977256999674557,
      "loss": 0.8592,
      "step": 11720
    },
    {
      "epoch": 0.2059376042416475,
      "grad_norm": 0.053822986577173826,
      "learning_rate": 0.00019977181445524027,
      "loss": 0.8561,
      "step": 11730
    },
    {
      "epoch": 0.2061131691216489,
      "grad_norm": 0.06508445265780063,
      "learning_rate": 0.00019977105766227714,
      "loss": 0.8616,
      "step": 11740
    },
    {
      "epoch": 0.20628873400165032,
      "grad_norm": 0.06383980581889599,
      "learning_rate": 0.00019977029961786583,
      "loss": 0.865,
      "step": 11750
    },
    {
      "epoch": 0.20646429888165171,
      "grad_norm": 0.06884783399451656,
      "learning_rate": 0.00019976954032201587,
      "loss": 0.857,
      "step": 11760
    },
    {
      "epoch": 0.2066398637616531,
      "grad_norm": 0.06528916759534233,
      "learning_rate": 0.0001997687797747369,
      "loss": 0.8554,
      "step": 11770
    },
    {
      "epoch": 0.20681542864165453,
      "grad_norm": 0.07876543663359256,
      "learning_rate": 0.00019976801797603856,
      "loss": 0.8631,
      "step": 11780
    },
    {
      "epoch": 0.20699099352165592,
      "grad_norm": 0.07281560909354284,
      "learning_rate": 0.0001997672549259305,
      "loss": 0.8466,
      "step": 11790
    },
    {
      "epoch": 0.20716655840165732,
      "grad_norm": 0.055311249116328634,
      "learning_rate": 0.00019976649062442242,
      "loss": 0.8504,
      "step": 11800
    },
    {
      "epoch": 0.20734212328165874,
      "grad_norm": 0.09716586697776519,
      "learning_rate": 0.00019976572507152395,
      "loss": 0.8555,
      "step": 11810
    },
    {
      "epoch": 0.20751768816166014,
      "grad_norm": 0.06113500635489527,
      "learning_rate": 0.00019976495826724483,
      "loss": 0.8647,
      "step": 11820
    },
    {
      "epoch": 0.20769325304166156,
      "grad_norm": 0.07027966528696562,
      "learning_rate": 0.00019976419021159475,
      "loss": 0.8542,
      "step": 11830
    },
    {
      "epoch": 0.20786881792166295,
      "grad_norm": 0.07100812153254037,
      "learning_rate": 0.00019976342090458345,
      "loss": 0.8556,
      "step": 11840
    },
    {
      "epoch": 0.20804438280166435,
      "grad_norm": 0.0772208897526919,
      "learning_rate": 0.00019976265034622067,
      "loss": 0.8631,
      "step": 11850
    },
    {
      "epoch": 0.20821994768166577,
      "grad_norm": 0.07608892142316544,
      "learning_rate": 0.00019976187853651623,
      "loss": 0.8646,
      "step": 11860
    },
    {
      "epoch": 0.20839551256166716,
      "grad_norm": 0.07452072461172539,
      "learning_rate": 0.0001997611054754798,
      "loss": 0.8641,
      "step": 11870
    },
    {
      "epoch": 0.20857107744166856,
      "grad_norm": 0.08598566015782368,
      "learning_rate": 0.00019976033116312126,
      "loss": 0.8574,
      "step": 11880
    },
    {
      "epoch": 0.20874664232166998,
      "grad_norm": 0.07191413271910875,
      "learning_rate": 0.00019975955559945038,
      "loss": 0.8643,
      "step": 11890
    },
    {
      "epoch": 0.20892220720167137,
      "grad_norm": 0.1335582315344332,
      "learning_rate": 0.00019975877878447703,
      "loss": 0.8595,
      "step": 11900
    },
    {
      "epoch": 0.2090977720816728,
      "grad_norm": 0.09884260485481354,
      "learning_rate": 0.00019975800071821105,
      "loss": 0.8642,
      "step": 11910
    },
    {
      "epoch": 0.2092733369616742,
      "grad_norm": 0.0840633742921738,
      "learning_rate": 0.0001997572214006622,
      "loss": 0.8587,
      "step": 11920
    },
    {
      "epoch": 0.20944890184167558,
      "grad_norm": 0.07193175031098976,
      "learning_rate": 0.0001997564408318405,
      "loss": 0.8714,
      "step": 11930
    },
    {
      "epoch": 0.209624466721677,
      "grad_norm": 0.07566148052754874,
      "learning_rate": 0.00019975565901175573,
      "loss": 0.8676,
      "step": 11940
    },
    {
      "epoch": 0.2098000316016784,
      "grad_norm": 0.07856779267741669,
      "learning_rate": 0.00019975487594041784,
      "loss": 0.8666,
      "step": 11950
    },
    {
      "epoch": 0.2099755964816798,
      "grad_norm": 0.06873151821347684,
      "learning_rate": 0.00019975409161783676,
      "loss": 0.8564,
      "step": 11960
    },
    {
      "epoch": 0.21015116136168122,
      "grad_norm": 0.0749725187738329,
      "learning_rate": 0.0001997533060440224,
      "loss": 0.8613,
      "step": 11970
    },
    {
      "epoch": 0.2103267262416826,
      "grad_norm": 0.06714006955178443,
      "learning_rate": 0.00019975251921898474,
      "loss": 0.8574,
      "step": 11980
    },
    {
      "epoch": 0.21050229112168403,
      "grad_norm": 0.07947337013478468,
      "learning_rate": 0.0001997517311427337,
      "loss": 0.8624,
      "step": 11990
    },
    {
      "epoch": 0.21067785600168543,
      "grad_norm": 0.08850951615386116,
      "learning_rate": 0.00019975094181527933,
      "loss": 0.861,
      "step": 12000
    },
    {
      "epoch": 0.21085342088168682,
      "grad_norm": 0.08457137148210023,
      "learning_rate": 0.00019975015123663162,
      "loss": 0.8663,
      "step": 12010
    },
    {
      "epoch": 0.21102898576168824,
      "grad_norm": 0.06436894671536832,
      "learning_rate": 0.00019974935940680053,
      "loss": 0.8672,
      "step": 12020
    },
    {
      "epoch": 0.21120455064168964,
      "grad_norm": 0.06571936647156754,
      "learning_rate": 0.00019974856632579613,
      "loss": 0.8657,
      "step": 12030
    },
    {
      "epoch": 0.21138011552169103,
      "grad_norm": 0.06963931656300218,
      "learning_rate": 0.0001997477719936285,
      "loss": 0.8597,
      "step": 12040
    },
    {
      "epoch": 0.21155568040169245,
      "grad_norm": 0.08080260915798766,
      "learning_rate": 0.00019974697641030766,
      "loss": 0.8608,
      "step": 12050
    },
    {
      "epoch": 0.21173124528169385,
      "grad_norm": 0.10101367578936203,
      "learning_rate": 0.00019974617957584366,
      "loss": 0.86,
      "step": 12060
    },
    {
      "epoch": 0.21190681016169524,
      "grad_norm": 0.08444507697876563,
      "learning_rate": 0.00019974538149024673,
      "loss": 0.8619,
      "step": 12070
    },
    {
      "epoch": 0.21208237504169666,
      "grad_norm": 0.07394267130563466,
      "learning_rate": 0.00019974458215352682,
      "loss": 0.8586,
      "step": 12080
    },
    {
      "epoch": 0.21225793992169806,
      "grad_norm": 0.1029191545630217,
      "learning_rate": 0.00019974378156569416,
      "loss": 0.8596,
      "step": 12090
    },
    {
      "epoch": 0.21243350480169948,
      "grad_norm": 0.08594674102552544,
      "learning_rate": 0.00019974297972675885,
      "loss": 0.8558,
      "step": 12100
    },
    {
      "epoch": 0.21260906968170087,
      "grad_norm": 0.0927348594819347,
      "learning_rate": 0.0001997421766367311,
      "loss": 0.867,
      "step": 12110
    },
    {
      "epoch": 0.21278463456170227,
      "grad_norm": 0.08588534924362012,
      "learning_rate": 0.000199741372295621,
      "loss": 0.8629,
      "step": 12120
    },
    {
      "epoch": 0.2129601994417037,
      "grad_norm": 0.06768947614597055,
      "learning_rate": 0.00019974056670343882,
      "loss": 0.8609,
      "step": 12130
    },
    {
      "epoch": 0.21313576432170508,
      "grad_norm": 0.0680775308835976,
      "learning_rate": 0.00019973975986019472,
      "loss": 0.8595,
      "step": 12140
    },
    {
      "epoch": 0.21331132920170648,
      "grad_norm": 0.06114018108191955,
      "learning_rate": 0.00019973895176589895,
      "loss": 0.8708,
      "step": 12150
    },
    {
      "epoch": 0.2134868940817079,
      "grad_norm": 0.07272345753240134,
      "learning_rate": 0.00019973814242056173,
      "loss": 0.858,
      "step": 12160
    },
    {
      "epoch": 0.2136624589617093,
      "grad_norm": 0.06421246506344783,
      "learning_rate": 0.00019973733182419333,
      "loss": 0.8577,
      "step": 12170
    },
    {
      "epoch": 0.21383802384171072,
      "grad_norm": 0.07966401394686959,
      "learning_rate": 0.000199736519976804,
      "loss": 0.8577,
      "step": 12180
    },
    {
      "epoch": 0.2140135887217121,
      "grad_norm": 0.064671842191974,
      "learning_rate": 0.00019973570687840405,
      "loss": 0.8652,
      "step": 12190
    },
    {
      "epoch": 0.2141891536017135,
      "grad_norm": 0.07796808793021955,
      "learning_rate": 0.00019973489252900377,
      "loss": 0.8635,
      "step": 12200
    },
    {
      "epoch": 0.21436471848171493,
      "grad_norm": 0.05500961954160095,
      "learning_rate": 0.00019973407692861346,
      "loss": 0.8657,
      "step": 12210
    },
    {
      "epoch": 0.21454028336171632,
      "grad_norm": 0.08682530409346874,
      "learning_rate": 0.0001997332600772435,
      "loss": 0.8627,
      "step": 12220
    },
    {
      "epoch": 0.21471584824171772,
      "grad_norm": 0.0707245697063651,
      "learning_rate": 0.0001997324419749042,
      "loss": 0.8571,
      "step": 12230
    },
    {
      "epoch": 0.21489141312171914,
      "grad_norm": 0.09458535843728595,
      "learning_rate": 0.00019973162262160594,
      "loss": 0.8656,
      "step": 12240
    },
    {
      "epoch": 0.21506697800172053,
      "grad_norm": 0.08256330933791954,
      "learning_rate": 0.0001997308020173591,
      "loss": 0.8599,
      "step": 12250
    },
    {
      "epoch": 0.21524254288172193,
      "grad_norm": 0.12645388072969208,
      "learning_rate": 0.00019972998016217408,
      "loss": 0.8644,
      "step": 12260
    },
    {
      "epoch": 0.21541810776172335,
      "grad_norm": 0.06369345063309637,
      "learning_rate": 0.00019972915705606127,
      "loss": 0.8574,
      "step": 12270
    },
    {
      "epoch": 0.21559367264172474,
      "grad_norm": 0.06748337639120253,
      "learning_rate": 0.00019972833269903115,
      "loss": 0.8524,
      "step": 12280
    },
    {
      "epoch": 0.21576923752172616,
      "grad_norm": 0.06381258224809104,
      "learning_rate": 0.00019972750709109413,
      "loss": 0.868,
      "step": 12290
    },
    {
      "epoch": 0.21594480240172756,
      "grad_norm": 0.08841314655687235,
      "learning_rate": 0.00019972668023226063,
      "loss": 0.8642,
      "step": 12300
    },
    {
      "epoch": 0.21612036728172895,
      "grad_norm": 0.06914650960148769,
      "learning_rate": 0.00019972585212254125,
      "loss": 0.8642,
      "step": 12310
    },
    {
      "epoch": 0.21629593216173038,
      "grad_norm": 0.07303558688391464,
      "learning_rate": 0.00019972502276194637,
      "loss": 0.8618,
      "step": 12320
    },
    {
      "epoch": 0.21647149704173177,
      "grad_norm": 0.07985173937184216,
      "learning_rate": 0.0001997241921504865,
      "loss": 0.8656,
      "step": 12330
    },
    {
      "epoch": 0.21664706192173316,
      "grad_norm": 0.09654733341280401,
      "learning_rate": 0.0001997233602881722,
      "loss": 0.864,
      "step": 12340
    },
    {
      "epoch": 0.21682262680173459,
      "grad_norm": 0.06110114775268228,
      "learning_rate": 0.000199722527175014,
      "loss": 0.86,
      "step": 12350
    },
    {
      "epoch": 0.21699819168173598,
      "grad_norm": 0.07355724160720263,
      "learning_rate": 0.0001997216928110225,
      "loss": 0.8624,
      "step": 12360
    },
    {
      "epoch": 0.2171737565617374,
      "grad_norm": 0.0519312159382691,
      "learning_rate": 0.0001997208571962082,
      "loss": 0.8604,
      "step": 12370
    },
    {
      "epoch": 0.2173493214417388,
      "grad_norm": 0.07711396829642897,
      "learning_rate": 0.00019972002033058175,
      "loss": 0.8609,
      "step": 12380
    },
    {
      "epoch": 0.2175248863217402,
      "grad_norm": 0.06496701853996166,
      "learning_rate": 0.00019971918221415368,
      "loss": 0.8612,
      "step": 12390
    },
    {
      "epoch": 0.2177004512017416,
      "grad_norm": 0.07147394500416392,
      "learning_rate": 0.0001997183428469347,
      "loss": 0.86,
      "step": 12400
    },
    {
      "epoch": 0.217876016081743,
      "grad_norm": 0.08072310608274053,
      "learning_rate": 0.00019971750222893536,
      "loss": 0.865,
      "step": 12410
    },
    {
      "epoch": 0.2180515809617444,
      "grad_norm": 0.061596285709073675,
      "learning_rate": 0.00019971666036016635,
      "loss": 0.8583,
      "step": 12420
    },
    {
      "epoch": 0.21822714584174582,
      "grad_norm": 0.07435532849128003,
      "learning_rate": 0.00019971581724063836,
      "loss": 0.8668,
      "step": 12430
    },
    {
      "epoch": 0.21840271072174722,
      "grad_norm": 0.07860869212431895,
      "learning_rate": 0.00019971497287036202,
      "loss": 0.8552,
      "step": 12440
    },
    {
      "epoch": 0.21857827560174864,
      "grad_norm": 0.07657208275967126,
      "learning_rate": 0.00019971412724934805,
      "loss": 0.856,
      "step": 12450
    },
    {
      "epoch": 0.21875384048175003,
      "grad_norm": 0.057294883032555526,
      "learning_rate": 0.00019971328037760722,
      "loss": 0.8731,
      "step": 12460
    },
    {
      "epoch": 0.21892940536175143,
      "grad_norm": 0.06237115675749068,
      "learning_rate": 0.00019971243225515016,
      "loss": 0.8689,
      "step": 12470
    },
    {
      "epoch": 0.21910497024175285,
      "grad_norm": 0.07578260272482426,
      "learning_rate": 0.00019971158288198767,
      "loss": 0.8623,
      "step": 12480
    },
    {
      "epoch": 0.21928053512175424,
      "grad_norm": 0.07574743210172352,
      "learning_rate": 0.00019971073225813054,
      "loss": 0.8592,
      "step": 12490
    },
    {
      "epoch": 0.21945610000175564,
      "grad_norm": 0.06403675321513853,
      "learning_rate": 0.00019970988038358947,
      "loss": 0.8643,
      "step": 12500
    },
    {
      "epoch": 0.21963166488175706,
      "grad_norm": 0.05222275074125202,
      "learning_rate": 0.00019970902725837533,
      "loss": 0.8596,
      "step": 12510
    },
    {
      "epoch": 0.21980722976175845,
      "grad_norm": 0.0952052446680442,
      "learning_rate": 0.0001997081728824989,
      "loss": 0.8668,
      "step": 12520
    },
    {
      "epoch": 0.21998279464175985,
      "grad_norm": 0.05646238437575283,
      "learning_rate": 0.00019970731725597098,
      "loss": 0.8697,
      "step": 12530
    },
    {
      "epoch": 0.22015835952176127,
      "grad_norm": 0.08764024551743951,
      "learning_rate": 0.00019970646037880242,
      "loss": 0.8566,
      "step": 12540
    },
    {
      "epoch": 0.22033392440176267,
      "grad_norm": 0.0765120318770736,
      "learning_rate": 0.0001997056022510041,
      "loss": 0.8567,
      "step": 12550
    },
    {
      "epoch": 0.2205094892817641,
      "grad_norm": 0.0851652080867763,
      "learning_rate": 0.00019970474287258692,
      "loss": 0.8604,
      "step": 12560
    },
    {
      "epoch": 0.22068505416176548,
      "grad_norm": 0.06982621236973152,
      "learning_rate": 0.0001997038822435617,
      "loss": 0.8584,
      "step": 12570
    },
    {
      "epoch": 0.22086061904176688,
      "grad_norm": 0.07995651231900665,
      "learning_rate": 0.00019970302036393937,
      "loss": 0.8638,
      "step": 12580
    },
    {
      "epoch": 0.2210361839217683,
      "grad_norm": 0.06480861525741188,
      "learning_rate": 0.00019970215723373087,
      "loss": 0.858,
      "step": 12590
    },
    {
      "epoch": 0.2212117488017697,
      "grad_norm": 0.07618966053396582,
      "learning_rate": 0.00019970129285294712,
      "loss": 0.8584,
      "step": 12600
    },
    {
      "epoch": 0.2213873136817711,
      "grad_norm": 0.049028905927910045,
      "learning_rate": 0.00019970042722159908,
      "loss": 0.86,
      "step": 12610
    },
    {
      "epoch": 0.2215628785617725,
      "grad_norm": 0.06602833397277362,
      "learning_rate": 0.0001996995603396977,
      "loss": 0.8631,
      "step": 12620
    },
    {
      "epoch": 0.2217384434417739,
      "grad_norm": 0.06458237149426653,
      "learning_rate": 0.00019969869220725397,
      "loss": 0.8551,
      "step": 12630
    },
    {
      "epoch": 0.22191400832177532,
      "grad_norm": 0.07705687012468981,
      "learning_rate": 0.00019969782282427888,
      "loss": 0.8578,
      "step": 12640
    },
    {
      "epoch": 0.22208957320177672,
      "grad_norm": 0.06688938074370633,
      "learning_rate": 0.00019969695219078352,
      "loss": 0.8566,
      "step": 12650
    },
    {
      "epoch": 0.2222651380817781,
      "grad_norm": 0.07245910137021633,
      "learning_rate": 0.0001996960803067788,
      "loss": 0.8579,
      "step": 12660
    },
    {
      "epoch": 0.22244070296177954,
      "grad_norm": 0.05366959463862416,
      "learning_rate": 0.00019969520717227587,
      "loss": 0.8657,
      "step": 12670
    },
    {
      "epoch": 0.22261626784178093,
      "grad_norm": 0.0671181956840021,
      "learning_rate": 0.0001996943327872857,
      "loss": 0.8582,
      "step": 12680
    },
    {
      "epoch": 0.22279183272178232,
      "grad_norm": 0.06790553814166925,
      "learning_rate": 0.00019969345715181947,
      "loss": 0.8598,
      "step": 12690
    },
    {
      "epoch": 0.22296739760178375,
      "grad_norm": 0.08797915280816915,
      "learning_rate": 0.0001996925802658882,
      "loss": 0.8552,
      "step": 12700
    },
    {
      "epoch": 0.22314296248178514,
      "grad_norm": 0.08025084909632477,
      "learning_rate": 0.00019969170212950303,
      "loss": 0.8596,
      "step": 12710
    },
    {
      "epoch": 0.22331852736178656,
      "grad_norm": 0.06973590331103854,
      "learning_rate": 0.00019969082274267508,
      "loss": 0.8561,
      "step": 12720
    },
    {
      "epoch": 0.22349409224178796,
      "grad_norm": 0.10077642434926494,
      "learning_rate": 0.00019968994210541548,
      "loss": 0.8638,
      "step": 12730
    },
    {
      "epoch": 0.22366965712178935,
      "grad_norm": 0.0684252453798082,
      "learning_rate": 0.00019968906021773539,
      "loss": 0.8571,
      "step": 12740
    },
    {
      "epoch": 0.22384522200179077,
      "grad_norm": 0.05483887410424092,
      "learning_rate": 0.00019968817707964603,
      "loss": 0.8582,
      "step": 12750
    },
    {
      "epoch": 0.22402078688179217,
      "grad_norm": 0.08146992809927943,
      "learning_rate": 0.00019968729269115854,
      "loss": 0.8673,
      "step": 12760
    },
    {
      "epoch": 0.22419635176179356,
      "grad_norm": 0.06169269993917642,
      "learning_rate": 0.00019968640705228413,
      "loss": 0.855,
      "step": 12770
    },
    {
      "epoch": 0.22437191664179498,
      "grad_norm": 0.11292028485617943,
      "learning_rate": 0.00019968552016303405,
      "loss": 0.8597,
      "step": 12780
    },
    {
      "epoch": 0.22454748152179638,
      "grad_norm": 0.08681227251317894,
      "learning_rate": 0.00019968463202341946,
      "loss": 0.8586,
      "step": 12790
    },
    {
      "epoch": 0.22472304640179777,
      "grad_norm": 0.07958678278893613,
      "learning_rate": 0.0001996837426334517,
      "loss": 0.859,
      "step": 12800
    },
    {
      "epoch": 0.2248986112817992,
      "grad_norm": 0.08250478589411411,
      "learning_rate": 0.00019968285199314202,
      "loss": 0.8555,
      "step": 12810
    },
    {
      "epoch": 0.2250741761618006,
      "grad_norm": 0.06746346418802676,
      "learning_rate": 0.0001996819601025017,
      "loss": 0.8577,
      "step": 12820
    },
    {
      "epoch": 0.225249741041802,
      "grad_norm": 0.0660527416449139,
      "learning_rate": 0.000199681066961542,
      "loss": 0.8591,
      "step": 12830
    },
    {
      "epoch": 0.2254253059218034,
      "grad_norm": 0.05955655621897878,
      "learning_rate": 0.0001996801725702743,
      "loss": 0.8551,
      "step": 12840
    },
    {
      "epoch": 0.2256008708018048,
      "grad_norm": 0.06336162683197806,
      "learning_rate": 0.0001996792769287099,
      "loss": 0.8566,
      "step": 12850
    },
    {
      "epoch": 0.22577643568180622,
      "grad_norm": 0.09230503136106817,
      "learning_rate": 0.00019967838003686014,
      "loss": 0.8508,
      "step": 12860
    },
    {
      "epoch": 0.22595200056180761,
      "grad_norm": 0.08172690090519831,
      "learning_rate": 0.0001996774818947364,
      "loss": 0.8538,
      "step": 12870
    },
    {
      "epoch": 0.226127565441809,
      "grad_norm": 0.06363898315501915,
      "learning_rate": 0.00019967658250235004,
      "loss": 0.8656,
      "step": 12880
    },
    {
      "epoch": 0.22630313032181043,
      "grad_norm": 0.05474744709508439,
      "learning_rate": 0.00019967568185971245,
      "loss": 0.8625,
      "step": 12890
    },
    {
      "epoch": 0.22647869520181182,
      "grad_norm": 0.07630171477988831,
      "learning_rate": 0.0001996747799668351,
      "loss": 0.861,
      "step": 12900
    },
    {
      "epoch": 0.22665426008181325,
      "grad_norm": 0.08011111797927585,
      "learning_rate": 0.00019967387682372938,
      "loss": 0.8581,
      "step": 12910
    },
    {
      "epoch": 0.22682982496181464,
      "grad_norm": 0.10084510686913023,
      "learning_rate": 0.0001996729724304067,
      "loss": 0.8472,
      "step": 12920
    },
    {
      "epoch": 0.22700538984181604,
      "grad_norm": 0.06920932098291765,
      "learning_rate": 0.00019967206678687857,
      "loss": 0.8629,
      "step": 12930
    },
    {
      "epoch": 0.22718095472181746,
      "grad_norm": 0.09757733013923876,
      "learning_rate": 0.00019967115989315642,
      "loss": 0.8625,
      "step": 12940
    },
    {
      "epoch": 0.22735651960181885,
      "grad_norm": 0.06528613200478335,
      "learning_rate": 0.00019967025174925178,
      "loss": 0.8539,
      "step": 12950
    },
    {
      "epoch": 0.22753208448182025,
      "grad_norm": 0.07619606879973376,
      "learning_rate": 0.00019966934235517617,
      "loss": 0.8573,
      "step": 12960
    },
    {
      "epoch": 0.22770764936182167,
      "grad_norm": 0.059010347342820214,
      "learning_rate": 0.00019966843171094102,
      "loss": 0.8656,
      "step": 12970
    },
    {
      "epoch": 0.22788321424182306,
      "grad_norm": 0.05274377167490917,
      "learning_rate": 0.000199667519816558,
      "loss": 0.8572,
      "step": 12980
    },
    {
      "epoch": 0.22805877912182446,
      "grad_norm": 0.07310528239007595,
      "learning_rate": 0.00019966660667203852,
      "loss": 0.8648,
      "step": 12990
    },
    {
      "epoch": 0.22823434400182588,
      "grad_norm": 0.09604751435095704,
      "learning_rate": 0.00019966569227739426,
      "loss": 0.8553,
      "step": 13000
    },
    {
      "epoch": 0.22840990888182727,
      "grad_norm": 0.07868724830386786,
      "learning_rate": 0.00019966477663263679,
      "loss": 0.8638,
      "step": 13010
    },
    {
      "epoch": 0.2285854737618287,
      "grad_norm": 0.0807939148269027,
      "learning_rate": 0.00019966385973777765,
      "loss": 0.859,
      "step": 13020
    },
    {
      "epoch": 0.2287610386418301,
      "grad_norm": 0.06024600585240205,
      "learning_rate": 0.00019966294159282853,
      "loss": 0.8583,
      "step": 13030
    },
    {
      "epoch": 0.22893660352183148,
      "grad_norm": 0.05430868582058863,
      "learning_rate": 0.00019966202219780102,
      "loss": 0.8612,
      "step": 13040
    },
    {
      "epoch": 0.2291121684018329,
      "grad_norm": 0.059345834687725835,
      "learning_rate": 0.0001996611015527068,
      "loss": 0.8614,
      "step": 13050
    },
    {
      "epoch": 0.2292877332818343,
      "grad_norm": 0.07445285317788591,
      "learning_rate": 0.00019966017965755746,
      "loss": 0.8554,
      "step": 13060
    },
    {
      "epoch": 0.2294632981618357,
      "grad_norm": 0.08559837819380844,
      "learning_rate": 0.0001996592565123648,
      "loss": 0.8597,
      "step": 13070
    },
    {
      "epoch": 0.22963886304183712,
      "grad_norm": 0.10292629028492462,
      "learning_rate": 0.0001996583321171404,
      "loss": 0.8587,
      "step": 13080
    },
    {
      "epoch": 0.2298144279218385,
      "grad_norm": 0.09883399129712846,
      "learning_rate": 0.00019965740647189605,
      "loss": 0.8628,
      "step": 13090
    },
    {
      "epoch": 0.22998999280183993,
      "grad_norm": 0.06443772944297897,
      "learning_rate": 0.00019965647957664347,
      "loss": 0.8571,
      "step": 13100
    },
    {
      "epoch": 0.23016555768184133,
      "grad_norm": 0.08160493221035,
      "learning_rate": 0.00019965555143139436,
      "loss": 0.8638,
      "step": 13110
    },
    {
      "epoch": 0.23034112256184272,
      "grad_norm": 0.06517730571015831,
      "learning_rate": 0.0001996546220361605,
      "loss": 0.8633,
      "step": 13120
    },
    {
      "epoch": 0.23051668744184414,
      "grad_norm": 0.06574610269686215,
      "learning_rate": 0.00019965369139095368,
      "loss": 0.8684,
      "step": 13130
    },
    {
      "epoch": 0.23069225232184554,
      "grad_norm": 0.06441751554963561,
      "learning_rate": 0.0001996527594957857,
      "loss": 0.8664,
      "step": 13140
    },
    {
      "epoch": 0.23086781720184693,
      "grad_norm": 0.059154123938783665,
      "learning_rate": 0.00019965182635066836,
      "loss": 0.8596,
      "step": 13150
    },
    {
      "epoch": 0.23104338208184835,
      "grad_norm": 0.09180230055547017,
      "learning_rate": 0.00019965089195561343,
      "loss": 0.8657,
      "step": 13160
    },
    {
      "epoch": 0.23121894696184975,
      "grad_norm": 0.07403483364189829,
      "learning_rate": 0.0001996499563106328,
      "loss": 0.8632,
      "step": 13170
    },
    {
      "epoch": 0.23139451184185117,
      "grad_norm": 0.11631070950437476,
      "learning_rate": 0.00019964901941573834,
      "loss": 0.8585,
      "step": 13180
    },
    {
      "epoch": 0.23157007672185256,
      "grad_norm": 0.08959191594891236,
      "learning_rate": 0.0001996480812709419,
      "loss": 0.8671,
      "step": 13190
    },
    {
      "epoch": 0.23174564160185396,
      "grad_norm": 0.06440500271621241,
      "learning_rate": 0.00019964714187625535,
      "loss": 0.8567,
      "step": 13200
    },
    {
      "epoch": 0.23192120648185538,
      "grad_norm": 0.08028648687117654,
      "learning_rate": 0.00019964620123169062,
      "loss": 0.8535,
      "step": 13210
    },
    {
      "epoch": 0.23209677136185677,
      "grad_norm": 0.0908538771144053,
      "learning_rate": 0.00019964525933725962,
      "loss": 0.8603,
      "step": 13220
    },
    {
      "epoch": 0.23227233624185817,
      "grad_norm": 0.061861057542006596,
      "learning_rate": 0.00019964431619297424,
      "loss": 0.8639,
      "step": 13230
    },
    {
      "epoch": 0.2324479011218596,
      "grad_norm": 0.07480407813581291,
      "learning_rate": 0.00019964337179884648,
      "loss": 0.8595,
      "step": 13240
    },
    {
      "epoch": 0.23262346600186098,
      "grad_norm": 0.09391368857339884,
      "learning_rate": 0.00019964242615488833,
      "loss": 0.8603,
      "step": 13250
    },
    {
      "epoch": 0.23279903088186238,
      "grad_norm": 0.07434536311714536,
      "learning_rate": 0.0001996414792611117,
      "loss": 0.8604,
      "step": 13260
    },
    {
      "epoch": 0.2329745957618638,
      "grad_norm": 0.0713160482851545,
      "learning_rate": 0.00019964053111752865,
      "loss": 0.8569,
      "step": 13270
    },
    {
      "epoch": 0.2331501606418652,
      "grad_norm": 0.06815794896120089,
      "learning_rate": 0.00019963958172415115,
      "loss": 0.8671,
      "step": 13280
    },
    {
      "epoch": 0.23332572552186662,
      "grad_norm": 0.09860457256832716,
      "learning_rate": 0.00019963863108099123,
      "loss": 0.8578,
      "step": 13290
    },
    {
      "epoch": 0.233501290401868,
      "grad_norm": 0.0817229894738131,
      "learning_rate": 0.00019963767918806098,
      "loss": 0.8501,
      "step": 13300
    },
    {
      "epoch": 0.2336768552818694,
      "grad_norm": 0.06633499747993062,
      "learning_rate": 0.0001996367260453724,
      "loss": 0.8583,
      "step": 13310
    },
    {
      "epoch": 0.23385242016187083,
      "grad_norm": 0.06309723602880987,
      "learning_rate": 0.00019963577165293763,
      "loss": 0.86,
      "step": 13320
    },
    {
      "epoch": 0.23402798504187222,
      "grad_norm": 0.08032290564622807,
      "learning_rate": 0.0001996348160107687,
      "loss": 0.8606,
      "step": 13330
    },
    {
      "epoch": 0.23420354992187362,
      "grad_norm": 0.08485843992118826,
      "learning_rate": 0.0001996338591188778,
      "loss": 0.8666,
      "step": 13340
    },
    {
      "epoch": 0.23437911480187504,
      "grad_norm": 0.059068062789073256,
      "learning_rate": 0.00019963290097727693,
      "loss": 0.8603,
      "step": 13350
    },
    {
      "epoch": 0.23455467968187643,
      "grad_norm": 0.0770980751586183,
      "learning_rate": 0.00019963194158597838,
      "loss": 0.8573,
      "step": 13360
    },
    {
      "epoch": 0.23473024456187785,
      "grad_norm": 0.057733363725603934,
      "learning_rate": 0.00019963098094499416,
      "loss": 0.8554,
      "step": 13370
    },
    {
      "epoch": 0.23490580944187925,
      "grad_norm": 0.059937879996093374,
      "learning_rate": 0.00019963001905433657,
      "loss": 0.8592,
      "step": 13380
    },
    {
      "epoch": 0.23508137432188064,
      "grad_norm": 0.0632224404908158,
      "learning_rate": 0.00019962905591401766,
      "loss": 0.8578,
      "step": 13390
    },
    {
      "epoch": 0.23525693920188206,
      "grad_norm": 0.08170670551388819,
      "learning_rate": 0.00019962809152404977,
      "loss": 0.8566,
      "step": 13400
    },
    {
      "epoch": 0.23543250408188346,
      "grad_norm": 0.07861434380510587,
      "learning_rate": 0.00019962712588444503,
      "loss": 0.8617,
      "step": 13410
    },
    {
      "epoch": 0.23560806896188485,
      "grad_norm": 0.08039197674765287,
      "learning_rate": 0.0001996261589952157,
      "loss": 0.8576,
      "step": 13420
    },
    {
      "epoch": 0.23578363384188628,
      "grad_norm": 0.09240022885979778,
      "learning_rate": 0.00019962519085637407,
      "loss": 0.865,
      "step": 13430
    },
    {
      "epoch": 0.23595919872188767,
      "grad_norm": 0.07986316395702651,
      "learning_rate": 0.00019962422146793234,
      "loss": 0.8565,
      "step": 13440
    },
    {
      "epoch": 0.2361347636018891,
      "grad_norm": 0.06574640524381067,
      "learning_rate": 0.00019962325082990284,
      "loss": 0.8602,
      "step": 13450
    },
    {
      "epoch": 0.23631032848189049,
      "grad_norm": 0.09117917135106308,
      "learning_rate": 0.00019962227894229783,
      "loss": 0.8567,
      "step": 13460
    },
    {
      "epoch": 0.23648589336189188,
      "grad_norm": 0.05627202574634041,
      "learning_rate": 0.00019962130580512965,
      "loss": 0.8615,
      "step": 13470
    },
    {
      "epoch": 0.2366614582418933,
      "grad_norm": 0.05519988425747287,
      "learning_rate": 0.00019962033141841064,
      "loss": 0.8668,
      "step": 13480
    },
    {
      "epoch": 0.2368370231218947,
      "grad_norm": 0.06553942555227904,
      "learning_rate": 0.0001996193557821531,
      "loss": 0.8623,
      "step": 13490
    },
    {
      "epoch": 0.2370125880018961,
      "grad_norm": 0.07534788219067134,
      "learning_rate": 0.00019961837889636942,
      "loss": 0.8549,
      "step": 13500
    },
    {
      "epoch": 0.2371881528818975,
      "grad_norm": 0.11385461067284922,
      "learning_rate": 0.00019961740076107203,
      "loss": 0.8501,
      "step": 13510
    },
    {
      "epoch": 0.2373637177618989,
      "grad_norm": 0.06704085408675474,
      "learning_rate": 0.00019961642137627324,
      "loss": 0.8582,
      "step": 13520
    },
    {
      "epoch": 0.2375392826419003,
      "grad_norm": 0.08298062875200847,
      "learning_rate": 0.00019961544074198552,
      "loss": 0.857,
      "step": 13530
    },
    {
      "epoch": 0.23771484752190172,
      "grad_norm": 0.06494741095446872,
      "learning_rate": 0.00019961445885822123,
      "loss": 0.8561,
      "step": 13540
    },
    {
      "epoch": 0.23789041240190312,
      "grad_norm": 0.053319854546262105,
      "learning_rate": 0.00019961347572499286,
      "loss": 0.8657,
      "step": 13550
    },
    {
      "epoch": 0.23806597728190454,
      "grad_norm": 0.052530665080692034,
      "learning_rate": 0.00019961249134231284,
      "loss": 0.8636,
      "step": 13560
    },
    {
      "epoch": 0.23824154216190593,
      "grad_norm": 0.10420331434630822,
      "learning_rate": 0.0001996115057101937,
      "loss": 0.8605,
      "step": 13570
    },
    {
      "epoch": 0.23841710704190733,
      "grad_norm": 0.07513008704617051,
      "learning_rate": 0.00019961051882864782,
      "loss": 0.8626,
      "step": 13580
    },
    {
      "epoch": 0.23859267192190875,
      "grad_norm": 0.07083012319519245,
      "learning_rate": 0.00019960953069768781,
      "loss": 0.8547,
      "step": 13590
    },
    {
      "epoch": 0.23876823680191014,
      "grad_norm": 0.07477453054855489,
      "learning_rate": 0.0001996085413173261,
      "loss": 0.8615,
      "step": 13600
    },
    {
      "epoch": 0.23894380168191154,
      "grad_norm": 0.07064858969473756,
      "learning_rate": 0.00019960755068757532,
      "loss": 0.8597,
      "step": 13610
    },
    {
      "epoch": 0.23911936656191296,
      "grad_norm": 0.06942207105254398,
      "learning_rate": 0.00019960655880844796,
      "loss": 0.856,
      "step": 13620
    },
    {
      "epoch": 0.23929493144191435,
      "grad_norm": 0.05685954542760512,
      "learning_rate": 0.00019960556567995662,
      "loss": 0.8626,
      "step": 13630
    },
    {
      "epoch": 0.23947049632191578,
      "grad_norm": 0.06726683239476987,
      "learning_rate": 0.00019960457130211386,
      "loss": 0.8635,
      "step": 13640
    },
    {
      "epoch": 0.23964606120191717,
      "grad_norm": 0.06784527467547917,
      "learning_rate": 0.00019960357567493228,
      "loss": 0.8678,
      "step": 13650
    },
    {
      "epoch": 0.23982162608191857,
      "grad_norm": 0.07891725868316742,
      "learning_rate": 0.0001996025787984245,
      "loss": 0.8626,
      "step": 13660
    },
    {
      "epoch": 0.23999719096192,
      "grad_norm": 0.08630732144929593,
      "learning_rate": 0.00019960158067260312,
      "loss": 0.8585,
      "step": 13670
    },
    {
      "epoch": 0.24017275584192138,
      "grad_norm": 0.06865507076556479,
      "learning_rate": 0.00019960058129748085,
      "loss": 0.8639,
      "step": 13680
    },
    {
      "epoch": 0.24034832072192278,
      "grad_norm": 0.06533006848821106,
      "learning_rate": 0.00019959958067307032,
      "loss": 0.8584,
      "step": 13690
    },
    {
      "epoch": 0.2405238856019242,
      "grad_norm": 0.09544176297944504,
      "learning_rate": 0.0001995985787993842,
      "loss": 0.8646,
      "step": 13700
    },
    {
      "epoch": 0.2406994504819256,
      "grad_norm": 0.07944397864382828,
      "learning_rate": 0.00019959757567643519,
      "loss": 0.8632,
      "step": 13710
    },
    {
      "epoch": 0.240875015361927,
      "grad_norm": 0.0724567204717732,
      "learning_rate": 0.00019959657130423598,
      "loss": 0.8555,
      "step": 13720
    },
    {
      "epoch": 0.2410505802419284,
      "grad_norm": 0.07475838880427423,
      "learning_rate": 0.00019959556568279932,
      "loss": 0.869,
      "step": 13730
    },
    {
      "epoch": 0.2412261451219298,
      "grad_norm": 0.0800508272004524,
      "learning_rate": 0.000199594558812138,
      "loss": 0.8662,
      "step": 13740
    },
    {
      "epoch": 0.24140171000193122,
      "grad_norm": 0.05910746291538407,
      "learning_rate": 0.00019959355069226463,
      "loss": 0.867,
      "step": 13750
    },
    {
      "epoch": 0.24157727488193262,
      "grad_norm": 0.06406163985301921,
      "learning_rate": 0.00019959254132319215,
      "loss": 0.8638,
      "step": 13760
    },
    {
      "epoch": 0.241752839761934,
      "grad_norm": 0.07003991908556866,
      "learning_rate": 0.00019959153070493325,
      "loss": 0.8642,
      "step": 13770
    },
    {
      "epoch": 0.24192840464193544,
      "grad_norm": 0.0948217313141039,
      "learning_rate": 0.00019959051883750077,
      "loss": 0.8598,
      "step": 13780
    },
    {
      "epoch": 0.24210396952193683,
      "grad_norm": 0.06983295127303583,
      "learning_rate": 0.00019958950572090752,
      "loss": 0.8567,
      "step": 13790
    },
    {
      "epoch": 0.24227953440193822,
      "grad_norm": 0.055347991214125467,
      "learning_rate": 0.00019958849135516635,
      "loss": 0.8616,
      "step": 13800
    },
    {
      "epoch": 0.24245509928193965,
      "grad_norm": 0.05283474101240031,
      "learning_rate": 0.00019958747574029006,
      "loss": 0.8635,
      "step": 13810
    },
    {
      "epoch": 0.24263066416194104,
      "grad_norm": 0.09519078802638069,
      "learning_rate": 0.0001995864588762916,
      "loss": 0.8598,
      "step": 13820
    },
    {
      "epoch": 0.24280622904194246,
      "grad_norm": 0.07189266335151935,
      "learning_rate": 0.00019958544076318377,
      "loss": 0.8677,
      "step": 13830
    },
    {
      "epoch": 0.24298179392194386,
      "grad_norm": 0.055292437018689106,
      "learning_rate": 0.00019958442140097956,
      "loss": 0.8526,
      "step": 13840
    },
    {
      "epoch": 0.24315735880194525,
      "grad_norm": 0.08381916632876933,
      "learning_rate": 0.00019958340078969183,
      "loss": 0.8661,
      "step": 13850
    },
    {
      "epoch": 0.24333292368194667,
      "grad_norm": 0.10620086717544801,
      "learning_rate": 0.00019958237892933348,
      "loss": 0.8608,
      "step": 13860
    },
    {
      "epoch": 0.24350848856194807,
      "grad_norm": 0.07391684586630098,
      "learning_rate": 0.00019958135581991753,
      "loss": 0.8531,
      "step": 13870
    },
    {
      "epoch": 0.24368405344194946,
      "grad_norm": 0.06797082916127659,
      "learning_rate": 0.00019958033146145692,
      "loss": 0.859,
      "step": 13880
    },
    {
      "epoch": 0.24385961832195088,
      "grad_norm": 0.062085425271156326,
      "learning_rate": 0.00019957930585396457,
      "loss": 0.8607,
      "step": 13890
    },
    {
      "epoch": 0.24403518320195228,
      "grad_norm": 0.06447247822345926,
      "learning_rate": 0.00019957827899745354,
      "loss": 0.8604,
      "step": 13900
    },
    {
      "epoch": 0.2442107480819537,
      "grad_norm": 0.06929102607381335,
      "learning_rate": 0.00019957725089193685,
      "loss": 0.8651,
      "step": 13910
    },
    {
      "epoch": 0.2443863129619551,
      "grad_norm": 0.08457797576370008,
      "learning_rate": 0.00019957622153742748,
      "loss": 0.8667,
      "step": 13920
    },
    {
      "epoch": 0.2445618778419565,
      "grad_norm": 0.07543369237807052,
      "learning_rate": 0.00019957519093393846,
      "loss": 0.8518,
      "step": 13930
    },
    {
      "epoch": 0.2447374427219579,
      "grad_norm": 0.05197133800374198,
      "learning_rate": 0.00019957415908148293,
      "loss": 0.8661,
      "step": 13940
    },
    {
      "epoch": 0.2449130076019593,
      "grad_norm": 0.06474494494041087,
      "learning_rate": 0.0001995731259800739,
      "loss": 0.8597,
      "step": 13950
    },
    {
      "epoch": 0.2450885724819607,
      "grad_norm": 0.05507811973767036,
      "learning_rate": 0.00019957209162972444,
      "loss": 0.8558,
      "step": 13960
    },
    {
      "epoch": 0.24526413736196212,
      "grad_norm": 0.06449006102083141,
      "learning_rate": 0.00019957105603044773,
      "loss": 0.8606,
      "step": 13970
    },
    {
      "epoch": 0.24543970224196351,
      "grad_norm": 0.06948425427336218,
      "learning_rate": 0.00019957001918225685,
      "loss": 0.8644,
      "step": 13980
    },
    {
      "epoch": 0.2456152671219649,
      "grad_norm": 0.07111062370139126,
      "learning_rate": 0.0001995689810851649,
      "loss": 0.8595,
      "step": 13990
    },
    {
      "epoch": 0.24579083200196633,
      "grad_norm": 0.08079611216972363,
      "learning_rate": 0.00019956794173918508,
      "loss": 0.8565,
      "step": 14000
    },
    {
      "epoch": 0.24596639688196772,
      "grad_norm": 0.06636650925602623,
      "learning_rate": 0.00019956690114433058,
      "loss": 0.8627,
      "step": 14010
    },
    {
      "epoch": 0.24614196176196915,
      "grad_norm": 0.06888002936063668,
      "learning_rate": 0.00019956585930061451,
      "loss": 0.8692,
      "step": 14020
    },
    {
      "epoch": 0.24631752664197054,
      "grad_norm": 0.0675547970497489,
      "learning_rate": 0.00019956481620805012,
      "loss": 0.8604,
      "step": 14030
    },
    {
      "epoch": 0.24649309152197194,
      "grad_norm": 0.06275054616704875,
      "learning_rate": 0.00019956377186665066,
      "loss": 0.8476,
      "step": 14040
    },
    {
      "epoch": 0.24666865640197336,
      "grad_norm": 0.08469841729288999,
      "learning_rate": 0.00019956272627642927,
      "loss": 0.8508,
      "step": 14050
    },
    {
      "epoch": 0.24684422128197475,
      "grad_norm": 0.06782748577518789,
      "learning_rate": 0.00019956167943739926,
      "loss": 0.8629,
      "step": 14060
    },
    {
      "epoch": 0.24701978616197615,
      "grad_norm": 0.1257202632698329,
      "learning_rate": 0.0001995606313495739,
      "loss": 0.8518,
      "step": 14070
    },
    {
      "epoch": 0.24719535104197757,
      "grad_norm": 0.08676896461838673,
      "learning_rate": 0.00019955958201296642,
      "loss": 0.8543,
      "step": 14080
    },
    {
      "epoch": 0.24737091592197896,
      "grad_norm": 0.11054184338165894,
      "learning_rate": 0.00019955853142759018,
      "loss": 0.8539,
      "step": 14090
    },
    {
      "epoch": 0.24754648080198038,
      "grad_norm": 0.06297914852774192,
      "learning_rate": 0.00019955747959345845,
      "loss": 0.8617,
      "step": 14100
    },
    {
      "epoch": 0.24772204568198178,
      "grad_norm": 0.059152322453887136,
      "learning_rate": 0.00019955642651058457,
      "loss": 0.8619,
      "step": 14110
    },
    {
      "epoch": 0.24789761056198317,
      "grad_norm": 0.0758873849376312,
      "learning_rate": 0.00019955537217898186,
      "loss": 0.8521,
      "step": 14120
    },
    {
      "epoch": 0.2480731754419846,
      "grad_norm": 0.051644108807882866,
      "learning_rate": 0.00019955431659866372,
      "loss": 0.8574,
      "step": 14130
    },
    {
      "epoch": 0.248248740321986,
      "grad_norm": 0.06901620691290002,
      "learning_rate": 0.00019955325976964348,
      "loss": 0.8534,
      "step": 14140
    },
    {
      "epoch": 0.24842430520198738,
      "grad_norm": 0.06283408010309022,
      "learning_rate": 0.00019955220169193456,
      "loss": 0.8548,
      "step": 14150
    },
    {
      "epoch": 0.2485998700819888,
      "grad_norm": 0.09762046283936589,
      "learning_rate": 0.00019955114236555035,
      "loss": 0.854,
      "step": 14160
    },
    {
      "epoch": 0.2487754349619902,
      "grad_norm": 0.07580064981869453,
      "learning_rate": 0.0001995500817905043,
      "loss": 0.8611,
      "step": 14170
    },
    {
      "epoch": 0.24895099984199162,
      "grad_norm": 0.07540993018317861,
      "learning_rate": 0.0001995490199668098,
      "loss": 0.8566,
      "step": 14180
    },
    {
      "epoch": 0.24912656472199302,
      "grad_norm": 0.08381911954618285,
      "learning_rate": 0.00019954795689448033,
      "loss": 0.8656,
      "step": 14190
    },
    {
      "epoch": 0.2493021296019944,
      "grad_norm": 0.08774567498476624,
      "learning_rate": 0.0001995468925735294,
      "loss": 0.8635,
      "step": 14200
    },
    {
      "epoch": 0.24947769448199583,
      "grad_norm": 0.0707726016112555,
      "learning_rate": 0.0001995458270039704,
      "loss": 0.8467,
      "step": 14210
    },
    {
      "epoch": 0.24965325936199723,
      "grad_norm": 0.08976732565043521,
      "learning_rate": 0.00019954476018581695,
      "loss": 0.8575,
      "step": 14220
    },
    {
      "epoch": 0.24982882424199862,
      "grad_norm": 0.07250109380827384,
      "learning_rate": 0.00019954369211908245,
      "loss": 0.8583,
      "step": 14230
    },
    {
      "epoch": 0.250004389122,
      "grad_norm": 0.055044823812773445,
      "learning_rate": 0.00019954262280378045,
      "loss": 0.8598,
      "step": 14240
    },
    {
      "epoch": 0.25017995400200144,
      "grad_norm": 0.07227214383739743,
      "learning_rate": 0.0001995415522399246,
      "loss": 0.8591,
      "step": 14250
    },
    {
      "epoch": 0.25035551888200286,
      "grad_norm": 0.0838073762374175,
      "learning_rate": 0.00019954048042752838,
      "loss": 0.8681,
      "step": 14260
    },
    {
      "epoch": 0.2505310837620042,
      "grad_norm": 0.06698892016465113,
      "learning_rate": 0.00019953940736660542,
      "loss": 0.8545,
      "step": 14270
    },
    {
      "epoch": 0.25070664864200565,
      "grad_norm": 0.0752340954482605,
      "learning_rate": 0.00019953833305716922,
      "loss": 0.853,
      "step": 14280
    },
    {
      "epoch": 0.25088221352200707,
      "grad_norm": 0.08350924890195188,
      "learning_rate": 0.00019953725749923348,
      "loss": 0.8627,
      "step": 14290
    },
    {
      "epoch": 0.25105777840200844,
      "grad_norm": 0.08450552930801697,
      "learning_rate": 0.0001995361806928118,
      "loss": 0.8551,
      "step": 14300
    },
    {
      "epoch": 0.25123334328200986,
      "grad_norm": 0.056231253469683766,
      "learning_rate": 0.00019953510263791784,
      "loss": 0.8516,
      "step": 14310
    },
    {
      "epoch": 0.2514089081620113,
      "grad_norm": 0.08440568554315536,
      "learning_rate": 0.00019953402333456525,
      "loss": 0.8627,
      "step": 14320
    },
    {
      "epoch": 0.2515844730420127,
      "grad_norm": 0.06958882240933123,
      "learning_rate": 0.0001995329427827677,
      "loss": 0.8623,
      "step": 14330
    },
    {
      "epoch": 0.25176003792201407,
      "grad_norm": 0.08809798929105665,
      "learning_rate": 0.00019953186098253887,
      "loss": 0.8567,
      "step": 14340
    },
    {
      "epoch": 0.2519356028020155,
      "grad_norm": 0.06397476853674225,
      "learning_rate": 0.00019953077793389247,
      "loss": 0.8696,
      "step": 14350
    },
    {
      "epoch": 0.2521111676820169,
      "grad_norm": 0.06087596368285615,
      "learning_rate": 0.00019952969363684227,
      "loss": 0.8554,
      "step": 14360
    },
    {
      "epoch": 0.2522867325620183,
      "grad_norm": 0.09796475742085116,
      "learning_rate": 0.0001995286080914019,
      "loss": 0.8549,
      "step": 14370
    },
    {
      "epoch": 0.2524622974420197,
      "grad_norm": 0.09323122966284848,
      "learning_rate": 0.00019952752129758525,
      "loss": 0.8562,
      "step": 14380
    },
    {
      "epoch": 0.2526378623220211,
      "grad_norm": 0.08104986213499887,
      "learning_rate": 0.000199526433255406,
      "loss": 0.8678,
      "step": 14390
    },
    {
      "epoch": 0.2528134272020225,
      "grad_norm": 0.0823092086972603,
      "learning_rate": 0.00019952534396487794,
      "loss": 0.8643,
      "step": 14400
    },
    {
      "epoch": 0.2529889920820239,
      "grad_norm": 0.06541238804776654,
      "learning_rate": 0.0001995242534260149,
      "loss": 0.8691,
      "step": 14410
    },
    {
      "epoch": 0.25316455696202533,
      "grad_norm": 0.06811580378784352,
      "learning_rate": 0.0001995231616388307,
      "loss": 0.8599,
      "step": 14420
    },
    {
      "epoch": 0.2533401218420267,
      "grad_norm": 0.052965301341770216,
      "learning_rate": 0.00019952206860333915,
      "loss": 0.8525,
      "step": 14430
    },
    {
      "epoch": 0.2535156867220281,
      "grad_norm": 0.08451420049721323,
      "learning_rate": 0.00019952097431955411,
      "loss": 0.8503,
      "step": 14440
    },
    {
      "epoch": 0.25369125160202954,
      "grad_norm": 0.10248483474458604,
      "learning_rate": 0.00019951987878748945,
      "loss": 0.862,
      "step": 14450
    },
    {
      "epoch": 0.2538668164820309,
      "grad_norm": 0.10038642706372371,
      "learning_rate": 0.00019951878200715904,
      "loss": 0.8652,
      "step": 14460
    },
    {
      "epoch": 0.25404238136203233,
      "grad_norm": 0.08031892442777946,
      "learning_rate": 0.00019951768397857676,
      "loss": 0.868,
      "step": 14470
    },
    {
      "epoch": 0.25421794624203375,
      "grad_norm": 0.09205453785237903,
      "learning_rate": 0.00019951658470175655,
      "loss": 0.8555,
      "step": 14480
    },
    {
      "epoch": 0.2543935111220351,
      "grad_norm": 0.060170175291020624,
      "learning_rate": 0.00019951548417671236,
      "loss": 0.8531,
      "step": 14490
    },
    {
      "epoch": 0.25456907600203654,
      "grad_norm": 0.08382637825923747,
      "learning_rate": 0.00019951438240345808,
      "loss": 0.8618,
      "step": 14500
    },
    {
      "epoch": 0.25474464088203796,
      "grad_norm": 0.0695712522637437,
      "learning_rate": 0.0001995132793820077,
      "loss": 0.8562,
      "step": 14510
    },
    {
      "epoch": 0.2549202057620394,
      "grad_norm": 0.0654489707017867,
      "learning_rate": 0.0001995121751123752,
      "loss": 0.8581,
      "step": 14520
    },
    {
      "epoch": 0.25509577064204075,
      "grad_norm": 0.068870240021273,
      "learning_rate": 0.00019951106959457455,
      "loss": 0.8598,
      "step": 14530
    },
    {
      "epoch": 0.2552713355220422,
      "grad_norm": 0.089192001381446,
      "learning_rate": 0.0001995099628286198,
      "loss": 0.8593,
      "step": 14540
    },
    {
      "epoch": 0.2554469004020436,
      "grad_norm": 0.07295650383382928,
      "learning_rate": 0.00019950885481452487,
      "loss": 0.8613,
      "step": 14550
    },
    {
      "epoch": 0.25562246528204496,
      "grad_norm": 0.08360592222962818,
      "learning_rate": 0.00019950774555230395,
      "loss": 0.8548,
      "step": 14560
    },
    {
      "epoch": 0.2557980301620464,
      "grad_norm": 0.08391169673849941,
      "learning_rate": 0.00019950663504197095,
      "loss": 0.8606,
      "step": 14570
    },
    {
      "epoch": 0.2559735950420478,
      "grad_norm": 0.10918252013656823,
      "learning_rate": 0.00019950552328354003,
      "loss": 0.8544,
      "step": 14580
    },
    {
      "epoch": 0.2561491599220492,
      "grad_norm": 0.08638902306163822,
      "learning_rate": 0.00019950441027702528,
      "loss": 0.856,
      "step": 14590
    },
    {
      "epoch": 0.2563247248020506,
      "grad_norm": 0.08630070903540663,
      "learning_rate": 0.00019950329602244076,
      "loss": 0.8591,
      "step": 14600
    },
    {
      "epoch": 0.256500289682052,
      "grad_norm": 0.08364492262500373,
      "learning_rate": 0.00019950218051980058,
      "loss": 0.8617,
      "step": 14610
    },
    {
      "epoch": 0.2566758545620534,
      "grad_norm": 0.05698496216139443,
      "learning_rate": 0.0001995010637691189,
      "loss": 0.8552,
      "step": 14620
    },
    {
      "epoch": 0.2568514194420548,
      "grad_norm": 0.06101855631838769,
      "learning_rate": 0.00019949994577040986,
      "loss": 0.8611,
      "step": 14630
    },
    {
      "epoch": 0.25702698432205623,
      "grad_norm": 0.06442491170136191,
      "learning_rate": 0.00019949882652368765,
      "loss": 0.8584,
      "step": 14640
    },
    {
      "epoch": 0.2572025492020576,
      "grad_norm": 0.0769865238760441,
      "learning_rate": 0.00019949770602896641,
      "loss": 0.8613,
      "step": 14650
    },
    {
      "epoch": 0.257378114082059,
      "grad_norm": 0.07344353739916515,
      "learning_rate": 0.00019949658428626037,
      "loss": 0.8579,
      "step": 14660
    },
    {
      "epoch": 0.25755367896206044,
      "grad_norm": 0.0674799196446192,
      "learning_rate": 0.0001994954612955837,
      "loss": 0.8661,
      "step": 14670
    },
    {
      "epoch": 0.25772924384206186,
      "grad_norm": 0.05989849292668042,
      "learning_rate": 0.00019949433705695072,
      "loss": 0.855,
      "step": 14680
    },
    {
      "epoch": 0.25790480872206323,
      "grad_norm": 0.08201718793868297,
      "learning_rate": 0.00019949321157037558,
      "loss": 0.8547,
      "step": 14690
    },
    {
      "epoch": 0.25808037360206465,
      "grad_norm": 0.10871927434689527,
      "learning_rate": 0.00019949208483587257,
      "loss": 0.8671,
      "step": 14700
    },
    {
      "epoch": 0.25825593848206607,
      "grad_norm": 0.07312195730501901,
      "learning_rate": 0.00019949095685345598,
      "loss": 0.8591,
      "step": 14710
    },
    {
      "epoch": 0.25843150336206744,
      "grad_norm": 0.06476372827452581,
      "learning_rate": 0.0001994898276231401,
      "loss": 0.8596,
      "step": 14720
    },
    {
      "epoch": 0.25860706824206886,
      "grad_norm": 0.06798667236769548,
      "learning_rate": 0.0001994886971449392,
      "loss": 0.8637,
      "step": 14730
    },
    {
      "epoch": 0.2587826331220703,
      "grad_norm": 0.064407349275021,
      "learning_rate": 0.00019948756541886766,
      "loss": 0.8557,
      "step": 14740
    },
    {
      "epoch": 0.25895819800207165,
      "grad_norm": 0.1185406828061444,
      "learning_rate": 0.00019948643244493977,
      "loss": 0.8576,
      "step": 14750
    },
    {
      "epoch": 0.25913376288207307,
      "grad_norm": 0.09598548925354039,
      "learning_rate": 0.0001994852982231699,
      "loss": 0.8503,
      "step": 14760
    },
    {
      "epoch": 0.2593093277620745,
      "grad_norm": 0.05763865742883013,
      "learning_rate": 0.00019948416275357244,
      "loss": 0.8626,
      "step": 14770
    },
    {
      "epoch": 0.25948489264207586,
      "grad_norm": 0.07027125494343812,
      "learning_rate": 0.00019948302603616175,
      "loss": 0.8672,
      "step": 14780
    },
    {
      "epoch": 0.2596604575220773,
      "grad_norm": 0.06563743906802047,
      "learning_rate": 0.00019948188807095225,
      "loss": 0.8627,
      "step": 14790
    },
    {
      "epoch": 0.2598360224020787,
      "grad_norm": 0.06751699640995153,
      "learning_rate": 0.00019948074885795834,
      "loss": 0.8561,
      "step": 14800
    },
    {
      "epoch": 0.26001158728208007,
      "grad_norm": 0.06416503390544841,
      "learning_rate": 0.00019947960839719454,
      "loss": 0.857,
      "step": 14810
    },
    {
      "epoch": 0.2601871521620815,
      "grad_norm": 0.10598730482068526,
      "learning_rate": 0.00019947846668867515,
      "loss": 0.8628,
      "step": 14820
    },
    {
      "epoch": 0.2603627170420829,
      "grad_norm": 0.08307887657328619,
      "learning_rate": 0.00019947732373241475,
      "loss": 0.8635,
      "step": 14830
    },
    {
      "epoch": 0.2605382819220843,
      "grad_norm": 0.07348607297169166,
      "learning_rate": 0.00019947617952842777,
      "loss": 0.8604,
      "step": 14840
    },
    {
      "epoch": 0.2607138468020857,
      "grad_norm": 0.07471403751065663,
      "learning_rate": 0.00019947503407672873,
      "loss": 0.8563,
      "step": 14850
    },
    {
      "epoch": 0.2608894116820871,
      "grad_norm": 0.0653038817964391,
      "learning_rate": 0.00019947388737733216,
      "loss": 0.8604,
      "step": 14860
    },
    {
      "epoch": 0.26106497656208855,
      "grad_norm": 0.08734429902691533,
      "learning_rate": 0.0001994727394302525,
      "loss": 0.8601,
      "step": 14870
    },
    {
      "epoch": 0.2612405414420899,
      "grad_norm": 0.09714174951971054,
      "learning_rate": 0.0001994715902355044,
      "loss": 0.8602,
      "step": 14880
    },
    {
      "epoch": 0.26141610632209134,
      "grad_norm": 0.05213582636621975,
      "learning_rate": 0.0001994704397931024,
      "loss": 0.8603,
      "step": 14890
    },
    {
      "epoch": 0.26159167120209276,
      "grad_norm": 0.09453060688591909,
      "learning_rate": 0.00019946928810306104,
      "loss": 0.8575,
      "step": 14900
    },
    {
      "epoch": 0.2617672360820941,
      "grad_norm": 0.06696302555144164,
      "learning_rate": 0.00019946813516539494,
      "loss": 0.8515,
      "step": 14910
    },
    {
      "epoch": 0.26194280096209555,
      "grad_norm": 0.07007177645375523,
      "learning_rate": 0.00019946698098011867,
      "loss": 0.8558,
      "step": 14920
    },
    {
      "epoch": 0.26211836584209697,
      "grad_norm": 0.07095748350563354,
      "learning_rate": 0.0001994658255472469,
      "loss": 0.8573,
      "step": 14930
    },
    {
      "epoch": 0.26229393072209833,
      "grad_norm": 0.08723730767261309,
      "learning_rate": 0.00019946466886679423,
      "loss": 0.8576,
      "step": 14940
    },
    {
      "epoch": 0.26246949560209976,
      "grad_norm": 0.06257371483317639,
      "learning_rate": 0.00019946351093877532,
      "loss": 0.8601,
      "step": 14950
    },
    {
      "epoch": 0.2626450604821012,
      "grad_norm": 0.07005616301223214,
      "learning_rate": 0.0001994623517632049,
      "loss": 0.8617,
      "step": 14960
    },
    {
      "epoch": 0.26282062536210254,
      "grad_norm": 0.10495167338404755,
      "learning_rate": 0.00019946119134009758,
      "loss": 0.8626,
      "step": 14970
    },
    {
      "epoch": 0.26299619024210397,
      "grad_norm": 0.07526928806416286,
      "learning_rate": 0.0001994600296694681,
      "loss": 0.8576,
      "step": 14980
    },
    {
      "epoch": 0.2631717551221054,
      "grad_norm": 0.09613534910303849,
      "learning_rate": 0.00019945886675133118,
      "loss": 0.8664,
      "step": 14990
    },
    {
      "epoch": 0.26334732000210676,
      "grad_norm": 0.07195775617282611,
      "learning_rate": 0.00019945770258570156,
      "loss": 0.8672,
      "step": 15000
    },
    {
      "epoch": 0.2635228848821082,
      "grad_norm": 0.0792644418192245,
      "learning_rate": 0.00019945653717259399,
      "loss": 0.8518,
      "step": 15010
    },
    {
      "epoch": 0.2636984497621096,
      "grad_norm": 0.08147981421247041,
      "learning_rate": 0.00019945537051202318,
      "loss": 0.8598,
      "step": 15020
    },
    {
      "epoch": 0.26387401464211097,
      "grad_norm": 0.06340849234358589,
      "learning_rate": 0.000199454202604004,
      "loss": 0.856,
      "step": 15030
    },
    {
      "epoch": 0.2640495795221124,
      "grad_norm": 0.04734024661268947,
      "learning_rate": 0.0001994530334485512,
      "loss": 0.861,
      "step": 15040
    },
    {
      "epoch": 0.2642251444021138,
      "grad_norm": 0.05809592723200135,
      "learning_rate": 0.00019945186304567956,
      "loss": 0.8604,
      "step": 15050
    },
    {
      "epoch": 0.26440070928211523,
      "grad_norm": 0.0760507148983711,
      "learning_rate": 0.00019945069139540398,
      "loss": 0.8486,
      "step": 15060
    },
    {
      "epoch": 0.2645762741621166,
      "grad_norm": 0.0809802331922669,
      "learning_rate": 0.00019944951849773924,
      "loss": 0.8625,
      "step": 15070
    },
    {
      "epoch": 0.264751839042118,
      "grad_norm": 0.0688588726644685,
      "learning_rate": 0.00019944834435270026,
      "loss": 0.8592,
      "step": 15080
    },
    {
      "epoch": 0.26492740392211944,
      "grad_norm": 0.08616665668888285,
      "learning_rate": 0.0001994471689603019,
      "loss": 0.8584,
      "step": 15090
    },
    {
      "epoch": 0.2651029688021208,
      "grad_norm": 0.08110949314089932,
      "learning_rate": 0.00019944599232055903,
      "loss": 0.8567,
      "step": 15100
    },
    {
      "epoch": 0.26527853368212223,
      "grad_norm": 0.07019644854833353,
      "learning_rate": 0.00019944481443348656,
      "loss": 0.857,
      "step": 15110
    },
    {
      "epoch": 0.26545409856212365,
      "grad_norm": 0.07357093181321922,
      "learning_rate": 0.00019944363529909943,
      "loss": 0.86,
      "step": 15120
    },
    {
      "epoch": 0.265629663442125,
      "grad_norm": 0.0817515902034547,
      "learning_rate": 0.00019944245491741258,
      "loss": 0.8679,
      "step": 15130
    },
    {
      "epoch": 0.26580522832212644,
      "grad_norm": 0.061286659231383246,
      "learning_rate": 0.00019944127328844097,
      "loss": 0.8588,
      "step": 15140
    },
    {
      "epoch": 0.26598079320212786,
      "grad_norm": 0.059173874132725804,
      "learning_rate": 0.00019944009041219954,
      "loss": 0.8604,
      "step": 15150
    },
    {
      "epoch": 0.26615635808212923,
      "grad_norm": 0.05197657774103479,
      "learning_rate": 0.0001994389062887033,
      "loss": 0.8554,
      "step": 15160
    },
    {
      "epoch": 0.26633192296213065,
      "grad_norm": 0.06808451894132131,
      "learning_rate": 0.00019943772091796726,
      "loss": 0.8608,
      "step": 15170
    },
    {
      "epoch": 0.2665074878421321,
      "grad_norm": 0.07047614635428266,
      "learning_rate": 0.0001994365343000064,
      "loss": 0.8555,
      "step": 15180
    },
    {
      "epoch": 0.26668305272213344,
      "grad_norm": 0.10508749893678086,
      "learning_rate": 0.00019943534643483585,
      "loss": 0.8618,
      "step": 15190
    },
    {
      "epoch": 0.26685861760213486,
      "grad_norm": 0.07136833586693185,
      "learning_rate": 0.00019943415732247055,
      "loss": 0.8591,
      "step": 15200
    },
    {
      "epoch": 0.2670341824821363,
      "grad_norm": 0.09983142544668047,
      "learning_rate": 0.00019943296696292562,
      "loss": 0.8585,
      "step": 15210
    },
    {
      "epoch": 0.26720974736213765,
      "grad_norm": 0.07251035341402826,
      "learning_rate": 0.00019943177535621613,
      "loss": 0.8621,
      "step": 15220
    },
    {
      "epoch": 0.2673853122421391,
      "grad_norm": 0.08439254746939306,
      "learning_rate": 0.00019943058250235717,
      "loss": 0.8585,
      "step": 15230
    },
    {
      "epoch": 0.2675608771221405,
      "grad_norm": 0.07012011605321045,
      "learning_rate": 0.00019942938840136388,
      "loss": 0.8533,
      "step": 15240
    },
    {
      "epoch": 0.2677364420021419,
      "grad_norm": 0.06644274019649771,
      "learning_rate": 0.00019942819305325137,
      "loss": 0.8554,
      "step": 15250
    },
    {
      "epoch": 0.2679120068821433,
      "grad_norm": 0.05553434813813748,
      "learning_rate": 0.00019942699645803482,
      "loss": 0.8567,
      "step": 15260
    },
    {
      "epoch": 0.2680875717621447,
      "grad_norm": 0.0615237408809731,
      "learning_rate": 0.00019942579861572933,
      "loss": 0.8576,
      "step": 15270
    },
    {
      "epoch": 0.2682631366421461,
      "grad_norm": 0.06356950031888757,
      "learning_rate": 0.0001994245995263501,
      "loss": 0.855,
      "step": 15280
    },
    {
      "epoch": 0.2684387015221475,
      "grad_norm": 0.05983926636846021,
      "learning_rate": 0.00019942339918991239,
      "loss": 0.8617,
      "step": 15290
    },
    {
      "epoch": 0.2686142664021489,
      "grad_norm": 0.07037096956391803,
      "learning_rate": 0.0001994221976064313,
      "loss": 0.8588,
      "step": 15300
    },
    {
      "epoch": 0.26878983128215034,
      "grad_norm": 0.06530545571099387,
      "learning_rate": 0.0001994209947759221,
      "loss": 0.8593,
      "step": 15310
    },
    {
      "epoch": 0.2689653961621517,
      "grad_norm": 0.059316112291684,
      "learning_rate": 0.00019941979069840003,
      "loss": 0.8662,
      "step": 15320
    },
    {
      "epoch": 0.2691409610421531,
      "grad_norm": 0.05477020282891137,
      "learning_rate": 0.00019941858537388037,
      "loss": 0.8613,
      "step": 15330
    },
    {
      "epoch": 0.26931652592215455,
      "grad_norm": 0.07017035524547499,
      "learning_rate": 0.00019941737880237837,
      "loss": 0.8565,
      "step": 15340
    },
    {
      "epoch": 0.2694920908021559,
      "grad_norm": 0.08967510436016346,
      "learning_rate": 0.00019941617098390933,
      "loss": 0.8558,
      "step": 15350
    },
    {
      "epoch": 0.26966765568215734,
      "grad_norm": 0.07798973540389004,
      "learning_rate": 0.00019941496191848852,
      "loss": 0.858,
      "step": 15360
    },
    {
      "epoch": 0.26984322056215876,
      "grad_norm": 0.06308067321717561,
      "learning_rate": 0.00019941375160613127,
      "loss": 0.8579,
      "step": 15370
    },
    {
      "epoch": 0.2700187854421601,
      "grad_norm": 0.08192827114395146,
      "learning_rate": 0.00019941254004685293,
      "loss": 0.8614,
      "step": 15380
    },
    {
      "epoch": 0.27019435032216155,
      "grad_norm": 0.05796423307785897,
      "learning_rate": 0.0001994113272406689,
      "loss": 0.8588,
      "step": 15390
    },
    {
      "epoch": 0.27036991520216297,
      "grad_norm": 0.061589070551759294,
      "learning_rate": 0.00019941011318759446,
      "loss": 0.8695,
      "step": 15400
    },
    {
      "epoch": 0.2705454800821644,
      "grad_norm": 0.07508114376145081,
      "learning_rate": 0.000199408897887645,
      "loss": 0.8664,
      "step": 15410
    },
    {
      "epoch": 0.27072104496216576,
      "grad_norm": 0.05938097155689713,
      "learning_rate": 0.00019940768134083596,
      "loss": 0.8591,
      "step": 15420
    },
    {
      "epoch": 0.2708966098421672,
      "grad_norm": 0.060566614201024854,
      "learning_rate": 0.00019940646354718274,
      "loss": 0.8593,
      "step": 15430
    },
    {
      "epoch": 0.2710721747221686,
      "grad_norm": 0.06092729151042539,
      "learning_rate": 0.00019940524450670077,
      "loss": 0.86,
      "step": 15440
    },
    {
      "epoch": 0.27124773960216997,
      "grad_norm": 0.09361074441279993,
      "learning_rate": 0.0001994040242194055,
      "loss": 0.8575,
      "step": 15450
    },
    {
      "epoch": 0.2714233044821714,
      "grad_norm": 0.08385137239710018,
      "learning_rate": 0.0001994028026853124,
      "loss": 0.8603,
      "step": 15460
    },
    {
      "epoch": 0.2715988693621728,
      "grad_norm": 0.06041815489567256,
      "learning_rate": 0.00019940157990443687,
      "loss": 0.8659,
      "step": 15470
    },
    {
      "epoch": 0.2717744342421742,
      "grad_norm": 0.05850804898985812,
      "learning_rate": 0.0001994003558767945,
      "loss": 0.8491,
      "step": 15480
    },
    {
      "epoch": 0.2719499991221756,
      "grad_norm": 0.06034553249868876,
      "learning_rate": 0.00019939913060240078,
      "loss": 0.873,
      "step": 15490
    },
    {
      "epoch": 0.272125564002177,
      "grad_norm": 0.06293535078141517,
      "learning_rate": 0.00019939790408127122,
      "loss": 0.8595,
      "step": 15500
    },
    {
      "epoch": 0.2723011288821784,
      "grad_norm": 0.0811121688892563,
      "learning_rate": 0.00019939667631342135,
      "loss": 0.8642,
      "step": 15510
    },
    {
      "epoch": 0.2724766937621798,
      "grad_norm": 0.06196933978325333,
      "learning_rate": 0.0001993954472988667,
      "loss": 0.858,
      "step": 15520
    },
    {
      "epoch": 0.27265225864218123,
      "grad_norm": 0.08701626229158647,
      "learning_rate": 0.00019939421703762288,
      "loss": 0.8583,
      "step": 15530
    },
    {
      "epoch": 0.2728278235221826,
      "grad_norm": 0.0763265601821762,
      "learning_rate": 0.0001993929855297055,
      "loss": 0.8569,
      "step": 15540
    },
    {
      "epoch": 0.273003388402184,
      "grad_norm": 0.09040556133898896,
      "learning_rate": 0.00019939175277513014,
      "loss": 0.8571,
      "step": 15550
    },
    {
      "epoch": 0.27317895328218544,
      "grad_norm": 0.10787783472187551,
      "learning_rate": 0.0001993905187739124,
      "loss": 0.8605,
      "step": 15560
    },
    {
      "epoch": 0.2733545181621868,
      "grad_norm": 0.05170503509826312,
      "learning_rate": 0.00019938928352606795,
      "loss": 0.8618,
      "step": 15570
    },
    {
      "epoch": 0.27353008304218823,
      "grad_norm": 0.0863280097679521,
      "learning_rate": 0.00019938804703161238,
      "loss": 0.8552,
      "step": 15580
    },
    {
      "epoch": 0.27370564792218965,
      "grad_norm": 0.0691094489284862,
      "learning_rate": 0.00019938680929056145,
      "loss": 0.8576,
      "step": 15590
    },
    {
      "epoch": 0.2738812128021911,
      "grad_norm": 0.063285310270822,
      "learning_rate": 0.00019938557030293075,
      "loss": 0.8642,
      "step": 15600
    },
    {
      "epoch": 0.27405677768219244,
      "grad_norm": 0.06332654858749048,
      "learning_rate": 0.00019938433006873603,
      "loss": 0.8619,
      "step": 15610
    },
    {
      "epoch": 0.27423234256219386,
      "grad_norm": 0.07587451469937638,
      "learning_rate": 0.00019938308858799298,
      "loss": 0.8627,
      "step": 15620
    },
    {
      "epoch": 0.2744079074421953,
      "grad_norm": 0.05670961516455953,
      "learning_rate": 0.00019938184586071737,
      "loss": 0.8561,
      "step": 15630
    },
    {
      "epoch": 0.27458347232219665,
      "grad_norm": 0.09779480659004015,
      "learning_rate": 0.0001993806018869249,
      "loss": 0.8614,
      "step": 15640
    },
    {
      "epoch": 0.2747590372021981,
      "grad_norm": 0.05622871037263839,
      "learning_rate": 0.00019937935666663133,
      "loss": 0.8564,
      "step": 15650
    },
    {
      "epoch": 0.2749346020821995,
      "grad_norm": 0.06997749513049606,
      "learning_rate": 0.00019937811019985247,
      "loss": 0.8535,
      "step": 15660
    },
    {
      "epoch": 0.27511016696220086,
      "grad_norm": 0.08100381783597384,
      "learning_rate": 0.00019937686248660413,
      "loss": 0.8561,
      "step": 15670
    },
    {
      "epoch": 0.2752857318422023,
      "grad_norm": 0.07749711760124976,
      "learning_rate": 0.0001993756135269021,
      "loss": 0.8556,
      "step": 15680
    },
    {
      "epoch": 0.2754612967222037,
      "grad_norm": 0.056564541601300394,
      "learning_rate": 0.00019937436332076217,
      "loss": 0.8585,
      "step": 15690
    },
    {
      "epoch": 0.2756368616022051,
      "grad_norm": 0.06651248399908018,
      "learning_rate": 0.00019937311186820017,
      "loss": 0.8625,
      "step": 15700
    },
    {
      "epoch": 0.2758124264822065,
      "grad_norm": 0.08598846150848699,
      "learning_rate": 0.00019937185916923201,
      "loss": 0.8623,
      "step": 15710
    },
    {
      "epoch": 0.2759879913622079,
      "grad_norm": 0.07862610369990877,
      "learning_rate": 0.00019937060522387358,
      "loss": 0.8566,
      "step": 15720
    },
    {
      "epoch": 0.2761635562422093,
      "grad_norm": 0.06960192857765705,
      "learning_rate": 0.0001993693500321407,
      "loss": 0.8614,
      "step": 15730
    },
    {
      "epoch": 0.2763391211222107,
      "grad_norm": 0.06732708537020778,
      "learning_rate": 0.00019936809359404926,
      "loss": 0.8585,
      "step": 15740
    },
    {
      "epoch": 0.27651468600221213,
      "grad_norm": 0.07534180119172672,
      "learning_rate": 0.00019936683590961526,
      "loss": 0.8596,
      "step": 15750
    },
    {
      "epoch": 0.2766902508822135,
      "grad_norm": 0.07088513712059127,
      "learning_rate": 0.00019936557697885463,
      "loss": 0.8656,
      "step": 15760
    },
    {
      "epoch": 0.2768658157622149,
      "grad_norm": 0.06972115601487668,
      "learning_rate": 0.00019936431680178324,
      "loss": 0.866,
      "step": 15770
    },
    {
      "epoch": 0.27704138064221634,
      "grad_norm": 0.07324793150015474,
      "learning_rate": 0.00019936305537841714,
      "loss": 0.8625,
      "step": 15780
    },
    {
      "epoch": 0.27721694552221776,
      "grad_norm": 0.07253579492218197,
      "learning_rate": 0.00019936179270877222,
      "loss": 0.8601,
      "step": 15790
    },
    {
      "epoch": 0.27739251040221913,
      "grad_norm": 0.0764964208725961,
      "learning_rate": 0.00019936052879286457,
      "loss": 0.8595,
      "step": 15800
    },
    {
      "epoch": 0.27756807528222055,
      "grad_norm": 0.06281302842248587,
      "learning_rate": 0.0001993592636307102,
      "loss": 0.8578,
      "step": 15810
    },
    {
      "epoch": 0.27774364016222197,
      "grad_norm": 0.07124531523815639,
      "learning_rate": 0.00019935799722232508,
      "loss": 0.8581,
      "step": 15820
    },
    {
      "epoch": 0.27791920504222334,
      "grad_norm": 0.07789560430737721,
      "learning_rate": 0.00019935672956772528,
      "loss": 0.8679,
      "step": 15830
    },
    {
      "epoch": 0.27809476992222476,
      "grad_norm": 0.07194036810212937,
      "learning_rate": 0.00019935546066692688,
      "loss": 0.8584,
      "step": 15840
    },
    {
      "epoch": 0.2782703348022262,
      "grad_norm": 0.06334317552377043,
      "learning_rate": 0.00019935419051994596,
      "loss": 0.8579,
      "step": 15850
    },
    {
      "epoch": 0.27844589968222755,
      "grad_norm": 0.07865531965696003,
      "learning_rate": 0.00019935291912679854,
      "loss": 0.8501,
      "step": 15860
    },
    {
      "epoch": 0.27862146456222897,
      "grad_norm": 0.08826749624793569,
      "learning_rate": 0.00019935164648750083,
      "loss": 0.8595,
      "step": 15870
    },
    {
      "epoch": 0.2787970294422304,
      "grad_norm": 0.06218307312058364,
      "learning_rate": 0.00019935037260206893,
      "loss": 0.8514,
      "step": 15880
    },
    {
      "epoch": 0.27897259432223176,
      "grad_norm": 0.07610431204157117,
      "learning_rate": 0.0001993490974705189,
      "loss": 0.8627,
      "step": 15890
    },
    {
      "epoch": 0.2791481592022332,
      "grad_norm": 0.0729606425154137,
      "learning_rate": 0.00019934782109286698,
      "loss": 0.8524,
      "step": 15900
    },
    {
      "epoch": 0.2793237240822346,
      "grad_norm": 0.06563619915488823,
      "learning_rate": 0.00019934654346912933,
      "loss": 0.8568,
      "step": 15910
    },
    {
      "epoch": 0.27949928896223597,
      "grad_norm": 0.0733487232391486,
      "learning_rate": 0.00019934526459932216,
      "loss": 0.8575,
      "step": 15920
    },
    {
      "epoch": 0.2796748538422374,
      "grad_norm": 0.07966694020584535,
      "learning_rate": 0.0001993439844834616,
      "loss": 0.8587,
      "step": 15930
    },
    {
      "epoch": 0.2798504187222388,
      "grad_norm": 0.08177977120850058,
      "learning_rate": 0.00019934270312156388,
      "loss": 0.8643,
      "step": 15940
    },
    {
      "epoch": 0.2800259836022402,
      "grad_norm": 0.07766388085649538,
      "learning_rate": 0.0001993414205136453,
      "loss": 0.8665,
      "step": 15950
    },
    {
      "epoch": 0.2802015484822416,
      "grad_norm": 0.09774588948139021,
      "learning_rate": 0.00019934013665972203,
      "loss": 0.8571,
      "step": 15960
    },
    {
      "epoch": 0.280377113362243,
      "grad_norm": 0.08825931357494128,
      "learning_rate": 0.00019933885155981045,
      "loss": 0.8643,
      "step": 15970
    },
    {
      "epoch": 0.28055267824224445,
      "grad_norm": 0.060966301096905894,
      "learning_rate": 0.00019933756521392675,
      "loss": 0.8532,
      "step": 15980
    },
    {
      "epoch": 0.2807282431222458,
      "grad_norm": 0.05961521169022152,
      "learning_rate": 0.00019933627762208723,
      "loss": 0.861,
      "step": 15990
    },
    {
      "epoch": 0.28090380800224724,
      "grad_norm": 0.06176946856910091,
      "learning_rate": 0.00019933498878430824,
      "loss": 0.8616,
      "step": 16000
    },
    {
      "epoch": 0.28107937288224866,
      "grad_norm": 0.07452369399288046,
      "learning_rate": 0.00019933369870060613,
      "loss": 0.8587,
      "step": 16010
    },
    {
      "epoch": 0.28125493776225,
      "grad_norm": 0.07487060103204875,
      "learning_rate": 0.0001993324073709972,
      "loss": 0.8659,
      "step": 16020
    },
    {
      "epoch": 0.28143050264225145,
      "grad_norm": 0.08865705455486707,
      "learning_rate": 0.00019933111479549778,
      "loss": 0.8575,
      "step": 16030
    },
    {
      "epoch": 0.28160606752225287,
      "grad_norm": 0.06922714460907559,
      "learning_rate": 0.00019932982097412428,
      "loss": 0.8653,
      "step": 16040
    },
    {
      "epoch": 0.28178163240225423,
      "grad_norm": 0.05603644345233025,
      "learning_rate": 0.00019932852590689317,
      "loss": 0.8525,
      "step": 16050
    },
    {
      "epoch": 0.28195719728225566,
      "grad_norm": 0.08778585024463534,
      "learning_rate": 0.00019932722959382077,
      "loss": 0.8585,
      "step": 16060
    },
    {
      "epoch": 0.2821327621622571,
      "grad_norm": 0.07017282190283822,
      "learning_rate": 0.0001993259320349235,
      "loss": 0.8588,
      "step": 16070
    },
    {
      "epoch": 0.28230832704225844,
      "grad_norm": 0.09574002950131953,
      "learning_rate": 0.00019932463323021787,
      "loss": 0.8645,
      "step": 16080
    },
    {
      "epoch": 0.28248389192225987,
      "grad_norm": 0.08459664422323991,
      "learning_rate": 0.00019932333317972028,
      "loss": 0.8596,
      "step": 16090
    },
    {
      "epoch": 0.2826594568022613,
      "grad_norm": 0.05714972287323693,
      "learning_rate": 0.00019932203188344718,
      "loss": 0.859,
      "step": 16100
    },
    {
      "epoch": 0.28283502168226266,
      "grad_norm": 0.055828747395216934,
      "learning_rate": 0.00019932072934141513,
      "loss": 0.8623,
      "step": 16110
    },
    {
      "epoch": 0.2830105865622641,
      "grad_norm": 0.107651825727822,
      "learning_rate": 0.00019931942555364055,
      "loss": 0.8573,
      "step": 16120
    },
    {
      "epoch": 0.2831861514422655,
      "grad_norm": 0.0799038212972373,
      "learning_rate": 0.00019931812052014004,
      "loss": 0.8524,
      "step": 16130
    },
    {
      "epoch": 0.2833617163222669,
      "grad_norm": 0.04896050515286679,
      "learning_rate": 0.0001993168142409301,
      "loss": 0.8573,
      "step": 16140
    },
    {
      "epoch": 0.2835372812022683,
      "grad_norm": 0.07057822790250297,
      "learning_rate": 0.00019931550671602733,
      "loss": 0.8576,
      "step": 16150
    },
    {
      "epoch": 0.2837128460822697,
      "grad_norm": 0.07031964074150342,
      "learning_rate": 0.0001993141979454482,
      "loss": 0.8584,
      "step": 16160
    },
    {
      "epoch": 0.28388841096227113,
      "grad_norm": 0.10147315193914687,
      "learning_rate": 0.0001993128879292093,
      "loss": 0.8567,
      "step": 16170
    },
    {
      "epoch": 0.2840639758422725,
      "grad_norm": 0.07753844648971414,
      "learning_rate": 0.0001993115766673273,
      "loss": 0.8574,
      "step": 16180
    },
    {
      "epoch": 0.2842395407222739,
      "grad_norm": 0.06569563339503433,
      "learning_rate": 0.0001993102641598188,
      "loss": 0.8591,
      "step": 16190
    },
    {
      "epoch": 0.28441510560227534,
      "grad_norm": 0.06536824680205941,
      "learning_rate": 0.0001993089504067004,
      "loss": 0.8569,
      "step": 16200
    },
    {
      "epoch": 0.2845906704822767,
      "grad_norm": 0.05978093871844364,
      "learning_rate": 0.00019930763540798872,
      "loss": 0.8647,
      "step": 16210
    },
    {
      "epoch": 0.28476623536227813,
      "grad_norm": 0.05891315416868927,
      "learning_rate": 0.00019930631916370052,
      "loss": 0.855,
      "step": 16220
    },
    {
      "epoch": 0.28494180024227955,
      "grad_norm": 0.08247141480163506,
      "learning_rate": 0.00019930500167385243,
      "loss": 0.87,
      "step": 16230
    },
    {
      "epoch": 0.2851173651222809,
      "grad_norm": 0.0738842759669392,
      "learning_rate": 0.00019930368293846106,
      "loss": 0.8497,
      "step": 16240
    },
    {
      "epoch": 0.28529293000228234,
      "grad_norm": 0.06795391646433357,
      "learning_rate": 0.00019930236295754322,
      "loss": 0.8611,
      "step": 16250
    },
    {
      "epoch": 0.28546849488228376,
      "grad_norm": 0.06347159406758775,
      "learning_rate": 0.00019930104173111561,
      "loss": 0.8612,
      "step": 16260
    },
    {
      "epoch": 0.28564405976228513,
      "grad_norm": 0.0672827112361949,
      "learning_rate": 0.00019929971925919496,
      "loss": 0.8609,
      "step": 16270
    },
    {
      "epoch": 0.28581962464228655,
      "grad_norm": 0.07063282346389584,
      "learning_rate": 0.00019929839554179803,
      "loss": 0.8665,
      "step": 16280
    },
    {
      "epoch": 0.285995189522288,
      "grad_norm": 0.058121357999863066,
      "learning_rate": 0.00019929707057894154,
      "loss": 0.8535,
      "step": 16290
    },
    {
      "epoch": 0.28617075440228934,
      "grad_norm": 0.062092745301839745,
      "learning_rate": 0.00019929574437064238,
      "loss": 0.8622,
      "step": 16300
    },
    {
      "epoch": 0.28634631928229076,
      "grad_norm": 0.08651549075544836,
      "learning_rate": 0.00019929441691691728,
      "loss": 0.8499,
      "step": 16310
    },
    {
      "epoch": 0.2865218841622922,
      "grad_norm": 0.06277721894575615,
      "learning_rate": 0.0001992930882177831,
      "loss": 0.8589,
      "step": 16320
    },
    {
      "epoch": 0.2866974490422936,
      "grad_norm": 0.07148467120649954,
      "learning_rate": 0.00019929175827325664,
      "loss": 0.8584,
      "step": 16330
    },
    {
      "epoch": 0.286873013922295,
      "grad_norm": 0.06975523540808198,
      "learning_rate": 0.00019929042708335476,
      "loss": 0.8615,
      "step": 16340
    },
    {
      "epoch": 0.2870485788022964,
      "grad_norm": 0.07329949169858321,
      "learning_rate": 0.00019928909464809436,
      "loss": 0.8596,
      "step": 16350
    },
    {
      "epoch": 0.2872241436822978,
      "grad_norm": 0.07214151524210616,
      "learning_rate": 0.00019928776096749226,
      "loss": 0.8644,
      "step": 16360
    },
    {
      "epoch": 0.2873997085622992,
      "grad_norm": 0.06071910447017084,
      "learning_rate": 0.00019928642604156538,
      "loss": 0.8569,
      "step": 16370
    },
    {
      "epoch": 0.2875752734423006,
      "grad_norm": 0.08380390160438374,
      "learning_rate": 0.0001992850898703307,
      "loss": 0.8589,
      "step": 16380
    },
    {
      "epoch": 0.287750838322302,
      "grad_norm": 0.06769896118293908,
      "learning_rate": 0.00019928375245380507,
      "loss": 0.852,
      "step": 16390
    },
    {
      "epoch": 0.2879264032023034,
      "grad_norm": 0.06060509382591061,
      "learning_rate": 0.00019928241379200544,
      "loss": 0.8546,
      "step": 16400
    },
    {
      "epoch": 0.2881019680823048,
      "grad_norm": 0.11592813564545848,
      "learning_rate": 0.00019928107388494882,
      "loss": 0.8555,
      "step": 16410
    },
    {
      "epoch": 0.28827753296230624,
      "grad_norm": 0.06779115149536825,
      "learning_rate": 0.00019927973273265213,
      "loss": 0.8568,
      "step": 16420
    },
    {
      "epoch": 0.2884530978423076,
      "grad_norm": 0.06632098100976402,
      "learning_rate": 0.0001992783903351324,
      "loss": 0.8563,
      "step": 16430
    },
    {
      "epoch": 0.288628662722309,
      "grad_norm": 0.06046294947863675,
      "learning_rate": 0.00019927704669240665,
      "loss": 0.8562,
      "step": 16440
    },
    {
      "epoch": 0.28880422760231045,
      "grad_norm": 0.06413339569445198,
      "learning_rate": 0.00019927570180449186,
      "loss": 0.8667,
      "step": 16450
    },
    {
      "epoch": 0.2889797924823118,
      "grad_norm": 0.06722794680718401,
      "learning_rate": 0.0001992743556714051,
      "loss": 0.8583,
      "step": 16460
    },
    {
      "epoch": 0.28915535736231324,
      "grad_norm": 0.06132257937348581,
      "learning_rate": 0.00019927300829316344,
      "loss": 0.8631,
      "step": 16470
    },
    {
      "epoch": 0.28933092224231466,
      "grad_norm": 0.07916561129219288,
      "learning_rate": 0.0001992716596697839,
      "loss": 0.8673,
      "step": 16480
    },
    {
      "epoch": 0.289506487122316,
      "grad_norm": 0.06322339176536311,
      "learning_rate": 0.0001992703098012836,
      "loss": 0.8655,
      "step": 16490
    },
    {
      "epoch": 0.28968205200231745,
      "grad_norm": 0.06698421710085138,
      "learning_rate": 0.00019926895868767967,
      "loss": 0.8627,
      "step": 16500
    },
    {
      "epoch": 0.28985761688231887,
      "grad_norm": 0.08615878226914123,
      "learning_rate": 0.00019926760632898917,
      "loss": 0.8657,
      "step": 16510
    },
    {
      "epoch": 0.2900331817623203,
      "grad_norm": 0.059103537065474984,
      "learning_rate": 0.0001992662527252293,
      "loss": 0.8662,
      "step": 16520
    },
    {
      "epoch": 0.29020874664232166,
      "grad_norm": 0.0647505348110548,
      "learning_rate": 0.00019926489787641715,
      "loss": 0.8605,
      "step": 16530
    },
    {
      "epoch": 0.2903843115223231,
      "grad_norm": 0.0659775997450497,
      "learning_rate": 0.0001992635417825699,
      "loss": 0.8534,
      "step": 16540
    },
    {
      "epoch": 0.2905598764023245,
      "grad_norm": 0.06431457970931055,
      "learning_rate": 0.00019926218444370477,
      "loss": 0.8607,
      "step": 16550
    },
    {
      "epoch": 0.29073544128232587,
      "grad_norm": 0.06582377814753178,
      "learning_rate": 0.00019926082585983892,
      "loss": 0.8652,
      "step": 16560
    },
    {
      "epoch": 0.2909110061623273,
      "grad_norm": 0.09443996030285609,
      "learning_rate": 0.0001992594660309896,
      "loss": 0.8589,
      "step": 16570
    },
    {
      "epoch": 0.2910865710423287,
      "grad_norm": 0.04913185577601539,
      "learning_rate": 0.00019925810495717396,
      "loss": 0.8622,
      "step": 16580
    },
    {
      "epoch": 0.2912621359223301,
      "grad_norm": 0.09867460473950097,
      "learning_rate": 0.00019925674263840932,
      "loss": 0.8544,
      "step": 16590
    },
    {
      "epoch": 0.2914377008023315,
      "grad_norm": 0.0577415072302105,
      "learning_rate": 0.00019925537907471293,
      "loss": 0.8579,
      "step": 16600
    },
    {
      "epoch": 0.2916132656823329,
      "grad_norm": 0.07736453174061995,
      "learning_rate": 0.00019925401426610205,
      "loss": 0.8627,
      "step": 16610
    },
    {
      "epoch": 0.2917888305623343,
      "grad_norm": 0.07465201078768827,
      "learning_rate": 0.000199252648212594,
      "loss": 0.8567,
      "step": 16620
    },
    {
      "epoch": 0.2919643954423357,
      "grad_norm": 0.07763032841320909,
      "learning_rate": 0.00019925128091420607,
      "loss": 0.8588,
      "step": 16630
    },
    {
      "epoch": 0.29213996032233713,
      "grad_norm": 0.08377313427211736,
      "learning_rate": 0.00019924991237095554,
      "loss": 0.8637,
      "step": 16640
    },
    {
      "epoch": 0.2923155252023385,
      "grad_norm": 0.06547195441916155,
      "learning_rate": 0.0001992485425828598,
      "loss": 0.8529,
      "step": 16650
    },
    {
      "epoch": 0.2924910900823399,
      "grad_norm": 0.06699176033960474,
      "learning_rate": 0.0001992471715499362,
      "loss": 0.8572,
      "step": 16660
    },
    {
      "epoch": 0.29266665496234134,
      "grad_norm": 0.06947559305268568,
      "learning_rate": 0.00019924579927220213,
      "loss": 0.8589,
      "step": 16670
    },
    {
      "epoch": 0.2928422198423427,
      "grad_norm": 0.06478367052411628,
      "learning_rate": 0.00019924442574967491,
      "loss": 0.8602,
      "step": 16680
    },
    {
      "epoch": 0.29301778472234413,
      "grad_norm": 0.07899488728921693,
      "learning_rate": 0.00019924305098237207,
      "loss": 0.8638,
      "step": 16690
    },
    {
      "epoch": 0.29319334960234555,
      "grad_norm": 0.0704216527676942,
      "learning_rate": 0.00019924167497031088,
      "loss": 0.8601,
      "step": 16700
    },
    {
      "epoch": 0.293368914482347,
      "grad_norm": 0.06696349547699972,
      "learning_rate": 0.00019924029771350884,
      "loss": 0.8495,
      "step": 16710
    },
    {
      "epoch": 0.29354447936234834,
      "grad_norm": 0.06861727127727847,
      "learning_rate": 0.0001992389192119834,
      "loss": 0.8667,
      "step": 16720
    },
    {
      "epoch": 0.29372004424234976,
      "grad_norm": 0.09100653536155183,
      "learning_rate": 0.00019923753946575205,
      "loss": 0.8613,
      "step": 16730
    },
    {
      "epoch": 0.2938956091223512,
      "grad_norm": 0.08516117245094779,
      "learning_rate": 0.00019923615847483226,
      "loss": 0.8548,
      "step": 16740
    },
    {
      "epoch": 0.29407117400235255,
      "grad_norm": 0.08357640798814285,
      "learning_rate": 0.00019923477623924146,
      "loss": 0.8479,
      "step": 16750
    },
    {
      "epoch": 0.294246738882354,
      "grad_norm": 0.09654185042046867,
      "learning_rate": 0.00019923339275899725,
      "loss": 0.8557,
      "step": 16760
    },
    {
      "epoch": 0.2944223037623554,
      "grad_norm": 0.06481294399969062,
      "learning_rate": 0.00019923200803411716,
      "loss": 0.8598,
      "step": 16770
    },
    {
      "epoch": 0.29459786864235676,
      "grad_norm": 0.08702100131877259,
      "learning_rate": 0.00019923062206461868,
      "loss": 0.859,
      "step": 16780
    },
    {
      "epoch": 0.2947734335223582,
      "grad_norm": 0.07875094923759217,
      "learning_rate": 0.00019922923485051936,
      "loss": 0.8569,
      "step": 16790
    },
    {
      "epoch": 0.2949489984023596,
      "grad_norm": 0.07803535413144201,
      "learning_rate": 0.00019922784639183683,
      "loss": 0.8574,
      "step": 16800
    },
    {
      "epoch": 0.295124563282361,
      "grad_norm": 0.08663842115177851,
      "learning_rate": 0.00019922645668858866,
      "loss": 0.851,
      "step": 16810
    },
    {
      "epoch": 0.2953001281623624,
      "grad_norm": 0.06155337318504414,
      "learning_rate": 0.00019922506574079245,
      "loss": 0.8577,
      "step": 16820
    },
    {
      "epoch": 0.2954756930423638,
      "grad_norm": 0.07154136568963751,
      "learning_rate": 0.00019922367354846586,
      "loss": 0.8507,
      "step": 16830
    },
    {
      "epoch": 0.2956512579223652,
      "grad_norm": 0.06545651725789856,
      "learning_rate": 0.00019922228011162648,
      "loss": 0.8514,
      "step": 16840
    },
    {
      "epoch": 0.2958268228023666,
      "grad_norm": 0.08859357522122482,
      "learning_rate": 0.000199220885430292,
      "loss": 0.8692,
      "step": 16850
    },
    {
      "epoch": 0.29600238768236803,
      "grad_norm": 0.07397315354605344,
      "learning_rate": 0.00019921948950448007,
      "loss": 0.8639,
      "step": 16860
    },
    {
      "epoch": 0.29617795256236945,
      "grad_norm": 0.06291299047675504,
      "learning_rate": 0.0001992180923342084,
      "loss": 0.8586,
      "step": 16870
    },
    {
      "epoch": 0.2963535174423708,
      "grad_norm": 0.07169404656720314,
      "learning_rate": 0.00019921669391949467,
      "loss": 0.8639,
      "step": 16880
    },
    {
      "epoch": 0.29652908232237224,
      "grad_norm": 0.0744899877435283,
      "learning_rate": 0.0001992152942603566,
      "loss": 0.845,
      "step": 16890
    },
    {
      "epoch": 0.29670464720237366,
      "grad_norm": 0.06687562088109748,
      "learning_rate": 0.0001992138933568119,
      "loss": 0.8582,
      "step": 16900
    },
    {
      "epoch": 0.29688021208237503,
      "grad_norm": 0.07270881120810614,
      "learning_rate": 0.00019921249120887839,
      "loss": 0.8524,
      "step": 16910
    },
    {
      "epoch": 0.29705577696237645,
      "grad_norm": 0.0678215044805766,
      "learning_rate": 0.00019921108781657376,
      "loss": 0.8636,
      "step": 16920
    },
    {
      "epoch": 0.29723134184237787,
      "grad_norm": 0.08610893492573472,
      "learning_rate": 0.00019920968317991587,
      "loss": 0.8653,
      "step": 16930
    },
    {
      "epoch": 0.29740690672237924,
      "grad_norm": 0.06309889571476862,
      "learning_rate": 0.0001992082772989224,
      "loss": 0.8644,
      "step": 16940
    },
    {
      "epoch": 0.29758247160238066,
      "grad_norm": 0.05470421567029037,
      "learning_rate": 0.0001992068701736113,
      "loss": 0.8593,
      "step": 16950
    },
    {
      "epoch": 0.2977580364823821,
      "grad_norm": 0.04860957472320607,
      "learning_rate": 0.00019920546180400033,
      "loss": 0.8564,
      "step": 16960
    },
    {
      "epoch": 0.29793360136238345,
      "grad_norm": 0.0702526392554524,
      "learning_rate": 0.0001992040521901073,
      "loss": 0.8558,
      "step": 16970
    },
    {
      "epoch": 0.29810916624238487,
      "grad_norm": 0.08985493456627004,
      "learning_rate": 0.00019920264133195013,
      "loss": 0.8532,
      "step": 16980
    },
    {
      "epoch": 0.2982847311223863,
      "grad_norm": 0.06304539931811852,
      "learning_rate": 0.0001992012292295467,
      "loss": 0.8592,
      "step": 16990
    },
    {
      "epoch": 0.29846029600238766,
      "grad_norm": 0.09218348123781539,
      "learning_rate": 0.00019919981588291486,
      "loss": 0.8501,
      "step": 17000
    },
    {
      "epoch": 0.2986358608823891,
      "grad_norm": 0.09462626201981794,
      "learning_rate": 0.0001991984012920725,
      "loss": 0.8673,
      "step": 17010
    },
    {
      "epoch": 0.2988114257623905,
      "grad_norm": 0.11136589147732778,
      "learning_rate": 0.00019919698545703762,
      "loss": 0.8671,
      "step": 17020
    },
    {
      "epoch": 0.29898699064239187,
      "grad_norm": 0.07055317378741263,
      "learning_rate": 0.00019919556837782806,
      "loss": 0.8658,
      "step": 17030
    },
    {
      "epoch": 0.2991625555223933,
      "grad_norm": 0.06353913678846304,
      "learning_rate": 0.0001991941500544619,
      "loss": 0.8591,
      "step": 17040
    },
    {
      "epoch": 0.2993381204023947,
      "grad_norm": 0.08783528045065599,
      "learning_rate": 0.000199192730486957,
      "loss": 0.8613,
      "step": 17050
    },
    {
      "epoch": 0.29951368528239614,
      "grad_norm": 0.08438569670921732,
      "learning_rate": 0.00019919130967533134,
      "loss": 0.8528,
      "step": 17060
    },
    {
      "epoch": 0.2996892501623975,
      "grad_norm": 0.049188022654736624,
      "learning_rate": 0.00019918988761960304,
      "loss": 0.8613,
      "step": 17070
    },
    {
      "epoch": 0.2998648150423989,
      "grad_norm": 0.06887747059256775,
      "learning_rate": 0.00019918846431979005,
      "loss": 0.8596,
      "step": 17080
    },
    {
      "epoch": 0.30004037992240035,
      "grad_norm": 0.06696502181644078,
      "learning_rate": 0.00019918703977591038,
      "loss": 0.8536,
      "step": 17090
    },
    {
      "epoch": 0.3002159448024017,
      "grad_norm": 0.07167892019821537,
      "learning_rate": 0.0001991856139879821,
      "loss": 0.854,
      "step": 17100
    },
    {
      "epoch": 0.30039150968240313,
      "grad_norm": 0.07750390849353687,
      "learning_rate": 0.00019918418695602326,
      "loss": 0.8554,
      "step": 17110
    },
    {
      "epoch": 0.30056707456240456,
      "grad_norm": 0.0668620297136799,
      "learning_rate": 0.00019918275868005195,
      "loss": 0.8585,
      "step": 17120
    },
    {
      "epoch": 0.3007426394424059,
      "grad_norm": 0.052513058674325125,
      "learning_rate": 0.00019918132916008628,
      "loss": 0.8551,
      "step": 17130
    },
    {
      "epoch": 0.30091820432240735,
      "grad_norm": 0.08099888815070892,
      "learning_rate": 0.00019917989839614435,
      "loss": 0.8733,
      "step": 17140
    },
    {
      "epoch": 0.30109376920240877,
      "grad_norm": 0.05718829265862999,
      "learning_rate": 0.00019917846638824426,
      "loss": 0.8623,
      "step": 17150
    },
    {
      "epoch": 0.30126933408241013,
      "grad_norm": 0.0812392235842015,
      "learning_rate": 0.00019917703313640425,
      "loss": 0.8498,
      "step": 17160
    },
    {
      "epoch": 0.30144489896241156,
      "grad_norm": 0.08263407315827623,
      "learning_rate": 0.00019917559864064238,
      "loss": 0.8556,
      "step": 17170
    },
    {
      "epoch": 0.301620463842413,
      "grad_norm": 0.07967774944594185,
      "learning_rate": 0.00019917416290097685,
      "loss": 0.8657,
      "step": 17180
    },
    {
      "epoch": 0.30179602872241434,
      "grad_norm": 0.06284616202748283,
      "learning_rate": 0.00019917272591742587,
      "loss": 0.8598,
      "step": 17190
    },
    {
      "epoch": 0.30197159360241577,
      "grad_norm": 0.10478372134480675,
      "learning_rate": 0.0001991712876900077,
      "loss": 0.8608,
      "step": 17200
    },
    {
      "epoch": 0.3021471584824172,
      "grad_norm": 0.06489720637914502,
      "learning_rate": 0.0001991698482187404,
      "loss": 0.8619,
      "step": 17210
    },
    {
      "epoch": 0.30232272336241856,
      "grad_norm": 0.07534906959035347,
      "learning_rate": 0.00019916840750364235,
      "loss": 0.8638,
      "step": 17220
    },
    {
      "epoch": 0.30249828824242,
      "grad_norm": 0.11156549485828265,
      "learning_rate": 0.00019916696554473176,
      "loss": 0.849,
      "step": 17230
    },
    {
      "epoch": 0.3026738531224214,
      "grad_norm": 0.10062595698844387,
      "learning_rate": 0.0001991655223420269,
      "loss": 0.8525,
      "step": 17240
    },
    {
      "epoch": 0.3028494180024228,
      "grad_norm": 0.07823919621851963,
      "learning_rate": 0.00019916407789554607,
      "loss": 0.8562,
      "step": 17250
    },
    {
      "epoch": 0.3030249828824242,
      "grad_norm": 0.08952754681649663,
      "learning_rate": 0.00019916263220530754,
      "loss": 0.8575,
      "step": 17260
    },
    {
      "epoch": 0.3032005477624256,
      "grad_norm": 0.07502104115937305,
      "learning_rate": 0.00019916118527132966,
      "loss": 0.8625,
      "step": 17270
    },
    {
      "epoch": 0.30337611264242703,
      "grad_norm": 0.06532144739713754,
      "learning_rate": 0.00019915973709363078,
      "loss": 0.8509,
      "step": 17280
    },
    {
      "epoch": 0.3035516775224284,
      "grad_norm": 0.06847630610909773,
      "learning_rate": 0.0001991582876722292,
      "loss": 0.8577,
      "step": 17290
    },
    {
      "epoch": 0.3037272424024298,
      "grad_norm": 0.07179995021054224,
      "learning_rate": 0.00019915683700714328,
      "loss": 0.8547,
      "step": 17300
    },
    {
      "epoch": 0.30390280728243124,
      "grad_norm": 0.08723655881858587,
      "learning_rate": 0.00019915538509839148,
      "loss": 0.8524,
      "step": 17310
    },
    {
      "epoch": 0.3040783721624326,
      "grad_norm": 0.07555707490277051,
      "learning_rate": 0.0001991539319459921,
      "loss": 0.8668,
      "step": 17320
    },
    {
      "epoch": 0.30425393704243403,
      "grad_norm": 0.06800552983974828,
      "learning_rate": 0.00019915247754996357,
      "loss": 0.8519,
      "step": 17330
    },
    {
      "epoch": 0.30442950192243545,
      "grad_norm": 0.059510888652118094,
      "learning_rate": 0.00019915102191032436,
      "loss": 0.849,
      "step": 17340
    },
    {
      "epoch": 0.3046050668024368,
      "grad_norm": 0.0658050726810535,
      "learning_rate": 0.0001991495650270929,
      "loss": 0.8503,
      "step": 17350
    },
    {
      "epoch": 0.30478063168243824,
      "grad_norm": 0.1026816509322282,
      "learning_rate": 0.0001991481069002876,
      "loss": 0.857,
      "step": 17360
    },
    {
      "epoch": 0.30495619656243966,
      "grad_norm": 0.059308694420321384,
      "learning_rate": 0.000199146647529927,
      "loss": 0.8588,
      "step": 17370
    },
    {
      "epoch": 0.30513176144244103,
      "grad_norm": 0.09572537693407832,
      "learning_rate": 0.0001991451869160296,
      "loss": 0.8623,
      "step": 17380
    },
    {
      "epoch": 0.30530732632244245,
      "grad_norm": 0.07795378892056264,
      "learning_rate": 0.00019914372505861386,
      "loss": 0.8653,
      "step": 17390
    },
    {
      "epoch": 0.3054828912024439,
      "grad_norm": 0.05674074853667246,
      "learning_rate": 0.0001991422619576983,
      "loss": 0.8601,
      "step": 17400
    },
    {
      "epoch": 0.30565845608244524,
      "grad_norm": 0.07886007423232559,
      "learning_rate": 0.00019914079761330146,
      "loss": 0.8608,
      "step": 17410
    },
    {
      "epoch": 0.30583402096244666,
      "grad_norm": 0.060614888083388524,
      "learning_rate": 0.0001991393320254419,
      "loss": 0.8571,
      "step": 17420
    },
    {
      "epoch": 0.3060095858424481,
      "grad_norm": 0.06769047754781277,
      "learning_rate": 0.00019913786519413823,
      "loss": 0.8597,
      "step": 17430
    },
    {
      "epoch": 0.3061851507224495,
      "grad_norm": 0.10008464608534992,
      "learning_rate": 0.00019913639711940898,
      "loss": 0.8618,
      "step": 17440
    },
    {
      "epoch": 0.3063607156024509,
      "grad_norm": 0.10026201757117098,
      "learning_rate": 0.00019913492780127278,
      "loss": 0.8491,
      "step": 17450
    },
    {
      "epoch": 0.3065362804824523,
      "grad_norm": 0.10362134221346893,
      "learning_rate": 0.0001991334572397482,
      "loss": 0.8538,
      "step": 17460
    },
    {
      "epoch": 0.3067118453624537,
      "grad_norm": 0.05478538140852639,
      "learning_rate": 0.00019913198543485396,
      "loss": 0.8608,
      "step": 17470
    },
    {
      "epoch": 0.3068874102424551,
      "grad_norm": 0.06509852282065093,
      "learning_rate": 0.00019913051238660865,
      "loss": 0.86,
      "step": 17480
    },
    {
      "epoch": 0.3070629751224565,
      "grad_norm": 0.058674758172782095,
      "learning_rate": 0.00019912903809503094,
      "loss": 0.8526,
      "step": 17490
    },
    {
      "epoch": 0.3072385400024579,
      "grad_norm": 0.05536564637052904,
      "learning_rate": 0.00019912756256013946,
      "loss": 0.8603,
      "step": 17500
    },
    {
      "epoch": 0.3074141048824593,
      "grad_norm": 0.07654518144074346,
      "learning_rate": 0.000199126085781953,
      "loss": 0.8574,
      "step": 17510
    },
    {
      "epoch": 0.3075896697624607,
      "grad_norm": 0.06509695118041121,
      "learning_rate": 0.00019912460776049025,
      "loss": 0.8568,
      "step": 17520
    },
    {
      "epoch": 0.30776523464246214,
      "grad_norm": 0.06748571986621985,
      "learning_rate": 0.0001991231284957699,
      "loss": 0.8613,
      "step": 17530
    },
    {
      "epoch": 0.3079407995224635,
      "grad_norm": 0.07202895615321345,
      "learning_rate": 0.00019912164798781067,
      "loss": 0.8696,
      "step": 17540
    },
    {
      "epoch": 0.3081163644024649,
      "grad_norm": 0.08791907986656636,
      "learning_rate": 0.0001991201662366314,
      "loss": 0.8609,
      "step": 17550
    },
    {
      "epoch": 0.30829192928246635,
      "grad_norm": 0.08585888626608079,
      "learning_rate": 0.00019911868324225078,
      "loss": 0.8602,
      "step": 17560
    },
    {
      "epoch": 0.3084674941624677,
      "grad_norm": 0.06459070830729048,
      "learning_rate": 0.00019911719900468766,
      "loss": 0.8509,
      "step": 17570
    },
    {
      "epoch": 0.30864305904246914,
      "grad_norm": 0.06815495059520948,
      "learning_rate": 0.00019911571352396084,
      "loss": 0.8577,
      "step": 17580
    },
    {
      "epoch": 0.30881862392247056,
      "grad_norm": 0.06269367507707141,
      "learning_rate": 0.0001991142268000891,
      "loss": 0.854,
      "step": 17590
    },
    {
      "epoch": 0.308994188802472,
      "grad_norm": 0.07107986896930625,
      "learning_rate": 0.0001991127388330913,
      "loss": 0.8622,
      "step": 17600
    },
    {
      "epoch": 0.30916975368247335,
      "grad_norm": 0.10679766494016368,
      "learning_rate": 0.00019911124962298632,
      "loss": 0.8645,
      "step": 17610
    },
    {
      "epoch": 0.30934531856247477,
      "grad_norm": 0.07042292576948937,
      "learning_rate": 0.000199109759169793,
      "loss": 0.8527,
      "step": 17620
    },
    {
      "epoch": 0.3095208834424762,
      "grad_norm": 0.06506002728383325,
      "learning_rate": 0.0001991082674735302,
      "loss": 0.8634,
      "step": 17630
    },
    {
      "epoch": 0.30969644832247756,
      "grad_norm": 0.08377054110816094,
      "learning_rate": 0.00019910677453421687,
      "loss": 0.86,
      "step": 17640
    },
    {
      "epoch": 0.309872013202479,
      "grad_norm": 0.05981415279865544,
      "learning_rate": 0.0001991052803518719,
      "loss": 0.859,
      "step": 17650
    },
    {
      "epoch": 0.3100475780824804,
      "grad_norm": 0.07425654299348483,
      "learning_rate": 0.00019910378492651423,
      "loss": 0.8516,
      "step": 17660
    },
    {
      "epoch": 0.31022314296248177,
      "grad_norm": 0.08713602233396292,
      "learning_rate": 0.0001991022882581628,
      "loss": 0.8534,
      "step": 17670
    },
    {
      "epoch": 0.3103987078424832,
      "grad_norm": 0.10221955303412454,
      "learning_rate": 0.0001991007903468366,
      "loss": 0.8561,
      "step": 17680
    },
    {
      "epoch": 0.3105742727224846,
      "grad_norm": 0.0696522437510081,
      "learning_rate": 0.00019909929119255455,
      "loss": 0.8634,
      "step": 17690
    },
    {
      "epoch": 0.310749837602486,
      "grad_norm": 0.06316718621017117,
      "learning_rate": 0.00019909779079533568,
      "loss": 0.8625,
      "step": 17700
    },
    {
      "epoch": 0.3109254024824874,
      "grad_norm": 0.06028646350881662,
      "learning_rate": 0.00019909628915519903,
      "loss": 0.8566,
      "step": 17710
    },
    {
      "epoch": 0.3111009673624888,
      "grad_norm": 0.07510277518592742,
      "learning_rate": 0.0001990947862721636,
      "loss": 0.8627,
      "step": 17720
    },
    {
      "epoch": 0.3112765322424902,
      "grad_norm": 0.07040723154333153,
      "learning_rate": 0.0001990932821462484,
      "loss": 0.8586,
      "step": 17730
    },
    {
      "epoch": 0.3114520971224916,
      "grad_norm": 0.07085258611017076,
      "learning_rate": 0.00019909177677747254,
      "loss": 0.8504,
      "step": 17740
    },
    {
      "epoch": 0.31162766200249303,
      "grad_norm": 0.07237573593719533,
      "learning_rate": 0.000199090270165855,
      "loss": 0.8566,
      "step": 17750
    },
    {
      "epoch": 0.3118032268824944,
      "grad_norm": 0.08416570120820809,
      "learning_rate": 0.00019908876231141504,
      "loss": 0.8522,
      "step": 17760
    },
    {
      "epoch": 0.3119787917624958,
      "grad_norm": 0.07602909786548825,
      "learning_rate": 0.0001990872532141716,
      "loss": 0.8503,
      "step": 17770
    },
    {
      "epoch": 0.31215435664249724,
      "grad_norm": 0.11361552170769419,
      "learning_rate": 0.00019908574287414387,
      "loss": 0.8622,
      "step": 17780
    },
    {
      "epoch": 0.31232992152249867,
      "grad_norm": 0.05732009291944238,
      "learning_rate": 0.00019908423129135101,
      "loss": 0.8592,
      "step": 17790
    },
    {
      "epoch": 0.31250548640250003,
      "grad_norm": 0.07676230019903203,
      "learning_rate": 0.0001990827184658121,
      "loss": 0.867,
      "step": 17800
    },
    {
      "epoch": 0.31268105128250145,
      "grad_norm": 0.11433071449743344,
      "learning_rate": 0.00019908120439754637,
      "loss": 0.8577,
      "step": 17810
    },
    {
      "epoch": 0.3128566161625029,
      "grad_norm": 0.07224020343558185,
      "learning_rate": 0.000199079689086573,
      "loss": 0.8635,
      "step": 17820
    },
    {
      "epoch": 0.31303218104250424,
      "grad_norm": 0.06299465391345403,
      "learning_rate": 0.00019907817253291116,
      "loss": 0.8575,
      "step": 17830
    },
    {
      "epoch": 0.31320774592250566,
      "grad_norm": 0.0862599795849425,
      "learning_rate": 0.00019907665473658005,
      "loss": 0.8557,
      "step": 17840
    },
    {
      "epoch": 0.3133833108025071,
      "grad_norm": 0.06093418427124292,
      "learning_rate": 0.00019907513569759896,
      "loss": 0.8508,
      "step": 17850
    },
    {
      "epoch": 0.31355887568250845,
      "grad_norm": 0.05569976847357369,
      "learning_rate": 0.0001990736154159871,
      "loss": 0.8612,
      "step": 17860
    },
    {
      "epoch": 0.3137344405625099,
      "grad_norm": 0.07777719209226125,
      "learning_rate": 0.00019907209389176373,
      "loss": 0.8552,
      "step": 17870
    },
    {
      "epoch": 0.3139100054425113,
      "grad_norm": 0.11126040938286483,
      "learning_rate": 0.00019907057112494812,
      "loss": 0.8563,
      "step": 17880
    },
    {
      "epoch": 0.31408557032251266,
      "grad_norm": 0.07111194458004376,
      "learning_rate": 0.00019906904711555962,
      "loss": 0.859,
      "step": 17890
    },
    {
      "epoch": 0.3142611352025141,
      "grad_norm": 0.06964000774004818,
      "learning_rate": 0.0001990675218636175,
      "loss": 0.8602,
      "step": 17900
    },
    {
      "epoch": 0.3144367000825155,
      "grad_norm": 0.07183807291149802,
      "learning_rate": 0.00019906599536914104,
      "loss": 0.855,
      "step": 17910
    },
    {
      "epoch": 0.3146122649625169,
      "grad_norm": 0.06799537300445023,
      "learning_rate": 0.00019906446763214963,
      "loss": 0.8601,
      "step": 17920
    },
    {
      "epoch": 0.3147878298425183,
      "grad_norm": 0.05767954635122441,
      "learning_rate": 0.00019906293865266264,
      "loss": 0.8646,
      "step": 17930
    },
    {
      "epoch": 0.3149633947225197,
      "grad_norm": 0.06938395262092172,
      "learning_rate": 0.00019906140843069941,
      "loss": 0.8513,
      "step": 17940
    },
    {
      "epoch": 0.3151389596025211,
      "grad_norm": 0.07486366767299105,
      "learning_rate": 0.00019905987696627935,
      "loss": 0.859,
      "step": 17950
    },
    {
      "epoch": 0.3153145244825225,
      "grad_norm": 0.05086325506096764,
      "learning_rate": 0.00019905834425942184,
      "loss": 0.8601,
      "step": 17960
    },
    {
      "epoch": 0.31549008936252393,
      "grad_norm": 0.06417470099364238,
      "learning_rate": 0.00019905681031014635,
      "loss": 0.8596,
      "step": 17970
    },
    {
      "epoch": 0.31566565424252535,
      "grad_norm": 0.04795626461432263,
      "learning_rate": 0.00019905527511847226,
      "loss": 0.8547,
      "step": 17980
    },
    {
      "epoch": 0.3158412191225267,
      "grad_norm": 0.06989275421344479,
      "learning_rate": 0.00019905373868441906,
      "loss": 0.8629,
      "step": 17990
    },
    {
      "epoch": 0.31601678400252814,
      "grad_norm": 0.061028882691268965,
      "learning_rate": 0.00019905220100800616,
      "loss": 0.8486,
      "step": 18000
    },
    {
      "epoch": 0.31619234888252956,
      "grad_norm": 0.06198511050375044,
      "learning_rate": 0.0001990506620892531,
      "loss": 0.8643,
      "step": 18010
    },
    {
      "epoch": 0.31636791376253093,
      "grad_norm": 0.06156197078219974,
      "learning_rate": 0.00019904912192817938,
      "loss": 0.86,
      "step": 18020
    },
    {
      "epoch": 0.31654347864253235,
      "grad_norm": 0.1182654345658973,
      "learning_rate": 0.00019904758052480446,
      "loss": 0.856,
      "step": 18030
    },
    {
      "epoch": 0.31671904352253377,
      "grad_norm": 0.07426350090433857,
      "learning_rate": 0.00019904603787914793,
      "loss": 0.8577,
      "step": 18040
    },
    {
      "epoch": 0.31689460840253514,
      "grad_norm": 0.059168244742340954,
      "learning_rate": 0.0001990444939912293,
      "loss": 0.8699,
      "step": 18050
    },
    {
      "epoch": 0.31707017328253656,
      "grad_norm": 0.07263772935116553,
      "learning_rate": 0.00019904294886106814,
      "loss": 0.856,
      "step": 18060
    },
    {
      "epoch": 0.317245738162538,
      "grad_norm": 0.07141097643541501,
      "learning_rate": 0.00019904140248868404,
      "loss": 0.858,
      "step": 18070
    },
    {
      "epoch": 0.31742130304253935,
      "grad_norm": 0.05894409185367452,
      "learning_rate": 0.00019903985487409656,
      "loss": 0.8541,
      "step": 18080
    },
    {
      "epoch": 0.31759686792254077,
      "grad_norm": 0.05895798345974111,
      "learning_rate": 0.00019903830601732537,
      "loss": 0.852,
      "step": 18090
    },
    {
      "epoch": 0.3177724328025422,
      "grad_norm": 0.08563914289492862,
      "learning_rate": 0.00019903675591838999,
      "loss": 0.858,
      "step": 18100
    },
    {
      "epoch": 0.31794799768254356,
      "grad_norm": 0.07373284071021617,
      "learning_rate": 0.00019903520457731016,
      "loss": 0.8574,
      "step": 18110
    },
    {
      "epoch": 0.318123562562545,
      "grad_norm": 0.058213175727576075,
      "learning_rate": 0.0001990336519941055,
      "loss": 0.8627,
      "step": 18120
    },
    {
      "epoch": 0.3182991274425464,
      "grad_norm": 0.054839151913072176,
      "learning_rate": 0.00019903209816879566,
      "loss": 0.8596,
      "step": 18130
    },
    {
      "epoch": 0.31847469232254777,
      "grad_norm": 0.05646908482736233,
      "learning_rate": 0.00019903054310140038,
      "loss": 0.8623,
      "step": 18140
    },
    {
      "epoch": 0.3186502572025492,
      "grad_norm": 0.06568910156076978,
      "learning_rate": 0.00019902898679193933,
      "loss": 0.8627,
      "step": 18150
    },
    {
      "epoch": 0.3188258220825506,
      "grad_norm": 0.06944476454711082,
      "learning_rate": 0.00019902742924043224,
      "loss": 0.8668,
      "step": 18160
    },
    {
      "epoch": 0.31900138696255204,
      "grad_norm": 0.0827206568014568,
      "learning_rate": 0.00019902587044689878,
      "loss": 0.8664,
      "step": 18170
    },
    {
      "epoch": 0.3191769518425534,
      "grad_norm": 0.0593173034798514,
      "learning_rate": 0.00019902431041135882,
      "loss": 0.8599,
      "step": 18180
    },
    {
      "epoch": 0.3193525167225548,
      "grad_norm": 0.0680994657334782,
      "learning_rate": 0.00019902274913383204,
      "loss": 0.8575,
      "step": 18190
    },
    {
      "epoch": 0.31952808160255625,
      "grad_norm": 0.09619008431631026,
      "learning_rate": 0.00019902118661433823,
      "loss": 0.8597,
      "step": 18200
    },
    {
      "epoch": 0.3197036464825576,
      "grad_norm": 0.07539639264146879,
      "learning_rate": 0.0001990196228528972,
      "loss": 0.8551,
      "step": 18210
    },
    {
      "epoch": 0.31987921136255903,
      "grad_norm": 0.06975892124818792,
      "learning_rate": 0.00019901805784952877,
      "loss": 0.8602,
      "step": 18220
    },
    {
      "epoch": 0.32005477624256046,
      "grad_norm": 0.07999303788266972,
      "learning_rate": 0.00019901649160425274,
      "loss": 0.8567,
      "step": 18230
    },
    {
      "epoch": 0.3202303411225618,
      "grad_norm": 0.06975631504386215,
      "learning_rate": 0.00019901492411708897,
      "loss": 0.8554,
      "step": 18240
    },
    {
      "epoch": 0.32040590600256325,
      "grad_norm": 0.059367621579359084,
      "learning_rate": 0.0001990133553880574,
      "loss": 0.8576,
      "step": 18250
    },
    {
      "epoch": 0.32058147088256467,
      "grad_norm": 0.09477396874749154,
      "learning_rate": 0.00019901178541717774,
      "loss": 0.8558,
      "step": 18260
    },
    {
      "epoch": 0.32075703576256603,
      "grad_norm": 0.05565276169402761,
      "learning_rate": 0.00019901021420447003,
      "loss": 0.855,
      "step": 18270
    },
    {
      "epoch": 0.32093260064256746,
      "grad_norm": 0.07134908873078358,
      "learning_rate": 0.0001990086417499541,
      "loss": 0.8544,
      "step": 18280
    },
    {
      "epoch": 0.3211081655225689,
      "grad_norm": 0.06623312429410394,
      "learning_rate": 0.00019900706805364991,
      "loss": 0.853,
      "step": 18290
    },
    {
      "epoch": 0.32128373040257024,
      "grad_norm": 0.0761714554282892,
      "learning_rate": 0.00019900549311557738,
      "loss": 0.8639,
      "step": 18300
    },
    {
      "epoch": 0.32145929528257167,
      "grad_norm": 0.07523319931073931,
      "learning_rate": 0.00019900391693575647,
      "loss": 0.8564,
      "step": 18310
    },
    {
      "epoch": 0.3216348601625731,
      "grad_norm": 0.08698971223671158,
      "learning_rate": 0.00019900233951420716,
      "loss": 0.8529,
      "step": 18320
    },
    {
      "epoch": 0.3218104250425745,
      "grad_norm": 0.06765231040231456,
      "learning_rate": 0.00019900076085094942,
      "loss": 0.8584,
      "step": 18330
    },
    {
      "epoch": 0.3219859899225759,
      "grad_norm": 0.06027251232694273,
      "learning_rate": 0.00019899918094600324,
      "loss": 0.8656,
      "step": 18340
    },
    {
      "epoch": 0.3221615548025773,
      "grad_norm": 0.08446034534489896,
      "learning_rate": 0.0001989975997993887,
      "loss": 0.8601,
      "step": 18350
    },
    {
      "epoch": 0.3223371196825787,
      "grad_norm": 0.06157426466354918,
      "learning_rate": 0.00019899601741112574,
      "loss": 0.8596,
      "step": 18360
    },
    {
      "epoch": 0.3225126845625801,
      "grad_norm": 0.06931464666964303,
      "learning_rate": 0.0001989944337812345,
      "loss": 0.8585,
      "step": 18370
    },
    {
      "epoch": 0.3226882494425815,
      "grad_norm": 0.058178952835094506,
      "learning_rate": 0.00019899284890973498,
      "loss": 0.8492,
      "step": 18380
    },
    {
      "epoch": 0.32286381432258293,
      "grad_norm": 0.06517554404899108,
      "learning_rate": 0.00019899126279664727,
      "loss": 0.856,
      "step": 18390
    },
    {
      "epoch": 0.3230393792025843,
      "grad_norm": 0.07420983071030114,
      "learning_rate": 0.0001989896754419915,
      "loss": 0.8608,
      "step": 18400
    },
    {
      "epoch": 0.3232149440825857,
      "grad_norm": 0.049508529215607104,
      "learning_rate": 0.00019898808684578776,
      "loss": 0.8589,
      "step": 18410
    },
    {
      "epoch": 0.32339050896258714,
      "grad_norm": 0.05673206488508253,
      "learning_rate": 0.0001989864970080562,
      "loss": 0.8678,
      "step": 18420
    },
    {
      "epoch": 0.3235660738425885,
      "grad_norm": 0.05363425592466746,
      "learning_rate": 0.0001989849059288169,
      "loss": 0.8602,
      "step": 18430
    },
    {
      "epoch": 0.32374163872258993,
      "grad_norm": 0.07498576446654881,
      "learning_rate": 0.0001989833136080901,
      "loss": 0.8573,
      "step": 18440
    },
    {
      "epoch": 0.32391720360259135,
      "grad_norm": 0.07388062209430764,
      "learning_rate": 0.0001989817200458959,
      "loss": 0.8577,
      "step": 18450
    },
    {
      "epoch": 0.3240927684825927,
      "grad_norm": 0.052702807122935805,
      "learning_rate": 0.0001989801252422546,
      "loss": 0.8598,
      "step": 18460
    },
    {
      "epoch": 0.32426833336259414,
      "grad_norm": 0.06101344716131673,
      "learning_rate": 0.00019897852919718628,
      "loss": 0.8519,
      "step": 18470
    },
    {
      "epoch": 0.32444389824259556,
      "grad_norm": 0.08190698317015752,
      "learning_rate": 0.0001989769319107112,
      "loss": 0.8559,
      "step": 18480
    },
    {
      "epoch": 0.32461946312259693,
      "grad_norm": 0.06471139644855006,
      "learning_rate": 0.00019897533338284967,
      "loss": 0.8524,
      "step": 18490
    },
    {
      "epoch": 0.32479502800259835,
      "grad_norm": 0.06993863208512949,
      "learning_rate": 0.00019897373361362186,
      "loss": 0.8556,
      "step": 18500
    },
    {
      "epoch": 0.3249705928825998,
      "grad_norm": 0.05685831199441382,
      "learning_rate": 0.00019897213260304807,
      "loss": 0.8563,
      "step": 18510
    },
    {
      "epoch": 0.3251461577626012,
      "grad_norm": 0.082169275270938,
      "learning_rate": 0.00019897053035114858,
      "loss": 0.8509,
      "step": 18520
    },
    {
      "epoch": 0.32532172264260256,
      "grad_norm": 0.06795800317773747,
      "learning_rate": 0.00019896892685794372,
      "loss": 0.8611,
      "step": 18530
    },
    {
      "epoch": 0.325497287522604,
      "grad_norm": 0.06533294230706752,
      "learning_rate": 0.00019896732212345373,
      "loss": 0.8603,
      "step": 18540
    },
    {
      "epoch": 0.3256728524026054,
      "grad_norm": 0.0663909903654868,
      "learning_rate": 0.00019896571614769903,
      "loss": 0.8545,
      "step": 18550
    },
    {
      "epoch": 0.3258484172826068,
      "grad_norm": 0.058432993993101257,
      "learning_rate": 0.00019896410893069995,
      "loss": 0.8624,
      "step": 18560
    },
    {
      "epoch": 0.3260239821626082,
      "grad_norm": 0.057338400120719515,
      "learning_rate": 0.00019896250047247679,
      "loss": 0.8596,
      "step": 18570
    },
    {
      "epoch": 0.3261995470426096,
      "grad_norm": 0.0691883331033922,
      "learning_rate": 0.00019896089077305,
      "loss": 0.8537,
      "step": 18580
    },
    {
      "epoch": 0.326375111922611,
      "grad_norm": 0.1076136534823399,
      "learning_rate": 0.00019895927983243993,
      "loss": 0.8613,
      "step": 18590
    },
    {
      "epoch": 0.3265506768026124,
      "grad_norm": 0.06863770809012484,
      "learning_rate": 0.00019895766765066702,
      "loss": 0.8539,
      "step": 18600
    },
    {
      "epoch": 0.3267262416826138,
      "grad_norm": 0.09068751545991117,
      "learning_rate": 0.00019895605422775168,
      "loss": 0.8557,
      "step": 18610
    },
    {
      "epoch": 0.3269018065626152,
      "grad_norm": 0.06686752968304399,
      "learning_rate": 0.00019895443956371439,
      "loss": 0.8661,
      "step": 18620
    },
    {
      "epoch": 0.3270773714426166,
      "grad_norm": 0.055359035040326764,
      "learning_rate": 0.00019895282365857558,
      "loss": 0.8527,
      "step": 18630
    },
    {
      "epoch": 0.32725293632261804,
      "grad_norm": 0.0625405413744945,
      "learning_rate": 0.0001989512065123557,
      "loss": 0.861,
      "step": 18640
    },
    {
      "epoch": 0.3274285012026194,
      "grad_norm": 0.06414350771746707,
      "learning_rate": 0.00019894958812507525,
      "loss": 0.8556,
      "step": 18650
    },
    {
      "epoch": 0.3276040660826208,
      "grad_norm": 0.08495493156166287,
      "learning_rate": 0.00019894796849675476,
      "loss": 0.8586,
      "step": 18660
    },
    {
      "epoch": 0.32777963096262225,
      "grad_norm": 0.08596165564328906,
      "learning_rate": 0.00019894634762741473,
      "loss": 0.8568,
      "step": 18670
    },
    {
      "epoch": 0.3279551958426236,
      "grad_norm": 0.055024184339675646,
      "learning_rate": 0.0001989447255170757,
      "loss": 0.8529,
      "step": 18680
    },
    {
      "epoch": 0.32813076072262504,
      "grad_norm": 0.07209922716629169,
      "learning_rate": 0.00019894310216575826,
      "loss": 0.8538,
      "step": 18690
    },
    {
      "epoch": 0.32830632560262646,
      "grad_norm": 0.10023394572185754,
      "learning_rate": 0.00019894147757348293,
      "loss": 0.8593,
      "step": 18700
    },
    {
      "epoch": 0.3284818904826279,
      "grad_norm": 0.07267567278231189,
      "learning_rate": 0.00019893985174027027,
      "loss": 0.8468,
      "step": 18710
    },
    {
      "epoch": 0.32865745536262925,
      "grad_norm": 0.0738679250746881,
      "learning_rate": 0.00019893822466614098,
      "loss": 0.8634,
      "step": 18720
    },
    {
      "epoch": 0.32883302024263067,
      "grad_norm": 0.07264093967782095,
      "learning_rate": 0.00019893659635111558,
      "loss": 0.851,
      "step": 18730
    },
    {
      "epoch": 0.3290085851226321,
      "grad_norm": 0.0580287917377172,
      "learning_rate": 0.00019893496679521472,
      "loss": 0.8499,
      "step": 18740
    },
    {
      "epoch": 0.32918415000263346,
      "grad_norm": 0.09794938207603877,
      "learning_rate": 0.0001989333359984591,
      "loss": 0.8554,
      "step": 18750
    },
    {
      "epoch": 0.3293597148826349,
      "grad_norm": 0.06515141338612652,
      "learning_rate": 0.00019893170396086934,
      "loss": 0.8617,
      "step": 18760
    },
    {
      "epoch": 0.3295352797626363,
      "grad_norm": 0.0657968709223574,
      "learning_rate": 0.0001989300706824661,
      "loss": 0.8533,
      "step": 18770
    },
    {
      "epoch": 0.32971084464263767,
      "grad_norm": 0.06306917805254626,
      "learning_rate": 0.00019892843616327007,
      "loss": 0.8617,
      "step": 18780
    },
    {
      "epoch": 0.3298864095226391,
      "grad_norm": 0.06915814028661635,
      "learning_rate": 0.000198926800403302,
      "loss": 0.8633,
      "step": 18790
    },
    {
      "epoch": 0.3300619744026405,
      "grad_norm": 0.059676357862892604,
      "learning_rate": 0.00019892516340258264,
      "loss": 0.8546,
      "step": 18800
    },
    {
      "epoch": 0.3302375392826419,
      "grad_norm": 0.07677420656760534,
      "learning_rate": 0.00019892352516113267,
      "loss": 0.8458,
      "step": 18810
    },
    {
      "epoch": 0.3304131041626433,
      "grad_norm": 0.07882154446475968,
      "learning_rate": 0.00019892188567897289,
      "loss": 0.8473,
      "step": 18820
    },
    {
      "epoch": 0.3305886690426447,
      "grad_norm": 0.07975205553621646,
      "learning_rate": 0.00019892024495612401,
      "loss": 0.8604,
      "step": 18830
    },
    {
      "epoch": 0.3307642339226461,
      "grad_norm": 0.06523200004511905,
      "learning_rate": 0.0001989186029926069,
      "loss": 0.8581,
      "step": 18840
    },
    {
      "epoch": 0.3309397988026475,
      "grad_norm": 0.0614540754826004,
      "learning_rate": 0.00019891695978844232,
      "loss": 0.8632,
      "step": 18850
    },
    {
      "epoch": 0.33111536368264893,
      "grad_norm": 0.05899052228208036,
      "learning_rate": 0.00019891531534365107,
      "loss": 0.8599,
      "step": 18860
    },
    {
      "epoch": 0.3312909285626503,
      "grad_norm": 0.07394025237206421,
      "learning_rate": 0.00019891366965825403,
      "loss": 0.8672,
      "step": 18870
    },
    {
      "epoch": 0.3314664934426517,
      "grad_norm": 0.07548975259731722,
      "learning_rate": 0.000198912022732272,
      "loss": 0.8623,
      "step": 18880
    },
    {
      "epoch": 0.33164205832265314,
      "grad_norm": 0.06802945886093213,
      "learning_rate": 0.0001989103745657259,
      "loss": 0.8553,
      "step": 18890
    },
    {
      "epoch": 0.33181762320265457,
      "grad_norm": 0.08529910965996217,
      "learning_rate": 0.00019890872515863662,
      "loss": 0.8533,
      "step": 18900
    },
    {
      "epoch": 0.33199318808265593,
      "grad_norm": 0.07900977298707702,
      "learning_rate": 0.00019890707451102497,
      "loss": 0.8672,
      "step": 18910
    },
    {
      "epoch": 0.33216875296265735,
      "grad_norm": 0.09257008185764286,
      "learning_rate": 0.00019890542262291193,
      "loss": 0.86,
      "step": 18920
    },
    {
      "epoch": 0.3323443178426588,
      "grad_norm": 0.06244187074627285,
      "learning_rate": 0.00019890376949431845,
      "loss": 0.8652,
      "step": 18930
    },
    {
      "epoch": 0.33251988272266014,
      "grad_norm": 0.05624041558005601,
      "learning_rate": 0.00019890211512526545,
      "loss": 0.8615,
      "step": 18940
    },
    {
      "epoch": 0.33269544760266156,
      "grad_norm": 0.0617480131881405,
      "learning_rate": 0.00019890045951577386,
      "loss": 0.8482,
      "step": 18950
    },
    {
      "epoch": 0.332871012482663,
      "grad_norm": 0.06337386002724951,
      "learning_rate": 0.00019889880266586472,
      "loss": 0.8603,
      "step": 18960
    },
    {
      "epoch": 0.33304657736266435,
      "grad_norm": 0.07428524293212364,
      "learning_rate": 0.00019889714457555898,
      "loss": 0.866,
      "step": 18970
    },
    {
      "epoch": 0.3332221422426658,
      "grad_norm": 0.06460208542113219,
      "learning_rate": 0.00019889548524487763,
      "loss": 0.8565,
      "step": 18980
    },
    {
      "epoch": 0.3333977071226672,
      "grad_norm": 0.07133436468416435,
      "learning_rate": 0.00019889382467384173,
      "loss": 0.8584,
      "step": 18990
    },
    {
      "epoch": 0.33357327200266856,
      "grad_norm": 0.049696927161112826,
      "learning_rate": 0.0001988921628624723,
      "loss": 0.8702,
      "step": 19000
    },
    {
      "epoch": 0.33374883688267,
      "grad_norm": 0.06120569653143817,
      "learning_rate": 0.00019889049981079043,
      "loss": 0.8523,
      "step": 19010
    },
    {
      "epoch": 0.3339244017626714,
      "grad_norm": 0.06875605222135577,
      "learning_rate": 0.00019888883551881718,
      "loss": 0.8634,
      "step": 19020
    },
    {
      "epoch": 0.3340999666426728,
      "grad_norm": 0.07120057991833055,
      "learning_rate": 0.00019888716998657359,
      "loss": 0.8621,
      "step": 19030
    },
    {
      "epoch": 0.3342755315226742,
      "grad_norm": 0.07578951118508193,
      "learning_rate": 0.00019888550321408083,
      "loss": 0.8508,
      "step": 19040
    },
    {
      "epoch": 0.3344510964026756,
      "grad_norm": 0.09799541178275413,
      "learning_rate": 0.0001988838352013599,
      "loss": 0.8547,
      "step": 19050
    },
    {
      "epoch": 0.33462666128267704,
      "grad_norm": 0.06273564983196227,
      "learning_rate": 0.0001988821659484321,
      "loss": 0.8587,
      "step": 19060
    },
    {
      "epoch": 0.3348022261626784,
      "grad_norm": 0.07694243429423718,
      "learning_rate": 0.0001988804954553185,
      "loss": 0.8525,
      "step": 19070
    },
    {
      "epoch": 0.33497779104267983,
      "grad_norm": 0.062721001798842,
      "learning_rate": 0.00019887882372204022,
      "loss": 0.8516,
      "step": 19080
    },
    {
      "epoch": 0.33515335592268125,
      "grad_norm": 0.0818155119618455,
      "learning_rate": 0.00019887715074861853,
      "loss": 0.8588,
      "step": 19090
    },
    {
      "epoch": 0.3353289208026826,
      "grad_norm": 0.06557288465529827,
      "learning_rate": 0.00019887547653507455,
      "loss": 0.859,
      "step": 19100
    },
    {
      "epoch": 0.33550448568268404,
      "grad_norm": 0.06265468053727054,
      "learning_rate": 0.00019887380108142954,
      "loss": 0.8532,
      "step": 19110
    },
    {
      "epoch": 0.33568005056268546,
      "grad_norm": 0.08459570665241795,
      "learning_rate": 0.0001988721243877047,
      "loss": 0.8565,
      "step": 19120
    },
    {
      "epoch": 0.33585561544268683,
      "grad_norm": 0.07117709860027395,
      "learning_rate": 0.0001988704464539213,
      "loss": 0.8646,
      "step": 19130
    },
    {
      "epoch": 0.33603118032268825,
      "grad_norm": 0.06918889042790631,
      "learning_rate": 0.0001988687672801006,
      "loss": 0.8543,
      "step": 19140
    },
    {
      "epoch": 0.33620674520268967,
      "grad_norm": 0.05641660473064654,
      "learning_rate": 0.00019886708686626386,
      "loss": 0.8648,
      "step": 19150
    },
    {
      "epoch": 0.33638231008269104,
      "grad_norm": 0.06356431171466186,
      "learning_rate": 0.00019886540521243232,
      "loss": 0.8561,
      "step": 19160
    },
    {
      "epoch": 0.33655787496269246,
      "grad_norm": 0.06176508724472949,
      "learning_rate": 0.0001988637223186274,
      "loss": 0.854,
      "step": 19170
    },
    {
      "epoch": 0.3367334398426939,
      "grad_norm": 0.07941477904117106,
      "learning_rate": 0.0001988620381848703,
      "loss": 0.866,
      "step": 19180
    },
    {
      "epoch": 0.33690900472269525,
      "grad_norm": 0.097539352141863,
      "learning_rate": 0.00019886035281118248,
      "loss": 0.8571,
      "step": 19190
    },
    {
      "epoch": 0.33708456960269667,
      "grad_norm": 0.07822545800772135,
      "learning_rate": 0.00019885866619758522,
      "loss": 0.858,
      "step": 19200
    },
    {
      "epoch": 0.3372601344826981,
      "grad_norm": 0.07492605204460843,
      "learning_rate": 0.00019885697834409987,
      "loss": 0.8634,
      "step": 19210
    },
    {
      "epoch": 0.33743569936269946,
      "grad_norm": 0.0725855075048947,
      "learning_rate": 0.0001988552892507479,
      "loss": 0.8578,
      "step": 19220
    },
    {
      "epoch": 0.3376112642427009,
      "grad_norm": 0.06839137658634518,
      "learning_rate": 0.00019885359891755065,
      "loss": 0.8541,
      "step": 19230
    },
    {
      "epoch": 0.3377868291227023,
      "grad_norm": 0.08010319760528196,
      "learning_rate": 0.00019885190734452953,
      "loss": 0.8597,
      "step": 19240
    },
    {
      "epoch": 0.3379623940027037,
      "grad_norm": 0.1042104552223733,
      "learning_rate": 0.000198850214531706,
      "loss": 0.8602,
      "step": 19250
    },
    {
      "epoch": 0.3381379588827051,
      "grad_norm": 0.09303863814319994,
      "learning_rate": 0.0001988485204791015,
      "loss": 0.8599,
      "step": 19260
    },
    {
      "epoch": 0.3383135237627065,
      "grad_norm": 0.058872347367429065,
      "learning_rate": 0.00019884682518673752,
      "loss": 0.8597,
      "step": 19270
    },
    {
      "epoch": 0.33848908864270794,
      "grad_norm": 0.07542841309523092,
      "learning_rate": 0.00019884512865463546,
      "loss": 0.85,
      "step": 19280
    },
    {
      "epoch": 0.3386646535227093,
      "grad_norm": 0.07476841687870499,
      "learning_rate": 0.00019884343088281693,
      "loss": 0.8629,
      "step": 19290
    },
    {
      "epoch": 0.3388402184027107,
      "grad_norm": 0.07080523910555533,
      "learning_rate": 0.00019884173187130334,
      "loss": 0.8579,
      "step": 19300
    },
    {
      "epoch": 0.33901578328271215,
      "grad_norm": 0.06323463238820282,
      "learning_rate": 0.00019884003162011623,
      "loss": 0.8608,
      "step": 19310
    },
    {
      "epoch": 0.3391913481627135,
      "grad_norm": 0.06578883049271064,
      "learning_rate": 0.00019883833012927722,
      "loss": 0.8576,
      "step": 19320
    },
    {
      "epoch": 0.33936691304271493,
      "grad_norm": 0.0762327459427793,
      "learning_rate": 0.00019883662739880777,
      "loss": 0.8589,
      "step": 19330
    },
    {
      "epoch": 0.33954247792271636,
      "grad_norm": 0.0639584680801323,
      "learning_rate": 0.0001988349234287295,
      "loss": 0.862,
      "step": 19340
    },
    {
      "epoch": 0.3397180428027177,
      "grad_norm": 0.06860161690537506,
      "learning_rate": 0.00019883321821906406,
      "loss": 0.8633,
      "step": 19350
    },
    {
      "epoch": 0.33989360768271915,
      "grad_norm": 0.061219235355922934,
      "learning_rate": 0.00019883151176983295,
      "loss": 0.8637,
      "step": 19360
    },
    {
      "epoch": 0.34006917256272057,
      "grad_norm": 0.07219105485277513,
      "learning_rate": 0.00019882980408105784,
      "loss": 0.8638,
      "step": 19370
    },
    {
      "epoch": 0.34024473744272193,
      "grad_norm": 0.068214182492769,
      "learning_rate": 0.0001988280951527604,
      "loss": 0.8619,
      "step": 19380
    },
    {
      "epoch": 0.34042030232272336,
      "grad_norm": 0.06545387018373347,
      "learning_rate": 0.0001988263849849622,
      "loss": 0.8586,
      "step": 19390
    },
    {
      "epoch": 0.3405958672027248,
      "grad_norm": 0.07469779262237132,
      "learning_rate": 0.00019882467357768495,
      "loss": 0.8505,
      "step": 19400
    },
    {
      "epoch": 0.34077143208272614,
      "grad_norm": 0.06814175260656055,
      "learning_rate": 0.00019882296093095037,
      "loss": 0.8543,
      "step": 19410
    },
    {
      "epoch": 0.34094699696272757,
      "grad_norm": 0.06865376790400311,
      "learning_rate": 0.0001988212470447801,
      "loss": 0.8587,
      "step": 19420
    },
    {
      "epoch": 0.341122561842729,
      "grad_norm": 0.06299276954007664,
      "learning_rate": 0.00019881953191919593,
      "loss": 0.8587,
      "step": 19430
    },
    {
      "epoch": 0.3412981267227304,
      "grad_norm": 0.07484274614655176,
      "learning_rate": 0.00019881781555421949,
      "loss": 0.8614,
      "step": 19440
    },
    {
      "epoch": 0.3414736916027318,
      "grad_norm": 0.06984967965563989,
      "learning_rate": 0.00019881609794987262,
      "loss": 0.8613,
      "step": 19450
    },
    {
      "epoch": 0.3416492564827332,
      "grad_norm": 0.0780934288468703,
      "learning_rate": 0.00019881437910617702,
      "loss": 0.8531,
      "step": 19460
    },
    {
      "epoch": 0.3418248213627346,
      "grad_norm": 0.06033108308601824,
      "learning_rate": 0.00019881265902315447,
      "loss": 0.8571,
      "step": 19470
    },
    {
      "epoch": 0.342000386242736,
      "grad_norm": 0.055020852395432036,
      "learning_rate": 0.00019881093770082685,
      "loss": 0.8545,
      "step": 19480
    },
    {
      "epoch": 0.3421759511227374,
      "grad_norm": 0.06502461248464224,
      "learning_rate": 0.0001988092151392158,
      "loss": 0.8554,
      "step": 19490
    },
    {
      "epoch": 0.34235151600273883,
      "grad_norm": 0.08431985430216779,
      "learning_rate": 0.00019880749133834337,
      "loss": 0.861,
      "step": 19500
    },
    {
      "epoch": 0.3425270808827402,
      "grad_norm": 0.07290201849200763,
      "learning_rate": 0.00019880576629823117,
      "loss": 0.8596,
      "step": 19510
    },
    {
      "epoch": 0.3427026457627416,
      "grad_norm": 0.06599842265496005,
      "learning_rate": 0.00019880404001890124,
      "loss": 0.8536,
      "step": 19520
    },
    {
      "epoch": 0.34287821064274304,
      "grad_norm": 0.05105415106065,
      "learning_rate": 0.00019880231250037536,
      "loss": 0.8627,
      "step": 19530
    },
    {
      "epoch": 0.3430537755227444,
      "grad_norm": 0.07552749113078049,
      "learning_rate": 0.00019880058374267546,
      "loss": 0.8594,
      "step": 19540
    },
    {
      "epoch": 0.34322934040274583,
      "grad_norm": 0.07541343507894413,
      "learning_rate": 0.00019879885374582336,
      "loss": 0.8473,
      "step": 19550
    },
    {
      "epoch": 0.34340490528274725,
      "grad_norm": 0.08796876798333103,
      "learning_rate": 0.0001987971225098411,
      "loss": 0.8567,
      "step": 19560
    },
    {
      "epoch": 0.3435804701627486,
      "grad_norm": 0.05992163733142649,
      "learning_rate": 0.00019879539003475052,
      "loss": 0.8606,
      "step": 19570
    },
    {
      "epoch": 0.34375603504275004,
      "grad_norm": 0.0779392219255425,
      "learning_rate": 0.00019879365632057363,
      "loss": 0.8497,
      "step": 19580
    },
    {
      "epoch": 0.34393159992275146,
      "grad_norm": 0.07657760176427332,
      "learning_rate": 0.00019879192136733234,
      "loss": 0.8592,
      "step": 19590
    },
    {
      "epoch": 0.34410716480275283,
      "grad_norm": 0.09636905389619554,
      "learning_rate": 0.00019879018517504865,
      "loss": 0.8472,
      "step": 19600
    },
    {
      "epoch": 0.34428272968275425,
      "grad_norm": 0.09385584026768841,
      "learning_rate": 0.00019878844774374462,
      "loss": 0.8585,
      "step": 19610
    },
    {
      "epoch": 0.3444582945627557,
      "grad_norm": 0.06769690963694532,
      "learning_rate": 0.00019878670907344222,
      "loss": 0.8627,
      "step": 19620
    },
    {
      "epoch": 0.3446338594427571,
      "grad_norm": 0.0539661532233032,
      "learning_rate": 0.00019878496916416345,
      "loss": 0.8626,
      "step": 19630
    },
    {
      "epoch": 0.34480942432275846,
      "grad_norm": 0.05589894544911685,
      "learning_rate": 0.0001987832280159304,
      "loss": 0.8616,
      "step": 19640
    },
    {
      "epoch": 0.3449849892027599,
      "grad_norm": 0.0662786838428017,
      "learning_rate": 0.00019878148562876507,
      "loss": 0.8509,
      "step": 19650
    },
    {
      "epoch": 0.3451605540827613,
      "grad_norm": 0.06874158970078188,
      "learning_rate": 0.00019877974200268963,
      "loss": 0.851,
      "step": 19660
    },
    {
      "epoch": 0.3453361189627627,
      "grad_norm": 0.12275611337537763,
      "learning_rate": 0.0001987779971377261,
      "loss": 0.8567,
      "step": 19670
    },
    {
      "epoch": 0.3455116838427641,
      "grad_norm": 0.06640712973763466,
      "learning_rate": 0.0001987762510338966,
      "loss": 0.8545,
      "step": 19680
    },
    {
      "epoch": 0.3456872487227655,
      "grad_norm": 0.0664307905848072,
      "learning_rate": 0.00019877450369122326,
      "loss": 0.8594,
      "step": 19690
    },
    {
      "epoch": 0.3458628136027669,
      "grad_norm": 0.061385686251929666,
      "learning_rate": 0.00019877275510972822,
      "loss": 0.8556,
      "step": 19700
    },
    {
      "epoch": 0.3460383784827683,
      "grad_norm": 0.07348603857578694,
      "learning_rate": 0.00019877100528943362,
      "loss": 0.8589,
      "step": 19710
    },
    {
      "epoch": 0.3462139433627697,
      "grad_norm": 0.08435252076769903,
      "learning_rate": 0.00019876925423036168,
      "loss": 0.8619,
      "step": 19720
    },
    {
      "epoch": 0.3463895082427711,
      "grad_norm": 0.05701628579623089,
      "learning_rate": 0.00019876750193253452,
      "loss": 0.8617,
      "step": 19730
    },
    {
      "epoch": 0.3465650731227725,
      "grad_norm": 0.053009458911936376,
      "learning_rate": 0.0001987657483959744,
      "loss": 0.8629,
      "step": 19740
    },
    {
      "epoch": 0.34674063800277394,
      "grad_norm": 0.06494154887299836,
      "learning_rate": 0.00019876399362070348,
      "loss": 0.8567,
      "step": 19750
    },
    {
      "epoch": 0.3469162028827753,
      "grad_norm": 0.06252552744175231,
      "learning_rate": 0.00019876223760674407,
      "loss": 0.8585,
      "step": 19760
    },
    {
      "epoch": 0.3470917677627767,
      "grad_norm": 0.06468028372497749,
      "learning_rate": 0.00019876048035411833,
      "loss": 0.8595,
      "step": 19770
    },
    {
      "epoch": 0.34726733264277815,
      "grad_norm": 0.07421886930242856,
      "learning_rate": 0.00019875872186284857,
      "loss": 0.8552,
      "step": 19780
    },
    {
      "epoch": 0.34744289752277957,
      "grad_norm": 0.07157750975157508,
      "learning_rate": 0.0001987569621329571,
      "loss": 0.8604,
      "step": 19790
    },
    {
      "epoch": 0.34761846240278094,
      "grad_norm": 0.07600286289357912,
      "learning_rate": 0.00019875520116446616,
      "loss": 0.8664,
      "step": 19800
    },
    {
      "epoch": 0.34779402728278236,
      "grad_norm": 0.0925284187542818,
      "learning_rate": 0.0001987534389573981,
      "loss": 0.8605,
      "step": 19810
    },
    {
      "epoch": 0.3479695921627838,
      "grad_norm": 0.07120781055885743,
      "learning_rate": 0.00019875167551177523,
      "loss": 0.8613,
      "step": 19820
    },
    {
      "epoch": 0.34814515704278515,
      "grad_norm": 0.07128265897297884,
      "learning_rate": 0.0001987499108276199,
      "loss": 0.86,
      "step": 19830
    },
    {
      "epoch": 0.34832072192278657,
      "grad_norm": 0.07530020697530185,
      "learning_rate": 0.00019874814490495447,
      "loss": 0.8601,
      "step": 19840
    },
    {
      "epoch": 0.348496286802788,
      "grad_norm": 0.09638692670439293,
      "learning_rate": 0.0001987463777438013,
      "loss": 0.8589,
      "step": 19850
    },
    {
      "epoch": 0.34867185168278936,
      "grad_norm": 0.07599327635980532,
      "learning_rate": 0.0001987446093441828,
      "loss": 0.8557,
      "step": 19860
    },
    {
      "epoch": 0.3488474165627908,
      "grad_norm": 0.0642043758751764,
      "learning_rate": 0.00019874283970612136,
      "loss": 0.8537,
      "step": 19870
    },
    {
      "epoch": 0.3490229814427922,
      "grad_norm": 0.0991161460276962,
      "learning_rate": 0.00019874106882963942,
      "loss": 0.8599,
      "step": 19880
    },
    {
      "epoch": 0.34919854632279357,
      "grad_norm": 0.07033014614232033,
      "learning_rate": 0.00019873929671475937,
      "loss": 0.8641,
      "step": 19890
    },
    {
      "epoch": 0.349374111202795,
      "grad_norm": 0.05675025439440612,
      "learning_rate": 0.00019873752336150373,
      "loss": 0.8585,
      "step": 19900
    },
    {
      "epoch": 0.3495496760827964,
      "grad_norm": 0.05574624689821413,
      "learning_rate": 0.00019873574876989492,
      "loss": 0.8567,
      "step": 19910
    },
    {
      "epoch": 0.3497252409627978,
      "grad_norm": 0.06063040586012082,
      "learning_rate": 0.00019873397293995544,
      "loss": 0.8499,
      "step": 19920
    },
    {
      "epoch": 0.3499008058427992,
      "grad_norm": 0.06134876430093596,
      "learning_rate": 0.00019873219587170782,
      "loss": 0.8604,
      "step": 19930
    },
    {
      "epoch": 0.3500763707228006,
      "grad_norm": 0.06534915883068793,
      "learning_rate": 0.0001987304175651745,
      "loss": 0.856,
      "step": 19940
    },
    {
      "epoch": 0.350251935602802,
      "grad_norm": 0.07218691768826466,
      "learning_rate": 0.00019872863802037813,
      "loss": 0.875,
      "step": 19950
    },
    {
      "epoch": 0.3504275004828034,
      "grad_norm": 0.05975619007980095,
      "learning_rate": 0.00019872685723734112,
      "loss": 0.8587,
      "step": 19960
    },
    {
      "epoch": 0.35060306536280483,
      "grad_norm": 0.08557476009712692,
      "learning_rate": 0.00019872507521608615,
      "loss": 0.8574,
      "step": 19970
    },
    {
      "epoch": 0.35077863024280626,
      "grad_norm": 0.06472602172752759,
      "learning_rate": 0.00019872329195663572,
      "loss": 0.8583,
      "step": 19980
    },
    {
      "epoch": 0.3509541951228076,
      "grad_norm": 0.05756598727329916,
      "learning_rate": 0.00019872150745901246,
      "loss": 0.8569,
      "step": 19990
    },
    {
      "epoch": 0.35112976000280904,
      "grad_norm": 0.07935423533285582,
      "learning_rate": 0.00019871972172323899,
      "loss": 0.8586,
      "step": 20000
    },
    {
      "epoch": 0.35130532488281047,
      "grad_norm": 0.06027886554525522,
      "learning_rate": 0.0001987179347493379,
      "loss": 0.8562,
      "step": 20010
    },
    {
      "epoch": 0.35148088976281183,
      "grad_norm": 0.061797041752572425,
      "learning_rate": 0.00019871614653733184,
      "loss": 0.857,
      "step": 20020
    },
    {
      "epoch": 0.35165645464281325,
      "grad_norm": 0.07176249771241038,
      "learning_rate": 0.00019871435708724346,
      "loss": 0.8571,
      "step": 20030
    },
    {
      "epoch": 0.3518320195228147,
      "grad_norm": 0.062783318398205,
      "learning_rate": 0.0001987125663990955,
      "loss": 0.8605,
      "step": 20040
    },
    {
      "epoch": 0.35200758440281604,
      "grad_norm": 0.05826237709703035,
      "learning_rate": 0.00019871077447291055,
      "loss": 0.8565,
      "step": 20050
    },
    {
      "epoch": 0.35218314928281746,
      "grad_norm": 0.0538422186561053,
      "learning_rate": 0.0001987089813087114,
      "loss": 0.8605,
      "step": 20060
    },
    {
      "epoch": 0.3523587141628189,
      "grad_norm": 0.06453328614478401,
      "learning_rate": 0.0001987071869065207,
      "loss": 0.8624,
      "step": 20070
    },
    {
      "epoch": 0.35253427904282025,
      "grad_norm": 0.0769593029665899,
      "learning_rate": 0.00019870539126636122,
      "loss": 0.8562,
      "step": 20080
    },
    {
      "epoch": 0.3527098439228217,
      "grad_norm": 0.07390313686004522,
      "learning_rate": 0.0001987035943882557,
      "loss": 0.859,
      "step": 20090
    },
    {
      "epoch": 0.3528854088028231,
      "grad_norm": 0.07484257631919211,
      "learning_rate": 0.00019870179627222697,
      "loss": 0.8549,
      "step": 20100
    },
    {
      "epoch": 0.35306097368282446,
      "grad_norm": 0.05233942075967626,
      "learning_rate": 0.0001986999969182977,
      "loss": 0.8504,
      "step": 20110
    },
    {
      "epoch": 0.3532365385628259,
      "grad_norm": 0.06873565051060167,
      "learning_rate": 0.0001986981963264908,
      "loss": 0.8566,
      "step": 20120
    },
    {
      "epoch": 0.3534121034428273,
      "grad_norm": 0.05326195061645983,
      "learning_rate": 0.00019869639449682897,
      "loss": 0.8592,
      "step": 20130
    },
    {
      "epoch": 0.3535876683228287,
      "grad_norm": 0.06556523802328988,
      "learning_rate": 0.0001986945914293351,
      "loss": 0.8624,
      "step": 20140
    },
    {
      "epoch": 0.3537632332028301,
      "grad_norm": 0.07242523536634146,
      "learning_rate": 0.00019869278712403207,
      "loss": 0.8599,
      "step": 20150
    },
    {
      "epoch": 0.3539387980828315,
      "grad_norm": 0.07754953118306936,
      "learning_rate": 0.0001986909815809427,
      "loss": 0.8593,
      "step": 20160
    },
    {
      "epoch": 0.35411436296283294,
      "grad_norm": 0.05789654239359886,
      "learning_rate": 0.00019868917480008984,
      "loss": 0.8579,
      "step": 20170
    },
    {
      "epoch": 0.3542899278428343,
      "grad_norm": 0.058689837448439815,
      "learning_rate": 0.00019868736678149644,
      "loss": 0.8601,
      "step": 20180
    },
    {
      "epoch": 0.35446549272283573,
      "grad_norm": 0.0681252476981887,
      "learning_rate": 0.00019868555752518537,
      "loss": 0.8584,
      "step": 20190
    },
    {
      "epoch": 0.35464105760283715,
      "grad_norm": 0.07910975863758687,
      "learning_rate": 0.00019868374703117957,
      "loss": 0.861,
      "step": 20200
    },
    {
      "epoch": 0.3548166224828385,
      "grad_norm": 0.06361999003084971,
      "learning_rate": 0.000198681935299502,
      "loss": 0.8548,
      "step": 20210
    },
    {
      "epoch": 0.35499218736283994,
      "grad_norm": 0.06954391710823767,
      "learning_rate": 0.00019868012233017557,
      "loss": 0.858,
      "step": 20220
    },
    {
      "epoch": 0.35516775224284136,
      "grad_norm": 0.053769468743871444,
      "learning_rate": 0.00019867830812322324,
      "loss": 0.8611,
      "step": 20230
    },
    {
      "epoch": 0.35534331712284273,
      "grad_norm": 0.051781819891275706,
      "learning_rate": 0.00019867649267866805,
      "loss": 0.862,
      "step": 20240
    },
    {
      "epoch": 0.35551888200284415,
      "grad_norm": 0.0836105578843198,
      "learning_rate": 0.00019867467599653298,
      "loss": 0.8543,
      "step": 20250
    },
    {
      "epoch": 0.35569444688284557,
      "grad_norm": 0.08230228515551685,
      "learning_rate": 0.00019867285807684105,
      "loss": 0.8584,
      "step": 20260
    },
    {
      "epoch": 0.35587001176284694,
      "grad_norm": 0.07077824414648133,
      "learning_rate": 0.00019867103891961526,
      "loss": 0.8557,
      "step": 20270
    },
    {
      "epoch": 0.35604557664284836,
      "grad_norm": 0.06628259096472172,
      "learning_rate": 0.0001986692185248787,
      "loss": 0.8557,
      "step": 20280
    },
    {
      "epoch": 0.3562211415228498,
      "grad_norm": 0.07641370070077073,
      "learning_rate": 0.00019866739689265443,
      "loss": 0.8589,
      "step": 20290
    },
    {
      "epoch": 0.35639670640285115,
      "grad_norm": 0.09757237773052772,
      "learning_rate": 0.00019866557402296552,
      "loss": 0.8627,
      "step": 20300
    },
    {
      "epoch": 0.35657227128285257,
      "grad_norm": 0.07236062508102792,
      "learning_rate": 0.0001986637499158351,
      "loss": 0.862,
      "step": 20310
    },
    {
      "epoch": 0.356747836162854,
      "grad_norm": 0.05841148482985717,
      "learning_rate": 0.00019866192457128622,
      "loss": 0.8531,
      "step": 20320
    },
    {
      "epoch": 0.3569234010428554,
      "grad_norm": 0.07915689839779634,
      "learning_rate": 0.00019866009798934206,
      "loss": 0.8466,
      "step": 20330
    },
    {
      "epoch": 0.3570989659228568,
      "grad_norm": 0.06630181529700638,
      "learning_rate": 0.00019865827017002572,
      "loss": 0.8525,
      "step": 20340
    },
    {
      "epoch": 0.3572745308028582,
      "grad_norm": 0.062441598034236906,
      "learning_rate": 0.00019865644111336037,
      "loss": 0.8597,
      "step": 20350
    },
    {
      "epoch": 0.3574500956828596,
      "grad_norm": 0.09197148012406059,
      "learning_rate": 0.00019865461081936922,
      "loss": 0.8517,
      "step": 20360
    },
    {
      "epoch": 0.357625660562861,
      "grad_norm": 0.07851641792021084,
      "learning_rate": 0.00019865277928807543,
      "loss": 0.8653,
      "step": 20370
    },
    {
      "epoch": 0.3578012254428624,
      "grad_norm": 0.07460828633647006,
      "learning_rate": 0.0001986509465195022,
      "loss": 0.8531,
      "step": 20380
    },
    {
      "epoch": 0.35797679032286384,
      "grad_norm": 0.06621448048041624,
      "learning_rate": 0.0001986491125136728,
      "loss": 0.8648,
      "step": 20390
    },
    {
      "epoch": 0.3581523552028652,
      "grad_norm": 0.057390429353180096,
      "learning_rate": 0.0001986472772706104,
      "loss": 0.8598,
      "step": 20400
    },
    {
      "epoch": 0.3583279200828666,
      "grad_norm": 0.06221512188821132,
      "learning_rate": 0.00019864544079033826,
      "loss": 0.8513,
      "step": 20410
    },
    {
      "epoch": 0.35850348496286805,
      "grad_norm": 0.052469336455054895,
      "learning_rate": 0.00019864360307287972,
      "loss": 0.8581,
      "step": 20420
    },
    {
      "epoch": 0.3586790498428694,
      "grad_norm": 0.0504191238558091,
      "learning_rate": 0.00019864176411825798,
      "loss": 0.8617,
      "step": 20430
    },
    {
      "epoch": 0.35885461472287083,
      "grad_norm": 0.06414887709086453,
      "learning_rate": 0.0001986399239264964,
      "loss": 0.8458,
      "step": 20440
    },
    {
      "epoch": 0.35903017960287226,
      "grad_norm": 0.0685322691889545,
      "learning_rate": 0.00019863808249761825,
      "loss": 0.8521,
      "step": 20450
    },
    {
      "epoch": 0.3592057444828736,
      "grad_norm": 0.07533151216400781,
      "learning_rate": 0.00019863623983164692,
      "loss": 0.8597,
      "step": 20460
    },
    {
      "epoch": 0.35938130936287505,
      "grad_norm": 0.06722008188016607,
      "learning_rate": 0.00019863439592860565,
      "loss": 0.8575,
      "step": 20470
    },
    {
      "epoch": 0.35955687424287647,
      "grad_norm": 0.05414428543190899,
      "learning_rate": 0.00019863255078851795,
      "loss": 0.8582,
      "step": 20480
    },
    {
      "epoch": 0.35973243912287783,
      "grad_norm": 0.07133769921784942,
      "learning_rate": 0.00019863070441140707,
      "loss": 0.8614,
      "step": 20490
    },
    {
      "epoch": 0.35990800400287926,
      "grad_norm": 0.05138855622320601,
      "learning_rate": 0.00019862885679729645,
      "loss": 0.853,
      "step": 20500
    },
    {
      "epoch": 0.3600835688828807,
      "grad_norm": 0.09435590937062452,
      "learning_rate": 0.00019862700794620958,
      "loss": 0.8593,
      "step": 20510
    },
    {
      "epoch": 0.3602591337628821,
      "grad_norm": 0.06812291235095772,
      "learning_rate": 0.0001986251578581697,
      "loss": 0.8647,
      "step": 20520
    },
    {
      "epoch": 0.36043469864288347,
      "grad_norm": 0.06524320192133441,
      "learning_rate": 0.00019862330653320042,
      "loss": 0.8504,
      "step": 20530
    },
    {
      "epoch": 0.3606102635228849,
      "grad_norm": 0.08785004000841906,
      "learning_rate": 0.0001986214539713251,
      "loss": 0.8526,
      "step": 20540
    },
    {
      "epoch": 0.3607858284028863,
      "grad_norm": 0.07101622580846258,
      "learning_rate": 0.00019861960017256733,
      "loss": 0.8634,
      "step": 20550
    },
    {
      "epoch": 0.3609613932828877,
      "grad_norm": 0.06330315061637946,
      "learning_rate": 0.00019861774513695043,
      "loss": 0.8511,
      "step": 20560
    },
    {
      "epoch": 0.3611369581628891,
      "grad_norm": 0.0652338557445132,
      "learning_rate": 0.00019861588886449803,
      "loss": 0.8681,
      "step": 20570
    },
    {
      "epoch": 0.3613125230428905,
      "grad_norm": 0.10066426516875414,
      "learning_rate": 0.00019861403135523363,
      "loss": 0.8515,
      "step": 20580
    },
    {
      "epoch": 0.3614880879228919,
      "grad_norm": 0.045605147875744316,
      "learning_rate": 0.00019861217260918072,
      "loss": 0.8552,
      "step": 20590
    },
    {
      "epoch": 0.3616636528028933,
      "grad_norm": 0.06137757647397811,
      "learning_rate": 0.00019861031262636286,
      "loss": 0.8543,
      "step": 20600
    },
    {
      "epoch": 0.36183921768289473,
      "grad_norm": 0.05466464941380702,
      "learning_rate": 0.00019860845140680367,
      "loss": 0.8609,
      "step": 20610
    },
    {
      "epoch": 0.3620147825628961,
      "grad_norm": 0.06211208453644534,
      "learning_rate": 0.00019860658895052668,
      "loss": 0.8542,
      "step": 20620
    },
    {
      "epoch": 0.3621903474428975,
      "grad_norm": 0.08191909271875586,
      "learning_rate": 0.0001986047252575555,
      "loss": 0.8585,
      "step": 20630
    },
    {
      "epoch": 0.36236591232289894,
      "grad_norm": 0.07477485575549135,
      "learning_rate": 0.0001986028603279137,
      "loss": 0.8566,
      "step": 20640
    },
    {
      "epoch": 0.3625414772029003,
      "grad_norm": 0.0609639343855269,
      "learning_rate": 0.00019860099416162501,
      "loss": 0.8624,
      "step": 20650
    },
    {
      "epoch": 0.36271704208290173,
      "grad_norm": 0.061352977298533236,
      "learning_rate": 0.00019859912675871302,
      "loss": 0.8591,
      "step": 20660
    },
    {
      "epoch": 0.36289260696290315,
      "grad_norm": 0.06861804094859952,
      "learning_rate": 0.00019859725811920137,
      "loss": 0.8555,
      "step": 20670
    },
    {
      "epoch": 0.3630681718429045,
      "grad_norm": 0.060921008196608596,
      "learning_rate": 0.00019859538824311377,
      "loss": 0.8583,
      "step": 20680
    },
    {
      "epoch": 0.36324373672290594,
      "grad_norm": 0.07007639029575313,
      "learning_rate": 0.00019859351713047386,
      "loss": 0.8622,
      "step": 20690
    },
    {
      "epoch": 0.36341930160290736,
      "grad_norm": 0.07136377048766915,
      "learning_rate": 0.00019859164478130544,
      "loss": 0.8601,
      "step": 20700
    },
    {
      "epoch": 0.3635948664829088,
      "grad_norm": 0.08082491643464151,
      "learning_rate": 0.00019858977119563214,
      "loss": 0.8547,
      "step": 20710
    },
    {
      "epoch": 0.36377043136291015,
      "grad_norm": 0.06168096771557879,
      "learning_rate": 0.0001985878963734777,
      "loss": 0.8566,
      "step": 20720
    },
    {
      "epoch": 0.3639459962429116,
      "grad_norm": 0.059272326271018645,
      "learning_rate": 0.00019858602031486597,
      "loss": 0.8503,
      "step": 20730
    },
    {
      "epoch": 0.364121561122913,
      "grad_norm": 0.07280361445766313,
      "learning_rate": 0.00019858414301982064,
      "loss": 0.8532,
      "step": 20740
    },
    {
      "epoch": 0.36429712600291436,
      "grad_norm": 0.06923731370291544,
      "learning_rate": 0.00019858226448836555,
      "loss": 0.8447,
      "step": 20750
    },
    {
      "epoch": 0.3644726908829158,
      "grad_norm": 0.09024439616340618,
      "learning_rate": 0.00019858038472052444,
      "loss": 0.8567,
      "step": 20760
    },
    {
      "epoch": 0.3646482557629172,
      "grad_norm": 0.06630720028593005,
      "learning_rate": 0.00019857850371632113,
      "loss": 0.8564,
      "step": 20770
    },
    {
      "epoch": 0.3648238206429186,
      "grad_norm": 0.059560027169951,
      "learning_rate": 0.0001985766214757795,
      "loss": 0.853,
      "step": 20780
    },
    {
      "epoch": 0.36499938552292,
      "grad_norm": 0.07936113175026589,
      "learning_rate": 0.00019857473799892337,
      "loss": 0.8593,
      "step": 20790
    },
    {
      "epoch": 0.3651749504029214,
      "grad_norm": 0.06009347204594745,
      "learning_rate": 0.00019857285328577659,
      "loss": 0.8549,
      "step": 20800
    },
    {
      "epoch": 0.3653505152829228,
      "grad_norm": 0.052488318479547436,
      "learning_rate": 0.0001985709673363631,
      "loss": 0.8625,
      "step": 20810
    },
    {
      "epoch": 0.3655260801629242,
      "grad_norm": 0.08310975066718011,
      "learning_rate": 0.0001985690801507067,
      "loss": 0.855,
      "step": 20820
    },
    {
      "epoch": 0.3657016450429256,
      "grad_norm": 0.09194066940430959,
      "learning_rate": 0.00019856719172883143,
      "loss": 0.8541,
      "step": 20830
    },
    {
      "epoch": 0.365877209922927,
      "grad_norm": 0.09419886219868105,
      "learning_rate": 0.00019856530207076108,
      "loss": 0.8552,
      "step": 20840
    },
    {
      "epoch": 0.3660527748029284,
      "grad_norm": 0.09113880642862149,
      "learning_rate": 0.0001985634111765197,
      "loss": 0.8521,
      "step": 20850
    },
    {
      "epoch": 0.36622833968292984,
      "grad_norm": 0.11714406946777135,
      "learning_rate": 0.00019856151904613117,
      "loss": 0.8486,
      "step": 20860
    },
    {
      "epoch": 0.3664039045629312,
      "grad_norm": 0.09361727897457355,
      "learning_rate": 0.00019855962567961953,
      "loss": 0.8549,
      "step": 20870
    },
    {
      "epoch": 0.3665794694429326,
      "grad_norm": 0.06762462946141516,
      "learning_rate": 0.00019855773107700867,
      "loss": 0.8615,
      "step": 20880
    },
    {
      "epoch": 0.36675503432293405,
      "grad_norm": 0.052170002929892864,
      "learning_rate": 0.00019855583523832273,
      "loss": 0.8635,
      "step": 20890
    },
    {
      "epoch": 0.36693059920293547,
      "grad_norm": 0.061353774280918726,
      "learning_rate": 0.00019855393816358563,
      "loss": 0.8541,
      "step": 20900
    },
    {
      "epoch": 0.36710616408293684,
      "grad_norm": 0.06546734598510343,
      "learning_rate": 0.00019855203985282146,
      "loss": 0.8658,
      "step": 20910
    },
    {
      "epoch": 0.36728172896293826,
      "grad_norm": 0.08629601450004108,
      "learning_rate": 0.0001985501403060542,
      "loss": 0.8616,
      "step": 20920
    },
    {
      "epoch": 0.3674572938429397,
      "grad_norm": 0.061456611201421756,
      "learning_rate": 0.00019854823952330798,
      "loss": 0.8641,
      "step": 20930
    },
    {
      "epoch": 0.36763285872294105,
      "grad_norm": 0.06567825956902973,
      "learning_rate": 0.0001985463375046069,
      "loss": 0.8591,
      "step": 20940
    },
    {
      "epoch": 0.36780842360294247,
      "grad_norm": 0.06449822421912224,
      "learning_rate": 0.00019854443424997505,
      "loss": 0.8625,
      "step": 20950
    },
    {
      "epoch": 0.3679839884829439,
      "grad_norm": 0.07966224644673892,
      "learning_rate": 0.00019854252975943647,
      "loss": 0.8585,
      "step": 20960
    },
    {
      "epoch": 0.36815955336294526,
      "grad_norm": 0.06390297806942022,
      "learning_rate": 0.0001985406240330154,
      "loss": 0.8594,
      "step": 20970
    },
    {
      "epoch": 0.3683351182429467,
      "grad_norm": 0.07063656669323085,
      "learning_rate": 0.00019853871707073585,
      "loss": 0.8569,
      "step": 20980
    },
    {
      "epoch": 0.3685106831229481,
      "grad_norm": 0.06747201721028045,
      "learning_rate": 0.00019853680887262212,
      "loss": 0.8534,
      "step": 20990
    },
    {
      "epoch": 0.36868624800294947,
      "grad_norm": 0.11369187023359388,
      "learning_rate": 0.00019853489943869832,
      "loss": 0.8575,
      "step": 21000
    },
    {
      "epoch": 0.3688618128829509,
      "grad_norm": 0.06647586403393553,
      "learning_rate": 0.00019853298876898863,
      "loss": 0.8511,
      "step": 21010
    },
    {
      "epoch": 0.3690373777629523,
      "grad_norm": 0.054195001151210224,
      "learning_rate": 0.0001985310768635173,
      "loss": 0.8544,
      "step": 21020
    },
    {
      "epoch": 0.3692129426429537,
      "grad_norm": 0.05363407736344607,
      "learning_rate": 0.00019852916372230852,
      "loss": 0.8645,
      "step": 21030
    },
    {
      "epoch": 0.3693885075229551,
      "grad_norm": 0.08611382219943738,
      "learning_rate": 0.0001985272493453866,
      "loss": 0.8636,
      "step": 21040
    },
    {
      "epoch": 0.3695640724029565,
      "grad_norm": 0.10107962828797619,
      "learning_rate": 0.00019852533373277567,
      "loss": 0.8601,
      "step": 21050
    },
    {
      "epoch": 0.36973963728295794,
      "grad_norm": 0.06454278566893266,
      "learning_rate": 0.0001985234168845001,
      "loss": 0.8569,
      "step": 21060
    },
    {
      "epoch": 0.3699152021629593,
      "grad_norm": 0.07356993685733872,
      "learning_rate": 0.00019852149880058412,
      "loss": 0.854,
      "step": 21070
    },
    {
      "epoch": 0.37009076704296073,
      "grad_norm": 0.051505103504157726,
      "learning_rate": 0.00019851957948105212,
      "loss": 0.8571,
      "step": 21080
    },
    {
      "epoch": 0.37026633192296216,
      "grad_norm": 0.06192958159405417,
      "learning_rate": 0.0001985176589259283,
      "loss": 0.8572,
      "step": 21090
    },
    {
      "epoch": 0.3704418968029635,
      "grad_norm": 0.06626773876582008,
      "learning_rate": 0.00019851573713523707,
      "loss": 0.8645,
      "step": 21100
    },
    {
      "epoch": 0.37061746168296494,
      "grad_norm": 0.08097625460621588,
      "learning_rate": 0.00019851381410900275,
      "loss": 0.8518,
      "step": 21110
    },
    {
      "epoch": 0.37079302656296637,
      "grad_norm": 0.09360490054101624,
      "learning_rate": 0.00019851188984724974,
      "loss": 0.8532,
      "step": 21120
    },
    {
      "epoch": 0.37096859144296773,
      "grad_norm": 0.08246888161615051,
      "learning_rate": 0.00019850996435000242,
      "loss": 0.8599,
      "step": 21130
    },
    {
      "epoch": 0.37114415632296915,
      "grad_norm": 0.068838910529555,
      "learning_rate": 0.00019850803761728512,
      "loss": 0.8493,
      "step": 21140
    },
    {
      "epoch": 0.3713197212029706,
      "grad_norm": 0.0782346845729182,
      "learning_rate": 0.00019850610964912234,
      "loss": 0.8554,
      "step": 21150
    },
    {
      "epoch": 0.37149528608297194,
      "grad_norm": 0.06917961834006454,
      "learning_rate": 0.00019850418044553843,
      "loss": 0.8468,
      "step": 21160
    },
    {
      "epoch": 0.37167085096297336,
      "grad_norm": 0.059836160652915284,
      "learning_rate": 0.00019850225000655786,
      "loss": 0.8561,
      "step": 21170
    },
    {
      "epoch": 0.3718464158429748,
      "grad_norm": 0.06287104467293583,
      "learning_rate": 0.0001985003183322051,
      "loss": 0.8478,
      "step": 21180
    },
    {
      "epoch": 0.37202198072297615,
      "grad_norm": 0.07072310927351382,
      "learning_rate": 0.00019849838542250462,
      "loss": 0.8567,
      "step": 21190
    },
    {
      "epoch": 0.3721975456029776,
      "grad_norm": 0.06533132003157183,
      "learning_rate": 0.0001984964512774809,
      "loss": 0.8669,
      "step": 21200
    },
    {
      "epoch": 0.372373110482979,
      "grad_norm": 0.07364924707063636,
      "learning_rate": 0.00019849451589715848,
      "loss": 0.8506,
      "step": 21210
    },
    {
      "epoch": 0.37254867536298036,
      "grad_norm": 0.0651331914491308,
      "learning_rate": 0.00019849257928156186,
      "loss": 0.8538,
      "step": 21220
    },
    {
      "epoch": 0.3727242402429818,
      "grad_norm": 0.07513384761459455,
      "learning_rate": 0.0001984906414307156,
      "loss": 0.8455,
      "step": 21230
    },
    {
      "epoch": 0.3728998051229832,
      "grad_norm": 0.06373187625050931,
      "learning_rate": 0.0001984887023446442,
      "loss": 0.8601,
      "step": 21240
    },
    {
      "epoch": 0.37307537000298463,
      "grad_norm": 0.05347574943371268,
      "learning_rate": 0.00019848676202337227,
      "loss": 0.8624,
      "step": 21250
    },
    {
      "epoch": 0.373250934882986,
      "grad_norm": 0.06623527392777927,
      "learning_rate": 0.00019848482046692435,
      "loss": 0.8561,
      "step": 21260
    },
    {
      "epoch": 0.3734264997629874,
      "grad_norm": 0.05664131753698621,
      "learning_rate": 0.00019848287767532506,
      "loss": 0.8676,
      "step": 21270
    },
    {
      "epoch": 0.37360206464298884,
      "grad_norm": 0.05824008572306137,
      "learning_rate": 0.0001984809336485991,
      "loss": 0.8597,
      "step": 21280
    },
    {
      "epoch": 0.3737776295229902,
      "grad_norm": 0.07433870865810295,
      "learning_rate": 0.00019847898838677095,
      "loss": 0.8462,
      "step": 21290
    },
    {
      "epoch": 0.37395319440299163,
      "grad_norm": 0.07267265418783667,
      "learning_rate": 0.0001984770418898654,
      "loss": 0.866,
      "step": 21300
    },
    {
      "epoch": 0.37412875928299305,
      "grad_norm": 0.0828913885238659,
      "learning_rate": 0.00019847509415790702,
      "loss": 0.8632,
      "step": 21310
    },
    {
      "epoch": 0.3743043241629944,
      "grad_norm": 0.060059736444323567,
      "learning_rate": 0.0001984731451909205,
      "loss": 0.8559,
      "step": 21320
    },
    {
      "epoch": 0.37447988904299584,
      "grad_norm": 0.06761030589463742,
      "learning_rate": 0.00019847119498893058,
      "loss": 0.8549,
      "step": 21330
    },
    {
      "epoch": 0.37465545392299726,
      "grad_norm": 0.06441231019183383,
      "learning_rate": 0.00019846924355196194,
      "loss": 0.8591,
      "step": 21340
    },
    {
      "epoch": 0.37483101880299863,
      "grad_norm": 0.09252416471506107,
      "learning_rate": 0.0001984672908800393,
      "loss": 0.8601,
      "step": 21350
    },
    {
      "epoch": 0.37500658368300005,
      "grad_norm": 0.07169519748551925,
      "learning_rate": 0.00019846533697318735,
      "loss": 0.8566,
      "step": 21360
    },
    {
      "epoch": 0.37518214856300147,
      "grad_norm": 0.05729009952544185,
      "learning_rate": 0.00019846338183143096,
      "loss": 0.8526,
      "step": 21370
    },
    {
      "epoch": 0.37535771344300284,
      "grad_norm": 0.04695375908852543,
      "learning_rate": 0.00019846142545479483,
      "loss": 0.866,
      "step": 21380
    },
    {
      "epoch": 0.37553327832300426,
      "grad_norm": 0.06351044343605845,
      "learning_rate": 0.00019845946784330376,
      "loss": 0.8544,
      "step": 21390
    },
    {
      "epoch": 0.3757088432030057,
      "grad_norm": 0.06420551934979944,
      "learning_rate": 0.00019845750899698258,
      "loss": 0.8588,
      "step": 21400
    },
    {
      "epoch": 0.37588440808300705,
      "grad_norm": 0.05098219700887573,
      "learning_rate": 0.00019845554891585604,
      "loss": 0.8703,
      "step": 21410
    },
    {
      "epoch": 0.37605997296300847,
      "grad_norm": 0.061368845639855486,
      "learning_rate": 0.00019845358759994905,
      "loss": 0.864,
      "step": 21420
    },
    {
      "epoch": 0.3762355378430099,
      "grad_norm": 0.06839721721981801,
      "learning_rate": 0.0001984516250492864,
      "loss": 0.8572,
      "step": 21430
    },
    {
      "epoch": 0.3764111027230113,
      "grad_norm": 0.09748133537404186,
      "learning_rate": 0.00019844966126389302,
      "loss": 0.8529,
      "step": 21440
    },
    {
      "epoch": 0.3765866676030127,
      "grad_norm": 0.09211973957222723,
      "learning_rate": 0.00019844769624379372,
      "loss": 0.861,
      "step": 21450
    },
    {
      "epoch": 0.3767622324830141,
      "grad_norm": 0.06414265671058293,
      "learning_rate": 0.00019844572998901344,
      "loss": 0.8548,
      "step": 21460
    },
    {
      "epoch": 0.3769377973630155,
      "grad_norm": 0.07306796978890948,
      "learning_rate": 0.0001984437624995771,
      "loss": 0.87,
      "step": 21470
    },
    {
      "epoch": 0.3771133622430169,
      "grad_norm": 0.06974604823330724,
      "learning_rate": 0.0001984417937755096,
      "loss": 0.8542,
      "step": 21480
    },
    {
      "epoch": 0.3772889271230183,
      "grad_norm": 0.0627619367922476,
      "learning_rate": 0.00019843982381683592,
      "loss": 0.8565,
      "step": 21490
    },
    {
      "epoch": 0.37746449200301974,
      "grad_norm": 0.06649293911201172,
      "learning_rate": 0.000198437852623581,
      "loss": 0.8527,
      "step": 21500
    },
    {
      "epoch": 0.3776400568830211,
      "grad_norm": 0.068434194766881,
      "learning_rate": 0.00019843588019576977,
      "loss": 0.8536,
      "step": 21510
    },
    {
      "epoch": 0.3778156217630225,
      "grad_norm": 0.07152553983546923,
      "learning_rate": 0.00019843390653342732,
      "loss": 0.8598,
      "step": 21520
    },
    {
      "epoch": 0.37799118664302395,
      "grad_norm": 0.07897481837040127,
      "learning_rate": 0.00019843193163657854,
      "loss": 0.8635,
      "step": 21530
    },
    {
      "epoch": 0.3781667515230253,
      "grad_norm": 0.0604294316830433,
      "learning_rate": 0.00019842995550524857,
      "loss": 0.8513,
      "step": 21540
    },
    {
      "epoch": 0.37834231640302673,
      "grad_norm": 0.12348898178664286,
      "learning_rate": 0.00019842797813946237,
      "loss": 0.859,
      "step": 21550
    },
    {
      "epoch": 0.37851788128302816,
      "grad_norm": 0.10442615986459526,
      "learning_rate": 0.000198425999539245,
      "loss": 0.8564,
      "step": 21560
    },
    {
      "epoch": 0.3786934461630295,
      "grad_norm": 0.0729075213909541,
      "learning_rate": 0.00019842401970462156,
      "loss": 0.8572,
      "step": 21570
    },
    {
      "epoch": 0.37886901104303095,
      "grad_norm": 0.06698044564629187,
      "learning_rate": 0.00019842203863561713,
      "loss": 0.8588,
      "step": 21580
    },
    {
      "epoch": 0.37904457592303237,
      "grad_norm": 0.05945893581426337,
      "learning_rate": 0.00019842005633225677,
      "loss": 0.8588,
      "step": 21590
    },
    {
      "epoch": 0.37922014080303373,
      "grad_norm": 0.07234279952243321,
      "learning_rate": 0.00019841807279456566,
      "loss": 0.8581,
      "step": 21600
    },
    {
      "epoch": 0.37939570568303516,
      "grad_norm": 0.06653919519593295,
      "learning_rate": 0.00019841608802256887,
      "loss": 0.8643,
      "step": 21610
    },
    {
      "epoch": 0.3795712705630366,
      "grad_norm": 0.08774200752167353,
      "learning_rate": 0.00019841410201629157,
      "loss": 0.8569,
      "step": 21620
    },
    {
      "epoch": 0.379746835443038,
      "grad_norm": 0.07013883991986533,
      "learning_rate": 0.00019841211477575896,
      "loss": 0.8544,
      "step": 21630
    },
    {
      "epoch": 0.37992240032303937,
      "grad_norm": 0.07454836373073823,
      "learning_rate": 0.00019841012630099616,
      "loss": 0.8547,
      "step": 21640
    },
    {
      "epoch": 0.3800979652030408,
      "grad_norm": 0.06913589721720555,
      "learning_rate": 0.0001984081365920284,
      "loss": 0.86,
      "step": 21650
    },
    {
      "epoch": 0.3802735300830422,
      "grad_norm": 0.09597324611263876,
      "learning_rate": 0.00019840614564888088,
      "loss": 0.8578,
      "step": 21660
    },
    {
      "epoch": 0.3804490949630436,
      "grad_norm": 0.05161470921538061,
      "learning_rate": 0.00019840415347157882,
      "loss": 0.8522,
      "step": 21670
    },
    {
      "epoch": 0.380624659843045,
      "grad_norm": 0.067454923160964,
      "learning_rate": 0.00019840216006014748,
      "loss": 0.8496,
      "step": 21680
    },
    {
      "epoch": 0.3808002247230464,
      "grad_norm": 0.06804149091312632,
      "learning_rate": 0.0001984001654146121,
      "loss": 0.852,
      "step": 21690
    },
    {
      "epoch": 0.3809757896030478,
      "grad_norm": 0.08742635083079356,
      "learning_rate": 0.00019839816953499796,
      "loss": 0.8524,
      "step": 21700
    },
    {
      "epoch": 0.3811513544830492,
      "grad_norm": 0.08243442393559469,
      "learning_rate": 0.00019839617242133038,
      "loss": 0.8592,
      "step": 21710
    },
    {
      "epoch": 0.38132691936305063,
      "grad_norm": 0.0658712765669098,
      "learning_rate": 0.00019839417407363456,
      "loss": 0.854,
      "step": 21720
    },
    {
      "epoch": 0.381502484243052,
      "grad_norm": 0.06042899135510992,
      "learning_rate": 0.00019839217449193594,
      "loss": 0.8555,
      "step": 21730
    },
    {
      "epoch": 0.3816780491230534,
      "grad_norm": 0.05263623340379115,
      "learning_rate": 0.0001983901736762598,
      "loss": 0.8575,
      "step": 21740
    },
    {
      "epoch": 0.38185361400305484,
      "grad_norm": 0.061425149239162875,
      "learning_rate": 0.00019838817162663147,
      "loss": 0.8562,
      "step": 21750
    },
    {
      "epoch": 0.3820291788830562,
      "grad_norm": 0.06723810192075078,
      "learning_rate": 0.00019838616834307632,
      "loss": 0.8561,
      "step": 21760
    },
    {
      "epoch": 0.38220474376305763,
      "grad_norm": 0.06120698886984292,
      "learning_rate": 0.0001983841638256198,
      "loss": 0.847,
      "step": 21770
    },
    {
      "epoch": 0.38238030864305905,
      "grad_norm": 0.05112279863930128,
      "learning_rate": 0.00019838215807428723,
      "loss": 0.853,
      "step": 21780
    },
    {
      "epoch": 0.3825558735230605,
      "grad_norm": 0.08475370036103823,
      "learning_rate": 0.0001983801510891041,
      "loss": 0.8541,
      "step": 21790
    },
    {
      "epoch": 0.38273143840306184,
      "grad_norm": 0.08191707729787495,
      "learning_rate": 0.00019837814287009578,
      "loss": 0.8581,
      "step": 21800
    },
    {
      "epoch": 0.38290700328306326,
      "grad_norm": 0.062182770724212605,
      "learning_rate": 0.0001983761334172877,
      "loss": 0.8662,
      "step": 21810
    },
    {
      "epoch": 0.3830825681630647,
      "grad_norm": 0.05955800143147683,
      "learning_rate": 0.00019837412273070534,
      "loss": 0.8691,
      "step": 21820
    },
    {
      "epoch": 0.38325813304306605,
      "grad_norm": 0.05968948344729945,
      "learning_rate": 0.00019837211081037424,
      "loss": 0.8573,
      "step": 21830
    },
    {
      "epoch": 0.3834336979230675,
      "grad_norm": 0.06967898929879139,
      "learning_rate": 0.00019837009765631978,
      "loss": 0.8516,
      "step": 21840
    },
    {
      "epoch": 0.3836092628030689,
      "grad_norm": 0.08202778179809123,
      "learning_rate": 0.00019836808326856753,
      "loss": 0.8575,
      "step": 21850
    },
    {
      "epoch": 0.38378482768307026,
      "grad_norm": 0.07447009445489626,
      "learning_rate": 0.00019836606764714302,
      "loss": 0.8619,
      "step": 21860
    },
    {
      "epoch": 0.3839603925630717,
      "grad_norm": 0.06088996716384084,
      "learning_rate": 0.00019836405079207175,
      "loss": 0.8566,
      "step": 21870
    },
    {
      "epoch": 0.3841359574430731,
      "grad_norm": 0.07542266248653058,
      "learning_rate": 0.00019836203270337933,
      "loss": 0.8632,
      "step": 21880
    },
    {
      "epoch": 0.3843115223230745,
      "grad_norm": 0.0575769895311918,
      "learning_rate": 0.00019836001338109128,
      "loss": 0.8512,
      "step": 21890
    },
    {
      "epoch": 0.3844870872030759,
      "grad_norm": 0.07302798104513017,
      "learning_rate": 0.00019835799282523318,
      "loss": 0.8643,
      "step": 21900
    },
    {
      "epoch": 0.3846626520830773,
      "grad_norm": 0.0777591160239891,
      "learning_rate": 0.00019835597103583068,
      "loss": 0.8596,
      "step": 21910
    },
    {
      "epoch": 0.3848382169630787,
      "grad_norm": 0.07105437521817472,
      "learning_rate": 0.00019835394801290937,
      "loss": 0.8633,
      "step": 21920
    },
    {
      "epoch": 0.3850137818430801,
      "grad_norm": 0.04866051436927219,
      "learning_rate": 0.00019835192375649487,
      "loss": 0.8382,
      "step": 21930
    },
    {
      "epoch": 0.3851893467230815,
      "grad_norm": 0.061221708716225454,
      "learning_rate": 0.00019834989826661285,
      "loss": 0.8593,
      "step": 21940
    },
    {
      "epoch": 0.3853649116030829,
      "grad_norm": 0.059668558885333914,
      "learning_rate": 0.00019834787154328893,
      "loss": 0.8589,
      "step": 21950
    },
    {
      "epoch": 0.3855404764830843,
      "grad_norm": 0.06464661351471321,
      "learning_rate": 0.00019834584358654887,
      "loss": 0.8585,
      "step": 21960
    },
    {
      "epoch": 0.38571604136308574,
      "grad_norm": 0.0699951449185999,
      "learning_rate": 0.00019834381439641827,
      "loss": 0.8567,
      "step": 21970
    },
    {
      "epoch": 0.38589160624308716,
      "grad_norm": 0.0766102167523704,
      "learning_rate": 0.00019834178397292292,
      "loss": 0.8565,
      "step": 21980
    },
    {
      "epoch": 0.3860671711230885,
      "grad_norm": 0.06317028557785732,
      "learning_rate": 0.00019833975231608848,
      "loss": 0.8562,
      "step": 21990
    },
    {
      "epoch": 0.38624273600308995,
      "grad_norm": 0.062228491986103335,
      "learning_rate": 0.00019833771942594076,
      "loss": 0.8518,
      "step": 22000
    },
    {
      "epoch": 0.38641830088309137,
      "grad_norm": 0.07941131441190327,
      "learning_rate": 0.00019833568530250545,
      "loss": 0.8634,
      "step": 22010
    },
    {
      "epoch": 0.38659386576309274,
      "grad_norm": 0.08101349752320992,
      "learning_rate": 0.00019833364994580835,
      "loss": 0.8598,
      "step": 22020
    },
    {
      "epoch": 0.38676943064309416,
      "grad_norm": 0.08730677003659564,
      "learning_rate": 0.00019833161335587528,
      "loss": 0.8576,
      "step": 22030
    },
    {
      "epoch": 0.3869449955230956,
      "grad_norm": 0.07674069550870512,
      "learning_rate": 0.000198329575532732,
      "loss": 0.8679,
      "step": 22040
    },
    {
      "epoch": 0.38712056040309695,
      "grad_norm": 0.08539629431414497,
      "learning_rate": 0.00019832753647640433,
      "loss": 0.8606,
      "step": 22050
    },
    {
      "epoch": 0.38729612528309837,
      "grad_norm": 0.06752679539515945,
      "learning_rate": 0.00019832549618691817,
      "loss": 0.8593,
      "step": 22060
    },
    {
      "epoch": 0.3874716901630998,
      "grad_norm": 0.08454307736531158,
      "learning_rate": 0.00019832345466429924,
      "loss": 0.8533,
      "step": 22070
    },
    {
      "epoch": 0.38764725504310116,
      "grad_norm": 0.08392256778495387,
      "learning_rate": 0.00019832141190857358,
      "loss": 0.8552,
      "step": 22080
    },
    {
      "epoch": 0.3878228199231026,
      "grad_norm": 0.07883678788043542,
      "learning_rate": 0.00019831936791976694,
      "loss": 0.8568,
      "step": 22090
    },
    {
      "epoch": 0.387998384803104,
      "grad_norm": 0.06000413353228194,
      "learning_rate": 0.0001983173226979052,
      "loss": 0.8474,
      "step": 22100
    },
    {
      "epoch": 0.38817394968310537,
      "grad_norm": 0.06310719080970618,
      "learning_rate": 0.00019831527624301437,
      "loss": 0.8515,
      "step": 22110
    },
    {
      "epoch": 0.3883495145631068,
      "grad_norm": 0.06069011599970246,
      "learning_rate": 0.00019831322855512037,
      "loss": 0.8522,
      "step": 22120
    },
    {
      "epoch": 0.3885250794431082,
      "grad_norm": 0.07004895507552614,
      "learning_rate": 0.00019831117963424908,
      "loss": 0.8578,
      "step": 22130
    },
    {
      "epoch": 0.3887006443231096,
      "grad_norm": 0.07915452781639763,
      "learning_rate": 0.00019830912948042652,
      "loss": 0.8547,
      "step": 22140
    },
    {
      "epoch": 0.388876209203111,
      "grad_norm": 0.0737491989109706,
      "learning_rate": 0.00019830707809367863,
      "loss": 0.858,
      "step": 22150
    },
    {
      "epoch": 0.3890517740831124,
      "grad_norm": 0.05073966065280343,
      "learning_rate": 0.00019830502547403143,
      "loss": 0.8622,
      "step": 22160
    },
    {
      "epoch": 0.38922733896311384,
      "grad_norm": 0.056876122013873404,
      "learning_rate": 0.00019830297162151087,
      "loss": 0.8551,
      "step": 22170
    },
    {
      "epoch": 0.3894029038431152,
      "grad_norm": 0.08502826017405937,
      "learning_rate": 0.000198300916536143,
      "loss": 0.8574,
      "step": 22180
    },
    {
      "epoch": 0.38957846872311663,
      "grad_norm": 0.07577126998531585,
      "learning_rate": 0.00019829886021795393,
      "loss": 0.8488,
      "step": 22190
    },
    {
      "epoch": 0.38975403360311806,
      "grad_norm": 0.06777178163113586,
      "learning_rate": 0.00019829680266696963,
      "loss": 0.8591,
      "step": 22200
    },
    {
      "epoch": 0.3899295984831194,
      "grad_norm": 0.0833149312326189,
      "learning_rate": 0.0001982947438832162,
      "loss": 0.8595,
      "step": 22210
    },
    {
      "epoch": 0.39010516336312084,
      "grad_norm": 0.09386928974660287,
      "learning_rate": 0.00019829268386671971,
      "loss": 0.857,
      "step": 22220
    },
    {
      "epoch": 0.39028072824312227,
      "grad_norm": 0.10050378435036543,
      "learning_rate": 0.00019829062261750627,
      "loss": 0.862,
      "step": 22230
    },
    {
      "epoch": 0.39045629312312363,
      "grad_norm": 0.05842977046027029,
      "learning_rate": 0.000198288560135602,
      "loss": 0.8676,
      "step": 22240
    },
    {
      "epoch": 0.39063185800312505,
      "grad_norm": 0.05776647969457122,
      "learning_rate": 0.00019828649642103302,
      "loss": 0.8602,
      "step": 22250
    },
    {
      "epoch": 0.3908074228831265,
      "grad_norm": 0.06435581739826694,
      "learning_rate": 0.00019828443147382546,
      "loss": 0.8607,
      "step": 22260
    },
    {
      "epoch": 0.39098298776312784,
      "grad_norm": 0.06046419926844219,
      "learning_rate": 0.00019828236529400556,
      "loss": 0.8581,
      "step": 22270
    },
    {
      "epoch": 0.39115855264312926,
      "grad_norm": 0.07230999646359845,
      "learning_rate": 0.0001982802978815994,
      "loss": 0.86,
      "step": 22280
    },
    {
      "epoch": 0.3913341175231307,
      "grad_norm": 0.07432619716115772,
      "learning_rate": 0.00019827822923663326,
      "loss": 0.8581,
      "step": 22290
    },
    {
      "epoch": 0.39150968240313205,
      "grad_norm": 0.06391885689081613,
      "learning_rate": 0.00019827615935913328,
      "loss": 0.853,
      "step": 22300
    },
    {
      "epoch": 0.3916852472831335,
      "grad_norm": 0.07544428942790078,
      "learning_rate": 0.00019827408824912575,
      "loss": 0.8641,
      "step": 22310
    },
    {
      "epoch": 0.3918608121631349,
      "grad_norm": 0.08299436477060301,
      "learning_rate": 0.00019827201590663684,
      "loss": 0.8575,
      "step": 22320
    },
    {
      "epoch": 0.39203637704313626,
      "grad_norm": 0.06463450008870716,
      "learning_rate": 0.0001982699423316929,
      "loss": 0.8593,
      "step": 22330
    },
    {
      "epoch": 0.3922119419231377,
      "grad_norm": 0.06684675988496451,
      "learning_rate": 0.00019826786752432009,
      "loss": 0.8563,
      "step": 22340
    },
    {
      "epoch": 0.3923875068031391,
      "grad_norm": 0.06631351982775205,
      "learning_rate": 0.0001982657914845448,
      "loss": 0.8604,
      "step": 22350
    },
    {
      "epoch": 0.39256307168314053,
      "grad_norm": 0.06694191320663952,
      "learning_rate": 0.00019826371421239328,
      "loss": 0.8561,
      "step": 22360
    },
    {
      "epoch": 0.3927386365631419,
      "grad_norm": 0.07105473389214038,
      "learning_rate": 0.00019826163570789185,
      "loss": 0.8534,
      "step": 22370
    },
    {
      "epoch": 0.3929142014431433,
      "grad_norm": 0.09865078226629627,
      "learning_rate": 0.00019825955597106687,
      "loss": 0.8633,
      "step": 22380
    },
    {
      "epoch": 0.39308976632314474,
      "grad_norm": 0.06563134360469504,
      "learning_rate": 0.00019825747500194467,
      "loss": 0.8516,
      "step": 22390
    },
    {
      "epoch": 0.3932653312031461,
      "grad_norm": 0.08067568616037082,
      "learning_rate": 0.0001982553928005516,
      "loss": 0.8506,
      "step": 22400
    },
    {
      "epoch": 0.39344089608314753,
      "grad_norm": 0.07496220565530316,
      "learning_rate": 0.0001982533093669141,
      "loss": 0.8496,
      "step": 22410
    },
    {
      "epoch": 0.39361646096314895,
      "grad_norm": 0.06445599681336384,
      "learning_rate": 0.0001982512247010585,
      "loss": 0.8565,
      "step": 22420
    },
    {
      "epoch": 0.3937920258431503,
      "grad_norm": 0.07657176962003114,
      "learning_rate": 0.00019824913880301127,
      "loss": 0.8491,
      "step": 22430
    },
    {
      "epoch": 0.39396759072315174,
      "grad_norm": 0.06882316680340718,
      "learning_rate": 0.00019824705167279883,
      "loss": 0.8539,
      "step": 22440
    },
    {
      "epoch": 0.39414315560315316,
      "grad_norm": 0.05671388949866655,
      "learning_rate": 0.00019824496331044756,
      "loss": 0.8563,
      "step": 22450
    },
    {
      "epoch": 0.39431872048315453,
      "grad_norm": 0.07974667067703264,
      "learning_rate": 0.000198242873715984,
      "loss": 0.8594,
      "step": 22460
    },
    {
      "epoch": 0.39449428536315595,
      "grad_norm": 0.08188042810165573,
      "learning_rate": 0.00019824078288943457,
      "loss": 0.8538,
      "step": 22470
    },
    {
      "epoch": 0.39466985024315737,
      "grad_norm": 0.0733855376192077,
      "learning_rate": 0.0001982386908308258,
      "loss": 0.8584,
      "step": 22480
    },
    {
      "epoch": 0.39484541512315874,
      "grad_norm": 0.06036357156630562,
      "learning_rate": 0.00019823659754018418,
      "loss": 0.8635,
      "step": 22490
    },
    {
      "epoch": 0.39502098000316016,
      "grad_norm": 0.06661629026615645,
      "learning_rate": 0.00019823450301753625,
      "loss": 0.8564,
      "step": 22500
    },
    {
      "epoch": 0.3951965448831616,
      "grad_norm": 0.06231781656975422,
      "learning_rate": 0.00019823240726290852,
      "loss": 0.8561,
      "step": 22510
    },
    {
      "epoch": 0.395372109763163,
      "grad_norm": 0.06096240379003488,
      "learning_rate": 0.00019823031027632756,
      "loss": 0.8565,
      "step": 22520
    },
    {
      "epoch": 0.39554767464316437,
      "grad_norm": 0.09393680630296437,
      "learning_rate": 0.00019822821205781993,
      "loss": 0.8535,
      "step": 22530
    },
    {
      "epoch": 0.3957232395231658,
      "grad_norm": 0.05414479707211924,
      "learning_rate": 0.0001982261126074122,
      "loss": 0.8596,
      "step": 22540
    },
    {
      "epoch": 0.3958988044031672,
      "grad_norm": 0.14926894859863835,
      "learning_rate": 0.000198224011925131,
      "loss": 0.8441,
      "step": 22550
    },
    {
      "epoch": 0.3960743692831686,
      "grad_norm": 0.059054374111373424,
      "learning_rate": 0.00019822191001100297,
      "loss": 0.8644,
      "step": 22560
    },
    {
      "epoch": 0.39624993416317,
      "grad_norm": 0.07342881158190863,
      "learning_rate": 0.00019821980686505466,
      "loss": 0.8573,
      "step": 22570
    },
    {
      "epoch": 0.3964254990431714,
      "grad_norm": 0.08730806487700825,
      "learning_rate": 0.00019821770248731283,
      "loss": 0.8601,
      "step": 22580
    },
    {
      "epoch": 0.3966010639231728,
      "grad_norm": 0.07579893119437758,
      "learning_rate": 0.000198215596877804,
      "loss": 0.8607,
      "step": 22590
    },
    {
      "epoch": 0.3967766288031742,
      "grad_norm": 0.06688473267349664,
      "learning_rate": 0.00019821349003655496,
      "loss": 0.8569,
      "step": 22600
    },
    {
      "epoch": 0.39695219368317564,
      "grad_norm": 0.12019625690946904,
      "learning_rate": 0.00019821138196359235,
      "loss": 0.8521,
      "step": 22610
    },
    {
      "epoch": 0.397127758563177,
      "grad_norm": 0.07462574683190502,
      "learning_rate": 0.00019820927265894295,
      "loss": 0.859,
      "step": 22620
    },
    {
      "epoch": 0.3973033234431784,
      "grad_norm": 0.08351665320837431,
      "learning_rate": 0.0001982071621226334,
      "loss": 0.8663,
      "step": 22630
    },
    {
      "epoch": 0.39747888832317985,
      "grad_norm": 0.06282382221284032,
      "learning_rate": 0.0001982050503546905,
      "loss": 0.85,
      "step": 22640
    },
    {
      "epoch": 0.3976544532031812,
      "grad_norm": 0.0761864320388309,
      "learning_rate": 0.00019820293735514093,
      "loss": 0.8541,
      "step": 22650
    },
    {
      "epoch": 0.39783001808318263,
      "grad_norm": 0.09154468750326782,
      "learning_rate": 0.00019820082312401157,
      "loss": 0.8494,
      "step": 22660
    },
    {
      "epoch": 0.39800558296318406,
      "grad_norm": 0.06062876496083987,
      "learning_rate": 0.00019819870766132913,
      "loss": 0.8577,
      "step": 22670
    },
    {
      "epoch": 0.3981811478431854,
      "grad_norm": 0.06463480748692994,
      "learning_rate": 0.0001981965909671204,
      "loss": 0.8555,
      "step": 22680
    },
    {
      "epoch": 0.39835671272318685,
      "grad_norm": 0.06335097480852701,
      "learning_rate": 0.0001981944730414123,
      "loss": 0.8605,
      "step": 22690
    },
    {
      "epoch": 0.39853227760318827,
      "grad_norm": 0.05695027002412183,
      "learning_rate": 0.00019819235388423157,
      "loss": 0.8648,
      "step": 22700
    },
    {
      "epoch": 0.3987078424831897,
      "grad_norm": 0.0838035235423582,
      "learning_rate": 0.0001981902334956051,
      "loss": 0.8485,
      "step": 22710
    },
    {
      "epoch": 0.39888340736319106,
      "grad_norm": 0.07827330896906573,
      "learning_rate": 0.00019818811187555975,
      "loss": 0.854,
      "step": 22720
    },
    {
      "epoch": 0.3990589722431925,
      "grad_norm": 0.09509872634496237,
      "learning_rate": 0.00019818598902412237,
      "loss": 0.8533,
      "step": 22730
    },
    {
      "epoch": 0.3992345371231939,
      "grad_norm": 0.06138830162742917,
      "learning_rate": 0.0001981838649413199,
      "loss": 0.8575,
      "step": 22740
    },
    {
      "epoch": 0.39941010200319527,
      "grad_norm": 0.06111454213300817,
      "learning_rate": 0.00019818173962717924,
      "loss": 0.859,
      "step": 22750
    },
    {
      "epoch": 0.3995856668831967,
      "grad_norm": 0.06751742782045675,
      "learning_rate": 0.00019817961308172727,
      "loss": 0.8699,
      "step": 22760
    },
    {
      "epoch": 0.3997612317631981,
      "grad_norm": 0.059671829447977956,
      "learning_rate": 0.00019817748530499103,
      "loss": 0.8533,
      "step": 22770
    },
    {
      "epoch": 0.3999367966431995,
      "grad_norm": 0.047728806272552125,
      "learning_rate": 0.00019817535629699742,
      "loss": 0.861,
      "step": 22780
    },
    {
      "epoch": 0.4001123615232009,
      "grad_norm": 0.08476539347533435,
      "learning_rate": 0.00019817322605777339,
      "loss": 0.8515,
      "step": 22790
    },
    {
      "epoch": 0.4002879264032023,
      "grad_norm": 0.06427466247493895,
      "learning_rate": 0.00019817109458734598,
      "loss": 0.8586,
      "step": 22800
    },
    {
      "epoch": 0.4004634912832037,
      "grad_norm": 0.058135872337657137,
      "learning_rate": 0.0001981689618857422,
      "loss": 0.8649,
      "step": 22810
    },
    {
      "epoch": 0.4006390561632051,
      "grad_norm": 0.05866619117213911,
      "learning_rate": 0.00019816682795298905,
      "loss": 0.8576,
      "step": 22820
    },
    {
      "epoch": 0.40081462104320653,
      "grad_norm": 0.1164585560052983,
      "learning_rate": 0.00019816469278911356,
      "loss": 0.8561,
      "step": 22830
    },
    {
      "epoch": 0.4009901859232079,
      "grad_norm": 0.05889345340400301,
      "learning_rate": 0.00019816255639414277,
      "loss": 0.8581,
      "step": 22840
    },
    {
      "epoch": 0.4011657508032093,
      "grad_norm": 0.07101314173848436,
      "learning_rate": 0.0001981604187681038,
      "loss": 0.8677,
      "step": 22850
    },
    {
      "epoch": 0.40134131568321074,
      "grad_norm": 0.0635159803219572,
      "learning_rate": 0.00019815827991102368,
      "loss": 0.8463,
      "step": 22860
    },
    {
      "epoch": 0.4015168805632121,
      "grad_norm": 0.06035743295557798,
      "learning_rate": 0.00019815613982292953,
      "loss": 0.8527,
      "step": 22870
    },
    {
      "epoch": 0.40169244544321353,
      "grad_norm": 0.06510149830873506,
      "learning_rate": 0.0001981539985038485,
      "loss": 0.8608,
      "step": 22880
    },
    {
      "epoch": 0.40186801032321495,
      "grad_norm": 0.08104299272364077,
      "learning_rate": 0.00019815185595380765,
      "loss": 0.8584,
      "step": 22890
    },
    {
      "epoch": 0.4020435752032164,
      "grad_norm": 0.08294908559809008,
      "learning_rate": 0.0001981497121728342,
      "loss": 0.8553,
      "step": 22900
    },
    {
      "epoch": 0.40221914008321774,
      "grad_norm": 0.054311103935317795,
      "learning_rate": 0.00019814756716095527,
      "loss": 0.8617,
      "step": 22910
    },
    {
      "epoch": 0.40239470496321916,
      "grad_norm": 0.06874681358245362,
      "learning_rate": 0.00019814542091819802,
      "loss": 0.848,
      "step": 22920
    },
    {
      "epoch": 0.4025702698432206,
      "grad_norm": 0.08566011020106062,
      "learning_rate": 0.0001981432734445897,
      "loss": 0.8619,
      "step": 22930
    },
    {
      "epoch": 0.40274583472322195,
      "grad_norm": 0.06129256068942554,
      "learning_rate": 0.00019814112474015748,
      "loss": 0.863,
      "step": 22940
    },
    {
      "epoch": 0.4029213996032234,
      "grad_norm": 0.06576799839777153,
      "learning_rate": 0.00019813897480492856,
      "loss": 0.8493,
      "step": 22950
    },
    {
      "epoch": 0.4030969644832248,
      "grad_norm": 0.06379649249292493,
      "learning_rate": 0.00019813682363893024,
      "loss": 0.8494,
      "step": 22960
    },
    {
      "epoch": 0.40327252936322616,
      "grad_norm": 0.05767916102258527,
      "learning_rate": 0.00019813467124218974,
      "loss": 0.8577,
      "step": 22970
    },
    {
      "epoch": 0.4034480942432276,
      "grad_norm": 0.09955837052707728,
      "learning_rate": 0.00019813251761473437,
      "loss": 0.8577,
      "step": 22980
    },
    {
      "epoch": 0.403623659123229,
      "grad_norm": 0.07192973605523031,
      "learning_rate": 0.00019813036275659137,
      "loss": 0.8677,
      "step": 22990
    },
    {
      "epoch": 0.4037992240032304,
      "grad_norm": 0.07040881016822204,
      "learning_rate": 0.00019812820666778805,
      "loss": 0.8615,
      "step": 23000
    },
    {
      "epoch": 0.4039747888832318,
      "grad_norm": 0.06434516149443532,
      "learning_rate": 0.00019812604934835176,
      "loss": 0.859,
      "step": 23010
    },
    {
      "epoch": 0.4041503537632332,
      "grad_norm": 0.0728057841446281,
      "learning_rate": 0.0001981238907983098,
      "loss": 0.8629,
      "step": 23020
    },
    {
      "epoch": 0.4043259186432346,
      "grad_norm": 0.06446347140312143,
      "learning_rate": 0.0001981217310176895,
      "loss": 0.856,
      "step": 23030
    },
    {
      "epoch": 0.404501483523236,
      "grad_norm": 0.06610854442636781,
      "learning_rate": 0.00019811957000651827,
      "loss": 0.8616,
      "step": 23040
    },
    {
      "epoch": 0.4046770484032374,
      "grad_norm": 0.07597897873744168,
      "learning_rate": 0.0001981174077648235,
      "loss": 0.8611,
      "step": 23050
    },
    {
      "epoch": 0.4048526132832388,
      "grad_norm": 0.08304798114336405,
      "learning_rate": 0.00019811524429263252,
      "loss": 0.856,
      "step": 23060
    },
    {
      "epoch": 0.4050281781632402,
      "grad_norm": 0.06690170552580567,
      "learning_rate": 0.0001981130795899728,
      "loss": 0.8633,
      "step": 23070
    },
    {
      "epoch": 0.40520374304324164,
      "grad_norm": 0.06214814791360192,
      "learning_rate": 0.00019811091365687174,
      "loss": 0.8597,
      "step": 23080
    },
    {
      "epoch": 0.40537930792324306,
      "grad_norm": 0.0682531381496847,
      "learning_rate": 0.00019810874649335683,
      "loss": 0.8604,
      "step": 23090
    },
    {
      "epoch": 0.4055548728032444,
      "grad_norm": 0.07244788784277467,
      "learning_rate": 0.00019810657809945546,
      "loss": 0.8569,
      "step": 23100
    },
    {
      "epoch": 0.40573043768324585,
      "grad_norm": 0.053262554953853104,
      "learning_rate": 0.00019810440847519516,
      "loss": 0.8486,
      "step": 23110
    },
    {
      "epoch": 0.40590600256324727,
      "grad_norm": 0.06049155014562685,
      "learning_rate": 0.0001981022376206034,
      "loss": 0.8599,
      "step": 23120
    },
    {
      "epoch": 0.40608156744324864,
      "grad_norm": 0.07696558092019924,
      "learning_rate": 0.00019810006553570763,
      "loss": 0.8545,
      "step": 23130
    },
    {
      "epoch": 0.40625713232325006,
      "grad_norm": 0.08459049347914498,
      "learning_rate": 0.00019809789222053546,
      "loss": 0.8471,
      "step": 23140
    },
    {
      "epoch": 0.4064326972032515,
      "grad_norm": 0.07671822724747164,
      "learning_rate": 0.00019809571767511437,
      "loss": 0.8593,
      "step": 23150
    },
    {
      "epoch": 0.40660826208325285,
      "grad_norm": 0.08290844686529619,
      "learning_rate": 0.00019809354189947195,
      "loss": 0.8521,
      "step": 23160
    },
    {
      "epoch": 0.40678382696325427,
      "grad_norm": 0.06079841000562817,
      "learning_rate": 0.00019809136489363574,
      "loss": 0.8587,
      "step": 23170
    },
    {
      "epoch": 0.4069593918432557,
      "grad_norm": 0.06199859737677302,
      "learning_rate": 0.00019808918665763329,
      "loss": 0.8634,
      "step": 23180
    },
    {
      "epoch": 0.40713495672325706,
      "grad_norm": 0.06752971372726763,
      "learning_rate": 0.00019808700719149228,
      "loss": 0.8534,
      "step": 23190
    },
    {
      "epoch": 0.4073105216032585,
      "grad_norm": 0.05583540013997792,
      "learning_rate": 0.00019808482649524029,
      "loss": 0.8592,
      "step": 23200
    },
    {
      "epoch": 0.4074860864832599,
      "grad_norm": 0.08183497137954088,
      "learning_rate": 0.0001980826445689049,
      "loss": 0.8527,
      "step": 23210
    },
    {
      "epoch": 0.40766165136326127,
      "grad_norm": 0.06773310707747451,
      "learning_rate": 0.00019808046141251383,
      "loss": 0.8553,
      "step": 23220
    },
    {
      "epoch": 0.4078372162432627,
      "grad_norm": 0.06276681207705936,
      "learning_rate": 0.0001980782770260947,
      "loss": 0.8567,
      "step": 23230
    },
    {
      "epoch": 0.4080127811232641,
      "grad_norm": 0.06662299174138643,
      "learning_rate": 0.00019807609140967518,
      "loss": 0.8571,
      "step": 23240
    },
    {
      "epoch": 0.40818834600326553,
      "grad_norm": 0.08451406513935299,
      "learning_rate": 0.00019807390456328297,
      "loss": 0.8571,
      "step": 23250
    },
    {
      "epoch": 0.4083639108832669,
      "grad_norm": 0.055398298747036834,
      "learning_rate": 0.00019807171648694583,
      "loss": 0.8476,
      "step": 23260
    },
    {
      "epoch": 0.4085394757632683,
      "grad_norm": 0.07378088655209167,
      "learning_rate": 0.00019806952718069143,
      "loss": 0.8567,
      "step": 23270
    },
    {
      "epoch": 0.40871504064326974,
      "grad_norm": 0.07688034950647044,
      "learning_rate": 0.00019806733664454748,
      "loss": 0.8696,
      "step": 23280
    },
    {
      "epoch": 0.4088906055232711,
      "grad_norm": 0.057814325841919374,
      "learning_rate": 0.0001980651448785418,
      "loss": 0.8604,
      "step": 23290
    },
    {
      "epoch": 0.40906617040327253,
      "grad_norm": 0.07056077654422281,
      "learning_rate": 0.00019806295188270215,
      "loss": 0.8577,
      "step": 23300
    },
    {
      "epoch": 0.40924173528327396,
      "grad_norm": 0.05818722992951602,
      "learning_rate": 0.00019806075765705625,
      "loss": 0.8459,
      "step": 23310
    },
    {
      "epoch": 0.4094173001632753,
      "grad_norm": 0.08275198239650826,
      "learning_rate": 0.000198058562201632,
      "loss": 0.8522,
      "step": 23320
    },
    {
      "epoch": 0.40959286504327674,
      "grad_norm": 0.06688531292575486,
      "learning_rate": 0.0001980563655164571,
      "loss": 0.8532,
      "step": 23330
    },
    {
      "epoch": 0.40976842992327817,
      "grad_norm": 0.0554119671518036,
      "learning_rate": 0.00019805416760155947,
      "loss": 0.8577,
      "step": 23340
    },
    {
      "epoch": 0.40994399480327953,
      "grad_norm": 0.052037903308334826,
      "learning_rate": 0.000198051968456967,
      "loss": 0.8611,
      "step": 23350
    },
    {
      "epoch": 0.41011955968328095,
      "grad_norm": 0.08773803305464657,
      "learning_rate": 0.0001980497680827074,
      "loss": 0.8583,
      "step": 23360
    },
    {
      "epoch": 0.4102951245632824,
      "grad_norm": 0.10496633317947089,
      "learning_rate": 0.00019804756647880868,
      "loss": 0.8664,
      "step": 23370
    },
    {
      "epoch": 0.41047068944328374,
      "grad_norm": 0.07247083140870463,
      "learning_rate": 0.0001980453636452987,
      "loss": 0.8525,
      "step": 23380
    },
    {
      "epoch": 0.41064625432328516,
      "grad_norm": 0.05729821609218918,
      "learning_rate": 0.00019804315958220537,
      "loss": 0.8538,
      "step": 23390
    },
    {
      "epoch": 0.4108218192032866,
      "grad_norm": 0.08807434262535596,
      "learning_rate": 0.0001980409542895566,
      "loss": 0.8604,
      "step": 23400
    },
    {
      "epoch": 0.41099738408328795,
      "grad_norm": 0.07372357930039504,
      "learning_rate": 0.00019803874776738034,
      "loss": 0.8563,
      "step": 23410
    },
    {
      "epoch": 0.4111729489632894,
      "grad_norm": 0.0983103111612344,
      "learning_rate": 0.00019803654001570456,
      "loss": 0.8482,
      "step": 23420
    },
    {
      "epoch": 0.4113485138432908,
      "grad_norm": 0.0633742441224773,
      "learning_rate": 0.0001980343310345572,
      "loss": 0.8551,
      "step": 23430
    },
    {
      "epoch": 0.4115240787232922,
      "grad_norm": 0.06111901912931253,
      "learning_rate": 0.00019803212082396626,
      "loss": 0.8564,
      "step": 23440
    },
    {
      "epoch": 0.4116996436032936,
      "grad_norm": 0.06890547478422795,
      "learning_rate": 0.00019802990938395977,
      "loss": 0.8615,
      "step": 23450
    },
    {
      "epoch": 0.411875208483295,
      "grad_norm": 0.06676508183337429,
      "learning_rate": 0.00019802769671456577,
      "loss": 0.8573,
      "step": 23460
    },
    {
      "epoch": 0.41205077336329643,
      "grad_norm": 0.057337838080242474,
      "learning_rate": 0.0001980254828158122,
      "loss": 0.8526,
      "step": 23470
    },
    {
      "epoch": 0.4122263382432978,
      "grad_norm": 0.06936469850866271,
      "learning_rate": 0.0001980232676877272,
      "loss": 0.8556,
      "step": 23480
    },
    {
      "epoch": 0.4124019031232992,
      "grad_norm": 0.07213550100036975,
      "learning_rate": 0.0001980210513303388,
      "loss": 0.8565,
      "step": 23490
    },
    {
      "epoch": 0.41257746800330064,
      "grad_norm": 0.06588595064391055,
      "learning_rate": 0.00019801883374367507,
      "loss": 0.852,
      "step": 23500
    },
    {
      "epoch": 0.412753032883302,
      "grad_norm": 0.06792203364631942,
      "learning_rate": 0.00019801661492776411,
      "loss": 0.8616,
      "step": 23510
    },
    {
      "epoch": 0.41292859776330343,
      "grad_norm": 0.06392304677940171,
      "learning_rate": 0.0001980143948826341,
      "loss": 0.857,
      "step": 23520
    },
    {
      "epoch": 0.41310416264330485,
      "grad_norm": 0.06556644318623903,
      "learning_rate": 0.00019801217360831307,
      "loss": 0.8504,
      "step": 23530
    },
    {
      "epoch": 0.4132797275233062,
      "grad_norm": 0.058328382888417414,
      "learning_rate": 0.00019800995110482922,
      "loss": 0.8546,
      "step": 23540
    },
    {
      "epoch": 0.41345529240330764,
      "grad_norm": 0.04814563181147783,
      "learning_rate": 0.0001980077273722107,
      "loss": 0.858,
      "step": 23550
    },
    {
      "epoch": 0.41363085728330906,
      "grad_norm": 0.0778680371884089,
      "learning_rate": 0.00019800550241048568,
      "loss": 0.8562,
      "step": 23560
    },
    {
      "epoch": 0.41380642216331043,
      "grad_norm": 0.07468572742662002,
      "learning_rate": 0.00019800327621968236,
      "loss": 0.852,
      "step": 23570
    },
    {
      "epoch": 0.41398198704331185,
      "grad_norm": 0.05148370277107974,
      "learning_rate": 0.0001980010487998289,
      "loss": 0.855,
      "step": 23580
    },
    {
      "epoch": 0.41415755192331327,
      "grad_norm": 0.0996252565913877,
      "learning_rate": 0.00019799882015095362,
      "loss": 0.8484,
      "step": 23590
    },
    {
      "epoch": 0.41433311680331464,
      "grad_norm": 0.09525798434414152,
      "learning_rate": 0.00019799659027308465,
      "loss": 0.8574,
      "step": 23600
    },
    {
      "epoch": 0.41450868168331606,
      "grad_norm": 0.10306924100013512,
      "learning_rate": 0.00019799435916625032,
      "loss": 0.8542,
      "step": 23610
    },
    {
      "epoch": 0.4146842465633175,
      "grad_norm": 0.06390334941123062,
      "learning_rate": 0.0001979921268304789,
      "loss": 0.8524,
      "step": 23620
    },
    {
      "epoch": 0.4148598114433189,
      "grad_norm": 0.07701891841391194,
      "learning_rate": 0.0001979898932657986,
      "loss": 0.85,
      "step": 23630
    },
    {
      "epoch": 0.41503537632332027,
      "grad_norm": 0.09441398741056316,
      "learning_rate": 0.00019798765847223778,
      "loss": 0.8486,
      "step": 23640
    },
    {
      "epoch": 0.4152109412033217,
      "grad_norm": 0.05119082010188207,
      "learning_rate": 0.00019798542244982474,
      "loss": 0.8615,
      "step": 23650
    },
    {
      "epoch": 0.4153865060833231,
      "grad_norm": 0.05558081175280894,
      "learning_rate": 0.0001979831851985878,
      "loss": 0.8568,
      "step": 23660
    },
    {
      "epoch": 0.4155620709633245,
      "grad_norm": 0.06970046611831716,
      "learning_rate": 0.00019798094671855533,
      "loss": 0.8566,
      "step": 23670
    },
    {
      "epoch": 0.4157376358433259,
      "grad_norm": 0.07299640180291771,
      "learning_rate": 0.00019797870700975564,
      "loss": 0.8594,
      "step": 23680
    },
    {
      "epoch": 0.4159132007233273,
      "grad_norm": 0.06899319342372974,
      "learning_rate": 0.0001979764660722172,
      "loss": 0.8503,
      "step": 23690
    },
    {
      "epoch": 0.4160887656033287,
      "grad_norm": 0.07570135450171334,
      "learning_rate": 0.00019797422390596832,
      "loss": 0.8531,
      "step": 23700
    },
    {
      "epoch": 0.4162643304833301,
      "grad_norm": 0.06165927780660987,
      "learning_rate": 0.00019797198051103745,
      "loss": 0.8657,
      "step": 23710
    },
    {
      "epoch": 0.41643989536333154,
      "grad_norm": 0.07822418431374392,
      "learning_rate": 0.00019796973588745299,
      "loss": 0.866,
      "step": 23720
    },
    {
      "epoch": 0.4166154602433329,
      "grad_norm": 0.05807130075435765,
      "learning_rate": 0.0001979674900352434,
      "loss": 0.8645,
      "step": 23730
    },
    {
      "epoch": 0.4167910251233343,
      "grad_norm": 0.04995825621534916,
      "learning_rate": 0.00019796524295443712,
      "loss": 0.849,
      "step": 23740
    },
    {
      "epoch": 0.41696659000333575,
      "grad_norm": 0.06756585321124141,
      "learning_rate": 0.00019796299464506264,
      "loss": 0.8536,
      "step": 23750
    },
    {
      "epoch": 0.4171421548833371,
      "grad_norm": 0.06853167609873134,
      "learning_rate": 0.0001979607451071484,
      "loss": 0.8577,
      "step": 23760
    },
    {
      "epoch": 0.41731771976333853,
      "grad_norm": 0.07273940248672528,
      "learning_rate": 0.000197958494340723,
      "loss": 0.8583,
      "step": 23770
    },
    {
      "epoch": 0.41749328464333996,
      "grad_norm": 0.049919461322572155,
      "learning_rate": 0.00019795624234581486,
      "loss": 0.8582,
      "step": 23780
    },
    {
      "epoch": 0.4176688495233413,
      "grad_norm": 0.0563739857635172,
      "learning_rate": 0.00019795398912245254,
      "loss": 0.8559,
      "step": 23790
    },
    {
      "epoch": 0.41784441440334275,
      "grad_norm": 0.07150476180406265,
      "learning_rate": 0.00019795173467066458,
      "loss": 0.8554,
      "step": 23800
    },
    {
      "epoch": 0.41801997928334417,
      "grad_norm": 0.09252759659178214,
      "learning_rate": 0.00019794947899047963,
      "loss": 0.857,
      "step": 23810
    },
    {
      "epoch": 0.4181955441633456,
      "grad_norm": 0.06186291373486963,
      "learning_rate": 0.00019794722208192617,
      "loss": 0.8532,
      "step": 23820
    },
    {
      "epoch": 0.41837110904334696,
      "grad_norm": 0.11780658268729939,
      "learning_rate": 0.00019794496394503276,
      "loss": 0.8605,
      "step": 23830
    },
    {
      "epoch": 0.4185466739233484,
      "grad_norm": 0.0690304600417895,
      "learning_rate": 0.00019794270457982818,
      "loss": 0.8611,
      "step": 23840
    },
    {
      "epoch": 0.4187222388033498,
      "grad_norm": 0.0663812311824687,
      "learning_rate": 0.00019794044398634088,
      "loss": 0.8556,
      "step": 23850
    },
    {
      "epoch": 0.41889780368335117,
      "grad_norm": 0.11918954700709052,
      "learning_rate": 0.0001979381821645996,
      "loss": 0.8617,
      "step": 23860
    },
    {
      "epoch": 0.4190733685633526,
      "grad_norm": 0.07570436378591087,
      "learning_rate": 0.00019793591911463298,
      "loss": 0.853,
      "step": 23870
    },
    {
      "epoch": 0.419248933443354,
      "grad_norm": 0.07316220843400868,
      "learning_rate": 0.00019793365483646968,
      "loss": 0.8539,
      "step": 23880
    },
    {
      "epoch": 0.4194244983233554,
      "grad_norm": 0.06552309597771855,
      "learning_rate": 0.0001979313893301384,
      "loss": 0.8576,
      "step": 23890
    },
    {
      "epoch": 0.4196000632033568,
      "grad_norm": 0.07657497613704695,
      "learning_rate": 0.00019792912259566784,
      "loss": 0.8537,
      "step": 23900
    },
    {
      "epoch": 0.4197756280833582,
      "grad_norm": 0.06425175581475873,
      "learning_rate": 0.0001979268546330867,
      "loss": 0.8521,
      "step": 23910
    },
    {
      "epoch": 0.4199511929633596,
      "grad_norm": 0.0859316711663665,
      "learning_rate": 0.00019792458544242377,
      "loss": 0.8533,
      "step": 23920
    },
    {
      "epoch": 0.420126757843361,
      "grad_norm": 0.06280677874823465,
      "learning_rate": 0.00019792231502370778,
      "loss": 0.8522,
      "step": 23930
    },
    {
      "epoch": 0.42030232272336243,
      "grad_norm": 0.07362405229663191,
      "learning_rate": 0.00019792004337696744,
      "loss": 0.855,
      "step": 23940
    },
    {
      "epoch": 0.4204778876033638,
      "grad_norm": 0.10061992476206615,
      "learning_rate": 0.00019791777050223158,
      "loss": 0.8511,
      "step": 23950
    },
    {
      "epoch": 0.4206534524833652,
      "grad_norm": 0.06408570687176796,
      "learning_rate": 0.00019791549639952902,
      "loss": 0.8502,
      "step": 23960
    },
    {
      "epoch": 0.42082901736336664,
      "grad_norm": 0.06815544627426995,
      "learning_rate": 0.0001979132210688885,
      "loss": 0.8522,
      "step": 23970
    },
    {
      "epoch": 0.42100458224336806,
      "grad_norm": 0.07302871359728337,
      "learning_rate": 0.00019791094451033895,
      "loss": 0.8537,
      "step": 23980
    },
    {
      "epoch": 0.42118014712336943,
      "grad_norm": 0.06933045327223912,
      "learning_rate": 0.00019790866672390913,
      "loss": 0.8615,
      "step": 23990
    },
    {
      "epoch": 0.42135571200337085,
      "grad_norm": 0.09606811423185657,
      "learning_rate": 0.00019790638770962794,
      "loss": 0.8451,
      "step": 24000
    },
    {
      "epoch": 0.4215312768833723,
      "grad_norm": 0.08293317891856797,
      "learning_rate": 0.00019790410746752427,
      "loss": 0.8557,
      "step": 24010
    },
    {
      "epoch": 0.42170684176337364,
      "grad_norm": 0.08624193602795534,
      "learning_rate": 0.00019790182599762693,
      "loss": 0.8536,
      "step": 24020
    },
    {
      "epoch": 0.42188240664337506,
      "grad_norm": 0.06311840964815427,
      "learning_rate": 0.0001978995432999649,
      "loss": 0.8555,
      "step": 24030
    },
    {
      "epoch": 0.4220579715233765,
      "grad_norm": 0.05856072285662311,
      "learning_rate": 0.00019789725937456715,
      "loss": 0.8514,
      "step": 24040
    },
    {
      "epoch": 0.42223353640337785,
      "grad_norm": 0.059146073676464815,
      "learning_rate": 0.00019789497422146248,
      "loss": 0.8561,
      "step": 24050
    },
    {
      "epoch": 0.4224091012833793,
      "grad_norm": 0.07215384845719484,
      "learning_rate": 0.00019789268784067994,
      "loss": 0.8539,
      "step": 24060
    },
    {
      "epoch": 0.4225846661633807,
      "grad_norm": 0.0641524063642619,
      "learning_rate": 0.00019789040023224848,
      "loss": 0.8577,
      "step": 24070
    },
    {
      "epoch": 0.42276023104338206,
      "grad_norm": 0.06857747151826586,
      "learning_rate": 0.00019788811139619708,
      "loss": 0.8555,
      "step": 24080
    },
    {
      "epoch": 0.4229357959233835,
      "grad_norm": 0.0622270363425274,
      "learning_rate": 0.00019788582133255472,
      "loss": 0.8623,
      "step": 24090
    },
    {
      "epoch": 0.4231113608033849,
      "grad_norm": 0.05556365992968483,
      "learning_rate": 0.00019788353004135045,
      "loss": 0.8534,
      "step": 24100
    },
    {
      "epoch": 0.4232869256833863,
      "grad_norm": 0.06184879503846866,
      "learning_rate": 0.00019788123752261327,
      "loss": 0.8519,
      "step": 24110
    },
    {
      "epoch": 0.4234624905633877,
      "grad_norm": 0.08369938904904543,
      "learning_rate": 0.00019787894377637226,
      "loss": 0.8544,
      "step": 24120
    },
    {
      "epoch": 0.4236380554433891,
      "grad_norm": 0.06664902020457696,
      "learning_rate": 0.0001978766488026565,
      "loss": 0.8552,
      "step": 24130
    },
    {
      "epoch": 0.4238136203233905,
      "grad_norm": 0.07079367411446787,
      "learning_rate": 0.000197874352601495,
      "loss": 0.8606,
      "step": 24140
    },
    {
      "epoch": 0.4239891852033919,
      "grad_norm": 0.07556567699870627,
      "learning_rate": 0.00019787205517291689,
      "loss": 0.8575,
      "step": 24150
    },
    {
      "epoch": 0.4241647500833933,
      "grad_norm": 0.07569680511052704,
      "learning_rate": 0.00019786975651695126,
      "loss": 0.849,
      "step": 24160
    },
    {
      "epoch": 0.42434031496339475,
      "grad_norm": 0.08157585971048129,
      "learning_rate": 0.00019786745663362727,
      "loss": 0.8575,
      "step": 24170
    },
    {
      "epoch": 0.4245158798433961,
      "grad_norm": 0.046014629669633336,
      "learning_rate": 0.00019786515552297403,
      "loss": 0.8566,
      "step": 24180
    },
    {
      "epoch": 0.42469144472339754,
      "grad_norm": 0.0661294329747174,
      "learning_rate": 0.0001978628531850207,
      "loss": 0.8569,
      "step": 24190
    },
    {
      "epoch": 0.42486700960339896,
      "grad_norm": 0.07763000325479355,
      "learning_rate": 0.00019786054961979646,
      "loss": 0.8464,
      "step": 24200
    },
    {
      "epoch": 0.4250425744834003,
      "grad_norm": 0.07449889897229006,
      "learning_rate": 0.0001978582448273305,
      "loss": 0.8556,
      "step": 24210
    },
    {
      "epoch": 0.42521813936340175,
      "grad_norm": 0.05360801043814137,
      "learning_rate": 0.00019785593880765202,
      "loss": 0.8596,
      "step": 24220
    },
    {
      "epoch": 0.42539370424340317,
      "grad_norm": 0.06589127621124076,
      "learning_rate": 0.00019785363156079021,
      "loss": 0.8566,
      "step": 24230
    },
    {
      "epoch": 0.42556926912340454,
      "grad_norm": 0.06407633192586366,
      "learning_rate": 0.00019785132308677438,
      "loss": 0.8507,
      "step": 24240
    },
    {
      "epoch": 0.42574483400340596,
      "grad_norm": 0.07199896498509271,
      "learning_rate": 0.00019784901338563363,
      "loss": 0.8514,
      "step": 24250
    },
    {
      "epoch": 0.4259203988834074,
      "grad_norm": 0.05660652428213764,
      "learning_rate": 0.0001978467024573974,
      "loss": 0.8595,
      "step": 24260
    },
    {
      "epoch": 0.42609596376340875,
      "grad_norm": 0.05541259350959857,
      "learning_rate": 0.00019784439030209484,
      "loss": 0.8508,
      "step": 24270
    },
    {
      "epoch": 0.42627152864341017,
      "grad_norm": 0.05970333823542039,
      "learning_rate": 0.00019784207691975534,
      "loss": 0.867,
      "step": 24280
    },
    {
      "epoch": 0.4264470935234116,
      "grad_norm": 0.06804668792554719,
      "learning_rate": 0.00019783976231040816,
      "loss": 0.8565,
      "step": 24290
    },
    {
      "epoch": 0.42662265840341296,
      "grad_norm": 0.10429915867835482,
      "learning_rate": 0.0001978374464740826,
      "loss": 0.8483,
      "step": 24300
    },
    {
      "epoch": 0.4267982232834144,
      "grad_norm": 0.05944180200713313,
      "learning_rate": 0.00019783512941080807,
      "loss": 0.8538,
      "step": 24310
    },
    {
      "epoch": 0.4269737881634158,
      "grad_norm": 0.05330582456695055,
      "learning_rate": 0.00019783281112061384,
      "loss": 0.8638,
      "step": 24320
    },
    {
      "epoch": 0.42714935304341717,
      "grad_norm": 0.05397754504394192,
      "learning_rate": 0.0001978304916035294,
      "loss": 0.8604,
      "step": 24330
    },
    {
      "epoch": 0.4273249179234186,
      "grad_norm": 0.05588666231909027,
      "learning_rate": 0.00019782817085958404,
      "loss": 0.8606,
      "step": 24340
    },
    {
      "epoch": 0.42750048280342,
      "grad_norm": 0.08578383176992048,
      "learning_rate": 0.0001978258488888072,
      "loss": 0.8574,
      "step": 24350
    },
    {
      "epoch": 0.42767604768342143,
      "grad_norm": 0.06438015996916513,
      "learning_rate": 0.0001978235256912283,
      "loss": 0.8559,
      "step": 24360
    },
    {
      "epoch": 0.4278516125634228,
      "grad_norm": 0.07389407006925401,
      "learning_rate": 0.00019782120126687674,
      "loss": 0.8598,
      "step": 24370
    },
    {
      "epoch": 0.4280271774434242,
      "grad_norm": 0.0719480130256265,
      "learning_rate": 0.00019781887561578208,
      "loss": 0.8595,
      "step": 24380
    },
    {
      "epoch": 0.42820274232342564,
      "grad_norm": 0.06631055254040422,
      "learning_rate": 0.00019781654873797367,
      "loss": 0.8577,
      "step": 24390
    },
    {
      "epoch": 0.428378307203427,
      "grad_norm": 0.06158103187432808,
      "learning_rate": 0.00019781422063348104,
      "loss": 0.8492,
      "step": 24400
    },
    {
      "epoch": 0.42855387208342843,
      "grad_norm": 0.05956110387560395,
      "learning_rate": 0.00019781189130233366,
      "loss": 0.8576,
      "step": 24410
    },
    {
      "epoch": 0.42872943696342986,
      "grad_norm": 0.08201524126342943,
      "learning_rate": 0.00019780956074456107,
      "loss": 0.8527,
      "step": 24420
    },
    {
      "epoch": 0.4289050018434312,
      "grad_norm": 0.08272910798076559,
      "learning_rate": 0.0001978072289601928,
      "loss": 0.8533,
      "step": 24430
    },
    {
      "epoch": 0.42908056672343264,
      "grad_norm": 0.07973105340418299,
      "learning_rate": 0.00019780489594925838,
      "loss": 0.8546,
      "step": 24440
    },
    {
      "epoch": 0.42925613160343407,
      "grad_norm": 0.05737166704713955,
      "learning_rate": 0.0001978025617117874,
      "loss": 0.853,
      "step": 24450
    },
    {
      "epoch": 0.42943169648343543,
      "grad_norm": 0.08757422527546417,
      "learning_rate": 0.00019780022624780937,
      "loss": 0.863,
      "step": 24460
    },
    {
      "epoch": 0.42960726136343685,
      "grad_norm": 0.06569540755359904,
      "learning_rate": 0.00019779788955735396,
      "loss": 0.8547,
      "step": 24470
    },
    {
      "epoch": 0.4297828262434383,
      "grad_norm": 0.09818175352279988,
      "learning_rate": 0.00019779555164045073,
      "loss": 0.8554,
      "step": 24480
    },
    {
      "epoch": 0.42995839112343964,
      "grad_norm": 0.066972859743233,
      "learning_rate": 0.00019779321249712925,
      "loss": 0.849,
      "step": 24490
    },
    {
      "epoch": 0.43013395600344106,
      "grad_norm": 0.06215760135249169,
      "learning_rate": 0.0001977908721274193,
      "loss": 0.862,
      "step": 24500
    },
    {
      "epoch": 0.4303095208834425,
      "grad_norm": 0.06780205736970095,
      "learning_rate": 0.0001977885305313504,
      "loss": 0.8589,
      "step": 24510
    },
    {
      "epoch": 0.43048508576344385,
      "grad_norm": 0.07969857247895247,
      "learning_rate": 0.00019778618770895226,
      "loss": 0.8557,
      "step": 24520
    },
    {
      "epoch": 0.4306606506434453,
      "grad_norm": 0.05729831808063077,
      "learning_rate": 0.00019778384366025463,
      "loss": 0.8659,
      "step": 24530
    },
    {
      "epoch": 0.4308362155234467,
      "grad_norm": 0.08522407823094233,
      "learning_rate": 0.00019778149838528707,
      "loss": 0.8569,
      "step": 24540
    },
    {
      "epoch": 0.4310117804034481,
      "grad_norm": 0.06291241280985671,
      "learning_rate": 0.00019777915188407946,
      "loss": 0.8476,
      "step": 24550
    },
    {
      "epoch": 0.4311873452834495,
      "grad_norm": 0.064078293755528,
      "learning_rate": 0.0001977768041566614,
      "loss": 0.8569,
      "step": 24560
    },
    {
      "epoch": 0.4313629101634509,
      "grad_norm": 0.07934553902905082,
      "learning_rate": 0.0001977744552030627,
      "loss": 0.8614,
      "step": 24570
    },
    {
      "epoch": 0.43153847504345233,
      "grad_norm": 0.06585583672621256,
      "learning_rate": 0.0001977721050233131,
      "loss": 0.8577,
      "step": 24580
    },
    {
      "epoch": 0.4317140399234537,
      "grad_norm": 0.08440985147938669,
      "learning_rate": 0.00019776975361744241,
      "loss": 0.8607,
      "step": 24590
    },
    {
      "epoch": 0.4318896048034551,
      "grad_norm": 0.07097727632745177,
      "learning_rate": 0.00019776740098548037,
      "loss": 0.8579,
      "step": 24600
    },
    {
      "epoch": 0.43206516968345654,
      "grad_norm": 0.05153926768134646,
      "learning_rate": 0.00019776504712745683,
      "loss": 0.8558,
      "step": 24610
    },
    {
      "epoch": 0.4322407345634579,
      "grad_norm": 0.06209039092930526,
      "learning_rate": 0.0001977626920434016,
      "loss": 0.8637,
      "step": 24620
    },
    {
      "epoch": 0.43241629944345933,
      "grad_norm": 0.066468961530916,
      "learning_rate": 0.00019776033573334448,
      "loss": 0.8539,
      "step": 24630
    },
    {
      "epoch": 0.43259186432346075,
      "grad_norm": 0.07869399171339564,
      "learning_rate": 0.0001977579781973154,
      "loss": 0.8691,
      "step": 24640
    },
    {
      "epoch": 0.4327674292034621,
      "grad_norm": 0.06666957214204834,
      "learning_rate": 0.00019775561943534417,
      "loss": 0.8532,
      "step": 24650
    },
    {
      "epoch": 0.43294299408346354,
      "grad_norm": 0.05606149810756,
      "learning_rate": 0.0001977532594474607,
      "loss": 0.8595,
      "step": 24660
    },
    {
      "epoch": 0.43311855896346496,
      "grad_norm": 0.06433590181686363,
      "learning_rate": 0.00019775089823369487,
      "loss": 0.8584,
      "step": 24670
    },
    {
      "epoch": 0.43329412384346633,
      "grad_norm": 0.06980280422862005,
      "learning_rate": 0.00019774853579407667,
      "loss": 0.8553,
      "step": 24680
    },
    {
      "epoch": 0.43346968872346775,
      "grad_norm": 0.07591350929000831,
      "learning_rate": 0.00019774617212863594,
      "loss": 0.8562,
      "step": 24690
    },
    {
      "epoch": 0.43364525360346917,
      "grad_norm": 0.07207398510280955,
      "learning_rate": 0.0001977438072374027,
      "loss": 0.851,
      "step": 24700
    },
    {
      "epoch": 0.4338208184834706,
      "grad_norm": 0.07377362253681613,
      "learning_rate": 0.00019774144112040686,
      "loss": 0.8557,
      "step": 24710
    },
    {
      "epoch": 0.43399638336347196,
      "grad_norm": 0.07853337544157342,
      "learning_rate": 0.0001977390737776784,
      "loss": 0.857,
      "step": 24720
    },
    {
      "epoch": 0.4341719482434734,
      "grad_norm": 0.06417901015734073,
      "learning_rate": 0.0001977367052092474,
      "loss": 0.8487,
      "step": 24730
    },
    {
      "epoch": 0.4343475131234748,
      "grad_norm": 0.08066824091943287,
      "learning_rate": 0.00019773433541514374,
      "loss": 0.8554,
      "step": 24740
    },
    {
      "epoch": 0.43452307800347617,
      "grad_norm": 0.07794390508993439,
      "learning_rate": 0.00019773196439539756,
      "loss": 0.8526,
      "step": 24750
    },
    {
      "epoch": 0.4346986428834776,
      "grad_norm": 0.06903696165643244,
      "learning_rate": 0.0001977295921500388,
      "loss": 0.8532,
      "step": 24760
    },
    {
      "epoch": 0.434874207763479,
      "grad_norm": 0.05875076967552137,
      "learning_rate": 0.00019772721867909762,
      "loss": 0.8551,
      "step": 24770
    },
    {
      "epoch": 0.4350497726434804,
      "grad_norm": 0.04352285108165102,
      "learning_rate": 0.00019772484398260403,
      "loss": 0.8601,
      "step": 24780
    },
    {
      "epoch": 0.4352253375234818,
      "grad_norm": 0.058180277393845424,
      "learning_rate": 0.0001977224680605881,
      "loss": 0.8563,
      "step": 24790
    },
    {
      "epoch": 0.4354009024034832,
      "grad_norm": 0.07881379105896263,
      "learning_rate": 0.00019772009091307999,
      "loss": 0.8545,
      "step": 24800
    },
    {
      "epoch": 0.4355764672834846,
      "grad_norm": 0.0724083084779048,
      "learning_rate": 0.0001977177125401098,
      "loss": 0.8519,
      "step": 24810
    },
    {
      "epoch": 0.435752032163486,
      "grad_norm": 0.0605052291363654,
      "learning_rate": 0.0001977153329417076,
      "loss": 0.862,
      "step": 24820
    },
    {
      "epoch": 0.43592759704348744,
      "grad_norm": 0.0749075434916469,
      "learning_rate": 0.00019771295211790365,
      "loss": 0.856,
      "step": 24830
    },
    {
      "epoch": 0.4361031619234888,
      "grad_norm": 0.0738487440995315,
      "learning_rate": 0.00019771057006872804,
      "loss": 0.8589,
      "step": 24840
    },
    {
      "epoch": 0.4362787268034902,
      "grad_norm": 0.062425040392400495,
      "learning_rate": 0.000197708186794211,
      "loss": 0.8549,
      "step": 24850
    },
    {
      "epoch": 0.43645429168349165,
      "grad_norm": 0.08140037484256551,
      "learning_rate": 0.00019770580229438268,
      "loss": 0.8507,
      "step": 24860
    },
    {
      "epoch": 0.436629856563493,
      "grad_norm": 0.057053769269239205,
      "learning_rate": 0.0001977034165692733,
      "loss": 0.8623,
      "step": 24870
    },
    {
      "epoch": 0.43680542144349443,
      "grad_norm": 0.08048341813481333,
      "learning_rate": 0.0001977010296189131,
      "loss": 0.8595,
      "step": 24880
    },
    {
      "epoch": 0.43698098632349586,
      "grad_norm": 0.08047620808888586,
      "learning_rate": 0.00019769864144333235,
      "loss": 0.8588,
      "step": 24890
    },
    {
      "epoch": 0.4371565512034973,
      "grad_norm": 0.07657836732185166,
      "learning_rate": 0.00019769625204256125,
      "loss": 0.8659,
      "step": 24900
    },
    {
      "epoch": 0.43733211608349865,
      "grad_norm": 0.06542083584403642,
      "learning_rate": 0.00019769386141663014,
      "loss": 0.8507,
      "step": 24910
    },
    {
      "epoch": 0.43750768096350007,
      "grad_norm": 0.07001080707681787,
      "learning_rate": 0.00019769146956556924,
      "loss": 0.86,
      "step": 24920
    },
    {
      "epoch": 0.4376832458435015,
      "grad_norm": 0.06424802703006366,
      "learning_rate": 0.0001976890764894089,
      "loss": 0.8542,
      "step": 24930
    },
    {
      "epoch": 0.43785881072350286,
      "grad_norm": 0.06748385489499933,
      "learning_rate": 0.00019768668218817943,
      "loss": 0.8548,
      "step": 24940
    },
    {
      "epoch": 0.4380343756035043,
      "grad_norm": 0.06929843485827702,
      "learning_rate": 0.00019768428666191117,
      "loss": 0.8536,
      "step": 24950
    },
    {
      "epoch": 0.4382099404835057,
      "grad_norm": 0.0467498812454557,
      "learning_rate": 0.00019768188991063447,
      "loss": 0.8525,
      "step": 24960
    },
    {
      "epoch": 0.43838550536350707,
      "grad_norm": 0.06193147680844693,
      "learning_rate": 0.00019767949193437968,
      "loss": 0.8467,
      "step": 24970
    },
    {
      "epoch": 0.4385610702435085,
      "grad_norm": 0.05893836999071017,
      "learning_rate": 0.00019767709273317726,
      "loss": 0.849,
      "step": 24980
    },
    {
      "epoch": 0.4387366351235099,
      "grad_norm": 0.08345229110028242,
      "learning_rate": 0.00019767469230705747,
      "loss": 0.8597,
      "step": 24990
    },
    {
      "epoch": 0.4389122000035113,
      "grad_norm": 0.0712881624204767,
      "learning_rate": 0.00019767229065605085,
      "loss": 0.8591,
      "step": 25000
    },
    {
      "epoch": 0.4390877648835127,
      "grad_norm": 0.06603382551614456,
      "learning_rate": 0.00019766988778018772,
      "loss": 0.8564,
      "step": 25010
    },
    {
      "epoch": 0.4392633297635141,
      "grad_norm": 0.0842370316671665,
      "learning_rate": 0.00019766748367949866,
      "loss": 0.8519,
      "step": 25020
    },
    {
      "epoch": 0.4394388946435155,
      "grad_norm": 0.06305657141043847,
      "learning_rate": 0.000197665078354014,
      "loss": 0.8519,
      "step": 25030
    },
    {
      "epoch": 0.4396144595235169,
      "grad_norm": 0.0704706897482048,
      "learning_rate": 0.0001976626718037643,
      "loss": 0.8558,
      "step": 25040
    },
    {
      "epoch": 0.43979002440351833,
      "grad_norm": 0.07219280069631484,
      "learning_rate": 0.00019766026402878001,
      "loss": 0.8476,
      "step": 25050
    },
    {
      "epoch": 0.4399655892835197,
      "grad_norm": 0.060793173723770286,
      "learning_rate": 0.00019765785502909165,
      "loss": 0.8554,
      "step": 25060
    },
    {
      "epoch": 0.4401411541635211,
      "grad_norm": 0.0640619126753617,
      "learning_rate": 0.00019765544480472976,
      "loss": 0.855,
      "step": 25070
    },
    {
      "epoch": 0.44031671904352254,
      "grad_norm": 0.07624633153976491,
      "learning_rate": 0.00019765303335572484,
      "loss": 0.8611,
      "step": 25080
    },
    {
      "epoch": 0.44049228392352396,
      "grad_norm": 0.07693357677421128,
      "learning_rate": 0.00019765062068210747,
      "loss": 0.8544,
      "step": 25090
    },
    {
      "epoch": 0.44066784880352533,
      "grad_norm": 0.06625169153846602,
      "learning_rate": 0.00019764820678390822,
      "loss": 0.8585,
      "step": 25100
    },
    {
      "epoch": 0.44084341368352675,
      "grad_norm": 0.067079881020378,
      "learning_rate": 0.00019764579166115767,
      "loss": 0.8582,
      "step": 25110
    },
    {
      "epoch": 0.4410189785635282,
      "grad_norm": 0.06620556557505951,
      "learning_rate": 0.0001976433753138864,
      "loss": 0.8635,
      "step": 25120
    },
    {
      "epoch": 0.44119454344352954,
      "grad_norm": 0.058579842998731535,
      "learning_rate": 0.00019764095774212507,
      "loss": 0.8527,
      "step": 25130
    },
    {
      "epoch": 0.44137010832353096,
      "grad_norm": 0.05750106598772731,
      "learning_rate": 0.00019763853894590424,
      "loss": 0.8583,
      "step": 25140
    },
    {
      "epoch": 0.4415456732035324,
      "grad_norm": 0.1491795576181004,
      "learning_rate": 0.00019763611892525466,
      "loss": 0.8576,
      "step": 25150
    },
    {
      "epoch": 0.44172123808353375,
      "grad_norm": 0.08800184052776842,
      "learning_rate": 0.00019763369768020688,
      "loss": 0.8594,
      "step": 25160
    },
    {
      "epoch": 0.4418968029635352,
      "grad_norm": 0.13714712337089133,
      "learning_rate": 0.00019763127521079163,
      "loss": 0.8619,
      "step": 25170
    },
    {
      "epoch": 0.4420723678435366,
      "grad_norm": 0.06827995960148658,
      "learning_rate": 0.00019762885151703964,
      "loss": 0.8592,
      "step": 25180
    },
    {
      "epoch": 0.44224793272353796,
      "grad_norm": 0.08026320836140254,
      "learning_rate": 0.0001976264265989816,
      "loss": 0.8558,
      "step": 25190
    },
    {
      "epoch": 0.4424234976035394,
      "grad_norm": 0.06525160807289564,
      "learning_rate": 0.00019762400045664818,
      "loss": 0.8488,
      "step": 25200
    },
    {
      "epoch": 0.4425990624835408,
      "grad_norm": 0.09080795514481625,
      "learning_rate": 0.00019762157309007015,
      "loss": 0.8561,
      "step": 25210
    },
    {
      "epoch": 0.4427746273635422,
      "grad_norm": 0.09034003716921575,
      "learning_rate": 0.00019761914449927827,
      "loss": 0.8529,
      "step": 25220
    },
    {
      "epoch": 0.4429501922435436,
      "grad_norm": 0.058979142958439014,
      "learning_rate": 0.00019761671468430334,
      "loss": 0.8559,
      "step": 25230
    },
    {
      "epoch": 0.443125757123545,
      "grad_norm": 0.07504305655782251,
      "learning_rate": 0.0001976142836451761,
      "loss": 0.8559,
      "step": 25240
    },
    {
      "epoch": 0.4433013220035464,
      "grad_norm": 0.06808968071728645,
      "learning_rate": 0.00019761185138192738,
      "loss": 0.8612,
      "step": 25250
    },
    {
      "epoch": 0.4434768868835478,
      "grad_norm": 0.056719983063948945,
      "learning_rate": 0.00019760941789458798,
      "loss": 0.8468,
      "step": 25260
    },
    {
      "epoch": 0.4436524517635492,
      "grad_norm": 0.06982229476127343,
      "learning_rate": 0.00019760698318318874,
      "loss": 0.8533,
      "step": 25270
    },
    {
      "epoch": 0.44382801664355065,
      "grad_norm": 0.06469580772177011,
      "learning_rate": 0.00019760454724776052,
      "loss": 0.8538,
      "step": 25280
    },
    {
      "epoch": 0.444003581523552,
      "grad_norm": 0.06502776524755217,
      "learning_rate": 0.00019760211008833413,
      "loss": 0.8442,
      "step": 25290
    },
    {
      "epoch": 0.44417914640355344,
      "grad_norm": 0.059322857751290876,
      "learning_rate": 0.00019759967170494055,
      "loss": 0.8598,
      "step": 25300
    },
    {
      "epoch": 0.44435471128355486,
      "grad_norm": 0.0697747681813306,
      "learning_rate": 0.0001975972320976106,
      "loss": 0.8594,
      "step": 25310
    },
    {
      "epoch": 0.4445302761635562,
      "grad_norm": 0.05632276779391216,
      "learning_rate": 0.00019759479126637517,
      "loss": 0.8613,
      "step": 25320
    },
    {
      "epoch": 0.44470584104355765,
      "grad_norm": 0.06558599804187533,
      "learning_rate": 0.00019759234921126523,
      "loss": 0.8553,
      "step": 25330
    },
    {
      "epoch": 0.44488140592355907,
      "grad_norm": 0.06306544317595814,
      "learning_rate": 0.00019758990593231175,
      "loss": 0.8582,
      "step": 25340
    },
    {
      "epoch": 0.44505697080356044,
      "grad_norm": 0.060705804720247336,
      "learning_rate": 0.00019758746142954563,
      "loss": 0.8505,
      "step": 25350
    },
    {
      "epoch": 0.44523253568356186,
      "grad_norm": 0.0696494391347431,
      "learning_rate": 0.0001975850157029979,
      "loss": 0.8589,
      "step": 25360
    },
    {
      "epoch": 0.4454081005635633,
      "grad_norm": 0.06679106668523975,
      "learning_rate": 0.00019758256875269944,
      "loss": 0.8581,
      "step": 25370
    },
    {
      "epoch": 0.44558366544356465,
      "grad_norm": 0.06706216017145482,
      "learning_rate": 0.0001975801205786814,
      "loss": 0.8585,
      "step": 25380
    },
    {
      "epoch": 0.44575923032356607,
      "grad_norm": 0.055244073934571036,
      "learning_rate": 0.00019757767118097465,
      "loss": 0.8458,
      "step": 25390
    },
    {
      "epoch": 0.4459347952035675,
      "grad_norm": 0.0800053013612004,
      "learning_rate": 0.00019757522055961033,
      "loss": 0.853,
      "step": 25400
    },
    {
      "epoch": 0.44611036008356886,
      "grad_norm": 0.062089781646429115,
      "learning_rate": 0.00019757276871461944,
      "loss": 0.8544,
      "step": 25410
    },
    {
      "epoch": 0.4462859249635703,
      "grad_norm": 0.061152738431304275,
      "learning_rate": 0.0001975703156460331,
      "loss": 0.865,
      "step": 25420
    },
    {
      "epoch": 0.4464614898435717,
      "grad_norm": 0.0606194377806782,
      "learning_rate": 0.00019756786135388232,
      "loss": 0.8521,
      "step": 25430
    },
    {
      "epoch": 0.4466370547235731,
      "grad_norm": 0.06931937526935042,
      "learning_rate": 0.00019756540583819825,
      "loss": 0.8621,
      "step": 25440
    },
    {
      "epoch": 0.4468126196035745,
      "grad_norm": 0.06297235433212978,
      "learning_rate": 0.00019756294909901198,
      "loss": 0.8586,
      "step": 25450
    },
    {
      "epoch": 0.4469881844835759,
      "grad_norm": 0.06696533071229899,
      "learning_rate": 0.00019756049113635462,
      "loss": 0.8474,
      "step": 25460
    },
    {
      "epoch": 0.44716374936357733,
      "grad_norm": 0.06908007197884486,
      "learning_rate": 0.00019755803195025735,
      "loss": 0.8524,
      "step": 25470
    },
    {
      "epoch": 0.4473393142435787,
      "grad_norm": 0.06397453880434888,
      "learning_rate": 0.00019755557154075128,
      "loss": 0.8565,
      "step": 25480
    },
    {
      "epoch": 0.4475148791235801,
      "grad_norm": 0.056189337422617905,
      "learning_rate": 0.00019755310990786766,
      "loss": 0.8593,
      "step": 25490
    },
    {
      "epoch": 0.44769044400358154,
      "grad_norm": 0.06394630411033435,
      "learning_rate": 0.00019755064705163762,
      "loss": 0.8486,
      "step": 25500
    },
    {
      "epoch": 0.4478660088835829,
      "grad_norm": 0.0791930741062487,
      "learning_rate": 0.00019754818297209234,
      "loss": 0.8651,
      "step": 25510
    },
    {
      "epoch": 0.44804157376358433,
      "grad_norm": 0.080293269794989,
      "learning_rate": 0.00019754571766926314,
      "loss": 0.8591,
      "step": 25520
    },
    {
      "epoch": 0.44821713864358576,
      "grad_norm": 0.07119814004271376,
      "learning_rate": 0.00019754325114318115,
      "loss": 0.851,
      "step": 25530
    },
    {
      "epoch": 0.4483927035235871,
      "grad_norm": 0.0647928172840431,
      "learning_rate": 0.00019754078339387764,
      "loss": 0.8578,
      "step": 25540
    },
    {
      "epoch": 0.44856826840358854,
      "grad_norm": 0.06020072054287736,
      "learning_rate": 0.00019753831442138395,
      "loss": 0.8562,
      "step": 25550
    },
    {
      "epoch": 0.44874383328358997,
      "grad_norm": 0.057633668422608676,
      "learning_rate": 0.00019753584422573133,
      "loss": 0.8602,
      "step": 25560
    },
    {
      "epoch": 0.44891939816359133,
      "grad_norm": 0.08223096601723473,
      "learning_rate": 0.00019753337280695103,
      "loss": 0.8649,
      "step": 25570
    },
    {
      "epoch": 0.44909496304359275,
      "grad_norm": 0.08187224625018363,
      "learning_rate": 0.0001975309001650744,
      "loss": 0.8622,
      "step": 25580
    },
    {
      "epoch": 0.4492705279235942,
      "grad_norm": 0.06923147383890592,
      "learning_rate": 0.00019752842630013276,
      "loss": 0.8584,
      "step": 25590
    },
    {
      "epoch": 0.44944609280359554,
      "grad_norm": 0.0667927222782772,
      "learning_rate": 0.00019752595121215748,
      "loss": 0.8546,
      "step": 25600
    },
    {
      "epoch": 0.44962165768359696,
      "grad_norm": 0.12862016482726973,
      "learning_rate": 0.00019752347490117986,
      "loss": 0.8578,
      "step": 25610
    },
    {
      "epoch": 0.4497972225635984,
      "grad_norm": 0.06599878704117441,
      "learning_rate": 0.00019752099736723136,
      "loss": 0.8602,
      "step": 25620
    },
    {
      "epoch": 0.4499727874435998,
      "grad_norm": 0.06941542330152119,
      "learning_rate": 0.0001975185186103433,
      "loss": 0.8492,
      "step": 25630
    },
    {
      "epoch": 0.4501483523236012,
      "grad_norm": 0.07208484446943192,
      "learning_rate": 0.0001975160386305471,
      "loss": 0.8578,
      "step": 25640
    },
    {
      "epoch": 0.4503239172036026,
      "grad_norm": 0.06892861296834842,
      "learning_rate": 0.0001975135574278742,
      "loss": 0.8615,
      "step": 25650
    },
    {
      "epoch": 0.450499482083604,
      "grad_norm": 0.07325382632386332,
      "learning_rate": 0.0001975110750023561,
      "loss": 0.8582,
      "step": 25660
    },
    {
      "epoch": 0.4506750469636054,
      "grad_norm": 0.07046486183242773,
      "learning_rate": 0.00019750859135402412,
      "loss": 0.8603,
      "step": 25670
    },
    {
      "epoch": 0.4508506118436068,
      "grad_norm": 0.07717062669980136,
      "learning_rate": 0.00019750610648290977,
      "loss": 0.8605,
      "step": 25680
    },
    {
      "epoch": 0.45102617672360823,
      "grad_norm": 0.06852262225060798,
      "learning_rate": 0.00019750362038904458,
      "loss": 0.8555,
      "step": 25690
    },
    {
      "epoch": 0.4512017416036096,
      "grad_norm": 0.06855582232572903,
      "learning_rate": 0.00019750113307246,
      "loss": 0.8485,
      "step": 25700
    },
    {
      "epoch": 0.451377306483611,
      "grad_norm": 0.05936174089489748,
      "learning_rate": 0.00019749864453318762,
      "loss": 0.861,
      "step": 25710
    },
    {
      "epoch": 0.45155287136361244,
      "grad_norm": 0.05475049004134085,
      "learning_rate": 0.0001974961547712589,
      "loss": 0.8525,
      "step": 25720
    },
    {
      "epoch": 0.4517284362436138,
      "grad_norm": 0.06623534532309994,
      "learning_rate": 0.0001974936637867054,
      "loss": 0.8579,
      "step": 25730
    },
    {
      "epoch": 0.45190400112361523,
      "grad_norm": 0.06592493272481496,
      "learning_rate": 0.00019749117157955866,
      "loss": 0.8658,
      "step": 25740
    },
    {
      "epoch": 0.45207956600361665,
      "grad_norm": 0.0743160955462744,
      "learning_rate": 0.00019748867814985033,
      "loss": 0.861,
      "step": 25750
    },
    {
      "epoch": 0.452255130883618,
      "grad_norm": 0.06241954908832853,
      "learning_rate": 0.00019748618349761193,
      "loss": 0.8637,
      "step": 25760
    },
    {
      "epoch": 0.45243069576361944,
      "grad_norm": 0.10865288492804304,
      "learning_rate": 0.0001974836876228751,
      "loss": 0.8519,
      "step": 25770
    },
    {
      "epoch": 0.45260626064362086,
      "grad_norm": 0.06881262795555637,
      "learning_rate": 0.0001974811905256715,
      "loss": 0.8538,
      "step": 25780
    },
    {
      "epoch": 0.45278182552362223,
      "grad_norm": 0.06536248621961203,
      "learning_rate": 0.00019747869220603266,
      "loss": 0.8476,
      "step": 25790
    },
    {
      "epoch": 0.45295739040362365,
      "grad_norm": 0.06724865553377461,
      "learning_rate": 0.00019747619266399032,
      "loss": 0.8538,
      "step": 25800
    },
    {
      "epoch": 0.45313295528362507,
      "grad_norm": 0.07225148646015678,
      "learning_rate": 0.00019747369189957614,
      "loss": 0.8637,
      "step": 25810
    },
    {
      "epoch": 0.4533085201636265,
      "grad_norm": 0.08821475884262162,
      "learning_rate": 0.00019747118991282178,
      "loss": 0.8535,
      "step": 25820
    },
    {
      "epoch": 0.45348408504362786,
      "grad_norm": 0.050823989860526976,
      "learning_rate": 0.00019746868670375895,
      "loss": 0.8483,
      "step": 25830
    },
    {
      "epoch": 0.4536596499236293,
      "grad_norm": 0.06272377424180864,
      "learning_rate": 0.0001974661822724194,
      "loss": 0.8536,
      "step": 25840
    },
    {
      "epoch": 0.4538352148036307,
      "grad_norm": 0.060015026135441164,
      "learning_rate": 0.00019746367661883482,
      "loss": 0.8616,
      "step": 25850
    },
    {
      "epoch": 0.45401077968363207,
      "grad_norm": 0.06258378552554607,
      "learning_rate": 0.00019746116974303697,
      "loss": 0.8581,
      "step": 25860
    },
    {
      "epoch": 0.4541863445636335,
      "grad_norm": 0.08861856660405056,
      "learning_rate": 0.0001974586616450576,
      "loss": 0.858,
      "step": 25870
    },
    {
      "epoch": 0.4543619094436349,
      "grad_norm": 0.06511130204204882,
      "learning_rate": 0.0001974561523249285,
      "loss": 0.8456,
      "step": 25880
    },
    {
      "epoch": 0.4545374743236363,
      "grad_norm": 0.057123727995082245,
      "learning_rate": 0.00019745364178268146,
      "loss": 0.8523,
      "step": 25890
    },
    {
      "epoch": 0.4547130392036377,
      "grad_norm": 0.07292345221291806,
      "learning_rate": 0.00019745113001834835,
      "loss": 0.8552,
      "step": 25900
    },
    {
      "epoch": 0.4548886040836391,
      "grad_norm": 0.06204831190602485,
      "learning_rate": 0.00019744861703196087,
      "loss": 0.8614,
      "step": 25910
    },
    {
      "epoch": 0.4550641689636405,
      "grad_norm": 0.06643268127765942,
      "learning_rate": 0.00019744610282355096,
      "loss": 0.8556,
      "step": 25920
    },
    {
      "epoch": 0.4552397338436419,
      "grad_norm": 0.09204164757507034,
      "learning_rate": 0.00019744358739315045,
      "loss": 0.8629,
      "step": 25930
    },
    {
      "epoch": 0.45541529872364334,
      "grad_norm": 0.06915176108425833,
      "learning_rate": 0.00019744107074079118,
      "loss": 0.8616,
      "step": 25940
    },
    {
      "epoch": 0.4555908636036447,
      "grad_norm": 0.05846330624786993,
      "learning_rate": 0.00019743855286650508,
      "loss": 0.8567,
      "step": 25950
    },
    {
      "epoch": 0.4557664284836461,
      "grad_norm": 0.0709093899290539,
      "learning_rate": 0.00019743603377032402,
      "loss": 0.847,
      "step": 25960
    },
    {
      "epoch": 0.45594199336364755,
      "grad_norm": 0.07469314654917064,
      "learning_rate": 0.00019743351345227994,
      "loss": 0.8459,
      "step": 25970
    },
    {
      "epoch": 0.4561175582436489,
      "grad_norm": 0.059695903654907644,
      "learning_rate": 0.00019743099191240474,
      "loss": 0.8515,
      "step": 25980
    },
    {
      "epoch": 0.45629312312365033,
      "grad_norm": 0.06198680149354743,
      "learning_rate": 0.0001974284691507304,
      "loss": 0.8537,
      "step": 25990
    },
    {
      "epoch": 0.45646868800365176,
      "grad_norm": 0.0791742065237304,
      "learning_rate": 0.0001974259451672889,
      "loss": 0.8585,
      "step": 26000
    },
    {
      "epoch": 0.4566442528836532,
      "grad_norm": 0.05781224818517847,
      "learning_rate": 0.0001974234199621122,
      "loss": 0.8505,
      "step": 26010
    },
    {
      "epoch": 0.45681981776365455,
      "grad_norm": 0.09219073236747584,
      "learning_rate": 0.00019742089353523226,
      "loss": 0.8541,
      "step": 26020
    },
    {
      "epoch": 0.45699538264365597,
      "grad_norm": 0.06185860386241468,
      "learning_rate": 0.00019741836588668113,
      "loss": 0.8562,
      "step": 26030
    },
    {
      "epoch": 0.4571709475236574,
      "grad_norm": 0.06329335041108722,
      "learning_rate": 0.0001974158370164908,
      "loss": 0.8528,
      "step": 26040
    },
    {
      "epoch": 0.45734651240365876,
      "grad_norm": 0.07781871297151056,
      "learning_rate": 0.00019741330692469339,
      "loss": 0.8512,
      "step": 26050
    },
    {
      "epoch": 0.4575220772836602,
      "grad_norm": 0.06439762780958869,
      "learning_rate": 0.00019741077561132085,
      "loss": 0.8655,
      "step": 26060
    },
    {
      "epoch": 0.4576976421636616,
      "grad_norm": 0.057798955374480515,
      "learning_rate": 0.00019740824307640535,
      "loss": 0.8631,
      "step": 26070
    },
    {
      "epoch": 0.45787320704366297,
      "grad_norm": 0.07058121455304425,
      "learning_rate": 0.00019740570931997892,
      "loss": 0.849,
      "step": 26080
    },
    {
      "epoch": 0.4580487719236644,
      "grad_norm": 0.05803130591280722,
      "learning_rate": 0.00019740317434207368,
      "loss": 0.8623,
      "step": 26090
    },
    {
      "epoch": 0.4582243368036658,
      "grad_norm": 0.10659324908157354,
      "learning_rate": 0.00019740063814272172,
      "loss": 0.8393,
      "step": 26100
    },
    {
      "epoch": 0.4583999016836672,
      "grad_norm": 0.04964848965698683,
      "learning_rate": 0.00019739810072195521,
      "loss": 0.8481,
      "step": 26110
    },
    {
      "epoch": 0.4585754665636686,
      "grad_norm": 0.05937466818366005,
      "learning_rate": 0.0001973955620798063,
      "loss": 0.8556,
      "step": 26120
    },
    {
      "epoch": 0.45875103144367,
      "grad_norm": 0.08309738465008282,
      "learning_rate": 0.00019739302221630716,
      "loss": 0.8603,
      "step": 26130
    },
    {
      "epoch": 0.4589265963236714,
      "grad_norm": 0.07569544799658894,
      "learning_rate": 0.00019739048113148993,
      "loss": 0.8556,
      "step": 26140
    },
    {
      "epoch": 0.4591021612036728,
      "grad_norm": 0.0581931298140712,
      "learning_rate": 0.00019738793882538683,
      "loss": 0.8597,
      "step": 26150
    },
    {
      "epoch": 0.45927772608367423,
      "grad_norm": 0.06024352766630603,
      "learning_rate": 0.00019738539529803008,
      "loss": 0.8528,
      "step": 26160
    },
    {
      "epoch": 0.45945329096367565,
      "grad_norm": 0.09786796744735765,
      "learning_rate": 0.0001973828505494519,
      "loss": 0.8504,
      "step": 26170
    },
    {
      "epoch": 0.459628855843677,
      "grad_norm": 0.07645646939371104,
      "learning_rate": 0.00019738030457968454,
      "loss": 0.8581,
      "step": 26180
    },
    {
      "epoch": 0.45980442072367844,
      "grad_norm": 0.07815478775210909,
      "learning_rate": 0.00019737775738876023,
      "loss": 0.862,
      "step": 26190
    },
    {
      "epoch": 0.45997998560367986,
      "grad_norm": 0.07306504737766241,
      "learning_rate": 0.00019737520897671127,
      "loss": 0.856,
      "step": 26200
    },
    {
      "epoch": 0.46015555048368123,
      "grad_norm": 0.06297350356356136,
      "learning_rate": 0.00019737265934356994,
      "loss": 0.8588,
      "step": 26210
    },
    {
      "epoch": 0.46033111536368265,
      "grad_norm": 0.06785527930383016,
      "learning_rate": 0.00019737010848936857,
      "loss": 0.8607,
      "step": 26220
    },
    {
      "epoch": 0.4605066802436841,
      "grad_norm": 0.09250933185283432,
      "learning_rate": 0.00019736755641413942,
      "loss": 0.8523,
      "step": 26230
    },
    {
      "epoch": 0.46068224512368544,
      "grad_norm": 0.06397657965565698,
      "learning_rate": 0.00019736500311791488,
      "loss": 0.8567,
      "step": 26240
    },
    {
      "epoch": 0.46085781000368686,
      "grad_norm": 0.06191457613190606,
      "learning_rate": 0.00019736244860072728,
      "loss": 0.8575,
      "step": 26250
    },
    {
      "epoch": 0.4610333748836883,
      "grad_norm": 0.05887282523356087,
      "learning_rate": 0.00019735989286260902,
      "loss": 0.8456,
      "step": 26260
    },
    {
      "epoch": 0.46120893976368965,
      "grad_norm": 0.06348652134994523,
      "learning_rate": 0.0001973573359035924,
      "loss": 0.8532,
      "step": 26270
    },
    {
      "epoch": 0.4613845046436911,
      "grad_norm": 0.06438693255766571,
      "learning_rate": 0.0001973547777237099,
      "loss": 0.8592,
      "step": 26280
    },
    {
      "epoch": 0.4615600695236925,
      "grad_norm": 0.07480862042205318,
      "learning_rate": 0.00019735221832299388,
      "loss": 0.8536,
      "step": 26290
    },
    {
      "epoch": 0.46173563440369386,
      "grad_norm": 0.0615523423254598,
      "learning_rate": 0.0001973496577014768,
      "loss": 0.8531,
      "step": 26300
    },
    {
      "epoch": 0.4619111992836953,
      "grad_norm": 0.06752819426740302,
      "learning_rate": 0.0001973470958591911,
      "loss": 0.8525,
      "step": 26310
    },
    {
      "epoch": 0.4620867641636967,
      "grad_norm": 0.06567431646929696,
      "learning_rate": 0.0001973445327961692,
      "loss": 0.8634,
      "step": 26320
    },
    {
      "epoch": 0.4622623290436981,
      "grad_norm": 0.05264530011156955,
      "learning_rate": 0.0001973419685124436,
      "loss": 0.8678,
      "step": 26330
    },
    {
      "epoch": 0.4624378939236995,
      "grad_norm": 0.06406744085634797,
      "learning_rate": 0.00019733940300804682,
      "loss": 0.8518,
      "step": 26340
    },
    {
      "epoch": 0.4626134588037009,
      "grad_norm": 0.05975998140019014,
      "learning_rate": 0.00019733683628301137,
      "loss": 0.8536,
      "step": 26350
    },
    {
      "epoch": 0.46278902368370234,
      "grad_norm": 0.06342105122601151,
      "learning_rate": 0.0001973342683373697,
      "loss": 0.852,
      "step": 26360
    },
    {
      "epoch": 0.4629645885637037,
      "grad_norm": 0.06292407016298075,
      "learning_rate": 0.0001973316991711544,
      "loss": 0.8506,
      "step": 26370
    },
    {
      "epoch": 0.4631401534437051,
      "grad_norm": 0.06554698926571184,
      "learning_rate": 0.00019732912878439802,
      "loss": 0.8485,
      "step": 26380
    },
    {
      "epoch": 0.46331571832370655,
      "grad_norm": 0.06864888219883872,
      "learning_rate": 0.0001973265571771331,
      "loss": 0.8635,
      "step": 26390
    },
    {
      "epoch": 0.4634912832037079,
      "grad_norm": 0.060651055027513584,
      "learning_rate": 0.00019732398434939223,
      "loss": 0.8538,
      "step": 26400
    },
    {
      "epoch": 0.46366684808370934,
      "grad_norm": 0.10985393004152066,
      "learning_rate": 0.00019732141030120802,
      "loss": 0.8566,
      "step": 26410
    },
    {
      "epoch": 0.46384241296371076,
      "grad_norm": 0.07303041532196781,
      "learning_rate": 0.00019731883503261313,
      "loss": 0.8537,
      "step": 26420
    },
    {
      "epoch": 0.4640179778437121,
      "grad_norm": 0.06324763680670217,
      "learning_rate": 0.0001973162585436401,
      "loss": 0.8513,
      "step": 26430
    },
    {
      "epoch": 0.46419354272371355,
      "grad_norm": 0.058332092625942314,
      "learning_rate": 0.0001973136808343216,
      "loss": 0.8558,
      "step": 26440
    },
    {
      "epoch": 0.46436910760371497,
      "grad_norm": 0.060225211905783914,
      "learning_rate": 0.00019731110190469032,
      "loss": 0.8534,
      "step": 26450
    },
    {
      "epoch": 0.46454467248371634,
      "grad_norm": 0.06090528901855355,
      "learning_rate": 0.00019730852175477894,
      "loss": 0.8496,
      "step": 26460
    },
    {
      "epoch": 0.46472023736371776,
      "grad_norm": 0.06304923506048156,
      "learning_rate": 0.0001973059403846201,
      "loss": 0.8663,
      "step": 26470
    },
    {
      "epoch": 0.4648958022437192,
      "grad_norm": 0.07332038216050225,
      "learning_rate": 0.0001973033577942466,
      "loss": 0.8594,
      "step": 26480
    },
    {
      "epoch": 0.46507136712372055,
      "grad_norm": 0.07054149587920959,
      "learning_rate": 0.00019730077398369102,
      "loss": 0.8431,
      "step": 26490
    },
    {
      "epoch": 0.46524693200372197,
      "grad_norm": 0.07373895429775396,
      "learning_rate": 0.00019729818895298625,
      "loss": 0.8494,
      "step": 26500
    },
    {
      "epoch": 0.4654224968837234,
      "grad_norm": 0.06070831681420259,
      "learning_rate": 0.0001972956027021649,
      "loss": 0.8566,
      "step": 26510
    },
    {
      "epoch": 0.46559806176372476,
      "grad_norm": 0.05727110036337638,
      "learning_rate": 0.00019729301523125985,
      "loss": 0.8478,
      "step": 26520
    },
    {
      "epoch": 0.4657736266437262,
      "grad_norm": 0.05911256446091189,
      "learning_rate": 0.00019729042654030384,
      "loss": 0.8527,
      "step": 26530
    },
    {
      "epoch": 0.4659491915237276,
      "grad_norm": 0.06314722630865474,
      "learning_rate": 0.00019728783662932965,
      "loss": 0.8509,
      "step": 26540
    },
    {
      "epoch": 0.466124756403729,
      "grad_norm": 0.06277261603433518,
      "learning_rate": 0.00019728524549837015,
      "loss": 0.8627,
      "step": 26550
    },
    {
      "epoch": 0.4663003212837304,
      "grad_norm": 0.07187657135366095,
      "learning_rate": 0.00019728265314745811,
      "loss": 0.8576,
      "step": 26560
    },
    {
      "epoch": 0.4664758861637318,
      "grad_norm": 0.061149791830974735,
      "learning_rate": 0.0001972800595766264,
      "loss": 0.8602,
      "step": 26570
    },
    {
      "epoch": 0.46665145104373323,
      "grad_norm": 0.04997806297382646,
      "learning_rate": 0.0001972774647859079,
      "loss": 0.8608,
      "step": 26580
    },
    {
      "epoch": 0.4668270159237346,
      "grad_norm": 0.08171766670386364,
      "learning_rate": 0.0001972748687753355,
      "loss": 0.8607,
      "step": 26590
    },
    {
      "epoch": 0.467002580803736,
      "grad_norm": 0.08464067451262,
      "learning_rate": 0.000197272271544942,
      "loss": 0.8611,
      "step": 26600
    },
    {
      "epoch": 0.46717814568373744,
      "grad_norm": 0.0707141193108053,
      "learning_rate": 0.0001972696730947604,
      "loss": 0.8515,
      "step": 26610
    },
    {
      "epoch": 0.4673537105637388,
      "grad_norm": 0.07434070586382739,
      "learning_rate": 0.0001972670734248236,
      "loss": 0.8548,
      "step": 26620
    },
    {
      "epoch": 0.46752927544374023,
      "grad_norm": 0.09037047372551983,
      "learning_rate": 0.00019726447253516454,
      "loss": 0.8531,
      "step": 26630
    },
    {
      "epoch": 0.46770484032374166,
      "grad_norm": 0.07113035733731866,
      "learning_rate": 0.00019726187042581612,
      "loss": 0.8518,
      "step": 26640
    },
    {
      "epoch": 0.467880405203743,
      "grad_norm": 0.06434064330213746,
      "learning_rate": 0.00019725926709681136,
      "loss": 0.8613,
      "step": 26650
    },
    {
      "epoch": 0.46805597008374444,
      "grad_norm": 0.060887858542256756,
      "learning_rate": 0.00019725666254818324,
      "loss": 0.8571,
      "step": 26660
    },
    {
      "epoch": 0.46823153496374587,
      "grad_norm": 0.058992315120566435,
      "learning_rate": 0.0001972540567799648,
      "loss": 0.8517,
      "step": 26670
    },
    {
      "epoch": 0.46840709984374723,
      "grad_norm": 0.07839284333847493,
      "learning_rate": 0.000197251449792189,
      "loss": 0.8549,
      "step": 26680
    },
    {
      "epoch": 0.46858266472374865,
      "grad_norm": 0.0507488083887886,
      "learning_rate": 0.00019724884158488882,
      "loss": 0.8539,
      "step": 26690
    },
    {
      "epoch": 0.4687582296037501,
      "grad_norm": 0.06119576619958442,
      "learning_rate": 0.00019724623215809748,
      "loss": 0.8564,
      "step": 26700
    },
    {
      "epoch": 0.46893379448375144,
      "grad_norm": 0.06408336786298548,
      "learning_rate": 0.00019724362151184784,
      "loss": 0.8579,
      "step": 26710
    },
    {
      "epoch": 0.46910935936375286,
      "grad_norm": 0.0644146252790462,
      "learning_rate": 0.00019724100964617312,
      "loss": 0.8463,
      "step": 26720
    },
    {
      "epoch": 0.4692849242437543,
      "grad_norm": 0.06743610101294686,
      "learning_rate": 0.00019723839656110635,
      "loss": 0.8529,
      "step": 26730
    },
    {
      "epoch": 0.4694604891237557,
      "grad_norm": 0.06915855696397694,
      "learning_rate": 0.00019723578225668063,
      "loss": 0.8577,
      "step": 26740
    },
    {
      "epoch": 0.4696360540037571,
      "grad_norm": 0.06283346400551916,
      "learning_rate": 0.0001972331667329291,
      "loss": 0.8577,
      "step": 26750
    },
    {
      "epoch": 0.4698116188837585,
      "grad_norm": 0.05420639805557297,
      "learning_rate": 0.00019723054998988494,
      "loss": 0.8616,
      "step": 26760
    },
    {
      "epoch": 0.4699871837637599,
      "grad_norm": 0.07025774787070745,
      "learning_rate": 0.00019722793202758127,
      "loss": 0.8581,
      "step": 26770
    },
    {
      "epoch": 0.4701627486437613,
      "grad_norm": 0.0697601324510459,
      "learning_rate": 0.00019722531284605127,
      "loss": 0.8615,
      "step": 26780
    },
    {
      "epoch": 0.4703383135237627,
      "grad_norm": 0.06383171459732533,
      "learning_rate": 0.0001972226924453281,
      "loss": 0.8588,
      "step": 26790
    },
    {
      "epoch": 0.47051387840376413,
      "grad_norm": 0.10879940764412253,
      "learning_rate": 0.00019722007082544494,
      "loss": 0.8551,
      "step": 26800
    },
    {
      "epoch": 0.4706894432837655,
      "grad_norm": 0.06624317600283815,
      "learning_rate": 0.00019721744798643508,
      "loss": 0.8528,
      "step": 26810
    },
    {
      "epoch": 0.4708650081637669,
      "grad_norm": 0.07607252469210495,
      "learning_rate": 0.0001972148239283317,
      "loss": 0.8534,
      "step": 26820
    },
    {
      "epoch": 0.47104057304376834,
      "grad_norm": 0.059611879922200045,
      "learning_rate": 0.00019721219865116807,
      "loss": 0.8538,
      "step": 26830
    },
    {
      "epoch": 0.4712161379237697,
      "grad_norm": 0.05322570517771284,
      "learning_rate": 0.00019720957215497745,
      "loss": 0.8476,
      "step": 26840
    },
    {
      "epoch": 0.47139170280377113,
      "grad_norm": 0.0654096482193431,
      "learning_rate": 0.0001972069444397931,
      "loss": 0.8564,
      "step": 26850
    },
    {
      "epoch": 0.47156726768377255,
      "grad_norm": 0.055160666191559925,
      "learning_rate": 0.0001972043155056483,
      "loss": 0.8605,
      "step": 26860
    },
    {
      "epoch": 0.4717428325637739,
      "grad_norm": 0.09544192896123628,
      "learning_rate": 0.00019720168535257643,
      "loss": 0.8559,
      "step": 26870
    },
    {
      "epoch": 0.47191839744377534,
      "grad_norm": 0.07500152821653201,
      "learning_rate": 0.00019719905398061073,
      "loss": 0.8496,
      "step": 26880
    },
    {
      "epoch": 0.47209396232377676,
      "grad_norm": 0.0685291226268722,
      "learning_rate": 0.0001971964213897846,
      "loss": 0.8539,
      "step": 26890
    },
    {
      "epoch": 0.4722695272037782,
      "grad_norm": 0.0635625280811844,
      "learning_rate": 0.0001971937875801314,
      "loss": 0.8635,
      "step": 26900
    },
    {
      "epoch": 0.47244509208377955,
      "grad_norm": 0.07064911737222981,
      "learning_rate": 0.00019719115255168444,
      "loss": 0.8588,
      "step": 26910
    },
    {
      "epoch": 0.47262065696378097,
      "grad_norm": 0.06217224542783314,
      "learning_rate": 0.0001971885163044772,
      "loss": 0.8565,
      "step": 26920
    },
    {
      "epoch": 0.4727962218437824,
      "grad_norm": 0.05841006781324115,
      "learning_rate": 0.00019718587883854293,
      "loss": 0.8547,
      "step": 26930
    },
    {
      "epoch": 0.47297178672378376,
      "grad_norm": 0.06904368806585524,
      "learning_rate": 0.0001971832401539152,
      "loss": 0.8536,
      "step": 26940
    },
    {
      "epoch": 0.4731473516037852,
      "grad_norm": 0.06311542859313705,
      "learning_rate": 0.0001971806002506274,
      "loss": 0.8592,
      "step": 26950
    },
    {
      "epoch": 0.4733229164837866,
      "grad_norm": 0.062353835494114176,
      "learning_rate": 0.00019717795912871293,
      "loss": 0.8549,
      "step": 26960
    },
    {
      "epoch": 0.47349848136378797,
      "grad_norm": 0.11503668036406806,
      "learning_rate": 0.00019717531678820527,
      "loss": 0.8593,
      "step": 26970
    },
    {
      "epoch": 0.4736740462437894,
      "grad_norm": 0.08185983466603101,
      "learning_rate": 0.00019717267322913795,
      "loss": 0.858,
      "step": 26980
    },
    {
      "epoch": 0.4738496111237908,
      "grad_norm": 0.06630980104330357,
      "learning_rate": 0.0001971700284515444,
      "loss": 0.8648,
      "step": 26990
    },
    {
      "epoch": 0.4740251760037922,
      "grad_norm": 0.09181406606766367,
      "learning_rate": 0.0001971673824554582,
      "loss": 0.8567,
      "step": 27000
    },
    {
      "epoch": 0.4742007408837936,
      "grad_norm": 0.062316612121533686,
      "learning_rate": 0.0001971647352409128,
      "loss": 0.8524,
      "step": 27010
    },
    {
      "epoch": 0.474376305763795,
      "grad_norm": 0.06353189441594131,
      "learning_rate": 0.00019716208680794177,
      "loss": 0.8514,
      "step": 27020
    },
    {
      "epoch": 0.4745518706437964,
      "grad_norm": 0.0840507419055481,
      "learning_rate": 0.00019715943715657868,
      "loss": 0.8596,
      "step": 27030
    },
    {
      "epoch": 0.4747274355237978,
      "grad_norm": 0.09631547288685952,
      "learning_rate": 0.0001971567862868571,
      "loss": 0.8564,
      "step": 27040
    },
    {
      "epoch": 0.47490300040379924,
      "grad_norm": 0.07430686719049828,
      "learning_rate": 0.00019715413419881062,
      "loss": 0.8562,
      "step": 27050
    },
    {
      "epoch": 0.4750785652838006,
      "grad_norm": 0.10026830517995503,
      "learning_rate": 0.00019715148089247283,
      "loss": 0.8588,
      "step": 27060
    },
    {
      "epoch": 0.475254130163802,
      "grad_norm": 0.07478864479035628,
      "learning_rate": 0.00019714882636787733,
      "loss": 0.859,
      "step": 27070
    },
    {
      "epoch": 0.47542969504380345,
      "grad_norm": 0.04822948606426243,
      "learning_rate": 0.00019714617062505774,
      "loss": 0.8623,
      "step": 27080
    },
    {
      "epoch": 0.47560525992380487,
      "grad_norm": 0.07100643247143651,
      "learning_rate": 0.00019714351366404779,
      "loss": 0.8606,
      "step": 27090
    },
    {
      "epoch": 0.47578082480380623,
      "grad_norm": 0.0881956980849901,
      "learning_rate": 0.00019714085548488107,
      "loss": 0.8504,
      "step": 27100
    },
    {
      "epoch": 0.47595638968380766,
      "grad_norm": 0.07650538858460322,
      "learning_rate": 0.0001971381960875913,
      "loss": 0.8587,
      "step": 27110
    },
    {
      "epoch": 0.4761319545638091,
      "grad_norm": 0.06326272642744339,
      "learning_rate": 0.0001971355354722122,
      "loss": 0.8621,
      "step": 27120
    },
    {
      "epoch": 0.47630751944381045,
      "grad_norm": 0.07650012627697049,
      "learning_rate": 0.0001971328736387774,
      "loss": 0.8538,
      "step": 27130
    },
    {
      "epoch": 0.47648308432381187,
      "grad_norm": 0.06561713987030827,
      "learning_rate": 0.0001971302105873207,
      "loss": 0.8626,
      "step": 27140
    },
    {
      "epoch": 0.4766586492038133,
      "grad_norm": 0.0557605189838291,
      "learning_rate": 0.00019712754631787574,
      "loss": 0.856,
      "step": 27150
    },
    {
      "epoch": 0.47683421408381466,
      "grad_norm": 0.07248665010368488,
      "learning_rate": 0.0001971248808304764,
      "loss": 0.8549,
      "step": 27160
    },
    {
      "epoch": 0.4770097789638161,
      "grad_norm": 0.059109716405381874,
      "learning_rate": 0.00019712221412515637,
      "loss": 0.8551,
      "step": 27170
    },
    {
      "epoch": 0.4771853438438175,
      "grad_norm": 0.07786637259915112,
      "learning_rate": 0.00019711954620194947,
      "loss": 0.8521,
      "step": 27180
    },
    {
      "epoch": 0.47736090872381887,
      "grad_norm": 0.07036313806291657,
      "learning_rate": 0.00019711687706088952,
      "loss": 0.8618,
      "step": 27190
    },
    {
      "epoch": 0.4775364736038203,
      "grad_norm": 0.07562785145597291,
      "learning_rate": 0.0001971142067020103,
      "loss": 0.8493,
      "step": 27200
    },
    {
      "epoch": 0.4777120384838217,
      "grad_norm": 0.06102242095308004,
      "learning_rate": 0.00019711153512534568,
      "loss": 0.8611,
      "step": 27210
    },
    {
      "epoch": 0.4778876033638231,
      "grad_norm": 0.0638127087130811,
      "learning_rate": 0.00019710886233092947,
      "loss": 0.8575,
      "step": 27220
    },
    {
      "epoch": 0.4780631682438245,
      "grad_norm": 0.1099226829128162,
      "learning_rate": 0.00019710618831879554,
      "loss": 0.8591,
      "step": 27230
    },
    {
      "epoch": 0.4782387331238259,
      "grad_norm": 0.08905341714928514,
      "learning_rate": 0.0001971035130889778,
      "loss": 0.8594,
      "step": 27240
    },
    {
      "epoch": 0.4784142980038273,
      "grad_norm": 0.07175823802811242,
      "learning_rate": 0.00019710083664151016,
      "loss": 0.8552,
      "step": 27250
    },
    {
      "epoch": 0.4785898628838287,
      "grad_norm": 0.0739457202478879,
      "learning_rate": 0.00019709815897642647,
      "loss": 0.8602,
      "step": 27260
    },
    {
      "epoch": 0.47876542776383013,
      "grad_norm": 0.06758692472465767,
      "learning_rate": 0.0001970954800937607,
      "loss": 0.8492,
      "step": 27270
    },
    {
      "epoch": 0.47894099264383155,
      "grad_norm": 0.07294576388880668,
      "learning_rate": 0.00019709279999354678,
      "loss": 0.8585,
      "step": 27280
    },
    {
      "epoch": 0.4791165575238329,
      "grad_norm": 0.09820665008744509,
      "learning_rate": 0.00019709011867581868,
      "loss": 0.8489,
      "step": 27290
    },
    {
      "epoch": 0.47929212240383434,
      "grad_norm": 0.08618211523350933,
      "learning_rate": 0.00019708743614061034,
      "loss": 0.8518,
      "step": 27300
    },
    {
      "epoch": 0.47946768728383576,
      "grad_norm": 0.060274246487335,
      "learning_rate": 0.00019708475238795577,
      "loss": 0.8534,
      "step": 27310
    },
    {
      "epoch": 0.47964325216383713,
      "grad_norm": 0.05898601147409525,
      "learning_rate": 0.00019708206741788896,
      "loss": 0.8542,
      "step": 27320
    },
    {
      "epoch": 0.47981881704383855,
      "grad_norm": 0.054838399993351876,
      "learning_rate": 0.000197079381230444,
      "loss": 0.8582,
      "step": 27330
    },
    {
      "epoch": 0.47999438192384,
      "grad_norm": 0.06785639807392535,
      "learning_rate": 0.0001970766938256548,
      "loss": 0.854,
      "step": 27340
    },
    {
      "epoch": 0.48016994680384134,
      "grad_norm": 0.07638838428275395,
      "learning_rate": 0.0001970740052035555,
      "loss": 0.859,
      "step": 27350
    },
    {
      "epoch": 0.48034551168384276,
      "grad_norm": 0.06953102734728361,
      "learning_rate": 0.00019707131536418016,
      "loss": 0.8519,
      "step": 27360
    },
    {
      "epoch": 0.4805210765638442,
      "grad_norm": 0.05665040596576454,
      "learning_rate": 0.00019706862430756284,
      "loss": 0.8555,
      "step": 27370
    },
    {
      "epoch": 0.48069664144384555,
      "grad_norm": 0.07172261542305274,
      "learning_rate": 0.00019706593203373763,
      "loss": 0.8589,
      "step": 27380
    },
    {
      "epoch": 0.480872206323847,
      "grad_norm": 0.08768823651411391,
      "learning_rate": 0.00019706323854273863,
      "loss": 0.8563,
      "step": 27390
    },
    {
      "epoch": 0.4810477712038484,
      "grad_norm": 0.06113865244925923,
      "learning_rate": 0.0001970605438346,
      "loss": 0.8537,
      "step": 27400
    },
    {
      "epoch": 0.48122333608384976,
      "grad_norm": 0.06504890774484924,
      "learning_rate": 0.00019705784790935587,
      "loss": 0.8625,
      "step": 27410
    },
    {
      "epoch": 0.4813989009638512,
      "grad_norm": 0.05688937528245086,
      "learning_rate": 0.0001970551507670404,
      "loss": 0.8583,
      "step": 27420
    },
    {
      "epoch": 0.4815744658438526,
      "grad_norm": 0.07186803308600347,
      "learning_rate": 0.00019705245240768775,
      "loss": 0.8496,
      "step": 27430
    },
    {
      "epoch": 0.481750030723854,
      "grad_norm": 0.08790078781507331,
      "learning_rate": 0.0001970497528313321,
      "loss": 0.8472,
      "step": 27440
    },
    {
      "epoch": 0.4819255956038554,
      "grad_norm": 0.05386567030297408,
      "learning_rate": 0.00019704705203800775,
      "loss": 0.8501,
      "step": 27450
    },
    {
      "epoch": 0.4821011604838568,
      "grad_norm": 0.06268386956392248,
      "learning_rate": 0.0001970443500277488,
      "loss": 0.8592,
      "step": 27460
    },
    {
      "epoch": 0.48227672536385824,
      "grad_norm": 0.05708046906377315,
      "learning_rate": 0.0001970416468005895,
      "loss": 0.8659,
      "step": 27470
    },
    {
      "epoch": 0.4824522902438596,
      "grad_norm": 0.06958606396959711,
      "learning_rate": 0.0001970389423565642,
      "loss": 0.8545,
      "step": 27480
    },
    {
      "epoch": 0.482627855123861,
      "grad_norm": 0.06313474593042838,
      "learning_rate": 0.00019703623669570702,
      "loss": 0.8505,
      "step": 27490
    },
    {
      "epoch": 0.48280342000386245,
      "grad_norm": 0.06427527764303687,
      "learning_rate": 0.00019703352981805235,
      "loss": 0.8567,
      "step": 27500
    },
    {
      "epoch": 0.4829789848838638,
      "grad_norm": 0.06765557713218012,
      "learning_rate": 0.00019703082172363445,
      "loss": 0.8584,
      "step": 27510
    },
    {
      "epoch": 0.48315454976386524,
      "grad_norm": 0.054144149276880514,
      "learning_rate": 0.00019702811241248763,
      "loss": 0.8558,
      "step": 27520
    },
    {
      "epoch": 0.48333011464386666,
      "grad_norm": 0.061267536834216214,
      "learning_rate": 0.00019702540188464625,
      "loss": 0.8595,
      "step": 27530
    },
    {
      "epoch": 0.483505679523868,
      "grad_norm": 0.05504511996244035,
      "learning_rate": 0.0001970226901401446,
      "loss": 0.8561,
      "step": 27540
    },
    {
      "epoch": 0.48368124440386945,
      "grad_norm": 0.07838870881142505,
      "learning_rate": 0.00019701997717901703,
      "loss": 0.8612,
      "step": 27550
    },
    {
      "epoch": 0.48385680928387087,
      "grad_norm": 0.08555503188066257,
      "learning_rate": 0.00019701726300129797,
      "loss": 0.8585,
      "step": 27560
    },
    {
      "epoch": 0.48403237416387224,
      "grad_norm": 0.08683377504321554,
      "learning_rate": 0.00019701454760702182,
      "loss": 0.8546,
      "step": 27570
    },
    {
      "epoch": 0.48420793904387366,
      "grad_norm": 0.07944836326217591,
      "learning_rate": 0.00019701183099622293,
      "loss": 0.854,
      "step": 27580
    },
    {
      "epoch": 0.4843835039238751,
      "grad_norm": 0.0564204983108016,
      "learning_rate": 0.0001970091131689357,
      "loss": 0.8519,
      "step": 27590
    },
    {
      "epoch": 0.48455906880387645,
      "grad_norm": 0.07428200695757439,
      "learning_rate": 0.00019700639412519464,
      "loss": 0.8477,
      "step": 27600
    },
    {
      "epoch": 0.48473463368387787,
      "grad_norm": 0.06540456733060104,
      "learning_rate": 0.0001970036738650342,
      "loss": 0.8591,
      "step": 27610
    },
    {
      "epoch": 0.4849101985638793,
      "grad_norm": 0.08680075211634407,
      "learning_rate": 0.00019700095238848876,
      "loss": 0.8515,
      "step": 27620
    },
    {
      "epoch": 0.4850857634438807,
      "grad_norm": 0.1318464429460853,
      "learning_rate": 0.00019699822969559288,
      "loss": 0.8566,
      "step": 27630
    },
    {
      "epoch": 0.4852613283238821,
      "grad_norm": 0.06630031888808405,
      "learning_rate": 0.00019699550578638103,
      "loss": 0.8579,
      "step": 27640
    },
    {
      "epoch": 0.4854368932038835,
      "grad_norm": 0.07240842569717922,
      "learning_rate": 0.00019699278066088774,
      "loss": 0.8572,
      "step": 27650
    },
    {
      "epoch": 0.4856124580838849,
      "grad_norm": 0.05770079674679219,
      "learning_rate": 0.0001969900543191475,
      "loss": 0.86,
      "step": 27660
    },
    {
      "epoch": 0.4857880229638863,
      "grad_norm": 0.06509623105911508,
      "learning_rate": 0.0001969873267611949,
      "loss": 0.8605,
      "step": 27670
    },
    {
      "epoch": 0.4859635878438877,
      "grad_norm": 0.09301881989420197,
      "learning_rate": 0.00019698459798706444,
      "loss": 0.8602,
      "step": 27680
    },
    {
      "epoch": 0.48613915272388913,
      "grad_norm": 0.0747832824346993,
      "learning_rate": 0.00019698186799679073,
      "loss": 0.851,
      "step": 27690
    },
    {
      "epoch": 0.4863147176038905,
      "grad_norm": 0.06613081410063684,
      "learning_rate": 0.00019697913679040839,
      "loss": 0.8589,
      "step": 27700
    },
    {
      "epoch": 0.4864902824838919,
      "grad_norm": 0.054508370336264315,
      "learning_rate": 0.00019697640436795198,
      "loss": 0.8495,
      "step": 27710
    },
    {
      "epoch": 0.48666584736389334,
      "grad_norm": 0.057978159646401316,
      "learning_rate": 0.0001969736707294561,
      "loss": 0.8634,
      "step": 27720
    },
    {
      "epoch": 0.4868414122438947,
      "grad_norm": 0.07618244382054118,
      "learning_rate": 0.00019697093587495542,
      "loss": 0.85,
      "step": 27730
    },
    {
      "epoch": 0.48701697712389613,
      "grad_norm": 0.08086187047914759,
      "learning_rate": 0.00019696819980448467,
      "loss": 0.8646,
      "step": 27740
    },
    {
      "epoch": 0.48719254200389756,
      "grad_norm": 0.07924945099493652,
      "learning_rate": 0.00019696546251807837,
      "loss": 0.8544,
      "step": 27750
    },
    {
      "epoch": 0.4873681068838989,
      "grad_norm": 0.05977372430135445,
      "learning_rate": 0.00019696272401577126,
      "loss": 0.8593,
      "step": 27760
    },
    {
      "epoch": 0.48754367176390034,
      "grad_norm": 0.05557158049491057,
      "learning_rate": 0.00019695998429759807,
      "loss": 0.8533,
      "step": 27770
    },
    {
      "epoch": 0.48771923664390177,
      "grad_norm": 0.06082241780043993,
      "learning_rate": 0.00019695724336359348,
      "loss": 0.8619,
      "step": 27780
    },
    {
      "epoch": 0.48789480152390313,
      "grad_norm": 0.07461557549725935,
      "learning_rate": 0.00019695450121379223,
      "loss": 0.8512,
      "step": 27790
    },
    {
      "epoch": 0.48807036640390455,
      "grad_norm": 0.0642805376579669,
      "learning_rate": 0.00019695175784822906,
      "loss": 0.85,
      "step": 27800
    },
    {
      "epoch": 0.488245931283906,
      "grad_norm": 0.08563438884587328,
      "learning_rate": 0.00019694901326693874,
      "loss": 0.8519,
      "step": 27810
    },
    {
      "epoch": 0.4884214961639074,
      "grad_norm": 0.04716129367903034,
      "learning_rate": 0.00019694626746995602,
      "loss": 0.8589,
      "step": 27820
    },
    {
      "epoch": 0.48859706104390876,
      "grad_norm": 0.06252324871832242,
      "learning_rate": 0.0001969435204573157,
      "loss": 0.8532,
      "step": 27830
    },
    {
      "epoch": 0.4887726259239102,
      "grad_norm": 0.05710053110990763,
      "learning_rate": 0.00019694077222905262,
      "loss": 0.854,
      "step": 27840
    },
    {
      "epoch": 0.4889481908039116,
      "grad_norm": 0.06776119869554335,
      "learning_rate": 0.00019693802278520153,
      "loss": 0.8555,
      "step": 27850
    },
    {
      "epoch": 0.489123755683913,
      "grad_norm": 0.06046355107025973,
      "learning_rate": 0.0001969352721257973,
      "loss": 0.8527,
      "step": 27860
    },
    {
      "epoch": 0.4892993205639144,
      "grad_norm": 0.07847880543809231,
      "learning_rate": 0.00019693252025087483,
      "loss": 0.8545,
      "step": 27870
    },
    {
      "epoch": 0.4894748854439158,
      "grad_norm": 0.0979912048807461,
      "learning_rate": 0.00019692976716046892,
      "loss": 0.8568,
      "step": 27880
    },
    {
      "epoch": 0.4896504503239172,
      "grad_norm": 0.05352378443811399,
      "learning_rate": 0.00019692701285461443,
      "loss": 0.8562,
      "step": 27890
    },
    {
      "epoch": 0.4898260152039186,
      "grad_norm": 0.05975911268630526,
      "learning_rate": 0.00019692425733334635,
      "loss": 0.8582,
      "step": 27900
    },
    {
      "epoch": 0.49000158008392003,
      "grad_norm": 0.0847764870368552,
      "learning_rate": 0.00019692150059669953,
      "loss": 0.8582,
      "step": 27910
    },
    {
      "epoch": 0.4901771449639214,
      "grad_norm": 0.06975054035546162,
      "learning_rate": 0.00019691874264470894,
      "loss": 0.8536,
      "step": 27920
    },
    {
      "epoch": 0.4903527098439228,
      "grad_norm": 0.07462167687298878,
      "learning_rate": 0.0001969159834774095,
      "loss": 0.8587,
      "step": 27930
    },
    {
      "epoch": 0.49052827472392424,
      "grad_norm": 0.054980067801393334,
      "learning_rate": 0.00019691322309483612,
      "loss": 0.8579,
      "step": 27940
    },
    {
      "epoch": 0.4907038396039256,
      "grad_norm": 0.05476062022442028,
      "learning_rate": 0.00019691046149702385,
      "loss": 0.8602,
      "step": 27950
    },
    {
      "epoch": 0.49087940448392703,
      "grad_norm": 0.06465711858667579,
      "learning_rate": 0.00019690769868400767,
      "loss": 0.8584,
      "step": 27960
    },
    {
      "epoch": 0.49105496936392845,
      "grad_norm": 0.045391642822662545,
      "learning_rate": 0.00019690493465582257,
      "loss": 0.8553,
      "step": 27970
    },
    {
      "epoch": 0.4912305342439298,
      "grad_norm": 0.07291529003120864,
      "learning_rate": 0.00019690216941250352,
      "loss": 0.8524,
      "step": 27980
    },
    {
      "epoch": 0.49140609912393124,
      "grad_norm": 0.06918945937199074,
      "learning_rate": 0.00019689940295408565,
      "loss": 0.8599,
      "step": 27990
    },
    {
      "epoch": 0.49158166400393266,
      "grad_norm": 0.08755967155698391,
      "learning_rate": 0.00019689663528060393,
      "loss": 0.8567,
      "step": 28000
    },
    {
      "epoch": 0.4917572288839341,
      "grad_norm": 0.07434169871817878,
      "learning_rate": 0.00019689386639209347,
      "loss": 0.8525,
      "step": 28010
    },
    {
      "epoch": 0.49193279376393545,
      "grad_norm": 0.06072088719711004,
      "learning_rate": 0.00019689109628858936,
      "loss": 0.8628,
      "step": 28020
    },
    {
      "epoch": 0.49210835864393687,
      "grad_norm": 0.07200055159366285,
      "learning_rate": 0.0001968883249701267,
      "loss": 0.8508,
      "step": 28030
    },
    {
      "epoch": 0.4922839235239383,
      "grad_norm": 0.08129810905823791,
      "learning_rate": 0.00019688555243674055,
      "loss": 0.857,
      "step": 28040
    },
    {
      "epoch": 0.49245948840393966,
      "grad_norm": 0.076158473551463,
      "learning_rate": 0.0001968827786884661,
      "loss": 0.8522,
      "step": 28050
    },
    {
      "epoch": 0.4926350532839411,
      "grad_norm": 0.07272978604364178,
      "learning_rate": 0.00019688000372533842,
      "loss": 0.8523,
      "step": 28060
    },
    {
      "epoch": 0.4928106181639425,
      "grad_norm": 0.07464715123310973,
      "learning_rate": 0.00019687722754739277,
      "loss": 0.8479,
      "step": 28070
    },
    {
      "epoch": 0.49298618304394387,
      "grad_norm": 0.051572127263426,
      "learning_rate": 0.00019687445015466426,
      "loss": 0.8656,
      "step": 28080
    },
    {
      "epoch": 0.4931617479239453,
      "grad_norm": 0.06988047618828351,
      "learning_rate": 0.00019687167154718808,
      "loss": 0.8563,
      "step": 28090
    },
    {
      "epoch": 0.4933373128039467,
      "grad_norm": 0.12961497372747008,
      "learning_rate": 0.00019686889172499948,
      "loss": 0.8505,
      "step": 28100
    },
    {
      "epoch": 0.4935128776839481,
      "grad_norm": 0.08134407597794163,
      "learning_rate": 0.00019686611068813362,
      "loss": 0.8475,
      "step": 28110
    },
    {
      "epoch": 0.4936884425639495,
      "grad_norm": 0.08796502108985815,
      "learning_rate": 0.00019686332843662577,
      "loss": 0.8603,
      "step": 28120
    },
    {
      "epoch": 0.4938640074439509,
      "grad_norm": 0.05270731126698327,
      "learning_rate": 0.00019686054497051116,
      "loss": 0.8589,
      "step": 28130
    },
    {
      "epoch": 0.4940395723239523,
      "grad_norm": 0.08023002171770685,
      "learning_rate": 0.00019685776028982506,
      "loss": 0.8511,
      "step": 28140
    },
    {
      "epoch": 0.4942151372039537,
      "grad_norm": 0.09513636342523608,
      "learning_rate": 0.00019685497439460284,
      "loss": 0.8591,
      "step": 28150
    },
    {
      "epoch": 0.49439070208395514,
      "grad_norm": 0.09196266949727114,
      "learning_rate": 0.00019685218728487965,
      "loss": 0.8582,
      "step": 28160
    },
    {
      "epoch": 0.49456626696395656,
      "grad_norm": 0.08144858461720979,
      "learning_rate": 0.00019684939896069093,
      "loss": 0.8538,
      "step": 28170
    },
    {
      "epoch": 0.4947418318439579,
      "grad_norm": 0.05386335071486513,
      "learning_rate": 0.00019684660942207192,
      "loss": 0.8566,
      "step": 28180
    },
    {
      "epoch": 0.49491739672395935,
      "grad_norm": 0.0635646125774291,
      "learning_rate": 0.00019684381866905799,
      "loss": 0.852,
      "step": 28190
    },
    {
      "epoch": 0.49509296160396077,
      "grad_norm": 0.06157510930610143,
      "learning_rate": 0.00019684102670168455,
      "loss": 0.8519,
      "step": 28200
    },
    {
      "epoch": 0.49526852648396213,
      "grad_norm": 0.09124931737547673,
      "learning_rate": 0.00019683823351998692,
      "loss": 0.8608,
      "step": 28210
    },
    {
      "epoch": 0.49544409136396356,
      "grad_norm": 0.0785034765854294,
      "learning_rate": 0.00019683543912400048,
      "loss": 0.8577,
      "step": 28220
    },
    {
      "epoch": 0.495619656243965,
      "grad_norm": 0.06061286037670452,
      "learning_rate": 0.0001968326435137607,
      "loss": 0.8649,
      "step": 28230
    },
    {
      "epoch": 0.49579522112396635,
      "grad_norm": 0.05003654296561814,
      "learning_rate": 0.00019682984668930293,
      "loss": 0.8526,
      "step": 28240
    },
    {
      "epoch": 0.49597078600396777,
      "grad_norm": 0.1003020117510621,
      "learning_rate": 0.00019682704865066263,
      "loss": 0.8532,
      "step": 28250
    },
    {
      "epoch": 0.4961463508839692,
      "grad_norm": 0.0743356000140512,
      "learning_rate": 0.0001968242493978753,
      "loss": 0.8498,
      "step": 28260
    },
    {
      "epoch": 0.49632191576397056,
      "grad_norm": 0.07484948549150736,
      "learning_rate": 0.00019682144893097634,
      "loss": 0.8569,
      "step": 28270
    },
    {
      "epoch": 0.496497480643972,
      "grad_norm": 0.047856338197692215,
      "learning_rate": 0.00019681864725000126,
      "loss": 0.8632,
      "step": 28280
    },
    {
      "epoch": 0.4966730455239734,
      "grad_norm": 0.05817185207131404,
      "learning_rate": 0.0001968158443549855,
      "loss": 0.857,
      "step": 28290
    },
    {
      "epoch": 0.49684861040397477,
      "grad_norm": 0.06772494924207881,
      "learning_rate": 0.0001968130402459647,
      "loss": 0.8576,
      "step": 28300
    },
    {
      "epoch": 0.4970241752839762,
      "grad_norm": 0.09495119743544886,
      "learning_rate": 0.0001968102349229743,
      "loss": 0.8505,
      "step": 28310
    },
    {
      "epoch": 0.4971997401639776,
      "grad_norm": 0.08529626702908841,
      "learning_rate": 0.00019680742838604987,
      "loss": 0.8576,
      "step": 28320
    },
    {
      "epoch": 0.497375305043979,
      "grad_norm": 0.0642344918507555,
      "learning_rate": 0.00019680462063522696,
      "loss": 0.86,
      "step": 28330
    },
    {
      "epoch": 0.4975508699239804,
      "grad_norm": 0.054977266448035134,
      "learning_rate": 0.00019680181167054114,
      "loss": 0.8504,
      "step": 28340
    },
    {
      "epoch": 0.4977264348039818,
      "grad_norm": 0.07306428823664031,
      "learning_rate": 0.000196799001492028,
      "loss": 0.8528,
      "step": 28350
    },
    {
      "epoch": 0.49790199968398324,
      "grad_norm": 0.0920889170344716,
      "learning_rate": 0.00019679619009972313,
      "loss": 0.8635,
      "step": 28360
    },
    {
      "epoch": 0.4980775645639846,
      "grad_norm": 0.0765013935219488,
      "learning_rate": 0.00019679337749366224,
      "loss": 0.8551,
      "step": 28370
    },
    {
      "epoch": 0.49825312944398603,
      "grad_norm": 0.06356673850019326,
      "learning_rate": 0.0001967905636738808,
      "loss": 0.8542,
      "step": 28380
    },
    {
      "epoch": 0.49842869432398745,
      "grad_norm": 0.0571872077185213,
      "learning_rate": 0.00019678774864041463,
      "loss": 0.8645,
      "step": 28390
    },
    {
      "epoch": 0.4986042592039888,
      "grad_norm": 0.07088453104304986,
      "learning_rate": 0.00019678493239329932,
      "loss": 0.8545,
      "step": 28400
    },
    {
      "epoch": 0.49877982408399024,
      "grad_norm": 0.07034833057149031,
      "learning_rate": 0.00019678211493257057,
      "loss": 0.8504,
      "step": 28410
    },
    {
      "epoch": 0.49895538896399166,
      "grad_norm": 0.06082019132869783,
      "learning_rate": 0.00019677929625826402,
      "loss": 0.8522,
      "step": 28420
    },
    {
      "epoch": 0.49913095384399303,
      "grad_norm": 0.07014849848342347,
      "learning_rate": 0.0001967764763704155,
      "loss": 0.8541,
      "step": 28430
    },
    {
      "epoch": 0.49930651872399445,
      "grad_norm": 0.0695813818672704,
      "learning_rate": 0.0001967736552690606,
      "loss": 0.8549,
      "step": 28440
    },
    {
      "epoch": 0.4994820836039959,
      "grad_norm": 0.061296807033246845,
      "learning_rate": 0.00019677083295423518,
      "loss": 0.8569,
      "step": 28450
    },
    {
      "epoch": 0.49965764848399724,
      "grad_norm": 0.05164469332146675,
      "learning_rate": 0.00019676800942597492,
      "loss": 0.856,
      "step": 28460
    },
    {
      "epoch": 0.49983321336399866,
      "grad_norm": 0.06187045413124916,
      "learning_rate": 0.00019676518468431561,
      "loss": 0.8566,
      "step": 28470
    },
    {
      "epoch": 0.500008778244,
      "grad_norm": 0.07080208278739544,
      "learning_rate": 0.0001967623587292931,
      "loss": 0.858,
      "step": 28480
    },
    {
      "epoch": 0.5001843431240015,
      "grad_norm": 0.06924266251622617,
      "learning_rate": 0.00019675953156094318,
      "loss": 0.8522,
      "step": 28490
    },
    {
      "epoch": 0.5003599080040029,
      "grad_norm": 0.07479032458144634,
      "learning_rate": 0.0001967567031793016,
      "loss": 0.8476,
      "step": 28500
    },
    {
      "epoch": 0.5005354728840042,
      "grad_norm": 0.0697578274623891,
      "learning_rate": 0.00019675387358440423,
      "loss": 0.8545,
      "step": 28510
    },
    {
      "epoch": 0.5007110377640057,
      "grad_norm": 0.07392040612534022,
      "learning_rate": 0.00019675104277628693,
      "loss": 0.8532,
      "step": 28520
    },
    {
      "epoch": 0.5008866026440071,
      "grad_norm": 0.053056300256200964,
      "learning_rate": 0.00019674821075498557,
      "loss": 0.8567,
      "step": 28530
    },
    {
      "epoch": 0.5010621675240085,
      "grad_norm": 0.06085034952845698,
      "learning_rate": 0.00019674537752053606,
      "loss": 0.8531,
      "step": 28540
    },
    {
      "epoch": 0.5012377324040099,
      "grad_norm": 0.08072573550764119,
      "learning_rate": 0.00019674254307297423,
      "loss": 0.8544,
      "step": 28550
    },
    {
      "epoch": 0.5014132972840113,
      "grad_norm": 0.052338672417524,
      "learning_rate": 0.00019673970741233607,
      "loss": 0.8577,
      "step": 28560
    },
    {
      "epoch": 0.5015888621640127,
      "grad_norm": 0.08874564303887145,
      "learning_rate": 0.00019673687053865742,
      "loss": 0.859,
      "step": 28570
    },
    {
      "epoch": 0.5017644270440141,
      "grad_norm": 0.05031177349109472,
      "learning_rate": 0.00019673403245197427,
      "loss": 0.8612,
      "step": 28580
    },
    {
      "epoch": 0.5019399919240155,
      "grad_norm": 0.07104443204574167,
      "learning_rate": 0.00019673119315232265,
      "loss": 0.8486,
      "step": 28590
    },
    {
      "epoch": 0.5021155568040169,
      "grad_norm": 0.07150738079363231,
      "learning_rate": 0.00019672835263973842,
      "loss": 0.8614,
      "step": 28600
    },
    {
      "epoch": 0.5022911216840183,
      "grad_norm": 0.10211901546966073,
      "learning_rate": 0.0001967255109142576,
      "loss": 0.8521,
      "step": 28610
    },
    {
      "epoch": 0.5024666865640197,
      "grad_norm": 0.05247101739565363,
      "learning_rate": 0.00019672266797591626,
      "loss": 0.8528,
      "step": 28620
    },
    {
      "epoch": 0.5026422514440212,
      "grad_norm": 0.056019336449479755,
      "learning_rate": 0.00019671982382475032,
      "loss": 0.8495,
      "step": 28630
    },
    {
      "epoch": 0.5028178163240226,
      "grad_norm": 0.06005855638448258,
      "learning_rate": 0.0001967169784607959,
      "loss": 0.8453,
      "step": 28640
    },
    {
      "epoch": 0.5029933812040239,
      "grad_norm": 0.06388575917662044,
      "learning_rate": 0.000196714131884089,
      "loss": 0.8582,
      "step": 28650
    },
    {
      "epoch": 0.5031689460840254,
      "grad_norm": 0.06636219963542278,
      "learning_rate": 0.00019671128409466573,
      "loss": 0.8516,
      "step": 28660
    },
    {
      "epoch": 0.5033445109640268,
      "grad_norm": 0.07276744060792363,
      "learning_rate": 0.0001967084350925621,
      "loss": 0.8526,
      "step": 28670
    },
    {
      "epoch": 0.5035200758440281,
      "grad_norm": 0.06609971040239032,
      "learning_rate": 0.0001967055848778143,
      "loss": 0.857,
      "step": 28680
    },
    {
      "epoch": 0.5036956407240296,
      "grad_norm": 0.0539923723641777,
      "learning_rate": 0.0001967027334504584,
      "loss": 0.8623,
      "step": 28690
    },
    {
      "epoch": 0.503871205604031,
      "grad_norm": 0.06104817812943375,
      "learning_rate": 0.0001966998808105305,
      "loss": 0.8606,
      "step": 28700
    },
    {
      "epoch": 0.5040467704840323,
      "grad_norm": 0.05873168959172005,
      "learning_rate": 0.00019669702695806676,
      "loss": 0.8607,
      "step": 28710
    },
    {
      "epoch": 0.5042223353640338,
      "grad_norm": 0.05607111484339036,
      "learning_rate": 0.0001966941718931034,
      "loss": 0.8607,
      "step": 28720
    },
    {
      "epoch": 0.5043979002440352,
      "grad_norm": 0.10451295084450993,
      "learning_rate": 0.0001966913156156765,
      "loss": 0.8548,
      "step": 28730
    },
    {
      "epoch": 0.5045734651240366,
      "grad_norm": 0.08709305290308908,
      "learning_rate": 0.0001966884581258223,
      "loss": 0.8511,
      "step": 28740
    },
    {
      "epoch": 0.504749030004038,
      "grad_norm": 0.07016128947802401,
      "learning_rate": 0.00019668559942357698,
      "loss": 0.8521,
      "step": 28750
    },
    {
      "epoch": 0.5049245948840394,
      "grad_norm": 0.05707602524864542,
      "learning_rate": 0.0001966827395089768,
      "loss": 0.863,
      "step": 28760
    },
    {
      "epoch": 0.5051001597640408,
      "grad_norm": 0.05913228237151569,
      "learning_rate": 0.00019667987838205799,
      "loss": 0.8518,
      "step": 28770
    },
    {
      "epoch": 0.5052757246440422,
      "grad_norm": 0.052697905057215604,
      "learning_rate": 0.00019667701604285677,
      "loss": 0.8579,
      "step": 28780
    },
    {
      "epoch": 0.5054512895240436,
      "grad_norm": 0.08582442836068573,
      "learning_rate": 0.0001966741524914094,
      "loss": 0.8536,
      "step": 28790
    },
    {
      "epoch": 0.505626854404045,
      "grad_norm": 0.06950675657062483,
      "learning_rate": 0.0001966712877277522,
      "loss": 0.8558,
      "step": 28800
    },
    {
      "epoch": 0.5058024192840465,
      "grad_norm": 0.06060124014249393,
      "learning_rate": 0.00019666842175192147,
      "loss": 0.8545,
      "step": 28810
    },
    {
      "epoch": 0.5059779841640478,
      "grad_norm": 0.06256384173547858,
      "learning_rate": 0.0001966655545639535,
      "loss": 0.8571,
      "step": 28820
    },
    {
      "epoch": 0.5061535490440492,
      "grad_norm": 0.07757160036396411,
      "learning_rate": 0.00019666268616388454,
      "loss": 0.8552,
      "step": 28830
    },
    {
      "epoch": 0.5063291139240507,
      "grad_norm": 0.08470328858370337,
      "learning_rate": 0.00019665981655175106,
      "loss": 0.8562,
      "step": 28840
    },
    {
      "epoch": 0.506504678804052,
      "grad_norm": 0.05498181256171088,
      "learning_rate": 0.0001966569457275894,
      "loss": 0.8587,
      "step": 28850
    },
    {
      "epoch": 0.5066802436840534,
      "grad_norm": 0.07268950193000177,
      "learning_rate": 0.0001966540736914359,
      "loss": 0.8567,
      "step": 28860
    },
    {
      "epoch": 0.5068558085640549,
      "grad_norm": 0.07515760630461223,
      "learning_rate": 0.00019665120044332692,
      "loss": 0.8579,
      "step": 28870
    },
    {
      "epoch": 0.5070313734440562,
      "grad_norm": 0.04815890347876562,
      "learning_rate": 0.00019664832598329893,
      "loss": 0.8578,
      "step": 28880
    },
    {
      "epoch": 0.5072069383240576,
      "grad_norm": 0.06363460532772033,
      "learning_rate": 0.00019664545031138831,
      "loss": 0.8499,
      "step": 28890
    },
    {
      "epoch": 0.5073825032040591,
      "grad_norm": 0.0693817560485975,
      "learning_rate": 0.0001966425734276315,
      "loss": 0.8513,
      "step": 28900
    },
    {
      "epoch": 0.5075580680840605,
      "grad_norm": 0.06500542995974883,
      "learning_rate": 0.00019663969533206496,
      "loss": 0.8609,
      "step": 28910
    },
    {
      "epoch": 0.5077336329640618,
      "grad_norm": 0.06763513668813761,
      "learning_rate": 0.00019663681602472512,
      "loss": 0.8584,
      "step": 28920
    },
    {
      "epoch": 0.5079091978440633,
      "grad_norm": 0.07890750280203841,
      "learning_rate": 0.00019663393550564852,
      "loss": 0.8565,
      "step": 28930
    },
    {
      "epoch": 0.5080847627240647,
      "grad_norm": 0.08023626402463954,
      "learning_rate": 0.00019663105377487164,
      "loss": 0.8604,
      "step": 28940
    },
    {
      "epoch": 0.508260327604066,
      "grad_norm": 0.060796979829604914,
      "learning_rate": 0.00019662817083243098,
      "loss": 0.8572,
      "step": 28950
    },
    {
      "epoch": 0.5084358924840675,
      "grad_norm": 0.08642459421522944,
      "learning_rate": 0.00019662528667836304,
      "loss": 0.8513,
      "step": 28960
    },
    {
      "epoch": 0.5086114573640689,
      "grad_norm": 0.05317917639198606,
      "learning_rate": 0.00019662240131270442,
      "loss": 0.8541,
      "step": 28970
    },
    {
      "epoch": 0.5087870222440702,
      "grad_norm": 0.058196154945924866,
      "learning_rate": 0.00019661951473549163,
      "loss": 0.8621,
      "step": 28980
    },
    {
      "epoch": 0.5089625871240717,
      "grad_norm": 0.07306531977394666,
      "learning_rate": 0.0001966166269467613,
      "loss": 0.8586,
      "step": 28990
    },
    {
      "epoch": 0.5091381520040731,
      "grad_norm": 0.057092530454340304,
      "learning_rate": 0.00019661373794654995,
      "loss": 0.8596,
      "step": 29000
    },
    {
      "epoch": 0.5093137168840746,
      "grad_norm": 0.07431827672749158,
      "learning_rate": 0.0001966108477348942,
      "loss": 0.8559,
      "step": 29010
    },
    {
      "epoch": 0.5094892817640759,
      "grad_norm": 0.08475306322635409,
      "learning_rate": 0.0001966079563118307,
      "loss": 0.8561,
      "step": 29020
    },
    {
      "epoch": 0.5096648466440773,
      "grad_norm": 0.06745497913107734,
      "learning_rate": 0.00019660506367739607,
      "loss": 0.8577,
      "step": 29030
    },
    {
      "epoch": 0.5098404115240788,
      "grad_norm": 0.10322730831876117,
      "learning_rate": 0.000196602169831627,
      "loss": 0.8573,
      "step": 29040
    },
    {
      "epoch": 0.5100159764040801,
      "grad_norm": 0.0575897692328454,
      "learning_rate": 0.00019659927477456003,
      "loss": 0.855,
      "step": 29050
    },
    {
      "epoch": 0.5101915412840815,
      "grad_norm": 0.05472248268909245,
      "learning_rate": 0.00019659637850623203,
      "loss": 0.8524,
      "step": 29060
    },
    {
      "epoch": 0.510367106164083,
      "grad_norm": 0.05995420307313575,
      "learning_rate": 0.00019659348102667951,
      "loss": 0.8648,
      "step": 29070
    },
    {
      "epoch": 0.5105426710440844,
      "grad_norm": 0.07341081443038537,
      "learning_rate": 0.00019659058233593933,
      "loss": 0.8601,
      "step": 29080
    },
    {
      "epoch": 0.5107182359240857,
      "grad_norm": 0.07080951659032095,
      "learning_rate": 0.0001965876824340481,
      "loss": 0.8434,
      "step": 29090
    },
    {
      "epoch": 0.5108938008040872,
      "grad_norm": 0.07071285257480808,
      "learning_rate": 0.00019658478132104264,
      "loss": 0.8471,
      "step": 29100
    },
    {
      "epoch": 0.5110693656840886,
      "grad_norm": 0.09505061167855376,
      "learning_rate": 0.00019658187899695967,
      "loss": 0.8484,
      "step": 29110
    },
    {
      "epoch": 0.5112449305640899,
      "grad_norm": 0.0820942882120441,
      "learning_rate": 0.00019657897546183597,
      "loss": 0.8474,
      "step": 29120
    },
    {
      "epoch": 0.5114204954440914,
      "grad_norm": 0.062400303136284654,
      "learning_rate": 0.00019657607071570836,
      "loss": 0.864,
      "step": 29130
    },
    {
      "epoch": 0.5115960603240928,
      "grad_norm": 0.07034939096987408,
      "learning_rate": 0.00019657316475861362,
      "loss": 0.8592,
      "step": 29140
    },
    {
      "epoch": 0.5117716252040941,
      "grad_norm": 0.0687911901552609,
      "learning_rate": 0.0001965702575905885,
      "loss": 0.8526,
      "step": 29150
    },
    {
      "epoch": 0.5119471900840956,
      "grad_norm": 0.0577520726358231,
      "learning_rate": 0.00019656734921166995,
      "loss": 0.8552,
      "step": 29160
    },
    {
      "epoch": 0.512122754964097,
      "grad_norm": 0.06779404537847444,
      "learning_rate": 0.00019656443962189478,
      "loss": 0.8576,
      "step": 29170
    },
    {
      "epoch": 0.5122983198440983,
      "grad_norm": 0.056556139365457336,
      "learning_rate": 0.00019656152882129984,
      "loss": 0.8479,
      "step": 29180
    },
    {
      "epoch": 0.5124738847240998,
      "grad_norm": 0.05799334908046857,
      "learning_rate": 0.000196558616809922,
      "loss": 0.8614,
      "step": 29190
    },
    {
      "epoch": 0.5126494496041012,
      "grad_norm": 0.072061132236632,
      "learning_rate": 0.0001965557035877982,
      "loss": 0.8551,
      "step": 29200
    },
    {
      "epoch": 0.5128250144841026,
      "grad_norm": 0.0621255207250955,
      "learning_rate": 0.00019655278915496526,
      "loss": 0.8544,
      "step": 29210
    },
    {
      "epoch": 0.513000579364104,
      "grad_norm": 0.07131089839899371,
      "learning_rate": 0.00019654987351146023,
      "loss": 0.8572,
      "step": 29220
    },
    {
      "epoch": 0.5131761442441054,
      "grad_norm": 0.08437397224596317,
      "learning_rate": 0.00019654695665731996,
      "loss": 0.8485,
      "step": 29230
    },
    {
      "epoch": 0.5133517091241068,
      "grad_norm": 0.0870093742261487,
      "learning_rate": 0.00019654403859258146,
      "loss": 0.8558,
      "step": 29240
    },
    {
      "epoch": 0.5135272740041082,
      "grad_norm": 0.06494989286053843,
      "learning_rate": 0.00019654111931728164,
      "loss": 0.8536,
      "step": 29250
    },
    {
      "epoch": 0.5137028388841096,
      "grad_norm": 0.059563080023402645,
      "learning_rate": 0.0001965381988314576,
      "loss": 0.8616,
      "step": 29260
    },
    {
      "epoch": 0.513878403764111,
      "grad_norm": 0.058950337559111624,
      "learning_rate": 0.0001965352771351462,
      "loss": 0.8474,
      "step": 29270
    },
    {
      "epoch": 0.5140539686441125,
      "grad_norm": 0.06257656365168376,
      "learning_rate": 0.00019653235422838458,
      "loss": 0.8568,
      "step": 29280
    },
    {
      "epoch": 0.5142295335241138,
      "grad_norm": 0.07248588077789765,
      "learning_rate": 0.00019652943011120967,
      "loss": 0.8531,
      "step": 29290
    },
    {
      "epoch": 0.5144050984041152,
      "grad_norm": 0.06747371814498279,
      "learning_rate": 0.00019652650478365862,
      "loss": 0.8489,
      "step": 29300
    },
    {
      "epoch": 0.5145806632841167,
      "grad_norm": 0.06494721004421337,
      "learning_rate": 0.00019652357824576845,
      "loss": 0.8592,
      "step": 29310
    },
    {
      "epoch": 0.514756228164118,
      "grad_norm": 0.07534076794443097,
      "learning_rate": 0.0001965206504975762,
      "loss": 0.8531,
      "step": 29320
    },
    {
      "epoch": 0.5149317930441194,
      "grad_norm": 0.05588054727341398,
      "learning_rate": 0.000196517721539119,
      "loss": 0.8573,
      "step": 29330
    },
    {
      "epoch": 0.5151073579241209,
      "grad_norm": 0.07040679177811111,
      "learning_rate": 0.00019651479137043398,
      "loss": 0.8549,
      "step": 29340
    },
    {
      "epoch": 0.5152829228041222,
      "grad_norm": 0.06258757124726429,
      "learning_rate": 0.0001965118599915583,
      "loss": 0.848,
      "step": 29350
    },
    {
      "epoch": 0.5154584876841237,
      "grad_norm": 0.0743451128721037,
      "learning_rate": 0.00019650892740252895,
      "loss": 0.8619,
      "step": 29360
    },
    {
      "epoch": 0.5156340525641251,
      "grad_norm": 0.0625483912213372,
      "learning_rate": 0.00019650599360338322,
      "loss": 0.8502,
      "step": 29370
    },
    {
      "epoch": 0.5158096174441265,
      "grad_norm": 0.06596670612278381,
      "learning_rate": 0.00019650305859415828,
      "loss": 0.8517,
      "step": 29380
    },
    {
      "epoch": 0.5159851823241279,
      "grad_norm": 0.06149868108193842,
      "learning_rate": 0.00019650012237489126,
      "loss": 0.8593,
      "step": 29390
    },
    {
      "epoch": 0.5161607472041293,
      "grad_norm": 0.06414077129217328,
      "learning_rate": 0.0001964971849456194,
      "loss": 0.8573,
      "step": 29400
    },
    {
      "epoch": 0.5163363120841307,
      "grad_norm": 0.05483397338438271,
      "learning_rate": 0.00019649424630637988,
      "loss": 0.8667,
      "step": 29410
    },
    {
      "epoch": 0.5165118769641321,
      "grad_norm": 0.06486684862629184,
      "learning_rate": 0.00019649130645720998,
      "loss": 0.8464,
      "step": 29420
    },
    {
      "epoch": 0.5166874418441335,
      "grad_norm": 0.06471348189800105,
      "learning_rate": 0.00019648836539814691,
      "loss": 0.8532,
      "step": 29430
    },
    {
      "epoch": 0.5168630067241349,
      "grad_norm": 0.06002676805311119,
      "learning_rate": 0.00019648542312922799,
      "loss": 0.8608,
      "step": 29440
    },
    {
      "epoch": 0.5170385716041364,
      "grad_norm": 0.0729591386512088,
      "learning_rate": 0.0001964824796504904,
      "loss": 0.8536,
      "step": 29450
    },
    {
      "epoch": 0.5172141364841377,
      "grad_norm": 0.06549644377653978,
      "learning_rate": 0.00019647953496197153,
      "loss": 0.8535,
      "step": 29460
    },
    {
      "epoch": 0.5173897013641391,
      "grad_norm": 0.07681369193735942,
      "learning_rate": 0.00019647658906370865,
      "loss": 0.859,
      "step": 29470
    },
    {
      "epoch": 0.5175652662441406,
      "grad_norm": 0.05636037831782027,
      "learning_rate": 0.0001964736419557391,
      "loss": 0.8605,
      "step": 29480
    },
    {
      "epoch": 0.5177408311241419,
      "grad_norm": 0.05264371463297658,
      "learning_rate": 0.0001964706936381002,
      "loss": 0.8548,
      "step": 29490
    },
    {
      "epoch": 0.5179163960041433,
      "grad_norm": 0.06909741107248228,
      "learning_rate": 0.0001964677441108293,
      "loss": 0.8569,
      "step": 29500
    },
    {
      "epoch": 0.5180919608841448,
      "grad_norm": 0.08323226397303046,
      "learning_rate": 0.00019646479337396378,
      "loss": 0.8614,
      "step": 29510
    },
    {
      "epoch": 0.5182675257641461,
      "grad_norm": 0.06503661042138759,
      "learning_rate": 0.00019646184142754107,
      "loss": 0.8495,
      "step": 29520
    },
    {
      "epoch": 0.5184430906441475,
      "grad_norm": 0.05664063271067172,
      "learning_rate": 0.0001964588882715985,
      "loss": 0.8568,
      "step": 29530
    },
    {
      "epoch": 0.518618655524149,
      "grad_norm": 0.06746343422238885,
      "learning_rate": 0.0001964559339061735,
      "loss": 0.847,
      "step": 29540
    },
    {
      "epoch": 0.5187942204041504,
      "grad_norm": 0.0603684990626405,
      "learning_rate": 0.0001964529783313036,
      "loss": 0.855,
      "step": 29550
    },
    {
      "epoch": 0.5189697852841517,
      "grad_norm": 0.06908034842187188,
      "learning_rate": 0.0001964500215470261,
      "loss": 0.8478,
      "step": 29560
    },
    {
      "epoch": 0.5191453501641532,
      "grad_norm": 0.04839023508258588,
      "learning_rate": 0.00019644706355337855,
      "loss": 0.8545,
      "step": 29570
    },
    {
      "epoch": 0.5193209150441546,
      "grad_norm": 0.0787462272429842,
      "learning_rate": 0.00019644410435039843,
      "loss": 0.8535,
      "step": 29580
    },
    {
      "epoch": 0.5194964799241559,
      "grad_norm": 0.07626882626330007,
      "learning_rate": 0.00019644114393812322,
      "loss": 0.8545,
      "step": 29590
    },
    {
      "epoch": 0.5196720448041574,
      "grad_norm": 0.0729947224820148,
      "learning_rate": 0.0001964381823165904,
      "loss": 0.8582,
      "step": 29600
    },
    {
      "epoch": 0.5198476096841588,
      "grad_norm": 0.0686030957666668,
      "learning_rate": 0.00019643521948583756,
      "loss": 0.8505,
      "step": 29610
    },
    {
      "epoch": 0.5200231745641601,
      "grad_norm": 0.053398033909572984,
      "learning_rate": 0.0001964322554459022,
      "loss": 0.8585,
      "step": 29620
    },
    {
      "epoch": 0.5201987394441616,
      "grad_norm": 0.06755765140566246,
      "learning_rate": 0.0001964292901968218,
      "loss": 0.8583,
      "step": 29630
    },
    {
      "epoch": 0.520374304324163,
      "grad_norm": 0.06553384308571403,
      "learning_rate": 0.00019642632373863405,
      "loss": 0.8566,
      "step": 29640
    },
    {
      "epoch": 0.5205498692041644,
      "grad_norm": 0.059842833668989986,
      "learning_rate": 0.0001964233560713765,
      "loss": 0.8625,
      "step": 29650
    },
    {
      "epoch": 0.5207254340841658,
      "grad_norm": 0.07104715243923304,
      "learning_rate": 0.00019642038719508672,
      "loss": 0.8531,
      "step": 29660
    },
    {
      "epoch": 0.5209009989641672,
      "grad_norm": 0.06124429862190087,
      "learning_rate": 0.00019641741710980237,
      "loss": 0.8532,
      "step": 29670
    },
    {
      "epoch": 0.5210765638441686,
      "grad_norm": 0.0558207752265672,
      "learning_rate": 0.00019641444581556107,
      "loss": 0.8578,
      "step": 29680
    },
    {
      "epoch": 0.52125212872417,
      "grad_norm": 0.07738167998989112,
      "learning_rate": 0.0001964114733124004,
      "loss": 0.8619,
      "step": 29690
    },
    {
      "epoch": 0.5214276936041714,
      "grad_norm": 0.056507839073384664,
      "learning_rate": 0.00019640849960035812,
      "loss": 0.8572,
      "step": 29700
    },
    {
      "epoch": 0.5216032584841728,
      "grad_norm": 0.09869250331127731,
      "learning_rate": 0.00019640552467947185,
      "loss": 0.8501,
      "step": 29710
    },
    {
      "epoch": 0.5217788233641742,
      "grad_norm": 0.0625304730553803,
      "learning_rate": 0.0001964025485497793,
      "loss": 0.8536,
      "step": 29720
    },
    {
      "epoch": 0.5219543882441756,
      "grad_norm": 0.08749242438326142,
      "learning_rate": 0.00019639957121131817,
      "loss": 0.8533,
      "step": 29730
    },
    {
      "epoch": 0.5221299531241771,
      "grad_norm": 0.05239805550305343,
      "learning_rate": 0.00019639659266412624,
      "loss": 0.8575,
      "step": 29740
    },
    {
      "epoch": 0.5223055180041785,
      "grad_norm": 0.08100977759558976,
      "learning_rate": 0.00019639361290824114,
      "loss": 0.857,
      "step": 29750
    },
    {
      "epoch": 0.5224810828841798,
      "grad_norm": 0.06140604991752547,
      "learning_rate": 0.00019639063194370072,
      "loss": 0.8603,
      "step": 29760
    },
    {
      "epoch": 0.5226566477641813,
      "grad_norm": 0.09263348759768697,
      "learning_rate": 0.00019638764977054264,
      "loss": 0.8555,
      "step": 29770
    },
    {
      "epoch": 0.5228322126441827,
      "grad_norm": 0.061244343164706755,
      "learning_rate": 0.0001963846663888048,
      "loss": 0.8556,
      "step": 29780
    },
    {
      "epoch": 0.523007777524184,
      "grad_norm": 0.06864376755632137,
      "learning_rate": 0.000196381681798525,
      "loss": 0.8526,
      "step": 29790
    },
    {
      "epoch": 0.5231833424041855,
      "grad_norm": 0.053288175408355204,
      "learning_rate": 0.000196378695999741,
      "loss": 0.857,
      "step": 29800
    },
    {
      "epoch": 0.5233589072841869,
      "grad_norm": 0.05846528181562085,
      "learning_rate": 0.0001963757089924906,
      "loss": 0.8467,
      "step": 29810
    },
    {
      "epoch": 0.5235344721641882,
      "grad_norm": 0.07747761914494161,
      "learning_rate": 0.0001963727207768117,
      "loss": 0.8422,
      "step": 29820
    },
    {
      "epoch": 0.5237100370441897,
      "grad_norm": 0.06579598774786691,
      "learning_rate": 0.00019636973135274215,
      "loss": 0.858,
      "step": 29830
    },
    {
      "epoch": 0.5238856019241911,
      "grad_norm": 0.09312063864804114,
      "learning_rate": 0.00019636674072031983,
      "loss": 0.8552,
      "step": 29840
    },
    {
      "epoch": 0.5240611668041925,
      "grad_norm": 0.08032511618278704,
      "learning_rate": 0.00019636374887958263,
      "loss": 0.8537,
      "step": 29850
    },
    {
      "epoch": 0.5242367316841939,
      "grad_norm": 0.07415918580168729,
      "learning_rate": 0.00019636075583056849,
      "loss": 0.865,
      "step": 29860
    },
    {
      "epoch": 0.5244122965641953,
      "grad_norm": 0.05753984393376343,
      "learning_rate": 0.00019635776157331524,
      "loss": 0.85,
      "step": 29870
    },
    {
      "epoch": 0.5245878614441967,
      "grad_norm": 0.05604963030753635,
      "learning_rate": 0.0001963547661078609,
      "loss": 0.8659,
      "step": 29880
    },
    {
      "epoch": 0.5247634263241981,
      "grad_norm": 0.06785437646367144,
      "learning_rate": 0.0001963517694342434,
      "loss": 0.8591,
      "step": 29890
    },
    {
      "epoch": 0.5249389912041995,
      "grad_norm": 0.09506306814963333,
      "learning_rate": 0.00019634877155250068,
      "loss": 0.8546,
      "step": 29900
    },
    {
      "epoch": 0.5251145560842009,
      "grad_norm": 0.06974750303142745,
      "learning_rate": 0.00019634577246267082,
      "loss": 0.8591,
      "step": 29910
    },
    {
      "epoch": 0.5252901209642024,
      "grad_norm": 0.06729471802471466,
      "learning_rate": 0.0001963427721647917,
      "loss": 0.8595,
      "step": 29920
    },
    {
      "epoch": 0.5254656858442037,
      "grad_norm": 0.06782281515909792,
      "learning_rate": 0.00019633977065890138,
      "loss": 0.8566,
      "step": 29930
    },
    {
      "epoch": 0.5256412507242051,
      "grad_norm": 0.06525861458042002,
      "learning_rate": 0.0001963367679450379,
      "loss": 0.8521,
      "step": 29940
    },
    {
      "epoch": 0.5258168156042066,
      "grad_norm": 0.058256550401721216,
      "learning_rate": 0.00019633376402323932,
      "loss": 0.856,
      "step": 29950
    },
    {
      "epoch": 0.5259923804842079,
      "grad_norm": 0.06186496869053203,
      "learning_rate": 0.00019633075889354367,
      "loss": 0.8598,
      "step": 29960
    },
    {
      "epoch": 0.5261679453642093,
      "grad_norm": 0.06698776523393773,
      "learning_rate": 0.00019632775255598898,
      "loss": 0.8539,
      "step": 29970
    },
    {
      "epoch": 0.5263435102442108,
      "grad_norm": 0.074650276538104,
      "learning_rate": 0.00019632474501061348,
      "loss": 0.8621,
      "step": 29980
    },
    {
      "epoch": 0.5265190751242121,
      "grad_norm": 0.056689804999970596,
      "learning_rate": 0.00019632173625745516,
      "loss": 0.8576,
      "step": 29990
    },
    {
      "epoch": 0.5266946400042135,
      "grad_norm": 0.06061149881854521,
      "learning_rate": 0.00019631872629655216,
      "loss": 0.8573,
      "step": 30000
    },
    {
      "epoch": 0.526870204884215,
      "grad_norm": 0.07078620941328169,
      "learning_rate": 0.00019631571512794261,
      "loss": 0.8547,
      "step": 30010
    },
    {
      "epoch": 0.5270457697642164,
      "grad_norm": 0.06992207484519342,
      "learning_rate": 0.00019631270275166472,
      "loss": 0.8515,
      "step": 30020
    },
    {
      "epoch": 0.5272213346442177,
      "grad_norm": 0.049878999241611315,
      "learning_rate": 0.00019630968916775662,
      "loss": 0.8561,
      "step": 30030
    },
    {
      "epoch": 0.5273968995242192,
      "grad_norm": 0.07422145303211666,
      "learning_rate": 0.00019630667437625648,
      "loss": 0.8544,
      "step": 30040
    },
    {
      "epoch": 0.5275724644042206,
      "grad_norm": 0.08524446501905307,
      "learning_rate": 0.0001963036583772025,
      "loss": 0.8527,
      "step": 30050
    },
    {
      "epoch": 0.5277480292842219,
      "grad_norm": 0.057215501819116554,
      "learning_rate": 0.00019630064117063292,
      "loss": 0.8648,
      "step": 30060
    },
    {
      "epoch": 0.5279235941642234,
      "grad_norm": 0.06796894687160122,
      "learning_rate": 0.00019629762275658597,
      "loss": 0.8472,
      "step": 30070
    },
    {
      "epoch": 0.5280991590442248,
      "grad_norm": 0.0668299732384758,
      "learning_rate": 0.00019629460313509985,
      "loss": 0.8589,
      "step": 30080
    },
    {
      "epoch": 0.5282747239242263,
      "grad_norm": 0.06565511938649976,
      "learning_rate": 0.00019629158230621285,
      "loss": 0.8599,
      "step": 30090
    },
    {
      "epoch": 0.5284502888042276,
      "grad_norm": 0.05845835196901997,
      "learning_rate": 0.00019628856026996324,
      "loss": 0.8583,
      "step": 30100
    },
    {
      "epoch": 0.528625853684229,
      "grad_norm": 0.0706525655887078,
      "learning_rate": 0.0001962855370263893,
      "loss": 0.8586,
      "step": 30110
    },
    {
      "epoch": 0.5288014185642305,
      "grad_norm": 0.10266506983966323,
      "learning_rate": 0.00019628251257552935,
      "loss": 0.8512,
      "step": 30120
    },
    {
      "epoch": 0.5289769834442318,
      "grad_norm": 0.0885022510929142,
      "learning_rate": 0.00019627948691742172,
      "loss": 0.8633,
      "step": 30130
    },
    {
      "epoch": 0.5291525483242332,
      "grad_norm": 0.09832188174023573,
      "learning_rate": 0.0001962764600521047,
      "loss": 0.857,
      "step": 30140
    },
    {
      "epoch": 0.5293281132042347,
      "grad_norm": 0.07396252400816616,
      "learning_rate": 0.0001962734319796167,
      "loss": 0.8496,
      "step": 30150
    },
    {
      "epoch": 0.529503678084236,
      "grad_norm": 0.0650703873310173,
      "learning_rate": 0.00019627040269999603,
      "loss": 0.8504,
      "step": 30160
    },
    {
      "epoch": 0.5296792429642374,
      "grad_norm": 0.06432984117066809,
      "learning_rate": 0.00019626737221328113,
      "loss": 0.851,
      "step": 30170
    },
    {
      "epoch": 0.5298548078442389,
      "grad_norm": 0.06143618294659299,
      "learning_rate": 0.00019626434051951038,
      "loss": 0.8623,
      "step": 30180
    },
    {
      "epoch": 0.5300303727242403,
      "grad_norm": 0.05380577223124103,
      "learning_rate": 0.00019626130761872212,
      "loss": 0.8573,
      "step": 30190
    },
    {
      "epoch": 0.5302059376042416,
      "grad_norm": 0.07578576334936009,
      "learning_rate": 0.00019625827351095487,
      "loss": 0.8552,
      "step": 30200
    },
    {
      "epoch": 0.5303815024842431,
      "grad_norm": 0.07081403704294216,
      "learning_rate": 0.00019625523819624704,
      "loss": 0.8544,
      "step": 30210
    },
    {
      "epoch": 0.5305570673642445,
      "grad_norm": 0.06998939231584354,
      "learning_rate": 0.00019625220167463706,
      "loss": 0.8647,
      "step": 30220
    },
    {
      "epoch": 0.5307326322442458,
      "grad_norm": 0.05728505838075256,
      "learning_rate": 0.00019624916394616346,
      "loss": 0.858,
      "step": 30230
    },
    {
      "epoch": 0.5309081971242473,
      "grad_norm": 0.057459633872014006,
      "learning_rate": 0.00019624612501086472,
      "loss": 0.8579,
      "step": 30240
    },
    {
      "epoch": 0.5310837620042487,
      "grad_norm": 0.05466761837763058,
      "learning_rate": 0.00019624308486877927,
      "loss": 0.8549,
      "step": 30250
    },
    {
      "epoch": 0.53125932688425,
      "grad_norm": 0.058019312422236666,
      "learning_rate": 0.00019624004351994568,
      "loss": 0.8593,
      "step": 30260
    },
    {
      "epoch": 0.5314348917642515,
      "grad_norm": 0.06419225351849087,
      "learning_rate": 0.00019623700096440253,
      "loss": 0.8499,
      "step": 30270
    },
    {
      "epoch": 0.5316104566442529,
      "grad_norm": 0.07475582388121765,
      "learning_rate": 0.0001962339572021883,
      "loss": 0.8598,
      "step": 30280
    },
    {
      "epoch": 0.5317860215242542,
      "grad_norm": 0.05967688844256034,
      "learning_rate": 0.0001962309122333416,
      "loss": 0.8574,
      "step": 30290
    },
    {
      "epoch": 0.5319615864042557,
      "grad_norm": 0.053410180101682234,
      "learning_rate": 0.00019622786605790098,
      "loss": 0.8519,
      "step": 30300
    },
    {
      "epoch": 0.5321371512842571,
      "grad_norm": 0.057012365140871726,
      "learning_rate": 0.00019622481867590503,
      "loss": 0.8552,
      "step": 30310
    },
    {
      "epoch": 0.5323127161642585,
      "grad_norm": 0.07147933977304959,
      "learning_rate": 0.00019622177008739243,
      "loss": 0.8542,
      "step": 30320
    },
    {
      "epoch": 0.5324882810442599,
      "grad_norm": 0.07653244976396063,
      "learning_rate": 0.0001962187202924017,
      "loss": 0.856,
      "step": 30330
    },
    {
      "epoch": 0.5326638459242613,
      "grad_norm": 0.08102451666489424,
      "learning_rate": 0.00019621566929097154,
      "loss": 0.8543,
      "step": 30340
    },
    {
      "epoch": 0.5328394108042627,
      "grad_norm": 0.06737488582096392,
      "learning_rate": 0.00019621261708314063,
      "loss": 0.859,
      "step": 30350
    },
    {
      "epoch": 0.5330149756842641,
      "grad_norm": 0.09027076721190101,
      "learning_rate": 0.00019620956366894757,
      "loss": 0.8566,
      "step": 30360
    },
    {
      "epoch": 0.5331905405642655,
      "grad_norm": 0.0672266587577479,
      "learning_rate": 0.00019620650904843113,
      "loss": 0.8581,
      "step": 30370
    },
    {
      "epoch": 0.5333661054442669,
      "grad_norm": 0.06824188088027007,
      "learning_rate": 0.00019620345322162995,
      "loss": 0.8496,
      "step": 30380
    },
    {
      "epoch": 0.5335416703242684,
      "grad_norm": 0.060706502560839856,
      "learning_rate": 0.00019620039618858278,
      "loss": 0.8591,
      "step": 30390
    },
    {
      "epoch": 0.5337172352042697,
      "grad_norm": 0.06759943096142017,
      "learning_rate": 0.00019619733794932836,
      "loss": 0.8565,
      "step": 30400
    },
    {
      "epoch": 0.5338928000842711,
      "grad_norm": 0.061694404328551264,
      "learning_rate": 0.00019619427850390542,
      "loss": 0.8606,
      "step": 30410
    },
    {
      "epoch": 0.5340683649642726,
      "grad_norm": 0.06231748044261503,
      "learning_rate": 0.0001961912178523527,
      "loss": 0.8553,
      "step": 30420
    },
    {
      "epoch": 0.5342439298442739,
      "grad_norm": 0.06729876285049984,
      "learning_rate": 0.00019618815599470904,
      "loss": 0.8565,
      "step": 30430
    },
    {
      "epoch": 0.5344194947242753,
      "grad_norm": 0.05956723198355466,
      "learning_rate": 0.0001961850929310132,
      "loss": 0.8541,
      "step": 30440
    },
    {
      "epoch": 0.5345950596042768,
      "grad_norm": 0.06320112852304388,
      "learning_rate": 0.00019618202866130397,
      "loss": 0.8584,
      "step": 30450
    },
    {
      "epoch": 0.5347706244842781,
      "grad_norm": 0.0815508741179819,
      "learning_rate": 0.0001961789631856202,
      "loss": 0.8506,
      "step": 30460
    },
    {
      "epoch": 0.5349461893642796,
      "grad_norm": 0.07613068479830905,
      "learning_rate": 0.00019617589650400074,
      "loss": 0.8582,
      "step": 30470
    },
    {
      "epoch": 0.535121754244281,
      "grad_norm": 0.07708838474025331,
      "learning_rate": 0.0001961728286164844,
      "loss": 0.8587,
      "step": 30480
    },
    {
      "epoch": 0.5352973191242824,
      "grad_norm": 0.06766788017720894,
      "learning_rate": 0.00019616975952311012,
      "loss": 0.8545,
      "step": 30490
    },
    {
      "epoch": 0.5354728840042838,
      "grad_norm": 0.05073342521741752,
      "learning_rate": 0.00019616668922391674,
      "loss": 0.8557,
      "step": 30500
    },
    {
      "epoch": 0.5356484488842852,
      "grad_norm": 0.06159435314787109,
      "learning_rate": 0.00019616361771894315,
      "loss": 0.8606,
      "step": 30510
    },
    {
      "epoch": 0.5358240137642866,
      "grad_norm": 0.06280479216584198,
      "learning_rate": 0.0001961605450082283,
      "loss": 0.8538,
      "step": 30520
    },
    {
      "epoch": 0.535999578644288,
      "grad_norm": 0.06705575635542556,
      "learning_rate": 0.0001961574710918111,
      "loss": 0.8602,
      "step": 30530
    },
    {
      "epoch": 0.5361751435242894,
      "grad_norm": 0.05553967410853268,
      "learning_rate": 0.00019615439596973048,
      "loss": 0.8528,
      "step": 30540
    },
    {
      "epoch": 0.5363507084042908,
      "grad_norm": 0.10522535807607154,
      "learning_rate": 0.00019615131964202545,
      "loss": 0.8495,
      "step": 30550
    },
    {
      "epoch": 0.5365262732842923,
      "grad_norm": 0.06779916831188469,
      "learning_rate": 0.000196148242108735,
      "loss": 0.8565,
      "step": 30560
    },
    {
      "epoch": 0.5367018381642936,
      "grad_norm": 0.06841085906933309,
      "learning_rate": 0.00019614516336989804,
      "loss": 0.859,
      "step": 30570
    },
    {
      "epoch": 0.536877403044295,
      "grad_norm": 0.07590716610877227,
      "learning_rate": 0.00019614208342555364,
      "loss": 0.8464,
      "step": 30580
    },
    {
      "epoch": 0.5370529679242965,
      "grad_norm": 0.0534862862559443,
      "learning_rate": 0.0001961390022757408,
      "loss": 0.8525,
      "step": 30590
    },
    {
      "epoch": 0.5372285328042978,
      "grad_norm": 0.05944938344393981,
      "learning_rate": 0.0001961359199204986,
      "loss": 0.8605,
      "step": 30600
    },
    {
      "epoch": 0.5374040976842992,
      "grad_norm": 0.06265636178463135,
      "learning_rate": 0.00019613283635986603,
      "loss": 0.8582,
      "step": 30610
    },
    {
      "epoch": 0.5375796625643007,
      "grad_norm": 0.06544692966884402,
      "learning_rate": 0.0001961297515938822,
      "loss": 0.8565,
      "step": 30620
    },
    {
      "epoch": 0.537755227444302,
      "grad_norm": 0.09928856325097558,
      "learning_rate": 0.00019612666562258618,
      "loss": 0.8549,
      "step": 30630
    },
    {
      "epoch": 0.5379307923243034,
      "grad_norm": 0.057894183102515484,
      "learning_rate": 0.00019612357844601708,
      "loss": 0.8524,
      "step": 30640
    },
    {
      "epoch": 0.5381063572043049,
      "grad_norm": 0.04813293914371265,
      "learning_rate": 0.00019612049006421402,
      "loss": 0.85,
      "step": 30650
    },
    {
      "epoch": 0.5382819220843063,
      "grad_norm": 0.04955068903160701,
      "learning_rate": 0.0001961174004772161,
      "loss": 0.8529,
      "step": 30660
    },
    {
      "epoch": 0.5384574869643076,
      "grad_norm": 0.062329597648845825,
      "learning_rate": 0.00019611430968506253,
      "loss": 0.8518,
      "step": 30670
    },
    {
      "epoch": 0.5386330518443091,
      "grad_norm": 0.06497113296450348,
      "learning_rate": 0.00019611121768779236,
      "loss": 0.8582,
      "step": 30680
    },
    {
      "epoch": 0.5388086167243105,
      "grad_norm": 0.061216124242441945,
      "learning_rate": 0.0001961081244854449,
      "loss": 0.847,
      "step": 30690
    },
    {
      "epoch": 0.5389841816043118,
      "grad_norm": 0.05353900989879404,
      "learning_rate": 0.00019610503007805923,
      "loss": 0.8606,
      "step": 30700
    },
    {
      "epoch": 0.5391597464843133,
      "grad_norm": 0.05783572518683303,
      "learning_rate": 0.00019610193446567458,
      "loss": 0.8602,
      "step": 30710
    },
    {
      "epoch": 0.5393353113643147,
      "grad_norm": 0.056585790841109485,
      "learning_rate": 0.00019609883764833027,
      "loss": 0.8557,
      "step": 30720
    },
    {
      "epoch": 0.539510876244316,
      "grad_norm": 0.05221170562159371,
      "learning_rate": 0.00019609573962606542,
      "loss": 0.8523,
      "step": 30730
    },
    {
      "epoch": 0.5396864411243175,
      "grad_norm": 0.05834363663663778,
      "learning_rate": 0.00019609264039891934,
      "loss": 0.8512,
      "step": 30740
    },
    {
      "epoch": 0.5398620060043189,
      "grad_norm": 0.0809738568144274,
      "learning_rate": 0.00019608953996693126,
      "loss": 0.8556,
      "step": 30750
    },
    {
      "epoch": 0.5400375708843203,
      "grad_norm": 0.0696679952259909,
      "learning_rate": 0.00019608643833014051,
      "loss": 0.8551,
      "step": 30760
    },
    {
      "epoch": 0.5402131357643217,
      "grad_norm": 0.06414910892818133,
      "learning_rate": 0.00019608333548858632,
      "loss": 0.8562,
      "step": 30770
    },
    {
      "epoch": 0.5403887006443231,
      "grad_norm": 0.05473219230677461,
      "learning_rate": 0.0001960802314423081,
      "loss": 0.8621,
      "step": 30780
    },
    {
      "epoch": 0.5405642655243245,
      "grad_norm": 0.053320432735048544,
      "learning_rate": 0.0001960771261913451,
      "loss": 0.8539,
      "step": 30790
    },
    {
      "epoch": 0.5407398304043259,
      "grad_norm": 0.06235709531183975,
      "learning_rate": 0.0001960740197357367,
      "loss": 0.8553,
      "step": 30800
    },
    {
      "epoch": 0.5409153952843273,
      "grad_norm": 0.05978238263731294,
      "learning_rate": 0.00019607091207552224,
      "loss": 0.8578,
      "step": 30810
    },
    {
      "epoch": 0.5410909601643288,
      "grad_norm": 0.06387667704979794,
      "learning_rate": 0.00019606780321074107,
      "loss": 0.8571,
      "step": 30820
    },
    {
      "epoch": 0.5412665250443301,
      "grad_norm": 0.07710736470789495,
      "learning_rate": 0.00019606469314143265,
      "loss": 0.8603,
      "step": 30830
    },
    {
      "epoch": 0.5414420899243315,
      "grad_norm": 0.06521402622978321,
      "learning_rate": 0.00019606158186763635,
      "loss": 0.8526,
      "step": 30840
    },
    {
      "epoch": 0.541617654804333,
      "grad_norm": 0.06252391322302357,
      "learning_rate": 0.0001960584693893916,
      "loss": 0.8559,
      "step": 30850
    },
    {
      "epoch": 0.5417932196843344,
      "grad_norm": 0.061241669425140914,
      "learning_rate": 0.00019605535570673782,
      "loss": 0.8604,
      "step": 30860
    },
    {
      "epoch": 0.5419687845643357,
      "grad_norm": 0.06626495373155282,
      "learning_rate": 0.00019605224081971446,
      "loss": 0.8569,
      "step": 30870
    },
    {
      "epoch": 0.5421443494443372,
      "grad_norm": 0.06358856057575256,
      "learning_rate": 0.00019604912472836097,
      "loss": 0.8563,
      "step": 30880
    },
    {
      "epoch": 0.5423199143243386,
      "grad_norm": 0.09675419054697534,
      "learning_rate": 0.00019604600743271686,
      "loss": 0.8613,
      "step": 30890
    },
    {
      "epoch": 0.5424954792043399,
      "grad_norm": 0.059113408003792355,
      "learning_rate": 0.00019604288893282164,
      "loss": 0.8547,
      "step": 30900
    },
    {
      "epoch": 0.5426710440843414,
      "grad_norm": 0.06022339738180512,
      "learning_rate": 0.00019603976922871478,
      "loss": 0.8612,
      "step": 30910
    },
    {
      "epoch": 0.5428466089643428,
      "grad_norm": 0.056798099601534854,
      "learning_rate": 0.00019603664832043587,
      "loss": 0.851,
      "step": 30920
    },
    {
      "epoch": 0.5430221738443441,
      "grad_norm": 0.056533749081201935,
      "learning_rate": 0.00019603352620802436,
      "loss": 0.8486,
      "step": 30930
    },
    {
      "epoch": 0.5431977387243456,
      "grad_norm": 0.09510156735930125,
      "learning_rate": 0.0001960304028915199,
      "loss": 0.8565,
      "step": 30940
    },
    {
      "epoch": 0.543373303604347,
      "grad_norm": 0.04229844551432287,
      "learning_rate": 0.000196027278370962,
      "loss": 0.8603,
      "step": 30950
    },
    {
      "epoch": 0.5435488684843484,
      "grad_norm": 0.060239811817676535,
      "learning_rate": 0.00019602415264639028,
      "loss": 0.8614,
      "step": 30960
    },
    {
      "epoch": 0.5437244333643498,
      "grad_norm": 0.06762920515169867,
      "learning_rate": 0.00019602102571784436,
      "loss": 0.8472,
      "step": 30970
    },
    {
      "epoch": 0.5438999982443512,
      "grad_norm": 0.05664952702937626,
      "learning_rate": 0.00019601789758536377,
      "loss": 0.8568,
      "step": 30980
    },
    {
      "epoch": 0.5440755631243526,
      "grad_norm": 0.08823481670097164,
      "learning_rate": 0.00019601476824898827,
      "loss": 0.8516,
      "step": 30990
    },
    {
      "epoch": 0.544251128004354,
      "grad_norm": 0.05490832266266027,
      "learning_rate": 0.00019601163770875744,
      "loss": 0.8586,
      "step": 31000
    },
    {
      "epoch": 0.5444266928843554,
      "grad_norm": 0.07650527537709481,
      "learning_rate": 0.00019600850596471092,
      "loss": 0.8545,
      "step": 31010
    },
    {
      "epoch": 0.5446022577643568,
      "grad_norm": 0.0547505465974548,
      "learning_rate": 0.00019600537301688842,
      "loss": 0.8509,
      "step": 31020
    },
    {
      "epoch": 0.5447778226443583,
      "grad_norm": 0.08565909626390215,
      "learning_rate": 0.00019600223886532964,
      "loss": 0.8537,
      "step": 31030
    },
    {
      "epoch": 0.5449533875243596,
      "grad_norm": 0.07066931427355183,
      "learning_rate": 0.00019599910351007432,
      "loss": 0.8649,
      "step": 31040
    },
    {
      "epoch": 0.545128952404361,
      "grad_norm": 0.08158795817982192,
      "learning_rate": 0.00019599596695116215,
      "loss": 0.8574,
      "step": 31050
    },
    {
      "epoch": 0.5453045172843625,
      "grad_norm": 0.05932502816743643,
      "learning_rate": 0.00019599282918863283,
      "loss": 0.8603,
      "step": 31060
    },
    {
      "epoch": 0.5454800821643638,
      "grad_norm": 0.05370838824681865,
      "learning_rate": 0.00019598969022252617,
      "loss": 0.853,
      "step": 31070
    },
    {
      "epoch": 0.5456556470443652,
      "grad_norm": 0.06261855347059044,
      "learning_rate": 0.00019598655005288192,
      "loss": 0.8519,
      "step": 31080
    },
    {
      "epoch": 0.5458312119243667,
      "grad_norm": 0.054504655176170175,
      "learning_rate": 0.00019598340867973992,
      "loss": 0.8515,
      "step": 31090
    },
    {
      "epoch": 0.546006776804368,
      "grad_norm": 0.06834320859411637,
      "learning_rate": 0.0001959802661031399,
      "loss": 0.8553,
      "step": 31100
    },
    {
      "epoch": 0.5461823416843694,
      "grad_norm": 0.06619250273961198,
      "learning_rate": 0.00019597712232312172,
      "loss": 0.8577,
      "step": 31110
    },
    {
      "epoch": 0.5463579065643709,
      "grad_norm": 0.08430469435717552,
      "learning_rate": 0.0001959739773397252,
      "loss": 0.8488,
      "step": 31120
    },
    {
      "epoch": 0.5465334714443723,
      "grad_norm": 0.062691773801042,
      "learning_rate": 0.00019597083115299016,
      "loss": 0.8537,
      "step": 31130
    },
    {
      "epoch": 0.5467090363243736,
      "grad_norm": 0.07636042514329008,
      "learning_rate": 0.00019596768376295654,
      "loss": 0.8483,
      "step": 31140
    },
    {
      "epoch": 0.5468846012043751,
      "grad_norm": 0.05347562696605515,
      "learning_rate": 0.0001959645351696641,
      "loss": 0.8542,
      "step": 31150
    },
    {
      "epoch": 0.5470601660843765,
      "grad_norm": 0.08984153980795019,
      "learning_rate": 0.00019596138537315283,
      "loss": 0.8557,
      "step": 31160
    },
    {
      "epoch": 0.5472357309643778,
      "grad_norm": 0.05278466441422529,
      "learning_rate": 0.0001959582343734626,
      "loss": 0.8553,
      "step": 31170
    },
    {
      "epoch": 0.5474112958443793,
      "grad_norm": 0.06887020280374957,
      "learning_rate": 0.00019595508217063335,
      "loss": 0.8497,
      "step": 31180
    },
    {
      "epoch": 0.5475868607243807,
      "grad_norm": 0.06647247352971479,
      "learning_rate": 0.00019595192876470502,
      "loss": 0.8562,
      "step": 31190
    },
    {
      "epoch": 0.5477624256043822,
      "grad_norm": 0.06881807765364689,
      "learning_rate": 0.00019594877415571752,
      "loss": 0.8591,
      "step": 31200
    },
    {
      "epoch": 0.5479379904843835,
      "grad_norm": 0.07713915799901103,
      "learning_rate": 0.00019594561834371092,
      "loss": 0.8543,
      "step": 31210
    },
    {
      "epoch": 0.5481135553643849,
      "grad_norm": 0.07984303996443193,
      "learning_rate": 0.0001959424613287251,
      "loss": 0.8526,
      "step": 31220
    },
    {
      "epoch": 0.5482891202443864,
      "grad_norm": 0.08804034265050545,
      "learning_rate": 0.00019593930311080009,
      "loss": 0.8562,
      "step": 31230
    },
    {
      "epoch": 0.5484646851243877,
      "grad_norm": 0.06776319926171526,
      "learning_rate": 0.0001959361436899759,
      "loss": 0.8568,
      "step": 31240
    },
    {
      "epoch": 0.5486402500043891,
      "grad_norm": 0.06731456618341315,
      "learning_rate": 0.00019593298306629263,
      "loss": 0.8527,
      "step": 31250
    },
    {
      "epoch": 0.5488158148843906,
      "grad_norm": 0.07191946155220058,
      "learning_rate": 0.00019592982123979024,
      "loss": 0.851,
      "step": 31260
    },
    {
      "epoch": 0.5489913797643919,
      "grad_norm": 0.06603551674135963,
      "learning_rate": 0.00019592665821050882,
      "loss": 0.8536,
      "step": 31270
    },
    {
      "epoch": 0.5491669446443933,
      "grad_norm": 0.06994125865469039,
      "learning_rate": 0.0001959234939784884,
      "loss": 0.8555,
      "step": 31280
    },
    {
      "epoch": 0.5493425095243948,
      "grad_norm": 0.06710442787109057,
      "learning_rate": 0.0001959203285437692,
      "loss": 0.857,
      "step": 31290
    },
    {
      "epoch": 0.5495180744043962,
      "grad_norm": 0.07214762271055486,
      "learning_rate": 0.0001959171619063912,
      "loss": 0.8611,
      "step": 31300
    },
    {
      "epoch": 0.5496936392843975,
      "grad_norm": 0.07544033548932451,
      "learning_rate": 0.00019591399406639458,
      "loss": 0.8567,
      "step": 31310
    },
    {
      "epoch": 0.549869204164399,
      "grad_norm": 0.080836020629921,
      "learning_rate": 0.0001959108250238195,
      "loss": 0.8541,
      "step": 31320
    },
    {
      "epoch": 0.5500447690444004,
      "grad_norm": 0.05929326937113656,
      "learning_rate": 0.00019590765477870602,
      "loss": 0.8553,
      "step": 31330
    },
    {
      "epoch": 0.5502203339244017,
      "grad_norm": 0.06155410598352605,
      "learning_rate": 0.00019590448333109439,
      "loss": 0.8508,
      "step": 31340
    },
    {
      "epoch": 0.5503958988044032,
      "grad_norm": 0.05443807569705259,
      "learning_rate": 0.00019590131068102478,
      "loss": 0.8521,
      "step": 31350
    },
    {
      "epoch": 0.5505714636844046,
      "grad_norm": 0.06063489287317628,
      "learning_rate": 0.00019589813682853734,
      "loss": 0.856,
      "step": 31360
    },
    {
      "epoch": 0.5507470285644059,
      "grad_norm": 0.07695275882028457,
      "learning_rate": 0.00019589496177367235,
      "loss": 0.8562,
      "step": 31370
    },
    {
      "epoch": 0.5509225934444074,
      "grad_norm": 0.08067880955928376,
      "learning_rate": 0.00019589178551647,
      "loss": 0.8564,
      "step": 31380
    },
    {
      "epoch": 0.5510981583244088,
      "grad_norm": 0.06923804892431346,
      "learning_rate": 0.00019588860805697054,
      "loss": 0.8496,
      "step": 31390
    },
    {
      "epoch": 0.5512737232044101,
      "grad_norm": 0.052691206177765765,
      "learning_rate": 0.00019588542939521426,
      "loss": 0.8654,
      "step": 31400
    },
    {
      "epoch": 0.5514492880844116,
      "grad_norm": 0.06393804859254164,
      "learning_rate": 0.00019588224953124135,
      "loss": 0.8518,
      "step": 31410
    },
    {
      "epoch": 0.551624852964413,
      "grad_norm": 0.07425297745799572,
      "learning_rate": 0.0001958790684650922,
      "loss": 0.8544,
      "step": 31420
    },
    {
      "epoch": 0.5518004178444144,
      "grad_norm": 0.07324577509927807,
      "learning_rate": 0.00019587588619680702,
      "loss": 0.8629,
      "step": 31430
    },
    {
      "epoch": 0.5519759827244158,
      "grad_norm": 0.06402725019450788,
      "learning_rate": 0.00019587270272642621,
      "loss": 0.856,
      "step": 31440
    },
    {
      "epoch": 0.5521515476044172,
      "grad_norm": 0.06376624815374939,
      "learning_rate": 0.00019586951805399008,
      "loss": 0.8654,
      "step": 31450
    },
    {
      "epoch": 0.5523271124844186,
      "grad_norm": 0.07172595595444321,
      "learning_rate": 0.00019586633217953897,
      "loss": 0.8539,
      "step": 31460
    },
    {
      "epoch": 0.55250267736442,
      "grad_norm": 0.06224971416214019,
      "learning_rate": 0.00019586314510311323,
      "loss": 0.8503,
      "step": 31470
    },
    {
      "epoch": 0.5526782422444214,
      "grad_norm": 0.060967924833527316,
      "learning_rate": 0.0001958599568247533,
      "loss": 0.8563,
      "step": 31480
    },
    {
      "epoch": 0.5528538071244228,
      "grad_norm": 0.052385261010453606,
      "learning_rate": 0.00019585676734449952,
      "loss": 0.8516,
      "step": 31490
    },
    {
      "epoch": 0.5530293720044243,
      "grad_norm": 0.05416649705429727,
      "learning_rate": 0.00019585357666239233,
      "loss": 0.8473,
      "step": 31500
    },
    {
      "epoch": 0.5532049368844256,
      "grad_norm": 0.04253422872447094,
      "learning_rate": 0.0001958503847784721,
      "loss": 0.8617,
      "step": 31510
    },
    {
      "epoch": 0.553380501764427,
      "grad_norm": 0.060460168502510465,
      "learning_rate": 0.00019584719169277935,
      "loss": 0.854,
      "step": 31520
    },
    {
      "epoch": 0.5535560666444285,
      "grad_norm": 0.0620545493050864,
      "learning_rate": 0.0001958439974053545,
      "loss": 0.8445,
      "step": 31530
    },
    {
      "epoch": 0.5537316315244298,
      "grad_norm": 0.058353293539114066,
      "learning_rate": 0.00019584080191623805,
      "loss": 0.8569,
      "step": 31540
    },
    {
      "epoch": 0.5539071964044313,
      "grad_norm": 0.05856229996948684,
      "learning_rate": 0.0001958376052254704,
      "loss": 0.8616,
      "step": 31550
    },
    {
      "epoch": 0.5540827612844327,
      "grad_norm": 0.05200711375220617,
      "learning_rate": 0.00019583440733309218,
      "loss": 0.8587,
      "step": 31560
    },
    {
      "epoch": 0.554258326164434,
      "grad_norm": 0.06762062500426555,
      "learning_rate": 0.00019583120823914383,
      "loss": 0.8613,
      "step": 31570
    },
    {
      "epoch": 0.5544338910444355,
      "grad_norm": 0.05762786111197334,
      "learning_rate": 0.0001958280079436659,
      "loss": 0.8658,
      "step": 31580
    },
    {
      "epoch": 0.5546094559244369,
      "grad_norm": 0.08732367330497617,
      "learning_rate": 0.0001958248064466989,
      "loss": 0.8621,
      "step": 31590
    },
    {
      "epoch": 0.5547850208044383,
      "grad_norm": 0.059696370308523546,
      "learning_rate": 0.00019582160374828343,
      "loss": 0.8555,
      "step": 31600
    },
    {
      "epoch": 0.5549605856844397,
      "grad_norm": 0.0560633481066052,
      "learning_rate": 0.00019581839984846014,
      "loss": 0.8561,
      "step": 31610
    },
    {
      "epoch": 0.5551361505644411,
      "grad_norm": 0.1431558658988991,
      "learning_rate": 0.00019581519474726948,
      "loss": 0.8601,
      "step": 31620
    },
    {
      "epoch": 0.5553117154444425,
      "grad_norm": 0.07375584953127019,
      "learning_rate": 0.00019581198844475218,
      "loss": 0.8558,
      "step": 31630
    },
    {
      "epoch": 0.5554872803244439,
      "grad_norm": 0.05725049858196563,
      "learning_rate": 0.00019580878094094877,
      "loss": 0.8549,
      "step": 31640
    },
    {
      "epoch": 0.5556628452044453,
      "grad_norm": 0.0882034299725775,
      "learning_rate": 0.00019580557223589995,
      "loss": 0.8577,
      "step": 31650
    },
    {
      "epoch": 0.5558384100844467,
      "grad_norm": 0.07256546527978786,
      "learning_rate": 0.00019580236232964637,
      "loss": 0.8492,
      "step": 31660
    },
    {
      "epoch": 0.5560139749644482,
      "grad_norm": 0.06413020605031637,
      "learning_rate": 0.00019579915122222867,
      "loss": 0.8546,
      "step": 31670
    },
    {
      "epoch": 0.5561895398444495,
      "grad_norm": 0.07044355118818768,
      "learning_rate": 0.0001957959389136876,
      "loss": 0.8568,
      "step": 31680
    },
    {
      "epoch": 0.5563651047244509,
      "grad_norm": 0.06980349348485593,
      "learning_rate": 0.00019579272540406377,
      "loss": 0.8645,
      "step": 31690
    },
    {
      "epoch": 0.5565406696044524,
      "grad_norm": 0.058262753208645726,
      "learning_rate": 0.00019578951069339798,
      "loss": 0.861,
      "step": 31700
    },
    {
      "epoch": 0.5567162344844537,
      "grad_norm": 0.08796901541864095,
      "learning_rate": 0.00019578629478173088,
      "loss": 0.8609,
      "step": 31710
    },
    {
      "epoch": 0.5568917993644551,
      "grad_norm": 0.06268754264687475,
      "learning_rate": 0.0001957830776691033,
      "loss": 0.8647,
      "step": 31720
    },
    {
      "epoch": 0.5570673642444566,
      "grad_norm": 0.07622015059842348,
      "learning_rate": 0.0001957798593555559,
      "loss": 0.8478,
      "step": 31730
    },
    {
      "epoch": 0.5572429291244579,
      "grad_norm": 0.08102497416470163,
      "learning_rate": 0.0001957766398411296,
      "loss": 0.8481,
      "step": 31740
    },
    {
      "epoch": 0.5574184940044593,
      "grad_norm": 0.06949774285406789,
      "learning_rate": 0.0001957734191258651,
      "loss": 0.8587,
      "step": 31750
    },
    {
      "epoch": 0.5575940588844608,
      "grad_norm": 0.08512747486995045,
      "learning_rate": 0.00019577019720980315,
      "loss": 0.8451,
      "step": 31760
    },
    {
      "epoch": 0.5577696237644622,
      "grad_norm": 0.08706121732980912,
      "learning_rate": 0.00019576697409298468,
      "loss": 0.8507,
      "step": 31770
    },
    {
      "epoch": 0.5579451886444635,
      "grad_norm": 0.06143680051100377,
      "learning_rate": 0.0001957637497754505,
      "loss": 0.8573,
      "step": 31780
    },
    {
      "epoch": 0.558120753524465,
      "grad_norm": 0.054589618921446016,
      "learning_rate": 0.00019576052425724142,
      "loss": 0.856,
      "step": 31790
    },
    {
      "epoch": 0.5582963184044664,
      "grad_norm": 0.08196412932858765,
      "learning_rate": 0.00019575729753839834,
      "loss": 0.8513,
      "step": 31800
    },
    {
      "epoch": 0.5584718832844677,
      "grad_norm": 0.0541628824593687,
      "learning_rate": 0.00019575406961896215,
      "loss": 0.8523,
      "step": 31810
    },
    {
      "epoch": 0.5586474481644692,
      "grad_norm": 0.0698271835997056,
      "learning_rate": 0.00019575084049897374,
      "loss": 0.8574,
      "step": 31820
    },
    {
      "epoch": 0.5588230130444706,
      "grad_norm": 0.06193716612672238,
      "learning_rate": 0.000195747610178474,
      "loss": 0.8515,
      "step": 31830
    },
    {
      "epoch": 0.5589985779244719,
      "grad_norm": 0.08167504284755635,
      "learning_rate": 0.0001957443786575039,
      "loss": 0.855,
      "step": 31840
    },
    {
      "epoch": 0.5591741428044734,
      "grad_norm": 0.0642995610333742,
      "learning_rate": 0.00019574114593610438,
      "loss": 0.8605,
      "step": 31850
    },
    {
      "epoch": 0.5593497076844748,
      "grad_norm": 0.06712826869952879,
      "learning_rate": 0.00019573791201431636,
      "loss": 0.8557,
      "step": 31860
    },
    {
      "epoch": 0.5595252725644762,
      "grad_norm": 0.06389846978137712,
      "learning_rate": 0.00019573467689218088,
      "loss": 0.8542,
      "step": 31870
    },
    {
      "epoch": 0.5597008374444776,
      "grad_norm": 0.06442824021700329,
      "learning_rate": 0.00019573144056973885,
      "loss": 0.8557,
      "step": 31880
    },
    {
      "epoch": 0.559876402324479,
      "grad_norm": 0.058089365076181794,
      "learning_rate": 0.00019572820304703133,
      "loss": 0.8496,
      "step": 31890
    },
    {
      "epoch": 0.5600519672044804,
      "grad_norm": 0.08149158635271352,
      "learning_rate": 0.00019572496432409926,
      "loss": 0.8615,
      "step": 31900
    },
    {
      "epoch": 0.5602275320844818,
      "grad_norm": 0.07666930655525285,
      "learning_rate": 0.00019572172440098382,
      "loss": 0.8579,
      "step": 31910
    },
    {
      "epoch": 0.5604030969644832,
      "grad_norm": 0.10159118756022271,
      "learning_rate": 0.00019571848327772591,
      "loss": 0.858,
      "step": 31920
    },
    {
      "epoch": 0.5605786618444847,
      "grad_norm": 0.08841384980410794,
      "learning_rate": 0.00019571524095436674,
      "loss": 0.8604,
      "step": 31930
    },
    {
      "epoch": 0.560754226724486,
      "grad_norm": 0.06606160188179354,
      "learning_rate": 0.00019571199743094727,
      "loss": 0.8524,
      "step": 31940
    },
    {
      "epoch": 0.5609297916044874,
      "grad_norm": 0.09045001375640095,
      "learning_rate": 0.00019570875270750863,
      "loss": 0.8483,
      "step": 31950
    },
    {
      "epoch": 0.5611053564844889,
      "grad_norm": 0.06985870669150247,
      "learning_rate": 0.00019570550678409197,
      "loss": 0.8552,
      "step": 31960
    },
    {
      "epoch": 0.5612809213644903,
      "grad_norm": 0.05399414284616059,
      "learning_rate": 0.00019570225966073835,
      "loss": 0.8584,
      "step": 31970
    },
    {
      "epoch": 0.5614564862444916,
      "grad_norm": 0.05778512582234772,
      "learning_rate": 0.000195699011337489,
      "loss": 0.8516,
      "step": 31980
    },
    {
      "epoch": 0.5616320511244931,
      "grad_norm": 0.06241525841741541,
      "learning_rate": 0.000195695761814385,
      "loss": 0.855,
      "step": 31990
    },
    {
      "epoch": 0.5618076160044945,
      "grad_norm": 0.10594391546805544,
      "learning_rate": 0.00019569251109146756,
      "loss": 0.857,
      "step": 32000
    },
    {
      "epoch": 0.5619831808844958,
      "grad_norm": 0.09370594884544348,
      "learning_rate": 0.00019568925916877786,
      "loss": 0.8493,
      "step": 32010
    },
    {
      "epoch": 0.5621587457644973,
      "grad_norm": 0.07148149568791162,
      "learning_rate": 0.00019568600604635708,
      "loss": 0.8589,
      "step": 32020
    },
    {
      "epoch": 0.5623343106444987,
      "grad_norm": 0.05994117207920045,
      "learning_rate": 0.00019568275172424646,
      "loss": 0.8563,
      "step": 32030
    },
    {
      "epoch": 0.5625098755245,
      "grad_norm": 0.06913380368814628,
      "learning_rate": 0.00019567949620248726,
      "loss": 0.8669,
      "step": 32040
    },
    {
      "epoch": 0.5626854404045015,
      "grad_norm": 0.06356107213516923,
      "learning_rate": 0.00019567623948112066,
      "loss": 0.8581,
      "step": 32050
    },
    {
      "epoch": 0.5628610052845029,
      "grad_norm": 0.06112409139027663,
      "learning_rate": 0.000195672981560188,
      "loss": 0.8532,
      "step": 32060
    },
    {
      "epoch": 0.5630365701645043,
      "grad_norm": 0.05747512904435455,
      "learning_rate": 0.00019566972243973047,
      "loss": 0.8546,
      "step": 32070
    },
    {
      "epoch": 0.5632121350445057,
      "grad_norm": 0.1052427911599045,
      "learning_rate": 0.00019566646211978945,
      "loss": 0.8503,
      "step": 32080
    },
    {
      "epoch": 0.5633876999245071,
      "grad_norm": 0.1009751274224381,
      "learning_rate": 0.0001956632006004062,
      "loss": 0.8578,
      "step": 32090
    },
    {
      "epoch": 0.5635632648045085,
      "grad_norm": 0.054740770294274876,
      "learning_rate": 0.00019565993788162207,
      "loss": 0.8561,
      "step": 32100
    },
    {
      "epoch": 0.56373882968451,
      "grad_norm": 0.06933507814274649,
      "learning_rate": 0.00019565667396347837,
      "loss": 0.8556,
      "step": 32110
    },
    {
      "epoch": 0.5639143945645113,
      "grad_norm": 0.06608990849334938,
      "learning_rate": 0.00019565340884601645,
      "loss": 0.8588,
      "step": 32120
    },
    {
      "epoch": 0.5640899594445127,
      "grad_norm": 0.0725117537252476,
      "learning_rate": 0.00019565014252927775,
      "loss": 0.8547,
      "step": 32130
    },
    {
      "epoch": 0.5642655243245142,
      "grad_norm": 0.07125325370309396,
      "learning_rate": 0.00019564687501330357,
      "loss": 0.8505,
      "step": 32140
    },
    {
      "epoch": 0.5644410892045155,
      "grad_norm": 0.09132015742005194,
      "learning_rate": 0.00019564360629813535,
      "loss": 0.8448,
      "step": 32150
    },
    {
      "epoch": 0.5646166540845169,
      "grad_norm": 0.06305975153096847,
      "learning_rate": 0.0001956403363838145,
      "loss": 0.8565,
      "step": 32160
    },
    {
      "epoch": 0.5647922189645184,
      "grad_norm": 0.06922877442031199,
      "learning_rate": 0.00019563706527038247,
      "loss": 0.8493,
      "step": 32170
    },
    {
      "epoch": 0.5649677838445197,
      "grad_norm": 0.05361829627271556,
      "learning_rate": 0.00019563379295788063,
      "loss": 0.851,
      "step": 32180
    },
    {
      "epoch": 0.5651433487245211,
      "grad_norm": 0.07667918495855176,
      "learning_rate": 0.00019563051944635057,
      "loss": 0.8575,
      "step": 32190
    },
    {
      "epoch": 0.5653189136045226,
      "grad_norm": 0.05309647160987069,
      "learning_rate": 0.00019562724473583365,
      "loss": 0.8518,
      "step": 32200
    },
    {
      "epoch": 0.5654944784845239,
      "grad_norm": 0.05995795900810519,
      "learning_rate": 0.00019562396882637142,
      "loss": 0.8519,
      "step": 32210
    },
    {
      "epoch": 0.5656700433645253,
      "grad_norm": 0.06778232524308747,
      "learning_rate": 0.00019562069171800536,
      "loss": 0.8459,
      "step": 32220
    },
    {
      "epoch": 0.5658456082445268,
      "grad_norm": 0.10834944396195287,
      "learning_rate": 0.000195617413410777,
      "loss": 0.8448,
      "step": 32230
    },
    {
      "epoch": 0.5660211731245282,
      "grad_norm": 0.06556827610495528,
      "learning_rate": 0.0001956141339047279,
      "loss": 0.8515,
      "step": 32240
    },
    {
      "epoch": 0.5661967380045295,
      "grad_norm": 0.09547206963110622,
      "learning_rate": 0.00019561085319989954,
      "loss": 0.8509,
      "step": 32250
    },
    {
      "epoch": 0.566372302884531,
      "grad_norm": 0.056858823761605094,
      "learning_rate": 0.00019560757129633358,
      "loss": 0.8582,
      "step": 32260
    },
    {
      "epoch": 0.5665478677645324,
      "grad_norm": 0.08110861930422189,
      "learning_rate": 0.00019560428819407157,
      "loss": 0.8537,
      "step": 32270
    },
    {
      "epoch": 0.5667234326445338,
      "grad_norm": 0.07549000548632909,
      "learning_rate": 0.00019560100389315506,
      "loss": 0.864,
      "step": 32280
    },
    {
      "epoch": 0.5668989975245352,
      "grad_norm": 0.058631737890289484,
      "learning_rate": 0.00019559771839362575,
      "loss": 0.8567,
      "step": 32290
    },
    {
      "epoch": 0.5670745624045366,
      "grad_norm": 0.071333860048227,
      "learning_rate": 0.00019559443169552516,
      "loss": 0.8561,
      "step": 32300
    },
    {
      "epoch": 0.567250127284538,
      "grad_norm": 0.06298551312018201,
      "learning_rate": 0.00019559114379889502,
      "loss": 0.8554,
      "step": 32310
    },
    {
      "epoch": 0.5674256921645394,
      "grad_norm": 0.050211034339646285,
      "learning_rate": 0.00019558785470377696,
      "loss": 0.8517,
      "step": 32320
    },
    {
      "epoch": 0.5676012570445408,
      "grad_norm": 0.10202404898623746,
      "learning_rate": 0.00019558456441021266,
      "loss": 0.8536,
      "step": 32330
    },
    {
      "epoch": 0.5677768219245423,
      "grad_norm": 0.05383635805431459,
      "learning_rate": 0.0001955812729182438,
      "loss": 0.8639,
      "step": 32340
    },
    {
      "epoch": 0.5679523868045436,
      "grad_norm": 0.09278525788411675,
      "learning_rate": 0.0001955779802279121,
      "loss": 0.8601,
      "step": 32350
    },
    {
      "epoch": 0.568127951684545,
      "grad_norm": 0.0697810754164993,
      "learning_rate": 0.00019557468633925922,
      "loss": 0.8554,
      "step": 32360
    },
    {
      "epoch": 0.5683035165645465,
      "grad_norm": 0.062017064714340125,
      "learning_rate": 0.00019557139125232697,
      "loss": 0.851,
      "step": 32370
    },
    {
      "epoch": 0.5684790814445478,
      "grad_norm": 0.05202628934185076,
      "learning_rate": 0.00019556809496715707,
      "loss": 0.8536,
      "step": 32380
    },
    {
      "epoch": 0.5686546463245492,
      "grad_norm": 0.07075653120874596,
      "learning_rate": 0.00019556479748379126,
      "loss": 0.8619,
      "step": 32390
    },
    {
      "epoch": 0.5688302112045507,
      "grad_norm": 0.06935967163033377,
      "learning_rate": 0.00019556149880227134,
      "loss": 0.8487,
      "step": 32400
    },
    {
      "epoch": 0.569005776084552,
      "grad_norm": 0.06541204725450121,
      "learning_rate": 0.00019555819892263916,
      "loss": 0.8576,
      "step": 32410
    },
    {
      "epoch": 0.5691813409645534,
      "grad_norm": 0.07174605139084908,
      "learning_rate": 0.00019555489784493644,
      "loss": 0.853,
      "step": 32420
    },
    {
      "epoch": 0.5693569058445549,
      "grad_norm": 0.06249874602560976,
      "learning_rate": 0.00019555159556920506,
      "loss": 0.8591,
      "step": 32430
    },
    {
      "epoch": 0.5695324707245563,
      "grad_norm": 0.06694068397331837,
      "learning_rate": 0.00019554829209548683,
      "loss": 0.8498,
      "step": 32440
    },
    {
      "epoch": 0.5697080356045576,
      "grad_norm": 0.06226893561156828,
      "learning_rate": 0.0001955449874238236,
      "loss": 0.8587,
      "step": 32450
    },
    {
      "epoch": 0.5698836004845591,
      "grad_norm": 0.08524853511409376,
      "learning_rate": 0.00019554168155425726,
      "loss": 0.8578,
      "step": 32460
    },
    {
      "epoch": 0.5700591653645605,
      "grad_norm": 0.05711813520034612,
      "learning_rate": 0.0001955383744868297,
      "loss": 0.8552,
      "step": 32470
    },
    {
      "epoch": 0.5702347302445618,
      "grad_norm": 0.06821456133069702,
      "learning_rate": 0.00019553506622158287,
      "loss": 0.8546,
      "step": 32480
    },
    {
      "epoch": 0.5704102951245633,
      "grad_norm": 0.0704241835836233,
      "learning_rate": 0.0001955317567585586,
      "loss": 0.8551,
      "step": 32490
    },
    {
      "epoch": 0.5705858600045647,
      "grad_norm": 0.050187446766258664,
      "learning_rate": 0.00019552844609779883,
      "loss": 0.8514,
      "step": 32500
    },
    {
      "epoch": 0.570761424884566,
      "grad_norm": 0.06157481285610321,
      "learning_rate": 0.00019552513423934557,
      "loss": 0.8566,
      "step": 32510
    },
    {
      "epoch": 0.5709369897645675,
      "grad_norm": 0.05869478772363935,
      "learning_rate": 0.00019552182118324075,
      "loss": 0.8522,
      "step": 32520
    },
    {
      "epoch": 0.5711125546445689,
      "grad_norm": 0.09613495052058811,
      "learning_rate": 0.00019551850692952626,
      "loss": 0.8543,
      "step": 32530
    },
    {
      "epoch": 0.5712881195245703,
      "grad_norm": 0.06834327998859761,
      "learning_rate": 0.00019551519147824426,
      "loss": 0.8601,
      "step": 32540
    },
    {
      "epoch": 0.5714636844045717,
      "grad_norm": 0.06322959783553499,
      "learning_rate": 0.0001955118748294366,
      "loss": 0.8614,
      "step": 32550
    },
    {
      "epoch": 0.5716392492845731,
      "grad_norm": 0.08449027209011815,
      "learning_rate": 0.00019550855698314542,
      "loss": 0.8527,
      "step": 32560
    },
    {
      "epoch": 0.5718148141645745,
      "grad_norm": 0.06708895108790885,
      "learning_rate": 0.00019550523793941273,
      "loss": 0.8573,
      "step": 32570
    },
    {
      "epoch": 0.571990379044576,
      "grad_norm": 0.056215528053975644,
      "learning_rate": 0.00019550191769828054,
      "loss": 0.8484,
      "step": 32580
    },
    {
      "epoch": 0.5721659439245773,
      "grad_norm": 0.06035692082036451,
      "learning_rate": 0.00019549859625979092,
      "loss": 0.8513,
      "step": 32590
    },
    {
      "epoch": 0.5723415088045787,
      "grad_norm": 0.06773662296722908,
      "learning_rate": 0.000195495273623986,
      "loss": 0.8559,
      "step": 32600
    },
    {
      "epoch": 0.5725170736845802,
      "grad_norm": 0.057970265147435716,
      "learning_rate": 0.00019549194979090784,
      "loss": 0.8572,
      "step": 32610
    },
    {
      "epoch": 0.5726926385645815,
      "grad_norm": 0.06149947485892703,
      "learning_rate": 0.0001954886247605986,
      "loss": 0.8629,
      "step": 32620
    },
    {
      "epoch": 0.5728682034445829,
      "grad_norm": 0.06418158817260643,
      "learning_rate": 0.0001954852985331003,
      "loss": 0.8541,
      "step": 32630
    },
    {
      "epoch": 0.5730437683245844,
      "grad_norm": 0.06820565420635122,
      "learning_rate": 0.00019548197110845524,
      "loss": 0.8548,
      "step": 32640
    },
    {
      "epoch": 0.5732193332045857,
      "grad_norm": 0.0850417782025672,
      "learning_rate": 0.00019547864248670545,
      "loss": 0.8585,
      "step": 32650
    },
    {
      "epoch": 0.5733948980845872,
      "grad_norm": 0.05512428362927731,
      "learning_rate": 0.00019547531266789317,
      "loss": 0.8652,
      "step": 32660
    },
    {
      "epoch": 0.5735704629645886,
      "grad_norm": 0.05663140257434919,
      "learning_rate": 0.00019547198165206057,
      "loss": 0.8535,
      "step": 32670
    },
    {
      "epoch": 0.57374602784459,
      "grad_norm": 0.045479998677152536,
      "learning_rate": 0.00019546864943924987,
      "loss": 0.8523,
      "step": 32680
    },
    {
      "epoch": 0.5739215927245914,
      "grad_norm": 0.056360798377025616,
      "learning_rate": 0.00019546531602950324,
      "loss": 0.8536,
      "step": 32690
    },
    {
      "epoch": 0.5740971576045928,
      "grad_norm": 0.06728035242933054,
      "learning_rate": 0.000195461981422863,
      "loss": 0.8612,
      "step": 32700
    },
    {
      "epoch": 0.5742727224845942,
      "grad_norm": 0.05527212856242969,
      "learning_rate": 0.0001954586456193713,
      "loss": 0.8554,
      "step": 32710
    },
    {
      "epoch": 0.5744482873645956,
      "grad_norm": 0.063170391597777,
      "learning_rate": 0.00019545530861907052,
      "loss": 0.8544,
      "step": 32720
    },
    {
      "epoch": 0.574623852244597,
      "grad_norm": 0.06047200528152274,
      "learning_rate": 0.00019545197042200284,
      "loss": 0.8587,
      "step": 32730
    },
    {
      "epoch": 0.5747994171245984,
      "grad_norm": 0.07210847914944883,
      "learning_rate": 0.00019544863102821062,
      "loss": 0.8529,
      "step": 32740
    },
    {
      "epoch": 0.5749749820045998,
      "grad_norm": 0.08803498701511522,
      "learning_rate": 0.0001954452904377361,
      "loss": 0.8577,
      "step": 32750
    },
    {
      "epoch": 0.5751505468846012,
      "grad_norm": 0.06343911457050976,
      "learning_rate": 0.00019544194865062165,
      "loss": 0.8572,
      "step": 32760
    },
    {
      "epoch": 0.5753261117646026,
      "grad_norm": 0.06144494067582817,
      "learning_rate": 0.00019543860566690965,
      "loss": 0.8538,
      "step": 32770
    },
    {
      "epoch": 0.575501676644604,
      "grad_norm": 0.05758203604263062,
      "learning_rate": 0.00019543526148664239,
      "loss": 0.8615,
      "step": 32780
    },
    {
      "epoch": 0.5756772415246054,
      "grad_norm": 0.07149944263812592,
      "learning_rate": 0.00019543191610986227,
      "loss": 0.8573,
      "step": 32790
    },
    {
      "epoch": 0.5758528064046068,
      "grad_norm": 0.07822340244045609,
      "learning_rate": 0.00019542856953661168,
      "loss": 0.8517,
      "step": 32800
    },
    {
      "epoch": 0.5760283712846083,
      "grad_norm": 0.06618174241528814,
      "learning_rate": 0.00019542522176693304,
      "loss": 0.857,
      "step": 32810
    },
    {
      "epoch": 0.5762039361646096,
      "grad_norm": 0.06891477095979336,
      "learning_rate": 0.0001954218728008687,
      "loss": 0.8531,
      "step": 32820
    },
    {
      "epoch": 0.576379501044611,
      "grad_norm": 0.07120260716374313,
      "learning_rate": 0.00019541852263846115,
      "loss": 0.8529,
      "step": 32830
    },
    {
      "epoch": 0.5765550659246125,
      "grad_norm": 0.0725433459095679,
      "learning_rate": 0.00019541517127975286,
      "loss": 0.8537,
      "step": 32840
    },
    {
      "epoch": 0.5767306308046138,
      "grad_norm": 0.06777706835747493,
      "learning_rate": 0.0001954118187247862,
      "loss": 0.8623,
      "step": 32850
    },
    {
      "epoch": 0.5769061956846152,
      "grad_norm": 0.05555952558711061,
      "learning_rate": 0.00019540846497360375,
      "loss": 0.8467,
      "step": 32860
    },
    {
      "epoch": 0.5770817605646167,
      "grad_norm": 0.07965141384146267,
      "learning_rate": 0.0001954051100262479,
      "loss": 0.8507,
      "step": 32870
    },
    {
      "epoch": 0.577257325444618,
      "grad_norm": 0.05288075455877257,
      "learning_rate": 0.00019540175388276128,
      "loss": 0.8501,
      "step": 32880
    },
    {
      "epoch": 0.5774328903246194,
      "grad_norm": 0.058908304935020514,
      "learning_rate": 0.0001953983965431863,
      "loss": 0.842,
      "step": 32890
    },
    {
      "epoch": 0.5776084552046209,
      "grad_norm": 0.05158073567780164,
      "learning_rate": 0.00019539503800756558,
      "loss": 0.8532,
      "step": 32900
    },
    {
      "epoch": 0.5777840200846223,
      "grad_norm": 0.06619999891439178,
      "learning_rate": 0.0001953916782759416,
      "loss": 0.861,
      "step": 32910
    },
    {
      "epoch": 0.5779595849646236,
      "grad_norm": 0.0658944623556457,
      "learning_rate": 0.000195388317348357,
      "loss": 0.8575,
      "step": 32920
    },
    {
      "epoch": 0.5781351498446251,
      "grad_norm": 0.06149888042412686,
      "learning_rate": 0.0001953849552248543,
      "loss": 0.851,
      "step": 32930
    },
    {
      "epoch": 0.5783107147246265,
      "grad_norm": 0.049793700414776314,
      "learning_rate": 0.00019538159190547614,
      "loss": 0.856,
      "step": 32940
    },
    {
      "epoch": 0.5784862796046278,
      "grad_norm": 0.07549503528031541,
      "learning_rate": 0.00019537822739026516,
      "loss": 0.8569,
      "step": 32950
    },
    {
      "epoch": 0.5786618444846293,
      "grad_norm": 0.05699737536255877,
      "learning_rate": 0.0001953748616792639,
      "loss": 0.8556,
      "step": 32960
    },
    {
      "epoch": 0.5788374093646307,
      "grad_norm": 0.0594727773713429,
      "learning_rate": 0.00019537149477251504,
      "loss": 0.8487,
      "step": 32970
    },
    {
      "epoch": 0.579012974244632,
      "grad_norm": 0.07689454575683302,
      "learning_rate": 0.00019536812667006128,
      "loss": 0.8558,
      "step": 32980
    },
    {
      "epoch": 0.5791885391246335,
      "grad_norm": 0.05671456229033563,
      "learning_rate": 0.00019536475737194528,
      "loss": 0.858,
      "step": 32990
    },
    {
      "epoch": 0.5793641040046349,
      "grad_norm": 0.05469247374942836,
      "learning_rate": 0.00019536138687820974,
      "loss": 0.8517,
      "step": 33000
    },
    {
      "epoch": 0.5795396688846364,
      "grad_norm": 0.07712834583253583,
      "learning_rate": 0.00019535801518889728,
      "loss": 0.8522,
      "step": 33010
    },
    {
      "epoch": 0.5797152337646377,
      "grad_norm": 0.06989569983795088,
      "learning_rate": 0.00019535464230405073,
      "loss": 0.8561,
      "step": 33020
    },
    {
      "epoch": 0.5798907986446391,
      "grad_norm": 0.07395283590236464,
      "learning_rate": 0.00019535126822371277,
      "loss": 0.8567,
      "step": 33030
    },
    {
      "epoch": 0.5800663635246406,
      "grad_norm": 0.06999036792271435,
      "learning_rate": 0.00019534789294792617,
      "loss": 0.8573,
      "step": 33040
    },
    {
      "epoch": 0.580241928404642,
      "grad_norm": 0.09420814162812476,
      "learning_rate": 0.00019534451647673366,
      "loss": 0.8517,
      "step": 33050
    },
    {
      "epoch": 0.5804174932846433,
      "grad_norm": 0.07120906120907611,
      "learning_rate": 0.00019534113881017809,
      "loss": 0.8592,
      "step": 33060
    },
    {
      "epoch": 0.5805930581646448,
      "grad_norm": 0.06391727084741361,
      "learning_rate": 0.00019533775994830217,
      "loss": 0.8515,
      "step": 33070
    },
    {
      "epoch": 0.5807686230446462,
      "grad_norm": 0.07043152272685764,
      "learning_rate": 0.00019533437989114875,
      "loss": 0.8566,
      "step": 33080
    },
    {
      "epoch": 0.5809441879246475,
      "grad_norm": 0.0614115125025493,
      "learning_rate": 0.0001953309986387607,
      "loss": 0.8485,
      "step": 33090
    },
    {
      "epoch": 0.581119752804649,
      "grad_norm": 0.06917892675822503,
      "learning_rate": 0.00019532761619118076,
      "loss": 0.85,
      "step": 33100
    },
    {
      "epoch": 0.5812953176846504,
      "grad_norm": 0.054761103618492304,
      "learning_rate": 0.00019532423254845188,
      "loss": 0.8577,
      "step": 33110
    },
    {
      "epoch": 0.5814708825646517,
      "grad_norm": 0.0681351329419022,
      "learning_rate": 0.0001953208477106169,
      "loss": 0.8587,
      "step": 33120
    },
    {
      "epoch": 0.5816464474446532,
      "grad_norm": 0.0706103397974869,
      "learning_rate": 0.00019531746167771868,
      "loss": 0.8648,
      "step": 33130
    },
    {
      "epoch": 0.5818220123246546,
      "grad_norm": 0.06694398238092401,
      "learning_rate": 0.00019531407444980017,
      "loss": 0.8509,
      "step": 33140
    },
    {
      "epoch": 0.581997577204656,
      "grad_norm": 0.05520543034692205,
      "learning_rate": 0.00019531068602690426,
      "loss": 0.8524,
      "step": 33150
    },
    {
      "epoch": 0.5821731420846574,
      "grad_norm": 0.0887314141279773,
      "learning_rate": 0.00019530729640907386,
      "loss": 0.8511,
      "step": 33160
    },
    {
      "epoch": 0.5823487069646588,
      "grad_norm": 0.06860135272727876,
      "learning_rate": 0.00019530390559635195,
      "loss": 0.8531,
      "step": 33170
    },
    {
      "epoch": 0.5825242718446602,
      "grad_norm": 0.06795289711491338,
      "learning_rate": 0.00019530051358878148,
      "loss": 0.8531,
      "step": 33180
    },
    {
      "epoch": 0.5826998367246616,
      "grad_norm": 0.07784408291646687,
      "learning_rate": 0.00019529712038640545,
      "loss": 0.8537,
      "step": 33190
    },
    {
      "epoch": 0.582875401604663,
      "grad_norm": 0.07968604495453765,
      "learning_rate": 0.0001952937259892668,
      "loss": 0.8538,
      "step": 33200
    },
    {
      "epoch": 0.5830509664846644,
      "grad_norm": 0.05766934665639535,
      "learning_rate": 0.0001952903303974086,
      "loss": 0.8729,
      "step": 33210
    },
    {
      "epoch": 0.5832265313646658,
      "grad_norm": 0.06313669813406546,
      "learning_rate": 0.00019528693361087383,
      "loss": 0.8596,
      "step": 33220
    },
    {
      "epoch": 0.5834020962446672,
      "grad_norm": 0.09774903074276924,
      "learning_rate": 0.00019528353562970555,
      "loss": 0.8574,
      "step": 33230
    },
    {
      "epoch": 0.5835776611246686,
      "grad_norm": 0.07867385189941656,
      "learning_rate": 0.0001952801364539468,
      "loss": 0.859,
      "step": 33240
    },
    {
      "epoch": 0.5837532260046701,
      "grad_norm": 0.06524542970636586,
      "learning_rate": 0.00019527673608364063,
      "loss": 0.8603,
      "step": 33250
    },
    {
      "epoch": 0.5839287908846714,
      "grad_norm": 0.05392323491940685,
      "learning_rate": 0.00019527333451883018,
      "loss": 0.8545,
      "step": 33260
    },
    {
      "epoch": 0.5841043557646728,
      "grad_norm": 0.061512778826081665,
      "learning_rate": 0.0001952699317595585,
      "loss": 0.8564,
      "step": 33270
    },
    {
      "epoch": 0.5842799206446743,
      "grad_norm": 0.07206003041884652,
      "learning_rate": 0.0001952665278058687,
      "loss": 0.8509,
      "step": 33280
    },
    {
      "epoch": 0.5844554855246756,
      "grad_norm": 0.06589693294560058,
      "learning_rate": 0.00019526312265780396,
      "loss": 0.8469,
      "step": 33290
    },
    {
      "epoch": 0.584631050404677,
      "grad_norm": 0.09184637925104366,
      "learning_rate": 0.0001952597163154074,
      "loss": 0.8552,
      "step": 33300
    },
    {
      "epoch": 0.5848066152846785,
      "grad_norm": 0.06652555081692095,
      "learning_rate": 0.00019525630877872214,
      "loss": 0.8542,
      "step": 33310
    },
    {
      "epoch": 0.5849821801646798,
      "grad_norm": 0.08885322685765253,
      "learning_rate": 0.00019525290004779138,
      "loss": 0.8506,
      "step": 33320
    },
    {
      "epoch": 0.5851577450446812,
      "grad_norm": 0.07731486012239168,
      "learning_rate": 0.00019524949012265832,
      "loss": 0.855,
      "step": 33330
    },
    {
      "epoch": 0.5853333099246827,
      "grad_norm": 0.05903663132414971,
      "learning_rate": 0.00019524607900336615,
      "loss": 0.8525,
      "step": 33340
    },
    {
      "epoch": 0.585508874804684,
      "grad_norm": 0.08349084797630148,
      "learning_rate": 0.0001952426666899581,
      "loss": 0.8507,
      "step": 33350
    },
    {
      "epoch": 0.5856844396846854,
      "grad_norm": 0.06649268638869657,
      "learning_rate": 0.00019523925318247742,
      "loss": 0.8565,
      "step": 33360
    },
    {
      "epoch": 0.5858600045646869,
      "grad_norm": 0.08653840606044026,
      "learning_rate": 0.0001952358384809673,
      "loss": 0.8552,
      "step": 33370
    },
    {
      "epoch": 0.5860355694446883,
      "grad_norm": 0.05117611212437168,
      "learning_rate": 0.0001952324225854711,
      "loss": 0.857,
      "step": 33380
    },
    {
      "epoch": 0.5862111343246897,
      "grad_norm": 0.05208601775011767,
      "learning_rate": 0.000195229005496032,
      "loss": 0.8547,
      "step": 33390
    },
    {
      "epoch": 0.5863866992046911,
      "grad_norm": 0.0699694107702731,
      "learning_rate": 0.00019522558721269342,
      "loss": 0.8538,
      "step": 33400
    },
    {
      "epoch": 0.5865622640846925,
      "grad_norm": 0.05411439753909007,
      "learning_rate": 0.0001952221677354985,
      "loss": 0.8567,
      "step": 33410
    },
    {
      "epoch": 0.586737828964694,
      "grad_norm": 0.07012237501175037,
      "learning_rate": 0.00019521874706449071,
      "loss": 0.8561,
      "step": 33420
    },
    {
      "epoch": 0.5869133938446953,
      "grad_norm": 0.08970227560560895,
      "learning_rate": 0.00019521532519971332,
      "loss": 0.8596,
      "step": 33430
    },
    {
      "epoch": 0.5870889587246967,
      "grad_norm": 0.07390837647039934,
      "learning_rate": 0.0001952119021412097,
      "loss": 0.8556,
      "step": 33440
    },
    {
      "epoch": 0.5872645236046982,
      "grad_norm": 0.06846710226296572,
      "learning_rate": 0.00019520847788902324,
      "loss": 0.8541,
      "step": 33450
    },
    {
      "epoch": 0.5874400884846995,
      "grad_norm": 0.08030473220427722,
      "learning_rate": 0.00019520505244319727,
      "loss": 0.8516,
      "step": 33460
    },
    {
      "epoch": 0.5876156533647009,
      "grad_norm": 0.06515571055005846,
      "learning_rate": 0.00019520162580377527,
      "loss": 0.854,
      "step": 33470
    },
    {
      "epoch": 0.5877912182447024,
      "grad_norm": 0.06341075296202849,
      "learning_rate": 0.00019519819797080058,
      "loss": 0.8563,
      "step": 33480
    },
    {
      "epoch": 0.5879667831247037,
      "grad_norm": 0.056854140427527455,
      "learning_rate": 0.0001951947689443167,
      "loss": 0.8563,
      "step": 33490
    },
    {
      "epoch": 0.5881423480047051,
      "grad_norm": 0.06403078479278276,
      "learning_rate": 0.00019519133872436705,
      "loss": 0.8647,
      "step": 33500
    },
    {
      "epoch": 0.5883179128847066,
      "grad_norm": 0.10009494403915625,
      "learning_rate": 0.0001951879073109951,
      "loss": 0.8536,
      "step": 33510
    },
    {
      "epoch": 0.588493477764708,
      "grad_norm": 0.0791051588423064,
      "learning_rate": 0.00019518447470424428,
      "loss": 0.8508,
      "step": 33520
    },
    {
      "epoch": 0.5886690426447093,
      "grad_norm": 0.07866982254835435,
      "learning_rate": 0.00019518104090415813,
      "loss": 0.8436,
      "step": 33530
    },
    {
      "epoch": 0.5888446075247108,
      "grad_norm": 0.061089543174193836,
      "learning_rate": 0.00019517760591078015,
      "loss": 0.8593,
      "step": 33540
    },
    {
      "epoch": 0.5890201724047122,
      "grad_norm": 0.13594521794257305,
      "learning_rate": 0.0001951741697241538,
      "loss": 0.865,
      "step": 33550
    },
    {
      "epoch": 0.5891957372847135,
      "grad_norm": 0.07158132895896371,
      "learning_rate": 0.00019517073234432273,
      "loss": 0.8536,
      "step": 33560
    },
    {
      "epoch": 0.589371302164715,
      "grad_norm": 0.12006549550629925,
      "learning_rate": 0.00019516729377133042,
      "loss": 0.8525,
      "step": 33570
    },
    {
      "epoch": 0.5895468670447164,
      "grad_norm": 0.06425124151914942,
      "learning_rate": 0.00019516385400522042,
      "loss": 0.8572,
      "step": 33580
    },
    {
      "epoch": 0.5897224319247177,
      "grad_norm": 0.06686111936501374,
      "learning_rate": 0.0001951604130460364,
      "loss": 0.8518,
      "step": 33590
    },
    {
      "epoch": 0.5898979968047192,
      "grad_norm": 0.060975112248772605,
      "learning_rate": 0.00019515697089382187,
      "loss": 0.8477,
      "step": 33600
    },
    {
      "epoch": 0.5900735616847206,
      "grad_norm": 0.05049322131511545,
      "learning_rate": 0.00019515352754862045,
      "loss": 0.8588,
      "step": 33610
    },
    {
      "epoch": 0.590249126564722,
      "grad_norm": 0.10198541671936864,
      "learning_rate": 0.00019515008301047583,
      "loss": 0.8526,
      "step": 33620
    },
    {
      "epoch": 0.5904246914447234,
      "grad_norm": 0.0712410478804826,
      "learning_rate": 0.00019514663727943158,
      "loss": 0.8554,
      "step": 33630
    },
    {
      "epoch": 0.5906002563247248,
      "grad_norm": 0.06458814114066187,
      "learning_rate": 0.00019514319035553144,
      "loss": 0.8482,
      "step": 33640
    },
    {
      "epoch": 0.5907758212047262,
      "grad_norm": 0.053621302605691055,
      "learning_rate": 0.00019513974223881898,
      "loss": 0.8471,
      "step": 33650
    },
    {
      "epoch": 0.5909513860847276,
      "grad_norm": 0.054680010622272554,
      "learning_rate": 0.00019513629292933798,
      "loss": 0.8526,
      "step": 33660
    },
    {
      "epoch": 0.591126950964729,
      "grad_norm": 0.06723672244578614,
      "learning_rate": 0.0001951328424271321,
      "loss": 0.8533,
      "step": 33670
    },
    {
      "epoch": 0.5913025158447304,
      "grad_norm": 0.06361498561948252,
      "learning_rate": 0.00019512939073224508,
      "loss": 0.859,
      "step": 33680
    },
    {
      "epoch": 0.5914780807247318,
      "grad_norm": 0.06683229989138366,
      "learning_rate": 0.00019512593784472065,
      "loss": 0.8536,
      "step": 33690
    },
    {
      "epoch": 0.5916536456047332,
      "grad_norm": 0.06050614450745206,
      "learning_rate": 0.0001951224837646025,
      "loss": 0.8588,
      "step": 33700
    },
    {
      "epoch": 0.5918292104847346,
      "grad_norm": 0.05907782241003819,
      "learning_rate": 0.00019511902849193448,
      "loss": 0.8536,
      "step": 33710
    },
    {
      "epoch": 0.5920047753647361,
      "grad_norm": 0.09188242198720971,
      "learning_rate": 0.00019511557202676031,
      "loss": 0.8516,
      "step": 33720
    },
    {
      "epoch": 0.5921803402447374,
      "grad_norm": 0.06739650667573739,
      "learning_rate": 0.0001951121143691238,
      "loss": 0.8562,
      "step": 33730
    },
    {
      "epoch": 0.5923559051247389,
      "grad_norm": 0.07094022648851389,
      "learning_rate": 0.0001951086555190688,
      "loss": 0.8586,
      "step": 33740
    },
    {
      "epoch": 0.5925314700047403,
      "grad_norm": 0.0679746640367729,
      "learning_rate": 0.00019510519547663908,
      "loss": 0.8556,
      "step": 33750
    },
    {
      "epoch": 0.5927070348847416,
      "grad_norm": 0.10123201053272536,
      "learning_rate": 0.0001951017342418785,
      "loss": 0.8537,
      "step": 33760
    },
    {
      "epoch": 0.5928825997647431,
      "grad_norm": 0.055839758649634655,
      "learning_rate": 0.0001950982718148309,
      "loss": 0.8571,
      "step": 33770
    },
    {
      "epoch": 0.5930581646447445,
      "grad_norm": 0.05432636230090095,
      "learning_rate": 0.00019509480819554013,
      "loss": 0.8586,
      "step": 33780
    },
    {
      "epoch": 0.5932337295247458,
      "grad_norm": 0.07366911562454,
      "learning_rate": 0.00019509134338405016,
      "loss": 0.8603,
      "step": 33790
    },
    {
      "epoch": 0.5934092944047473,
      "grad_norm": 0.06329128840980038,
      "learning_rate": 0.00019508787738040483,
      "loss": 0.8532,
      "step": 33800
    },
    {
      "epoch": 0.5935848592847487,
      "grad_norm": 0.06276336833351032,
      "learning_rate": 0.00019508441018464804,
      "loss": 0.8552,
      "step": 33810
    },
    {
      "epoch": 0.5937604241647501,
      "grad_norm": 0.08763719514226466,
      "learning_rate": 0.00019508094179682378,
      "loss": 0.8585,
      "step": 33820
    },
    {
      "epoch": 0.5939359890447515,
      "grad_norm": 0.08174460464418325,
      "learning_rate": 0.00019507747221697593,
      "loss": 0.8509,
      "step": 33830
    },
    {
      "epoch": 0.5941115539247529,
      "grad_norm": 0.06506027588637346,
      "learning_rate": 0.00019507400144514847,
      "loss": 0.8512,
      "step": 33840
    },
    {
      "epoch": 0.5942871188047543,
      "grad_norm": 0.05887661347839093,
      "learning_rate": 0.0001950705294813854,
      "loss": 0.8639,
      "step": 33850
    },
    {
      "epoch": 0.5944626836847557,
      "grad_norm": 0.05840276904719976,
      "learning_rate": 0.0001950670563257307,
      "loss": 0.8622,
      "step": 33860
    },
    {
      "epoch": 0.5946382485647571,
      "grad_norm": 0.08659359691907355,
      "learning_rate": 0.00019506358197822834,
      "loss": 0.8523,
      "step": 33870
    },
    {
      "epoch": 0.5948138134447585,
      "grad_norm": 0.053869451163014524,
      "learning_rate": 0.0001950601064389224,
      "loss": 0.8549,
      "step": 33880
    },
    {
      "epoch": 0.59498937832476,
      "grad_norm": 0.06038405341912577,
      "learning_rate": 0.00019505662970785686,
      "loss": 0.8532,
      "step": 33890
    },
    {
      "epoch": 0.5951649432047613,
      "grad_norm": 0.0721505272296486,
      "learning_rate": 0.00019505315178507576,
      "loss": 0.855,
      "step": 33900
    },
    {
      "epoch": 0.5953405080847627,
      "grad_norm": 0.053230407062687564,
      "learning_rate": 0.00019504967267062329,
      "loss": 0.8615,
      "step": 33910
    },
    {
      "epoch": 0.5955160729647642,
      "grad_norm": 0.058657199640471905,
      "learning_rate": 0.0001950461923645434,
      "loss": 0.8556,
      "step": 33920
    },
    {
      "epoch": 0.5956916378447655,
      "grad_norm": 0.07727005047861021,
      "learning_rate": 0.0001950427108668802,
      "loss": 0.8525,
      "step": 33930
    },
    {
      "epoch": 0.5958672027247669,
      "grad_norm": 0.07068961915817634,
      "learning_rate": 0.00019503922817767788,
      "loss": 0.8509,
      "step": 33940
    },
    {
      "epoch": 0.5960427676047684,
      "grad_norm": 0.05916249988616056,
      "learning_rate": 0.00019503574429698047,
      "loss": 0.8534,
      "step": 33950
    },
    {
      "epoch": 0.5962183324847697,
      "grad_norm": 0.06955834830224931,
      "learning_rate": 0.0001950322592248322,
      "loss": 0.845,
      "step": 33960
    },
    {
      "epoch": 0.5963938973647711,
      "grad_norm": 0.05876858293524209,
      "learning_rate": 0.00019502877296127716,
      "loss": 0.8579,
      "step": 33970
    },
    {
      "epoch": 0.5965694622447726,
      "grad_norm": 0.059142508073981366,
      "learning_rate": 0.00019502528550635956,
      "loss": 0.85,
      "step": 33980
    },
    {
      "epoch": 0.596745027124774,
      "grad_norm": 0.05969014428855633,
      "learning_rate": 0.00019502179686012357,
      "loss": 0.8676,
      "step": 33990
    },
    {
      "epoch": 0.5969205920047753,
      "grad_norm": 0.06916823034422771,
      "learning_rate": 0.0001950183070226134,
      "loss": 0.8557,
      "step": 34000
    },
    {
      "epoch": 0.5970961568847768,
      "grad_norm": 0.0681463783297539,
      "learning_rate": 0.00019501481599387325,
      "loss": 0.8543,
      "step": 34010
    },
    {
      "epoch": 0.5972717217647782,
      "grad_norm": 0.06169019852425541,
      "learning_rate": 0.00019501132377394736,
      "loss": 0.8568,
      "step": 34020
    },
    {
      "epoch": 0.5974472866447795,
      "grad_norm": 0.0698389529854056,
      "learning_rate": 0.00019500783036287996,
      "loss": 0.8583,
      "step": 34030
    },
    {
      "epoch": 0.597622851524781,
      "grad_norm": 0.07269948440082466,
      "learning_rate": 0.0001950043357607154,
      "loss": 0.8546,
      "step": 34040
    },
    {
      "epoch": 0.5977984164047824,
      "grad_norm": 0.07001628449510446,
      "learning_rate": 0.00019500083996749782,
      "loss": 0.8509,
      "step": 34050
    },
    {
      "epoch": 0.5979739812847837,
      "grad_norm": 0.061539304354296116,
      "learning_rate": 0.0001949973429832716,
      "loss": 0.8533,
      "step": 34060
    },
    {
      "epoch": 0.5981495461647852,
      "grad_norm": 0.07275834089086519,
      "learning_rate": 0.00019499384480808104,
      "loss": 0.8548,
      "step": 34070
    },
    {
      "epoch": 0.5983251110447866,
      "grad_norm": 0.06378460335538848,
      "learning_rate": 0.00019499034544197047,
      "loss": 0.8463,
      "step": 34080
    },
    {
      "epoch": 0.598500675924788,
      "grad_norm": 0.06908130484288214,
      "learning_rate": 0.00019498684488498418,
      "loss": 0.852,
      "step": 34090
    },
    {
      "epoch": 0.5986762408047894,
      "grad_norm": 0.06901830723010856,
      "learning_rate": 0.00019498334313716658,
      "loss": 0.8589,
      "step": 34100
    },
    {
      "epoch": 0.5988518056847908,
      "grad_norm": 0.08159285468925763,
      "learning_rate": 0.000194979840198562,
      "loss": 0.8457,
      "step": 34110
    },
    {
      "epoch": 0.5990273705647923,
      "grad_norm": 0.06328393113993441,
      "learning_rate": 0.00019497633606921483,
      "loss": 0.8521,
      "step": 34120
    },
    {
      "epoch": 0.5992029354447936,
      "grad_norm": 0.0617059580951153,
      "learning_rate": 0.00019497283074916945,
      "loss": 0.8536,
      "step": 34130
    },
    {
      "epoch": 0.599378500324795,
      "grad_norm": 0.06899821000403349,
      "learning_rate": 0.00019496932423847035,
      "loss": 0.8571,
      "step": 34140
    },
    {
      "epoch": 0.5995540652047965,
      "grad_norm": 0.06574458310330941,
      "learning_rate": 0.00019496581653716185,
      "loss": 0.8575,
      "step": 34150
    },
    {
      "epoch": 0.5997296300847978,
      "grad_norm": 0.05704013541866481,
      "learning_rate": 0.00019496230764528844,
      "loss": 0.8548,
      "step": 34160
    },
    {
      "epoch": 0.5999051949647992,
      "grad_norm": 0.06801725699137037,
      "learning_rate": 0.00019495879756289465,
      "loss": 0.8568,
      "step": 34170
    },
    {
      "epoch": 0.6000807598448007,
      "grad_norm": 0.07012249983770487,
      "learning_rate": 0.00019495528629002483,
      "loss": 0.8572,
      "step": 34180
    },
    {
      "epoch": 0.6002563247248021,
      "grad_norm": 0.06961452937468617,
      "learning_rate": 0.00019495177382672355,
      "loss": 0.8478,
      "step": 34190
    },
    {
      "epoch": 0.6004318896048034,
      "grad_norm": 0.05723780463908684,
      "learning_rate": 0.0001949482601730353,
      "loss": 0.8548,
      "step": 34200
    },
    {
      "epoch": 0.6006074544848049,
      "grad_norm": 0.0664632457918821,
      "learning_rate": 0.00019494474532900452,
      "loss": 0.8587,
      "step": 34210
    },
    {
      "epoch": 0.6007830193648063,
      "grad_norm": 0.08746435703327587,
      "learning_rate": 0.0001949412292946759,
      "loss": 0.8571,
      "step": 34220
    },
    {
      "epoch": 0.6009585842448076,
      "grad_norm": 0.07415728197339716,
      "learning_rate": 0.00019493771207009384,
      "loss": 0.8505,
      "step": 34230
    },
    {
      "epoch": 0.6011341491248091,
      "grad_norm": 0.0645957367885992,
      "learning_rate": 0.000194934193655303,
      "loss": 0.8542,
      "step": 34240
    },
    {
      "epoch": 0.6013097140048105,
      "grad_norm": 0.0753753670283995,
      "learning_rate": 0.0001949306740503479,
      "loss": 0.8427,
      "step": 34250
    },
    {
      "epoch": 0.6014852788848118,
      "grad_norm": 0.06909887879824832,
      "learning_rate": 0.00019492715325527318,
      "loss": 0.8472,
      "step": 34260
    },
    {
      "epoch": 0.6016608437648133,
      "grad_norm": 0.06510616957342819,
      "learning_rate": 0.00019492363127012342,
      "loss": 0.8538,
      "step": 34270
    },
    {
      "epoch": 0.6018364086448147,
      "grad_norm": 0.07478147354247672,
      "learning_rate": 0.00019492010809494324,
      "loss": 0.8555,
      "step": 34280
    },
    {
      "epoch": 0.6020119735248161,
      "grad_norm": 0.1040143627505095,
      "learning_rate": 0.0001949165837297773,
      "loss": 0.8466,
      "step": 34290
    },
    {
      "epoch": 0.6021875384048175,
      "grad_norm": 0.08853087611619244,
      "learning_rate": 0.00019491305817467025,
      "loss": 0.8503,
      "step": 34300
    },
    {
      "epoch": 0.6023631032848189,
      "grad_norm": 0.07051582668870532,
      "learning_rate": 0.00019490953142966677,
      "loss": 0.8504,
      "step": 34310
    },
    {
      "epoch": 0.6025386681648203,
      "grad_norm": 0.06507810491914909,
      "learning_rate": 0.00019490600349481152,
      "loss": 0.8515,
      "step": 34320
    },
    {
      "epoch": 0.6027142330448217,
      "grad_norm": 0.05803637568335623,
      "learning_rate": 0.0001949024743701492,
      "loss": 0.8496,
      "step": 34330
    },
    {
      "epoch": 0.6028897979248231,
      "grad_norm": 0.08804326703234838,
      "learning_rate": 0.0001948989440557245,
      "loss": 0.8557,
      "step": 34340
    },
    {
      "epoch": 0.6030653628048245,
      "grad_norm": 0.07521450401318783,
      "learning_rate": 0.00019489541255158225,
      "loss": 0.8478,
      "step": 34350
    },
    {
      "epoch": 0.603240927684826,
      "grad_norm": 0.06362222379227477,
      "learning_rate": 0.00019489187985776706,
      "loss": 0.8495,
      "step": 34360
    },
    {
      "epoch": 0.6034164925648273,
      "grad_norm": 0.08576794885034267,
      "learning_rate": 0.00019488834597432378,
      "loss": 0.8545,
      "step": 34370
    },
    {
      "epoch": 0.6035920574448287,
      "grad_norm": 0.05403595388340057,
      "learning_rate": 0.00019488481090129721,
      "loss": 0.8478,
      "step": 34380
    },
    {
      "epoch": 0.6037676223248302,
      "grad_norm": 0.08444872432734009,
      "learning_rate": 0.00019488127463873202,
      "loss": 0.8528,
      "step": 34390
    },
    {
      "epoch": 0.6039431872048315,
      "grad_norm": 0.07713042900071777,
      "learning_rate": 0.00019487773718667313,
      "loss": 0.8421,
      "step": 34400
    },
    {
      "epoch": 0.6041187520848329,
      "grad_norm": 0.06253681330025504,
      "learning_rate": 0.00019487419854516533,
      "loss": 0.8569,
      "step": 34410
    },
    {
      "epoch": 0.6042943169648344,
      "grad_norm": 0.07541534542829878,
      "learning_rate": 0.00019487065871425343,
      "loss": 0.85,
      "step": 34420
    },
    {
      "epoch": 0.6044698818448357,
      "grad_norm": 0.06569872882552927,
      "learning_rate": 0.0001948671176939823,
      "loss": 0.8516,
      "step": 34430
    },
    {
      "epoch": 0.6046454467248371,
      "grad_norm": 0.05787874501281186,
      "learning_rate": 0.0001948635754843968,
      "loss": 0.8519,
      "step": 34440
    },
    {
      "epoch": 0.6048210116048386,
      "grad_norm": 0.07088968423592686,
      "learning_rate": 0.00019486003208554178,
      "loss": 0.8557,
      "step": 34450
    },
    {
      "epoch": 0.60499657648484,
      "grad_norm": 0.07937747612236312,
      "learning_rate": 0.0001948564874974622,
      "loss": 0.8539,
      "step": 34460
    },
    {
      "epoch": 0.6051721413648414,
      "grad_norm": 0.054789960025789536,
      "learning_rate": 0.0001948529417202029,
      "loss": 0.8558,
      "step": 34470
    },
    {
      "epoch": 0.6053477062448428,
      "grad_norm": 0.06191223901418255,
      "learning_rate": 0.00019484939475380888,
      "loss": 0.8563,
      "step": 34480
    },
    {
      "epoch": 0.6055232711248442,
      "grad_norm": 0.07394923377594598,
      "learning_rate": 0.00019484584659832502,
      "loss": 0.8464,
      "step": 34490
    },
    {
      "epoch": 0.6056988360048456,
      "grad_norm": 0.08377041515233934,
      "learning_rate": 0.00019484229725379631,
      "loss": 0.8532,
      "step": 34500
    },
    {
      "epoch": 0.605874400884847,
      "grad_norm": 0.08768644896356383,
      "learning_rate": 0.0001948387467202677,
      "loss": 0.8554,
      "step": 34510
    },
    {
      "epoch": 0.6060499657648484,
      "grad_norm": 0.09512010864008254,
      "learning_rate": 0.00019483519499778417,
      "loss": 0.8465,
      "step": 34520
    },
    {
      "epoch": 0.6062255306448499,
      "grad_norm": 0.05139461283749981,
      "learning_rate": 0.00019483164208639075,
      "loss": 0.8521,
      "step": 34530
    },
    {
      "epoch": 0.6064010955248512,
      "grad_norm": 0.07612331646637036,
      "learning_rate": 0.00019482808798613246,
      "loss": 0.8545,
      "step": 34540
    },
    {
      "epoch": 0.6065766604048526,
      "grad_norm": 0.08350886131162212,
      "learning_rate": 0.0001948245326970543,
      "loss": 0.8518,
      "step": 34550
    },
    {
      "epoch": 0.6067522252848541,
      "grad_norm": 0.08362189520514308,
      "learning_rate": 0.00019482097621920132,
      "loss": 0.855,
      "step": 34560
    },
    {
      "epoch": 0.6069277901648554,
      "grad_norm": 0.10478827675198307,
      "learning_rate": 0.0001948174185526186,
      "loss": 0.8552,
      "step": 34570
    },
    {
      "epoch": 0.6071033550448568,
      "grad_norm": 0.057904228374658286,
      "learning_rate": 0.0001948138596973512,
      "loss": 0.8462,
      "step": 34580
    },
    {
      "epoch": 0.6072789199248583,
      "grad_norm": 0.09665187461508506,
      "learning_rate": 0.00019481029965344422,
      "loss": 0.8619,
      "step": 34590
    },
    {
      "epoch": 0.6074544848048596,
      "grad_norm": 0.058454337765642655,
      "learning_rate": 0.00019480673842094278,
      "loss": 0.8562,
      "step": 34600
    },
    {
      "epoch": 0.607630049684861,
      "grad_norm": 0.07033780953395381,
      "learning_rate": 0.00019480317599989197,
      "loss": 0.8539,
      "step": 34610
    },
    {
      "epoch": 0.6078056145648625,
      "grad_norm": 0.07943137462688576,
      "learning_rate": 0.00019479961239033696,
      "loss": 0.856,
      "step": 34620
    },
    {
      "epoch": 0.6079811794448639,
      "grad_norm": 0.0707311613267164,
      "learning_rate": 0.00019479604759232283,
      "loss": 0.8555,
      "step": 34630
    },
    {
      "epoch": 0.6081567443248652,
      "grad_norm": 0.07987787880419214,
      "learning_rate": 0.00019479248160589487,
      "loss": 0.8531,
      "step": 34640
    },
    {
      "epoch": 0.6083323092048667,
      "grad_norm": 0.05356653201670992,
      "learning_rate": 0.00019478891443109816,
      "loss": 0.8574,
      "step": 34650
    },
    {
      "epoch": 0.6085078740848681,
      "grad_norm": 0.04843820949108811,
      "learning_rate": 0.00019478534606797796,
      "loss": 0.8537,
      "step": 34660
    },
    {
      "epoch": 0.6086834389648694,
      "grad_norm": 0.09059237229235718,
      "learning_rate": 0.0001947817765165794,
      "loss": 0.8569,
      "step": 34670
    },
    {
      "epoch": 0.6088590038448709,
      "grad_norm": 0.12780915295476045,
      "learning_rate": 0.00019477820577694778,
      "loss": 0.8624,
      "step": 34680
    },
    {
      "epoch": 0.6090345687248723,
      "grad_norm": 0.09966172911865735,
      "learning_rate": 0.00019477463384912834,
      "loss": 0.8524,
      "step": 34690
    },
    {
      "epoch": 0.6092101336048736,
      "grad_norm": 0.06560661431774106,
      "learning_rate": 0.0001947710607331663,
      "loss": 0.8469,
      "step": 34700
    },
    {
      "epoch": 0.6093856984848751,
      "grad_norm": 0.06268195356031811,
      "learning_rate": 0.00019476748642910694,
      "loss": 0.8544,
      "step": 34710
    },
    {
      "epoch": 0.6095612633648765,
      "grad_norm": 0.08819573541613727,
      "learning_rate": 0.00019476391093699557,
      "loss": 0.8509,
      "step": 34720
    },
    {
      "epoch": 0.6097368282448778,
      "grad_norm": 0.057939836604518694,
      "learning_rate": 0.00019476033425687748,
      "loss": 0.8565,
      "step": 34730
    },
    {
      "epoch": 0.6099123931248793,
      "grad_norm": 0.05143523506946456,
      "learning_rate": 0.000194756756388798,
      "loss": 0.8581,
      "step": 34740
    },
    {
      "epoch": 0.6100879580048807,
      "grad_norm": 0.06646143553453308,
      "learning_rate": 0.0001947531773328024,
      "loss": 0.8512,
      "step": 34750
    },
    {
      "epoch": 0.6102635228848821,
      "grad_norm": 0.08173241030253155,
      "learning_rate": 0.00019474959708893613,
      "loss": 0.8491,
      "step": 34760
    },
    {
      "epoch": 0.6104390877648835,
      "grad_norm": 0.055024221196336494,
      "learning_rate": 0.00019474601565724447,
      "loss": 0.857,
      "step": 34770
    },
    {
      "epoch": 0.6106146526448849,
      "grad_norm": 0.05243261163213456,
      "learning_rate": 0.00019474243303777282,
      "loss": 0.8516,
      "step": 34780
    },
    {
      "epoch": 0.6107902175248863,
      "grad_norm": 0.0636974240648765,
      "learning_rate": 0.00019473884923056658,
      "loss": 0.8612,
      "step": 34790
    },
    {
      "epoch": 0.6109657824048877,
      "grad_norm": 0.07280353527188747,
      "learning_rate": 0.00019473526423567116,
      "loss": 0.8549,
      "step": 34800
    },
    {
      "epoch": 0.6111413472848891,
      "grad_norm": 0.050835402437000354,
      "learning_rate": 0.000194731678053132,
      "loss": 0.8646,
      "step": 34810
    },
    {
      "epoch": 0.6113169121648905,
      "grad_norm": 0.06979680964693555,
      "learning_rate": 0.0001947280906829945,
      "loss": 0.85,
      "step": 34820
    },
    {
      "epoch": 0.611492477044892,
      "grad_norm": 0.052344970397456426,
      "learning_rate": 0.00019472450212530408,
      "loss": 0.8504,
      "step": 34830
    },
    {
      "epoch": 0.6116680419248933,
      "grad_norm": 0.05888793116942056,
      "learning_rate": 0.0001947209123801063,
      "loss": 0.8591,
      "step": 34840
    },
    {
      "epoch": 0.6118436068048948,
      "grad_norm": 0.06725615289430606,
      "learning_rate": 0.0001947173214474466,
      "loss": 0.8504,
      "step": 34850
    },
    {
      "epoch": 0.6120191716848962,
      "grad_norm": 0.10396665103145325,
      "learning_rate": 0.0001947137293273705,
      "loss": 0.8568,
      "step": 34860
    },
    {
      "epoch": 0.6121947365648975,
      "grad_norm": 0.05593176188823821,
      "learning_rate": 0.00019471013601992344,
      "loss": 0.8539,
      "step": 34870
    },
    {
      "epoch": 0.612370301444899,
      "grad_norm": 0.06518475816837115,
      "learning_rate": 0.00019470654152515103,
      "loss": 0.8526,
      "step": 34880
    },
    {
      "epoch": 0.6125458663249004,
      "grad_norm": 0.06985734451624626,
      "learning_rate": 0.0001947029458430988,
      "loss": 0.856,
      "step": 34890
    },
    {
      "epoch": 0.6127214312049017,
      "grad_norm": 0.0839693543407261,
      "learning_rate": 0.00019469934897381225,
      "loss": 0.8485,
      "step": 34900
    },
    {
      "epoch": 0.6128969960849032,
      "grad_norm": 0.05545623296106622,
      "learning_rate": 0.00019469575091733698,
      "loss": 0.8534,
      "step": 34910
    },
    {
      "epoch": 0.6130725609649046,
      "grad_norm": 0.0671854795455712,
      "learning_rate": 0.00019469215167371861,
      "loss": 0.8554,
      "step": 34920
    },
    {
      "epoch": 0.613248125844906,
      "grad_norm": 0.07516795658812246,
      "learning_rate": 0.00019468855124300273,
      "loss": 0.8539,
      "step": 34930
    },
    {
      "epoch": 0.6134236907249074,
      "grad_norm": 0.0657584907316818,
      "learning_rate": 0.00019468494962523496,
      "loss": 0.8533,
      "step": 34940
    },
    {
      "epoch": 0.6135992556049088,
      "grad_norm": 0.05710331700715575,
      "learning_rate": 0.00019468134682046087,
      "loss": 0.8542,
      "step": 34950
    },
    {
      "epoch": 0.6137748204849102,
      "grad_norm": 0.06165258854173056,
      "learning_rate": 0.00019467774282872622,
      "loss": 0.8501,
      "step": 34960
    },
    {
      "epoch": 0.6139503853649116,
      "grad_norm": 0.05925466068659386,
      "learning_rate": 0.0001946741376500766,
      "loss": 0.86,
      "step": 34970
    },
    {
      "epoch": 0.614125950244913,
      "grad_norm": 0.08297599360398715,
      "learning_rate": 0.0001946705312845577,
      "loss": 0.8483,
      "step": 34980
    },
    {
      "epoch": 0.6143015151249144,
      "grad_norm": 0.06848582810772601,
      "learning_rate": 0.0001946669237322152,
      "loss": 0.8524,
      "step": 34990
    },
    {
      "epoch": 0.6144770800049159,
      "grad_norm": 0.06415836359778315,
      "learning_rate": 0.00019466331499309484,
      "loss": 0.8539,
      "step": 35000
    },
    {
      "epoch": 0.6146526448849172,
      "grad_norm": 0.0699341589465008,
      "learning_rate": 0.00019465970506724235,
      "loss": 0.8462,
      "step": 35010
    },
    {
      "epoch": 0.6148282097649186,
      "grad_norm": 0.056854316111495426,
      "learning_rate": 0.0001946560939547034,
      "loss": 0.8502,
      "step": 35020
    },
    {
      "epoch": 0.6150037746449201,
      "grad_norm": 0.062496942147475074,
      "learning_rate": 0.0001946524816555238,
      "loss": 0.8525,
      "step": 35030
    },
    {
      "epoch": 0.6151793395249214,
      "grad_norm": 0.09785139797026242,
      "learning_rate": 0.00019464886816974934,
      "loss": 0.852,
      "step": 35040
    },
    {
      "epoch": 0.6153549044049228,
      "grad_norm": 0.05990644503817809,
      "learning_rate": 0.00019464525349742576,
      "loss": 0.8506,
      "step": 35050
    },
    {
      "epoch": 0.6155304692849243,
      "grad_norm": 0.05639479508290439,
      "learning_rate": 0.00019464163763859886,
      "loss": 0.8515,
      "step": 35060
    },
    {
      "epoch": 0.6157060341649256,
      "grad_norm": 0.06107081242997416,
      "learning_rate": 0.00019463802059331445,
      "loss": 0.8489,
      "step": 35070
    },
    {
      "epoch": 0.615881599044927,
      "grad_norm": 0.061323045401882895,
      "learning_rate": 0.00019463440236161842,
      "loss": 0.8605,
      "step": 35080
    },
    {
      "epoch": 0.6160571639249285,
      "grad_norm": 0.07425184240079015,
      "learning_rate": 0.00019463078294355653,
      "loss": 0.8502,
      "step": 35090
    },
    {
      "epoch": 0.6162327288049299,
      "grad_norm": 0.05686981818084413,
      "learning_rate": 0.00019462716233917471,
      "loss": 0.8518,
      "step": 35100
    },
    {
      "epoch": 0.6164082936849312,
      "grad_norm": 0.05353311030969614,
      "learning_rate": 0.00019462354054851876,
      "loss": 0.8634,
      "step": 35110
    },
    {
      "epoch": 0.6165838585649327,
      "grad_norm": 0.04669879776062381,
      "learning_rate": 0.00019461991757163463,
      "loss": 0.8483,
      "step": 35120
    },
    {
      "epoch": 0.6167594234449341,
      "grad_norm": 0.08159711820773151,
      "learning_rate": 0.0001946162934085682,
      "loss": 0.8529,
      "step": 35130
    },
    {
      "epoch": 0.6169349883249354,
      "grad_norm": 0.07477392707231849,
      "learning_rate": 0.0001946126680593654,
      "loss": 0.8571,
      "step": 35140
    },
    {
      "epoch": 0.6171105532049369,
      "grad_norm": 0.07083966007230696,
      "learning_rate": 0.00019460904152407215,
      "loss": 0.8509,
      "step": 35150
    },
    {
      "epoch": 0.6172861180849383,
      "grad_norm": 0.05950305218945452,
      "learning_rate": 0.00019460541380273439,
      "loss": 0.85,
      "step": 35160
    },
    {
      "epoch": 0.6174616829649396,
      "grad_norm": 0.05840870989158585,
      "learning_rate": 0.0001946017848953981,
      "loss": 0.8574,
      "step": 35170
    },
    {
      "epoch": 0.6176372478449411,
      "grad_norm": 0.0630857625531817,
      "learning_rate": 0.00019459815480210926,
      "loss": 0.8545,
      "step": 35180
    },
    {
      "epoch": 0.6178128127249425,
      "grad_norm": 0.11852715547995746,
      "learning_rate": 0.00019459452352291388,
      "loss": 0.8651,
      "step": 35190
    },
    {
      "epoch": 0.617988377604944,
      "grad_norm": 0.10820359625858686,
      "learning_rate": 0.00019459089105785792,
      "loss": 0.85,
      "step": 35200
    },
    {
      "epoch": 0.6181639424849453,
      "grad_norm": 0.07438104524092465,
      "learning_rate": 0.00019458725740698743,
      "loss": 0.8476,
      "step": 35210
    },
    {
      "epoch": 0.6183395073649467,
      "grad_norm": 0.06411775496824966,
      "learning_rate": 0.00019458362257034848,
      "loss": 0.8565,
      "step": 35220
    },
    {
      "epoch": 0.6185150722449482,
      "grad_norm": 0.06020952816501542,
      "learning_rate": 0.00019457998654798708,
      "loss": 0.8528,
      "step": 35230
    },
    {
      "epoch": 0.6186906371249495,
      "grad_norm": 0.06809620171079009,
      "learning_rate": 0.00019457634933994932,
      "loss": 0.8472,
      "step": 35240
    },
    {
      "epoch": 0.6188662020049509,
      "grad_norm": 0.07869469384170233,
      "learning_rate": 0.0001945727109462813,
      "loss": 0.8543,
      "step": 35250
    },
    {
      "epoch": 0.6190417668849524,
      "grad_norm": 0.0969606223284533,
      "learning_rate": 0.00019456907136702905,
      "loss": 0.856,
      "step": 35260
    },
    {
      "epoch": 0.6192173317649537,
      "grad_norm": 0.07657802789244514,
      "learning_rate": 0.00019456543060223877,
      "loss": 0.8468,
      "step": 35270
    },
    {
      "epoch": 0.6193928966449551,
      "grad_norm": 0.11840800771275703,
      "learning_rate": 0.00019456178865195656,
      "loss": 0.8569,
      "step": 35280
    },
    {
      "epoch": 0.6195684615249566,
      "grad_norm": 0.11782081658706912,
      "learning_rate": 0.00019455814551622855,
      "loss": 0.8507,
      "step": 35290
    },
    {
      "epoch": 0.619744026404958,
      "grad_norm": 0.051412046120139276,
      "learning_rate": 0.00019455450119510088,
      "loss": 0.8501,
      "step": 35300
    },
    {
      "epoch": 0.6199195912849593,
      "grad_norm": 0.060447155727302464,
      "learning_rate": 0.0001945508556886198,
      "loss": 0.8568,
      "step": 35310
    },
    {
      "epoch": 0.6200951561649608,
      "grad_norm": 0.056978118148601875,
      "learning_rate": 0.00019454720899683142,
      "loss": 0.8552,
      "step": 35320
    },
    {
      "epoch": 0.6202707210449622,
      "grad_norm": 0.07041942313975055,
      "learning_rate": 0.00019454356111978197,
      "loss": 0.8539,
      "step": 35330
    },
    {
      "epoch": 0.6204462859249635,
      "grad_norm": 0.06063322173155,
      "learning_rate": 0.00019453991205751767,
      "loss": 0.8517,
      "step": 35340
    },
    {
      "epoch": 0.620621850804965,
      "grad_norm": 0.054265375415042304,
      "learning_rate": 0.00019453626181008474,
      "loss": 0.8552,
      "step": 35350
    },
    {
      "epoch": 0.6207974156849664,
      "grad_norm": 0.06929662846308,
      "learning_rate": 0.0001945326103775295,
      "loss": 0.8401,
      "step": 35360
    },
    {
      "epoch": 0.6209729805649677,
      "grad_norm": 0.07075652924699205,
      "learning_rate": 0.00019452895775989816,
      "loss": 0.8524,
      "step": 35370
    },
    {
      "epoch": 0.6211485454449692,
      "grad_norm": 0.10783682228526828,
      "learning_rate": 0.00019452530395723693,
      "loss": 0.8513,
      "step": 35380
    },
    {
      "epoch": 0.6213241103249706,
      "grad_norm": 0.0828773076149495,
      "learning_rate": 0.00019452164896959222,
      "loss": 0.8552,
      "step": 35390
    },
    {
      "epoch": 0.621499675204972,
      "grad_norm": 0.06800421382588624,
      "learning_rate": 0.0001945179927970103,
      "loss": 0.8597,
      "step": 35400
    },
    {
      "epoch": 0.6216752400849734,
      "grad_norm": 0.09257435864103389,
      "learning_rate": 0.00019451433543953746,
      "loss": 0.8496,
      "step": 35410
    },
    {
      "epoch": 0.6218508049649748,
      "grad_norm": 0.0952315488951537,
      "learning_rate": 0.00019451067689722009,
      "loss": 0.85,
      "step": 35420
    },
    {
      "epoch": 0.6220263698449762,
      "grad_norm": 0.04250544748115005,
      "learning_rate": 0.00019450701717010454,
      "loss": 0.8576,
      "step": 35430
    },
    {
      "epoch": 0.6222019347249776,
      "grad_norm": 0.09417091775707515,
      "learning_rate": 0.00019450335625823714,
      "loss": 0.8535,
      "step": 35440
    },
    {
      "epoch": 0.622377499604979,
      "grad_norm": 0.06978252480478395,
      "learning_rate": 0.0001944996941616643,
      "loss": 0.8593,
      "step": 35450
    },
    {
      "epoch": 0.6225530644849804,
      "grad_norm": 0.0595719592262861,
      "learning_rate": 0.0001944960308804324,
      "loss": 0.8614,
      "step": 35460
    },
    {
      "epoch": 0.6227286293649819,
      "grad_norm": 0.06421082553200595,
      "learning_rate": 0.00019449236641458787,
      "loss": 0.8526,
      "step": 35470
    },
    {
      "epoch": 0.6229041942449832,
      "grad_norm": 0.04825250045597371,
      "learning_rate": 0.00019448870076417717,
      "loss": 0.8566,
      "step": 35480
    },
    {
      "epoch": 0.6230797591249846,
      "grad_norm": 0.057775415430948954,
      "learning_rate": 0.00019448503392924672,
      "loss": 0.8543,
      "step": 35490
    },
    {
      "epoch": 0.6232553240049861,
      "grad_norm": 0.06200843207263272,
      "learning_rate": 0.00019448136590984294,
      "loss": 0.8559,
      "step": 35500
    },
    {
      "epoch": 0.6234308888849874,
      "grad_norm": 0.06729193899831018,
      "learning_rate": 0.00019447769670601233,
      "loss": 0.8561,
      "step": 35510
    },
    {
      "epoch": 0.6236064537649888,
      "grad_norm": 0.06436754804426412,
      "learning_rate": 0.00019447402631780144,
      "loss": 0.8599,
      "step": 35520
    },
    {
      "epoch": 0.6237820186449903,
      "grad_norm": 0.06539739452354515,
      "learning_rate": 0.0001944703547452567,
      "loss": 0.8645,
      "step": 35530
    },
    {
      "epoch": 0.6239575835249916,
      "grad_norm": 0.053083807336956945,
      "learning_rate": 0.00019446668198842463,
      "loss": 0.8525,
      "step": 35540
    },
    {
      "epoch": 0.624133148404993,
      "grad_norm": 0.0749700342389716,
      "learning_rate": 0.00019446300804735176,
      "loss": 0.8522,
      "step": 35550
    },
    {
      "epoch": 0.6243087132849945,
      "grad_norm": 0.06392837979672508,
      "learning_rate": 0.00019445933292208472,
      "loss": 0.8644,
      "step": 35560
    },
    {
      "epoch": 0.6244842781649959,
      "grad_norm": 0.07479804828629034,
      "learning_rate": 0.00019445565661267003,
      "loss": 0.8543,
      "step": 35570
    },
    {
      "epoch": 0.6246598430449973,
      "grad_norm": 0.07714742521734454,
      "learning_rate": 0.00019445197911915422,
      "loss": 0.8567,
      "step": 35580
    },
    {
      "epoch": 0.6248354079249987,
      "grad_norm": 0.07212979777045361,
      "learning_rate": 0.00019444830044158394,
      "loss": 0.8586,
      "step": 35590
    },
    {
      "epoch": 0.6250109728050001,
      "grad_norm": 0.08013864706255329,
      "learning_rate": 0.00019444462058000578,
      "loss": 0.854,
      "step": 35600
    },
    {
      "epoch": 0.6251865376850015,
      "grad_norm": 0.08123172652829323,
      "learning_rate": 0.00019444093953446638,
      "loss": 0.862,
      "step": 35610
    },
    {
      "epoch": 0.6253621025650029,
      "grad_norm": 0.05798147529217454,
      "learning_rate": 0.00019443725730501233,
      "loss": 0.8573,
      "step": 35620
    },
    {
      "epoch": 0.6255376674450043,
      "grad_norm": 0.055017675113326975,
      "learning_rate": 0.00019443357389169037,
      "loss": 0.8507,
      "step": 35630
    },
    {
      "epoch": 0.6257132323250058,
      "grad_norm": 0.07599734709884388,
      "learning_rate": 0.0001944298892945471,
      "loss": 0.8515,
      "step": 35640
    },
    {
      "epoch": 0.6258887972050071,
      "grad_norm": 0.07687000643091679,
      "learning_rate": 0.00019442620351362925,
      "loss": 0.853,
      "step": 35650
    },
    {
      "epoch": 0.6260643620850085,
      "grad_norm": 0.07479447487599987,
      "learning_rate": 0.0001944225165489835,
      "loss": 0.8569,
      "step": 35660
    },
    {
      "epoch": 0.62623992696501,
      "grad_norm": 0.06641948546158757,
      "learning_rate": 0.00019441882840065652,
      "loss": 0.8523,
      "step": 35670
    },
    {
      "epoch": 0.6264154918450113,
      "grad_norm": 0.06156123953782701,
      "learning_rate": 0.0001944151390686951,
      "loss": 0.8596,
      "step": 35680
    },
    {
      "epoch": 0.6265910567250127,
      "grad_norm": 0.06228433821008845,
      "learning_rate": 0.00019441144855314598,
      "loss": 0.8556,
      "step": 35690
    },
    {
      "epoch": 0.6267666216050142,
      "grad_norm": 0.09107482336510285,
      "learning_rate": 0.00019440775685405587,
      "loss": 0.8605,
      "step": 35700
    },
    {
      "epoch": 0.6269421864850155,
      "grad_norm": 0.05458582870981273,
      "learning_rate": 0.0001944040639714716,
      "loss": 0.8515,
      "step": 35710
    },
    {
      "epoch": 0.6271177513650169,
      "grad_norm": 0.09173208687315132,
      "learning_rate": 0.00019440036990543995,
      "loss": 0.8489,
      "step": 35720
    },
    {
      "epoch": 0.6272933162450184,
      "grad_norm": 0.0681794116682946,
      "learning_rate": 0.0001943966746560077,
      "loss": 0.8537,
      "step": 35730
    },
    {
      "epoch": 0.6274688811250198,
      "grad_norm": 0.08587233644344451,
      "learning_rate": 0.00019439297822322168,
      "loss": 0.8461,
      "step": 35740
    },
    {
      "epoch": 0.6276444460050211,
      "grad_norm": 0.06356517416369904,
      "learning_rate": 0.00019438928060712873,
      "loss": 0.8459,
      "step": 35750
    },
    {
      "epoch": 0.6278200108850226,
      "grad_norm": 0.05177373409835948,
      "learning_rate": 0.00019438558180777572,
      "loss": 0.8524,
      "step": 35760
    },
    {
      "epoch": 0.627995575765024,
      "grad_norm": 0.06735479320037475,
      "learning_rate": 0.00019438188182520943,
      "loss": 0.8545,
      "step": 35770
    },
    {
      "epoch": 0.6281711406450253,
      "grad_norm": 0.04545458783677734,
      "learning_rate": 0.00019437818065947687,
      "loss": 0.8513,
      "step": 35780
    },
    {
      "epoch": 0.6283467055250268,
      "grad_norm": 0.0743030862743558,
      "learning_rate": 0.0001943744783106248,
      "loss": 0.8485,
      "step": 35790
    },
    {
      "epoch": 0.6285222704050282,
      "grad_norm": 0.06242227394235423,
      "learning_rate": 0.00019437077477870026,
      "loss": 0.8541,
      "step": 35800
    },
    {
      "epoch": 0.6286978352850295,
      "grad_norm": 0.057731829242700525,
      "learning_rate": 0.00019436707006375005,
      "loss": 0.851,
      "step": 35810
    },
    {
      "epoch": 0.628873400165031,
      "grad_norm": 0.062025293211090925,
      "learning_rate": 0.00019436336416582125,
      "loss": 0.8572,
      "step": 35820
    },
    {
      "epoch": 0.6290489650450324,
      "grad_norm": 0.0658025467520558,
      "learning_rate": 0.00019435965708496065,
      "loss": 0.8431,
      "step": 35830
    },
    {
      "epoch": 0.6292245299250337,
      "grad_norm": 0.057001104229437535,
      "learning_rate": 0.00019435594882121534,
      "loss": 0.8545,
      "step": 35840
    },
    {
      "epoch": 0.6294000948050352,
      "grad_norm": 0.05765310490469667,
      "learning_rate": 0.00019435223937463224,
      "loss": 0.8583,
      "step": 35850
    },
    {
      "epoch": 0.6295756596850366,
      "grad_norm": 0.07063938627794408,
      "learning_rate": 0.0001943485287452584,
      "loss": 0.8549,
      "step": 35860
    },
    {
      "epoch": 0.629751224565038,
      "grad_norm": 0.050729905105507095,
      "learning_rate": 0.0001943448169331408,
      "loss": 0.8576,
      "step": 35870
    },
    {
      "epoch": 0.6299267894450394,
      "grad_norm": 0.09129028499355137,
      "learning_rate": 0.0001943411039383265,
      "loss": 0.8517,
      "step": 35880
    },
    {
      "epoch": 0.6301023543250408,
      "grad_norm": 0.07732869895160206,
      "learning_rate": 0.00019433738976086252,
      "loss": 0.8514,
      "step": 35890
    },
    {
      "epoch": 0.6302779192050422,
      "grad_norm": 0.07663801325704016,
      "learning_rate": 0.00019433367440079593,
      "loss": 0.8494,
      "step": 35900
    },
    {
      "epoch": 0.6304534840850436,
      "grad_norm": 0.05256721152514931,
      "learning_rate": 0.00019432995785817378,
      "loss": 0.8594,
      "step": 35910
    },
    {
      "epoch": 0.630629048965045,
      "grad_norm": 0.07482238246995944,
      "learning_rate": 0.00019432624013304318,
      "loss": 0.859,
      "step": 35920
    },
    {
      "epoch": 0.6308046138450465,
      "grad_norm": 0.05195625898512678,
      "learning_rate": 0.0001943225212254512,
      "loss": 0.8439,
      "step": 35930
    },
    {
      "epoch": 0.6309801787250479,
      "grad_norm": 0.05836665698651033,
      "learning_rate": 0.00019431880113544506,
      "loss": 0.8556,
      "step": 35940
    },
    {
      "epoch": 0.6311557436050492,
      "grad_norm": 0.06637415270835795,
      "learning_rate": 0.00019431507986307178,
      "loss": 0.8505,
      "step": 35950
    },
    {
      "epoch": 0.6313313084850507,
      "grad_norm": 0.0653947570492513,
      "learning_rate": 0.00019431135740837854,
      "loss": 0.8558,
      "step": 35960
    },
    {
      "epoch": 0.6315068733650521,
      "grad_norm": 0.07278740518276308,
      "learning_rate": 0.00019430763377141256,
      "loss": 0.8619,
      "step": 35970
    },
    {
      "epoch": 0.6316824382450534,
      "grad_norm": 0.05890957119272087,
      "learning_rate": 0.00019430390895222094,
      "loss": 0.8538,
      "step": 35980
    },
    {
      "epoch": 0.6318580031250549,
      "grad_norm": 0.0746151396973331,
      "learning_rate": 0.00019430018295085095,
      "loss": 0.853,
      "step": 35990
    },
    {
      "epoch": 0.6320335680050563,
      "grad_norm": 0.056191638741705914,
      "learning_rate": 0.00019429645576734975,
      "loss": 0.8496,
      "step": 36000
    },
    {
      "epoch": 0.6322091328850576,
      "grad_norm": 0.05035942127316008,
      "learning_rate": 0.00019429272740176457,
      "loss": 0.8561,
      "step": 36010
    },
    {
      "epoch": 0.6323846977650591,
      "grad_norm": 0.07052213208075607,
      "learning_rate": 0.00019428899785414264,
      "loss": 0.8534,
      "step": 36020
    },
    {
      "epoch": 0.6325602626450605,
      "grad_norm": 0.05475529996117037,
      "learning_rate": 0.00019428526712453122,
      "loss": 0.8574,
      "step": 36030
    },
    {
      "epoch": 0.6327358275250619,
      "grad_norm": 0.08155595905734213,
      "learning_rate": 0.00019428153521297762,
      "loss": 0.8546,
      "step": 36040
    },
    {
      "epoch": 0.6329113924050633,
      "grad_norm": 0.07526565631650077,
      "learning_rate": 0.00019427780211952908,
      "loss": 0.8509,
      "step": 36050
    },
    {
      "epoch": 0.6330869572850647,
      "grad_norm": 0.04797682428152055,
      "learning_rate": 0.0001942740678442329,
      "loss": 0.853,
      "step": 36060
    },
    {
      "epoch": 0.6332625221650661,
      "grad_norm": 0.06358432686153061,
      "learning_rate": 0.00019427033238713645,
      "loss": 0.8561,
      "step": 36070
    },
    {
      "epoch": 0.6334380870450675,
      "grad_norm": 0.06505799840811059,
      "learning_rate": 0.00019426659574828695,
      "loss": 0.8546,
      "step": 36080
    },
    {
      "epoch": 0.6336136519250689,
      "grad_norm": 0.08030254511205151,
      "learning_rate": 0.00019426285792773185,
      "loss": 0.8567,
      "step": 36090
    },
    {
      "epoch": 0.6337892168050703,
      "grad_norm": 0.08157588032788955,
      "learning_rate": 0.00019425911892551845,
      "loss": 0.8501,
      "step": 36100
    },
    {
      "epoch": 0.6339647816850718,
      "grad_norm": 0.06870263335132609,
      "learning_rate": 0.00019425537874169412,
      "loss": 0.8508,
      "step": 36110
    },
    {
      "epoch": 0.6341403465650731,
      "grad_norm": 0.07160045281201786,
      "learning_rate": 0.00019425163737630632,
      "loss": 0.8475,
      "step": 36120
    },
    {
      "epoch": 0.6343159114450745,
      "grad_norm": 0.05822114303336733,
      "learning_rate": 0.00019424789482940235,
      "loss": 0.8555,
      "step": 36130
    },
    {
      "epoch": 0.634491476325076,
      "grad_norm": 0.05068342806106513,
      "learning_rate": 0.00019424415110102968,
      "loss": 0.8593,
      "step": 36140
    },
    {
      "epoch": 0.6346670412050773,
      "grad_norm": 0.061873293117991245,
      "learning_rate": 0.0001942404061912358,
      "loss": 0.8596,
      "step": 36150
    },
    {
      "epoch": 0.6348426060850787,
      "grad_norm": 0.051877590190828715,
      "learning_rate": 0.00019423666010006806,
      "loss": 0.8565,
      "step": 36160
    },
    {
      "epoch": 0.6350181709650802,
      "grad_norm": 0.06718572663230457,
      "learning_rate": 0.000194232912827574,
      "loss": 0.8592,
      "step": 36170
    },
    {
      "epoch": 0.6351937358450815,
      "grad_norm": 0.07219749948735053,
      "learning_rate": 0.00019422916437380106,
      "loss": 0.8443,
      "step": 36180
    },
    {
      "epoch": 0.6353693007250829,
      "grad_norm": 0.08080954873563949,
      "learning_rate": 0.0001942254147387967,
      "loss": 0.8612,
      "step": 36190
    },
    {
      "epoch": 0.6355448656050844,
      "grad_norm": 0.07246671536703048,
      "learning_rate": 0.00019422166392260852,
      "loss": 0.8605,
      "step": 36200
    },
    {
      "epoch": 0.6357204304850858,
      "grad_norm": 0.07713676763317144,
      "learning_rate": 0.00019421791192528396,
      "loss": 0.8456,
      "step": 36210
    },
    {
      "epoch": 0.6358959953650871,
      "grad_norm": 0.060114851381019865,
      "learning_rate": 0.0001942141587468706,
      "loss": 0.8473,
      "step": 36220
    },
    {
      "epoch": 0.6360715602450886,
      "grad_norm": 0.05181978553966608,
      "learning_rate": 0.00019421040438741597,
      "loss": 0.8511,
      "step": 36230
    },
    {
      "epoch": 0.63624712512509,
      "grad_norm": 0.06839396033294472,
      "learning_rate": 0.00019420664884696765,
      "loss": 0.8531,
      "step": 36240
    },
    {
      "epoch": 0.6364226900050913,
      "grad_norm": 0.06459867748021651,
      "learning_rate": 0.00019420289212557327,
      "loss": 0.8541,
      "step": 36250
    },
    {
      "epoch": 0.6365982548850928,
      "grad_norm": 0.0856833376884787,
      "learning_rate": 0.00019419913422328035,
      "loss": 0.8515,
      "step": 36260
    },
    {
      "epoch": 0.6367738197650942,
      "grad_norm": 0.06780356702193666,
      "learning_rate": 0.00019419537514013657,
      "loss": 0.8564,
      "step": 36270
    },
    {
      "epoch": 0.6369493846450955,
      "grad_norm": 0.07958979410284236,
      "learning_rate": 0.0001941916148761895,
      "loss": 0.8433,
      "step": 36280
    },
    {
      "epoch": 0.637124949525097,
      "grad_norm": 0.07850443744298666,
      "learning_rate": 0.0001941878534314868,
      "loss": 0.8458,
      "step": 36290
    },
    {
      "epoch": 0.6373005144050984,
      "grad_norm": 0.07707223822641375,
      "learning_rate": 0.00019418409080607617,
      "loss": 0.8516,
      "step": 36300
    },
    {
      "epoch": 0.6374760792850999,
      "grad_norm": 0.06214210032982156,
      "learning_rate": 0.0001941803270000052,
      "loss": 0.8515,
      "step": 36310
    },
    {
      "epoch": 0.6376516441651012,
      "grad_norm": 0.06355774981208874,
      "learning_rate": 0.00019417656201332168,
      "loss": 0.8497,
      "step": 36320
    },
    {
      "epoch": 0.6378272090451026,
      "grad_norm": 0.06067445704718926,
      "learning_rate": 0.00019417279584607323,
      "loss": 0.8514,
      "step": 36330
    },
    {
      "epoch": 0.6380027739251041,
      "grad_norm": 0.08770130500633533,
      "learning_rate": 0.0001941690284983076,
      "loss": 0.847,
      "step": 36340
    },
    {
      "epoch": 0.6381783388051054,
      "grad_norm": 0.06558576971820798,
      "learning_rate": 0.00019416525997007254,
      "loss": 0.8567,
      "step": 36350
    },
    {
      "epoch": 0.6383539036851068,
      "grad_norm": 0.05820107222837087,
      "learning_rate": 0.00019416149026141581,
      "loss": 0.8539,
      "step": 36360
    },
    {
      "epoch": 0.6385294685651083,
      "grad_norm": 0.05267804125794523,
      "learning_rate": 0.00019415771937238507,
      "loss": 0.8586,
      "step": 36370
    },
    {
      "epoch": 0.6387050334451096,
      "grad_norm": 0.08695477606851233,
      "learning_rate": 0.00019415394730302823,
      "loss": 0.851,
      "step": 36380
    },
    {
      "epoch": 0.638880598325111,
      "grad_norm": 0.06538505723661595,
      "learning_rate": 0.00019415017405339298,
      "loss": 0.8606,
      "step": 36390
    },
    {
      "epoch": 0.6390561632051125,
      "grad_norm": 0.06376383163971192,
      "learning_rate": 0.00019414639962352724,
      "loss": 0.8486,
      "step": 36400
    },
    {
      "epoch": 0.6392317280851139,
      "grad_norm": 0.04739020285717028,
      "learning_rate": 0.00019414262401347868,
      "loss": 0.8548,
      "step": 36410
    },
    {
      "epoch": 0.6394072929651152,
      "grad_norm": 0.06173938329739082,
      "learning_rate": 0.0001941388472232953,
      "loss": 0.8472,
      "step": 36420
    },
    {
      "epoch": 0.6395828578451167,
      "grad_norm": 0.04840986598326304,
      "learning_rate": 0.00019413506925302485,
      "loss": 0.8626,
      "step": 36430
    },
    {
      "epoch": 0.6397584227251181,
      "grad_norm": 0.06173354933683468,
      "learning_rate": 0.00019413129010271524,
      "loss": 0.8569,
      "step": 36440
    },
    {
      "epoch": 0.6399339876051194,
      "grad_norm": 0.043975034048418765,
      "learning_rate": 0.0001941275097724143,
      "loss": 0.8632,
      "step": 36450
    },
    {
      "epoch": 0.6401095524851209,
      "grad_norm": 0.07374775523059414,
      "learning_rate": 0.00019412372826216995,
      "loss": 0.8513,
      "step": 36460
    },
    {
      "epoch": 0.6402851173651223,
      "grad_norm": 0.07095778101191982,
      "learning_rate": 0.00019411994557203013,
      "loss": 0.8496,
      "step": 36470
    },
    {
      "epoch": 0.6404606822451236,
      "grad_norm": 0.0588199543716021,
      "learning_rate": 0.0001941161617020428,
      "loss": 0.8547,
      "step": 36480
    },
    {
      "epoch": 0.6406362471251251,
      "grad_norm": 0.05790530082306417,
      "learning_rate": 0.0001941123766522558,
      "loss": 0.8558,
      "step": 36490
    },
    {
      "epoch": 0.6408118120051265,
      "grad_norm": 0.0820573216689123,
      "learning_rate": 0.00019410859042271717,
      "loss": 0.851,
      "step": 36500
    },
    {
      "epoch": 0.6409873768851279,
      "grad_norm": 0.06959660918500564,
      "learning_rate": 0.00019410480301347484,
      "loss": 0.8515,
      "step": 36510
    },
    {
      "epoch": 0.6411629417651293,
      "grad_norm": 0.05388821726819028,
      "learning_rate": 0.0001941010144245768,
      "loss": 0.8506,
      "step": 36520
    },
    {
      "epoch": 0.6413385066451307,
      "grad_norm": 0.06397575300903521,
      "learning_rate": 0.00019409722465607113,
      "loss": 0.8553,
      "step": 36530
    },
    {
      "epoch": 0.6415140715251321,
      "grad_norm": 0.08310493491477118,
      "learning_rate": 0.00019409343370800573,
      "loss": 0.8484,
      "step": 36540
    },
    {
      "epoch": 0.6416896364051335,
      "grad_norm": 0.0793142164178133,
      "learning_rate": 0.00019408964158042867,
      "loss": 0.8501,
      "step": 36550
    },
    {
      "epoch": 0.6418652012851349,
      "grad_norm": 0.09197633468028114,
      "learning_rate": 0.00019408584827338804,
      "loss": 0.8474,
      "step": 36560
    },
    {
      "epoch": 0.6420407661651363,
      "grad_norm": 0.08012505785680628,
      "learning_rate": 0.00019408205378693185,
      "loss": 0.8558,
      "step": 36570
    },
    {
      "epoch": 0.6422163310451378,
      "grad_norm": 0.07553688354126001,
      "learning_rate": 0.0001940782581211082,
      "loss": 0.8617,
      "step": 36580
    },
    {
      "epoch": 0.6423918959251391,
      "grad_norm": 0.07068734646253913,
      "learning_rate": 0.00019407446127596523,
      "loss": 0.8555,
      "step": 36590
    },
    {
      "epoch": 0.6425674608051405,
      "grad_norm": 0.05181002428238948,
      "learning_rate": 0.00019407066325155095,
      "loss": 0.8485,
      "step": 36600
    },
    {
      "epoch": 0.642743025685142,
      "grad_norm": 0.0573593318759357,
      "learning_rate": 0.00019406686404791352,
      "loss": 0.8522,
      "step": 36610
    },
    {
      "epoch": 0.6429185905651433,
      "grad_norm": 0.07303750883194077,
      "learning_rate": 0.00019406306366510115,
      "loss": 0.8503,
      "step": 36620
    },
    {
      "epoch": 0.6430941554451447,
      "grad_norm": 0.0659211719524383,
      "learning_rate": 0.00019405926210316188,
      "loss": 0.8606,
      "step": 36630
    },
    {
      "epoch": 0.6432697203251462,
      "grad_norm": 0.0656886444006019,
      "learning_rate": 0.00019405545936214392,
      "loss": 0.8549,
      "step": 36640
    },
    {
      "epoch": 0.6434452852051475,
      "grad_norm": 0.062332585438194754,
      "learning_rate": 0.00019405165544209547,
      "loss": 0.8505,
      "step": 36650
    },
    {
      "epoch": 0.643620850085149,
      "grad_norm": 0.06841176714510738,
      "learning_rate": 0.0001940478503430647,
      "loss": 0.8551,
      "step": 36660
    },
    {
      "epoch": 0.6437964149651504,
      "grad_norm": 0.07150888512434354,
      "learning_rate": 0.00019404404406509986,
      "loss": 0.8561,
      "step": 36670
    },
    {
      "epoch": 0.6439719798451518,
      "grad_norm": 0.06379551699126415,
      "learning_rate": 0.0001940402366082491,
      "loss": 0.8476,
      "step": 36680
    },
    {
      "epoch": 0.6441475447251532,
      "grad_norm": 0.061140521720130285,
      "learning_rate": 0.00019403642797256073,
      "loss": 0.8533,
      "step": 36690
    },
    {
      "epoch": 0.6443231096051546,
      "grad_norm": 0.10444503389107626,
      "learning_rate": 0.00019403261815808306,
      "loss": 0.8493,
      "step": 36700
    },
    {
      "epoch": 0.644498674485156,
      "grad_norm": 0.0700846855138409,
      "learning_rate": 0.00019402880716486422,
      "loss": 0.8488,
      "step": 36710
    },
    {
      "epoch": 0.6446742393651574,
      "grad_norm": 0.06713513941886354,
      "learning_rate": 0.00019402499499295256,
      "loss": 0.8556,
      "step": 36720
    },
    {
      "epoch": 0.6448498042451588,
      "grad_norm": 0.061225876506984425,
      "learning_rate": 0.0001940211816423964,
      "loss": 0.849,
      "step": 36730
    },
    {
      "epoch": 0.6450253691251602,
      "grad_norm": 0.06802664143611026,
      "learning_rate": 0.00019401736711324404,
      "loss": 0.8557,
      "step": 36740
    },
    {
      "epoch": 0.6452009340051617,
      "grad_norm": 0.0842578402681984,
      "learning_rate": 0.0001940135514055438,
      "loss": 0.8554,
      "step": 36750
    },
    {
      "epoch": 0.645376498885163,
      "grad_norm": 0.048621232009313696,
      "learning_rate": 0.0001940097345193441,
      "loss": 0.8434,
      "step": 36760
    },
    {
      "epoch": 0.6455520637651644,
      "grad_norm": 0.09336965721386961,
      "learning_rate": 0.00019400591645469314,
      "loss": 0.8533,
      "step": 36770
    },
    {
      "epoch": 0.6457276286451659,
      "grad_norm": 0.06565779274386349,
      "learning_rate": 0.00019400209721163943,
      "loss": 0.8576,
      "step": 36780
    },
    {
      "epoch": 0.6459031935251672,
      "grad_norm": 0.06844731073120032,
      "learning_rate": 0.00019399827679023138,
      "loss": 0.8548,
      "step": 36790
    },
    {
      "epoch": 0.6460787584051686,
      "grad_norm": 0.06856352957673315,
      "learning_rate": 0.00019399445519051726,
      "loss": 0.8496,
      "step": 36800
    },
    {
      "epoch": 0.6462543232851701,
      "grad_norm": 0.053700853849284004,
      "learning_rate": 0.00019399063241254562,
      "loss": 0.8497,
      "step": 36810
    },
    {
      "epoch": 0.6464298881651714,
      "grad_norm": 0.06127983370055654,
      "learning_rate": 0.00019398680845636484,
      "loss": 0.8597,
      "step": 36820
    },
    {
      "epoch": 0.6466054530451728,
      "grad_norm": 0.08646198908870435,
      "learning_rate": 0.00019398298332202335,
      "loss": 0.8495,
      "step": 36830
    },
    {
      "epoch": 0.6467810179251743,
      "grad_norm": 0.08239851108001221,
      "learning_rate": 0.00019397915700956965,
      "loss": 0.8407,
      "step": 36840
    },
    {
      "epoch": 0.6469565828051757,
      "grad_norm": 0.06207546976664638,
      "learning_rate": 0.00019397532951905225,
      "loss": 0.845,
      "step": 36850
    },
    {
      "epoch": 0.647132147685177,
      "grad_norm": 0.08551758212191632,
      "learning_rate": 0.0001939715008505196,
      "loss": 0.8533,
      "step": 36860
    },
    {
      "epoch": 0.6473077125651785,
      "grad_norm": 0.0836928276485169,
      "learning_rate": 0.00019396767100402016,
      "loss": 0.8552,
      "step": 36870
    },
    {
      "epoch": 0.6474832774451799,
      "grad_norm": 0.07469543587171348,
      "learning_rate": 0.00019396383997960257,
      "loss": 0.8445,
      "step": 36880
    },
    {
      "epoch": 0.6476588423251812,
      "grad_norm": 0.05758658249101825,
      "learning_rate": 0.00019396000777731524,
      "loss": 0.8626,
      "step": 36890
    },
    {
      "epoch": 0.6478344072051827,
      "grad_norm": 0.05965928336856097,
      "learning_rate": 0.0001939561743972069,
      "loss": 0.8618,
      "step": 36900
    },
    {
      "epoch": 0.6480099720851841,
      "grad_norm": 0.12204435628010683,
      "learning_rate": 0.0001939523398393259,
      "loss": 0.8552,
      "step": 36910
    },
    {
      "epoch": 0.6481855369651854,
      "grad_norm": 0.06056068795873357,
      "learning_rate": 0.000193948504103721,
      "loss": 0.8563,
      "step": 36920
    },
    {
      "epoch": 0.6483611018451869,
      "grad_norm": 0.09568873045411537,
      "learning_rate": 0.00019394466719044077,
      "loss": 0.8559,
      "step": 36930
    },
    {
      "epoch": 0.6485366667251883,
      "grad_norm": 0.06076745235090655,
      "learning_rate": 0.00019394082909953374,
      "loss": 0.857,
      "step": 36940
    },
    {
      "epoch": 0.6487122316051896,
      "grad_norm": 0.06621688889768534,
      "learning_rate": 0.00019393698983104864,
      "loss": 0.8458,
      "step": 36950
    },
    {
      "epoch": 0.6488877964851911,
      "grad_norm": 0.04461924919043371,
      "learning_rate": 0.00019393314938503402,
      "loss": 0.8553,
      "step": 36960
    },
    {
      "epoch": 0.6490633613651925,
      "grad_norm": 0.061924014414468175,
      "learning_rate": 0.0001939293077615386,
      "loss": 0.8539,
      "step": 36970
    },
    {
      "epoch": 0.6492389262451939,
      "grad_norm": 0.06476495351455278,
      "learning_rate": 0.00019392546496061107,
      "loss": 0.8506,
      "step": 36980
    },
    {
      "epoch": 0.6494144911251953,
      "grad_norm": 0.06953577071739893,
      "learning_rate": 0.00019392162098230006,
      "loss": 0.8581,
      "step": 36990
    },
    {
      "epoch": 0.6495900560051967,
      "grad_norm": 0.0785118898999296,
      "learning_rate": 0.00019391777582665432,
      "loss": 0.8577,
      "step": 37000
    },
    {
      "epoch": 0.6497656208851981,
      "grad_norm": 0.05206107295997744,
      "learning_rate": 0.00019391392949372253,
      "loss": 0.8615,
      "step": 37010
    },
    {
      "epoch": 0.6499411857651995,
      "grad_norm": 0.06496737040581292,
      "learning_rate": 0.00019391008198355348,
      "loss": 0.861,
      "step": 37020
    },
    {
      "epoch": 0.6501167506452009,
      "grad_norm": 0.05692369045321024,
      "learning_rate": 0.00019390623329619585,
      "loss": 0.8527,
      "step": 37030
    },
    {
      "epoch": 0.6502923155252024,
      "grad_norm": 0.05305745381489499,
      "learning_rate": 0.00019390238343169848,
      "loss": 0.8547,
      "step": 37040
    },
    {
      "epoch": 0.6504678804052038,
      "grad_norm": 0.04733323803046007,
      "learning_rate": 0.00019389853239011007,
      "loss": 0.8553,
      "step": 37050
    },
    {
      "epoch": 0.6506434452852051,
      "grad_norm": 0.06540989739031151,
      "learning_rate": 0.00019389468017147945,
      "loss": 0.8561,
      "step": 37060
    },
    {
      "epoch": 0.6508190101652066,
      "grad_norm": 0.06167428874051344,
      "learning_rate": 0.00019389082677585546,
      "loss": 0.8565,
      "step": 37070
    },
    {
      "epoch": 0.650994575045208,
      "grad_norm": 0.06191360125354541,
      "learning_rate": 0.00019388697220328686,
      "loss": 0.8482,
      "step": 37080
    },
    {
      "epoch": 0.6511701399252093,
      "grad_norm": 0.052200485125201194,
      "learning_rate": 0.00019388311645382255,
      "loss": 0.8599,
      "step": 37090
    },
    {
      "epoch": 0.6513457048052108,
      "grad_norm": 0.05238262424281822,
      "learning_rate": 0.00019387925952751137,
      "loss": 0.8465,
      "step": 37100
    },
    {
      "epoch": 0.6515212696852122,
      "grad_norm": 0.0650624249456386,
      "learning_rate": 0.0001938754014244021,
      "loss": 0.8537,
      "step": 37110
    },
    {
      "epoch": 0.6516968345652135,
      "grad_norm": 0.060728436266565156,
      "learning_rate": 0.00019387154214454372,
      "loss": 0.852,
      "step": 37120
    },
    {
      "epoch": 0.651872399445215,
      "grad_norm": 0.07147667386646017,
      "learning_rate": 0.00019386768168798513,
      "loss": 0.8552,
      "step": 37130
    },
    {
      "epoch": 0.6520479643252164,
      "grad_norm": 0.07186009586091582,
      "learning_rate": 0.0001938638200547752,
      "loss": 0.8594,
      "step": 37140
    },
    {
      "epoch": 0.6522235292052178,
      "grad_norm": 0.07252579646041139,
      "learning_rate": 0.00019385995724496288,
      "loss": 0.8568,
      "step": 37150
    },
    {
      "epoch": 0.6523990940852192,
      "grad_norm": 0.04992076063858382,
      "learning_rate": 0.00019385609325859712,
      "loss": 0.8553,
      "step": 37160
    },
    {
      "epoch": 0.6525746589652206,
      "grad_norm": 0.1156216386538427,
      "learning_rate": 0.00019385222809572682,
      "loss": 0.8492,
      "step": 37170
    },
    {
      "epoch": 0.652750223845222,
      "grad_norm": 0.06597385893622341,
      "learning_rate": 0.00019384836175640103,
      "loss": 0.8597,
      "step": 37180
    },
    {
      "epoch": 0.6529257887252234,
      "grad_norm": 0.09201002113372127,
      "learning_rate": 0.0001938444942406687,
      "loss": 0.861,
      "step": 37190
    },
    {
      "epoch": 0.6531013536052248,
      "grad_norm": 0.05997204217787153,
      "learning_rate": 0.0001938406255485788,
      "loss": 0.8565,
      "step": 37200
    },
    {
      "epoch": 0.6532769184852262,
      "grad_norm": 0.06725761373097415,
      "learning_rate": 0.0001938367556801804,
      "loss": 0.8536,
      "step": 37210
    },
    {
      "epoch": 0.6534524833652277,
      "grad_norm": 0.048547706175447805,
      "learning_rate": 0.00019383288463552253,
      "loss": 0.8462,
      "step": 37220
    },
    {
      "epoch": 0.653628048245229,
      "grad_norm": 0.0854249613098066,
      "learning_rate": 0.0001938290124146542,
      "loss": 0.8504,
      "step": 37230
    },
    {
      "epoch": 0.6538036131252304,
      "grad_norm": 0.10695412726949444,
      "learning_rate": 0.00019382513901762451,
      "loss": 0.8563,
      "step": 37240
    },
    {
      "epoch": 0.6539791780052319,
      "grad_norm": 0.059549706245784706,
      "learning_rate": 0.00019382126444448253,
      "loss": 0.8528,
      "step": 37250
    },
    {
      "epoch": 0.6541547428852332,
      "grad_norm": 0.06480454683593952,
      "learning_rate": 0.0001938173886952773,
      "loss": 0.8552,
      "step": 37260
    },
    {
      "epoch": 0.6543303077652346,
      "grad_norm": 0.052992262233834436,
      "learning_rate": 0.00019381351177005798,
      "loss": 0.8503,
      "step": 37270
    },
    {
      "epoch": 0.6545058726452361,
      "grad_norm": 0.0840218871094458,
      "learning_rate": 0.00019380963366887372,
      "loss": 0.8532,
      "step": 37280
    },
    {
      "epoch": 0.6546814375252374,
      "grad_norm": 0.08335825080738776,
      "learning_rate": 0.00019380575439177357,
      "loss": 0.8578,
      "step": 37290
    },
    {
      "epoch": 0.6548570024052388,
      "grad_norm": 0.057238469421032806,
      "learning_rate": 0.00019380187393880672,
      "loss": 0.8564,
      "step": 37300
    },
    {
      "epoch": 0.6550325672852403,
      "grad_norm": 0.07772996567163555,
      "learning_rate": 0.00019379799231002235,
      "loss": 0.8573,
      "step": 37310
    },
    {
      "epoch": 0.6552081321652417,
      "grad_norm": 0.055074023597818295,
      "learning_rate": 0.00019379410950546967,
      "loss": 0.8515,
      "step": 37320
    },
    {
      "epoch": 0.655383697045243,
      "grad_norm": 0.074800004767543,
      "learning_rate": 0.0001937902255251978,
      "loss": 0.8528,
      "step": 37330
    },
    {
      "epoch": 0.6555592619252445,
      "grad_norm": 0.06819611774728637,
      "learning_rate": 0.000193786340369256,
      "loss": 0.8536,
      "step": 37340
    },
    {
      "epoch": 0.6557348268052459,
      "grad_norm": 0.0751267868098657,
      "learning_rate": 0.0001937824540376935,
      "loss": 0.8549,
      "step": 37350
    },
    {
      "epoch": 0.6559103916852472,
      "grad_norm": 0.05917351128923754,
      "learning_rate": 0.0001937785665305595,
      "loss": 0.8531,
      "step": 37360
    },
    {
      "epoch": 0.6560859565652487,
      "grad_norm": 0.052737832642472025,
      "learning_rate": 0.00019377467784790331,
      "loss": 0.8513,
      "step": 37370
    },
    {
      "epoch": 0.6562615214452501,
      "grad_norm": 0.061557127069232956,
      "learning_rate": 0.0001937707879897741,
      "loss": 0.8507,
      "step": 37380
    },
    {
      "epoch": 0.6564370863252516,
      "grad_norm": 0.06521211198992052,
      "learning_rate": 0.0001937668969562213,
      "loss": 0.8538,
      "step": 37390
    },
    {
      "epoch": 0.6566126512052529,
      "grad_norm": 0.07378682797359613,
      "learning_rate": 0.00019376300474729412,
      "loss": 0.8545,
      "step": 37400
    },
    {
      "epoch": 0.6567882160852543,
      "grad_norm": 0.07702464063341834,
      "learning_rate": 0.0001937591113630419,
      "loss": 0.8544,
      "step": 37410
    },
    {
      "epoch": 0.6569637809652558,
      "grad_norm": 0.08383112344379849,
      "learning_rate": 0.00019375521680351393,
      "loss": 0.8544,
      "step": 37420
    },
    {
      "epoch": 0.6571393458452571,
      "grad_norm": 0.07530485328552966,
      "learning_rate": 0.00019375132106875961,
      "loss": 0.8552,
      "step": 37430
    },
    {
      "epoch": 0.6573149107252585,
      "grad_norm": 0.0544413732347663,
      "learning_rate": 0.00019374742415882829,
      "loss": 0.8432,
      "step": 37440
    },
    {
      "epoch": 0.65749047560526,
      "grad_norm": 0.06160155077162333,
      "learning_rate": 0.0001937435260737693,
      "loss": 0.8484,
      "step": 37450
    },
    {
      "epoch": 0.6576660404852613,
      "grad_norm": 0.0952057739173672,
      "learning_rate": 0.0001937396268136321,
      "loss": 0.8582,
      "step": 37460
    },
    {
      "epoch": 0.6578416053652627,
      "grad_norm": 0.05244839661665075,
      "learning_rate": 0.00019373572637846603,
      "loss": 0.8535,
      "step": 37470
    },
    {
      "epoch": 0.6580171702452642,
      "grad_norm": 0.06187155209555895,
      "learning_rate": 0.0001937318247683205,
      "loss": 0.8552,
      "step": 37480
    },
    {
      "epoch": 0.6581927351252655,
      "grad_norm": 0.0584040938235431,
      "learning_rate": 0.00019372792198324502,
      "loss": 0.8559,
      "step": 37490
    },
    {
      "epoch": 0.6583683000052669,
      "grad_norm": 0.055474237687724116,
      "learning_rate": 0.00019372401802328901,
      "loss": 0.8438,
      "step": 37500
    },
    {
      "epoch": 0.6585438648852684,
      "grad_norm": 0.0524119338954262,
      "learning_rate": 0.00019372011288850189,
      "loss": 0.8484,
      "step": 37510
    },
    {
      "epoch": 0.6587194297652698,
      "grad_norm": 0.05103332671480185,
      "learning_rate": 0.00019371620657893316,
      "loss": 0.847,
      "step": 37520
    },
    {
      "epoch": 0.6588949946452711,
      "grad_norm": 0.09493167130321829,
      "learning_rate": 0.00019371229909463235,
      "loss": 0.8533,
      "step": 37530
    },
    {
      "epoch": 0.6590705595252726,
      "grad_norm": 0.06317861540815776,
      "learning_rate": 0.00019370839043564894,
      "loss": 0.8554,
      "step": 37540
    },
    {
      "epoch": 0.659246124405274,
      "grad_norm": 0.07241413700919462,
      "learning_rate": 0.00019370448060203246,
      "loss": 0.8489,
      "step": 37550
    },
    {
      "epoch": 0.6594216892852753,
      "grad_norm": 0.07442804515717467,
      "learning_rate": 0.0001937005695938324,
      "loss": 0.8592,
      "step": 37560
    },
    {
      "epoch": 0.6595972541652768,
      "grad_norm": 0.05907274198172103,
      "learning_rate": 0.00019369665741109842,
      "loss": 0.8528,
      "step": 37570
    },
    {
      "epoch": 0.6597728190452782,
      "grad_norm": 0.07008583989482692,
      "learning_rate": 0.00019369274405387998,
      "loss": 0.85,
      "step": 37580
    },
    {
      "epoch": 0.6599483839252795,
      "grad_norm": 0.06629954575937737,
      "learning_rate": 0.00019368882952222673,
      "loss": 0.8531,
      "step": 37590
    },
    {
      "epoch": 0.660123948805281,
      "grad_norm": 0.0657371451834434,
      "learning_rate": 0.00019368491381618828,
      "loss": 0.8632,
      "step": 37600
    },
    {
      "epoch": 0.6602995136852824,
      "grad_norm": 0.05998068569773772,
      "learning_rate": 0.00019368099693581415,
      "loss": 0.8439,
      "step": 37610
    },
    {
      "epoch": 0.6604750785652838,
      "grad_norm": 0.05304750448711977,
      "learning_rate": 0.00019367707888115406,
      "loss": 0.8536,
      "step": 37620
    },
    {
      "epoch": 0.6606506434452852,
      "grad_norm": 0.12268445650693684,
      "learning_rate": 0.0001936731596522576,
      "loss": 0.8542,
      "step": 37630
    },
    {
      "epoch": 0.6608262083252866,
      "grad_norm": 0.07169785483504959,
      "learning_rate": 0.00019366923924917445,
      "loss": 0.86,
      "step": 37640
    },
    {
      "epoch": 0.661001773205288,
      "grad_norm": 0.06573221563619855,
      "learning_rate": 0.0001936653176719543,
      "loss": 0.8487,
      "step": 37650
    },
    {
      "epoch": 0.6611773380852894,
      "grad_norm": 0.07629495591588048,
      "learning_rate": 0.00019366139492064678,
      "loss": 0.8511,
      "step": 37660
    },
    {
      "epoch": 0.6613529029652908,
      "grad_norm": 0.05404632399896313,
      "learning_rate": 0.00019365747099530164,
      "loss": 0.858,
      "step": 37670
    },
    {
      "epoch": 0.6615284678452922,
      "grad_norm": 0.05610280941789637,
      "learning_rate": 0.0001936535458959686,
      "loss": 0.859,
      "step": 37680
    },
    {
      "epoch": 0.6617040327252937,
      "grad_norm": 0.06168636363671716,
      "learning_rate": 0.0001936496196226974,
      "loss": 0.8589,
      "step": 37690
    },
    {
      "epoch": 0.661879597605295,
      "grad_norm": 0.05932972821608176,
      "learning_rate": 0.0001936456921755377,
      "loss": 0.8554,
      "step": 37700
    },
    {
      "epoch": 0.6620551624852964,
      "grad_norm": 0.04937703482295075,
      "learning_rate": 0.00019364176355453935,
      "loss": 0.8568,
      "step": 37710
    },
    {
      "epoch": 0.6622307273652979,
      "grad_norm": 0.08329938439856557,
      "learning_rate": 0.0001936378337597521,
      "loss": 0.8563,
      "step": 37720
    },
    {
      "epoch": 0.6624062922452992,
      "grad_norm": 0.05902381127889509,
      "learning_rate": 0.00019363390279122575,
      "loss": 0.8502,
      "step": 37730
    },
    {
      "epoch": 0.6625818571253006,
      "grad_norm": 0.05623210413911416,
      "learning_rate": 0.0001936299706490101,
      "loss": 0.8555,
      "step": 37740
    },
    {
      "epoch": 0.6627574220053021,
      "grad_norm": 0.05133706041623081,
      "learning_rate": 0.00019362603733315495,
      "loss": 0.8509,
      "step": 37750
    },
    {
      "epoch": 0.6629329868853034,
      "grad_norm": 0.11011880523108593,
      "learning_rate": 0.00019362210284371018,
      "loss": 0.8571,
      "step": 37760
    },
    {
      "epoch": 0.6631085517653049,
      "grad_norm": 0.056125700670241994,
      "learning_rate": 0.00019361816718072562,
      "loss": 0.8491,
      "step": 37770
    },
    {
      "epoch": 0.6632841166453063,
      "grad_norm": 0.06945032734786988,
      "learning_rate": 0.00019361423034425113,
      "loss": 0.8571,
      "step": 37780
    },
    {
      "epoch": 0.6634596815253077,
      "grad_norm": 0.0630911025901408,
      "learning_rate": 0.00019361029233433657,
      "loss": 0.8621,
      "step": 37790
    },
    {
      "epoch": 0.6636352464053091,
      "grad_norm": 0.08774168477957467,
      "learning_rate": 0.00019360635315103186,
      "loss": 0.8589,
      "step": 37800
    },
    {
      "epoch": 0.6638108112853105,
      "grad_norm": 0.07719629346190231,
      "learning_rate": 0.0001936024127943869,
      "loss": 0.8576,
      "step": 37810
    },
    {
      "epoch": 0.6639863761653119,
      "grad_norm": 0.0637031205071741,
      "learning_rate": 0.00019359847126445166,
      "loss": 0.8507,
      "step": 37820
    },
    {
      "epoch": 0.6641619410453133,
      "grad_norm": 0.07173186053671246,
      "learning_rate": 0.000193594528561276,
      "loss": 0.8553,
      "step": 37830
    },
    {
      "epoch": 0.6643375059253147,
      "grad_norm": 0.05762728689472479,
      "learning_rate": 0.00019359058468490997,
      "loss": 0.8554,
      "step": 37840
    },
    {
      "epoch": 0.6645130708053161,
      "grad_norm": 0.06257446837105776,
      "learning_rate": 0.00019358663963540344,
      "loss": 0.8581,
      "step": 37850
    },
    {
      "epoch": 0.6646886356853176,
      "grad_norm": 0.07737467716384697,
      "learning_rate": 0.00019358269341280648,
      "loss": 0.8556,
      "step": 37860
    },
    {
      "epoch": 0.6648642005653189,
      "grad_norm": 0.06481174655334365,
      "learning_rate": 0.00019357874601716902,
      "loss": 0.8572,
      "step": 37870
    },
    {
      "epoch": 0.6650397654453203,
      "grad_norm": 0.0849916077242286,
      "learning_rate": 0.00019357479744854113,
      "loss": 0.8572,
      "step": 37880
    },
    {
      "epoch": 0.6652153303253218,
      "grad_norm": 0.08019375369849202,
      "learning_rate": 0.0001935708477069728,
      "loss": 0.8484,
      "step": 37890
    },
    {
      "epoch": 0.6653908952053231,
      "grad_norm": 0.04663846251202346,
      "learning_rate": 0.00019356689679251408,
      "loss": 0.8467,
      "step": 37900
    },
    {
      "epoch": 0.6655664600853245,
      "grad_norm": 0.0815017942556124,
      "learning_rate": 0.00019356294470521506,
      "loss": 0.8512,
      "step": 37910
    },
    {
      "epoch": 0.665742024965326,
      "grad_norm": 0.05352204645032317,
      "learning_rate": 0.00019355899144512574,
      "loss": 0.8574,
      "step": 37920
    },
    {
      "epoch": 0.6659175898453273,
      "grad_norm": 0.05049978544171157,
      "learning_rate": 0.0001935550370122963,
      "loss": 0.8518,
      "step": 37930
    },
    {
      "epoch": 0.6660931547253287,
      "grad_norm": 0.07150430081331152,
      "learning_rate": 0.0001935510814067768,
      "loss": 0.8491,
      "step": 37940
    },
    {
      "epoch": 0.6662687196053302,
      "grad_norm": 0.06538086035413604,
      "learning_rate": 0.00019354712462861737,
      "loss": 0.844,
      "step": 37950
    },
    {
      "epoch": 0.6664442844853316,
      "grad_norm": 0.06699699572194485,
      "learning_rate": 0.0001935431666778681,
      "loss": 0.8547,
      "step": 37960
    },
    {
      "epoch": 0.6666198493653329,
      "grad_norm": 0.08190437415819164,
      "learning_rate": 0.00019353920755457921,
      "loss": 0.8564,
      "step": 37970
    },
    {
      "epoch": 0.6667954142453344,
      "grad_norm": 0.06734593343604438,
      "learning_rate": 0.00019353524725880083,
      "loss": 0.8523,
      "step": 37980
    },
    {
      "epoch": 0.6669709791253358,
      "grad_norm": 0.08204448515188016,
      "learning_rate": 0.0001935312857905831,
      "loss": 0.8575,
      "step": 37990
    },
    {
      "epoch": 0.6671465440053371,
      "grad_norm": 0.062112684606804865,
      "learning_rate": 0.00019352732314997627,
      "loss": 0.8532,
      "step": 38000
    },
    {
      "epoch": 0.6673221088853386,
      "grad_norm": 0.05615120423314835,
      "learning_rate": 0.0001935233593370305,
      "loss": 0.8469,
      "step": 38010
    },
    {
      "epoch": 0.66749767376534,
      "grad_norm": 0.08780748667504068,
      "learning_rate": 0.00019351939435179605,
      "loss": 0.8554,
      "step": 38020
    },
    {
      "epoch": 0.6676732386453413,
      "grad_norm": 0.07200007753811391,
      "learning_rate": 0.00019351542819432313,
      "loss": 0.8574,
      "step": 38030
    },
    {
      "epoch": 0.6678488035253428,
      "grad_norm": 0.06828041013773944,
      "learning_rate": 0.00019351146086466202,
      "loss": 0.8527,
      "step": 38040
    },
    {
      "epoch": 0.6680243684053442,
      "grad_norm": 0.06736967113016043,
      "learning_rate": 0.00019350749236286298,
      "loss": 0.8484,
      "step": 38050
    },
    {
      "epoch": 0.6681999332853455,
      "grad_norm": 0.06476182808661356,
      "learning_rate": 0.00019350352268897626,
      "loss": 0.8565,
      "step": 38060
    },
    {
      "epoch": 0.668375498165347,
      "grad_norm": 0.058755311279339584,
      "learning_rate": 0.00019349955184305218,
      "loss": 0.8556,
      "step": 38070
    },
    {
      "epoch": 0.6685510630453484,
      "grad_norm": 0.0931523051048159,
      "learning_rate": 0.00019349557982514107,
      "loss": 0.8471,
      "step": 38080
    },
    {
      "epoch": 0.6687266279253498,
      "grad_norm": 0.05676744715878483,
      "learning_rate": 0.00019349160663529325,
      "loss": 0.857,
      "step": 38090
    },
    {
      "epoch": 0.6689021928053512,
      "grad_norm": 0.05059909715748807,
      "learning_rate": 0.000193487632273559,
      "loss": 0.8565,
      "step": 38100
    },
    {
      "epoch": 0.6690777576853526,
      "grad_norm": 0.07809517341834823,
      "learning_rate": 0.00019348365673998874,
      "loss": 0.8529,
      "step": 38110
    },
    {
      "epoch": 0.6692533225653541,
      "grad_norm": 0.0584716014164441,
      "learning_rate": 0.00019347968003463283,
      "loss": 0.8583,
      "step": 38120
    },
    {
      "epoch": 0.6694288874453554,
      "grad_norm": 0.06662936901938107,
      "learning_rate": 0.00019347570215754166,
      "loss": 0.8603,
      "step": 38130
    },
    {
      "epoch": 0.6696044523253568,
      "grad_norm": 0.06586392234386014,
      "learning_rate": 0.0001934717231087656,
      "loss": 0.8608,
      "step": 38140
    },
    {
      "epoch": 0.6697800172053583,
      "grad_norm": 0.10623083761347614,
      "learning_rate": 0.0001934677428883551,
      "loss": 0.8532,
      "step": 38150
    },
    {
      "epoch": 0.6699555820853597,
      "grad_norm": 0.09638475995923605,
      "learning_rate": 0.0001934637614963606,
      "loss": 0.8516,
      "step": 38160
    },
    {
      "epoch": 0.670131146965361,
      "grad_norm": 0.058795422311076415,
      "learning_rate": 0.00019345977893283248,
      "loss": 0.8607,
      "step": 38170
    },
    {
      "epoch": 0.6703067118453625,
      "grad_norm": 0.05649260063757745,
      "learning_rate": 0.00019345579519782126,
      "loss": 0.8532,
      "step": 38180
    },
    {
      "epoch": 0.6704822767253639,
      "grad_norm": 0.07299982565979918,
      "learning_rate": 0.00019345181029137737,
      "loss": 0.8495,
      "step": 38190
    },
    {
      "epoch": 0.6706578416053652,
      "grad_norm": 0.05793285622496156,
      "learning_rate": 0.00019344782421355135,
      "loss": 0.8587,
      "step": 38200
    },
    {
      "epoch": 0.6708334064853667,
      "grad_norm": 0.05833669304085181,
      "learning_rate": 0.0001934438369643937,
      "loss": 0.8482,
      "step": 38210
    },
    {
      "epoch": 0.6710089713653681,
      "grad_norm": 0.0651532540557894,
      "learning_rate": 0.00019343984854395488,
      "loss": 0.8454,
      "step": 38220
    },
    {
      "epoch": 0.6711845362453694,
      "grad_norm": 0.07119110329659442,
      "learning_rate": 0.00019343585895228549,
      "loss": 0.8585,
      "step": 38230
    },
    {
      "epoch": 0.6713601011253709,
      "grad_norm": 0.04163655715027227,
      "learning_rate": 0.00019343186818943607,
      "loss": 0.8548,
      "step": 38240
    },
    {
      "epoch": 0.6715356660053723,
      "grad_norm": 0.061613283317418326,
      "learning_rate": 0.00019342787625545714,
      "loss": 0.8614,
      "step": 38250
    },
    {
      "epoch": 0.6717112308853737,
      "grad_norm": 0.05157890872554061,
      "learning_rate": 0.00019342388315039927,
      "loss": 0.8523,
      "step": 38260
    },
    {
      "epoch": 0.6718867957653751,
      "grad_norm": 0.061386481554547775,
      "learning_rate": 0.00019341988887431312,
      "loss": 0.8527,
      "step": 38270
    },
    {
      "epoch": 0.6720623606453765,
      "grad_norm": 0.0798198824291417,
      "learning_rate": 0.00019341589342724926,
      "loss": 0.851,
      "step": 38280
    },
    {
      "epoch": 0.6722379255253779,
      "grad_norm": 0.05031272671096633,
      "learning_rate": 0.00019341189680925836,
      "loss": 0.8616,
      "step": 38290
    },
    {
      "epoch": 0.6724134904053793,
      "grad_norm": 0.05479447089001072,
      "learning_rate": 0.00019340789902039097,
      "loss": 0.8547,
      "step": 38300
    },
    {
      "epoch": 0.6725890552853807,
      "grad_norm": 0.06697802113164197,
      "learning_rate": 0.0001934039000606978,
      "loss": 0.8555,
      "step": 38310
    },
    {
      "epoch": 0.6727646201653821,
      "grad_norm": 0.060141097433892836,
      "learning_rate": 0.00019339989993022948,
      "loss": 0.8584,
      "step": 38320
    },
    {
      "epoch": 0.6729401850453836,
      "grad_norm": 0.055329847827215,
      "learning_rate": 0.00019339589862903676,
      "loss": 0.8484,
      "step": 38330
    },
    {
      "epoch": 0.6731157499253849,
      "grad_norm": 0.060419160437882743,
      "learning_rate": 0.00019339189615717031,
      "loss": 0.8539,
      "step": 38340
    },
    {
      "epoch": 0.6732913148053863,
      "grad_norm": 0.054023037183733885,
      "learning_rate": 0.00019338789251468076,
      "loss": 0.8531,
      "step": 38350
    },
    {
      "epoch": 0.6734668796853878,
      "grad_norm": 0.08802178944611289,
      "learning_rate": 0.00019338388770161896,
      "loss": 0.861,
      "step": 38360
    },
    {
      "epoch": 0.6736424445653891,
      "grad_norm": 0.09752123999721078,
      "learning_rate": 0.00019337988171803559,
      "loss": 0.8555,
      "step": 38370
    },
    {
      "epoch": 0.6738180094453905,
      "grad_norm": 0.05774619508475413,
      "learning_rate": 0.0001933758745639814,
      "loss": 0.85,
      "step": 38380
    },
    {
      "epoch": 0.673993574325392,
      "grad_norm": 0.048937483933054374,
      "learning_rate": 0.0001933718662395072,
      "loss": 0.8502,
      "step": 38390
    },
    {
      "epoch": 0.6741691392053933,
      "grad_norm": 0.08287386001195325,
      "learning_rate": 0.00019336785674466375,
      "loss": 0.8516,
      "step": 38400
    },
    {
      "epoch": 0.6743447040853947,
      "grad_norm": 0.05978697925761371,
      "learning_rate": 0.00019336384607950181,
      "loss": 0.8505,
      "step": 38410
    },
    {
      "epoch": 0.6745202689653962,
      "grad_norm": 0.06886447323597256,
      "learning_rate": 0.00019335983424407228,
      "loss": 0.8558,
      "step": 38420
    },
    {
      "epoch": 0.6746958338453976,
      "grad_norm": 0.0950789971308266,
      "learning_rate": 0.00019335582123842596,
      "loss": 0.8633,
      "step": 38430
    },
    {
      "epoch": 0.6748713987253989,
      "grad_norm": 0.08032533504576125,
      "learning_rate": 0.00019335180706261365,
      "loss": 0.8586,
      "step": 38440
    },
    {
      "epoch": 0.6750469636054004,
      "grad_norm": 0.057238656686856876,
      "learning_rate": 0.00019334779171668623,
      "loss": 0.8472,
      "step": 38450
    },
    {
      "epoch": 0.6752225284854018,
      "grad_norm": 0.05968640403986131,
      "learning_rate": 0.00019334377520069463,
      "loss": 0.856,
      "step": 38460
    },
    {
      "epoch": 0.6753980933654031,
      "grad_norm": 0.0642107965970955,
      "learning_rate": 0.0001933397575146897,
      "loss": 0.8586,
      "step": 38470
    },
    {
      "epoch": 0.6755736582454046,
      "grad_norm": 0.06006490875712044,
      "learning_rate": 0.00019333573865872234,
      "loss": 0.8544,
      "step": 38480
    },
    {
      "epoch": 0.675749223125406,
      "grad_norm": 0.05519994110777395,
      "learning_rate": 0.00019333171863284345,
      "loss": 0.8537,
      "step": 38490
    },
    {
      "epoch": 0.6759247880054075,
      "grad_norm": 0.06055860848526909,
      "learning_rate": 0.00019332769743710403,
      "loss": 0.8539,
      "step": 38500
    },
    {
      "epoch": 0.6761003528854088,
      "grad_norm": 0.053168845089898914,
      "learning_rate": 0.000193323675071555,
      "loss": 0.8525,
      "step": 38510
    },
    {
      "epoch": 0.6762759177654102,
      "grad_norm": 0.06885724077961501,
      "learning_rate": 0.00019331965153624725,
      "loss": 0.85,
      "step": 38520
    },
    {
      "epoch": 0.6764514826454117,
      "grad_norm": 0.06850288855808581,
      "learning_rate": 0.00019331562683123186,
      "loss": 0.856,
      "step": 38530
    },
    {
      "epoch": 0.676627047525413,
      "grad_norm": 0.07292865868603024,
      "learning_rate": 0.0001933116009565598,
      "loss": 0.8535,
      "step": 38540
    },
    {
      "epoch": 0.6768026124054144,
      "grad_norm": 0.06846372645073082,
      "learning_rate": 0.00019330757391228205,
      "loss": 0.8505,
      "step": 38550
    },
    {
      "epoch": 0.6769781772854159,
      "grad_norm": 0.058091477946306376,
      "learning_rate": 0.0001933035456984496,
      "loss": 0.8495,
      "step": 38560
    },
    {
      "epoch": 0.6771537421654172,
      "grad_norm": 0.05989839426993255,
      "learning_rate": 0.0001932995163151136,
      "loss": 0.8499,
      "step": 38570
    },
    {
      "epoch": 0.6773293070454186,
      "grad_norm": 0.09240247662497077,
      "learning_rate": 0.000193295485762325,
      "loss": 0.8565,
      "step": 38580
    },
    {
      "epoch": 0.6775048719254201,
      "grad_norm": 0.057063204172556496,
      "learning_rate": 0.00019329145404013495,
      "loss": 0.8547,
      "step": 38590
    },
    {
      "epoch": 0.6776804368054214,
      "grad_norm": 0.15279920991728776,
      "learning_rate": 0.00019328742114859443,
      "loss": 0.8586,
      "step": 38600
    },
    {
      "epoch": 0.6778560016854228,
      "grad_norm": 0.08301381738396195,
      "learning_rate": 0.00019328338708775463,
      "loss": 0.8558,
      "step": 38610
    },
    {
      "epoch": 0.6780315665654243,
      "grad_norm": 0.07332676076301292,
      "learning_rate": 0.00019327935185766665,
      "loss": 0.8513,
      "step": 38620
    },
    {
      "epoch": 0.6782071314454257,
      "grad_norm": 0.04996566281264174,
      "learning_rate": 0.00019327531545838158,
      "loss": 0.8602,
      "step": 38630
    },
    {
      "epoch": 0.678382696325427,
      "grad_norm": 0.07444060929298968,
      "learning_rate": 0.0001932712778899506,
      "loss": 0.8489,
      "step": 38640
    },
    {
      "epoch": 0.6785582612054285,
      "grad_norm": 0.05838653074748976,
      "learning_rate": 0.00019326723915242478,
      "loss": 0.8592,
      "step": 38650
    },
    {
      "epoch": 0.6787338260854299,
      "grad_norm": 0.08236897116036301,
      "learning_rate": 0.0001932631992458554,
      "loss": 0.8452,
      "step": 38660
    },
    {
      "epoch": 0.6789093909654312,
      "grad_norm": 0.06238003302606617,
      "learning_rate": 0.00019325915817029364,
      "loss": 0.8512,
      "step": 38670
    },
    {
      "epoch": 0.6790849558454327,
      "grad_norm": 0.07658942518419591,
      "learning_rate": 0.0001932551159257906,
      "loss": 0.8503,
      "step": 38680
    },
    {
      "epoch": 0.6792605207254341,
      "grad_norm": 0.051652297465904655,
      "learning_rate": 0.0001932510725123976,
      "loss": 0.863,
      "step": 38690
    },
    {
      "epoch": 0.6794360856054354,
      "grad_norm": 0.058276152637775726,
      "learning_rate": 0.0001932470279301658,
      "loss": 0.8533,
      "step": 38700
    },
    {
      "epoch": 0.6796116504854369,
      "grad_norm": 0.05188917669731097,
      "learning_rate": 0.0001932429821791465,
      "loss": 0.8601,
      "step": 38710
    },
    {
      "epoch": 0.6797872153654383,
      "grad_norm": 0.0852786127616298,
      "learning_rate": 0.00019323893525939094,
      "loss": 0.8509,
      "step": 38720
    },
    {
      "epoch": 0.6799627802454397,
      "grad_norm": 0.05371347240418529,
      "learning_rate": 0.0001932348871709504,
      "loss": 0.8516,
      "step": 38730
    },
    {
      "epoch": 0.6801383451254411,
      "grad_norm": 0.06175767364634412,
      "learning_rate": 0.00019323083791387613,
      "loss": 0.846,
      "step": 38740
    },
    {
      "epoch": 0.6803139100054425,
      "grad_norm": 0.06482905751299804,
      "learning_rate": 0.0001932267874882195,
      "loss": 0.8532,
      "step": 38750
    },
    {
      "epoch": 0.6804894748854439,
      "grad_norm": 0.06701586287069444,
      "learning_rate": 0.0001932227358940318,
      "loss": 0.8584,
      "step": 38760
    },
    {
      "epoch": 0.6806650397654453,
      "grad_norm": 0.08841277703472471,
      "learning_rate": 0.00019321868313136432,
      "loss": 0.8506,
      "step": 38770
    },
    {
      "epoch": 0.6808406046454467,
      "grad_norm": 0.052896336535333564,
      "learning_rate": 0.00019321462920026848,
      "loss": 0.8558,
      "step": 38780
    },
    {
      "epoch": 0.6810161695254481,
      "grad_norm": 0.06982670010308453,
      "learning_rate": 0.0001932105741007956,
      "loss": 0.8562,
      "step": 38790
    },
    {
      "epoch": 0.6811917344054496,
      "grad_norm": 0.12828401747542187,
      "learning_rate": 0.0001932065178329971,
      "loss": 0.8518,
      "step": 38800
    },
    {
      "epoch": 0.6813672992854509,
      "grad_norm": 0.0736766727163479,
      "learning_rate": 0.00019320246039692434,
      "loss": 0.8486,
      "step": 38810
    },
    {
      "epoch": 0.6815428641654523,
      "grad_norm": 0.06060630103327168,
      "learning_rate": 0.0001931984017926287,
      "loss": 0.8561,
      "step": 38820
    },
    {
      "epoch": 0.6817184290454538,
      "grad_norm": 0.0596949079670509,
      "learning_rate": 0.00019319434202016172,
      "loss": 0.854,
      "step": 38830
    },
    {
      "epoch": 0.6818939939254551,
      "grad_norm": 0.0884984598032678,
      "learning_rate": 0.0001931902810795747,
      "loss": 0.8438,
      "step": 38840
    },
    {
      "epoch": 0.6820695588054566,
      "grad_norm": 0.07427328943601644,
      "learning_rate": 0.00019318621897091915,
      "loss": 0.8578,
      "step": 38850
    },
    {
      "epoch": 0.682245123685458,
      "grad_norm": 0.0695403498132366,
      "learning_rate": 0.00019318215569424655,
      "loss": 0.8587,
      "step": 38860
    },
    {
      "epoch": 0.6824206885654593,
      "grad_norm": 0.052921105612545385,
      "learning_rate": 0.00019317809124960836,
      "loss": 0.8624,
      "step": 38870
    },
    {
      "epoch": 0.6825962534454608,
      "grad_norm": 0.10602625417418367,
      "learning_rate": 0.0001931740256370561,
      "loss": 0.8525,
      "step": 38880
    },
    {
      "epoch": 0.6827718183254622,
      "grad_norm": 0.08029971683443676,
      "learning_rate": 0.00019316995885664126,
      "loss": 0.8611,
      "step": 38890
    },
    {
      "epoch": 0.6829473832054636,
      "grad_norm": 0.07532149671644263,
      "learning_rate": 0.00019316589090841537,
      "loss": 0.8571,
      "step": 38900
    },
    {
      "epoch": 0.683122948085465,
      "grad_norm": 0.0488963992628108,
      "learning_rate": 0.00019316182179243,
      "loss": 0.8439,
      "step": 38910
    },
    {
      "epoch": 0.6832985129654664,
      "grad_norm": 0.04996543073060992,
      "learning_rate": 0.00019315775150873668,
      "loss": 0.8519,
      "step": 38920
    },
    {
      "epoch": 0.6834740778454678,
      "grad_norm": 0.08368173784236026,
      "learning_rate": 0.00019315368005738698,
      "loss": 0.8602,
      "step": 38930
    },
    {
      "epoch": 0.6836496427254692,
      "grad_norm": 0.05969460573442689,
      "learning_rate": 0.0001931496074384325,
      "loss": 0.8509,
      "step": 38940
    },
    {
      "epoch": 0.6838252076054706,
      "grad_norm": 0.06950926342423024,
      "learning_rate": 0.00019314553365192482,
      "loss": 0.8543,
      "step": 38950
    },
    {
      "epoch": 0.684000772485472,
      "grad_norm": 0.05466953021066055,
      "learning_rate": 0.0001931414586979156,
      "loss": 0.8604,
      "step": 38960
    },
    {
      "epoch": 0.6841763373654735,
      "grad_norm": 0.0592583004826709,
      "learning_rate": 0.0001931373825764564,
      "loss": 0.8613,
      "step": 38970
    },
    {
      "epoch": 0.6843519022454748,
      "grad_norm": 0.070511066870787,
      "learning_rate": 0.00019313330528759892,
      "loss": 0.8455,
      "step": 38980
    },
    {
      "epoch": 0.6845274671254762,
      "grad_norm": 0.051798172212695205,
      "learning_rate": 0.00019312922683139479,
      "loss": 0.8526,
      "step": 38990
    },
    {
      "epoch": 0.6847030320054777,
      "grad_norm": 0.07215883328100364,
      "learning_rate": 0.0001931251472078957,
      "loss": 0.852,
      "step": 39000
    },
    {
      "epoch": 0.684878596885479,
      "grad_norm": 0.0618639252225575,
      "learning_rate": 0.00019312106641715336,
      "loss": 0.859,
      "step": 39010
    },
    {
      "epoch": 0.6850541617654804,
      "grad_norm": 0.05551352564423542,
      "learning_rate": 0.00019311698445921945,
      "loss": 0.8585,
      "step": 39020
    },
    {
      "epoch": 0.6852297266454819,
      "grad_norm": 0.058557227476278166,
      "learning_rate": 0.0001931129013341457,
      "loss": 0.8508,
      "step": 39030
    },
    {
      "epoch": 0.6854052915254832,
      "grad_norm": 0.0732357403262482,
      "learning_rate": 0.00019310881704198384,
      "loss": 0.8603,
      "step": 39040
    },
    {
      "epoch": 0.6855808564054846,
      "grad_norm": 0.09603189234764403,
      "learning_rate": 0.0001931047315827856,
      "loss": 0.8627,
      "step": 39050
    },
    {
      "epoch": 0.6857564212854861,
      "grad_norm": 0.05584783677663213,
      "learning_rate": 0.00019310064495660275,
      "loss": 0.8556,
      "step": 39060
    },
    {
      "epoch": 0.6859319861654875,
      "grad_norm": 0.0722215460733453,
      "learning_rate": 0.0001930965571634871,
      "loss": 0.8544,
      "step": 39070
    },
    {
      "epoch": 0.6861075510454888,
      "grad_norm": 0.0515256471561081,
      "learning_rate": 0.0001930924682034904,
      "loss": 0.8512,
      "step": 39080
    },
    {
      "epoch": 0.6862831159254903,
      "grad_norm": 0.06307840612849448,
      "learning_rate": 0.00019308837807666452,
      "loss": 0.8527,
      "step": 39090
    },
    {
      "epoch": 0.6864586808054917,
      "grad_norm": 0.07251188614187588,
      "learning_rate": 0.0001930842867830612,
      "loss": 0.8483,
      "step": 39100
    },
    {
      "epoch": 0.686634245685493,
      "grad_norm": 0.04592357363491942,
      "learning_rate": 0.00019308019432273232,
      "loss": 0.8508,
      "step": 39110
    },
    {
      "epoch": 0.6868098105654945,
      "grad_norm": 0.05842964633874537,
      "learning_rate": 0.00019307610069572976,
      "loss": 0.8631,
      "step": 39120
    },
    {
      "epoch": 0.6869853754454959,
      "grad_norm": 0.0548974517983271,
      "learning_rate": 0.00019307200590210533,
      "loss": 0.8468,
      "step": 39130
    },
    {
      "epoch": 0.6871609403254972,
      "grad_norm": 0.07734569009092715,
      "learning_rate": 0.00019306790994191096,
      "loss": 0.8581,
      "step": 39140
    },
    {
      "epoch": 0.6873365052054987,
      "grad_norm": 0.06702685900146772,
      "learning_rate": 0.0001930638128151985,
      "loss": 0.8541,
      "step": 39150
    },
    {
      "epoch": 0.6875120700855001,
      "grad_norm": 0.06147504787293771,
      "learning_rate": 0.0001930597145220199,
      "loss": 0.8593,
      "step": 39160
    },
    {
      "epoch": 0.6876876349655014,
      "grad_norm": 0.08401141176432524,
      "learning_rate": 0.00019305561506242707,
      "loss": 0.8422,
      "step": 39170
    },
    {
      "epoch": 0.6878631998455029,
      "grad_norm": 0.1019247445622885,
      "learning_rate": 0.00019305151443647196,
      "loss": 0.8534,
      "step": 39180
    },
    {
      "epoch": 0.6880387647255043,
      "grad_norm": 0.08465951681213887,
      "learning_rate": 0.0001930474126442065,
      "loss": 0.8588,
      "step": 39190
    },
    {
      "epoch": 0.6882143296055057,
      "grad_norm": 0.04819082361805634,
      "learning_rate": 0.00019304330968568273,
      "loss": 0.8538,
      "step": 39200
    },
    {
      "epoch": 0.6883898944855071,
      "grad_norm": 0.08349136568384237,
      "learning_rate": 0.00019303920556095253,
      "loss": 0.8564,
      "step": 39210
    },
    {
      "epoch": 0.6885654593655085,
      "grad_norm": 0.0803805851132133,
      "learning_rate": 0.000193035100270068,
      "loss": 0.8523,
      "step": 39220
    },
    {
      "epoch": 0.68874102424551,
      "grad_norm": 0.08243130545606239,
      "learning_rate": 0.00019303099381308108,
      "loss": 0.853,
      "step": 39230
    },
    {
      "epoch": 0.6889165891255113,
      "grad_norm": 0.05699464595008366,
      "learning_rate": 0.00019302688619004385,
      "loss": 0.8592,
      "step": 39240
    },
    {
      "epoch": 0.6890921540055127,
      "grad_norm": 0.058983319507713905,
      "learning_rate": 0.00019302277740100836,
      "loss": 0.8509,
      "step": 39250
    },
    {
      "epoch": 0.6892677188855142,
      "grad_norm": 0.05611598573331356,
      "learning_rate": 0.0001930186674460266,
      "loss": 0.852,
      "step": 39260
    },
    {
      "epoch": 0.6894432837655156,
      "grad_norm": 0.072694381629622,
      "learning_rate": 0.00019301455632515072,
      "loss": 0.8529,
      "step": 39270
    },
    {
      "epoch": 0.6896188486455169,
      "grad_norm": 0.06085516882022055,
      "learning_rate": 0.00019301044403843273,
      "loss": 0.8506,
      "step": 39280
    },
    {
      "epoch": 0.6897944135255184,
      "grad_norm": 0.051872929277386064,
      "learning_rate": 0.0001930063305859248,
      "loss": 0.8584,
      "step": 39290
    },
    {
      "epoch": 0.6899699784055198,
      "grad_norm": 0.06197476955839769,
      "learning_rate": 0.00019300221596767905,
      "loss": 0.8544,
      "step": 39300
    },
    {
      "epoch": 0.6901455432855211,
      "grad_norm": 0.054877418541637614,
      "learning_rate": 0.0001929981001837476,
      "loss": 0.8561,
      "step": 39310
    },
    {
      "epoch": 0.6903211081655226,
      "grad_norm": 0.05333720123494409,
      "learning_rate": 0.00019299398323418255,
      "loss": 0.8587,
      "step": 39320
    },
    {
      "epoch": 0.690496673045524,
      "grad_norm": 0.06290232228870606,
      "learning_rate": 0.00019298986511903613,
      "loss": 0.8525,
      "step": 39330
    },
    {
      "epoch": 0.6906722379255253,
      "grad_norm": 0.0667830221012249,
      "learning_rate": 0.00019298574583836048,
      "loss": 0.851,
      "step": 39340
    },
    {
      "epoch": 0.6908478028055268,
      "grad_norm": 0.07089261200270268,
      "learning_rate": 0.0001929816253922078,
      "loss": 0.8575,
      "step": 39350
    },
    {
      "epoch": 0.6910233676855282,
      "grad_norm": 0.06633403435046449,
      "learning_rate": 0.0001929775037806303,
      "loss": 0.859,
      "step": 39360
    },
    {
      "epoch": 0.6911989325655296,
      "grad_norm": 0.08159110217248701,
      "learning_rate": 0.00019297338100368022,
      "loss": 0.8515,
      "step": 39370
    },
    {
      "epoch": 0.691374497445531,
      "grad_norm": 0.05944260381423131,
      "learning_rate": 0.00019296925706140973,
      "loss": 0.848,
      "step": 39380
    },
    {
      "epoch": 0.6915500623255324,
      "grad_norm": 0.05161463902254573,
      "learning_rate": 0.00019296513195387116,
      "loss": 0.8547,
      "step": 39390
    },
    {
      "epoch": 0.6917256272055338,
      "grad_norm": 0.07701673482884352,
      "learning_rate": 0.00019296100568111672,
      "loss": 0.8582,
      "step": 39400
    },
    {
      "epoch": 0.6919011920855352,
      "grad_norm": 0.08289918849164897,
      "learning_rate": 0.00019295687824319875,
      "loss": 0.8509,
      "step": 39410
    },
    {
      "epoch": 0.6920767569655366,
      "grad_norm": 0.0815473795101243,
      "learning_rate": 0.00019295274964016947,
      "loss": 0.8497,
      "step": 39420
    },
    {
      "epoch": 0.692252321845538,
      "grad_norm": 0.06215755010616076,
      "learning_rate": 0.0001929486198720813,
      "loss": 0.8563,
      "step": 39430
    },
    {
      "epoch": 0.6924278867255395,
      "grad_norm": 0.10291679395820223,
      "learning_rate": 0.0001929444889389864,
      "loss": 0.8475,
      "step": 39440
    },
    {
      "epoch": 0.6926034516055408,
      "grad_norm": 0.090666628887683,
      "learning_rate": 0.00019294035684093726,
      "loss": 0.8494,
      "step": 39450
    },
    {
      "epoch": 0.6927790164855422,
      "grad_norm": 0.07973267476430718,
      "learning_rate": 0.00019293622357798615,
      "loss": 0.8572,
      "step": 39460
    },
    {
      "epoch": 0.6929545813655437,
      "grad_norm": 0.09042869485698973,
      "learning_rate": 0.00019293208915018549,
      "loss": 0.8468,
      "step": 39470
    },
    {
      "epoch": 0.693130146245545,
      "grad_norm": 0.07232202727710102,
      "learning_rate": 0.00019292795355758765,
      "loss": 0.8544,
      "step": 39480
    },
    {
      "epoch": 0.6933057111255464,
      "grad_norm": 0.05852719714458008,
      "learning_rate": 0.00019292381680024502,
      "loss": 0.8558,
      "step": 39490
    },
    {
      "epoch": 0.6934812760055479,
      "grad_norm": 0.08352401576331973,
      "learning_rate": 0.00019291967887820997,
      "loss": 0.8591,
      "step": 39500
    },
    {
      "epoch": 0.6936568408855492,
      "grad_norm": 0.059308469750079064,
      "learning_rate": 0.000192915539791535,
      "loss": 0.8533,
      "step": 39510
    },
    {
      "epoch": 0.6938324057655506,
      "grad_norm": 0.06687425893166533,
      "learning_rate": 0.00019291139954027254,
      "loss": 0.8549,
      "step": 39520
    },
    {
      "epoch": 0.6940079706455521,
      "grad_norm": 0.07771428369904236,
      "learning_rate": 0.000192907258124475,
      "loss": 0.849,
      "step": 39530
    },
    {
      "epoch": 0.6941835355255535,
      "grad_norm": 0.057288094928347084,
      "learning_rate": 0.00019290311554419492,
      "loss": 0.856,
      "step": 39540
    },
    {
      "epoch": 0.6943591004055548,
      "grad_norm": 0.0625256063533773,
      "learning_rate": 0.00019289897179948474,
      "loss": 0.848,
      "step": 39550
    },
    {
      "epoch": 0.6945346652855563,
      "grad_norm": 0.07113092426009918,
      "learning_rate": 0.00019289482689039696,
      "loss": 0.8514,
      "step": 39560
    },
    {
      "epoch": 0.6947102301655577,
      "grad_norm": 0.06180327669124641,
      "learning_rate": 0.00019289068081698415,
      "loss": 0.8557,
      "step": 39570
    },
    {
      "epoch": 0.6948857950455591,
      "grad_norm": 0.06398428514211486,
      "learning_rate": 0.00019288653357929877,
      "loss": 0.8529,
      "step": 39580
    },
    {
      "epoch": 0.6950613599255605,
      "grad_norm": 0.08926975517855078,
      "learning_rate": 0.0001928823851773934,
      "loss": 0.8507,
      "step": 39590
    },
    {
      "epoch": 0.6952369248055619,
      "grad_norm": 0.04769765364785748,
      "learning_rate": 0.0001928782356113206,
      "loss": 0.854,
      "step": 39600
    },
    {
      "epoch": 0.6954124896855634,
      "grad_norm": 0.05617235317217034,
      "learning_rate": 0.00019287408488113296,
      "loss": 0.8483,
      "step": 39610
    },
    {
      "epoch": 0.6955880545655647,
      "grad_norm": 0.06395089548322581,
      "learning_rate": 0.00019286993298688306,
      "loss": 0.8588,
      "step": 39620
    },
    {
      "epoch": 0.6957636194455661,
      "grad_norm": 0.06991060067644503,
      "learning_rate": 0.0001928657799286235,
      "loss": 0.8559,
      "step": 39630
    },
    {
      "epoch": 0.6959391843255676,
      "grad_norm": 0.07208240719484725,
      "learning_rate": 0.0001928616257064069,
      "loss": 0.8478,
      "step": 39640
    },
    {
      "epoch": 0.6961147492055689,
      "grad_norm": 0.0612707381454146,
      "learning_rate": 0.0001928574703202859,
      "loss": 0.8499,
      "step": 39650
    },
    {
      "epoch": 0.6962903140855703,
      "grad_norm": 0.06850549787667702,
      "learning_rate": 0.00019285331377031314,
      "loss": 0.8474,
      "step": 39660
    },
    {
      "epoch": 0.6964658789655718,
      "grad_norm": 0.057437290469697616,
      "learning_rate": 0.0001928491560565413,
      "loss": 0.8603,
      "step": 39670
    },
    {
      "epoch": 0.6966414438455731,
      "grad_norm": 0.07488015789621516,
      "learning_rate": 0.00019284499717902301,
      "loss": 0.8505,
      "step": 39680
    },
    {
      "epoch": 0.6968170087255745,
      "grad_norm": 0.07511974793172084,
      "learning_rate": 0.00019284083713781103,
      "loss": 0.853,
      "step": 39690
    },
    {
      "epoch": 0.696992573605576,
      "grad_norm": 0.10539632835165216,
      "learning_rate": 0.00019283667593295804,
      "loss": 0.8548,
      "step": 39700
    },
    {
      "epoch": 0.6971681384855773,
      "grad_norm": 0.08259317931424316,
      "learning_rate": 0.0001928325135645168,
      "loss": 0.8574,
      "step": 39710
    },
    {
      "epoch": 0.6973437033655787,
      "grad_norm": 0.05960858783279013,
      "learning_rate": 0.00019282835003253996,
      "loss": 0.8524,
      "step": 39720
    },
    {
      "epoch": 0.6975192682455802,
      "grad_norm": 0.0579568054693564,
      "learning_rate": 0.00019282418533708037,
      "loss": 0.8576,
      "step": 39730
    },
    {
      "epoch": 0.6976948331255816,
      "grad_norm": 0.06745398251273503,
      "learning_rate": 0.00019282001947819072,
      "loss": 0.859,
      "step": 39740
    },
    {
      "epoch": 0.6978703980055829,
      "grad_norm": 0.05242194629657755,
      "learning_rate": 0.00019281585245592387,
      "loss": 0.8665,
      "step": 39750
    },
    {
      "epoch": 0.6980459628855844,
      "grad_norm": 0.0686979600673577,
      "learning_rate": 0.00019281168427033254,
      "loss": 0.8511,
      "step": 39760
    },
    {
      "epoch": 0.6982215277655858,
      "grad_norm": 0.07325314077632923,
      "learning_rate": 0.00019280751492146958,
      "loss": 0.8423,
      "step": 39770
    },
    {
      "epoch": 0.6983970926455871,
      "grad_norm": 0.048946136820700134,
      "learning_rate": 0.0001928033444093878,
      "loss": 0.8529,
      "step": 39780
    },
    {
      "epoch": 0.6985726575255886,
      "grad_norm": 0.061469555929228786,
      "learning_rate": 0.00019279917273414004,
      "loss": 0.8551,
      "step": 39790
    },
    {
      "epoch": 0.69874822240559,
      "grad_norm": 0.09575740518183706,
      "learning_rate": 0.00019279499989577922,
      "loss": 0.8514,
      "step": 39800
    },
    {
      "epoch": 0.6989237872855913,
      "grad_norm": 0.06084184412577095,
      "learning_rate": 0.00019279082589435814,
      "loss": 0.858,
      "step": 39810
    },
    {
      "epoch": 0.6990993521655928,
      "grad_norm": 0.044651984636360866,
      "learning_rate": 0.00019278665072992968,
      "loss": 0.8623,
      "step": 39820
    },
    {
      "epoch": 0.6992749170455942,
      "grad_norm": 0.061281215682015444,
      "learning_rate": 0.00019278247440254677,
      "loss": 0.8467,
      "step": 39830
    },
    {
      "epoch": 0.6994504819255956,
      "grad_norm": 0.05270602271536164,
      "learning_rate": 0.00019277829691226233,
      "loss": 0.8538,
      "step": 39840
    },
    {
      "epoch": 0.699626046805597,
      "grad_norm": 0.0626273299082221,
      "learning_rate": 0.0001927741182591293,
      "loss": 0.8579,
      "step": 39850
    },
    {
      "epoch": 0.6998016116855984,
      "grad_norm": 0.06725882936533198,
      "learning_rate": 0.00019276993844320056,
      "loss": 0.8531,
      "step": 39860
    },
    {
      "epoch": 0.6999771765655998,
      "grad_norm": 0.059288342021241086,
      "learning_rate": 0.00019276575746452913,
      "loss": 0.8531,
      "step": 39870
    },
    {
      "epoch": 0.7001527414456012,
      "grad_norm": 0.06770062616155512,
      "learning_rate": 0.00019276157532316796,
      "loss": 0.8548,
      "step": 39880
    },
    {
      "epoch": 0.7003283063256026,
      "grad_norm": 0.06691647403903675,
      "learning_rate": 0.00019275739201917008,
      "loss": 0.8491,
      "step": 39890
    },
    {
      "epoch": 0.700503871205604,
      "grad_norm": 0.08820102267375056,
      "learning_rate": 0.0001927532075525884,
      "loss": 0.8547,
      "step": 39900
    },
    {
      "epoch": 0.7006794360856055,
      "grad_norm": 0.0786645396382141,
      "learning_rate": 0.00019274902192347606,
      "loss": 0.8528,
      "step": 39910
    },
    {
      "epoch": 0.7008550009656068,
      "grad_norm": 0.09000650980579282,
      "learning_rate": 0.000192744835131886,
      "loss": 0.8581,
      "step": 39920
    },
    {
      "epoch": 0.7010305658456083,
      "grad_norm": 0.062014816345771995,
      "learning_rate": 0.0001927406471778713,
      "loss": 0.8581,
      "step": 39930
    },
    {
      "epoch": 0.7012061307256097,
      "grad_norm": 0.08678221446084688,
      "learning_rate": 0.000192736458061485,
      "loss": 0.8553,
      "step": 39940
    },
    {
      "epoch": 0.701381695605611,
      "grad_norm": 0.05676554448753186,
      "learning_rate": 0.00019273226778278023,
      "loss": 0.8484,
      "step": 39950
    },
    {
      "epoch": 0.7015572604856125,
      "grad_norm": 0.05139759513784446,
      "learning_rate": 0.00019272807634181,
      "loss": 0.8612,
      "step": 39960
    },
    {
      "epoch": 0.7017328253656139,
      "grad_norm": 0.06034016056200693,
      "learning_rate": 0.00019272388373862748,
      "loss": 0.8588,
      "step": 39970
    },
    {
      "epoch": 0.7019083902456152,
      "grad_norm": 0.057607044031957794,
      "learning_rate": 0.00019271968997328578,
      "loss": 0.8536,
      "step": 39980
    },
    {
      "epoch": 0.7020839551256167,
      "grad_norm": 0.05030933331519665,
      "learning_rate": 0.00019271549504583802,
      "loss": 0.851,
      "step": 39990
    },
    {
      "epoch": 0.7022595200056181,
      "grad_norm": 0.07551186974064805,
      "learning_rate": 0.0001927112989563374,
      "loss": 0.8523,
      "step": 40000
    },
    {
      "epoch": 0.7024350848856195,
      "grad_norm": 0.06826745389743773,
      "learning_rate": 0.00019270710170483699,
      "loss": 0.8498,
      "step": 40010
    },
    {
      "epoch": 0.7026106497656209,
      "grad_norm": 0.0673760849795274,
      "learning_rate": 0.00019270290329139006,
      "loss": 0.8529,
      "step": 40020
    },
    {
      "epoch": 0.7027862146456223,
      "grad_norm": 0.07066615447766973,
      "learning_rate": 0.00019269870371604975,
      "loss": 0.8526,
      "step": 40030
    },
    {
      "epoch": 0.7029617795256237,
      "grad_norm": 0.06432651616368167,
      "learning_rate": 0.00019269450297886934,
      "loss": 0.8541,
      "step": 40040
    },
    {
      "epoch": 0.7031373444056251,
      "grad_norm": 0.06750360573297502,
      "learning_rate": 0.00019269030107990195,
      "loss": 0.8524,
      "step": 40050
    },
    {
      "epoch": 0.7033129092856265,
      "grad_norm": 0.08005612627467208,
      "learning_rate": 0.00019268609801920087,
      "loss": 0.8548,
      "step": 40060
    },
    {
      "epoch": 0.7034884741656279,
      "grad_norm": 0.0929884158060586,
      "learning_rate": 0.00019268189379681937,
      "loss": 0.8573,
      "step": 40070
    },
    {
      "epoch": 0.7036640390456294,
      "grad_norm": 0.06392614790820661,
      "learning_rate": 0.0001926776884128107,
      "loss": 0.8528,
      "step": 40080
    },
    {
      "epoch": 0.7038396039256307,
      "grad_norm": 0.08612564130215204,
      "learning_rate": 0.00019267348186722817,
      "loss": 0.8552,
      "step": 40090
    },
    {
      "epoch": 0.7040151688056321,
      "grad_norm": 0.07212923760919795,
      "learning_rate": 0.00019266927416012504,
      "loss": 0.8483,
      "step": 40100
    },
    {
      "epoch": 0.7041907336856336,
      "grad_norm": 0.05629359456703132,
      "learning_rate": 0.00019266506529155465,
      "loss": 0.8537,
      "step": 40110
    },
    {
      "epoch": 0.7043662985656349,
      "grad_norm": 0.08417697963957782,
      "learning_rate": 0.00019266085526157032,
      "loss": 0.8573,
      "step": 40120
    },
    {
      "epoch": 0.7045418634456363,
      "grad_norm": 0.07507479331321099,
      "learning_rate": 0.00019265664407022538,
      "loss": 0.8575,
      "step": 40130
    },
    {
      "epoch": 0.7047174283256378,
      "grad_norm": 0.05789201693726613,
      "learning_rate": 0.00019265243171757317,
      "loss": 0.8566,
      "step": 40140
    },
    {
      "epoch": 0.7048929932056391,
      "grad_norm": 0.06895235364978505,
      "learning_rate": 0.00019264821820366712,
      "loss": 0.848,
      "step": 40150
    },
    {
      "epoch": 0.7050685580856405,
      "grad_norm": 0.07878223614944117,
      "learning_rate": 0.00019264400352856056,
      "loss": 0.856,
      "step": 40160
    },
    {
      "epoch": 0.705244122965642,
      "grad_norm": 0.08936892313237853,
      "learning_rate": 0.0001926397876923069,
      "loss": 0.8591,
      "step": 40170
    },
    {
      "epoch": 0.7054196878456434,
      "grad_norm": 0.07582258065535043,
      "learning_rate": 0.00019263557069495958,
      "loss": 0.8625,
      "step": 40180
    },
    {
      "epoch": 0.7055952527256447,
      "grad_norm": 0.056070430770379445,
      "learning_rate": 0.00019263135253657206,
      "loss": 0.8545,
      "step": 40190
    },
    {
      "epoch": 0.7057708176056462,
      "grad_norm": 0.050227775411311015,
      "learning_rate": 0.00019262713321719767,
      "loss": 0.8532,
      "step": 40200
    },
    {
      "epoch": 0.7059463824856476,
      "grad_norm": 0.06359920532283089,
      "learning_rate": 0.00019262291273688996,
      "loss": 0.8475,
      "step": 40210
    },
    {
      "epoch": 0.7061219473656489,
      "grad_norm": 0.05808075924442314,
      "learning_rate": 0.00019261869109570243,
      "loss": 0.855,
      "step": 40220
    },
    {
      "epoch": 0.7062975122456504,
      "grad_norm": 0.07049556074470041,
      "learning_rate": 0.00019261446829368846,
      "loss": 0.8569,
      "step": 40230
    },
    {
      "epoch": 0.7064730771256518,
      "grad_norm": 0.0647314198312937,
      "learning_rate": 0.0001926102443309017,
      "loss": 0.8573,
      "step": 40240
    },
    {
      "epoch": 0.7066486420056531,
      "grad_norm": 0.07371664519931076,
      "learning_rate": 0.00019260601920739553,
      "loss": 0.8628,
      "step": 40250
    },
    {
      "epoch": 0.7068242068856546,
      "grad_norm": 0.056440609135191624,
      "learning_rate": 0.00019260179292322356,
      "loss": 0.8604,
      "step": 40260
    },
    {
      "epoch": 0.706999771765656,
      "grad_norm": 0.0547099350287864,
      "learning_rate": 0.00019259756547843927,
      "loss": 0.8625,
      "step": 40270
    },
    {
      "epoch": 0.7071753366456573,
      "grad_norm": 0.06745431809649541,
      "learning_rate": 0.00019259333687309636,
      "loss": 0.8555,
      "step": 40280
    },
    {
      "epoch": 0.7073509015256588,
      "grad_norm": 0.11555382225101422,
      "learning_rate": 0.00019258910710724824,
      "loss": 0.8495,
      "step": 40290
    },
    {
      "epoch": 0.7075264664056602,
      "grad_norm": 0.060307670027768136,
      "learning_rate": 0.00019258487618094863,
      "loss": 0.8572,
      "step": 40300
    },
    {
      "epoch": 0.7077020312856617,
      "grad_norm": 0.06341198875050888,
      "learning_rate": 0.00019258064409425108,
      "loss": 0.8619,
      "step": 40310
    },
    {
      "epoch": 0.707877596165663,
      "grad_norm": 0.07470955056721337,
      "learning_rate": 0.0001925764108472092,
      "loss": 0.846,
      "step": 40320
    },
    {
      "epoch": 0.7080531610456644,
      "grad_norm": 0.062031412084366175,
      "learning_rate": 0.00019257217643987664,
      "loss": 0.8561,
      "step": 40330
    },
    {
      "epoch": 0.7082287259256659,
      "grad_norm": 0.05020595433020414,
      "learning_rate": 0.00019256794087230708,
      "loss": 0.8509,
      "step": 40340
    },
    {
      "epoch": 0.7084042908056672,
      "grad_norm": 0.056798834610652216,
      "learning_rate": 0.0001925637041445541,
      "loss": 0.854,
      "step": 40350
    },
    {
      "epoch": 0.7085798556856686,
      "grad_norm": 0.05160124986006178,
      "learning_rate": 0.0001925594662566715,
      "loss": 0.8539,
      "step": 40360
    },
    {
      "epoch": 0.7087554205656701,
      "grad_norm": 0.06313353980191602,
      "learning_rate": 0.0001925552272087129,
      "loss": 0.8524,
      "step": 40370
    },
    {
      "epoch": 0.7089309854456715,
      "grad_norm": 0.06994214543913452,
      "learning_rate": 0.00019255098700073203,
      "loss": 0.8502,
      "step": 40380
    },
    {
      "epoch": 0.7091065503256728,
      "grad_norm": 0.08532218635058296,
      "learning_rate": 0.00019254674563278258,
      "loss": 0.8554,
      "step": 40390
    },
    {
      "epoch": 0.7092821152056743,
      "grad_norm": 0.06381129781994219,
      "learning_rate": 0.00019254250310491838,
      "loss": 0.8567,
      "step": 40400
    },
    {
      "epoch": 0.7094576800856757,
      "grad_norm": 0.08768667690618347,
      "learning_rate": 0.00019253825941719307,
      "loss": 0.8529,
      "step": 40410
    },
    {
      "epoch": 0.709633244965677,
      "grad_norm": 0.06199940668579513,
      "learning_rate": 0.00019253401456966046,
      "loss": 0.8425,
      "step": 40420
    },
    {
      "epoch": 0.7098088098456785,
      "grad_norm": 0.07241266129129936,
      "learning_rate": 0.00019252976856237437,
      "loss": 0.8552,
      "step": 40430
    },
    {
      "epoch": 0.7099843747256799,
      "grad_norm": 0.05419095243465862,
      "learning_rate": 0.0001925255213953886,
      "loss": 0.8614,
      "step": 40440
    },
    {
      "epoch": 0.7101599396056812,
      "grad_norm": 0.07158035741887907,
      "learning_rate": 0.0001925212730687569,
      "loss": 0.8544,
      "step": 40450
    },
    {
      "epoch": 0.7103355044856827,
      "grad_norm": 0.06531853207314471,
      "learning_rate": 0.00019251702358253313,
      "loss": 0.8512,
      "step": 40460
    },
    {
      "epoch": 0.7105110693656841,
      "grad_norm": 0.06333540304633178,
      "learning_rate": 0.00019251277293677114,
      "loss": 0.856,
      "step": 40470
    },
    {
      "epoch": 0.7106866342456855,
      "grad_norm": 0.0917594856457596,
      "learning_rate": 0.00019250852113152474,
      "loss": 0.8567,
      "step": 40480
    },
    {
      "epoch": 0.7108621991256869,
      "grad_norm": 0.061034748858780415,
      "learning_rate": 0.00019250426816684788,
      "loss": 0.8446,
      "step": 40490
    },
    {
      "epoch": 0.7110377640056883,
      "grad_norm": 0.07071334088181173,
      "learning_rate": 0.0001925000140427944,
      "loss": 0.8589,
      "step": 40500
    },
    {
      "epoch": 0.7112133288856897,
      "grad_norm": 0.0835492689684211,
      "learning_rate": 0.00019249575875941822,
      "loss": 0.8462,
      "step": 40510
    },
    {
      "epoch": 0.7113888937656911,
      "grad_norm": 0.07838018562660146,
      "learning_rate": 0.0001924915023167732,
      "loss": 0.8528,
      "step": 40520
    },
    {
      "epoch": 0.7115644586456925,
      "grad_norm": 0.08609882896090715,
      "learning_rate": 0.00019248724471491332,
      "loss": 0.8494,
      "step": 40530
    },
    {
      "epoch": 0.7117400235256939,
      "grad_norm": 0.04554676864957363,
      "learning_rate": 0.00019248298595389254,
      "loss": 0.8571,
      "step": 40540
    },
    {
      "epoch": 0.7119155884056954,
      "grad_norm": 0.06619695297987577,
      "learning_rate": 0.0001924787260337648,
      "loss": 0.8491,
      "step": 40550
    },
    {
      "epoch": 0.7120911532856967,
      "grad_norm": 0.0710183811152742,
      "learning_rate": 0.00019247446495458404,
      "loss": 0.8494,
      "step": 40560
    },
    {
      "epoch": 0.7122667181656981,
      "grad_norm": 0.063792567317728,
      "learning_rate": 0.0001924702027164043,
      "loss": 0.8583,
      "step": 40570
    },
    {
      "epoch": 0.7124422830456996,
      "grad_norm": 0.06208690367423301,
      "learning_rate": 0.00019246593931927954,
      "loss": 0.8594,
      "step": 40580
    },
    {
      "epoch": 0.7126178479257009,
      "grad_norm": 0.07543496911794001,
      "learning_rate": 0.00019246167476326382,
      "loss": 0.8567,
      "step": 40590
    },
    {
      "epoch": 0.7127934128057023,
      "grad_norm": 0.05351122906322302,
      "learning_rate": 0.00019245740904841115,
      "loss": 0.8578,
      "step": 40600
    },
    {
      "epoch": 0.7129689776857038,
      "grad_norm": 0.08848229452504475,
      "learning_rate": 0.00019245314217477557,
      "loss": 0.8613,
      "step": 40610
    },
    {
      "epoch": 0.7131445425657051,
      "grad_norm": 0.05357183785745211,
      "learning_rate": 0.00019244887414241116,
      "loss": 0.8477,
      "step": 40620
    },
    {
      "epoch": 0.7133201074457065,
      "grad_norm": 0.07150899206688024,
      "learning_rate": 0.00019244460495137199,
      "loss": 0.8555,
      "step": 40630
    },
    {
      "epoch": 0.713495672325708,
      "grad_norm": 0.05604692932117298,
      "learning_rate": 0.00019244033460171214,
      "loss": 0.8466,
      "step": 40640
    },
    {
      "epoch": 0.7136712372057094,
      "grad_norm": 0.07155515810246776,
      "learning_rate": 0.00019243606309348577,
      "loss": 0.8503,
      "step": 40650
    },
    {
      "epoch": 0.7138468020857108,
      "grad_norm": 0.0598349392503002,
      "learning_rate": 0.0001924317904267469,
      "loss": 0.8503,
      "step": 40660
    },
    {
      "epoch": 0.7140223669657122,
      "grad_norm": 0.06432666955176702,
      "learning_rate": 0.0001924275166015497,
      "loss": 0.852,
      "step": 40670
    },
    {
      "epoch": 0.7141979318457136,
      "grad_norm": 0.057687519786230414,
      "learning_rate": 0.0001924232416179484,
      "loss": 0.8589,
      "step": 40680
    },
    {
      "epoch": 0.714373496725715,
      "grad_norm": 0.07593613338250556,
      "learning_rate": 0.0001924189654759971,
      "loss": 0.8557,
      "step": 40690
    },
    {
      "epoch": 0.7145490616057164,
      "grad_norm": 0.07712329896602356,
      "learning_rate": 0.00019241468817574996,
      "loss": 0.8536,
      "step": 40700
    },
    {
      "epoch": 0.7147246264857178,
      "grad_norm": 0.06960471940631713,
      "learning_rate": 0.00019241040971726124,
      "loss": 0.8495,
      "step": 40710
    },
    {
      "epoch": 0.7149001913657193,
      "grad_norm": 0.07526239216597744,
      "learning_rate": 0.0001924061301005851,
      "loss": 0.8497,
      "step": 40720
    },
    {
      "epoch": 0.7150757562457206,
      "grad_norm": 0.06826442429987162,
      "learning_rate": 0.00019240184932577576,
      "loss": 0.8549,
      "step": 40730
    },
    {
      "epoch": 0.715251321125722,
      "grad_norm": 0.09228528575022618,
      "learning_rate": 0.00019239756739288747,
      "loss": 0.8491,
      "step": 40740
    },
    {
      "epoch": 0.7154268860057235,
      "grad_norm": 0.059673747044249506,
      "learning_rate": 0.00019239328430197453,
      "loss": 0.8515,
      "step": 40750
    },
    {
      "epoch": 0.7156024508857248,
      "grad_norm": 0.10302134193070571,
      "learning_rate": 0.00019238900005309108,
      "loss": 0.8573,
      "step": 40760
    },
    {
      "epoch": 0.7157780157657262,
      "grad_norm": 0.06531163917460162,
      "learning_rate": 0.00019238471464629153,
      "loss": 0.8586,
      "step": 40770
    },
    {
      "epoch": 0.7159535806457277,
      "grad_norm": 0.05003231203496736,
      "learning_rate": 0.00019238042808163015,
      "loss": 0.8616,
      "step": 40780
    },
    {
      "epoch": 0.716129145525729,
      "grad_norm": 0.049590707256350126,
      "learning_rate": 0.00019237614035916119,
      "loss": 0.8572,
      "step": 40790
    },
    {
      "epoch": 0.7163047104057304,
      "grad_norm": 0.07198723891074335,
      "learning_rate": 0.00019237185147893905,
      "loss": 0.853,
      "step": 40800
    },
    {
      "epoch": 0.7164802752857319,
      "grad_norm": 0.05901154553512088,
      "learning_rate": 0.00019236756144101804,
      "loss": 0.8504,
      "step": 40810
    },
    {
      "epoch": 0.7166558401657332,
      "grad_norm": 0.06520520898641255,
      "learning_rate": 0.0001923632702454525,
      "loss": 0.8557,
      "step": 40820
    },
    {
      "epoch": 0.7168314050457346,
      "grad_norm": 0.09378214467176965,
      "learning_rate": 0.00019235897789229682,
      "loss": 0.8527,
      "step": 40830
    },
    {
      "epoch": 0.7170069699257361,
      "grad_norm": 0.06945690130699264,
      "learning_rate": 0.0001923546843816054,
      "loss": 0.8543,
      "step": 40840
    },
    {
      "epoch": 0.7171825348057375,
      "grad_norm": 0.07247302897478629,
      "learning_rate": 0.0001923503897134326,
      "loss": 0.8492,
      "step": 40850
    },
    {
      "epoch": 0.7173580996857388,
      "grad_norm": 0.06530772636123368,
      "learning_rate": 0.00019234609388783287,
      "loss": 0.8479,
      "step": 40860
    },
    {
      "epoch": 0.7175336645657403,
      "grad_norm": 0.055463666140670026,
      "learning_rate": 0.00019234179690486063,
      "loss": 0.8571,
      "step": 40870
    },
    {
      "epoch": 0.7177092294457417,
      "grad_norm": 0.06815727315668724,
      "learning_rate": 0.0001923374987645703,
      "loss": 0.8595,
      "step": 40880
    },
    {
      "epoch": 0.717884794325743,
      "grad_norm": 0.07188126822142968,
      "learning_rate": 0.00019233319946701638,
      "loss": 0.8629,
      "step": 40890
    },
    {
      "epoch": 0.7180603592057445,
      "grad_norm": 0.08179265584577546,
      "learning_rate": 0.00019232889901225325,
      "loss": 0.8444,
      "step": 40900
    },
    {
      "epoch": 0.7182359240857459,
      "grad_norm": 0.056096007811202275,
      "learning_rate": 0.00019232459740033554,
      "loss": 0.8606,
      "step": 40910
    },
    {
      "epoch": 0.7184114889657472,
      "grad_norm": 0.07064148818590282,
      "learning_rate": 0.00019232029463131766,
      "loss": 0.8517,
      "step": 40920
    },
    {
      "epoch": 0.7185870538457487,
      "grad_norm": 0.062401529905027814,
      "learning_rate": 0.00019231599070525414,
      "loss": 0.8565,
      "step": 40930
    },
    {
      "epoch": 0.7187626187257501,
      "grad_norm": 0.11392660224733912,
      "learning_rate": 0.0001923116856221995,
      "loss": 0.858,
      "step": 40940
    },
    {
      "epoch": 0.7189381836057515,
      "grad_norm": 0.0650536833659954,
      "learning_rate": 0.00019230737938220833,
      "loss": 0.8576,
      "step": 40950
    },
    {
      "epoch": 0.7191137484857529,
      "grad_norm": 0.07877526642704848,
      "learning_rate": 0.00019230307198533518,
      "loss": 0.8559,
      "step": 40960
    },
    {
      "epoch": 0.7192893133657543,
      "grad_norm": 0.08064863234330585,
      "learning_rate": 0.00019229876343163457,
      "loss": 0.8567,
      "step": 40970
    },
    {
      "epoch": 0.7194648782457557,
      "grad_norm": 0.0716722190696521,
      "learning_rate": 0.00019229445372116112,
      "loss": 0.8481,
      "step": 40980
    },
    {
      "epoch": 0.7196404431257571,
      "grad_norm": 0.05217606874592304,
      "learning_rate": 0.0001922901428539695,
      "loss": 0.8601,
      "step": 40990
    },
    {
      "epoch": 0.7198160080057585,
      "grad_norm": 0.08972730239930479,
      "learning_rate": 0.00019228583083011423,
      "loss": 0.8574,
      "step": 41000
    },
    {
      "epoch": 0.7199915728857599,
      "grad_norm": 0.055185017664135735,
      "learning_rate": 0.00019228151764965,
      "loss": 0.8563,
      "step": 41010
    },
    {
      "epoch": 0.7201671377657614,
      "grad_norm": 0.0642685641758122,
      "learning_rate": 0.00019227720331263146,
      "loss": 0.8543,
      "step": 41020
    },
    {
      "epoch": 0.7203427026457627,
      "grad_norm": 0.07126838313131871,
      "learning_rate": 0.00019227288781911323,
      "loss": 0.8576,
      "step": 41030
    },
    {
      "epoch": 0.7205182675257642,
      "grad_norm": 0.06266225332822965,
      "learning_rate": 0.00019226857116915006,
      "loss": 0.855,
      "step": 41040
    },
    {
      "epoch": 0.7206938324057656,
      "grad_norm": 0.05051464356879038,
      "learning_rate": 0.00019226425336279658,
      "loss": 0.86,
      "step": 41050
    },
    {
      "epoch": 0.7208693972857669,
      "grad_norm": 0.06850049425284685,
      "learning_rate": 0.0001922599344001075,
      "loss": 0.8493,
      "step": 41060
    },
    {
      "epoch": 0.7210449621657684,
      "grad_norm": 0.05414267756963482,
      "learning_rate": 0.00019225561428113758,
      "loss": 0.852,
      "step": 41070
    },
    {
      "epoch": 0.7212205270457698,
      "grad_norm": 0.09544593383682366,
      "learning_rate": 0.00019225129300594156,
      "loss": 0.8558,
      "step": 41080
    },
    {
      "epoch": 0.7213960919257711,
      "grad_norm": 0.06333672098790791,
      "learning_rate": 0.00019224697057457414,
      "loss": 0.8552,
      "step": 41090
    },
    {
      "epoch": 0.7215716568057726,
      "grad_norm": 0.0771624506047455,
      "learning_rate": 0.00019224264698709012,
      "loss": 0.8623,
      "step": 41100
    },
    {
      "epoch": 0.721747221685774,
      "grad_norm": 0.0735347773793183,
      "learning_rate": 0.00019223832224354428,
      "loss": 0.8481,
      "step": 41110
    },
    {
      "epoch": 0.7219227865657754,
      "grad_norm": 0.07847556874922292,
      "learning_rate": 0.00019223399634399143,
      "loss": 0.8569,
      "step": 41120
    },
    {
      "epoch": 0.7220983514457768,
      "grad_norm": 0.05263539827215551,
      "learning_rate": 0.00019222966928848635,
      "loss": 0.8531,
      "step": 41130
    },
    {
      "epoch": 0.7222739163257782,
      "grad_norm": 0.07260781811270274,
      "learning_rate": 0.00019222534107708386,
      "loss": 0.8536,
      "step": 41140
    },
    {
      "epoch": 0.7224494812057796,
      "grad_norm": 0.057075316471987825,
      "learning_rate": 0.00019222101170983882,
      "loss": 0.8535,
      "step": 41150
    },
    {
      "epoch": 0.722625046085781,
      "grad_norm": 0.07235300750646124,
      "learning_rate": 0.0001922166811868061,
      "loss": 0.8577,
      "step": 41160
    },
    {
      "epoch": 0.7228006109657824,
      "grad_norm": 0.055625380845334686,
      "learning_rate": 0.0001922123495080405,
      "loss": 0.8577,
      "step": 41170
    },
    {
      "epoch": 0.7229761758457838,
      "grad_norm": 0.062913286099711,
      "learning_rate": 0.000192208016673597,
      "loss": 0.8559,
      "step": 41180
    },
    {
      "epoch": 0.7231517407257853,
      "grad_norm": 0.08174737430690163,
      "learning_rate": 0.00019220368268353042,
      "loss": 0.846,
      "step": 41190
    },
    {
      "epoch": 0.7233273056057866,
      "grad_norm": 0.0478226750514807,
      "learning_rate": 0.0001921993475378957,
      "loss": 0.8535,
      "step": 41200
    },
    {
      "epoch": 0.723502870485788,
      "grad_norm": 0.06437284047010983,
      "learning_rate": 0.0001921950112367478,
      "loss": 0.8538,
      "step": 41210
    },
    {
      "epoch": 0.7236784353657895,
      "grad_norm": 0.07098144902359291,
      "learning_rate": 0.00019219067378014159,
      "loss": 0.8572,
      "step": 41220
    },
    {
      "epoch": 0.7238540002457908,
      "grad_norm": 0.09159410948177986,
      "learning_rate": 0.0001921863351681321,
      "loss": 0.8501,
      "step": 41230
    },
    {
      "epoch": 0.7240295651257922,
      "grad_norm": 0.0732180444619823,
      "learning_rate": 0.00019218199540077424,
      "loss": 0.8488,
      "step": 41240
    },
    {
      "epoch": 0.7242051300057937,
      "grad_norm": 0.06961050928183718,
      "learning_rate": 0.00019217765447812306,
      "loss": 0.8545,
      "step": 41250
    },
    {
      "epoch": 0.724380694885795,
      "grad_norm": 0.05305235105744925,
      "learning_rate": 0.0001921733124002335,
      "loss": 0.8592,
      "step": 41260
    },
    {
      "epoch": 0.7245562597657964,
      "grad_norm": 0.06442110220951865,
      "learning_rate": 0.00019216896916716058,
      "loss": 0.8618,
      "step": 41270
    },
    {
      "epoch": 0.7247318246457979,
      "grad_norm": 0.06681188729386295,
      "learning_rate": 0.0001921646247789594,
      "loss": 0.8425,
      "step": 41280
    },
    {
      "epoch": 0.7249073895257993,
      "grad_norm": 0.0988456661833109,
      "learning_rate": 0.00019216027923568493,
      "loss": 0.8596,
      "step": 41290
    },
    {
      "epoch": 0.7250829544058006,
      "grad_norm": 0.053590953122201794,
      "learning_rate": 0.0001921559325373922,
      "loss": 0.8567,
      "step": 41300
    },
    {
      "epoch": 0.7252585192858021,
      "grad_norm": 0.07281803549322106,
      "learning_rate": 0.0001921515846841364,
      "loss": 0.8586,
      "step": 41310
    },
    {
      "epoch": 0.7254340841658035,
      "grad_norm": 0.08233978364631804,
      "learning_rate": 0.00019214723567597254,
      "loss": 0.8605,
      "step": 41320
    },
    {
      "epoch": 0.7256096490458048,
      "grad_norm": 0.055531399492499973,
      "learning_rate": 0.0001921428855129557,
      "loss": 0.8605,
      "step": 41330
    },
    {
      "epoch": 0.7257852139258063,
      "grad_norm": 0.07327506071356443,
      "learning_rate": 0.0001921385341951411,
      "loss": 0.8567,
      "step": 41340
    },
    {
      "epoch": 0.7259607788058077,
      "grad_norm": 0.07579730737062462,
      "learning_rate": 0.00019213418172258373,
      "loss": 0.8542,
      "step": 41350
    },
    {
      "epoch": 0.726136343685809,
      "grad_norm": 0.059236666219940504,
      "learning_rate": 0.00019212982809533886,
      "loss": 0.8488,
      "step": 41360
    },
    {
      "epoch": 0.7263119085658105,
      "grad_norm": 0.07074983006810254,
      "learning_rate": 0.00019212547331346158,
      "loss": 0.8611,
      "step": 41370
    },
    {
      "epoch": 0.7264874734458119,
      "grad_norm": 0.0614739908076061,
      "learning_rate": 0.0001921211173770071,
      "loss": 0.8484,
      "step": 41380
    },
    {
      "epoch": 0.7266630383258134,
      "grad_norm": 0.07467274784910788,
      "learning_rate": 0.00019211676028603057,
      "loss": 0.8515,
      "step": 41390
    },
    {
      "epoch": 0.7268386032058147,
      "grad_norm": 0.06102583700383781,
      "learning_rate": 0.0001921124020405873,
      "loss": 0.8566,
      "step": 41400
    },
    {
      "epoch": 0.7270141680858161,
      "grad_norm": 0.05548945233312262,
      "learning_rate": 0.00019210804264073236,
      "loss": 0.8489,
      "step": 41410
    },
    {
      "epoch": 0.7271897329658176,
      "grad_norm": 0.06884314092630858,
      "learning_rate": 0.0001921036820865211,
      "loss": 0.8539,
      "step": 41420
    },
    {
      "epoch": 0.7273652978458189,
      "grad_norm": 0.06681588307556562,
      "learning_rate": 0.00019209932037800873,
      "loss": 0.8409,
      "step": 41430
    },
    {
      "epoch": 0.7275408627258203,
      "grad_norm": 0.07018868445921589,
      "learning_rate": 0.00019209495751525049,
      "loss": 0.8511,
      "step": 41440
    },
    {
      "epoch": 0.7277164276058218,
      "grad_norm": 0.08883011130974354,
      "learning_rate": 0.00019209059349830168,
      "loss": 0.8498,
      "step": 41450
    },
    {
      "epoch": 0.7278919924858231,
      "grad_norm": 0.05860884524557781,
      "learning_rate": 0.0001920862283272176,
      "loss": 0.8467,
      "step": 41460
    },
    {
      "epoch": 0.7280675573658245,
      "grad_norm": 0.062154192089645,
      "learning_rate": 0.00019208186200205354,
      "loss": 0.8534,
      "step": 41470
    },
    {
      "epoch": 0.728243122245826,
      "grad_norm": 0.06918745644328879,
      "learning_rate": 0.00019207749452286485,
      "loss": 0.8536,
      "step": 41480
    },
    {
      "epoch": 0.7284186871258274,
      "grad_norm": 0.06987809098147872,
      "learning_rate": 0.00019207312588970682,
      "loss": 0.8597,
      "step": 41490
    },
    {
      "epoch": 0.7285942520058287,
      "grad_norm": 0.06648024988295402,
      "learning_rate": 0.00019206875610263487,
      "loss": 0.8563,
      "step": 41500
    },
    {
      "epoch": 0.7287698168858302,
      "grad_norm": 0.06054428865001836,
      "learning_rate": 0.0001920643851617043,
      "loss": 0.8532,
      "step": 41510
    },
    {
      "epoch": 0.7289453817658316,
      "grad_norm": 0.08572841857322734,
      "learning_rate": 0.00019206001306697055,
      "loss": 0.8492,
      "step": 41520
    },
    {
      "epoch": 0.7291209466458329,
      "grad_norm": 0.06637976285418597,
      "learning_rate": 0.00019205563981848894,
      "loss": 0.8484,
      "step": 41530
    },
    {
      "epoch": 0.7292965115258344,
      "grad_norm": 0.08718326242308899,
      "learning_rate": 0.00019205126541631495,
      "loss": 0.8542,
      "step": 41540
    },
    {
      "epoch": 0.7294720764058358,
      "grad_norm": 0.06533664097368844,
      "learning_rate": 0.00019204688986050397,
      "loss": 0.8467,
      "step": 41550
    },
    {
      "epoch": 0.7296476412858371,
      "grad_norm": 0.06692804737566767,
      "learning_rate": 0.00019204251315111144,
      "loss": 0.8493,
      "step": 41560
    },
    {
      "epoch": 0.7298232061658386,
      "grad_norm": 0.06669041222204283,
      "learning_rate": 0.00019203813528819282,
      "loss": 0.8515,
      "step": 41570
    },
    {
      "epoch": 0.72999877104584,
      "grad_norm": 0.0667100629163934,
      "learning_rate": 0.00019203375627180357,
      "loss": 0.8568,
      "step": 41580
    },
    {
      "epoch": 0.7301743359258414,
      "grad_norm": 0.07918849344008952,
      "learning_rate": 0.00019202937610199924,
      "loss": 0.8514,
      "step": 41590
    },
    {
      "epoch": 0.7303499008058428,
      "grad_norm": 0.08090038972082882,
      "learning_rate": 0.0001920249947788352,
      "loss": 0.8575,
      "step": 41600
    },
    {
      "epoch": 0.7305254656858442,
      "grad_norm": 0.05902481529888745,
      "learning_rate": 0.00019202061230236706,
      "loss": 0.8541,
      "step": 41610
    },
    {
      "epoch": 0.7307010305658456,
      "grad_norm": 0.06554218887731347,
      "learning_rate": 0.00019201622867265033,
      "loss": 0.8432,
      "step": 41620
    },
    {
      "epoch": 0.730876595445847,
      "grad_norm": 0.05109700214498215,
      "learning_rate": 0.0001920118438897405,
      "loss": 0.8579,
      "step": 41630
    },
    {
      "epoch": 0.7310521603258484,
      "grad_norm": 0.0870847113093831,
      "learning_rate": 0.0001920074579536932,
      "loss": 0.855,
      "step": 41640
    },
    {
      "epoch": 0.7312277252058498,
      "grad_norm": 0.06672960728942325,
      "learning_rate": 0.00019200307086456392,
      "loss": 0.8517,
      "step": 41650
    },
    {
      "epoch": 0.7314032900858513,
      "grad_norm": 0.06963570618095959,
      "learning_rate": 0.00019199868262240836,
      "loss": 0.8569,
      "step": 41660
    },
    {
      "epoch": 0.7315788549658526,
      "grad_norm": 0.0604864717252325,
      "learning_rate": 0.00019199429322728202,
      "loss": 0.8529,
      "step": 41670
    },
    {
      "epoch": 0.731754419845854,
      "grad_norm": 0.07074809218564695,
      "learning_rate": 0.00019198990267924054,
      "loss": 0.8522,
      "step": 41680
    },
    {
      "epoch": 0.7319299847258555,
      "grad_norm": 0.07741907626664224,
      "learning_rate": 0.00019198551097833954,
      "loss": 0.8553,
      "step": 41690
    },
    {
      "epoch": 0.7321055496058568,
      "grad_norm": 0.07156291399585768,
      "learning_rate": 0.0001919811181246347,
      "loss": 0.8523,
      "step": 41700
    },
    {
      "epoch": 0.7322811144858582,
      "grad_norm": 0.05199088369172314,
      "learning_rate": 0.00019197672411818168,
      "loss": 0.8517,
      "step": 41710
    },
    {
      "epoch": 0.7324566793658597,
      "grad_norm": 0.07410704115284417,
      "learning_rate": 0.0001919723289590361,
      "loss": 0.8513,
      "step": 41720
    },
    {
      "epoch": 0.732632244245861,
      "grad_norm": 0.06347352899849315,
      "learning_rate": 0.0001919679326472537,
      "loss": 0.8587,
      "step": 41730
    },
    {
      "epoch": 0.7328078091258624,
      "grad_norm": 0.05807932068617638,
      "learning_rate": 0.00019196353518289013,
      "loss": 0.8533,
      "step": 41740
    },
    {
      "epoch": 0.7329833740058639,
      "grad_norm": 0.0706125815882134,
      "learning_rate": 0.00019195913656600114,
      "loss": 0.8676,
      "step": 41750
    },
    {
      "epoch": 0.7331589388858653,
      "grad_norm": 0.054199879633621544,
      "learning_rate": 0.00019195473679664245,
      "loss": 0.855,
      "step": 41760
    },
    {
      "epoch": 0.7333345037658667,
      "grad_norm": 0.07083921303970052,
      "learning_rate": 0.00019195033587486986,
      "loss": 0.856,
      "step": 41770
    },
    {
      "epoch": 0.7335100686458681,
      "grad_norm": 0.06807940444850266,
      "learning_rate": 0.00019194593380073905,
      "loss": 0.8524,
      "step": 41780
    },
    {
      "epoch": 0.7336856335258695,
      "grad_norm": 0.06582011766936863,
      "learning_rate": 0.00019194153057430584,
      "loss": 0.8515,
      "step": 41790
    },
    {
      "epoch": 0.7338611984058709,
      "grad_norm": 0.06458613831609501,
      "learning_rate": 0.00019193712619562599,
      "loss": 0.8534,
      "step": 41800
    },
    {
      "epoch": 0.7340367632858723,
      "grad_norm": 0.0710525368071661,
      "learning_rate": 0.00019193272066475535,
      "loss": 0.8547,
      "step": 41810
    },
    {
      "epoch": 0.7342123281658737,
      "grad_norm": 0.06533630147388196,
      "learning_rate": 0.00019192831398174965,
      "loss": 0.8525,
      "step": 41820
    },
    {
      "epoch": 0.7343878930458752,
      "grad_norm": 0.05766203568208778,
      "learning_rate": 0.00019192390614666485,
      "loss": 0.8569,
      "step": 41830
    },
    {
      "epoch": 0.7345634579258765,
      "grad_norm": 0.05928881481505614,
      "learning_rate": 0.00019191949715955672,
      "loss": 0.8584,
      "step": 41840
    },
    {
      "epoch": 0.7347390228058779,
      "grad_norm": 0.08220713814183263,
      "learning_rate": 0.00019191508702048112,
      "loss": 0.8479,
      "step": 41850
    },
    {
      "epoch": 0.7349145876858794,
      "grad_norm": 0.06903736561967637,
      "learning_rate": 0.00019191067572949398,
      "loss": 0.8462,
      "step": 41860
    },
    {
      "epoch": 0.7350901525658807,
      "grad_norm": 0.05535131479150352,
      "learning_rate": 0.00019190626328665112,
      "loss": 0.8499,
      "step": 41870
    },
    {
      "epoch": 0.7352657174458821,
      "grad_norm": 0.04418016080274009,
      "learning_rate": 0.00019190184969200848,
      "loss": 0.8482,
      "step": 41880
    },
    {
      "epoch": 0.7354412823258836,
      "grad_norm": 0.055365908014858176,
      "learning_rate": 0.000191897434945622,
      "loss": 0.853,
      "step": 41890
    },
    {
      "epoch": 0.7356168472058849,
      "grad_norm": 0.05648825865455968,
      "learning_rate": 0.00019189301904754758,
      "loss": 0.8432,
      "step": 41900
    },
    {
      "epoch": 0.7357924120858863,
      "grad_norm": 0.06141361672770195,
      "learning_rate": 0.00019188860199784123,
      "loss": 0.8439,
      "step": 41910
    },
    {
      "epoch": 0.7359679769658878,
      "grad_norm": 0.05237978612837326,
      "learning_rate": 0.00019188418379655883,
      "loss": 0.852,
      "step": 41920
    },
    {
      "epoch": 0.7361435418458891,
      "grad_norm": 0.06884684080682875,
      "learning_rate": 0.00019187976444375645,
      "loss": 0.8559,
      "step": 41930
    },
    {
      "epoch": 0.7363191067258905,
      "grad_norm": 0.06627078358003316,
      "learning_rate": 0.00019187534393949,
      "loss": 0.849,
      "step": 41940
    },
    {
      "epoch": 0.736494671605892,
      "grad_norm": 0.05516337570345281,
      "learning_rate": 0.00019187092228381553,
      "loss": 0.8541,
      "step": 41950
    },
    {
      "epoch": 0.7366702364858934,
      "grad_norm": 0.05411168139049187,
      "learning_rate": 0.00019186649947678907,
      "loss": 0.8534,
      "step": 41960
    },
    {
      "epoch": 0.7368458013658947,
      "grad_norm": 0.05030790422918472,
      "learning_rate": 0.00019186207551846664,
      "loss": 0.8491,
      "step": 41970
    },
    {
      "epoch": 0.7370213662458962,
      "grad_norm": 0.08649242063828952,
      "learning_rate": 0.0001918576504089043,
      "loss": 0.8517,
      "step": 41980
    },
    {
      "epoch": 0.7371969311258976,
      "grad_norm": 0.07146140215788527,
      "learning_rate": 0.0001918532241481581,
      "loss": 0.8518,
      "step": 41990
    },
    {
      "epoch": 0.7373724960058989,
      "grad_norm": 0.06318181150594072,
      "learning_rate": 0.00019184879673628415,
      "loss": 0.8495,
      "step": 42000
    },
    {
      "epoch": 0.7375480608859004,
      "grad_norm": 0.04911692271438476,
      "learning_rate": 0.00019184436817333854,
      "loss": 0.8478,
      "step": 42010
    },
    {
      "epoch": 0.7377236257659018,
      "grad_norm": 0.0743884540788653,
      "learning_rate": 0.0001918399384593774,
      "loss": 0.8518,
      "step": 42020
    },
    {
      "epoch": 0.7378991906459031,
      "grad_norm": 0.06131054850299415,
      "learning_rate": 0.00019183550759445682,
      "loss": 0.8494,
      "step": 42030
    },
    {
      "epoch": 0.7380747555259046,
      "grad_norm": 0.07204146136591467,
      "learning_rate": 0.00019183107557863292,
      "loss": 0.8526,
      "step": 42040
    },
    {
      "epoch": 0.738250320405906,
      "grad_norm": 0.0770639771637805,
      "learning_rate": 0.0001918266424119619,
      "loss": 0.8573,
      "step": 42050
    },
    {
      "epoch": 0.7384258852859074,
      "grad_norm": 0.056213196354656984,
      "learning_rate": 0.00019182220809449988,
      "loss": 0.8499,
      "step": 42060
    },
    {
      "epoch": 0.7386014501659088,
      "grad_norm": 0.07993380854865577,
      "learning_rate": 0.00019181777262630313,
      "loss": 0.8466,
      "step": 42070
    },
    {
      "epoch": 0.7387770150459102,
      "grad_norm": 0.06349126329786077,
      "learning_rate": 0.0001918133360074278,
      "loss": 0.8529,
      "step": 42080
    },
    {
      "epoch": 0.7389525799259116,
      "grad_norm": 0.05431233989208571,
      "learning_rate": 0.00019180889823793004,
      "loss": 0.8529,
      "step": 42090
    },
    {
      "epoch": 0.739128144805913,
      "grad_norm": 0.06015516864128651,
      "learning_rate": 0.00019180445931786617,
      "loss": 0.8455,
      "step": 42100
    },
    {
      "epoch": 0.7393037096859144,
      "grad_norm": 0.06118978558607309,
      "learning_rate": 0.0001918000192472924,
      "loss": 0.8612,
      "step": 42110
    },
    {
      "epoch": 0.7394792745659159,
      "grad_norm": 0.05491308317056076,
      "learning_rate": 0.000191795578026265,
      "loss": 0.8474,
      "step": 42120
    },
    {
      "epoch": 0.7396548394459173,
      "grad_norm": 0.05324015440791035,
      "learning_rate": 0.0001917911356548402,
      "loss": 0.847,
      "step": 42130
    },
    {
      "epoch": 0.7398304043259186,
      "grad_norm": 0.06510685189528716,
      "learning_rate": 0.00019178669213307432,
      "loss": 0.8558,
      "step": 42140
    },
    {
      "epoch": 0.7400059692059201,
      "grad_norm": 0.07212088995340919,
      "learning_rate": 0.00019178224746102364,
      "loss": 0.8496,
      "step": 42150
    },
    {
      "epoch": 0.7401815340859215,
      "grad_norm": 0.06690921465272831,
      "learning_rate": 0.00019177780163874452,
      "loss": 0.8613,
      "step": 42160
    },
    {
      "epoch": 0.7403570989659228,
      "grad_norm": 0.07419316223395012,
      "learning_rate": 0.00019177335466629322,
      "loss": 0.8554,
      "step": 42170
    },
    {
      "epoch": 0.7405326638459243,
      "grad_norm": 0.06705817201824564,
      "learning_rate": 0.0001917689065437261,
      "loss": 0.8562,
      "step": 42180
    },
    {
      "epoch": 0.7407082287259257,
      "grad_norm": 0.07961908843048297,
      "learning_rate": 0.00019176445727109962,
      "loss": 0.8523,
      "step": 42190
    },
    {
      "epoch": 0.740883793605927,
      "grad_norm": 0.0613690049009976,
      "learning_rate": 0.00019176000684847003,
      "loss": 0.8605,
      "step": 42200
    },
    {
      "epoch": 0.7410593584859285,
      "grad_norm": 0.05431219176933614,
      "learning_rate": 0.00019175555527589376,
      "loss": 0.8548,
      "step": 42210
    },
    {
      "epoch": 0.7412349233659299,
      "grad_norm": 0.07304118623601136,
      "learning_rate": 0.0001917511025534272,
      "loss": 0.8641,
      "step": 42220
    },
    {
      "epoch": 0.7414104882459313,
      "grad_norm": 0.07200687431119446,
      "learning_rate": 0.00019174664868112676,
      "loss": 0.8611,
      "step": 42230
    },
    {
      "epoch": 0.7415860531259327,
      "grad_norm": 0.05667991305945278,
      "learning_rate": 0.0001917421936590489,
      "loss": 0.8539,
      "step": 42240
    },
    {
      "epoch": 0.7417616180059341,
      "grad_norm": 0.05218953950687969,
      "learning_rate": 0.00019173773748725007,
      "loss": 0.853,
      "step": 42250
    },
    {
      "epoch": 0.7419371828859355,
      "grad_norm": 0.05464323829602822,
      "learning_rate": 0.00019173328016578673,
      "loss": 0.855,
      "step": 42260
    },
    {
      "epoch": 0.7421127477659369,
      "grad_norm": 0.14908557751885032,
      "learning_rate": 0.0001917288216947153,
      "loss": 0.8526,
      "step": 42270
    },
    {
      "epoch": 0.7422883126459383,
      "grad_norm": 0.04655072479145387,
      "learning_rate": 0.00019172436207409236,
      "loss": 0.8515,
      "step": 42280
    },
    {
      "epoch": 0.7424638775259397,
      "grad_norm": 0.09146094248614521,
      "learning_rate": 0.00019171990130397435,
      "loss": 0.8464,
      "step": 42290
    },
    {
      "epoch": 0.7426394424059412,
      "grad_norm": 0.06827389521718434,
      "learning_rate": 0.00019171543938441778,
      "loss": 0.8582,
      "step": 42300
    },
    {
      "epoch": 0.7428150072859425,
      "grad_norm": 0.04824347520185968,
      "learning_rate": 0.00019171097631547922,
      "loss": 0.8467,
      "step": 42310
    },
    {
      "epoch": 0.7429905721659439,
      "grad_norm": 0.08095592792119338,
      "learning_rate": 0.0001917065120972152,
      "loss": 0.8534,
      "step": 42320
    },
    {
      "epoch": 0.7431661370459454,
      "grad_norm": 0.058717642133453464,
      "learning_rate": 0.0001917020467296823,
      "loss": 0.8535,
      "step": 42330
    },
    {
      "epoch": 0.7433417019259467,
      "grad_norm": 0.060189738147158875,
      "learning_rate": 0.00019169758021293708,
      "loss": 0.859,
      "step": 42340
    },
    {
      "epoch": 0.7435172668059481,
      "grad_norm": 0.08681181231403931,
      "learning_rate": 0.00019169311254703614,
      "loss": 0.854,
      "step": 42350
    },
    {
      "epoch": 0.7436928316859496,
      "grad_norm": 0.06436478244627117,
      "learning_rate": 0.0001916886437320361,
      "loss": 0.8537,
      "step": 42360
    },
    {
      "epoch": 0.7438683965659509,
      "grad_norm": 0.057444850744712325,
      "learning_rate": 0.00019168417376799353,
      "loss": 0.8603,
      "step": 42370
    },
    {
      "epoch": 0.7440439614459523,
      "grad_norm": 0.061900612484627714,
      "learning_rate": 0.0001916797026549651,
      "loss": 0.8445,
      "step": 42380
    },
    {
      "epoch": 0.7442195263259538,
      "grad_norm": 0.06594312150000915,
      "learning_rate": 0.00019167523039300747,
      "loss": 0.8583,
      "step": 42390
    },
    {
      "epoch": 0.7443950912059552,
      "grad_norm": 0.061792541172385064,
      "learning_rate": 0.00019167075698217729,
      "loss": 0.8567,
      "step": 42400
    },
    {
      "epoch": 0.7445706560859565,
      "grad_norm": 0.08174376091114989,
      "learning_rate": 0.00019166628242253128,
      "loss": 0.8536,
      "step": 42410
    },
    {
      "epoch": 0.744746220965958,
      "grad_norm": 0.05322973512844875,
      "learning_rate": 0.00019166180671412606,
      "loss": 0.8528,
      "step": 42420
    },
    {
      "epoch": 0.7449217858459594,
      "grad_norm": 0.0697307552313755,
      "learning_rate": 0.00019165732985701838,
      "loss": 0.8487,
      "step": 42430
    },
    {
      "epoch": 0.7450973507259607,
      "grad_norm": 0.06935201258294037,
      "learning_rate": 0.0001916528518512649,
      "loss": 0.8502,
      "step": 42440
    },
    {
      "epoch": 0.7452729156059622,
      "grad_norm": 0.06823494772913313,
      "learning_rate": 0.0001916483726969225,
      "loss": 0.8485,
      "step": 42450
    },
    {
      "epoch": 0.7454484804859636,
      "grad_norm": 0.06359380951137288,
      "learning_rate": 0.00019164389239404784,
      "loss": 0.8535,
      "step": 42460
    },
    {
      "epoch": 0.7456240453659649,
      "grad_norm": 0.06571564164702312,
      "learning_rate": 0.00019163941094269764,
      "loss": 0.8476,
      "step": 42470
    },
    {
      "epoch": 0.7457996102459664,
      "grad_norm": 0.05341484680142433,
      "learning_rate": 0.00019163492834292875,
      "loss": 0.8551,
      "step": 42480
    },
    {
      "epoch": 0.7459751751259678,
      "grad_norm": 0.07029233566397124,
      "learning_rate": 0.00019163044459479794,
      "loss": 0.8584,
      "step": 42490
    },
    {
      "epoch": 0.7461507400059693,
      "grad_norm": 0.05802606678474267,
      "learning_rate": 0.00019162595969836205,
      "loss": 0.8553,
      "step": 42500
    },
    {
      "epoch": 0.7463263048859706,
      "grad_norm": 0.05695427603705178,
      "learning_rate": 0.00019162147365367786,
      "loss": 0.845,
      "step": 42510
    },
    {
      "epoch": 0.746501869765972,
      "grad_norm": 0.07250464756465212,
      "learning_rate": 0.00019161698646080226,
      "loss": 0.8487,
      "step": 42520
    },
    {
      "epoch": 0.7466774346459735,
      "grad_norm": 0.06152934063463218,
      "learning_rate": 0.00019161249811979207,
      "loss": 0.8547,
      "step": 42530
    },
    {
      "epoch": 0.7468529995259748,
      "grad_norm": 0.07044498816886662,
      "learning_rate": 0.00019160800863070413,
      "loss": 0.8545,
      "step": 42540
    },
    {
      "epoch": 0.7470285644059762,
      "grad_norm": 0.0974201836219756,
      "learning_rate": 0.00019160351799359538,
      "loss": 0.8491,
      "step": 42550
    },
    {
      "epoch": 0.7472041292859777,
      "grad_norm": 0.048148701164463764,
      "learning_rate": 0.0001915990262085227,
      "loss": 0.8552,
      "step": 42560
    },
    {
      "epoch": 0.747379694165979,
      "grad_norm": 0.06354992156007205,
      "learning_rate": 0.00019159453327554298,
      "loss": 0.8574,
      "step": 42570
    },
    {
      "epoch": 0.7475552590459804,
      "grad_norm": 0.05283883831345714,
      "learning_rate": 0.00019159003919471318,
      "loss": 0.8554,
      "step": 42580
    },
    {
      "epoch": 0.7477308239259819,
      "grad_norm": 0.05041263254292258,
      "learning_rate": 0.00019158554396609022,
      "loss": 0.8494,
      "step": 42590
    },
    {
      "epoch": 0.7479063888059833,
      "grad_norm": 0.06162119155343058,
      "learning_rate": 0.00019158104758973103,
      "loss": 0.8635,
      "step": 42600
    },
    {
      "epoch": 0.7480819536859846,
      "grad_norm": 0.06816861916329131,
      "learning_rate": 0.00019157655006569262,
      "loss": 0.852,
      "step": 42610
    },
    {
      "epoch": 0.7482575185659861,
      "grad_norm": 0.06504041313408275,
      "learning_rate": 0.00019157205139403195,
      "loss": 0.8509,
      "step": 42620
    },
    {
      "epoch": 0.7484330834459875,
      "grad_norm": 0.06753825041295056,
      "learning_rate": 0.00019156755157480603,
      "loss": 0.8505,
      "step": 42630
    },
    {
      "epoch": 0.7486086483259888,
      "grad_norm": 0.06268828063413207,
      "learning_rate": 0.00019156305060807191,
      "loss": 0.8539,
      "step": 42640
    },
    {
      "epoch": 0.7487842132059903,
      "grad_norm": 0.06372893977689925,
      "learning_rate": 0.00019155854849388658,
      "loss": 0.852,
      "step": 42650
    },
    {
      "epoch": 0.7489597780859917,
      "grad_norm": 0.06041978434754767,
      "learning_rate": 0.00019155404523230705,
      "loss": 0.8501,
      "step": 42660
    },
    {
      "epoch": 0.749135342965993,
      "grad_norm": 0.0604424298778141,
      "learning_rate": 0.00019154954082339042,
      "loss": 0.8566,
      "step": 42670
    },
    {
      "epoch": 0.7493109078459945,
      "grad_norm": 0.059110150962359775,
      "learning_rate": 0.00019154503526719378,
      "loss": 0.8575,
      "step": 42680
    },
    {
      "epoch": 0.7494864727259959,
      "grad_norm": 0.060011944583619185,
      "learning_rate": 0.00019154052856377418,
      "loss": 0.8436,
      "step": 42690
    },
    {
      "epoch": 0.7496620376059973,
      "grad_norm": 0.05819662913692565,
      "learning_rate": 0.0001915360207131887,
      "loss": 0.8464,
      "step": 42700
    },
    {
      "epoch": 0.7498376024859987,
      "grad_norm": 0.06300363679996465,
      "learning_rate": 0.00019153151171549451,
      "loss": 0.8598,
      "step": 42710
    },
    {
      "epoch": 0.7500131673660001,
      "grad_norm": 0.06886949302323106,
      "learning_rate": 0.00019152700157074872,
      "loss": 0.8514,
      "step": 42720
    },
    {
      "epoch": 0.7501887322460015,
      "grad_norm": 0.07243209440370177,
      "learning_rate": 0.00019152249027900846,
      "loss": 0.8495,
      "step": 42730
    },
    {
      "epoch": 0.7503642971260029,
      "grad_norm": 0.06880769420176851,
      "learning_rate": 0.0001915179778403309,
      "loss": 0.8512,
      "step": 42740
    },
    {
      "epoch": 0.7505398620060043,
      "grad_norm": 0.053201257541854584,
      "learning_rate": 0.00019151346425477322,
      "loss": 0.8553,
      "step": 42750
    },
    {
      "epoch": 0.7507154268860057,
      "grad_norm": 0.06682259123974979,
      "learning_rate": 0.00019150894952239263,
      "loss": 0.8669,
      "step": 42760
    },
    {
      "epoch": 0.7508909917660072,
      "grad_norm": 0.060071405810216943,
      "learning_rate": 0.00019150443364324626,
      "loss": 0.8538,
      "step": 42770
    },
    {
      "epoch": 0.7510665566460085,
      "grad_norm": 0.06637404349142398,
      "learning_rate": 0.0001914999166173914,
      "loss": 0.8485,
      "step": 42780
    },
    {
      "epoch": 0.7512421215260099,
      "grad_norm": 0.0970489456885728,
      "learning_rate": 0.00019149539844488524,
      "loss": 0.8542,
      "step": 42790
    },
    {
      "epoch": 0.7514176864060114,
      "grad_norm": 0.13836465412392815,
      "learning_rate": 0.00019149087912578503,
      "loss": 0.8512,
      "step": 42800
    },
    {
      "epoch": 0.7515932512860127,
      "grad_norm": 0.05112321718768301,
      "learning_rate": 0.00019148635866014804,
      "loss": 0.8486,
      "step": 42810
    },
    {
      "epoch": 0.7517688161660141,
      "grad_norm": 0.07153757206206239,
      "learning_rate": 0.00019148183704803155,
      "loss": 0.8563,
      "step": 42820
    },
    {
      "epoch": 0.7519443810460156,
      "grad_norm": 0.09040642235020295,
      "learning_rate": 0.00019147731428949282,
      "loss": 0.851,
      "step": 42830
    },
    {
      "epoch": 0.7521199459260169,
      "grad_norm": 0.05708291779687337,
      "learning_rate": 0.00019147279038458922,
      "loss": 0.8483,
      "step": 42840
    },
    {
      "epoch": 0.7522955108060184,
      "grad_norm": 0.048069040288879435,
      "learning_rate": 0.00019146826533337802,
      "loss": 0.8578,
      "step": 42850
    },
    {
      "epoch": 0.7524710756860198,
      "grad_norm": 0.051950612145056245,
      "learning_rate": 0.00019146373913591654,
      "loss": 0.8493,
      "step": 42860
    },
    {
      "epoch": 0.7526466405660212,
      "grad_norm": 0.06367300066290818,
      "learning_rate": 0.00019145921179226217,
      "loss": 0.8509,
      "step": 42870
    },
    {
      "epoch": 0.7528222054460226,
      "grad_norm": 0.05603438033476271,
      "learning_rate": 0.00019145468330247225,
      "loss": 0.85,
      "step": 42880
    },
    {
      "epoch": 0.752997770326024,
      "grad_norm": 0.05455794232045381,
      "learning_rate": 0.00019145015366660412,
      "loss": 0.8475,
      "step": 42890
    },
    {
      "epoch": 0.7531733352060254,
      "grad_norm": 0.1022504540453588,
      "learning_rate": 0.00019144562288471526,
      "loss": 0.849,
      "step": 42900
    },
    {
      "epoch": 0.7533489000860268,
      "grad_norm": 0.07481500100964095,
      "learning_rate": 0.00019144109095686298,
      "loss": 0.8482,
      "step": 42910
    },
    {
      "epoch": 0.7535244649660282,
      "grad_norm": 0.05393901041883641,
      "learning_rate": 0.00019143655788310474,
      "loss": 0.8558,
      "step": 42920
    },
    {
      "epoch": 0.7537000298460296,
      "grad_norm": 0.05751725945045717,
      "learning_rate": 0.000191432023663498,
      "loss": 0.8545,
      "step": 42930
    },
    {
      "epoch": 0.753875594726031,
      "grad_norm": 0.09838444129726343,
      "learning_rate": 0.0001914274882981002,
      "loss": 0.8523,
      "step": 42940
    },
    {
      "epoch": 0.7540511596060324,
      "grad_norm": 0.050764922188395206,
      "learning_rate": 0.00019142295178696878,
      "loss": 0.8499,
      "step": 42950
    },
    {
      "epoch": 0.7542267244860338,
      "grad_norm": 0.07497254169517446,
      "learning_rate": 0.0001914184141301612,
      "loss": 0.8624,
      "step": 42960
    },
    {
      "epoch": 0.7544022893660353,
      "grad_norm": 0.0610058609466335,
      "learning_rate": 0.00019141387532773508,
      "loss": 0.8547,
      "step": 42970
    },
    {
      "epoch": 0.7545778542460366,
      "grad_norm": 0.051617165619585126,
      "learning_rate": 0.00019140933537974776,
      "loss": 0.8565,
      "step": 42980
    },
    {
      "epoch": 0.754753419126038,
      "grad_norm": 0.05675798193752236,
      "learning_rate": 0.00019140479428625683,
      "loss": 0.851,
      "step": 42990
    },
    {
      "epoch": 0.7549289840060395,
      "grad_norm": 0.08674319007110397,
      "learning_rate": 0.00019140025204731983,
      "loss": 0.8541,
      "step": 43000
    },
    {
      "epoch": 0.7551045488860408,
      "grad_norm": 0.06139725990199388,
      "learning_rate": 0.00019139570866299433,
      "loss": 0.858,
      "step": 43010
    },
    {
      "epoch": 0.7552801137660422,
      "grad_norm": 0.05900332684373865,
      "learning_rate": 0.00019139116413333787,
      "loss": 0.8606,
      "step": 43020
    },
    {
      "epoch": 0.7554556786460437,
      "grad_norm": 0.11077976642458771,
      "learning_rate": 0.00019138661845840806,
      "loss": 0.8489,
      "step": 43030
    },
    {
      "epoch": 0.755631243526045,
      "grad_norm": 0.05868577398641113,
      "learning_rate": 0.00019138207163826244,
      "loss": 0.847,
      "step": 43040
    },
    {
      "epoch": 0.7558068084060464,
      "grad_norm": 0.04742798665858617,
      "learning_rate": 0.00019137752367295868,
      "loss": 0.8529,
      "step": 43050
    },
    {
      "epoch": 0.7559823732860479,
      "grad_norm": 0.07155013703865064,
      "learning_rate": 0.0001913729745625543,
      "loss": 0.8536,
      "step": 43060
    },
    {
      "epoch": 0.7561579381660493,
      "grad_norm": 0.09152997217188928,
      "learning_rate": 0.0001913684243071071,
      "loss": 0.8514,
      "step": 43070
    },
    {
      "epoch": 0.7563335030460506,
      "grad_norm": 0.0893512509356903,
      "learning_rate": 0.00019136387290667463,
      "loss": 0.8417,
      "step": 43080
    },
    {
      "epoch": 0.7565090679260521,
      "grad_norm": 0.053045948997844296,
      "learning_rate": 0.00019135932036131457,
      "loss": 0.8479,
      "step": 43090
    },
    {
      "epoch": 0.7566846328060535,
      "grad_norm": 0.06054906529493519,
      "learning_rate": 0.00019135476667108458,
      "loss": 0.8486,
      "step": 43100
    },
    {
      "epoch": 0.7568601976860548,
      "grad_norm": 0.05856809846315758,
      "learning_rate": 0.00019135021183604238,
      "loss": 0.8551,
      "step": 43110
    },
    {
      "epoch": 0.7570357625660563,
      "grad_norm": 0.06416588555842025,
      "learning_rate": 0.0001913456558562457,
      "loss": 0.8554,
      "step": 43120
    },
    {
      "epoch": 0.7572113274460577,
      "grad_norm": 0.08976666528875044,
      "learning_rate": 0.00019134109873175222,
      "loss": 0.8447,
      "step": 43130
    },
    {
      "epoch": 0.757386892326059,
      "grad_norm": 0.06181263925021499,
      "learning_rate": 0.00019133654046261974,
      "loss": 0.8536,
      "step": 43140
    },
    {
      "epoch": 0.7575624572060605,
      "grad_norm": 0.085170097956014,
      "learning_rate": 0.00019133198104890595,
      "loss": 0.8508,
      "step": 43150
    },
    {
      "epoch": 0.7577380220860619,
      "grad_norm": 0.05963783007742476,
      "learning_rate": 0.00019132742049066866,
      "loss": 0.8571,
      "step": 43160
    },
    {
      "epoch": 0.7579135869660633,
      "grad_norm": 0.05502548248638751,
      "learning_rate": 0.00019132285878796564,
      "loss": 0.8544,
      "step": 43170
    },
    {
      "epoch": 0.7580891518460647,
      "grad_norm": 0.06258926822093153,
      "learning_rate": 0.0001913182959408547,
      "loss": 0.8527,
      "step": 43180
    },
    {
      "epoch": 0.7582647167260661,
      "grad_norm": 0.075405304302208,
      "learning_rate": 0.00019131373194939362,
      "loss": 0.8599,
      "step": 43190
    },
    {
      "epoch": 0.7584402816060675,
      "grad_norm": 0.05102254964403599,
      "learning_rate": 0.00019130916681364023,
      "loss": 0.8598,
      "step": 43200
    },
    {
      "epoch": 0.758615846486069,
      "grad_norm": 0.0633227979244252,
      "learning_rate": 0.00019130460053365243,
      "loss": 0.8578,
      "step": 43210
    },
    {
      "epoch": 0.7587914113660703,
      "grad_norm": 0.07162051105052118,
      "learning_rate": 0.00019130003310948804,
      "loss": 0.8519,
      "step": 43220
    },
    {
      "epoch": 0.7589669762460718,
      "grad_norm": 0.0751667719740759,
      "learning_rate": 0.00019129546454120486,
      "loss": 0.8459,
      "step": 43230
    },
    {
      "epoch": 0.7591425411260732,
      "grad_norm": 0.06265616918771903,
      "learning_rate": 0.00019129089482886093,
      "loss": 0.8557,
      "step": 43240
    },
    {
      "epoch": 0.7593181060060745,
      "grad_norm": 0.07799194320498737,
      "learning_rate": 0.00019128632397251397,
      "loss": 0.8549,
      "step": 43250
    },
    {
      "epoch": 0.759493670886076,
      "grad_norm": 0.0535220205843802,
      "learning_rate": 0.00019128175197222205,
      "loss": 0.8584,
      "step": 43260
    },
    {
      "epoch": 0.7596692357660774,
      "grad_norm": 0.055485579533433284,
      "learning_rate": 0.00019127717882804298,
      "loss": 0.8561,
      "step": 43270
    },
    {
      "epoch": 0.7598448006460787,
      "grad_norm": 0.05889931543284124,
      "learning_rate": 0.0001912726045400348,
      "loss": 0.8514,
      "step": 43280
    },
    {
      "epoch": 0.7600203655260802,
      "grad_norm": 0.05550677484685841,
      "learning_rate": 0.0001912680291082554,
      "loss": 0.857,
      "step": 43290
    },
    {
      "epoch": 0.7601959304060816,
      "grad_norm": 0.06464537934148877,
      "learning_rate": 0.00019126345253276276,
      "loss": 0.8535,
      "step": 43300
    },
    {
      "epoch": 0.7603714952860829,
      "grad_norm": 0.06577644805195315,
      "learning_rate": 0.0001912588748136149,
      "loss": 0.8557,
      "step": 43310
    },
    {
      "epoch": 0.7605470601660844,
      "grad_norm": 0.04599574391884156,
      "learning_rate": 0.00019125429595086976,
      "loss": 0.8589,
      "step": 43320
    },
    {
      "epoch": 0.7607226250460858,
      "grad_norm": 0.0876172595765683,
      "learning_rate": 0.00019124971594458543,
      "loss": 0.8557,
      "step": 43330
    },
    {
      "epoch": 0.7608981899260872,
      "grad_norm": 0.06091396322675385,
      "learning_rate": 0.00019124513479481988,
      "loss": 0.8521,
      "step": 43340
    },
    {
      "epoch": 0.7610737548060886,
      "grad_norm": 0.0958461469225102,
      "learning_rate": 0.0001912405525016312,
      "loss": 0.8503,
      "step": 43350
    },
    {
      "epoch": 0.76124931968609,
      "grad_norm": 0.06139837545230879,
      "learning_rate": 0.0001912359690650774,
      "loss": 0.8577,
      "step": 43360
    },
    {
      "epoch": 0.7614248845660914,
      "grad_norm": 0.04180305019279672,
      "learning_rate": 0.0001912313844852166,
      "loss": 0.8575,
      "step": 43370
    },
    {
      "epoch": 0.7616004494460928,
      "grad_norm": 0.06424519899493925,
      "learning_rate": 0.00019122679876210685,
      "loss": 0.8598,
      "step": 43380
    },
    {
      "epoch": 0.7617760143260942,
      "grad_norm": 0.0555589852013795,
      "learning_rate": 0.00019122221189580627,
      "loss": 0.8522,
      "step": 43390
    },
    {
      "epoch": 0.7619515792060956,
      "grad_norm": 0.0641139232562304,
      "learning_rate": 0.000191217623886373,
      "loss": 0.8611,
      "step": 43400
    },
    {
      "epoch": 0.762127144086097,
      "grad_norm": 0.08552525915987269,
      "learning_rate": 0.00019121303473386512,
      "loss": 0.8505,
      "step": 43410
    },
    {
      "epoch": 0.7623027089660984,
      "grad_norm": 0.06770031122780662,
      "learning_rate": 0.0001912084444383408,
      "loss": 0.8593,
      "step": 43420
    },
    {
      "epoch": 0.7624782738460998,
      "grad_norm": 0.06834292333081758,
      "learning_rate": 0.0001912038529998582,
      "loss": 0.8515,
      "step": 43430
    },
    {
      "epoch": 0.7626538387261013,
      "grad_norm": 0.08720453934868087,
      "learning_rate": 0.00019119926041847552,
      "loss": 0.854,
      "step": 43440
    },
    {
      "epoch": 0.7628294036061026,
      "grad_norm": 0.06323105895939206,
      "learning_rate": 0.00019119466669425088,
      "loss": 0.8565,
      "step": 43450
    },
    {
      "epoch": 0.763004968486104,
      "grad_norm": 0.0545830584378564,
      "learning_rate": 0.00019119007182724255,
      "loss": 0.8535,
      "step": 43460
    },
    {
      "epoch": 0.7631805333661055,
      "grad_norm": 0.06923936294047123,
      "learning_rate": 0.00019118547581750873,
      "loss": 0.8561,
      "step": 43470
    },
    {
      "epoch": 0.7633560982461068,
      "grad_norm": 0.05234530076885749,
      "learning_rate": 0.00019118087866510763,
      "loss": 0.853,
      "step": 43480
    },
    {
      "epoch": 0.7635316631261082,
      "grad_norm": 0.07569626401341849,
      "learning_rate": 0.00019117628037009754,
      "loss": 0.8565,
      "step": 43490
    },
    {
      "epoch": 0.7637072280061097,
      "grad_norm": 0.06133328432504088,
      "learning_rate": 0.0001911716809325367,
      "loss": 0.8523,
      "step": 43500
    },
    {
      "epoch": 0.763882792886111,
      "grad_norm": 0.07274960535026694,
      "learning_rate": 0.00019116708035248333,
      "loss": 0.8567,
      "step": 43510
    },
    {
      "epoch": 0.7640583577661124,
      "grad_norm": 0.0509351578216849,
      "learning_rate": 0.00019116247862999584,
      "loss": 0.843,
      "step": 43520
    },
    {
      "epoch": 0.7642339226461139,
      "grad_norm": 0.05042322822634551,
      "learning_rate": 0.0001911578757651324,
      "loss": 0.8456,
      "step": 43530
    },
    {
      "epoch": 0.7644094875261153,
      "grad_norm": 0.05356871696092514,
      "learning_rate": 0.00019115327175795143,
      "loss": 0.8526,
      "step": 43540
    },
    {
      "epoch": 0.7645850524061166,
      "grad_norm": 0.07239790615644189,
      "learning_rate": 0.00019114866660851124,
      "loss": 0.8429,
      "step": 43550
    },
    {
      "epoch": 0.7647606172861181,
      "grad_norm": 0.07453555750697899,
      "learning_rate": 0.00019114406031687015,
      "loss": 0.852,
      "step": 43560
    },
    {
      "epoch": 0.7649361821661195,
      "grad_norm": 0.08721007310983454,
      "learning_rate": 0.00019113945288308653,
      "loss": 0.8492,
      "step": 43570
    },
    {
      "epoch": 0.765111747046121,
      "grad_norm": 0.05042548215832969,
      "learning_rate": 0.0001911348443072188,
      "loss": 0.8569,
      "step": 43580
    },
    {
      "epoch": 0.7652873119261223,
      "grad_norm": 0.05200012309722896,
      "learning_rate": 0.00019113023458932528,
      "loss": 0.8516,
      "step": 43590
    },
    {
      "epoch": 0.7654628768061237,
      "grad_norm": 0.05703222553952181,
      "learning_rate": 0.00019112562372946445,
      "loss": 0.8557,
      "step": 43600
    },
    {
      "epoch": 0.7656384416861252,
      "grad_norm": 0.06320190677517377,
      "learning_rate": 0.00019112101172769468,
      "loss": 0.8542,
      "step": 43610
    },
    {
      "epoch": 0.7658140065661265,
      "grad_norm": 0.059391985254567736,
      "learning_rate": 0.00019111639858407443,
      "loss": 0.8545,
      "step": 43620
    },
    {
      "epoch": 0.7659895714461279,
      "grad_norm": 0.07383790747672873,
      "learning_rate": 0.0001911117842986621,
      "loss": 0.8447,
      "step": 43630
    },
    {
      "epoch": 0.7661651363261294,
      "grad_norm": 0.058010825468149334,
      "learning_rate": 0.0001911071688715162,
      "loss": 0.8464,
      "step": 43640
    },
    {
      "epoch": 0.7663407012061307,
      "grad_norm": 0.06686043542023572,
      "learning_rate": 0.00019110255230269524,
      "loss": 0.8524,
      "step": 43650
    },
    {
      "epoch": 0.7665162660861321,
      "grad_norm": 0.07255565478487642,
      "learning_rate": 0.00019109793459225763,
      "loss": 0.8468,
      "step": 43660
    },
    {
      "epoch": 0.7666918309661336,
      "grad_norm": 0.06190737457435144,
      "learning_rate": 0.00019109331574026196,
      "loss": 0.8592,
      "step": 43670
    },
    {
      "epoch": 0.766867395846135,
      "grad_norm": 0.055201367509751395,
      "learning_rate": 0.00019108869574676671,
      "loss": 0.8465,
      "step": 43680
    },
    {
      "epoch": 0.7670429607261363,
      "grad_norm": 0.06813231032629243,
      "learning_rate": 0.0001910840746118304,
      "loss": 0.8474,
      "step": 43690
    },
    {
      "epoch": 0.7672185256061378,
      "grad_norm": 0.060602642276489045,
      "learning_rate": 0.00019107945233551164,
      "loss": 0.8546,
      "step": 43700
    },
    {
      "epoch": 0.7673940904861392,
      "grad_norm": 0.07468021258741334,
      "learning_rate": 0.00019107482891786892,
      "loss": 0.8459,
      "step": 43710
    },
    {
      "epoch": 0.7675696553661405,
      "grad_norm": 0.08507546848009234,
      "learning_rate": 0.00019107020435896082,
      "loss": 0.8485,
      "step": 43720
    },
    {
      "epoch": 0.767745220246142,
      "grad_norm": 0.06518135960371774,
      "learning_rate": 0.00019106557865884602,
      "loss": 0.8531,
      "step": 43730
    },
    {
      "epoch": 0.7679207851261434,
      "grad_norm": 0.06309733927149293,
      "learning_rate": 0.0001910609518175831,
      "loss": 0.8532,
      "step": 43740
    },
    {
      "epoch": 0.7680963500061447,
      "grad_norm": 0.06900257575400766,
      "learning_rate": 0.0001910563238352306,
      "loss": 0.8515,
      "step": 43750
    },
    {
      "epoch": 0.7682719148861462,
      "grad_norm": 0.052757516885484565,
      "learning_rate": 0.00019105169471184726,
      "loss": 0.8454,
      "step": 43760
    },
    {
      "epoch": 0.7684474797661476,
      "grad_norm": 0.06038105679676378,
      "learning_rate": 0.00019104706444749168,
      "loss": 0.8493,
      "step": 43770
    },
    {
      "epoch": 0.768623044646149,
      "grad_norm": 0.058891281806450056,
      "learning_rate": 0.00019104243304222251,
      "loss": 0.8584,
      "step": 43780
    },
    {
      "epoch": 0.7687986095261504,
      "grad_norm": 0.05486035308692534,
      "learning_rate": 0.00019103780049609847,
      "loss": 0.8541,
      "step": 43790
    },
    {
      "epoch": 0.7689741744061518,
      "grad_norm": 0.06314697880451292,
      "learning_rate": 0.00019103316680917824,
      "loss": 0.8507,
      "step": 43800
    },
    {
      "epoch": 0.7691497392861532,
      "grad_norm": 0.06004674140562878,
      "learning_rate": 0.00019102853198152056,
      "loss": 0.8576,
      "step": 43810
    },
    {
      "epoch": 0.7693253041661546,
      "grad_norm": 0.06310113365605874,
      "learning_rate": 0.0001910238960131841,
      "loss": 0.8518,
      "step": 43820
    },
    {
      "epoch": 0.769500869046156,
      "grad_norm": 0.05784739438217662,
      "learning_rate": 0.00019101925890422766,
      "loss": 0.8579,
      "step": 43830
    },
    {
      "epoch": 0.7696764339261574,
      "grad_norm": 0.06885898332132023,
      "learning_rate": 0.0001910146206547099,
      "loss": 0.8467,
      "step": 43840
    },
    {
      "epoch": 0.7698519988061588,
      "grad_norm": 0.08235296008850172,
      "learning_rate": 0.00019100998126468966,
      "loss": 0.8532,
      "step": 43850
    },
    {
      "epoch": 0.7700275636861602,
      "grad_norm": 0.05562841696630697,
      "learning_rate": 0.00019100534073422572,
      "loss": 0.8566,
      "step": 43860
    },
    {
      "epoch": 0.7702031285661616,
      "grad_norm": 0.0634956229979075,
      "learning_rate": 0.00019100069906337688,
      "loss": 0.8471,
      "step": 43870
    },
    {
      "epoch": 0.770378693446163,
      "grad_norm": 0.0692280618064705,
      "learning_rate": 0.00019099605625220195,
      "loss": 0.8531,
      "step": 43880
    },
    {
      "epoch": 0.7705542583261644,
      "grad_norm": 0.09234540335188632,
      "learning_rate": 0.0001909914123007597,
      "loss": 0.8587,
      "step": 43890
    },
    {
      "epoch": 0.7707298232061658,
      "grad_norm": 0.06055945697706788,
      "learning_rate": 0.000190986767209109,
      "loss": 0.8554,
      "step": 43900
    },
    {
      "epoch": 0.7709053880861673,
      "grad_norm": 0.08376353456706266,
      "learning_rate": 0.00019098212097730873,
      "loss": 0.8527,
      "step": 43910
    },
    {
      "epoch": 0.7710809529661686,
      "grad_norm": 0.07920234826178658,
      "learning_rate": 0.00019097747360541777,
      "loss": 0.8522,
      "step": 43920
    },
    {
      "epoch": 0.77125651784617,
      "grad_norm": 0.06156897503603408,
      "learning_rate": 0.0001909728250934949,
      "loss": 0.8529,
      "step": 43930
    },
    {
      "epoch": 0.7714320827261715,
      "grad_norm": 0.12418466254872228,
      "learning_rate": 0.00019096817544159917,
      "loss": 0.8498,
      "step": 43940
    },
    {
      "epoch": 0.7716076476061728,
      "grad_norm": 0.078904928168104,
      "learning_rate": 0.0001909635246497894,
      "loss": 0.8502,
      "step": 43950
    },
    {
      "epoch": 0.7717832124861743,
      "grad_norm": 0.08014012510684952,
      "learning_rate": 0.0001909588727181245,
      "loss": 0.8542,
      "step": 43960
    },
    {
      "epoch": 0.7719587773661757,
      "grad_norm": 0.07506632874216354,
      "learning_rate": 0.00019095421964666345,
      "loss": 0.8487,
      "step": 43970
    },
    {
      "epoch": 0.772134342246177,
      "grad_norm": 0.059462781597757565,
      "learning_rate": 0.0001909495654354652,
      "loss": 0.8515,
      "step": 43980
    },
    {
      "epoch": 0.7723099071261785,
      "grad_norm": 0.0631024043440751,
      "learning_rate": 0.0001909449100845887,
      "loss": 0.8487,
      "step": 43990
    },
    {
      "epoch": 0.7724854720061799,
      "grad_norm": 0.08129650861027936,
      "learning_rate": 0.00019094025359409303,
      "loss": 0.8604,
      "step": 44000
    },
    {
      "epoch": 0.7726610368861813,
      "grad_norm": 0.09414710673669144,
      "learning_rate": 0.00019093559596403703,
      "loss": 0.852,
      "step": 44010
    },
    {
      "epoch": 0.7728366017661827,
      "grad_norm": 0.08823169545683522,
      "learning_rate": 0.00019093093719447984,
      "loss": 0.854,
      "step": 44020
    },
    {
      "epoch": 0.7730121666461841,
      "grad_norm": 0.05759452026997099,
      "learning_rate": 0.0001909262772854804,
      "loss": 0.8476,
      "step": 44030
    },
    {
      "epoch": 0.7731877315261855,
      "grad_norm": 0.045540678865752415,
      "learning_rate": 0.00019092161623709782,
      "loss": 0.8601,
      "step": 44040
    },
    {
      "epoch": 0.773363296406187,
      "grad_norm": 0.07700498430403721,
      "learning_rate": 0.00019091695404939112,
      "loss": 0.8533,
      "step": 44050
    },
    {
      "epoch": 0.7735388612861883,
      "grad_norm": 0.07056329290668334,
      "learning_rate": 0.0001909122907224194,
      "loss": 0.8524,
      "step": 44060
    },
    {
      "epoch": 0.7737144261661897,
      "grad_norm": 0.05806047574749843,
      "learning_rate": 0.0001909076262562417,
      "loss": 0.8523,
      "step": 44070
    },
    {
      "epoch": 0.7738899910461912,
      "grad_norm": 0.0746612113288889,
      "learning_rate": 0.00019090296065091714,
      "loss": 0.8533,
      "step": 44080
    },
    {
      "epoch": 0.7740655559261925,
      "grad_norm": 0.05002577678142353,
      "learning_rate": 0.00019089829390650482,
      "loss": 0.8485,
      "step": 44090
    },
    {
      "epoch": 0.7742411208061939,
      "grad_norm": 0.07242245639658373,
      "learning_rate": 0.00019089362602306393,
      "loss": 0.8516,
      "step": 44100
    },
    {
      "epoch": 0.7744166856861954,
      "grad_norm": 0.08380074614653389,
      "learning_rate": 0.00019088895700065356,
      "loss": 0.8495,
      "step": 44110
    },
    {
      "epoch": 0.7745922505661967,
      "grad_norm": 0.06970305940441722,
      "learning_rate": 0.00019088428683933288,
      "loss": 0.8531,
      "step": 44120
    },
    {
      "epoch": 0.7747678154461981,
      "grad_norm": 0.055188050487441756,
      "learning_rate": 0.00019087961553916104,
      "loss": 0.8541,
      "step": 44130
    },
    {
      "epoch": 0.7749433803261996,
      "grad_norm": 0.06563713931196974,
      "learning_rate": 0.00019087494310019728,
      "loss": 0.8551,
      "step": 44140
    },
    {
      "epoch": 0.775118945206201,
      "grad_norm": 0.07128414529342648,
      "learning_rate": 0.00019087026952250072,
      "loss": 0.8471,
      "step": 44150
    },
    {
      "epoch": 0.7752945100862023,
      "grad_norm": 0.08953288783930324,
      "learning_rate": 0.00019086559480613067,
      "loss": 0.8519,
      "step": 44160
    },
    {
      "epoch": 0.7754700749662038,
      "grad_norm": 0.0655321909468844,
      "learning_rate": 0.0001908609189511463,
      "loss": 0.8573,
      "step": 44170
    },
    {
      "epoch": 0.7756456398462052,
      "grad_norm": 0.05446819197474134,
      "learning_rate": 0.00019085624195760683,
      "loss": 0.8462,
      "step": 44180
    },
    {
      "epoch": 0.7758212047262065,
      "grad_norm": 0.06705855184861718,
      "learning_rate": 0.0001908515638255716,
      "loss": 0.8502,
      "step": 44190
    },
    {
      "epoch": 0.775996769606208,
      "grad_norm": 0.06142147638291985,
      "learning_rate": 0.0001908468845550998,
      "loss": 0.8528,
      "step": 44200
    },
    {
      "epoch": 0.7761723344862094,
      "grad_norm": 0.055073382392248134,
      "learning_rate": 0.00019084220414625075,
      "loss": 0.8594,
      "step": 44210
    },
    {
      "epoch": 0.7763478993662107,
      "grad_norm": 0.054326307380996176,
      "learning_rate": 0.00019083752259908376,
      "loss": 0.8481,
      "step": 44220
    },
    {
      "epoch": 0.7765234642462122,
      "grad_norm": 0.04876220891614223,
      "learning_rate": 0.00019083283991365817,
      "loss": 0.8491,
      "step": 44230
    },
    {
      "epoch": 0.7766990291262136,
      "grad_norm": 0.06388136463932291,
      "learning_rate": 0.00019082815609003327,
      "loss": 0.8576,
      "step": 44240
    },
    {
      "epoch": 0.776874594006215,
      "grad_norm": 0.08918023260377533,
      "learning_rate": 0.0001908234711282684,
      "loss": 0.8521,
      "step": 44250
    },
    {
      "epoch": 0.7770501588862164,
      "grad_norm": 0.060720894170926756,
      "learning_rate": 0.00019081878502842295,
      "loss": 0.8597,
      "step": 44260
    },
    {
      "epoch": 0.7772257237662178,
      "grad_norm": 0.05536199159828857,
      "learning_rate": 0.0001908140977905563,
      "loss": 0.8598,
      "step": 44270
    },
    {
      "epoch": 0.7774012886462192,
      "grad_norm": 0.06761192197851912,
      "learning_rate": 0.00019080940941472777,
      "loss": 0.8542,
      "step": 44280
    },
    {
      "epoch": 0.7775768535262206,
      "grad_norm": 0.05491260735687094,
      "learning_rate": 0.00019080471990099685,
      "loss": 0.8598,
      "step": 44290
    },
    {
      "epoch": 0.777752418406222,
      "grad_norm": 0.058908477409939784,
      "learning_rate": 0.00019080002924942287,
      "loss": 0.8514,
      "step": 44300
    },
    {
      "epoch": 0.7779279832862235,
      "grad_norm": 0.06055027996783998,
      "learning_rate": 0.00019079533746006538,
      "loss": 0.8537,
      "step": 44310
    },
    {
      "epoch": 0.7781035481662248,
      "grad_norm": 0.060964463957886836,
      "learning_rate": 0.00019079064453298371,
      "loss": 0.8552,
      "step": 44320
    },
    {
      "epoch": 0.7782791130462262,
      "grad_norm": 0.05964295129499069,
      "learning_rate": 0.00019078595046823738,
      "loss": 0.8484,
      "step": 44330
    },
    {
      "epoch": 0.7784546779262277,
      "grad_norm": 0.08503996924511849,
      "learning_rate": 0.00019078125526588585,
      "loss": 0.8526,
      "step": 44340
    },
    {
      "epoch": 0.7786302428062291,
      "grad_norm": 0.07124893767224534,
      "learning_rate": 0.0001907765589259886,
      "loss": 0.8684,
      "step": 44350
    },
    {
      "epoch": 0.7788058076862304,
      "grad_norm": 0.06456637241861683,
      "learning_rate": 0.00019077186144860514,
      "loss": 0.8494,
      "step": 44360
    },
    {
      "epoch": 0.7789813725662319,
      "grad_norm": 0.05518608520345722,
      "learning_rate": 0.000190767162833795,
      "loss": 0.8549,
      "step": 44370
    },
    {
      "epoch": 0.7791569374462333,
      "grad_norm": 0.07328874866011087,
      "learning_rate": 0.0001907624630816177,
      "loss": 0.8551,
      "step": 44380
    },
    {
      "epoch": 0.7793325023262346,
      "grad_norm": 0.07853929037049312,
      "learning_rate": 0.0001907577621921328,
      "loss": 0.8534,
      "step": 44390
    },
    {
      "epoch": 0.7795080672062361,
      "grad_norm": 0.07230949648987116,
      "learning_rate": 0.00019075306016539984,
      "loss": 0.8546,
      "step": 44400
    },
    {
      "epoch": 0.7796836320862375,
      "grad_norm": 0.07984786011580942,
      "learning_rate": 0.00019074835700147843,
      "loss": 0.8556,
      "step": 44410
    },
    {
      "epoch": 0.7798591969662388,
      "grad_norm": 0.05159836833100242,
      "learning_rate": 0.00019074365270042807,
      "loss": 0.8604,
      "step": 44420
    },
    {
      "epoch": 0.7800347618462403,
      "grad_norm": 0.0595292797613431,
      "learning_rate": 0.0001907389472623085,
      "loss": 0.8593,
      "step": 44430
    },
    {
      "epoch": 0.7802103267262417,
      "grad_norm": 0.0817672021982629,
      "learning_rate": 0.00019073424068717925,
      "loss": 0.8515,
      "step": 44440
    },
    {
      "epoch": 0.780385891606243,
      "grad_norm": 0.061914671942300874,
      "learning_rate": 0.00019072953297509998,
      "loss": 0.8416,
      "step": 44450
    },
    {
      "epoch": 0.7805614564862445,
      "grad_norm": 0.07186404087381086,
      "learning_rate": 0.00019072482412613032,
      "loss": 0.8582,
      "step": 44460
    },
    {
      "epoch": 0.7807370213662459,
      "grad_norm": 0.06711762271432485,
      "learning_rate": 0.00019072011414032996,
      "loss": 0.8527,
      "step": 44470
    },
    {
      "epoch": 0.7809125862462473,
      "grad_norm": 0.0637638883152513,
      "learning_rate": 0.0001907154030177585,
      "loss": 0.8536,
      "step": 44480
    },
    {
      "epoch": 0.7810881511262487,
      "grad_norm": 0.11699546982772245,
      "learning_rate": 0.00019071069075847575,
      "loss": 0.8435,
      "step": 44490
    },
    {
      "epoch": 0.7812637160062501,
      "grad_norm": 0.080135721187763,
      "learning_rate": 0.00019070597736254135,
      "loss": 0.8468,
      "step": 44500
    },
    {
      "epoch": 0.7814392808862515,
      "grad_norm": 0.05959723430712761,
      "learning_rate": 0.00019070126283001498,
      "loss": 0.8521,
      "step": 44510
    },
    {
      "epoch": 0.781614845766253,
      "grad_norm": 0.09812216239104536,
      "learning_rate": 0.00019069654716095645,
      "loss": 0.8578,
      "step": 44520
    },
    {
      "epoch": 0.7817904106462543,
      "grad_norm": 0.10250255722309344,
      "learning_rate": 0.00019069183035542547,
      "loss": 0.8448,
      "step": 44530
    },
    {
      "epoch": 0.7819659755262557,
      "grad_norm": 0.05778334435135411,
      "learning_rate": 0.00019068711241348186,
      "loss": 0.8551,
      "step": 44540
    },
    {
      "epoch": 0.7821415404062572,
      "grad_norm": 0.052092199279936566,
      "learning_rate": 0.0001906823933351853,
      "loss": 0.8512,
      "step": 44550
    },
    {
      "epoch": 0.7823171052862585,
      "grad_norm": 0.052814972343927485,
      "learning_rate": 0.0001906776731205956,
      "loss": 0.8491,
      "step": 44560
    },
    {
      "epoch": 0.7824926701662599,
      "grad_norm": 0.059063193637705715,
      "learning_rate": 0.00019067295176977263,
      "loss": 0.8493,
      "step": 44570
    },
    {
      "epoch": 0.7826682350462614,
      "grad_norm": 0.0475682620187915,
      "learning_rate": 0.00019066822928277618,
      "loss": 0.8487,
      "step": 44580
    },
    {
      "epoch": 0.7828437999262627,
      "grad_norm": 0.04747962345145666,
      "learning_rate": 0.0001906635056596661,
      "loss": 0.8501,
      "step": 44590
    },
    {
      "epoch": 0.7830193648062641,
      "grad_norm": 0.0734651274214562,
      "learning_rate": 0.00019065878090050218,
      "loss": 0.8564,
      "step": 44600
    },
    {
      "epoch": 0.7831949296862656,
      "grad_norm": 0.07305158555643522,
      "learning_rate": 0.00019065405500534434,
      "loss": 0.8534,
      "step": 44610
    },
    {
      "epoch": 0.783370494566267,
      "grad_norm": 0.06325821037968286,
      "learning_rate": 0.00019064932797425242,
      "loss": 0.8499,
      "step": 44620
    },
    {
      "epoch": 0.7835460594462683,
      "grad_norm": 0.06737561997340104,
      "learning_rate": 0.00019064459980728634,
      "loss": 0.8563,
      "step": 44630
    },
    {
      "epoch": 0.7837216243262698,
      "grad_norm": 0.07220315677135555,
      "learning_rate": 0.00019063987050450603,
      "loss": 0.8529,
      "step": 44640
    },
    {
      "epoch": 0.7838971892062712,
      "grad_norm": 0.09375034289144679,
      "learning_rate": 0.00019063514006597136,
      "loss": 0.8556,
      "step": 44650
    },
    {
      "epoch": 0.7840727540862725,
      "grad_norm": 0.09103694773730486,
      "learning_rate": 0.00019063040849174229,
      "loss": 0.8526,
      "step": 44660
    },
    {
      "epoch": 0.784248318966274,
      "grad_norm": 0.05598711888784064,
      "learning_rate": 0.00019062567578187876,
      "loss": 0.8501,
      "step": 44670
    },
    {
      "epoch": 0.7844238838462754,
      "grad_norm": 0.06444181986830774,
      "learning_rate": 0.00019062094193644075,
      "loss": 0.8575,
      "step": 44680
    },
    {
      "epoch": 0.7845994487262768,
      "grad_norm": 0.07476056212521108,
      "learning_rate": 0.00019061620695548823,
      "loss": 0.8488,
      "step": 44690
    },
    {
      "epoch": 0.7847750136062782,
      "grad_norm": 0.0600374838034639,
      "learning_rate": 0.00019061147083908117,
      "loss": 0.8552,
      "step": 44700
    },
    {
      "epoch": 0.7849505784862796,
      "grad_norm": 0.07389107271981177,
      "learning_rate": 0.00019060673358727964,
      "loss": 0.8485,
      "step": 44710
    },
    {
      "epoch": 0.7851261433662811,
      "grad_norm": 0.05844123743816953,
      "learning_rate": 0.00019060199520014358,
      "loss": 0.8583,
      "step": 44720
    },
    {
      "epoch": 0.7853017082462824,
      "grad_norm": 0.07247169316153972,
      "learning_rate": 0.0001905972556777331,
      "loss": 0.8544,
      "step": 44730
    },
    {
      "epoch": 0.7854772731262838,
      "grad_norm": 0.11620982744580263,
      "learning_rate": 0.00019059251502010822,
      "loss": 0.8521,
      "step": 44740
    },
    {
      "epoch": 0.7856528380062853,
      "grad_norm": 0.07321410228053757,
      "learning_rate": 0.000190587773227329,
      "loss": 0.8526,
      "step": 44750
    },
    {
      "epoch": 0.7858284028862866,
      "grad_norm": 0.05665206385582303,
      "learning_rate": 0.00019058303029945552,
      "loss": 0.8504,
      "step": 44760
    },
    {
      "epoch": 0.786003967766288,
      "grad_norm": 0.04772442397535011,
      "learning_rate": 0.00019057828623654787,
      "loss": 0.8507,
      "step": 44770
    },
    {
      "epoch": 0.7861795326462895,
      "grad_norm": 0.041550287725110845,
      "learning_rate": 0.00019057354103866619,
      "loss": 0.8533,
      "step": 44780
    },
    {
      "epoch": 0.7863550975262908,
      "grad_norm": 0.09255241627069272,
      "learning_rate": 0.00019056879470587054,
      "loss": 0.8522,
      "step": 44790
    },
    {
      "epoch": 0.7865306624062922,
      "grad_norm": 0.05616334423840382,
      "learning_rate": 0.00019056404723822112,
      "loss": 0.852,
      "step": 44800
    },
    {
      "epoch": 0.7867062272862937,
      "grad_norm": 0.0624341867026592,
      "learning_rate": 0.00019055929863577806,
      "loss": 0.8483,
      "step": 44810
    },
    {
      "epoch": 0.7868817921662951,
      "grad_norm": 0.07407400294540183,
      "learning_rate": 0.00019055454889860152,
      "loss": 0.8564,
      "step": 44820
    },
    {
      "epoch": 0.7870573570462964,
      "grad_norm": 0.057714722404672215,
      "learning_rate": 0.00019054979802675165,
      "loss": 0.8503,
      "step": 44830
    },
    {
      "epoch": 0.7872329219262979,
      "grad_norm": 0.061720097165039976,
      "learning_rate": 0.0001905450460202887,
      "loss": 0.8556,
      "step": 44840
    },
    {
      "epoch": 0.7874084868062993,
      "grad_norm": 0.0706636622638105,
      "learning_rate": 0.0001905402928792729,
      "loss": 0.8545,
      "step": 44850
    },
    {
      "epoch": 0.7875840516863006,
      "grad_norm": 0.06084927031288848,
      "learning_rate": 0.00019053553860376438,
      "loss": 0.8503,
      "step": 44860
    },
    {
      "epoch": 0.7877596165663021,
      "grad_norm": 0.07015971200747814,
      "learning_rate": 0.00019053078319382341,
      "loss": 0.8449,
      "step": 44870
    },
    {
      "epoch": 0.7879351814463035,
      "grad_norm": 0.060582275749050486,
      "learning_rate": 0.0001905260266495103,
      "loss": 0.848,
      "step": 44880
    },
    {
      "epoch": 0.7881107463263048,
      "grad_norm": 0.08838952935912316,
      "learning_rate": 0.00019052126897088525,
      "loss": 0.853,
      "step": 44890
    },
    {
      "epoch": 0.7882863112063063,
      "grad_norm": 0.05070568625717274,
      "learning_rate": 0.0001905165101580086,
      "loss": 0.8551,
      "step": 44900
    },
    {
      "epoch": 0.7884618760863077,
      "grad_norm": 0.09791567411732402,
      "learning_rate": 0.00019051175021094058,
      "loss": 0.8611,
      "step": 44910
    },
    {
      "epoch": 0.7886374409663091,
      "grad_norm": 0.053820492202107176,
      "learning_rate": 0.00019050698912974156,
      "loss": 0.8479,
      "step": 44920
    },
    {
      "epoch": 0.7888130058463105,
      "grad_norm": 0.06491341562977362,
      "learning_rate": 0.00019050222691447178,
      "loss": 0.855,
      "step": 44930
    },
    {
      "epoch": 0.7889885707263119,
      "grad_norm": 0.06249694383187704,
      "learning_rate": 0.00019049746356519166,
      "loss": 0.8526,
      "step": 44940
    },
    {
      "epoch": 0.7891641356063133,
      "grad_norm": 0.07137985874023098,
      "learning_rate": 0.00019049269908196151,
      "loss": 0.8652,
      "step": 44950
    },
    {
      "epoch": 0.7893397004863147,
      "grad_norm": 0.10578055449367747,
      "learning_rate": 0.00019048793346484173,
      "loss": 0.8548,
      "step": 44960
    },
    {
      "epoch": 0.7895152653663161,
      "grad_norm": 0.07460445604909198,
      "learning_rate": 0.0001904831667138927,
      "loss": 0.852,
      "step": 44970
    },
    {
      "epoch": 0.7896908302463175,
      "grad_norm": 0.06207883253497817,
      "learning_rate": 0.00019047839882917477,
      "loss": 0.8489,
      "step": 44980
    },
    {
      "epoch": 0.789866395126319,
      "grad_norm": 0.09828170667825473,
      "learning_rate": 0.0001904736298107484,
      "loss": 0.8572,
      "step": 44990
    },
    {
      "epoch": 0.7900419600063203,
      "grad_norm": 0.05375212629075038,
      "learning_rate": 0.00019046885965867392,
      "loss": 0.8493,
      "step": 45000
    },
    {
      "epoch": 0.7902175248863217,
      "grad_norm": 0.06596709392702883,
      "learning_rate": 0.00019046408837301192,
      "loss": 0.856,
      "step": 45010
    },
    {
      "epoch": 0.7903930897663232,
      "grad_norm": 0.06212117036884291,
      "learning_rate": 0.00019045931595382276,
      "loss": 0.8509,
      "step": 45020
    },
    {
      "epoch": 0.7905686546463245,
      "grad_norm": 0.05318496501497332,
      "learning_rate": 0.00019045454240116693,
      "loss": 0.8561,
      "step": 45030
    },
    {
      "epoch": 0.790744219526326,
      "grad_norm": 0.06210173954896695,
      "learning_rate": 0.00019044976771510487,
      "loss": 0.8617,
      "step": 45040
    },
    {
      "epoch": 0.7909197844063274,
      "grad_norm": 0.054732684085863854,
      "learning_rate": 0.00019044499189569712,
      "loss": 0.8524,
      "step": 45050
    },
    {
      "epoch": 0.7910953492863287,
      "grad_norm": 0.058249813563060124,
      "learning_rate": 0.0001904402149430042,
      "loss": 0.8482,
      "step": 45060
    },
    {
      "epoch": 0.7912709141663302,
      "grad_norm": 0.0674440750207019,
      "learning_rate": 0.0001904354368570866,
      "loss": 0.8565,
      "step": 45070
    },
    {
      "epoch": 0.7914464790463316,
      "grad_norm": 0.05419804825371086,
      "learning_rate": 0.0001904306576380049,
      "loss": 0.8504,
      "step": 45080
    },
    {
      "epoch": 0.791622043926333,
      "grad_norm": 0.05087269315394542,
      "learning_rate": 0.0001904258772858196,
      "loss": 0.8525,
      "step": 45090
    },
    {
      "epoch": 0.7917976088063344,
      "grad_norm": 0.056906375804341436,
      "learning_rate": 0.0001904210958005913,
      "loss": 0.8543,
      "step": 45100
    },
    {
      "epoch": 0.7919731736863358,
      "grad_norm": 0.08422389516273134,
      "learning_rate": 0.0001904163131823806,
      "loss": 0.8471,
      "step": 45110
    },
    {
      "epoch": 0.7921487385663372,
      "grad_norm": 0.07700253492641647,
      "learning_rate": 0.00019041152943124806,
      "loss": 0.8547,
      "step": 45120
    },
    {
      "epoch": 0.7923243034463386,
      "grad_norm": 0.06427242874197718,
      "learning_rate": 0.0001904067445472543,
      "loss": 0.8481,
      "step": 45130
    },
    {
      "epoch": 0.79249986832634,
      "grad_norm": 0.08882837090291437,
      "learning_rate": 0.00019040195853045996,
      "loss": 0.8534,
      "step": 45140
    },
    {
      "epoch": 0.7926754332063414,
      "grad_norm": 0.05604645902201223,
      "learning_rate": 0.00019039717138092567,
      "loss": 0.8519,
      "step": 45150
    },
    {
      "epoch": 0.7928509980863429,
      "grad_norm": 0.07639771741746557,
      "learning_rate": 0.00019039238309871207,
      "loss": 0.8457,
      "step": 45160
    },
    {
      "epoch": 0.7930265629663442,
      "grad_norm": 0.08222794188389493,
      "learning_rate": 0.0001903875936838799,
      "loss": 0.8532,
      "step": 45170
    },
    {
      "epoch": 0.7932021278463456,
      "grad_norm": 0.05643927431072956,
      "learning_rate": 0.00019038280313648972,
      "loss": 0.8419,
      "step": 45180
    },
    {
      "epoch": 0.7933776927263471,
      "grad_norm": 0.06234762497080811,
      "learning_rate": 0.00019037801145660232,
      "loss": 0.8508,
      "step": 45190
    },
    {
      "epoch": 0.7935532576063484,
      "grad_norm": 0.06507530068841298,
      "learning_rate": 0.00019037321864427838,
      "loss": 0.8543,
      "step": 45200
    },
    {
      "epoch": 0.7937288224863498,
      "grad_norm": 0.06852527209216674,
      "learning_rate": 0.00019036842469957865,
      "loss": 0.8521,
      "step": 45210
    },
    {
      "epoch": 0.7939043873663513,
      "grad_norm": 0.09371737526516519,
      "learning_rate": 0.00019036362962256384,
      "loss": 0.8473,
      "step": 45220
    },
    {
      "epoch": 0.7940799522463526,
      "grad_norm": 0.06605842960590787,
      "learning_rate": 0.0001903588334132947,
      "loss": 0.8529,
      "step": 45230
    },
    {
      "epoch": 0.794255517126354,
      "grad_norm": 0.054179401607217834,
      "learning_rate": 0.00019035403607183205,
      "loss": 0.8527,
      "step": 45240
    },
    {
      "epoch": 0.7944310820063555,
      "grad_norm": 0.05315809359419304,
      "learning_rate": 0.0001903492375982366,
      "loss": 0.8514,
      "step": 45250
    },
    {
      "epoch": 0.7946066468863568,
      "grad_norm": 0.07844966882842982,
      "learning_rate": 0.00019034443799256922,
      "loss": 0.8512,
      "step": 45260
    },
    {
      "epoch": 0.7947822117663582,
      "grad_norm": 0.06736568743122677,
      "learning_rate": 0.00019033963725489068,
      "loss": 0.8518,
      "step": 45270
    },
    {
      "epoch": 0.7949577766463597,
      "grad_norm": 0.06033688617736483,
      "learning_rate": 0.0001903348353852618,
      "loss": 0.8651,
      "step": 45280
    },
    {
      "epoch": 0.7951333415263611,
      "grad_norm": 0.06366639060250276,
      "learning_rate": 0.00019033003238374345,
      "loss": 0.8549,
      "step": 45290
    },
    {
      "epoch": 0.7953089064063624,
      "grad_norm": 0.041650680603661304,
      "learning_rate": 0.00019032522825039648,
      "loss": 0.8524,
      "step": 45300
    },
    {
      "epoch": 0.7954844712863639,
      "grad_norm": 0.05917001545708143,
      "learning_rate": 0.0001903204229852817,
      "loss": 0.8537,
      "step": 45310
    },
    {
      "epoch": 0.7956600361663653,
      "grad_norm": 0.05103812962728192,
      "learning_rate": 0.0001903156165884601,
      "loss": 0.8583,
      "step": 45320
    },
    {
      "epoch": 0.7958356010463666,
      "grad_norm": 0.06186221253990774,
      "learning_rate": 0.0001903108090599925,
      "loss": 0.853,
      "step": 45330
    },
    {
      "epoch": 0.7960111659263681,
      "grad_norm": 0.06817425122842442,
      "learning_rate": 0.0001903060003999398,
      "loss": 0.8572,
      "step": 45340
    },
    {
      "epoch": 0.7961867308063695,
      "grad_norm": 0.054659338938784274,
      "learning_rate": 0.00019030119060836298,
      "loss": 0.8525,
      "step": 45350
    },
    {
      "epoch": 0.7963622956863708,
      "grad_norm": 0.05144454493612122,
      "learning_rate": 0.00019029637968532296,
      "loss": 0.8506,
      "step": 45360
    },
    {
      "epoch": 0.7965378605663723,
      "grad_norm": 0.069271371433093,
      "learning_rate": 0.0001902915676308807,
      "loss": 0.8577,
      "step": 45370
    },
    {
      "epoch": 0.7967134254463737,
      "grad_norm": 0.06832503731925324,
      "learning_rate": 0.00019028675444509718,
      "loss": 0.8497,
      "step": 45380
    },
    {
      "epoch": 0.7968889903263751,
      "grad_norm": 0.059334702988105836,
      "learning_rate": 0.00019028194012803338,
      "loss": 0.8487,
      "step": 45390
    },
    {
      "epoch": 0.7970645552063765,
      "grad_norm": 0.060935876237263126,
      "learning_rate": 0.00019027712467975028,
      "loss": 0.8552,
      "step": 45400
    },
    {
      "epoch": 0.7972401200863779,
      "grad_norm": 0.05617883710968035,
      "learning_rate": 0.00019027230810030884,
      "loss": 0.8544,
      "step": 45410
    },
    {
      "epoch": 0.7974156849663794,
      "grad_norm": 0.06569343048613024,
      "learning_rate": 0.0001902674903897702,
      "loss": 0.8496,
      "step": 45420
    },
    {
      "epoch": 0.7975912498463807,
      "grad_norm": 0.07334980849485803,
      "learning_rate": 0.00019026267154819535,
      "loss": 0.8555,
      "step": 45430
    },
    {
      "epoch": 0.7977668147263821,
      "grad_norm": 0.07515709474897475,
      "learning_rate": 0.00019025785157564536,
      "loss": 0.8575,
      "step": 45440
    },
    {
      "epoch": 0.7979423796063836,
      "grad_norm": 0.05625817737110515,
      "learning_rate": 0.00019025303047218124,
      "loss": 0.8521,
      "step": 45450
    },
    {
      "epoch": 0.798117944486385,
      "grad_norm": 0.056018961146904155,
      "learning_rate": 0.00019024820823786414,
      "loss": 0.8473,
      "step": 45460
    },
    {
      "epoch": 0.7982935093663863,
      "grad_norm": 0.07014023364251756,
      "learning_rate": 0.00019024338487275513,
      "loss": 0.8531,
      "step": 45470
    },
    {
      "epoch": 0.7984690742463878,
      "grad_norm": 0.0570567039416256,
      "learning_rate": 0.00019023856037691535,
      "loss": 0.8401,
      "step": 45480
    },
    {
      "epoch": 0.7986446391263892,
      "grad_norm": 0.07096554283058747,
      "learning_rate": 0.00019023373475040588,
      "loss": 0.8509,
      "step": 45490
    },
    {
      "epoch": 0.7988202040063905,
      "grad_norm": 0.05431789560114488,
      "learning_rate": 0.0001902289079932879,
      "loss": 0.8582,
      "step": 45500
    },
    {
      "epoch": 0.798995768886392,
      "grad_norm": 0.05689089539075336,
      "learning_rate": 0.0001902240801056226,
      "loss": 0.8533,
      "step": 45510
    },
    {
      "epoch": 0.7991713337663934,
      "grad_norm": 0.06861304325485264,
      "learning_rate": 0.00019021925108747106,
      "loss": 0.8502,
      "step": 45520
    },
    {
      "epoch": 0.7993468986463947,
      "grad_norm": 0.07353205105233206,
      "learning_rate": 0.00019021442093889446,
      "loss": 0.8604,
      "step": 45530
    },
    {
      "epoch": 0.7995224635263962,
      "grad_norm": 0.14135287015450984,
      "learning_rate": 0.00019020958965995413,
      "loss": 0.854,
      "step": 45540
    },
    {
      "epoch": 0.7996980284063976,
      "grad_norm": 0.06564832681142667,
      "learning_rate": 0.00019020475725071116,
      "loss": 0.8551,
      "step": 45550
    },
    {
      "epoch": 0.799873593286399,
      "grad_norm": 0.050451671238620155,
      "learning_rate": 0.00019019992371122684,
      "loss": 0.8571,
      "step": 45560
    },
    {
      "epoch": 0.8000491581664004,
      "grad_norm": 0.07270916534337413,
      "learning_rate": 0.0001901950890415624,
      "loss": 0.861,
      "step": 45570
    },
    {
      "epoch": 0.8002247230464018,
      "grad_norm": 0.0718088232340436,
      "learning_rate": 0.00019019025324177905,
      "loss": 0.8598,
      "step": 45580
    },
    {
      "epoch": 0.8004002879264032,
      "grad_norm": 0.07205916574707691,
      "learning_rate": 0.00019018541631193815,
      "loss": 0.8502,
      "step": 45590
    },
    {
      "epoch": 0.8005758528064046,
      "grad_norm": 0.07515963801180701,
      "learning_rate": 0.0001901805782521009,
      "loss": 0.857,
      "step": 45600
    },
    {
      "epoch": 0.800751417686406,
      "grad_norm": 0.06248362538628925,
      "learning_rate": 0.00019017573906232864,
      "loss": 0.8642,
      "step": 45610
    },
    {
      "epoch": 0.8009269825664074,
      "grad_norm": 0.06292600064644602,
      "learning_rate": 0.00019017089874268266,
      "loss": 0.8578,
      "step": 45620
    },
    {
      "epoch": 0.8011025474464089,
      "grad_norm": 0.0905720160353948,
      "learning_rate": 0.00019016605729322433,
      "loss": 0.8542,
      "step": 45630
    },
    {
      "epoch": 0.8012781123264102,
      "grad_norm": 0.06798408775831243,
      "learning_rate": 0.00019016121471401496,
      "loss": 0.8536,
      "step": 45640
    },
    {
      "epoch": 0.8014536772064116,
      "grad_norm": 0.06479405117241491,
      "learning_rate": 0.00019015637100511587,
      "loss": 0.8484,
      "step": 45650
    },
    {
      "epoch": 0.8016292420864131,
      "grad_norm": 0.06962596196800161,
      "learning_rate": 0.00019015152616658852,
      "loss": 0.8496,
      "step": 45660
    },
    {
      "epoch": 0.8018048069664144,
      "grad_norm": 0.05599595964530864,
      "learning_rate": 0.00019014668019849424,
      "loss": 0.8548,
      "step": 45670
    },
    {
      "epoch": 0.8019803718464158,
      "grad_norm": 0.06960025740083006,
      "learning_rate": 0.00019014183310089443,
      "loss": 0.8499,
      "step": 45680
    },
    {
      "epoch": 0.8021559367264173,
      "grad_norm": 0.07668702677062234,
      "learning_rate": 0.00019013698487385053,
      "loss": 0.8555,
      "step": 45690
    },
    {
      "epoch": 0.8023315016064186,
      "grad_norm": 0.060637761563868235,
      "learning_rate": 0.00019013213551742392,
      "loss": 0.8467,
      "step": 45700
    },
    {
      "epoch": 0.80250706648642,
      "grad_norm": 0.06515311993243673,
      "learning_rate": 0.00019012728503167608,
      "loss": 0.855,
      "step": 45710
    },
    {
      "epoch": 0.8026826313664215,
      "grad_norm": 0.0617348418924647,
      "learning_rate": 0.00019012243341666845,
      "loss": 0.8621,
      "step": 45720
    },
    {
      "epoch": 0.8028581962464229,
      "grad_norm": 0.07433422568913817,
      "learning_rate": 0.00019011758067246252,
      "loss": 0.8471,
      "step": 45730
    },
    {
      "epoch": 0.8030337611264242,
      "grad_norm": 0.062126874546975276,
      "learning_rate": 0.00019011272679911976,
      "loss": 0.8498,
      "step": 45740
    },
    {
      "epoch": 0.8032093260064257,
      "grad_norm": 0.07438048193656709,
      "learning_rate": 0.00019010787179670163,
      "loss": 0.8475,
      "step": 45750
    },
    {
      "epoch": 0.8033848908864271,
      "grad_norm": 0.06045005832650598,
      "learning_rate": 0.0001901030156652698,
      "loss": 0.8555,
      "step": 45760
    },
    {
      "epoch": 0.8035604557664285,
      "grad_norm": 0.062183013856550264,
      "learning_rate": 0.0001900981584048856,
      "loss": 0.857,
      "step": 45770
    },
    {
      "epoch": 0.8037360206464299,
      "grad_norm": 0.08471155438516917,
      "learning_rate": 0.00019009330001561065,
      "loss": 0.8529,
      "step": 45780
    },
    {
      "epoch": 0.8039115855264313,
      "grad_norm": 0.0688981906475038,
      "learning_rate": 0.00019008844049750654,
      "loss": 0.858,
      "step": 45790
    },
    {
      "epoch": 0.8040871504064327,
      "grad_norm": 0.05868312393779347,
      "learning_rate": 0.00019008357985063484,
      "loss": 0.8646,
      "step": 45800
    },
    {
      "epoch": 0.8042627152864341,
      "grad_norm": 0.08072604235374833,
      "learning_rate": 0.00019007871807505706,
      "loss": 0.8539,
      "step": 45810
    },
    {
      "epoch": 0.8044382801664355,
      "grad_norm": 0.06845785003273397,
      "learning_rate": 0.00019007385517083486,
      "loss": 0.8495,
      "step": 45820
    },
    {
      "epoch": 0.804613845046437,
      "grad_norm": 0.07219590692821656,
      "learning_rate": 0.00019006899113802988,
      "loss": 0.852,
      "step": 45830
    },
    {
      "epoch": 0.8047894099264383,
      "grad_norm": 0.06302554500108988,
      "learning_rate": 0.00019006412597670367,
      "loss": 0.8509,
      "step": 45840
    },
    {
      "epoch": 0.8049649748064397,
      "grad_norm": 0.09711393393366734,
      "learning_rate": 0.00019005925968691793,
      "loss": 0.8557,
      "step": 45850
    },
    {
      "epoch": 0.8051405396864412,
      "grad_norm": 0.06239515712198112,
      "learning_rate": 0.0001900543922687343,
      "loss": 0.8559,
      "step": 45860
    },
    {
      "epoch": 0.8053161045664425,
      "grad_norm": 0.06423551032948352,
      "learning_rate": 0.00019004952372221446,
      "loss": 0.8558,
      "step": 45870
    },
    {
      "epoch": 0.8054916694464439,
      "grad_norm": 0.06998223903054362,
      "learning_rate": 0.0001900446540474201,
      "loss": 0.8511,
      "step": 45880
    },
    {
      "epoch": 0.8056672343264454,
      "grad_norm": 0.07519542701299065,
      "learning_rate": 0.0001900397832444129,
      "loss": 0.8478,
      "step": 45890
    },
    {
      "epoch": 0.8058427992064467,
      "grad_norm": 0.055711281749932666,
      "learning_rate": 0.00019003491131325454,
      "loss": 0.856,
      "step": 45900
    },
    {
      "epoch": 0.8060183640864481,
      "grad_norm": 0.07984023744053496,
      "learning_rate": 0.00019003003825400684,
      "loss": 0.8529,
      "step": 45910
    },
    {
      "epoch": 0.8061939289664496,
      "grad_norm": 0.05884499549278109,
      "learning_rate": 0.00019002516406673147,
      "loss": 0.8518,
      "step": 45920
    },
    {
      "epoch": 0.806369493846451,
      "grad_norm": 0.09758703691653474,
      "learning_rate": 0.00019002028875149025,
      "loss": 0.8395,
      "step": 45930
    },
    {
      "epoch": 0.8065450587264523,
      "grad_norm": 0.05818485183307461,
      "learning_rate": 0.00019001541230834489,
      "loss": 0.8493,
      "step": 45940
    },
    {
      "epoch": 0.8067206236064538,
      "grad_norm": 0.07467009776160706,
      "learning_rate": 0.00019001053473735718,
      "loss": 0.8437,
      "step": 45950
    },
    {
      "epoch": 0.8068961884864552,
      "grad_norm": 0.06436428824673444,
      "learning_rate": 0.00019000565603858894,
      "loss": 0.8562,
      "step": 45960
    },
    {
      "epoch": 0.8070717533664565,
      "grad_norm": 0.059840978988384086,
      "learning_rate": 0.00019000077621210198,
      "loss": 0.8604,
      "step": 45970
    },
    {
      "epoch": 0.807247318246458,
      "grad_norm": 0.06514624371310407,
      "learning_rate": 0.00018999589525795814,
      "loss": 0.8568,
      "step": 45980
    },
    {
      "epoch": 0.8074228831264594,
      "grad_norm": 0.08648604944448783,
      "learning_rate": 0.0001899910131762192,
      "loss": 0.8502,
      "step": 45990
    },
    {
      "epoch": 0.8075984480064607,
      "grad_norm": 0.06462220560216052,
      "learning_rate": 0.00018998612996694712,
      "loss": 0.8519,
      "step": 46000
    },
    {
      "epoch": 0.8077740128864622,
      "grad_norm": 0.06157552225329187,
      "learning_rate": 0.0001899812456302037,
      "loss": 0.8623,
      "step": 46010
    },
    {
      "epoch": 0.8079495777664636,
      "grad_norm": 0.0592896795163793,
      "learning_rate": 0.00018997636016605089,
      "loss": 0.8482,
      "step": 46020
    },
    {
      "epoch": 0.808125142646465,
      "grad_norm": 0.0774906996768122,
      "learning_rate": 0.0001899714735745505,
      "loss": 0.85,
      "step": 46030
    },
    {
      "epoch": 0.8083007075264664,
      "grad_norm": 0.06641715380737508,
      "learning_rate": 0.0001899665858557645,
      "loss": 0.8584,
      "step": 46040
    },
    {
      "epoch": 0.8084762724064678,
      "grad_norm": 0.04446826120888274,
      "learning_rate": 0.00018996169700975478,
      "loss": 0.8557,
      "step": 46050
    },
    {
      "epoch": 0.8086518372864692,
      "grad_norm": 0.056947636781394105,
      "learning_rate": 0.00018995680703658334,
      "loss": 0.8457,
      "step": 46060
    },
    {
      "epoch": 0.8088274021664706,
      "grad_norm": 0.05127344980934396,
      "learning_rate": 0.0001899519159363121,
      "loss": 0.8577,
      "step": 46070
    },
    {
      "epoch": 0.809002967046472,
      "grad_norm": 0.06168734081101779,
      "learning_rate": 0.00018994702370900307,
      "loss": 0.8466,
      "step": 46080
    },
    {
      "epoch": 0.8091785319264734,
      "grad_norm": 0.07757152856414536,
      "learning_rate": 0.00018994213035471814,
      "loss": 0.8453,
      "step": 46090
    },
    {
      "epoch": 0.8093540968064749,
      "grad_norm": 0.0663690580983604,
      "learning_rate": 0.00018993723587351941,
      "loss": 0.8495,
      "step": 46100
    },
    {
      "epoch": 0.8095296616864762,
      "grad_norm": 0.05425441203936288,
      "learning_rate": 0.00018993234026546887,
      "loss": 0.855,
      "step": 46110
    },
    {
      "epoch": 0.8097052265664776,
      "grad_norm": 0.0661423712070176,
      "learning_rate": 0.0001899274435306285,
      "loss": 0.8519,
      "step": 46120
    },
    {
      "epoch": 0.8098807914464791,
      "grad_norm": 0.059588850809817774,
      "learning_rate": 0.0001899225456690604,
      "loss": 0.8546,
      "step": 46130
    },
    {
      "epoch": 0.8100563563264804,
      "grad_norm": 0.05379269660924585,
      "learning_rate": 0.00018991764668082657,
      "loss": 0.8504,
      "step": 46140
    },
    {
      "epoch": 0.8102319212064819,
      "grad_norm": 0.06511348152017496,
      "learning_rate": 0.00018991274656598916,
      "loss": 0.8482,
      "step": 46150
    },
    {
      "epoch": 0.8104074860864833,
      "grad_norm": 0.11496745206431154,
      "learning_rate": 0.00018990784532461018,
      "loss": 0.8448,
      "step": 46160
    },
    {
      "epoch": 0.8105830509664846,
      "grad_norm": 0.060175012557130585,
      "learning_rate": 0.00018990294295675178,
      "loss": 0.8555,
      "step": 46170
    },
    {
      "epoch": 0.8107586158464861,
      "grad_norm": 0.09133606199171163,
      "learning_rate": 0.00018989803946247604,
      "loss": 0.85,
      "step": 46180
    },
    {
      "epoch": 0.8109341807264875,
      "grad_norm": 0.06552692364055855,
      "learning_rate": 0.0001898931348418451,
      "loss": 0.8542,
      "step": 46190
    },
    {
      "epoch": 0.8111097456064889,
      "grad_norm": 0.0506313239200749,
      "learning_rate": 0.00018988822909492116,
      "loss": 0.8539,
      "step": 46200
    },
    {
      "epoch": 0.8112853104864903,
      "grad_norm": 0.055178451795513755,
      "learning_rate": 0.00018988332222176626,
      "loss": 0.8559,
      "step": 46210
    },
    {
      "epoch": 0.8114608753664917,
      "grad_norm": 0.08336674919628562,
      "learning_rate": 0.00018987841422244263,
      "loss": 0.8558,
      "step": 46220
    },
    {
      "epoch": 0.8116364402464931,
      "grad_norm": 0.06613109408248742,
      "learning_rate": 0.0001898735050970125,
      "loss": 0.8528,
      "step": 46230
    },
    {
      "epoch": 0.8118120051264945,
      "grad_norm": 0.05368596252693342,
      "learning_rate": 0.00018986859484553803,
      "loss": 0.8509,
      "step": 46240
    },
    {
      "epoch": 0.8119875700064959,
      "grad_norm": 0.07274600966979473,
      "learning_rate": 0.0001898636834680814,
      "loss": 0.845,
      "step": 46250
    },
    {
      "epoch": 0.8121631348864973,
      "grad_norm": 0.05408585885989997,
      "learning_rate": 0.0001898587709647049,
      "loss": 0.8664,
      "step": 46260
    },
    {
      "epoch": 0.8123386997664988,
      "grad_norm": 0.06368971279783388,
      "learning_rate": 0.0001898538573354707,
      "loss": 0.8506,
      "step": 46270
    },
    {
      "epoch": 0.8125142646465001,
      "grad_norm": 0.06683808065200275,
      "learning_rate": 0.0001898489425804411,
      "loss": 0.8559,
      "step": 46280
    },
    {
      "epoch": 0.8126898295265015,
      "grad_norm": 0.0664593878067276,
      "learning_rate": 0.00018984402669967838,
      "loss": 0.8481,
      "step": 46290
    },
    {
      "epoch": 0.812865394406503,
      "grad_norm": 0.09066985591978259,
      "learning_rate": 0.00018983910969324485,
      "loss": 0.8535,
      "step": 46300
    },
    {
      "epoch": 0.8130409592865043,
      "grad_norm": 0.07055927716437695,
      "learning_rate": 0.00018983419156120272,
      "loss": 0.8554,
      "step": 46310
    },
    {
      "epoch": 0.8132165241665057,
      "grad_norm": 0.0683822843313708,
      "learning_rate": 0.0001898292723036144,
      "loss": 0.8583,
      "step": 46320
    },
    {
      "epoch": 0.8133920890465072,
      "grad_norm": 0.07174044593998855,
      "learning_rate": 0.00018982435192054216,
      "loss": 0.8521,
      "step": 46330
    },
    {
      "epoch": 0.8135676539265085,
      "grad_norm": 0.09296313581021926,
      "learning_rate": 0.00018981943041204838,
      "loss": 0.8447,
      "step": 46340
    },
    {
      "epoch": 0.8137432188065099,
      "grad_norm": 0.07905175687380822,
      "learning_rate": 0.0001898145077781954,
      "loss": 0.8606,
      "step": 46350
    },
    {
      "epoch": 0.8139187836865114,
      "grad_norm": 0.0669024680036709,
      "learning_rate": 0.00018980958401904554,
      "loss": 0.8569,
      "step": 46360
    },
    {
      "epoch": 0.8140943485665127,
      "grad_norm": 0.07491727357700086,
      "learning_rate": 0.00018980465913466125,
      "loss": 0.8498,
      "step": 46370
    },
    {
      "epoch": 0.8142699134465141,
      "grad_norm": 0.08247727238434298,
      "learning_rate": 0.00018979973312510488,
      "loss": 0.8556,
      "step": 46380
    },
    {
      "epoch": 0.8144454783265156,
      "grad_norm": 0.07121264764651228,
      "learning_rate": 0.00018979480599043893,
      "loss": 0.857,
      "step": 46390
    },
    {
      "epoch": 0.814621043206517,
      "grad_norm": 0.05457497769029417,
      "learning_rate": 0.00018978987773072574,
      "loss": 0.8494,
      "step": 46400
    },
    {
      "epoch": 0.8147966080865183,
      "grad_norm": 0.05328847301085295,
      "learning_rate": 0.00018978494834602776,
      "loss": 0.848,
      "step": 46410
    },
    {
      "epoch": 0.8149721729665198,
      "grad_norm": 0.061759649156685546,
      "learning_rate": 0.00018978001783640752,
      "loss": 0.85,
      "step": 46420
    },
    {
      "epoch": 0.8151477378465212,
      "grad_norm": 0.0588616526254329,
      "learning_rate": 0.00018977508620192742,
      "loss": 0.8516,
      "step": 46430
    },
    {
      "epoch": 0.8153233027265225,
      "grad_norm": 0.06488098197148062,
      "learning_rate": 0.00018977015344264994,
      "loss": 0.851,
      "step": 46440
    },
    {
      "epoch": 0.815498867606524,
      "grad_norm": 0.06466759075810592,
      "learning_rate": 0.00018976521955863765,
      "loss": 0.8575,
      "step": 46450
    },
    {
      "epoch": 0.8156744324865254,
      "grad_norm": 0.06653023457695677,
      "learning_rate": 0.000189760284549953,
      "loss": 0.8594,
      "step": 46460
    },
    {
      "epoch": 0.8158499973665267,
      "grad_norm": 0.06644839093451302,
      "learning_rate": 0.00018975534841665848,
      "loss": 0.8525,
      "step": 46470
    },
    {
      "epoch": 0.8160255622465282,
      "grad_norm": 0.0491603682529446,
      "learning_rate": 0.00018975041115881674,
      "loss": 0.8547,
      "step": 46480
    },
    {
      "epoch": 0.8162011271265296,
      "grad_norm": 0.06412496344829843,
      "learning_rate": 0.00018974547277649027,
      "loss": 0.8576,
      "step": 46490
    },
    {
      "epoch": 0.8163766920065311,
      "grad_norm": 0.06739642489738054,
      "learning_rate": 0.00018974053326974163,
      "loss": 0.8583,
      "step": 46500
    },
    {
      "epoch": 0.8165522568865324,
      "grad_norm": 0.07602797873970307,
      "learning_rate": 0.00018973559263863347,
      "loss": 0.8549,
      "step": 46510
    },
    {
      "epoch": 0.8167278217665338,
      "grad_norm": 0.07027213539478283,
      "learning_rate": 0.00018973065088322832,
      "loss": 0.851,
      "step": 46520
    },
    {
      "epoch": 0.8169033866465353,
      "grad_norm": 0.06884519127119716,
      "learning_rate": 0.0001897257080035888,
      "loss": 0.8556,
      "step": 46530
    },
    {
      "epoch": 0.8170789515265366,
      "grad_norm": 0.0900592303468677,
      "learning_rate": 0.00018972076399977758,
      "loss": 0.8506,
      "step": 46540
    },
    {
      "epoch": 0.817254516406538,
      "grad_norm": 0.0786951425220957,
      "learning_rate": 0.0001897158188718573,
      "loss": 0.8538,
      "step": 46550
    },
    {
      "epoch": 0.8174300812865395,
      "grad_norm": 0.06821753008568109,
      "learning_rate": 0.00018971087261989057,
      "loss": 0.8489,
      "step": 46560
    },
    {
      "epoch": 0.8176056461665409,
      "grad_norm": 0.0554611762104938,
      "learning_rate": 0.00018970592524394003,
      "loss": 0.8476,
      "step": 46570
    },
    {
      "epoch": 0.8177812110465422,
      "grad_norm": 0.05375327186207906,
      "learning_rate": 0.00018970097674406847,
      "loss": 0.8532,
      "step": 46580
    },
    {
      "epoch": 0.8179567759265437,
      "grad_norm": 0.07623888841753557,
      "learning_rate": 0.00018969602712033855,
      "loss": 0.8531,
      "step": 46590
    },
    {
      "epoch": 0.8181323408065451,
      "grad_norm": 0.07209546840059503,
      "learning_rate": 0.00018969107637281296,
      "loss": 0.8457,
      "step": 46600
    },
    {
      "epoch": 0.8183079056865464,
      "grad_norm": 0.06387397050678097,
      "learning_rate": 0.00018968612450155444,
      "loss": 0.8464,
      "step": 46610
    },
    {
      "epoch": 0.8184834705665479,
      "grad_norm": 0.08632149518400693,
      "learning_rate": 0.0001896811715066257,
      "loss": 0.8427,
      "step": 46620
    },
    {
      "epoch": 0.8186590354465493,
      "grad_norm": 0.04925674758261725,
      "learning_rate": 0.00018967621738808952,
      "loss": 0.8559,
      "step": 46630
    },
    {
      "epoch": 0.8188346003265506,
      "grad_norm": 0.05249711930504795,
      "learning_rate": 0.00018967126214600868,
      "loss": 0.8567,
      "step": 46640
    },
    {
      "epoch": 0.8190101652065521,
      "grad_norm": 0.05559878862568373,
      "learning_rate": 0.00018966630578044592,
      "loss": 0.8555,
      "step": 46650
    },
    {
      "epoch": 0.8191857300865535,
      "grad_norm": 0.07547487101227267,
      "learning_rate": 0.00018966134829146415,
      "loss": 0.8582,
      "step": 46660
    },
    {
      "epoch": 0.8193612949665549,
      "grad_norm": 0.06570766306806376,
      "learning_rate": 0.00018965638967912603,
      "loss": 0.8467,
      "step": 46670
    },
    {
      "epoch": 0.8195368598465563,
      "grad_norm": 0.11626975920103669,
      "learning_rate": 0.00018965142994349453,
      "loss": 0.8511,
      "step": 46680
    },
    {
      "epoch": 0.8197124247265577,
      "grad_norm": 0.05474368290740394,
      "learning_rate": 0.0001896464690846324,
      "loss": 0.8477,
      "step": 46690
    },
    {
      "epoch": 0.8198879896065591,
      "grad_norm": 0.06385719410362763,
      "learning_rate": 0.0001896415071026025,
      "loss": 0.8533,
      "step": 46700
    },
    {
      "epoch": 0.8200635544865605,
      "grad_norm": 0.0923714158351842,
      "learning_rate": 0.00018963654399746773,
      "loss": 0.8491,
      "step": 46710
    },
    {
      "epoch": 0.8202391193665619,
      "grad_norm": 0.06365559006995453,
      "learning_rate": 0.00018963157976929093,
      "loss": 0.8432,
      "step": 46720
    },
    {
      "epoch": 0.8204146842465633,
      "grad_norm": 0.06302473207033202,
      "learning_rate": 0.00018962661441813505,
      "loss": 0.8525,
      "step": 46730
    },
    {
      "epoch": 0.8205902491265648,
      "grad_norm": 0.05058420249561633,
      "learning_rate": 0.000189621647944063,
      "loss": 0.8451,
      "step": 46740
    },
    {
      "epoch": 0.8207658140065661,
      "grad_norm": 0.07301534961703443,
      "learning_rate": 0.00018961668034713766,
      "loss": 0.8495,
      "step": 46750
    },
    {
      "epoch": 0.8209413788865675,
      "grad_norm": 0.06499321177916682,
      "learning_rate": 0.000189611711627422,
      "loss": 0.8519,
      "step": 46760
    },
    {
      "epoch": 0.821116943766569,
      "grad_norm": 0.05946538615108797,
      "learning_rate": 0.00018960674178497897,
      "loss": 0.851,
      "step": 46770
    },
    {
      "epoch": 0.8212925086465703,
      "grad_norm": 0.08544450713699828,
      "learning_rate": 0.00018960177081987157,
      "loss": 0.8444,
      "step": 46780
    },
    {
      "epoch": 0.8214680735265717,
      "grad_norm": 0.0857770588339982,
      "learning_rate": 0.0001895967987321627,
      "loss": 0.8528,
      "step": 46790
    },
    {
      "epoch": 0.8216436384065732,
      "grad_norm": 0.07769071916905061,
      "learning_rate": 0.00018959182552191547,
      "loss": 0.8602,
      "step": 46800
    },
    {
      "epoch": 0.8218192032865745,
      "grad_norm": 0.055583775372724104,
      "learning_rate": 0.0001895868511891928,
      "loss": 0.8554,
      "step": 46810
    },
    {
      "epoch": 0.8219947681665759,
      "grad_norm": 0.09093404250839464,
      "learning_rate": 0.00018958187573405773,
      "loss": 0.8537,
      "step": 46820
    },
    {
      "epoch": 0.8221703330465774,
      "grad_norm": 0.061864116126272406,
      "learning_rate": 0.00018957689915657336,
      "loss": 0.8533,
      "step": 46830
    },
    {
      "epoch": 0.8223458979265788,
      "grad_norm": 0.056774982103093626,
      "learning_rate": 0.00018957192145680266,
      "loss": 0.8544,
      "step": 46840
    },
    {
      "epoch": 0.8225214628065801,
      "grad_norm": 0.06072434128807315,
      "learning_rate": 0.00018956694263480879,
      "loss": 0.8528,
      "step": 46850
    },
    {
      "epoch": 0.8226970276865816,
      "grad_norm": 0.09045058604369324,
      "learning_rate": 0.00018956196269065474,
      "loss": 0.8635,
      "step": 46860
    },
    {
      "epoch": 0.822872592566583,
      "grad_norm": 0.12291024021519797,
      "learning_rate": 0.00018955698162440368,
      "loss": 0.854,
      "step": 46870
    },
    {
      "epoch": 0.8230481574465844,
      "grad_norm": 0.10524487235125672,
      "learning_rate": 0.00018955199943611865,
      "loss": 0.8613,
      "step": 46880
    },
    {
      "epoch": 0.8232237223265858,
      "grad_norm": 0.06086763701661653,
      "learning_rate": 0.00018954701612586286,
      "loss": 0.851,
      "step": 46890
    },
    {
      "epoch": 0.8233992872065872,
      "grad_norm": 0.06502430575816841,
      "learning_rate": 0.00018954203169369938,
      "loss": 0.8526,
      "step": 46900
    },
    {
      "epoch": 0.8235748520865886,
      "grad_norm": 0.07456525943384433,
      "learning_rate": 0.0001895370461396914,
      "loss": 0.8614,
      "step": 46910
    },
    {
      "epoch": 0.82375041696659,
      "grad_norm": 0.0639397467096804,
      "learning_rate": 0.00018953205946390208,
      "loss": 0.846,
      "step": 46920
    },
    {
      "epoch": 0.8239259818465914,
      "grad_norm": 0.05925219447346283,
      "learning_rate": 0.00018952707166639457,
      "loss": 0.848,
      "step": 46930
    },
    {
      "epoch": 0.8241015467265929,
      "grad_norm": 0.101827744217073,
      "learning_rate": 0.0001895220827472321,
      "loss": 0.8469,
      "step": 46940
    },
    {
      "epoch": 0.8242771116065942,
      "grad_norm": 0.054317499693273014,
      "learning_rate": 0.00018951709270647793,
      "loss": 0.8524,
      "step": 46950
    },
    {
      "epoch": 0.8244526764865956,
      "grad_norm": 0.0513431063817031,
      "learning_rate": 0.00018951210154419517,
      "loss": 0.8442,
      "step": 46960
    },
    {
      "epoch": 0.8246282413665971,
      "grad_norm": 0.09557636288301301,
      "learning_rate": 0.00018950710926044718,
      "loss": 0.8473,
      "step": 46970
    },
    {
      "epoch": 0.8248038062465984,
      "grad_norm": 0.10685392522584745,
      "learning_rate": 0.0001895021158552971,
      "loss": 0.8538,
      "step": 46980
    },
    {
      "epoch": 0.8249793711265998,
      "grad_norm": 0.055512834277074596,
      "learning_rate": 0.00018949712132880828,
      "loss": 0.8521,
      "step": 46990
    },
    {
      "epoch": 0.8251549360066013,
      "grad_norm": 0.058856514896368346,
      "learning_rate": 0.00018949212568104398,
      "loss": 0.8597,
      "step": 47000
    },
    {
      "epoch": 0.8253305008866026,
      "grad_norm": 0.05917809867777741,
      "learning_rate": 0.00018948712891206745,
      "loss": 0.8519,
      "step": 47010
    },
    {
      "epoch": 0.825506065766604,
      "grad_norm": 0.07189503053379322,
      "learning_rate": 0.0001894821310219421,
      "loss": 0.8535,
      "step": 47020
    },
    {
      "epoch": 0.8256816306466055,
      "grad_norm": 0.06764289270821282,
      "learning_rate": 0.0001894771320107311,
      "loss": 0.8509,
      "step": 47030
    },
    {
      "epoch": 0.8258571955266069,
      "grad_norm": 0.059823661055223,
      "learning_rate": 0.00018947213187849794,
      "loss": 0.8523,
      "step": 47040
    },
    {
      "epoch": 0.8260327604066082,
      "grad_norm": 0.05143293988546911,
      "learning_rate": 0.00018946713062530592,
      "loss": 0.851,
      "step": 47050
    },
    {
      "epoch": 0.8262083252866097,
      "grad_norm": 0.05213759215253858,
      "learning_rate": 0.0001894621282512184,
      "loss": 0.8541,
      "step": 47060
    },
    {
      "epoch": 0.8263838901666111,
      "grad_norm": 0.07920603739970973,
      "learning_rate": 0.00018945712475629874,
      "loss": 0.8582,
      "step": 47070
    },
    {
      "epoch": 0.8265594550466124,
      "grad_norm": 0.05306529917732379,
      "learning_rate": 0.00018945212014061035,
      "loss": 0.8489,
      "step": 47080
    },
    {
      "epoch": 0.8267350199266139,
      "grad_norm": 0.06388662872974599,
      "learning_rate": 0.00018944711440421665,
      "loss": 0.8565,
      "step": 47090
    },
    {
      "epoch": 0.8269105848066153,
      "grad_norm": 0.0733585877991864,
      "learning_rate": 0.00018944210754718107,
      "loss": 0.8536,
      "step": 47100
    },
    {
      "epoch": 0.8270861496866166,
      "grad_norm": 0.0893423101182945,
      "learning_rate": 0.00018943709956956706,
      "loss": 0.8499,
      "step": 47110
    },
    {
      "epoch": 0.8272617145666181,
      "grad_norm": 0.06117599564128448,
      "learning_rate": 0.000189432090471438,
      "loss": 0.8486,
      "step": 47120
    },
    {
      "epoch": 0.8274372794466195,
      "grad_norm": 0.05345631581234062,
      "learning_rate": 0.00018942708025285743,
      "loss": 0.8599,
      "step": 47130
    },
    {
      "epoch": 0.8276128443266209,
      "grad_norm": 0.07153033672454288,
      "learning_rate": 0.00018942206891388883,
      "loss": 0.8495,
      "step": 47140
    },
    {
      "epoch": 0.8277884092066223,
      "grad_norm": 0.0613240341966764,
      "learning_rate": 0.00018941705645459563,
      "loss": 0.851,
      "step": 47150
    },
    {
      "epoch": 0.8279639740866237,
      "grad_norm": 0.09618872654848758,
      "learning_rate": 0.00018941204287504143,
      "loss": 0.8582,
      "step": 47160
    },
    {
      "epoch": 0.8281395389666251,
      "grad_norm": 0.07457352372633538,
      "learning_rate": 0.00018940702817528965,
      "loss": 0.8568,
      "step": 47170
    },
    {
      "epoch": 0.8283151038466265,
      "grad_norm": 0.05121564347886171,
      "learning_rate": 0.00018940201235540392,
      "loss": 0.8468,
      "step": 47180
    },
    {
      "epoch": 0.8284906687266279,
      "grad_norm": 0.07277432216715846,
      "learning_rate": 0.00018939699541544773,
      "loss": 0.854,
      "step": 47190
    },
    {
      "epoch": 0.8286662336066293,
      "grad_norm": 0.08359123319715725,
      "learning_rate": 0.00018939197735548467,
      "loss": 0.8505,
      "step": 47200
    },
    {
      "epoch": 0.8288417984866308,
      "grad_norm": 0.05689633741843706,
      "learning_rate": 0.00018938695817557832,
      "loss": 0.8545,
      "step": 47210
    },
    {
      "epoch": 0.8290173633666321,
      "grad_norm": 0.05071649776419942,
      "learning_rate": 0.00018938193787579227,
      "loss": 0.848,
      "step": 47220
    },
    {
      "epoch": 0.8291929282466336,
      "grad_norm": 0.0570919688560704,
      "learning_rate": 0.00018937691645619012,
      "loss": 0.862,
      "step": 47230
    },
    {
      "epoch": 0.829368493126635,
      "grad_norm": 0.05575275238720907,
      "learning_rate": 0.00018937189391683552,
      "loss": 0.8506,
      "step": 47240
    },
    {
      "epoch": 0.8295440580066363,
      "grad_norm": 0.07774245537941749,
      "learning_rate": 0.00018936687025779205,
      "loss": 0.8538,
      "step": 47250
    },
    {
      "epoch": 0.8297196228866378,
      "grad_norm": 0.08521369517396116,
      "learning_rate": 0.00018936184547912343,
      "loss": 0.849,
      "step": 47260
    },
    {
      "epoch": 0.8298951877666392,
      "grad_norm": 0.06826567445218286,
      "learning_rate": 0.00018935681958089329,
      "loss": 0.8508,
      "step": 47270
    },
    {
      "epoch": 0.8300707526466405,
      "grad_norm": 0.05248410465172415,
      "learning_rate": 0.00018935179256316528,
      "loss": 0.8466,
      "step": 47280
    },
    {
      "epoch": 0.830246317526642,
      "grad_norm": 0.045421752031187895,
      "learning_rate": 0.0001893467644260032,
      "loss": 0.8534,
      "step": 47290
    },
    {
      "epoch": 0.8304218824066434,
      "grad_norm": 0.05964232454685994,
      "learning_rate": 0.00018934173516947062,
      "loss": 0.8588,
      "step": 47300
    },
    {
      "epoch": 0.8305974472866448,
      "grad_norm": 0.07603719215337497,
      "learning_rate": 0.00018933670479363135,
      "loss": 0.8491,
      "step": 47310
    },
    {
      "epoch": 0.8307730121666462,
      "grad_norm": 0.05563556164514465,
      "learning_rate": 0.0001893316732985491,
      "loss": 0.8471,
      "step": 47320
    },
    {
      "epoch": 0.8309485770466476,
      "grad_norm": 0.05345872921400778,
      "learning_rate": 0.00018932664068428763,
      "loss": 0.8577,
      "step": 47330
    },
    {
      "epoch": 0.831124141926649,
      "grad_norm": 0.07456320629325344,
      "learning_rate": 0.00018932160695091065,
      "loss": 0.8557,
      "step": 47340
    },
    {
      "epoch": 0.8312997068066504,
      "grad_norm": 0.06870046593907769,
      "learning_rate": 0.000189316572098482,
      "loss": 0.8585,
      "step": 47350
    },
    {
      "epoch": 0.8314752716866518,
      "grad_norm": 0.05672111949170691,
      "learning_rate": 0.0001893115361270655,
      "loss": 0.8484,
      "step": 47360
    },
    {
      "epoch": 0.8316508365666532,
      "grad_norm": 0.07277531710607255,
      "learning_rate": 0.00018930649903672487,
      "loss": 0.86,
      "step": 47370
    },
    {
      "epoch": 0.8318264014466547,
      "grad_norm": 0.06294454113462596,
      "learning_rate": 0.00018930146082752396,
      "loss": 0.8594,
      "step": 47380
    },
    {
      "epoch": 0.832001966326656,
      "grad_norm": 0.05524929616482924,
      "learning_rate": 0.00018929642149952667,
      "loss": 0.8546,
      "step": 47390
    },
    {
      "epoch": 0.8321775312066574,
      "grad_norm": 0.06522904174290492,
      "learning_rate": 0.00018929138105279678,
      "loss": 0.8508,
      "step": 47400
    },
    {
      "epoch": 0.8323530960866589,
      "grad_norm": 0.06582262596064668,
      "learning_rate": 0.00018928633948739814,
      "loss": 0.8495,
      "step": 47410
    },
    {
      "epoch": 0.8325286609666602,
      "grad_norm": 0.06557657297444502,
      "learning_rate": 0.00018928129680339468,
      "loss": 0.8525,
      "step": 47420
    },
    {
      "epoch": 0.8327042258466616,
      "grad_norm": 0.052968359982008134,
      "learning_rate": 0.00018927625300085027,
      "loss": 0.8488,
      "step": 47430
    },
    {
      "epoch": 0.8328797907266631,
      "grad_norm": 0.054994853056666855,
      "learning_rate": 0.0001892712080798288,
      "loss": 0.8493,
      "step": 47440
    },
    {
      "epoch": 0.8330553556066644,
      "grad_norm": 0.05280822653741672,
      "learning_rate": 0.00018926616204039423,
      "loss": 0.8633,
      "step": 47450
    },
    {
      "epoch": 0.8332309204866658,
      "grad_norm": 0.06455053264293038,
      "learning_rate": 0.00018926111488261048,
      "loss": 0.8575,
      "step": 47460
    },
    {
      "epoch": 0.8334064853666673,
      "grad_norm": 0.057322035361754384,
      "learning_rate": 0.00018925606660654147,
      "loss": 0.8454,
      "step": 47470
    },
    {
      "epoch": 0.8335820502466686,
      "grad_norm": 0.062192408093644715,
      "learning_rate": 0.0001892510172122512,
      "loss": 0.8495,
      "step": 47480
    },
    {
      "epoch": 0.83375761512667,
      "grad_norm": 0.07002624974889766,
      "learning_rate": 0.00018924596669980362,
      "loss": 0.855,
      "step": 47490
    },
    {
      "epoch": 0.8339331800066715,
      "grad_norm": 0.05100828877780506,
      "learning_rate": 0.0001892409150692627,
      "loss": 0.8544,
      "step": 47500
    },
    {
      "epoch": 0.8341087448866729,
      "grad_norm": 0.06548445013709236,
      "learning_rate": 0.0001892358623206925,
      "loss": 0.8563,
      "step": 47510
    },
    {
      "epoch": 0.8342843097666742,
      "grad_norm": 0.1091812068044391,
      "learning_rate": 0.00018923080845415703,
      "loss": 0.8509,
      "step": 47520
    },
    {
      "epoch": 0.8344598746466757,
      "grad_norm": 0.07494999014990837,
      "learning_rate": 0.00018922575346972027,
      "loss": 0.8451,
      "step": 47530
    },
    {
      "epoch": 0.8346354395266771,
      "grad_norm": 0.0718689644197126,
      "learning_rate": 0.00018922069736744632,
      "loss": 0.8464,
      "step": 47540
    },
    {
      "epoch": 0.8348110044066784,
      "grad_norm": 0.07561613869229418,
      "learning_rate": 0.00018921564014739926,
      "loss": 0.8585,
      "step": 47550
    },
    {
      "epoch": 0.8349865692866799,
      "grad_norm": 0.062157002888636216,
      "learning_rate": 0.0001892105818096431,
      "loss": 0.8494,
      "step": 47560
    },
    {
      "epoch": 0.8351621341666813,
      "grad_norm": 0.051097607232915204,
      "learning_rate": 0.00018920552235424196,
      "loss": 0.8651,
      "step": 47570
    },
    {
      "epoch": 0.8353376990466826,
      "grad_norm": 0.06537968911808296,
      "learning_rate": 0.00018920046178125995,
      "loss": 0.8592,
      "step": 47580
    },
    {
      "epoch": 0.8355132639266841,
      "grad_norm": 0.06643674928046758,
      "learning_rate": 0.0001891954000907612,
      "loss": 0.8511,
      "step": 47590
    },
    {
      "epoch": 0.8356888288066855,
      "grad_norm": 0.05715773863786495,
      "learning_rate": 0.0001891903372828098,
      "loss": 0.8492,
      "step": 47600
    },
    {
      "epoch": 0.835864393686687,
      "grad_norm": 0.05653626934545264,
      "learning_rate": 0.00018918527335746993,
      "loss": 0.8563,
      "step": 47610
    },
    {
      "epoch": 0.8360399585666883,
      "grad_norm": 0.06068449756037767,
      "learning_rate": 0.0001891802083148058,
      "loss": 0.8491,
      "step": 47620
    },
    {
      "epoch": 0.8362155234466897,
      "grad_norm": 0.07914548372875925,
      "learning_rate": 0.00018917514215488148,
      "loss": 0.8524,
      "step": 47630
    },
    {
      "epoch": 0.8363910883266912,
      "grad_norm": 0.0846850791275102,
      "learning_rate": 0.00018917007487776122,
      "loss": 0.8546,
      "step": 47640
    },
    {
      "epoch": 0.8365666532066925,
      "grad_norm": 0.05719690953990907,
      "learning_rate": 0.00018916500648350921,
      "loss": 0.8524,
      "step": 47650
    },
    {
      "epoch": 0.8367422180866939,
      "grad_norm": 0.0800927039186236,
      "learning_rate": 0.00018915993697218966,
      "loss": 0.8499,
      "step": 47660
    },
    {
      "epoch": 0.8369177829666954,
      "grad_norm": 0.07140091198003722,
      "learning_rate": 0.00018915486634386684,
      "loss": 0.8562,
      "step": 47670
    },
    {
      "epoch": 0.8370933478466968,
      "grad_norm": 0.07856612912579983,
      "learning_rate": 0.00018914979459860496,
      "loss": 0.8564,
      "step": 47680
    },
    {
      "epoch": 0.8372689127266981,
      "grad_norm": 0.07666135405575163,
      "learning_rate": 0.00018914472173646828,
      "loss": 0.8518,
      "step": 47690
    },
    {
      "epoch": 0.8374444776066996,
      "grad_norm": 0.07288172798478493,
      "learning_rate": 0.00018913964775752108,
      "loss": 0.8519,
      "step": 47700
    },
    {
      "epoch": 0.837620042486701,
      "grad_norm": 0.05239959855927765,
      "learning_rate": 0.0001891345726618277,
      "loss": 0.8538,
      "step": 47710
    },
    {
      "epoch": 0.8377956073667023,
      "grad_norm": 0.05630675859985191,
      "learning_rate": 0.00018912949644945233,
      "loss": 0.8484,
      "step": 47720
    },
    {
      "epoch": 0.8379711722467038,
      "grad_norm": 0.06699310627732896,
      "learning_rate": 0.00018912441912045935,
      "loss": 0.8635,
      "step": 47730
    },
    {
      "epoch": 0.8381467371267052,
      "grad_norm": 0.09937560022934527,
      "learning_rate": 0.00018911934067491313,
      "loss": 0.8474,
      "step": 47740
    },
    {
      "epoch": 0.8383223020067065,
      "grad_norm": 0.07188212124129935,
      "learning_rate": 0.00018911426111287795,
      "loss": 0.8534,
      "step": 47750
    },
    {
      "epoch": 0.838497866886708,
      "grad_norm": 0.07552528276141186,
      "learning_rate": 0.00018910918043441822,
      "loss": 0.862,
      "step": 47760
    },
    {
      "epoch": 0.8386734317667094,
      "grad_norm": 0.047259683727201254,
      "learning_rate": 0.00018910409863959828,
      "loss": 0.8501,
      "step": 47770
    },
    {
      "epoch": 0.8388489966467108,
      "grad_norm": 0.06876500706262438,
      "learning_rate": 0.0001890990157284825,
      "loss": 0.8476,
      "step": 47780
    },
    {
      "epoch": 0.8390245615267122,
      "grad_norm": 0.057148324237052905,
      "learning_rate": 0.00018909393170113534,
      "loss": 0.856,
      "step": 47790
    },
    {
      "epoch": 0.8392001264067136,
      "grad_norm": 0.05238798926996781,
      "learning_rate": 0.00018908884655762117,
      "loss": 0.8477,
      "step": 47800
    },
    {
      "epoch": 0.839375691286715,
      "grad_norm": 0.06748613564255113,
      "learning_rate": 0.0001890837602980044,
      "loss": 0.8536,
      "step": 47810
    },
    {
      "epoch": 0.8395512561667164,
      "grad_norm": 0.06378335427244072,
      "learning_rate": 0.00018907867292234956,
      "loss": 0.8627,
      "step": 47820
    },
    {
      "epoch": 0.8397268210467178,
      "grad_norm": 0.09252113317017736,
      "learning_rate": 0.00018907358443072102,
      "loss": 0.8478,
      "step": 47830
    },
    {
      "epoch": 0.8399023859267192,
      "grad_norm": 0.06819248528190816,
      "learning_rate": 0.0001890684948231833,
      "loss": 0.8494,
      "step": 47840
    },
    {
      "epoch": 0.8400779508067207,
      "grad_norm": 0.05787257774764161,
      "learning_rate": 0.00018906340409980085,
      "loss": 0.8549,
      "step": 47850
    },
    {
      "epoch": 0.840253515686722,
      "grad_norm": 0.056192234492235926,
      "learning_rate": 0.00018905831226063822,
      "loss": 0.8548,
      "step": 47860
    },
    {
      "epoch": 0.8404290805667234,
      "grad_norm": 0.05164059613417292,
      "learning_rate": 0.00018905321930575987,
      "loss": 0.8495,
      "step": 47870
    },
    {
      "epoch": 0.8406046454467249,
      "grad_norm": 0.06701307710100989,
      "learning_rate": 0.0001890481252352304,
      "loss": 0.8552,
      "step": 47880
    },
    {
      "epoch": 0.8407802103267262,
      "grad_norm": 0.053163364422305785,
      "learning_rate": 0.00018904303004911428,
      "loss": 0.8606,
      "step": 47890
    },
    {
      "epoch": 0.8409557752067276,
      "grad_norm": 0.08250539753370305,
      "learning_rate": 0.0001890379337474761,
      "loss": 0.8558,
      "step": 47900
    },
    {
      "epoch": 0.8411313400867291,
      "grad_norm": 0.06342956315242838,
      "learning_rate": 0.0001890328363303804,
      "loss": 0.857,
      "step": 47910
    },
    {
      "epoch": 0.8413069049667304,
      "grad_norm": 0.049565943568239484,
      "learning_rate": 0.00018902773779789182,
      "loss": 0.8477,
      "step": 47920
    },
    {
      "epoch": 0.8414824698467318,
      "grad_norm": 0.060826431292887354,
      "learning_rate": 0.00018902263815007492,
      "loss": 0.8514,
      "step": 47930
    },
    {
      "epoch": 0.8416580347267333,
      "grad_norm": 0.11343369692025528,
      "learning_rate": 0.00018901753738699434,
      "loss": 0.8531,
      "step": 47940
    },
    {
      "epoch": 0.8418335996067347,
      "grad_norm": 0.06746706707410802,
      "learning_rate": 0.00018901243550871472,
      "loss": 0.8535,
      "step": 47950
    },
    {
      "epoch": 0.8420091644867361,
      "grad_norm": 0.05956609643218065,
      "learning_rate": 0.00018900733251530063,
      "loss": 0.8578,
      "step": 47960
    },
    {
      "epoch": 0.8421847293667375,
      "grad_norm": 0.11950673032210715,
      "learning_rate": 0.00018900222840681678,
      "loss": 0.8495,
      "step": 47970
    },
    {
      "epoch": 0.8423602942467389,
      "grad_norm": 0.07703795610397475,
      "learning_rate": 0.00018899712318332782,
      "loss": 0.8551,
      "step": 47980
    },
    {
      "epoch": 0.8425358591267403,
      "grad_norm": 0.04675449686301453,
      "learning_rate": 0.00018899201684489846,
      "loss": 0.8519,
      "step": 47990
    },
    {
      "epoch": 0.8427114240067417,
      "grad_norm": 0.05998750487164444,
      "learning_rate": 0.00018898690939159337,
      "loss": 0.8574,
      "step": 48000
    },
    {
      "epoch": 0.8428869888867431,
      "grad_norm": 0.07729410397985514,
      "learning_rate": 0.00018898180082347726,
      "loss": 0.8586,
      "step": 48010
    },
    {
      "epoch": 0.8430625537667445,
      "grad_norm": 0.09954703419178214,
      "learning_rate": 0.00018897669114061494,
      "loss": 0.8451,
      "step": 48020
    },
    {
      "epoch": 0.8432381186467459,
      "grad_norm": 0.1058881403922801,
      "learning_rate": 0.00018897158034307103,
      "loss": 0.8469,
      "step": 48030
    },
    {
      "epoch": 0.8434136835267473,
      "grad_norm": 0.0475617511337497,
      "learning_rate": 0.00018896646843091035,
      "loss": 0.8527,
      "step": 48040
    },
    {
      "epoch": 0.8435892484067488,
      "grad_norm": 0.06207884135915811,
      "learning_rate": 0.00018896135540419766,
      "loss": 0.8488,
      "step": 48050
    },
    {
      "epoch": 0.8437648132867501,
      "grad_norm": 0.0664470276834934,
      "learning_rate": 0.00018895624126299772,
      "loss": 0.8459,
      "step": 48060
    },
    {
      "epoch": 0.8439403781667515,
      "grad_norm": 0.04562696837851839,
      "learning_rate": 0.00018895112600737538,
      "loss": 0.8457,
      "step": 48070
    },
    {
      "epoch": 0.844115943046753,
      "grad_norm": 0.09223218404544382,
      "learning_rate": 0.00018894600963739538,
      "loss": 0.8483,
      "step": 48080
    },
    {
      "epoch": 0.8442915079267543,
      "grad_norm": 0.07869072727331555,
      "learning_rate": 0.00018894089215312258,
      "loss": 0.8596,
      "step": 48090
    },
    {
      "epoch": 0.8444670728067557,
      "grad_norm": 0.06307147327563047,
      "learning_rate": 0.00018893577355462185,
      "loss": 0.8501,
      "step": 48100
    },
    {
      "epoch": 0.8446426376867572,
      "grad_norm": 0.08250733999034059,
      "learning_rate": 0.000188930653841958,
      "loss": 0.8589,
      "step": 48110
    },
    {
      "epoch": 0.8448182025667585,
      "grad_norm": 0.09203451978422257,
      "learning_rate": 0.00018892553301519594,
      "loss": 0.8544,
      "step": 48120
    },
    {
      "epoch": 0.8449937674467599,
      "grad_norm": 0.08590057011378348,
      "learning_rate": 0.0001889204110744005,
      "loss": 0.8499,
      "step": 48130
    },
    {
      "epoch": 0.8451693323267614,
      "grad_norm": 0.06210318313792687,
      "learning_rate": 0.00018891528801963663,
      "loss": 0.8539,
      "step": 48140
    },
    {
      "epoch": 0.8453448972067628,
      "grad_norm": 0.0709859624938539,
      "learning_rate": 0.00018891016385096918,
      "loss": 0.8419,
      "step": 48150
    },
    {
      "epoch": 0.8455204620867641,
      "grad_norm": 0.06541316223319384,
      "learning_rate": 0.00018890503856846313,
      "loss": 0.8556,
      "step": 48160
    },
    {
      "epoch": 0.8456960269667656,
      "grad_norm": 0.0905139313030146,
      "learning_rate": 0.00018889991217218338,
      "loss": 0.8507,
      "step": 48170
    },
    {
      "epoch": 0.845871591846767,
      "grad_norm": 0.05462449082966102,
      "learning_rate": 0.0001888947846621949,
      "loss": 0.8467,
      "step": 48180
    },
    {
      "epoch": 0.8460471567267683,
      "grad_norm": 0.06557041860948322,
      "learning_rate": 0.00018888965603856267,
      "loss": 0.848,
      "step": 48190
    },
    {
      "epoch": 0.8462227216067698,
      "grad_norm": 0.043518409712607335,
      "learning_rate": 0.0001888845263013516,
      "loss": 0.856,
      "step": 48200
    },
    {
      "epoch": 0.8463982864867712,
      "grad_norm": 0.06548515268715119,
      "learning_rate": 0.00018887939545062679,
      "loss": 0.8527,
      "step": 48210
    },
    {
      "epoch": 0.8465738513667725,
      "grad_norm": 0.07209592735140236,
      "learning_rate": 0.00018887426348645318,
      "loss": 0.8489,
      "step": 48220
    },
    {
      "epoch": 0.846749416246774,
      "grad_norm": 0.06115298539679447,
      "learning_rate": 0.00018886913040889582,
      "loss": 0.8521,
      "step": 48230
    },
    {
      "epoch": 0.8469249811267754,
      "grad_norm": 0.06507427688130167,
      "learning_rate": 0.00018886399621801972,
      "loss": 0.8587,
      "step": 48240
    },
    {
      "epoch": 0.8471005460067768,
      "grad_norm": 0.08069562362492906,
      "learning_rate": 0.00018885886091388995,
      "loss": 0.8613,
      "step": 48250
    },
    {
      "epoch": 0.8472761108867782,
      "grad_norm": 0.07723261800417765,
      "learning_rate": 0.0001888537244965716,
      "loss": 0.8609,
      "step": 48260
    },
    {
      "epoch": 0.8474516757667796,
      "grad_norm": 0.11518396115848109,
      "learning_rate": 0.0001888485869661297,
      "loss": 0.8545,
      "step": 48270
    },
    {
      "epoch": 0.847627240646781,
      "grad_norm": 0.06075170931459192,
      "learning_rate": 0.00018884344832262938,
      "loss": 0.8559,
      "step": 48280
    },
    {
      "epoch": 0.8478028055267824,
      "grad_norm": 0.054000722681732845,
      "learning_rate": 0.00018883830856613573,
      "loss": 0.854,
      "step": 48290
    },
    {
      "epoch": 0.8479783704067838,
      "grad_norm": 0.059091077545673165,
      "learning_rate": 0.0001888331676967139,
      "loss": 0.846,
      "step": 48300
    },
    {
      "epoch": 0.8481539352867852,
      "grad_norm": 0.06748344078166037,
      "learning_rate": 0.00018882802571442897,
      "loss": 0.854,
      "step": 48310
    },
    {
      "epoch": 0.8483295001667867,
      "grad_norm": 0.06652518294493157,
      "learning_rate": 0.0001888228826193461,
      "loss": 0.843,
      "step": 48320
    },
    {
      "epoch": 0.848505065046788,
      "grad_norm": 0.08033835461162335,
      "learning_rate": 0.00018881773841153052,
      "loss": 0.8552,
      "step": 48330
    },
    {
      "epoch": 0.8486806299267895,
      "grad_norm": 0.07236344982634965,
      "learning_rate": 0.00018881259309104736,
      "loss": 0.856,
      "step": 48340
    },
    {
      "epoch": 0.8488561948067909,
      "grad_norm": 0.07156636833862395,
      "learning_rate": 0.00018880744665796182,
      "loss": 0.8616,
      "step": 48350
    },
    {
      "epoch": 0.8490317596867922,
      "grad_norm": 0.07523247935301755,
      "learning_rate": 0.00018880229911233908,
      "loss": 0.8486,
      "step": 48360
    },
    {
      "epoch": 0.8492073245667937,
      "grad_norm": 0.07234265918420145,
      "learning_rate": 0.0001887971504542444,
      "loss": 0.8471,
      "step": 48370
    },
    {
      "epoch": 0.8493828894467951,
      "grad_norm": 0.0806312503917523,
      "learning_rate": 0.00018879200068374296,
      "loss": 0.843,
      "step": 48380
    },
    {
      "epoch": 0.8495584543267964,
      "grad_norm": 0.06892892441253065,
      "learning_rate": 0.0001887868498009001,
      "loss": 0.86,
      "step": 48390
    },
    {
      "epoch": 0.8497340192067979,
      "grad_norm": 0.06427252071329209,
      "learning_rate": 0.000188781697805781,
      "loss": 0.8521,
      "step": 48400
    },
    {
      "epoch": 0.8499095840867993,
      "grad_norm": 0.0602552606640196,
      "learning_rate": 0.00018877654469845096,
      "loss": 0.849,
      "step": 48410
    },
    {
      "epoch": 0.8500851489668007,
      "grad_norm": 0.052413305662846646,
      "learning_rate": 0.00018877139047897528,
      "loss": 0.8492,
      "step": 48420
    },
    {
      "epoch": 0.8502607138468021,
      "grad_norm": 0.06172543841830905,
      "learning_rate": 0.00018876623514741927,
      "loss": 0.848,
      "step": 48430
    },
    {
      "epoch": 0.8504362787268035,
      "grad_norm": 0.06468147301090896,
      "learning_rate": 0.00018876107870384825,
      "loss": 0.8594,
      "step": 48440
    },
    {
      "epoch": 0.8506118436068049,
      "grad_norm": 0.06481177404416283,
      "learning_rate": 0.00018875592114832754,
      "loss": 0.8512,
      "step": 48450
    },
    {
      "epoch": 0.8507874084868063,
      "grad_norm": 0.06758501189555792,
      "learning_rate": 0.00018875076248092248,
      "loss": 0.8483,
      "step": 48460
    },
    {
      "epoch": 0.8509629733668077,
      "grad_norm": 0.07721368121010488,
      "learning_rate": 0.00018874560270169845,
      "loss": 0.8454,
      "step": 48470
    },
    {
      "epoch": 0.8511385382468091,
      "grad_norm": 0.0720966680310079,
      "learning_rate": 0.0001887404418107208,
      "loss": 0.8574,
      "step": 48480
    },
    {
      "epoch": 0.8513141031268106,
      "grad_norm": 0.0892230020609019,
      "learning_rate": 0.00018873527980805494,
      "loss": 0.8523,
      "step": 48490
    },
    {
      "epoch": 0.8514896680068119,
      "grad_norm": 0.05738708443025915,
      "learning_rate": 0.0001887301166937663,
      "loss": 0.8582,
      "step": 48500
    },
    {
      "epoch": 0.8516652328868133,
      "grad_norm": 0.06409508425777181,
      "learning_rate": 0.00018872495246792027,
      "loss": 0.8604,
      "step": 48510
    },
    {
      "epoch": 0.8518407977668148,
      "grad_norm": 0.053805326394150886,
      "learning_rate": 0.00018871978713058224,
      "loss": 0.8574,
      "step": 48520
    },
    {
      "epoch": 0.8520163626468161,
      "grad_norm": 0.09158301595534583,
      "learning_rate": 0.0001887146206818177,
      "loss": 0.8502,
      "step": 48530
    },
    {
      "epoch": 0.8521919275268175,
      "grad_norm": 0.06418058719569335,
      "learning_rate": 0.00018870945312169212,
      "loss": 0.8513,
      "step": 48540
    },
    {
      "epoch": 0.852367492406819,
      "grad_norm": 0.0760978214966419,
      "learning_rate": 0.00018870428445027097,
      "loss": 0.8483,
      "step": 48550
    },
    {
      "epoch": 0.8525430572868203,
      "grad_norm": 0.06422963049028697,
      "learning_rate": 0.00018869911466761968,
      "loss": 0.8528,
      "step": 48560
    },
    {
      "epoch": 0.8527186221668217,
      "grad_norm": 0.0796374149606218,
      "learning_rate": 0.00018869394377380385,
      "loss": 0.859,
      "step": 48570
    },
    {
      "epoch": 0.8528941870468232,
      "grad_norm": 0.0545622434360017,
      "learning_rate": 0.00018868877176888893,
      "loss": 0.8487,
      "step": 48580
    },
    {
      "epoch": 0.8530697519268245,
      "grad_norm": 0.08658295846770604,
      "learning_rate": 0.00018868359865294046,
      "loss": 0.8544,
      "step": 48590
    },
    {
      "epoch": 0.8532453168068259,
      "grad_norm": 0.06468235458730677,
      "learning_rate": 0.000188678424426024,
      "loss": 0.8548,
      "step": 48600
    },
    {
      "epoch": 0.8534208816868274,
      "grad_norm": 0.06249217541891776,
      "learning_rate": 0.00018867324908820513,
      "loss": 0.8505,
      "step": 48610
    },
    {
      "epoch": 0.8535964465668288,
      "grad_norm": 0.06611652097596006,
      "learning_rate": 0.00018866807263954934,
      "loss": 0.8509,
      "step": 48620
    },
    {
      "epoch": 0.8537720114468301,
      "grad_norm": 0.0832027570688526,
      "learning_rate": 0.0001886628950801223,
      "loss": 0.8629,
      "step": 48630
    },
    {
      "epoch": 0.8539475763268316,
      "grad_norm": 0.05192774804154658,
      "learning_rate": 0.00018865771640998958,
      "loss": 0.8458,
      "step": 48640
    },
    {
      "epoch": 0.854123141206833,
      "grad_norm": 0.0635379745341515,
      "learning_rate": 0.00018865253662921677,
      "loss": 0.853,
      "step": 48650
    },
    {
      "epoch": 0.8542987060868343,
      "grad_norm": 0.07595029092779879,
      "learning_rate": 0.0001886473557378695,
      "loss": 0.8597,
      "step": 48660
    },
    {
      "epoch": 0.8544742709668358,
      "grad_norm": 0.06861939922776017,
      "learning_rate": 0.00018864217373601346,
      "loss": 0.8524,
      "step": 48670
    },
    {
      "epoch": 0.8546498358468372,
      "grad_norm": 0.046819125027656665,
      "learning_rate": 0.0001886369906237143,
      "loss": 0.8593,
      "step": 48680
    },
    {
      "epoch": 0.8548254007268387,
      "grad_norm": 0.07382798905766637,
      "learning_rate": 0.00018863180640103766,
      "loss": 0.8598,
      "step": 48690
    },
    {
      "epoch": 0.85500096560684,
      "grad_norm": 0.07950287960605623,
      "learning_rate": 0.00018862662106804926,
      "loss": 0.8509,
      "step": 48700
    },
    {
      "epoch": 0.8551765304868414,
      "grad_norm": 0.07595087780881953,
      "learning_rate": 0.00018862143462481474,
      "loss": 0.8516,
      "step": 48710
    },
    {
      "epoch": 0.8553520953668429,
      "grad_norm": 0.05673667081468035,
      "learning_rate": 0.00018861624707139986,
      "loss": 0.8515,
      "step": 48720
    },
    {
      "epoch": 0.8555276602468442,
      "grad_norm": 0.05210360874673874,
      "learning_rate": 0.00018861105840787038,
      "loss": 0.8463,
      "step": 48730
    },
    {
      "epoch": 0.8557032251268456,
      "grad_norm": 0.05929645073633453,
      "learning_rate": 0.00018860586863429194,
      "loss": 0.8571,
      "step": 48740
    },
    {
      "epoch": 0.8558787900068471,
      "grad_norm": 0.050854820294379224,
      "learning_rate": 0.0001886006777507304,
      "loss": 0.8509,
      "step": 48750
    },
    {
      "epoch": 0.8560543548868484,
      "grad_norm": 0.059718452203761346,
      "learning_rate": 0.00018859548575725146,
      "loss": 0.8576,
      "step": 48760
    },
    {
      "epoch": 0.8562299197668498,
      "grad_norm": 0.06350814382251778,
      "learning_rate": 0.0001885902926539209,
      "loss": 0.8478,
      "step": 48770
    },
    {
      "epoch": 0.8564054846468513,
      "grad_norm": 0.05088790850418774,
      "learning_rate": 0.00018858509844080462,
      "loss": 0.8517,
      "step": 48780
    },
    {
      "epoch": 0.8565810495268527,
      "grad_norm": 0.16240285065545768,
      "learning_rate": 0.0001885799031179683,
      "loss": 0.855,
      "step": 48790
    },
    {
      "epoch": 0.856756614406854,
      "grad_norm": 0.07697751034937841,
      "learning_rate": 0.00018857470668547782,
      "loss": 0.854,
      "step": 48800
    },
    {
      "epoch": 0.8569321792868555,
      "grad_norm": 0.06503953927298918,
      "learning_rate": 0.00018856950914339905,
      "loss": 0.8467,
      "step": 48810
    },
    {
      "epoch": 0.8571077441668569,
      "grad_norm": 0.0835643310578262,
      "learning_rate": 0.0001885643104917978,
      "loss": 0.8581,
      "step": 48820
    },
    {
      "epoch": 0.8572833090468582,
      "grad_norm": 0.06812650330842171,
      "learning_rate": 0.00018855911073074,
      "loss": 0.8479,
      "step": 48830
    },
    {
      "epoch": 0.8574588739268597,
      "grad_norm": 0.06480460144371245,
      "learning_rate": 0.00018855390986029143,
      "loss": 0.8587,
      "step": 48840
    },
    {
      "epoch": 0.8576344388068611,
      "grad_norm": 0.0542475871005369,
      "learning_rate": 0.00018854870788051806,
      "loss": 0.8453,
      "step": 48850
    },
    {
      "epoch": 0.8578100036868624,
      "grad_norm": 0.07079363065802077,
      "learning_rate": 0.0001885435047914858,
      "loss": 0.8447,
      "step": 48860
    },
    {
      "epoch": 0.8579855685668639,
      "grad_norm": 0.06282690046457404,
      "learning_rate": 0.00018853830059326052,
      "loss": 0.85,
      "step": 48870
    },
    {
      "epoch": 0.8581611334468653,
      "grad_norm": 0.057426133760762485,
      "learning_rate": 0.0001885330952859082,
      "loss": 0.8475,
      "step": 48880
    },
    {
      "epoch": 0.8583366983268667,
      "grad_norm": 0.06559675437379853,
      "learning_rate": 0.00018852788886949482,
      "loss": 0.8588,
      "step": 48890
    },
    {
      "epoch": 0.8585122632068681,
      "grad_norm": 0.04906119772798667,
      "learning_rate": 0.00018852268134408632,
      "loss": 0.8492,
      "step": 48900
    },
    {
      "epoch": 0.8586878280868695,
      "grad_norm": 0.086849101837544,
      "learning_rate": 0.00018851747270974862,
      "loss": 0.8495,
      "step": 48910
    },
    {
      "epoch": 0.8588633929668709,
      "grad_norm": 0.0530837738279259,
      "learning_rate": 0.00018851226296654782,
      "loss": 0.8592,
      "step": 48920
    },
    {
      "epoch": 0.8590389578468723,
      "grad_norm": 0.07078255096720595,
      "learning_rate": 0.00018850705211454985,
      "loss": 0.8569,
      "step": 48930
    },
    {
      "epoch": 0.8592145227268737,
      "grad_norm": 0.057499243823498976,
      "learning_rate": 0.00018850184015382077,
      "loss": 0.8509,
      "step": 48940
    },
    {
      "epoch": 0.8593900876068751,
      "grad_norm": 0.05513845421797004,
      "learning_rate": 0.00018849662708442656,
      "loss": 0.8522,
      "step": 48950
    },
    {
      "epoch": 0.8595656524868766,
      "grad_norm": 0.07073522021812581,
      "learning_rate": 0.00018849141290643338,
      "loss": 0.8495,
      "step": 48960
    },
    {
      "epoch": 0.8597412173668779,
      "grad_norm": 0.07383646586279723,
      "learning_rate": 0.0001884861976199072,
      "loss": 0.8468,
      "step": 48970
    },
    {
      "epoch": 0.8599167822468793,
      "grad_norm": 0.05664629210241765,
      "learning_rate": 0.00018848098122491413,
      "loss": 0.8537,
      "step": 48980
    },
    {
      "epoch": 0.8600923471268808,
      "grad_norm": 0.07063373916246846,
      "learning_rate": 0.00018847576372152028,
      "loss": 0.8573,
      "step": 48990
    },
    {
      "epoch": 0.8602679120068821,
      "grad_norm": 0.06586540224634967,
      "learning_rate": 0.00018847054510979174,
      "loss": 0.8567,
      "step": 49000
    },
    {
      "epoch": 0.8604434768868835,
      "grad_norm": 0.057944043844559534,
      "learning_rate": 0.0001884653253897946,
      "loss": 0.8557,
      "step": 49010
    },
    {
      "epoch": 0.860619041766885,
      "grad_norm": 0.05788324166384675,
      "learning_rate": 0.00018846010456159505,
      "loss": 0.8521,
      "step": 49020
    },
    {
      "epoch": 0.8607946066468863,
      "grad_norm": 0.08862259518119608,
      "learning_rate": 0.00018845488262525916,
      "loss": 0.8479,
      "step": 49030
    },
    {
      "epoch": 0.8609701715268877,
      "grad_norm": 0.05674427754528213,
      "learning_rate": 0.00018844965958085322,
      "loss": 0.8615,
      "step": 49040
    },
    {
      "epoch": 0.8611457364068892,
      "grad_norm": 0.054331443417079586,
      "learning_rate": 0.0001884444354284433,
      "loss": 0.8493,
      "step": 49050
    },
    {
      "epoch": 0.8613213012868906,
      "grad_norm": 0.055908327047893225,
      "learning_rate": 0.00018843921016809562,
      "loss": 0.85,
      "step": 49060
    },
    {
      "epoch": 0.861496866166892,
      "grad_norm": 0.07243997884314513,
      "learning_rate": 0.00018843398379987642,
      "loss": 0.8522,
      "step": 49070
    },
    {
      "epoch": 0.8616724310468934,
      "grad_norm": 0.10604106335356121,
      "learning_rate": 0.00018842875632385186,
      "loss": 0.8585,
      "step": 49080
    },
    {
      "epoch": 0.8618479959268948,
      "grad_norm": 0.05982790050313992,
      "learning_rate": 0.0001884235277400882,
      "loss": 0.8527,
      "step": 49090
    },
    {
      "epoch": 0.8620235608068962,
      "grad_norm": 0.0858648296732026,
      "learning_rate": 0.0001884182980486517,
      "loss": 0.8524,
      "step": 49100
    },
    {
      "epoch": 0.8621991256868976,
      "grad_norm": 0.07123577021516432,
      "learning_rate": 0.00018841306724960862,
      "loss": 0.8487,
      "step": 49110
    },
    {
      "epoch": 0.862374690566899,
      "grad_norm": 0.09009399942839286,
      "learning_rate": 0.0001884078353430252,
      "loss": 0.8384,
      "step": 49120
    },
    {
      "epoch": 0.8625502554469004,
      "grad_norm": 0.06553789082569897,
      "learning_rate": 0.0001884026023289678,
      "loss": 0.8548,
      "step": 49130
    },
    {
      "epoch": 0.8627258203269018,
      "grad_norm": 0.04590071150377029,
      "learning_rate": 0.0001883973682075026,
      "loss": 0.8532,
      "step": 49140
    },
    {
      "epoch": 0.8629013852069032,
      "grad_norm": 0.10180630883274064,
      "learning_rate": 0.00018839213297869607,
      "loss": 0.8596,
      "step": 49150
    },
    {
      "epoch": 0.8630769500869047,
      "grad_norm": 0.11495616293129815,
      "learning_rate": 0.00018838689664261443,
      "loss": 0.8519,
      "step": 49160
    },
    {
      "epoch": 0.863252514966906,
      "grad_norm": 0.09551178827087767,
      "learning_rate": 0.00018838165919932408,
      "loss": 0.8527,
      "step": 49170
    },
    {
      "epoch": 0.8634280798469074,
      "grad_norm": 0.05571187415674919,
      "learning_rate": 0.00018837642064889137,
      "loss": 0.8576,
      "step": 49180
    },
    {
      "epoch": 0.8636036447269089,
      "grad_norm": 0.05427015232116651,
      "learning_rate": 0.0001883711809913826,
      "loss": 0.8477,
      "step": 49190
    },
    {
      "epoch": 0.8637792096069102,
      "grad_norm": 0.05772438890735048,
      "learning_rate": 0.0001883659402268643,
      "loss": 0.8511,
      "step": 49200
    },
    {
      "epoch": 0.8639547744869116,
      "grad_norm": 0.051077930443854454,
      "learning_rate": 0.00018836069835540274,
      "loss": 0.8652,
      "step": 49210
    },
    {
      "epoch": 0.8641303393669131,
      "grad_norm": 0.05168386271773331,
      "learning_rate": 0.0001883554553770644,
      "loss": 0.8511,
      "step": 49220
    },
    {
      "epoch": 0.8643059042469144,
      "grad_norm": 0.06605208869396463,
      "learning_rate": 0.0001883502112919157,
      "loss": 0.8548,
      "step": 49230
    },
    {
      "epoch": 0.8644814691269158,
      "grad_norm": 0.04772102692795562,
      "learning_rate": 0.0001883449661000231,
      "loss": 0.8562,
      "step": 49240
    },
    {
      "epoch": 0.8646570340069173,
      "grad_norm": 0.060875442870127555,
      "learning_rate": 0.00018833971980145305,
      "loss": 0.8499,
      "step": 49250
    },
    {
      "epoch": 0.8648325988869187,
      "grad_norm": 0.10989091518484727,
      "learning_rate": 0.00018833447239627196,
      "loss": 0.8579,
      "step": 49260
    },
    {
      "epoch": 0.86500816376692,
      "grad_norm": 0.10878467812771798,
      "learning_rate": 0.00018832922388454645,
      "loss": 0.8545,
      "step": 49270
    },
    {
      "epoch": 0.8651837286469215,
      "grad_norm": 0.08285304686892267,
      "learning_rate": 0.00018832397426634285,
      "loss": 0.8608,
      "step": 49280
    },
    {
      "epoch": 0.8653592935269229,
      "grad_norm": 0.0544080436075462,
      "learning_rate": 0.00018831872354172782,
      "loss": 0.8512,
      "step": 49290
    },
    {
      "epoch": 0.8655348584069242,
      "grad_norm": 0.0735796272968038,
      "learning_rate": 0.00018831347171076782,
      "loss": 0.8551,
      "step": 49300
    },
    {
      "epoch": 0.8657104232869257,
      "grad_norm": 0.054824296040180485,
      "learning_rate": 0.0001883082187735294,
      "loss": 0.8542,
      "step": 49310
    },
    {
      "epoch": 0.8658859881669271,
      "grad_norm": 0.06255208373192539,
      "learning_rate": 0.00018830296473007908,
      "loss": 0.8592,
      "step": 49320
    },
    {
      "epoch": 0.8660615530469284,
      "grad_norm": 0.055137282736054054,
      "learning_rate": 0.00018829770958048349,
      "loss": 0.8528,
      "step": 49330
    },
    {
      "epoch": 0.8662371179269299,
      "grad_norm": 0.04610351608859931,
      "learning_rate": 0.00018829245332480915,
      "loss": 0.8451,
      "step": 49340
    },
    {
      "epoch": 0.8664126828069313,
      "grad_norm": 0.06055008103391821,
      "learning_rate": 0.00018828719596312277,
      "loss": 0.8421,
      "step": 49350
    },
    {
      "epoch": 0.8665882476869327,
      "grad_norm": 0.06462751261675172,
      "learning_rate": 0.00018828193749549082,
      "loss": 0.849,
      "step": 49360
    },
    {
      "epoch": 0.8667638125669341,
      "grad_norm": 0.05248056979369269,
      "learning_rate": 0.00018827667792198002,
      "loss": 0.8577,
      "step": 49370
    },
    {
      "epoch": 0.8669393774469355,
      "grad_norm": 0.1030962948795938,
      "learning_rate": 0.000188271417242657,
      "loss": 0.8541,
      "step": 49380
    },
    {
      "epoch": 0.8671149423269369,
      "grad_norm": 0.08054578339386094,
      "learning_rate": 0.0001882661554575884,
      "loss": 0.8589,
      "step": 49390
    },
    {
      "epoch": 0.8672905072069383,
      "grad_norm": 0.04183782322278747,
      "learning_rate": 0.00018826089256684085,
      "loss": 0.8534,
      "step": 49400
    },
    {
      "epoch": 0.8674660720869397,
      "grad_norm": 0.05853861649587684,
      "learning_rate": 0.0001882556285704811,
      "loss": 0.8579,
      "step": 49410
    },
    {
      "epoch": 0.8676416369669412,
      "grad_norm": 0.09210096345908551,
      "learning_rate": 0.00018825036346857582,
      "loss": 0.8469,
      "step": 49420
    },
    {
      "epoch": 0.8678172018469426,
      "grad_norm": 0.051598428433766434,
      "learning_rate": 0.0001882450972611917,
      "loss": 0.8556,
      "step": 49430
    },
    {
      "epoch": 0.8679927667269439,
      "grad_norm": 0.04558613423209515,
      "learning_rate": 0.00018823982994839548,
      "loss": 0.8552,
      "step": 49440
    },
    {
      "epoch": 0.8681683316069454,
      "grad_norm": 0.06059451115088109,
      "learning_rate": 0.00018823456153025387,
      "loss": 0.8444,
      "step": 49450
    },
    {
      "epoch": 0.8683438964869468,
      "grad_norm": 0.05398534645398694,
      "learning_rate": 0.0001882292920068337,
      "loss": 0.8544,
      "step": 49460
    },
    {
      "epoch": 0.8685194613669481,
      "grad_norm": 0.07005999080060873,
      "learning_rate": 0.00018822402137820164,
      "loss": 0.8557,
      "step": 49470
    },
    {
      "epoch": 0.8686950262469496,
      "grad_norm": 0.0692717821255926,
      "learning_rate": 0.00018821874964442453,
      "loss": 0.8545,
      "step": 49480
    },
    {
      "epoch": 0.868870591126951,
      "grad_norm": 0.08347241271354426,
      "learning_rate": 0.00018821347680556908,
      "loss": 0.8457,
      "step": 49490
    },
    {
      "epoch": 0.8690461560069523,
      "grad_norm": 0.06371701752484844,
      "learning_rate": 0.00018820820286170227,
      "loss": 0.8445,
      "step": 49500
    },
    {
      "epoch": 0.8692217208869538,
      "grad_norm": 0.14650549724108275,
      "learning_rate": 0.00018820292781289078,
      "loss": 0.8466,
      "step": 49510
    },
    {
      "epoch": 0.8693972857669552,
      "grad_norm": 0.057074211415284265,
      "learning_rate": 0.00018819765165920145,
      "loss": 0.8574,
      "step": 49520
    },
    {
      "epoch": 0.8695728506469566,
      "grad_norm": 0.07324111672422629,
      "learning_rate": 0.0001881923744007012,
      "loss": 0.8518,
      "step": 49530
    },
    {
      "epoch": 0.869748415526958,
      "grad_norm": 0.059184613491639296,
      "learning_rate": 0.00018818709603745683,
      "loss": 0.8576,
      "step": 49540
    },
    {
      "epoch": 0.8699239804069594,
      "grad_norm": 0.08598080862807804,
      "learning_rate": 0.00018818181656953523,
      "loss": 0.8505,
      "step": 49550
    },
    {
      "epoch": 0.8700995452869608,
      "grad_norm": 0.07715849359938544,
      "learning_rate": 0.00018817653599700337,
      "loss": 0.8519,
      "step": 49560
    },
    {
      "epoch": 0.8702751101669622,
      "grad_norm": 0.07413488146228027,
      "learning_rate": 0.00018817125431992804,
      "loss": 0.8565,
      "step": 49570
    },
    {
      "epoch": 0.8704506750469636,
      "grad_norm": 0.049328003297057436,
      "learning_rate": 0.00018816597153837624,
      "loss": 0.8463,
      "step": 49580
    },
    {
      "epoch": 0.870626239926965,
      "grad_norm": 0.07365703034633267,
      "learning_rate": 0.00018816068765241484,
      "loss": 0.8608,
      "step": 49590
    },
    {
      "epoch": 0.8708018048069665,
      "grad_norm": 0.07107788122916416,
      "learning_rate": 0.00018815540266211084,
      "loss": 0.852,
      "step": 49600
    },
    {
      "epoch": 0.8709773696869678,
      "grad_norm": 0.05561036358200915,
      "learning_rate": 0.0001881501165675312,
      "loss": 0.8519,
      "step": 49610
    },
    {
      "epoch": 0.8711529345669692,
      "grad_norm": 0.08174601995457047,
      "learning_rate": 0.00018814482936874288,
      "loss": 0.8487,
      "step": 49620
    },
    {
      "epoch": 0.8713284994469707,
      "grad_norm": 0.07060665284047575,
      "learning_rate": 0.00018813954106581285,
      "loss": 0.849,
      "step": 49630
    },
    {
      "epoch": 0.871504064326972,
      "grad_norm": 0.084614316957792,
      "learning_rate": 0.00018813425165880814,
      "loss": 0.8595,
      "step": 49640
    },
    {
      "epoch": 0.8716796292069734,
      "grad_norm": 0.05318389636104503,
      "learning_rate": 0.00018812896114779576,
      "loss": 0.8531,
      "step": 49650
    },
    {
      "epoch": 0.8718551940869749,
      "grad_norm": 0.05362971781934206,
      "learning_rate": 0.00018812366953284274,
      "loss": 0.8606,
      "step": 49660
    },
    {
      "epoch": 0.8720307589669762,
      "grad_norm": 0.06560190341048526,
      "learning_rate": 0.00018811837681401614,
      "loss": 0.8571,
      "step": 49670
    },
    {
      "epoch": 0.8722063238469776,
      "grad_norm": 0.06136277380120743,
      "learning_rate": 0.000188113082991383,
      "loss": 0.8499,
      "step": 49680
    },
    {
      "epoch": 0.8723818887269791,
      "grad_norm": 0.06328825139031169,
      "learning_rate": 0.0001881077880650104,
      "loss": 0.8492,
      "step": 49690
    },
    {
      "epoch": 0.8725574536069804,
      "grad_norm": 0.07188620823734722,
      "learning_rate": 0.00018810249203496545,
      "loss": 0.8503,
      "step": 49700
    },
    {
      "epoch": 0.8727330184869818,
      "grad_norm": 0.07767806861019007,
      "learning_rate": 0.00018809719490131525,
      "loss": 0.846,
      "step": 49710
    },
    {
      "epoch": 0.8729085833669833,
      "grad_norm": 0.0709882756334045,
      "learning_rate": 0.00018809189666412689,
      "loss": 0.8541,
      "step": 49720
    },
    {
      "epoch": 0.8730841482469847,
      "grad_norm": 0.05343518595704108,
      "learning_rate": 0.0001880865973234675,
      "loss": 0.8502,
      "step": 49730
    },
    {
      "epoch": 0.873259713126986,
      "grad_norm": 0.06184675795962728,
      "learning_rate": 0.0001880812968794042,
      "loss": 0.8576,
      "step": 49740
    },
    {
      "epoch": 0.8734352780069875,
      "grad_norm": 0.047430287909337204,
      "learning_rate": 0.00018807599533200419,
      "loss": 0.8613,
      "step": 49750
    },
    {
      "epoch": 0.8736108428869889,
      "grad_norm": 0.05475508217784738,
      "learning_rate": 0.00018807069268133466,
      "loss": 0.8512,
      "step": 49760
    },
    {
      "epoch": 0.8737864077669902,
      "grad_norm": 0.062775226102437,
      "learning_rate": 0.00018806538892746278,
      "loss": 0.8532,
      "step": 49770
    },
    {
      "epoch": 0.8739619726469917,
      "grad_norm": 0.05848647337023254,
      "learning_rate": 0.0001880600840704557,
      "loss": 0.8549,
      "step": 49780
    },
    {
      "epoch": 0.8741375375269931,
      "grad_norm": 0.09320631959790206,
      "learning_rate": 0.00018805477811038065,
      "loss": 0.8532,
      "step": 49790
    },
    {
      "epoch": 0.8743131024069946,
      "grad_norm": 0.07906292996812664,
      "learning_rate": 0.0001880494710473049,
      "loss": 0.8527,
      "step": 49800
    },
    {
      "epoch": 0.8744886672869959,
      "grad_norm": 0.059778882659791574,
      "learning_rate": 0.0001880441628812957,
      "loss": 0.8446,
      "step": 49810
    },
    {
      "epoch": 0.8746642321669973,
      "grad_norm": 0.0742125099729129,
      "learning_rate": 0.00018803885361242024,
      "loss": 0.8516,
      "step": 49820
    },
    {
      "epoch": 0.8748397970469988,
      "grad_norm": 0.05702249813898268,
      "learning_rate": 0.00018803354324074584,
      "loss": 0.8631,
      "step": 49830
    },
    {
      "epoch": 0.8750153619270001,
      "grad_norm": 0.07605071846905573,
      "learning_rate": 0.00018802823176633976,
      "loss": 0.8572,
      "step": 49840
    },
    {
      "epoch": 0.8751909268070015,
      "grad_norm": 0.0919338022832953,
      "learning_rate": 0.0001880229191892693,
      "loss": 0.8449,
      "step": 49850
    },
    {
      "epoch": 0.875366491687003,
      "grad_norm": 0.06261192103594504,
      "learning_rate": 0.00018801760550960178,
      "loss": 0.8497,
      "step": 49860
    },
    {
      "epoch": 0.8755420565670043,
      "grad_norm": 0.06571108046665391,
      "learning_rate": 0.00018801229072740451,
      "loss": 0.8559,
      "step": 49870
    },
    {
      "epoch": 0.8757176214470057,
      "grad_norm": 0.09122124643947176,
      "learning_rate": 0.00018800697484274486,
      "loss": 0.8493,
      "step": 49880
    },
    {
      "epoch": 0.8758931863270072,
      "grad_norm": 0.06766242905017707,
      "learning_rate": 0.00018800165785569017,
      "loss": 0.8638,
      "step": 49890
    },
    {
      "epoch": 0.8760687512070086,
      "grad_norm": 0.06343494440466516,
      "learning_rate": 0.00018799633976630776,
      "loss": 0.8512,
      "step": 49900
    },
    {
      "epoch": 0.8762443160870099,
      "grad_norm": 0.06071990326635008,
      "learning_rate": 0.00018799102057466508,
      "loss": 0.8475,
      "step": 49910
    },
    {
      "epoch": 0.8764198809670114,
      "grad_norm": 0.04962592179579309,
      "learning_rate": 0.00018798570028082952,
      "loss": 0.8505,
      "step": 49920
    },
    {
      "epoch": 0.8765954458470128,
      "grad_norm": 0.06787401383947338,
      "learning_rate": 0.00018798037888486844,
      "loss": 0.8485,
      "step": 49930
    },
    {
      "epoch": 0.8767710107270141,
      "grad_norm": 0.06459568030407734,
      "learning_rate": 0.00018797505638684927,
      "loss": 0.8595,
      "step": 49940
    },
    {
      "epoch": 0.8769465756070156,
      "grad_norm": 0.12618848710398378,
      "learning_rate": 0.00018796973278683952,
      "loss": 0.8517,
      "step": 49950
    },
    {
      "epoch": 0.877122140487017,
      "grad_norm": 0.07333752737059247,
      "learning_rate": 0.00018796440808490654,
      "loss": 0.855,
      "step": 49960
    },
    {
      "epoch": 0.8772977053670183,
      "grad_norm": 0.05114884299288853,
      "learning_rate": 0.00018795908228111787,
      "loss": 0.8485,
      "step": 49970
    },
    {
      "epoch": 0.8774732702470198,
      "grad_norm": 0.07377974190241317,
      "learning_rate": 0.00018795375537554093,
      "loss": 0.8559,
      "step": 49980
    },
    {
      "epoch": 0.8776488351270212,
      "grad_norm": 0.06300737898558817,
      "learning_rate": 0.0001879484273682433,
      "loss": 0.8484,
      "step": 49990
    },
    {
      "epoch": 0.8778244000070226,
      "grad_norm": 0.07291594808784989,
      "learning_rate": 0.0001879430982592924,
      "loss": 0.8571,
      "step": 50000
    },
    {
      "epoch": 0.877999964887024,
      "grad_norm": 0.05935889173319087,
      "learning_rate": 0.00018793776804875578,
      "loss": 0.8572,
      "step": 50010
    },
    {
      "epoch": 0.8781755297670254,
      "grad_norm": 0.05890201175848357,
      "learning_rate": 0.00018793243673670098,
      "loss": 0.8484,
      "step": 50020
    },
    {
      "epoch": 0.8783510946470268,
      "grad_norm": 0.09039413968526228,
      "learning_rate": 0.00018792710432319557,
      "loss": 0.8436,
      "step": 50030
    },
    {
      "epoch": 0.8785266595270282,
      "grad_norm": 0.060327694558921806,
      "learning_rate": 0.000187921770808307,
      "loss": 0.8511,
      "step": 50040
    },
    {
      "epoch": 0.8787022244070296,
      "grad_norm": 0.055699877980463885,
      "learning_rate": 0.00018791643619210302,
      "loss": 0.852,
      "step": 50050
    },
    {
      "epoch": 0.878877789287031,
      "grad_norm": 0.061669739866279036,
      "learning_rate": 0.00018791110047465114,
      "loss": 0.8456,
      "step": 50060
    },
    {
      "epoch": 0.8790533541670325,
      "grad_norm": 0.07473816716979069,
      "learning_rate": 0.00018790576365601894,
      "loss": 0.8538,
      "step": 50070
    },
    {
      "epoch": 0.8792289190470338,
      "grad_norm": 0.06806125926994747,
      "learning_rate": 0.00018790042573627405,
      "loss": 0.8474,
      "step": 50080
    },
    {
      "epoch": 0.8794044839270352,
      "grad_norm": 0.07498377086572416,
      "learning_rate": 0.0001878950867154841,
      "loss": 0.8513,
      "step": 50090
    },
    {
      "epoch": 0.8795800488070367,
      "grad_norm": 0.07729645614303624,
      "learning_rate": 0.00018788974659371676,
      "loss": 0.8494,
      "step": 50100
    },
    {
      "epoch": 0.879755613687038,
      "grad_norm": 0.05279082736160683,
      "learning_rate": 0.0001878844053710397,
      "loss": 0.8591,
      "step": 50110
    },
    {
      "epoch": 0.8799311785670394,
      "grad_norm": 0.07782662190638033,
      "learning_rate": 0.00018787906304752055,
      "loss": 0.848,
      "step": 50120
    },
    {
      "epoch": 0.8801067434470409,
      "grad_norm": 0.05097926675055242,
      "learning_rate": 0.00018787371962322702,
      "loss": 0.8507,
      "step": 50130
    },
    {
      "epoch": 0.8802823083270422,
      "grad_norm": 0.06309680642501553,
      "learning_rate": 0.00018786837509822684,
      "loss": 0.8568,
      "step": 50140
    },
    {
      "epoch": 0.8804578732070437,
      "grad_norm": 0.059906092486743764,
      "learning_rate": 0.00018786302947258767,
      "loss": 0.8513,
      "step": 50150
    },
    {
      "epoch": 0.8806334380870451,
      "grad_norm": 0.08045060186324872,
      "learning_rate": 0.00018785768274637724,
      "loss": 0.8539,
      "step": 50160
    },
    {
      "epoch": 0.8808090029670465,
      "grad_norm": 0.05841389111604709,
      "learning_rate": 0.0001878523349196634,
      "loss": 0.8527,
      "step": 50170
    },
    {
      "epoch": 0.8809845678470479,
      "grad_norm": 0.062025225657058695,
      "learning_rate": 0.0001878469859925138,
      "loss": 0.8508,
      "step": 50180
    },
    {
      "epoch": 0.8811601327270493,
      "grad_norm": 0.0563731986835495,
      "learning_rate": 0.00018784163596499625,
      "loss": 0.8486,
      "step": 50190
    },
    {
      "epoch": 0.8813356976070507,
      "grad_norm": 0.06720194979955521,
      "learning_rate": 0.0001878362848371785,
      "loss": 0.8495,
      "step": 50200
    },
    {
      "epoch": 0.8815112624870521,
      "grad_norm": 0.06572342232097442,
      "learning_rate": 0.00018783093260912837,
      "loss": 0.8455,
      "step": 50210
    },
    {
      "epoch": 0.8816868273670535,
      "grad_norm": 0.058304307072872104,
      "learning_rate": 0.0001878255792809137,
      "loss": 0.859,
      "step": 50220
    },
    {
      "epoch": 0.8818623922470549,
      "grad_norm": 0.05079681376879491,
      "learning_rate": 0.00018782022485260233,
      "loss": 0.863,
      "step": 50230
    },
    {
      "epoch": 0.8820379571270563,
      "grad_norm": 0.048737427759766604,
      "learning_rate": 0.00018781486932426206,
      "loss": 0.8578,
      "step": 50240
    },
    {
      "epoch": 0.8822135220070577,
      "grad_norm": 0.06922719626290813,
      "learning_rate": 0.0001878095126959607,
      "loss": 0.8516,
      "step": 50250
    },
    {
      "epoch": 0.8823890868870591,
      "grad_norm": 0.0658364786770367,
      "learning_rate": 0.00018780415496776626,
      "loss": 0.8513,
      "step": 50260
    },
    {
      "epoch": 0.8825646517670606,
      "grad_norm": 0.08600908390717305,
      "learning_rate": 0.0001877987961397465,
      "loss": 0.8562,
      "step": 50270
    },
    {
      "epoch": 0.8827402166470619,
      "grad_norm": 0.048597086375931145,
      "learning_rate": 0.00018779343621196936,
      "loss": 0.8582,
      "step": 50280
    },
    {
      "epoch": 0.8829157815270633,
      "grad_norm": 0.06655191472683604,
      "learning_rate": 0.00018778807518450276,
      "loss": 0.8511,
      "step": 50290
    },
    {
      "epoch": 0.8830913464070648,
      "grad_norm": 0.06844144405246301,
      "learning_rate": 0.00018778271305741463,
      "loss": 0.8572,
      "step": 50300
    },
    {
      "epoch": 0.8832669112870661,
      "grad_norm": 0.060747148464578835,
      "learning_rate": 0.00018777734983077283,
      "loss": 0.8544,
      "step": 50310
    },
    {
      "epoch": 0.8834424761670675,
      "grad_norm": 0.05463920115207388,
      "learning_rate": 0.00018777198550464543,
      "loss": 0.8422,
      "step": 50320
    },
    {
      "epoch": 0.883618041047069,
      "grad_norm": 0.07672073816462882,
      "learning_rate": 0.00018776662007910036,
      "loss": 0.8437,
      "step": 50330
    },
    {
      "epoch": 0.8837936059270703,
      "grad_norm": 0.08209000338779424,
      "learning_rate": 0.00018776125355420554,
      "loss": 0.8551,
      "step": 50340
    },
    {
      "epoch": 0.8839691708070717,
      "grad_norm": 0.0629576260133559,
      "learning_rate": 0.00018775588593002905,
      "loss": 0.8382,
      "step": 50350
    },
    {
      "epoch": 0.8841447356870732,
      "grad_norm": 0.07900218299967032,
      "learning_rate": 0.0001877505172066388,
      "loss": 0.8533,
      "step": 50360
    },
    {
      "epoch": 0.8843203005670746,
      "grad_norm": 0.06344414424730535,
      "learning_rate": 0.00018774514738410293,
      "loss": 0.8556,
      "step": 50370
    },
    {
      "epoch": 0.8844958654470759,
      "grad_norm": 0.06498147851017197,
      "learning_rate": 0.00018773977646248938,
      "loss": 0.8463,
      "step": 50380
    },
    {
      "epoch": 0.8846714303270774,
      "grad_norm": 0.056786249767475636,
      "learning_rate": 0.00018773440444186626,
      "loss": 0.8466,
      "step": 50390
    },
    {
      "epoch": 0.8848469952070788,
      "grad_norm": 0.0719349882029267,
      "learning_rate": 0.0001877290313223016,
      "loss": 0.8489,
      "step": 50400
    },
    {
      "epoch": 0.8850225600870801,
      "grad_norm": 0.08767560213562822,
      "learning_rate": 0.0001877236571038635,
      "loss": 0.8468,
      "step": 50410
    },
    {
      "epoch": 0.8851981249670816,
      "grad_norm": 0.07095625761398223,
      "learning_rate": 0.00018771828178662002,
      "loss": 0.8547,
      "step": 50420
    },
    {
      "epoch": 0.885373689847083,
      "grad_norm": 0.07453826509152775,
      "learning_rate": 0.00018771290537063933,
      "loss": 0.8574,
      "step": 50430
    },
    {
      "epoch": 0.8855492547270843,
      "grad_norm": 0.08598926144800784,
      "learning_rate": 0.0001877075278559895,
      "loss": 0.852,
      "step": 50440
    },
    {
      "epoch": 0.8857248196070858,
      "grad_norm": 0.05124587250438287,
      "learning_rate": 0.00018770214924273865,
      "loss": 0.8578,
      "step": 50450
    },
    {
      "epoch": 0.8859003844870872,
      "grad_norm": 0.058731166838281376,
      "learning_rate": 0.000187696769530955,
      "loss": 0.8614,
      "step": 50460
    },
    {
      "epoch": 0.8860759493670886,
      "grad_norm": 0.0588161563323831,
      "learning_rate": 0.00018769138872070664,
      "loss": 0.8582,
      "step": 50470
    },
    {
      "epoch": 0.88625151424709,
      "grad_norm": 0.061077447572731924,
      "learning_rate": 0.00018768600681206174,
      "loss": 0.8556,
      "step": 50480
    },
    {
      "epoch": 0.8864270791270914,
      "grad_norm": 0.08074596310023444,
      "learning_rate": 0.00018768062380508855,
      "loss": 0.8466,
      "step": 50490
    },
    {
      "epoch": 0.8866026440070928,
      "grad_norm": 0.05960405245509958,
      "learning_rate": 0.00018767523969985523,
      "loss": 0.8503,
      "step": 50500
    },
    {
      "epoch": 0.8867782088870942,
      "grad_norm": 0.05011262738212449,
      "learning_rate": 0.00018766985449643003,
      "loss": 0.8386,
      "step": 50510
    },
    {
      "epoch": 0.8869537737670956,
      "grad_norm": 0.07195515317225881,
      "learning_rate": 0.00018766446819488115,
      "loss": 0.8534,
      "step": 50520
    },
    {
      "epoch": 0.8871293386470971,
      "grad_norm": 0.057774067965602846,
      "learning_rate": 0.00018765908079527687,
      "loss": 0.8515,
      "step": 50530
    },
    {
      "epoch": 0.8873049035270985,
      "grad_norm": 0.055419256095997925,
      "learning_rate": 0.0001876536922976854,
      "loss": 0.8625,
      "step": 50540
    },
    {
      "epoch": 0.8874804684070998,
      "grad_norm": 0.05743085419748505,
      "learning_rate": 0.00018764830270217507,
      "loss": 0.8538,
      "step": 50550
    },
    {
      "epoch": 0.8876560332871013,
      "grad_norm": 0.0657729277511328,
      "learning_rate": 0.0001876429120088141,
      "loss": 0.8535,
      "step": 50560
    },
    {
      "epoch": 0.8878315981671027,
      "grad_norm": 0.0684849413667164,
      "learning_rate": 0.00018763752021767088,
      "loss": 0.8548,
      "step": 50570
    },
    {
      "epoch": 0.888007163047104,
      "grad_norm": 0.09752218769112139,
      "learning_rate": 0.0001876321273288136,
      "loss": 0.858,
      "step": 50580
    },
    {
      "epoch": 0.8881827279271055,
      "grad_norm": 0.055367785973738114,
      "learning_rate": 0.00018762673334231074,
      "loss": 0.8421,
      "step": 50590
    },
    {
      "epoch": 0.8883582928071069,
      "grad_norm": 0.06489605138795157,
      "learning_rate": 0.00018762133825823055,
      "loss": 0.8417,
      "step": 50600
    },
    {
      "epoch": 0.8885338576871082,
      "grad_norm": 0.059301369840885396,
      "learning_rate": 0.00018761594207664139,
      "loss": 0.8555,
      "step": 50610
    },
    {
      "epoch": 0.8887094225671097,
      "grad_norm": 0.08343677483142059,
      "learning_rate": 0.00018761054479761162,
      "loss": 0.8569,
      "step": 50620
    },
    {
      "epoch": 0.8888849874471111,
      "grad_norm": 0.07230571919006619,
      "learning_rate": 0.0001876051464212097,
      "loss": 0.8509,
      "step": 50630
    },
    {
      "epoch": 0.8890605523271125,
      "grad_norm": 0.06441246691466639,
      "learning_rate": 0.00018759974694750392,
      "loss": 0.8575,
      "step": 50640
    },
    {
      "epoch": 0.8892361172071139,
      "grad_norm": 0.07488378489975694,
      "learning_rate": 0.0001875943463765628,
      "loss": 0.8529,
      "step": 50650
    },
    {
      "epoch": 0.8894116820871153,
      "grad_norm": 0.09454160678605143,
      "learning_rate": 0.0001875889447084547,
      "loss": 0.8439,
      "step": 50660
    },
    {
      "epoch": 0.8895872469671167,
      "grad_norm": 0.055148941893185525,
      "learning_rate": 0.00018758354194324805,
      "loss": 0.8559,
      "step": 50670
    },
    {
      "epoch": 0.8897628118471181,
      "grad_norm": 0.07033487732007443,
      "learning_rate": 0.00018757813808101137,
      "loss": 0.8451,
      "step": 50680
    },
    {
      "epoch": 0.8899383767271195,
      "grad_norm": 0.054140078983721036,
      "learning_rate": 0.00018757273312181307,
      "loss": 0.8492,
      "step": 50690
    },
    {
      "epoch": 0.8901139416071209,
      "grad_norm": 0.07438392329279504,
      "learning_rate": 0.00018756732706572165,
      "loss": 0.8492,
      "step": 50700
    },
    {
      "epoch": 0.8902895064871224,
      "grad_norm": 0.06166652158062049,
      "learning_rate": 0.00018756191991280561,
      "loss": 0.8491,
      "step": 50710
    },
    {
      "epoch": 0.8904650713671237,
      "grad_norm": 0.06462639490047087,
      "learning_rate": 0.0001875565116631334,
      "loss": 0.842,
      "step": 50720
    },
    {
      "epoch": 0.8906406362471251,
      "grad_norm": 0.04833005144601604,
      "learning_rate": 0.00018755110231677367,
      "loss": 0.851,
      "step": 50730
    },
    {
      "epoch": 0.8908162011271266,
      "grad_norm": 0.05006604313706428,
      "learning_rate": 0.00018754569187379486,
      "loss": 0.85,
      "step": 50740
    },
    {
      "epoch": 0.8909917660071279,
      "grad_norm": 0.11286663825746192,
      "learning_rate": 0.00018754028033426558,
      "loss": 0.8491,
      "step": 50750
    },
    {
      "epoch": 0.8911673308871293,
      "grad_norm": 0.07399246488788726,
      "learning_rate": 0.0001875348676982543,
      "loss": 0.8483,
      "step": 50760
    },
    {
      "epoch": 0.8913428957671308,
      "grad_norm": 0.05764747832325426,
      "learning_rate": 0.00018752945396582972,
      "loss": 0.8516,
      "step": 50770
    },
    {
      "epoch": 0.8915184606471321,
      "grad_norm": 0.05413863695093473,
      "learning_rate": 0.00018752403913706035,
      "loss": 0.8521,
      "step": 50780
    },
    {
      "epoch": 0.8916940255271335,
      "grad_norm": 0.061223666129464796,
      "learning_rate": 0.0001875186232120148,
      "loss": 0.8492,
      "step": 50790
    },
    {
      "epoch": 0.891869590407135,
      "grad_norm": 0.07872445302798205,
      "learning_rate": 0.00018751320619076175,
      "loss": 0.851,
      "step": 50800
    },
    {
      "epoch": 0.8920451552871363,
      "grad_norm": 0.07440618035702816,
      "learning_rate": 0.00018750778807336978,
      "loss": 0.8539,
      "step": 50810
    },
    {
      "epoch": 0.8922207201671377,
      "grad_norm": 0.07819704553344695,
      "learning_rate": 0.00018750236885990758,
      "loss": 0.854,
      "step": 50820
    },
    {
      "epoch": 0.8923962850471392,
      "grad_norm": 0.05564502300218891,
      "learning_rate": 0.00018749694855044378,
      "loss": 0.8539,
      "step": 50830
    },
    {
      "epoch": 0.8925718499271406,
      "grad_norm": 0.07229397836362848,
      "learning_rate": 0.00018749152714504704,
      "loss": 0.8507,
      "step": 50840
    },
    {
      "epoch": 0.8927474148071419,
      "grad_norm": 0.06618759003450547,
      "learning_rate": 0.0001874861046437861,
      "loss": 0.853,
      "step": 50850
    },
    {
      "epoch": 0.8929229796871434,
      "grad_norm": 0.0662224243557811,
      "learning_rate": 0.0001874806810467296,
      "loss": 0.8496,
      "step": 50860
    },
    {
      "epoch": 0.8930985445671448,
      "grad_norm": 0.079592667915976,
      "learning_rate": 0.00018747525635394633,
      "loss": 0.8562,
      "step": 50870
    },
    {
      "epoch": 0.8932741094471462,
      "grad_norm": 0.0537661633281545,
      "learning_rate": 0.00018746983056550497,
      "loss": 0.8455,
      "step": 50880
    },
    {
      "epoch": 0.8934496743271476,
      "grad_norm": 0.07627089021108606,
      "learning_rate": 0.00018746440368147432,
      "loss": 0.8502,
      "step": 50890
    },
    {
      "epoch": 0.893625239207149,
      "grad_norm": 0.07235968532612647,
      "learning_rate": 0.00018745897570192306,
      "loss": 0.8555,
      "step": 50900
    },
    {
      "epoch": 0.8938008040871505,
      "grad_norm": 0.0635589775961434,
      "learning_rate": 0.00018745354662692005,
      "loss": 0.8473,
      "step": 50910
    },
    {
      "epoch": 0.8939763689671518,
      "grad_norm": 0.07101162408551163,
      "learning_rate": 0.000187448116456534,
      "loss": 0.8548,
      "step": 50920
    },
    {
      "epoch": 0.8941519338471532,
      "grad_norm": 0.07490168341255175,
      "learning_rate": 0.0001874426851908338,
      "loss": 0.8608,
      "step": 50930
    },
    {
      "epoch": 0.8943274987271547,
      "grad_norm": 0.060437415146500244,
      "learning_rate": 0.00018743725282988814,
      "loss": 0.8542,
      "step": 50940
    },
    {
      "epoch": 0.894503063607156,
      "grad_norm": 0.060413964959036344,
      "learning_rate": 0.00018743181937376597,
      "loss": 0.8517,
      "step": 50950
    },
    {
      "epoch": 0.8946786284871574,
      "grad_norm": 0.05364845008372144,
      "learning_rate": 0.00018742638482253606,
      "loss": 0.8424,
      "step": 50960
    },
    {
      "epoch": 0.8948541933671589,
      "grad_norm": 0.05161128907783689,
      "learning_rate": 0.00018742094917626728,
      "loss": 0.8536,
      "step": 50970
    },
    {
      "epoch": 0.8950297582471602,
      "grad_norm": 0.06004069975335875,
      "learning_rate": 0.00018741551243502857,
      "loss": 0.8489,
      "step": 50980
    },
    {
      "epoch": 0.8952053231271616,
      "grad_norm": 0.06174598660753231,
      "learning_rate": 0.0001874100745988887,
      "loss": 0.8528,
      "step": 50990
    },
    {
      "epoch": 0.8953808880071631,
      "grad_norm": 0.06817657238512229,
      "learning_rate": 0.00018740463566791667,
      "loss": 0.8482,
      "step": 51000
    },
    {
      "epoch": 0.8955564528871645,
      "grad_norm": 0.07788965075578769,
      "learning_rate": 0.0001873991956421813,
      "loss": 0.8592,
      "step": 51010
    },
    {
      "epoch": 0.8957320177671658,
      "grad_norm": 0.06630833845797687,
      "learning_rate": 0.00018739375452175158,
      "loss": 0.8533,
      "step": 51020
    },
    {
      "epoch": 0.8959075826471673,
      "grad_norm": 0.0894381070079055,
      "learning_rate": 0.00018738831230669643,
      "loss": 0.8605,
      "step": 51030
    },
    {
      "epoch": 0.8960831475271687,
      "grad_norm": 0.06316342236467318,
      "learning_rate": 0.00018738286899708484,
      "loss": 0.854,
      "step": 51040
    },
    {
      "epoch": 0.89625871240717,
      "grad_norm": 0.06632657528081777,
      "learning_rate": 0.00018737742459298568,
      "loss": 0.8563,
      "step": 51050
    },
    {
      "epoch": 0.8964342772871715,
      "grad_norm": 0.08668069840339993,
      "learning_rate": 0.00018737197909446805,
      "loss": 0.8599,
      "step": 51060
    },
    {
      "epoch": 0.8966098421671729,
      "grad_norm": 0.0752222136258506,
      "learning_rate": 0.00018736653250160085,
      "loss": 0.8524,
      "step": 51070
    },
    {
      "epoch": 0.8967854070471742,
      "grad_norm": 0.0715680055744508,
      "learning_rate": 0.0001873610848144532,
      "loss": 0.8535,
      "step": 51080
    },
    {
      "epoch": 0.8969609719271757,
      "grad_norm": 0.08335870179528601,
      "learning_rate": 0.00018735563603309397,
      "loss": 0.8565,
      "step": 51090
    },
    {
      "epoch": 0.8971365368071771,
      "grad_norm": 0.04910302970986397,
      "learning_rate": 0.00018735018615759233,
      "loss": 0.8543,
      "step": 51100
    },
    {
      "epoch": 0.8973121016871785,
      "grad_norm": 0.05952412430306795,
      "learning_rate": 0.0001873447351880173,
      "loss": 0.8531,
      "step": 51110
    },
    {
      "epoch": 0.8974876665671799,
      "grad_norm": 0.07575401254822917,
      "learning_rate": 0.00018733928312443787,
      "loss": 0.8511,
      "step": 51120
    },
    {
      "epoch": 0.8976632314471813,
      "grad_norm": 0.05587606010562313,
      "learning_rate": 0.00018733382996692324,
      "loss": 0.8447,
      "step": 51130
    },
    {
      "epoch": 0.8978387963271827,
      "grad_norm": 0.07973054913717782,
      "learning_rate": 0.00018732837571554237,
      "loss": 0.8558,
      "step": 51140
    },
    {
      "epoch": 0.8980143612071841,
      "grad_norm": 0.05054581546350068,
      "learning_rate": 0.0001873229203703645,
      "loss": 0.8575,
      "step": 51150
    },
    {
      "epoch": 0.8981899260871855,
      "grad_norm": 0.05747105530930816,
      "learning_rate": 0.00018731746393145865,
      "loss": 0.8554,
      "step": 51160
    },
    {
      "epoch": 0.8983654909671869,
      "grad_norm": 0.09504366680475498,
      "learning_rate": 0.000187312006398894,
      "loss": 0.8587,
      "step": 51170
    },
    {
      "epoch": 0.8985410558471884,
      "grad_norm": 0.04989276656763578,
      "learning_rate": 0.00018730654777273967,
      "loss": 0.857,
      "step": 51180
    },
    {
      "epoch": 0.8987166207271897,
      "grad_norm": 0.09837991227506145,
      "learning_rate": 0.00018730108805306483,
      "loss": 0.8524,
      "step": 51190
    },
    {
      "epoch": 0.8988921856071911,
      "grad_norm": 0.06676704434023024,
      "learning_rate": 0.0001872956272399387,
      "loss": 0.8508,
      "step": 51200
    },
    {
      "epoch": 0.8990677504871926,
      "grad_norm": 0.0479009075196238,
      "learning_rate": 0.00018729016533343044,
      "loss": 0.8557,
      "step": 51210
    },
    {
      "epoch": 0.8992433153671939,
      "grad_norm": 0.05177292152460493,
      "learning_rate": 0.00018728470233360923,
      "loss": 0.8519,
      "step": 51220
    },
    {
      "epoch": 0.8994188802471953,
      "grad_norm": 0.058260281732208236,
      "learning_rate": 0.00018727923824054428,
      "loss": 0.8526,
      "step": 51230
    },
    {
      "epoch": 0.8995944451271968,
      "grad_norm": 0.07283496311594259,
      "learning_rate": 0.00018727377305430488,
      "loss": 0.8533,
      "step": 51240
    },
    {
      "epoch": 0.8997700100071981,
      "grad_norm": 0.05769298961233435,
      "learning_rate": 0.00018726830677496024,
      "loss": 0.8583,
      "step": 51250
    },
    {
      "epoch": 0.8999455748871996,
      "grad_norm": 0.08543911422606003,
      "learning_rate": 0.00018726283940257961,
      "loss": 0.8479,
      "step": 51260
    },
    {
      "epoch": 0.900121139767201,
      "grad_norm": 0.05773565404703946,
      "learning_rate": 0.00018725737093723226,
      "loss": 0.8422,
      "step": 51270
    },
    {
      "epoch": 0.9002967046472024,
      "grad_norm": 0.061184935483674284,
      "learning_rate": 0.00018725190137898752,
      "loss": 0.8532,
      "step": 51280
    },
    {
      "epoch": 0.9004722695272038,
      "grad_norm": 0.08730325528106753,
      "learning_rate": 0.00018724643072791464,
      "loss": 0.8472,
      "step": 51290
    },
    {
      "epoch": 0.9006478344072052,
      "grad_norm": 0.05057529637453755,
      "learning_rate": 0.00018724095898408296,
      "loss": 0.8419,
      "step": 51300
    },
    {
      "epoch": 0.9008233992872066,
      "grad_norm": 0.0883348252168898,
      "learning_rate": 0.00018723548614756177,
      "loss": 0.8577,
      "step": 51310
    },
    {
      "epoch": 0.900998964167208,
      "grad_norm": 0.0951062085074144,
      "learning_rate": 0.00018723001221842048,
      "loss": 0.8584,
      "step": 51320
    },
    {
      "epoch": 0.9011745290472094,
      "grad_norm": 0.06836967423866752,
      "learning_rate": 0.0001872245371967284,
      "loss": 0.8558,
      "step": 51330
    },
    {
      "epoch": 0.9013500939272108,
      "grad_norm": 0.07200103460269672,
      "learning_rate": 0.0001872190610825549,
      "loss": 0.853,
      "step": 51340
    },
    {
      "epoch": 0.9015256588072122,
      "grad_norm": 0.081017221593092,
      "learning_rate": 0.0001872135838759694,
      "loss": 0.8524,
      "step": 51350
    },
    {
      "epoch": 0.9017012236872136,
      "grad_norm": 0.06077713550819009,
      "learning_rate": 0.00018720810557704124,
      "loss": 0.8509,
      "step": 51360
    },
    {
      "epoch": 0.901876788567215,
      "grad_norm": 0.07977634291198864,
      "learning_rate": 0.0001872026261858399,
      "loss": 0.8491,
      "step": 51370
    },
    {
      "epoch": 0.9020523534472165,
      "grad_norm": 0.0805959761413738,
      "learning_rate": 0.0001871971457024347,
      "loss": 0.8527,
      "step": 51380
    },
    {
      "epoch": 0.9022279183272178,
      "grad_norm": 0.058383068052664466,
      "learning_rate": 0.0001871916641268952,
      "loss": 0.8525,
      "step": 51390
    },
    {
      "epoch": 0.9024034832072192,
      "grad_norm": 0.06935404418889989,
      "learning_rate": 0.00018718618145929076,
      "loss": 0.8599,
      "step": 51400
    },
    {
      "epoch": 0.9025790480872207,
      "grad_norm": 0.07021228161858148,
      "learning_rate": 0.00018718069769969092,
      "loss": 0.8501,
      "step": 51410
    },
    {
      "epoch": 0.902754612967222,
      "grad_norm": 0.054957455753611203,
      "learning_rate": 0.00018717521284816507,
      "loss": 0.8427,
      "step": 51420
    },
    {
      "epoch": 0.9029301778472234,
      "grad_norm": 0.05722356238779132,
      "learning_rate": 0.00018716972690478284,
      "loss": 0.852,
      "step": 51430
    },
    {
      "epoch": 0.9031057427272249,
      "grad_norm": 0.057422061275585086,
      "learning_rate": 0.0001871642398696136,
      "loss": 0.8484,
      "step": 51440
    },
    {
      "epoch": 0.9032813076072262,
      "grad_norm": 0.071405557652343,
      "learning_rate": 0.00018715875174272692,
      "loss": 0.8492,
      "step": 51450
    },
    {
      "epoch": 0.9034568724872276,
      "grad_norm": 0.06098827878157956,
      "learning_rate": 0.00018715326252419237,
      "loss": 0.8539,
      "step": 51460
    },
    {
      "epoch": 0.9036324373672291,
      "grad_norm": 0.07434703139923607,
      "learning_rate": 0.00018714777221407944,
      "loss": 0.849,
      "step": 51470
    },
    {
      "epoch": 0.9038080022472305,
      "grad_norm": 0.05501481372601671,
      "learning_rate": 0.00018714228081245776,
      "loss": 0.852,
      "step": 51480
    },
    {
      "epoch": 0.9039835671272318,
      "grad_norm": 0.06745530243237322,
      "learning_rate": 0.00018713678831939687,
      "loss": 0.8528,
      "step": 51490
    },
    {
      "epoch": 0.9041591320072333,
      "grad_norm": 0.0917293059279573,
      "learning_rate": 0.00018713129473496635,
      "loss": 0.8521,
      "step": 51500
    },
    {
      "epoch": 0.9043346968872347,
      "grad_norm": 0.06385416650910296,
      "learning_rate": 0.0001871258000592358,
      "loss": 0.8591,
      "step": 51510
    },
    {
      "epoch": 0.904510261767236,
      "grad_norm": 0.0937803135721276,
      "learning_rate": 0.0001871203042922749,
      "loss": 0.8526,
      "step": 51520
    },
    {
      "epoch": 0.9046858266472375,
      "grad_norm": 0.07135346202535937,
      "learning_rate": 0.00018711480743415323,
      "loss": 0.8478,
      "step": 51530
    },
    {
      "epoch": 0.9048613915272389,
      "grad_norm": 0.06604332855784446,
      "learning_rate": 0.00018710930948494044,
      "loss": 0.856,
      "step": 51540
    },
    {
      "epoch": 0.9050369564072402,
      "grad_norm": 0.055563031904983684,
      "learning_rate": 0.00018710381044470618,
      "loss": 0.8511,
      "step": 51550
    },
    {
      "epoch": 0.9052125212872417,
      "grad_norm": 0.04691697813180817,
      "learning_rate": 0.00018709831031352018,
      "loss": 0.8537,
      "step": 51560
    },
    {
      "epoch": 0.9053880861672431,
      "grad_norm": 0.06974836996418313,
      "learning_rate": 0.00018709280909145204,
      "loss": 0.849,
      "step": 51570
    },
    {
      "epoch": 0.9055636510472445,
      "grad_norm": 0.05162057428800077,
      "learning_rate": 0.00018708730677857155,
      "loss": 0.8518,
      "step": 51580
    },
    {
      "epoch": 0.9057392159272459,
      "grad_norm": 0.06168068695560914,
      "learning_rate": 0.00018708180337494835,
      "loss": 0.8523,
      "step": 51590
    },
    {
      "epoch": 0.9059147808072473,
      "grad_norm": 0.06938583524653129,
      "learning_rate": 0.0001870762988806522,
      "loss": 0.8486,
      "step": 51600
    },
    {
      "epoch": 0.9060903456872488,
      "grad_norm": 0.0638898046567099,
      "learning_rate": 0.00018707079329575288,
      "loss": 0.8539,
      "step": 51610
    },
    {
      "epoch": 0.9062659105672501,
      "grad_norm": 0.07018915828672245,
      "learning_rate": 0.0001870652866203201,
      "loss": 0.8534,
      "step": 51620
    },
    {
      "epoch": 0.9064414754472515,
      "grad_norm": 0.046090136216283877,
      "learning_rate": 0.00018705977885442364,
      "loss": 0.857,
      "step": 51630
    },
    {
      "epoch": 0.906617040327253,
      "grad_norm": 0.07037915251362557,
      "learning_rate": 0.00018705426999813328,
      "loss": 0.8453,
      "step": 51640
    },
    {
      "epoch": 0.9067926052072544,
      "grad_norm": 0.07535866724888861,
      "learning_rate": 0.00018704876005151884,
      "loss": 0.8577,
      "step": 51650
    },
    {
      "epoch": 0.9069681700872557,
      "grad_norm": 0.07264116039965367,
      "learning_rate": 0.00018704324901465012,
      "loss": 0.8444,
      "step": 51660
    },
    {
      "epoch": 0.9071437349672572,
      "grad_norm": 0.07257380989003108,
      "learning_rate": 0.00018703773688759695,
      "loss": 0.8482,
      "step": 51670
    },
    {
      "epoch": 0.9073192998472586,
      "grad_norm": 0.06273954295336075,
      "learning_rate": 0.00018703222367042914,
      "loss": 0.8493,
      "step": 51680
    },
    {
      "epoch": 0.9074948647272599,
      "grad_norm": 0.06594074151088294,
      "learning_rate": 0.00018702670936321657,
      "loss": 0.8453,
      "step": 51690
    },
    {
      "epoch": 0.9076704296072614,
      "grad_norm": 0.054860343960016524,
      "learning_rate": 0.00018702119396602914,
      "loss": 0.8592,
      "step": 51700
    },
    {
      "epoch": 0.9078459944872628,
      "grad_norm": 0.07839057289268589,
      "learning_rate": 0.00018701567747893668,
      "loss": 0.8503,
      "step": 51710
    },
    {
      "epoch": 0.9080215593672641,
      "grad_norm": 0.11486753116904445,
      "learning_rate": 0.0001870101599020091,
      "loss": 0.8588,
      "step": 51720
    },
    {
      "epoch": 0.9081971242472656,
      "grad_norm": 0.05417022434882022,
      "learning_rate": 0.00018700464123531628,
      "loss": 0.8496,
      "step": 51730
    },
    {
      "epoch": 0.908372689127267,
      "grad_norm": 0.0743336384380771,
      "learning_rate": 0.0001869991214789282,
      "loss": 0.8463,
      "step": 51740
    },
    {
      "epoch": 0.9085482540072684,
      "grad_norm": 0.07421877683135258,
      "learning_rate": 0.00018699360063291481,
      "loss": 0.85,
      "step": 51750
    },
    {
      "epoch": 0.9087238188872698,
      "grad_norm": 0.0566229441644477,
      "learning_rate": 0.00018698807869734597,
      "loss": 0.8523,
      "step": 51760
    },
    {
      "epoch": 0.9088993837672712,
      "grad_norm": 0.06055307498137061,
      "learning_rate": 0.00018698255567229172,
      "loss": 0.8603,
      "step": 51770
    },
    {
      "epoch": 0.9090749486472726,
      "grad_norm": 0.05978067195809929,
      "learning_rate": 0.00018697703155782198,
      "loss": 0.8563,
      "step": 51780
    },
    {
      "epoch": 0.909250513527274,
      "grad_norm": 0.08855716942748568,
      "learning_rate": 0.0001869715063540068,
      "loss": 0.8521,
      "step": 51790
    },
    {
      "epoch": 0.9094260784072754,
      "grad_norm": 0.053334483267376886,
      "learning_rate": 0.0001869659800609161,
      "loss": 0.8526,
      "step": 51800
    },
    {
      "epoch": 0.9096016432872768,
      "grad_norm": 0.05719583992685535,
      "learning_rate": 0.00018696045267862002,
      "loss": 0.854,
      "step": 51810
    },
    {
      "epoch": 0.9097772081672783,
      "grad_norm": 0.058139479633446775,
      "learning_rate": 0.0001869549242071885,
      "loss": 0.8581,
      "step": 51820
    },
    {
      "epoch": 0.9099527730472796,
      "grad_norm": 0.09909422402032068,
      "learning_rate": 0.00018694939464669166,
      "loss": 0.852,
      "step": 51830
    },
    {
      "epoch": 0.910128337927281,
      "grad_norm": 0.08315666330915147,
      "learning_rate": 0.00018694386399719948,
      "loss": 0.8543,
      "step": 51840
    },
    {
      "epoch": 0.9103039028072825,
      "grad_norm": 0.07227655058015592,
      "learning_rate": 0.00018693833225878208,
      "loss": 0.8561,
      "step": 51850
    },
    {
      "epoch": 0.9104794676872838,
      "grad_norm": 0.07471051064194667,
      "learning_rate": 0.00018693279943150955,
      "loss": 0.8552,
      "step": 51860
    },
    {
      "epoch": 0.9106550325672852,
      "grad_norm": 0.06757823760933587,
      "learning_rate": 0.00018692726551545197,
      "loss": 0.8493,
      "step": 51870
    },
    {
      "epoch": 0.9108305974472867,
      "grad_norm": 0.07208701991578528,
      "learning_rate": 0.00018692173051067945,
      "loss": 0.8533,
      "step": 51880
    },
    {
      "epoch": 0.911006162327288,
      "grad_norm": 0.07689853664374548,
      "learning_rate": 0.00018691619441726218,
      "loss": 0.8501,
      "step": 51890
    },
    {
      "epoch": 0.9111817272072894,
      "grad_norm": 0.05229901664777416,
      "learning_rate": 0.0001869106572352702,
      "loss": 0.8502,
      "step": 51900
    },
    {
      "epoch": 0.9113572920872909,
      "grad_norm": 0.08048460942485705,
      "learning_rate": 0.0001869051189647738,
      "loss": 0.8595,
      "step": 51910
    },
    {
      "epoch": 0.9115328569672922,
      "grad_norm": 0.06252718945206524,
      "learning_rate": 0.00018689957960584303,
      "loss": 0.8522,
      "step": 51920
    },
    {
      "epoch": 0.9117084218472936,
      "grad_norm": 0.06420993926211635,
      "learning_rate": 0.00018689403915854815,
      "loss": 0.8556,
      "step": 51930
    },
    {
      "epoch": 0.9118839867272951,
      "grad_norm": 0.05972521515844701,
      "learning_rate": 0.00018688849762295932,
      "loss": 0.8507,
      "step": 51940
    },
    {
      "epoch": 0.9120595516072965,
      "grad_norm": 0.08916823560454058,
      "learning_rate": 0.00018688295499914677,
      "loss": 0.8533,
      "step": 51950
    },
    {
      "epoch": 0.9122351164872978,
      "grad_norm": 0.047221112158325876,
      "learning_rate": 0.0001868774112871807,
      "loss": 0.8509,
      "step": 51960
    },
    {
      "epoch": 0.9124106813672993,
      "grad_norm": 0.05885773791636452,
      "learning_rate": 0.0001868718664871314,
      "loss": 0.8502,
      "step": 51970
    },
    {
      "epoch": 0.9125862462473007,
      "grad_norm": 0.07393128542937837,
      "learning_rate": 0.0001868663205990691,
      "loss": 0.8506,
      "step": 51980
    },
    {
      "epoch": 0.9127618111273021,
      "grad_norm": 0.053474003204453106,
      "learning_rate": 0.00018686077362306403,
      "loss": 0.857,
      "step": 51990
    },
    {
      "epoch": 0.9129373760073035,
      "grad_norm": 0.07016446383066285,
      "learning_rate": 0.00018685522555918652,
      "loss": 0.8498,
      "step": 52000
    },
    {
      "epoch": 0.9131129408873049,
      "grad_norm": 0.06450010188599384,
      "learning_rate": 0.00018684967640750685,
      "loss": 0.8565,
      "step": 52010
    },
    {
      "epoch": 0.9132885057673064,
      "grad_norm": 0.059383057978195035,
      "learning_rate": 0.00018684412616809531,
      "loss": 0.8545,
      "step": 52020
    },
    {
      "epoch": 0.9134640706473077,
      "grad_norm": 0.06055350872877517,
      "learning_rate": 0.00018683857484102224,
      "loss": 0.8515,
      "step": 52030
    },
    {
      "epoch": 0.9136396355273091,
      "grad_norm": 0.07731295172914557,
      "learning_rate": 0.000186833022426358,
      "loss": 0.8608,
      "step": 52040
    },
    {
      "epoch": 0.9138152004073106,
      "grad_norm": 0.08358581128355994,
      "learning_rate": 0.0001868274689241729,
      "loss": 0.8487,
      "step": 52050
    },
    {
      "epoch": 0.9139907652873119,
      "grad_norm": 0.07780389240433859,
      "learning_rate": 0.00018682191433453733,
      "loss": 0.8532,
      "step": 52060
    },
    {
      "epoch": 0.9141663301673133,
      "grad_norm": 0.06371081858937198,
      "learning_rate": 0.00018681635865752167,
      "loss": 0.8527,
      "step": 52070
    },
    {
      "epoch": 0.9143418950473148,
      "grad_norm": 0.0649971497353415,
      "learning_rate": 0.00018681080189319625,
      "loss": 0.8564,
      "step": 52080
    },
    {
      "epoch": 0.9145174599273161,
      "grad_norm": 0.06240742809624101,
      "learning_rate": 0.0001868052440416316,
      "loss": 0.8484,
      "step": 52090
    },
    {
      "epoch": 0.9146930248073175,
      "grad_norm": 0.06926152662338432,
      "learning_rate": 0.00018679968510289797,
      "loss": 0.8468,
      "step": 52100
    },
    {
      "epoch": 0.914868589687319,
      "grad_norm": 0.07162782561706299,
      "learning_rate": 0.00018679412507706597,
      "loss": 0.8466,
      "step": 52110
    },
    {
      "epoch": 0.9150441545673204,
      "grad_norm": 0.06552992019817752,
      "learning_rate": 0.0001867885639642059,
      "loss": 0.8559,
      "step": 52120
    },
    {
      "epoch": 0.9152197194473217,
      "grad_norm": 0.06049683167712435,
      "learning_rate": 0.00018678300176438832,
      "loss": 0.8607,
      "step": 52130
    },
    {
      "epoch": 0.9153952843273232,
      "grad_norm": 0.05973556817573839,
      "learning_rate": 0.00018677743847768366,
      "loss": 0.8547,
      "step": 52140
    },
    {
      "epoch": 0.9155708492073246,
      "grad_norm": 0.06023576213489868,
      "learning_rate": 0.00018677187410416244,
      "loss": 0.8475,
      "step": 52150
    },
    {
      "epoch": 0.9157464140873259,
      "grad_norm": 0.09195817309164189,
      "learning_rate": 0.00018676630864389508,
      "loss": 0.8542,
      "step": 52160
    },
    {
      "epoch": 0.9159219789673274,
      "grad_norm": 0.05982592856083625,
      "learning_rate": 0.0001867607420969522,
      "loss": 0.8526,
      "step": 52170
    },
    {
      "epoch": 0.9160975438473288,
      "grad_norm": 0.07149588907916492,
      "learning_rate": 0.00018675517446340426,
      "loss": 0.8482,
      "step": 52180
    },
    {
      "epoch": 0.9162731087273301,
      "grad_norm": 0.08561797680253087,
      "learning_rate": 0.00018674960574332182,
      "loss": 0.8425,
      "step": 52190
    },
    {
      "epoch": 0.9164486736073316,
      "grad_norm": 0.07025912956294914,
      "learning_rate": 0.00018674403593677542,
      "loss": 0.8604,
      "step": 52200
    },
    {
      "epoch": 0.916624238487333,
      "grad_norm": 0.05957180592209723,
      "learning_rate": 0.0001867384650438357,
      "loss": 0.8562,
      "step": 52210
    },
    {
      "epoch": 0.9167998033673344,
      "grad_norm": 0.07559596662279224,
      "learning_rate": 0.00018673289306457322,
      "loss": 0.847,
      "step": 52220
    },
    {
      "epoch": 0.9169753682473358,
      "grad_norm": 0.054846149774135565,
      "learning_rate": 0.00018672731999905848,
      "loss": 0.8533,
      "step": 52230
    },
    {
      "epoch": 0.9171509331273372,
      "grad_norm": 0.09725345824588667,
      "learning_rate": 0.00018672174584736221,
      "loss": 0.854,
      "step": 52240
    },
    {
      "epoch": 0.9173264980073386,
      "grad_norm": 0.05208071014157818,
      "learning_rate": 0.00018671617060955499,
      "loss": 0.8494,
      "step": 52250
    },
    {
      "epoch": 0.91750206288734,
      "grad_norm": 0.06516121981306885,
      "learning_rate": 0.00018671059428570743,
      "loss": 0.8559,
      "step": 52260
    },
    {
      "epoch": 0.9176776277673414,
      "grad_norm": 0.06704205479096088,
      "learning_rate": 0.00018670501687589023,
      "loss": 0.8522,
      "step": 52270
    },
    {
      "epoch": 0.9178531926473428,
      "grad_norm": 0.1143323543513422,
      "learning_rate": 0.00018669943838017406,
      "loss": 0.8499,
      "step": 52280
    },
    {
      "epoch": 0.9180287575273443,
      "grad_norm": 0.06985447247199748,
      "learning_rate": 0.00018669385879862958,
      "loss": 0.8553,
      "step": 52290
    },
    {
      "epoch": 0.9182043224073456,
      "grad_norm": 0.06399618804346029,
      "learning_rate": 0.00018668827813132744,
      "loss": 0.8478,
      "step": 52300
    },
    {
      "epoch": 0.918379887287347,
      "grad_norm": 0.05594552247362219,
      "learning_rate": 0.00018668269637833846,
      "loss": 0.8505,
      "step": 52310
    },
    {
      "epoch": 0.9185554521673485,
      "grad_norm": 0.0626233503723022,
      "learning_rate": 0.00018667711353973325,
      "loss": 0.8485,
      "step": 52320
    },
    {
      "epoch": 0.9187310170473498,
      "grad_norm": 0.115000110876983,
      "learning_rate": 0.0001866715296155826,
      "loss": 0.8483,
      "step": 52330
    },
    {
      "epoch": 0.9189065819273513,
      "grad_norm": 0.061610844659925716,
      "learning_rate": 0.00018666594460595729,
      "loss": 0.8562,
      "step": 52340
    },
    {
      "epoch": 0.9190821468073527,
      "grad_norm": 0.06769035752719643,
      "learning_rate": 0.000186660358510928,
      "loss": 0.8502,
      "step": 52350
    },
    {
      "epoch": 0.919257711687354,
      "grad_norm": 0.08160933652269628,
      "learning_rate": 0.00018665477133056555,
      "loss": 0.8542,
      "step": 52360
    },
    {
      "epoch": 0.9194332765673555,
      "grad_norm": 0.06486866876619511,
      "learning_rate": 0.00018664918306494075,
      "loss": 0.8503,
      "step": 52370
    },
    {
      "epoch": 0.9196088414473569,
      "grad_norm": 0.056308127154514574,
      "learning_rate": 0.0001866435937141244,
      "loss": 0.8535,
      "step": 52380
    },
    {
      "epoch": 0.9197844063273583,
      "grad_norm": 0.07429429376100193,
      "learning_rate": 0.00018663800327818725,
      "loss": 0.8624,
      "step": 52390
    },
    {
      "epoch": 0.9199599712073597,
      "grad_norm": 0.06977378954071706,
      "learning_rate": 0.00018663241175720022,
      "loss": 0.8521,
      "step": 52400
    },
    {
      "epoch": 0.9201355360873611,
      "grad_norm": 0.048465608596834416,
      "learning_rate": 0.00018662681915123413,
      "loss": 0.8551,
      "step": 52410
    },
    {
      "epoch": 0.9203111009673625,
      "grad_norm": 0.047029091623290074,
      "learning_rate": 0.00018662122546035981,
      "loss": 0.8452,
      "step": 52420
    },
    {
      "epoch": 0.9204866658473639,
      "grad_norm": 0.0854078755673119,
      "learning_rate": 0.00018661563068464814,
      "loss": 0.8473,
      "step": 52430
    },
    {
      "epoch": 0.9206622307273653,
      "grad_norm": 0.08417946525633584,
      "learning_rate": 0.00018661003482417008,
      "loss": 0.8479,
      "step": 52440
    },
    {
      "epoch": 0.9208377956073667,
      "grad_norm": 0.064585257171811,
      "learning_rate": 0.00018660443787899645,
      "loss": 0.8552,
      "step": 52450
    },
    {
      "epoch": 0.9210133604873681,
      "grad_norm": 0.060016060643486774,
      "learning_rate": 0.00018659883984919817,
      "loss": 0.8496,
      "step": 52460
    },
    {
      "epoch": 0.9211889253673695,
      "grad_norm": 0.05490352423434604,
      "learning_rate": 0.00018659324073484618,
      "loss": 0.8514,
      "step": 52470
    },
    {
      "epoch": 0.9213644902473709,
      "grad_norm": 0.05534406922999104,
      "learning_rate": 0.00018658764053601146,
      "loss": 0.8553,
      "step": 52480
    },
    {
      "epoch": 0.9215400551273724,
      "grad_norm": 0.06739847464312901,
      "learning_rate": 0.00018658203925276488,
      "loss": 0.8543,
      "step": 52490
    },
    {
      "epoch": 0.9217156200073737,
      "grad_norm": 0.0642445125513078,
      "learning_rate": 0.0001865764368851775,
      "loss": 0.8545,
      "step": 52500
    },
    {
      "epoch": 0.9218911848873751,
      "grad_norm": 0.06617560007237798,
      "learning_rate": 0.00018657083343332024,
      "loss": 0.8489,
      "step": 52510
    },
    {
      "epoch": 0.9220667497673766,
      "grad_norm": 0.0627436530575247,
      "learning_rate": 0.00018656522889726413,
      "loss": 0.8451,
      "step": 52520
    },
    {
      "epoch": 0.9222423146473779,
      "grad_norm": 0.0572444305230071,
      "learning_rate": 0.0001865596232770802,
      "loss": 0.8556,
      "step": 52530
    },
    {
      "epoch": 0.9224178795273793,
      "grad_norm": 0.06124502415445265,
      "learning_rate": 0.0001865540165728394,
      "loss": 0.8526,
      "step": 52540
    },
    {
      "epoch": 0.9225934444073808,
      "grad_norm": 0.06357652859601352,
      "learning_rate": 0.00018654840878461283,
      "loss": 0.8566,
      "step": 52550
    },
    {
      "epoch": 0.9227690092873821,
      "grad_norm": 0.05407334702176449,
      "learning_rate": 0.0001865427999124715,
      "loss": 0.8602,
      "step": 52560
    },
    {
      "epoch": 0.9229445741673835,
      "grad_norm": 0.05730982697747094,
      "learning_rate": 0.00018653718995648657,
      "loss": 0.848,
      "step": 52570
    },
    {
      "epoch": 0.923120139047385,
      "grad_norm": 0.06802857569895472,
      "learning_rate": 0.000186531578916729,
      "loss": 0.858,
      "step": 52580
    },
    {
      "epoch": 0.9232957039273864,
      "grad_norm": 0.061650603649277415,
      "learning_rate": 0.00018652596679326995,
      "loss": 0.8531,
      "step": 52590
    },
    {
      "epoch": 0.9234712688073877,
      "grad_norm": 0.05546285941775846,
      "learning_rate": 0.00018652035358618049,
      "loss": 0.8555,
      "step": 52600
    },
    {
      "epoch": 0.9236468336873892,
      "grad_norm": 0.06077111367217984,
      "learning_rate": 0.00018651473929553179,
      "loss": 0.8511,
      "step": 52610
    },
    {
      "epoch": 0.9238223985673906,
      "grad_norm": 0.06595077701663499,
      "learning_rate": 0.00018650912392139491,
      "loss": 0.8487,
      "step": 52620
    },
    {
      "epoch": 0.9239979634473919,
      "grad_norm": 0.07486354344581231,
      "learning_rate": 0.00018650350746384108,
      "loss": 0.8525,
      "step": 52630
    },
    {
      "epoch": 0.9241735283273934,
      "grad_norm": 0.05326577515516487,
      "learning_rate": 0.00018649788992294142,
      "loss": 0.8543,
      "step": 52640
    },
    {
      "epoch": 0.9243490932073948,
      "grad_norm": 0.10552815903550246,
      "learning_rate": 0.0001864922712987671,
      "loss": 0.8451,
      "step": 52650
    },
    {
      "epoch": 0.9245246580873961,
      "grad_norm": 0.061556157793090725,
      "learning_rate": 0.0001864866515913893,
      "loss": 0.8531,
      "step": 52660
    },
    {
      "epoch": 0.9247002229673976,
      "grad_norm": 0.05995617920383463,
      "learning_rate": 0.0001864810308008793,
      "loss": 0.8515,
      "step": 52670
    },
    {
      "epoch": 0.924875787847399,
      "grad_norm": 0.045314775642775244,
      "learning_rate": 0.0001864754089273082,
      "loss": 0.858,
      "step": 52680
    },
    {
      "epoch": 0.9250513527274004,
      "grad_norm": 0.06332290020978949,
      "learning_rate": 0.0001864697859707473,
      "loss": 0.8542,
      "step": 52690
    },
    {
      "epoch": 0.9252269176074018,
      "grad_norm": 0.05300400333697996,
      "learning_rate": 0.00018646416193126783,
      "loss": 0.8574,
      "step": 52700
    },
    {
      "epoch": 0.9254024824874032,
      "grad_norm": 0.050860002156386086,
      "learning_rate": 0.00018645853680894107,
      "loss": 0.8493,
      "step": 52710
    },
    {
      "epoch": 0.9255780473674047,
      "grad_norm": 0.08232432974338241,
      "learning_rate": 0.00018645291060383825,
      "loss": 0.8463,
      "step": 52720
    },
    {
      "epoch": 0.925753612247406,
      "grad_norm": 0.06951914859635001,
      "learning_rate": 0.0001864472833160307,
      "loss": 0.8494,
      "step": 52730
    },
    {
      "epoch": 0.9259291771274074,
      "grad_norm": 0.07185248061447896,
      "learning_rate": 0.00018644165494558968,
      "loss": 0.8589,
      "step": 52740
    },
    {
      "epoch": 0.9261047420074089,
      "grad_norm": 0.06074647979859754,
      "learning_rate": 0.0001864360254925865,
      "loss": 0.853,
      "step": 52750
    },
    {
      "epoch": 0.9262803068874103,
      "grad_norm": 0.09250280504103318,
      "learning_rate": 0.00018643039495709252,
      "loss": 0.8531,
      "step": 52760
    },
    {
      "epoch": 0.9264558717674116,
      "grad_norm": 0.060545783661831984,
      "learning_rate": 0.00018642476333917906,
      "loss": 0.8533,
      "step": 52770
    },
    {
      "epoch": 0.9266314366474131,
      "grad_norm": 0.06100195944831669,
      "learning_rate": 0.00018641913063891747,
      "loss": 0.8504,
      "step": 52780
    },
    {
      "epoch": 0.9268070015274145,
      "grad_norm": 0.054341703083092045,
      "learning_rate": 0.0001864134968563791,
      "loss": 0.8534,
      "step": 52790
    },
    {
      "epoch": 0.9269825664074158,
      "grad_norm": 0.052229302309438404,
      "learning_rate": 0.0001864078619916354,
      "loss": 0.8552,
      "step": 52800
    },
    {
      "epoch": 0.9271581312874173,
      "grad_norm": 0.05071248319481811,
      "learning_rate": 0.0001864022260447577,
      "loss": 0.8545,
      "step": 52810
    },
    {
      "epoch": 0.9273336961674187,
      "grad_norm": 0.065340259361956,
      "learning_rate": 0.00018639658901581748,
      "loss": 0.855,
      "step": 52820
    },
    {
      "epoch": 0.92750926104742,
      "grad_norm": 0.05289256323777533,
      "learning_rate": 0.00018639095090488606,
      "loss": 0.8607,
      "step": 52830
    },
    {
      "epoch": 0.9276848259274215,
      "grad_norm": 0.11482795970416908,
      "learning_rate": 0.00018638531171203494,
      "loss": 0.8547,
      "step": 52840
    },
    {
      "epoch": 0.9278603908074229,
      "grad_norm": 0.09636992999317114,
      "learning_rate": 0.00018637967143733555,
      "loss": 0.8475,
      "step": 52850
    },
    {
      "epoch": 0.9280359556874243,
      "grad_norm": 0.09525292468686201,
      "learning_rate": 0.00018637403008085935,
      "loss": 0.8514,
      "step": 52860
    },
    {
      "epoch": 0.9282115205674257,
      "grad_norm": 0.0873736054843577,
      "learning_rate": 0.00018636838764267782,
      "loss": 0.8577,
      "step": 52870
    },
    {
      "epoch": 0.9283870854474271,
      "grad_norm": 0.09000526149286751,
      "learning_rate": 0.0001863627441228625,
      "loss": 0.8587,
      "step": 52880
    },
    {
      "epoch": 0.9285626503274285,
      "grad_norm": 0.06993243806401504,
      "learning_rate": 0.00018635709952148484,
      "loss": 0.8463,
      "step": 52890
    },
    {
      "epoch": 0.9287382152074299,
      "grad_norm": 0.06881223524912469,
      "learning_rate": 0.00018635145383861634,
      "loss": 0.8443,
      "step": 52900
    },
    {
      "epoch": 0.9289137800874313,
      "grad_norm": 0.06078141111334846,
      "learning_rate": 0.0001863458070743286,
      "loss": 0.8556,
      "step": 52910
    },
    {
      "epoch": 0.9290893449674327,
      "grad_norm": 0.0633397999146511,
      "learning_rate": 0.0001863401592286931,
      "loss": 0.8564,
      "step": 52920
    },
    {
      "epoch": 0.9292649098474342,
      "grad_norm": 0.047030001176000286,
      "learning_rate": 0.00018633451030178143,
      "loss": 0.8507,
      "step": 52930
    },
    {
      "epoch": 0.9294404747274355,
      "grad_norm": 0.05409209278655873,
      "learning_rate": 0.00018632886029366515,
      "loss": 0.8484,
      "step": 52940
    },
    {
      "epoch": 0.9296160396074369,
      "grad_norm": 0.05731771754108553,
      "learning_rate": 0.0001863232092044159,
      "loss": 0.8559,
      "step": 52950
    },
    {
      "epoch": 0.9297916044874384,
      "grad_norm": 0.06289936496465522,
      "learning_rate": 0.0001863175570341052,
      "loss": 0.8492,
      "step": 52960
    },
    {
      "epoch": 0.9299671693674397,
      "grad_norm": 0.05761260188624836,
      "learning_rate": 0.0001863119037828047,
      "loss": 0.8451,
      "step": 52970
    },
    {
      "epoch": 0.9301427342474411,
      "grad_norm": 0.060571632662124356,
      "learning_rate": 0.00018630624945058604,
      "loss": 0.8511,
      "step": 52980
    },
    {
      "epoch": 0.9303182991274426,
      "grad_norm": 0.06875302608137962,
      "learning_rate": 0.00018630059403752084,
      "loss": 0.855,
      "step": 52990
    },
    {
      "epoch": 0.9304938640074439,
      "grad_norm": 0.06422783737825864,
      "learning_rate": 0.00018629493754368078,
      "loss": 0.8579,
      "step": 53000
    },
    {
      "epoch": 0.9306694288874453,
      "grad_norm": 0.07847203680506977,
      "learning_rate": 0.00018628927996913752,
      "loss": 0.8544,
      "step": 53010
    },
    {
      "epoch": 0.9308449937674468,
      "grad_norm": 0.045521412305349236,
      "learning_rate": 0.00018628362131396274,
      "loss": 0.8483,
      "step": 53020
    },
    {
      "epoch": 0.9310205586474481,
      "grad_norm": 0.10020794378759817,
      "learning_rate": 0.0001862779615782281,
      "loss": 0.8593,
      "step": 53030
    },
    {
      "epoch": 0.9311961235274495,
      "grad_norm": 0.08643991627970603,
      "learning_rate": 0.00018627230076200538,
      "loss": 0.8583,
      "step": 53040
    },
    {
      "epoch": 0.931371688407451,
      "grad_norm": 0.06405156262170593,
      "learning_rate": 0.00018626663886536627,
      "loss": 0.8481,
      "step": 53050
    },
    {
      "epoch": 0.9315472532874524,
      "grad_norm": 0.09661888021791752,
      "learning_rate": 0.00018626097588838247,
      "loss": 0.8503,
      "step": 53060
    },
    {
      "epoch": 0.9317228181674538,
      "grad_norm": 0.052934344194335785,
      "learning_rate": 0.00018625531183112578,
      "loss": 0.8585,
      "step": 53070
    },
    {
      "epoch": 0.9318983830474552,
      "grad_norm": 0.08824453175264592,
      "learning_rate": 0.00018624964669366795,
      "loss": 0.8494,
      "step": 53080
    },
    {
      "epoch": 0.9320739479274566,
      "grad_norm": 0.05887456138387104,
      "learning_rate": 0.00018624398047608075,
      "loss": 0.8542,
      "step": 53090
    },
    {
      "epoch": 0.932249512807458,
      "grad_norm": 0.06436235901292511,
      "learning_rate": 0.000186238313178436,
      "loss": 0.8541,
      "step": 53100
    },
    {
      "epoch": 0.9324250776874594,
      "grad_norm": 0.07771575260231488,
      "learning_rate": 0.00018623264480080548,
      "loss": 0.8414,
      "step": 53110
    },
    {
      "epoch": 0.9326006425674608,
      "grad_norm": 0.05298388726971687,
      "learning_rate": 0.00018622697534326102,
      "loss": 0.8536,
      "step": 53120
    },
    {
      "epoch": 0.9327762074474623,
      "grad_norm": 0.05711778790601906,
      "learning_rate": 0.00018622130480587442,
      "loss": 0.85,
      "step": 53130
    },
    {
      "epoch": 0.9329517723274636,
      "grad_norm": 0.05921473950070494,
      "learning_rate": 0.0001862156331887176,
      "loss": 0.8514,
      "step": 53140
    },
    {
      "epoch": 0.933127337207465,
      "grad_norm": 0.10971490506829766,
      "learning_rate": 0.00018620996049186235,
      "loss": 0.8539,
      "step": 53150
    },
    {
      "epoch": 0.9333029020874665,
      "grad_norm": 0.08989465684946801,
      "learning_rate": 0.00018620428671538056,
      "loss": 0.8449,
      "step": 53160
    },
    {
      "epoch": 0.9334784669674678,
      "grad_norm": 0.05870288484527035,
      "learning_rate": 0.00018619861185934416,
      "loss": 0.8541,
      "step": 53170
    },
    {
      "epoch": 0.9336540318474692,
      "grad_norm": 0.07474682224435653,
      "learning_rate": 0.000186192935923825,
      "loss": 0.851,
      "step": 53180
    },
    {
      "epoch": 0.9338295967274707,
      "grad_norm": 0.057768874607653335,
      "learning_rate": 0.00018618725890889504,
      "loss": 0.8541,
      "step": 53190
    },
    {
      "epoch": 0.934005161607472,
      "grad_norm": 0.05308329857234633,
      "learning_rate": 0.00018618158081462613,
      "loss": 0.8543,
      "step": 53200
    },
    {
      "epoch": 0.9341807264874734,
      "grad_norm": 0.059126228287055695,
      "learning_rate": 0.00018617590164109028,
      "loss": 0.8455,
      "step": 53210
    },
    {
      "epoch": 0.9343562913674749,
      "grad_norm": 0.05655763787829159,
      "learning_rate": 0.00018617022138835946,
      "loss": 0.857,
      "step": 53220
    },
    {
      "epoch": 0.9345318562474763,
      "grad_norm": 0.08429012011807922,
      "learning_rate": 0.00018616454005650562,
      "loss": 0.8522,
      "step": 53230
    },
    {
      "epoch": 0.9347074211274776,
      "grad_norm": 0.0762823782775509,
      "learning_rate": 0.0001861588576456007,
      "loss": 0.8613,
      "step": 53240
    },
    {
      "epoch": 0.9348829860074791,
      "grad_norm": 0.08512664281103186,
      "learning_rate": 0.00018615317415571675,
      "loss": 0.8439,
      "step": 53250
    },
    {
      "epoch": 0.9350585508874805,
      "grad_norm": 0.11196634564362304,
      "learning_rate": 0.00018614748958692575,
      "loss": 0.855,
      "step": 53260
    },
    {
      "epoch": 0.9352341157674818,
      "grad_norm": 0.0637948750944818,
      "learning_rate": 0.00018614180393929978,
      "loss": 0.8489,
      "step": 53270
    },
    {
      "epoch": 0.9354096806474833,
      "grad_norm": 0.060337310996855825,
      "learning_rate": 0.00018613611721291079,
      "loss": 0.8487,
      "step": 53280
    },
    {
      "epoch": 0.9355852455274847,
      "grad_norm": 0.047710088107510325,
      "learning_rate": 0.0001861304294078309,
      "loss": 0.8516,
      "step": 53290
    },
    {
      "epoch": 0.935760810407486,
      "grad_norm": 0.06430674545917486,
      "learning_rate": 0.00018612474052413214,
      "loss": 0.8512,
      "step": 53300
    },
    {
      "epoch": 0.9359363752874875,
      "grad_norm": 0.0576265846839811,
      "learning_rate": 0.0001861190505618866,
      "loss": 0.8543,
      "step": 53310
    },
    {
      "epoch": 0.9361119401674889,
      "grad_norm": 0.06616870611682764,
      "learning_rate": 0.00018611335952116643,
      "loss": 0.8486,
      "step": 53320
    },
    {
      "epoch": 0.9362875050474903,
      "grad_norm": 0.07206016524489993,
      "learning_rate": 0.00018610766740204365,
      "loss": 0.8535,
      "step": 53330
    },
    {
      "epoch": 0.9364630699274917,
      "grad_norm": 0.059065362503811676,
      "learning_rate": 0.00018610197420459036,
      "loss": 0.8543,
      "step": 53340
    },
    {
      "epoch": 0.9366386348074931,
      "grad_norm": 0.0593593348028111,
      "learning_rate": 0.0001860962799288788,
      "loss": 0.8452,
      "step": 53350
    },
    {
      "epoch": 0.9368141996874945,
      "grad_norm": 0.07901411604625812,
      "learning_rate": 0.00018609058457498106,
      "loss": 0.8501,
      "step": 53360
    },
    {
      "epoch": 0.9369897645674959,
      "grad_norm": 0.05781975285094038,
      "learning_rate": 0.00018608488814296931,
      "loss": 0.8613,
      "step": 53370
    },
    {
      "epoch": 0.9371653294474973,
      "grad_norm": 0.10291495742599677,
      "learning_rate": 0.00018607919063291572,
      "loss": 0.8529,
      "step": 53380
    },
    {
      "epoch": 0.9373408943274987,
      "grad_norm": 0.05888431010241856,
      "learning_rate": 0.0001860734920448925,
      "loss": 0.8504,
      "step": 53390
    },
    {
      "epoch": 0.9375164592075002,
      "grad_norm": 0.05964158250294826,
      "learning_rate": 0.00018606779237897178,
      "loss": 0.8576,
      "step": 53400
    },
    {
      "epoch": 0.9376920240875015,
      "grad_norm": 0.050981753881048426,
      "learning_rate": 0.00018606209163522585,
      "loss": 0.8598,
      "step": 53410
    },
    {
      "epoch": 0.9378675889675029,
      "grad_norm": 0.07411119415150541,
      "learning_rate": 0.00018605638981372693,
      "loss": 0.8553,
      "step": 53420
    },
    {
      "epoch": 0.9380431538475044,
      "grad_norm": 0.06676003558843394,
      "learning_rate": 0.00018605068691454724,
      "loss": 0.858,
      "step": 53430
    },
    {
      "epoch": 0.9382187187275057,
      "grad_norm": 0.05707785842502874,
      "learning_rate": 0.00018604498293775906,
      "loss": 0.8531,
      "step": 53440
    },
    {
      "epoch": 0.9383942836075072,
      "grad_norm": 0.08261143100189929,
      "learning_rate": 0.00018603927788343462,
      "loss": 0.8507,
      "step": 53450
    },
    {
      "epoch": 0.9385698484875086,
      "grad_norm": 0.06949460310879552,
      "learning_rate": 0.00018603357175164624,
      "loss": 0.8468,
      "step": 53460
    },
    {
      "epoch": 0.9387454133675099,
      "grad_norm": 0.06286742420292224,
      "learning_rate": 0.00018602786454246622,
      "loss": 0.843,
      "step": 53470
    },
    {
      "epoch": 0.9389209782475114,
      "grad_norm": 0.06195322748459346,
      "learning_rate": 0.00018602215625596687,
      "loss": 0.8546,
      "step": 53480
    },
    {
      "epoch": 0.9390965431275128,
      "grad_norm": 0.07864660314444545,
      "learning_rate": 0.00018601644689222047,
      "loss": 0.8625,
      "step": 53490
    },
    {
      "epoch": 0.9392721080075142,
      "grad_norm": 0.06908025779358791,
      "learning_rate": 0.00018601073645129943,
      "loss": 0.851,
      "step": 53500
    },
    {
      "epoch": 0.9394476728875156,
      "grad_norm": 0.06386970298142383,
      "learning_rate": 0.00018600502493327603,
      "loss": 0.8549,
      "step": 53510
    },
    {
      "epoch": 0.939623237767517,
      "grad_norm": 0.08629233935476499,
      "learning_rate": 0.00018599931233822273,
      "loss": 0.8573,
      "step": 53520
    },
    {
      "epoch": 0.9397988026475184,
      "grad_norm": 0.0670888055577602,
      "learning_rate": 0.0001859935986662118,
      "loss": 0.8489,
      "step": 53530
    },
    {
      "epoch": 0.9399743675275198,
      "grad_norm": 0.0690897993475298,
      "learning_rate": 0.0001859878839173157,
      "loss": 0.8452,
      "step": 53540
    },
    {
      "epoch": 0.9401499324075212,
      "grad_norm": 0.07305620813712654,
      "learning_rate": 0.00018598216809160683,
      "loss": 0.8505,
      "step": 53550
    },
    {
      "epoch": 0.9403254972875226,
      "grad_norm": 0.05712524218761124,
      "learning_rate": 0.00018597645118915762,
      "loss": 0.8527,
      "step": 53560
    },
    {
      "epoch": 0.940501062167524,
      "grad_norm": 0.06925092978891154,
      "learning_rate": 0.00018597073321004046,
      "loss": 0.8555,
      "step": 53570
    },
    {
      "epoch": 0.9406766270475254,
      "grad_norm": 0.06234379525228078,
      "learning_rate": 0.00018596501415432781,
      "loss": 0.851,
      "step": 53580
    },
    {
      "epoch": 0.9408521919275268,
      "grad_norm": 0.06176448239982641,
      "learning_rate": 0.0001859592940220922,
      "loss": 0.8518,
      "step": 53590
    },
    {
      "epoch": 0.9410277568075283,
      "grad_norm": 0.06187415056590022,
      "learning_rate": 0.00018595357281340601,
      "loss": 0.8535,
      "step": 53600
    },
    {
      "epoch": 0.9412033216875296,
      "grad_norm": 0.06282952954781876,
      "learning_rate": 0.0001859478505283418,
      "loss": 0.853,
      "step": 53610
    },
    {
      "epoch": 0.941378886567531,
      "grad_norm": 0.058328737323775755,
      "learning_rate": 0.00018594212716697202,
      "loss": 0.8531,
      "step": 53620
    },
    {
      "epoch": 0.9415544514475325,
      "grad_norm": 0.05853801671200862,
      "learning_rate": 0.0001859364027293692,
      "loss": 0.8557,
      "step": 53630
    },
    {
      "epoch": 0.9417300163275338,
      "grad_norm": 0.06946704587874246,
      "learning_rate": 0.00018593067721560592,
      "loss": 0.8489,
      "step": 53640
    },
    {
      "epoch": 0.9419055812075352,
      "grad_norm": 0.05480089898797709,
      "learning_rate": 0.00018592495062575465,
      "loss": 0.8521,
      "step": 53650
    },
    {
      "epoch": 0.9420811460875367,
      "grad_norm": 0.05450382944626004,
      "learning_rate": 0.00018591922295988799,
      "loss": 0.8489,
      "step": 53660
    },
    {
      "epoch": 0.942256710967538,
      "grad_norm": 0.05349484467820491,
      "learning_rate": 0.00018591349421807852,
      "loss": 0.8484,
      "step": 53670
    },
    {
      "epoch": 0.9424322758475394,
      "grad_norm": 0.05031196838143375,
      "learning_rate": 0.0001859077644003988,
      "loss": 0.8556,
      "step": 53680
    },
    {
      "epoch": 0.9426078407275409,
      "grad_norm": 0.06816853456925509,
      "learning_rate": 0.00018590203350692138,
      "loss": 0.8471,
      "step": 53690
    },
    {
      "epoch": 0.9427834056075423,
      "grad_norm": 0.07050087788521052,
      "learning_rate": 0.00018589630153771896,
      "loss": 0.8528,
      "step": 53700
    },
    {
      "epoch": 0.9429589704875436,
      "grad_norm": 0.06468522008614216,
      "learning_rate": 0.00018589056849286414,
      "loss": 0.8487,
      "step": 53710
    },
    {
      "epoch": 0.9431345353675451,
      "grad_norm": 0.057937880956632326,
      "learning_rate": 0.00018588483437242953,
      "loss": 0.8566,
      "step": 53720
    },
    {
      "epoch": 0.9433101002475465,
      "grad_norm": 0.09630393964486945,
      "learning_rate": 0.00018587909917648784,
      "loss": 0.8539,
      "step": 53730
    },
    {
      "epoch": 0.9434856651275478,
      "grad_norm": 0.05994197846211531,
      "learning_rate": 0.00018587336290511164,
      "loss": 0.8518,
      "step": 53740
    },
    {
      "epoch": 0.9436612300075493,
      "grad_norm": 0.06616697370623892,
      "learning_rate": 0.0001858676255583737,
      "loss": 0.8521,
      "step": 53750
    },
    {
      "epoch": 0.9438367948875507,
      "grad_norm": 0.06129151863652551,
      "learning_rate": 0.00018586188713634667,
      "loss": 0.8493,
      "step": 53760
    },
    {
      "epoch": 0.944012359767552,
      "grad_norm": 0.09484877699954966,
      "learning_rate": 0.00018585614763910329,
      "loss": 0.8487,
      "step": 53770
    },
    {
      "epoch": 0.9441879246475535,
      "grad_norm": 0.07234781289816868,
      "learning_rate": 0.0001858504070667162,
      "loss": 0.8589,
      "step": 53780
    },
    {
      "epoch": 0.9443634895275549,
      "grad_norm": 0.05766179244566343,
      "learning_rate": 0.00018584466541925825,
      "loss": 0.8516,
      "step": 53790
    },
    {
      "epoch": 0.9445390544075564,
      "grad_norm": 0.08479379166612562,
      "learning_rate": 0.0001858389226968021,
      "loss": 0.8479,
      "step": 53800
    },
    {
      "epoch": 0.9447146192875577,
      "grad_norm": 0.07212385086767178,
      "learning_rate": 0.00018583317889942053,
      "loss": 0.8505,
      "step": 53810
    },
    {
      "epoch": 0.9448901841675591,
      "grad_norm": 0.06990406462922998,
      "learning_rate": 0.00018582743402718635,
      "loss": 0.8543,
      "step": 53820
    },
    {
      "epoch": 0.9450657490475606,
      "grad_norm": 0.053689948869049736,
      "learning_rate": 0.00018582168808017232,
      "loss": 0.8563,
      "step": 53830
    },
    {
      "epoch": 0.9452413139275619,
      "grad_norm": 0.056469375464078604,
      "learning_rate": 0.00018581594105845123,
      "loss": 0.8496,
      "step": 53840
    },
    {
      "epoch": 0.9454168788075633,
      "grad_norm": 0.06498431017811772,
      "learning_rate": 0.0001858101929620959,
      "loss": 0.8449,
      "step": 53850
    },
    {
      "epoch": 0.9455924436875648,
      "grad_norm": 0.061827404072297615,
      "learning_rate": 0.00018580444379117923,
      "loss": 0.8456,
      "step": 53860
    },
    {
      "epoch": 0.9457680085675662,
      "grad_norm": 0.09084074943204759,
      "learning_rate": 0.00018579869354577396,
      "loss": 0.8484,
      "step": 53870
    },
    {
      "epoch": 0.9459435734475675,
      "grad_norm": 0.06529414863926848,
      "learning_rate": 0.000185792942225953,
      "loss": 0.8533,
      "step": 53880
    },
    {
      "epoch": 0.946119138327569,
      "grad_norm": 0.0666975039986091,
      "learning_rate": 0.0001857871898317892,
      "loss": 0.8532,
      "step": 53890
    },
    {
      "epoch": 0.9462947032075704,
      "grad_norm": 0.06852919544980718,
      "learning_rate": 0.00018578143636335549,
      "loss": 0.8453,
      "step": 53900
    },
    {
      "epoch": 0.9464702680875717,
      "grad_norm": 0.06512431380162538,
      "learning_rate": 0.0001857756818207247,
      "loss": 0.8482,
      "step": 53910
    },
    {
      "epoch": 0.9466458329675732,
      "grad_norm": 0.05353773163831803,
      "learning_rate": 0.0001857699262039698,
      "loss": 0.8475,
      "step": 53920
    },
    {
      "epoch": 0.9468213978475746,
      "grad_norm": 0.05643220526997353,
      "learning_rate": 0.00018576416951316364,
      "loss": 0.8553,
      "step": 53930
    },
    {
      "epoch": 0.9469969627275759,
      "grad_norm": 0.06170024213185728,
      "learning_rate": 0.0001857584117483792,
      "loss": 0.8561,
      "step": 53940
    },
    {
      "epoch": 0.9471725276075774,
      "grad_norm": 0.05894453670279045,
      "learning_rate": 0.00018575265290968955,
      "loss": 0.8533,
      "step": 53950
    },
    {
      "epoch": 0.9473480924875788,
      "grad_norm": 0.0524419208916826,
      "learning_rate": 0.00018574689299716746,
      "loss": 0.8582,
      "step": 53960
    },
    {
      "epoch": 0.9475236573675802,
      "grad_norm": 0.0581366022017881,
      "learning_rate": 0.00018574113201088598,
      "loss": 0.8502,
      "step": 53970
    },
    {
      "epoch": 0.9476992222475816,
      "grad_norm": 0.07667456399782642,
      "learning_rate": 0.00018573536995091815,
      "loss": 0.8496,
      "step": 53980
    },
    {
      "epoch": 0.947874787127583,
      "grad_norm": 0.056329860137378694,
      "learning_rate": 0.00018572960681733695,
      "loss": 0.8525,
      "step": 53990
    },
    {
      "epoch": 0.9480503520075844,
      "grad_norm": 0.06094916864712991,
      "learning_rate": 0.00018572384261021538,
      "loss": 0.8527,
      "step": 54000
    },
    {
      "epoch": 0.9482259168875858,
      "grad_norm": 0.04871483392827813,
      "learning_rate": 0.00018571807732962648,
      "loss": 0.8603,
      "step": 54010
    },
    {
      "epoch": 0.9484014817675872,
      "grad_norm": 0.060722613845225516,
      "learning_rate": 0.00018571231097564333,
      "loss": 0.8588,
      "step": 54020
    },
    {
      "epoch": 0.9485770466475886,
      "grad_norm": 0.054809983642605795,
      "learning_rate": 0.00018570654354833893,
      "loss": 0.8555,
      "step": 54030
    },
    {
      "epoch": 0.94875261152759,
      "grad_norm": 0.06313626418320133,
      "learning_rate": 0.0001857007750477864,
      "loss": 0.8574,
      "step": 54040
    },
    {
      "epoch": 0.9489281764075914,
      "grad_norm": 0.08154454964409878,
      "learning_rate": 0.00018569500547405883,
      "loss": 0.8529,
      "step": 54050
    },
    {
      "epoch": 0.9491037412875928,
      "grad_norm": 0.07563764218638472,
      "learning_rate": 0.0001856892348272293,
      "loss": 0.8521,
      "step": 54060
    },
    {
      "epoch": 0.9492793061675943,
      "grad_norm": 0.05583690686888979,
      "learning_rate": 0.00018568346310737094,
      "loss": 0.8504,
      "step": 54070
    },
    {
      "epoch": 0.9494548710475956,
      "grad_norm": 0.06218101565954428,
      "learning_rate": 0.00018567769031455685,
      "loss": 0.8514,
      "step": 54080
    },
    {
      "epoch": 0.949630435927597,
      "grad_norm": 0.07477745365436439,
      "learning_rate": 0.00018567191644886023,
      "loss": 0.8595,
      "step": 54090
    },
    {
      "epoch": 0.9498060008075985,
      "grad_norm": 0.05087197673976684,
      "learning_rate": 0.00018566614151035418,
      "loss": 0.8401,
      "step": 54100
    },
    {
      "epoch": 0.9499815656875998,
      "grad_norm": 0.05719067534983043,
      "learning_rate": 0.0001856603654991119,
      "loss": 0.8565,
      "step": 54110
    },
    {
      "epoch": 0.9501571305676012,
      "grad_norm": 0.06247082065332925,
      "learning_rate": 0.00018565458841520658,
      "loss": 0.8518,
      "step": 54120
    },
    {
      "epoch": 0.9503326954476027,
      "grad_norm": 0.09013147914560854,
      "learning_rate": 0.00018564881025871137,
      "loss": 0.8481,
      "step": 54130
    },
    {
      "epoch": 0.950508260327604,
      "grad_norm": 0.05210122256084442,
      "learning_rate": 0.00018564303102969952,
      "loss": 0.857,
      "step": 54140
    },
    {
      "epoch": 0.9506838252076054,
      "grad_norm": 0.06274604484734791,
      "learning_rate": 0.00018563725072824426,
      "loss": 0.851,
      "step": 54150
    },
    {
      "epoch": 0.9508593900876069,
      "grad_norm": 0.0558571741402602,
      "learning_rate": 0.0001856314693544188,
      "loss": 0.8432,
      "step": 54160
    },
    {
      "epoch": 0.9510349549676083,
      "grad_norm": 0.05962600239569053,
      "learning_rate": 0.0001856256869082964,
      "loss": 0.8493,
      "step": 54170
    },
    {
      "epoch": 0.9512105198476097,
      "grad_norm": 0.08939146073703436,
      "learning_rate": 0.00018561990338995032,
      "loss": 0.8587,
      "step": 54180
    },
    {
      "epoch": 0.9513860847276111,
      "grad_norm": 0.06901872113649257,
      "learning_rate": 0.00018561411879945388,
      "loss": 0.855,
      "step": 54190
    },
    {
      "epoch": 0.9515616496076125,
      "grad_norm": 0.05574194587925945,
      "learning_rate": 0.00018560833313688028,
      "loss": 0.8526,
      "step": 54200
    },
    {
      "epoch": 0.951737214487614,
      "grad_norm": 0.08030951701240316,
      "learning_rate": 0.00018560254640230295,
      "loss": 0.8515,
      "step": 54210
    },
    {
      "epoch": 0.9519127793676153,
      "grad_norm": 0.058178354185772295,
      "learning_rate": 0.00018559675859579507,
      "loss": 0.8458,
      "step": 54220
    },
    {
      "epoch": 0.9520883442476167,
      "grad_norm": 0.049132377136920945,
      "learning_rate": 0.00018559096971743012,
      "loss": 0.853,
      "step": 54230
    },
    {
      "epoch": 0.9522639091276182,
      "grad_norm": 0.06534712095312196,
      "learning_rate": 0.00018558517976728135,
      "loss": 0.8476,
      "step": 54240
    },
    {
      "epoch": 0.9524394740076195,
      "grad_norm": 0.07729234952964717,
      "learning_rate": 0.00018557938874542214,
      "loss": 0.8539,
      "step": 54250
    },
    {
      "epoch": 0.9526150388876209,
      "grad_norm": 0.06239805454638391,
      "learning_rate": 0.00018557359665192586,
      "loss": 0.8521,
      "step": 54260
    },
    {
      "epoch": 0.9527906037676224,
      "grad_norm": 0.07006186060052604,
      "learning_rate": 0.00018556780348686592,
      "loss": 0.8482,
      "step": 54270
    },
    {
      "epoch": 0.9529661686476237,
      "grad_norm": 0.0685217225777155,
      "learning_rate": 0.00018556200925031564,
      "loss": 0.8465,
      "step": 54280
    },
    {
      "epoch": 0.9531417335276251,
      "grad_norm": 0.07819626186195873,
      "learning_rate": 0.00018555621394234853,
      "loss": 0.8572,
      "step": 54290
    },
    {
      "epoch": 0.9533172984076266,
      "grad_norm": 0.06911111388868735,
      "learning_rate": 0.00018555041756303803,
      "loss": 0.8558,
      "step": 54300
    },
    {
      "epoch": 0.953492863287628,
      "grad_norm": 0.07701689991933643,
      "learning_rate": 0.0001855446201124575,
      "loss": 0.8511,
      "step": 54310
    },
    {
      "epoch": 0.9536684281676293,
      "grad_norm": 0.04527688378177501,
      "learning_rate": 0.0001855388215906804,
      "loss": 0.855,
      "step": 54320
    },
    {
      "epoch": 0.9538439930476308,
      "grad_norm": 0.05829381617578348,
      "learning_rate": 0.00018553302199778027,
      "loss": 0.8575,
      "step": 54330
    },
    {
      "epoch": 0.9540195579276322,
      "grad_norm": 0.04960008216162136,
      "learning_rate": 0.00018552722133383055,
      "loss": 0.8488,
      "step": 54340
    },
    {
      "epoch": 0.9541951228076335,
      "grad_norm": 0.0802672650511748,
      "learning_rate": 0.0001855214195989047,
      "loss": 0.8575,
      "step": 54350
    },
    {
      "epoch": 0.954370687687635,
      "grad_norm": 0.05198426213321711,
      "learning_rate": 0.00018551561679307633,
      "loss": 0.8448,
      "step": 54360
    },
    {
      "epoch": 0.9545462525676364,
      "grad_norm": 0.055178446741323814,
      "learning_rate": 0.00018550981291641882,
      "loss": 0.8437,
      "step": 54370
    },
    {
      "epoch": 0.9547218174476377,
      "grad_norm": 0.0647293447298661,
      "learning_rate": 0.00018550400796900582,
      "loss": 0.8462,
      "step": 54380
    },
    {
      "epoch": 0.9548973823276392,
      "grad_norm": 0.0649594347273618,
      "learning_rate": 0.0001854982019509109,
      "loss": 0.8462,
      "step": 54390
    },
    {
      "epoch": 0.9550729472076406,
      "grad_norm": 0.04450647974129818,
      "learning_rate": 0.00018549239486220748,
      "loss": 0.8561,
      "step": 54400
    },
    {
      "epoch": 0.9552485120876419,
      "grad_norm": 0.07206989843055972,
      "learning_rate": 0.00018548658670296925,
      "loss": 0.8461,
      "step": 54410
    },
    {
      "epoch": 0.9554240769676434,
      "grad_norm": 0.055369992810811966,
      "learning_rate": 0.00018548077747326978,
      "loss": 0.8551,
      "step": 54420
    },
    {
      "epoch": 0.9555996418476448,
      "grad_norm": 0.07209220630463675,
      "learning_rate": 0.0001854749671731827,
      "loss": 0.8548,
      "step": 54430
    },
    {
      "epoch": 0.9557752067276462,
      "grad_norm": 0.09849552757073331,
      "learning_rate": 0.00018546915580278157,
      "loss": 0.8547,
      "step": 54440
    },
    {
      "epoch": 0.9559507716076476,
      "grad_norm": 0.054070577651868165,
      "learning_rate": 0.00018546334336214,
      "loss": 0.8576,
      "step": 54450
    },
    {
      "epoch": 0.956126336487649,
      "grad_norm": 0.07287897873717422,
      "learning_rate": 0.00018545752985133177,
      "loss": 0.8476,
      "step": 54460
    },
    {
      "epoch": 0.9563019013676504,
      "grad_norm": 0.10064046825563364,
      "learning_rate": 0.0001854517152704304,
      "loss": 0.8549,
      "step": 54470
    },
    {
      "epoch": 0.9564774662476518,
      "grad_norm": 0.06871207292336907,
      "learning_rate": 0.00018544589961950962,
      "loss": 0.853,
      "step": 54480
    },
    {
      "epoch": 0.9566530311276532,
      "grad_norm": 0.05466980330390416,
      "learning_rate": 0.0001854400828986431,
      "loss": 0.855,
      "step": 54490
    },
    {
      "epoch": 0.9568285960076546,
      "grad_norm": 0.06795797993875703,
      "learning_rate": 0.00018543426510790457,
      "loss": 0.8543,
      "step": 54500
    },
    {
      "epoch": 0.957004160887656,
      "grad_norm": 0.06274678685720017,
      "learning_rate": 0.00018542844624736769,
      "loss": 0.8471,
      "step": 54510
    },
    {
      "epoch": 0.9571797257676574,
      "grad_norm": 0.059728414483256616,
      "learning_rate": 0.00018542262631710625,
      "loss": 0.8483,
      "step": 54520
    },
    {
      "epoch": 0.9573552906476589,
      "grad_norm": 0.06107976261254306,
      "learning_rate": 0.0001854168053171939,
      "loss": 0.8442,
      "step": 54530
    },
    {
      "epoch": 0.9575308555276603,
      "grad_norm": 0.06593359989469894,
      "learning_rate": 0.0001854109832477045,
      "loss": 0.8525,
      "step": 54540
    },
    {
      "epoch": 0.9577064204076616,
      "grad_norm": 0.05250797592266925,
      "learning_rate": 0.00018540516010871175,
      "loss": 0.8501,
      "step": 54550
    },
    {
      "epoch": 0.9578819852876631,
      "grad_norm": 0.06143394036182521,
      "learning_rate": 0.00018539933590028943,
      "loss": 0.8537,
      "step": 54560
    },
    {
      "epoch": 0.9580575501676645,
      "grad_norm": 0.0612258751030492,
      "learning_rate": 0.00018539351062251137,
      "loss": 0.8418,
      "step": 54570
    },
    {
      "epoch": 0.9582331150476658,
      "grad_norm": 0.050165479540366535,
      "learning_rate": 0.00018538768427545137,
      "loss": 0.8544,
      "step": 54580
    },
    {
      "epoch": 0.9584086799276673,
      "grad_norm": 0.05846749718776558,
      "learning_rate": 0.00018538185685918322,
      "loss": 0.8515,
      "step": 54590
    },
    {
      "epoch": 0.9585842448076687,
      "grad_norm": 0.05349466226121646,
      "learning_rate": 0.00018537602837378076,
      "loss": 0.8499,
      "step": 54600
    },
    {
      "epoch": 0.95875980968767,
      "grad_norm": 0.055316818296106775,
      "learning_rate": 0.00018537019881931786,
      "loss": 0.8502,
      "step": 54610
    },
    {
      "epoch": 0.9589353745676715,
      "grad_norm": 0.060605810154365974,
      "learning_rate": 0.00018536436819586836,
      "loss": 0.8492,
      "step": 54620
    },
    {
      "epoch": 0.9591109394476729,
      "grad_norm": 0.06392038305309096,
      "learning_rate": 0.00018535853650350617,
      "loss": 0.8456,
      "step": 54630
    },
    {
      "epoch": 0.9592865043276743,
      "grad_norm": 0.060862025814821336,
      "learning_rate": 0.00018535270374230514,
      "loss": 0.8578,
      "step": 54640
    },
    {
      "epoch": 0.9594620692076757,
      "grad_norm": 0.05492652794475265,
      "learning_rate": 0.00018534686991233915,
      "loss": 0.8528,
      "step": 54650
    },
    {
      "epoch": 0.9596376340876771,
      "grad_norm": 0.10049920740794822,
      "learning_rate": 0.00018534103501368222,
      "loss": 0.8549,
      "step": 54660
    },
    {
      "epoch": 0.9598131989676785,
      "grad_norm": 0.07549946622447878,
      "learning_rate": 0.0001853351990464082,
      "loss": 0.8468,
      "step": 54670
    },
    {
      "epoch": 0.95998876384768,
      "grad_norm": 0.06684809260188489,
      "learning_rate": 0.00018532936201059102,
      "loss": 0.8465,
      "step": 54680
    },
    {
      "epoch": 0.9601643287276813,
      "grad_norm": 0.05879738598479774,
      "learning_rate": 0.00018532352390630467,
      "loss": 0.8485,
      "step": 54690
    },
    {
      "epoch": 0.9603398936076827,
      "grad_norm": 0.05643508034506602,
      "learning_rate": 0.0001853176847336231,
      "loss": 0.8413,
      "step": 54700
    },
    {
      "epoch": 0.9605154584876842,
      "grad_norm": 0.054607381972910964,
      "learning_rate": 0.00018531184449262033,
      "loss": 0.8543,
      "step": 54710
    },
    {
      "epoch": 0.9606910233676855,
      "grad_norm": 0.0684252012627183,
      "learning_rate": 0.00018530600318337033,
      "loss": 0.8514,
      "step": 54720
    },
    {
      "epoch": 0.9608665882476869,
      "grad_norm": 0.048011149614720576,
      "learning_rate": 0.00018530016080594707,
      "loss": 0.8494,
      "step": 54730
    },
    {
      "epoch": 0.9610421531276884,
      "grad_norm": 0.06692847368599607,
      "learning_rate": 0.00018529431736042466,
      "loss": 0.8496,
      "step": 54740
    },
    {
      "epoch": 0.9612177180076897,
      "grad_norm": 0.07911500075453089,
      "learning_rate": 0.00018528847284687708,
      "loss": 0.8567,
      "step": 54750
    },
    {
      "epoch": 0.9613932828876911,
      "grad_norm": 0.05960608570571123,
      "learning_rate": 0.0001852826272653784,
      "loss": 0.8483,
      "step": 54760
    },
    {
      "epoch": 0.9615688477676926,
      "grad_norm": 0.07189012577310204,
      "learning_rate": 0.0001852767806160027,
      "loss": 0.8625,
      "step": 54770
    },
    {
      "epoch": 0.961744412647694,
      "grad_norm": 0.11558333901357626,
      "learning_rate": 0.000185270932898824,
      "loss": 0.8523,
      "step": 54780
    },
    {
      "epoch": 0.9619199775276953,
      "grad_norm": 0.0741711994790164,
      "learning_rate": 0.00018526508411391648,
      "loss": 0.8493,
      "step": 54790
    },
    {
      "epoch": 0.9620955424076968,
      "grad_norm": 0.0867529806092186,
      "learning_rate": 0.00018525923426135414,
      "loss": 0.8451,
      "step": 54800
    },
    {
      "epoch": 0.9622711072876982,
      "grad_norm": 0.07353928300276961,
      "learning_rate": 0.00018525338334121117,
      "loss": 0.8532,
      "step": 54810
    },
    {
      "epoch": 0.9624466721676995,
      "grad_norm": 0.05812702526478087,
      "learning_rate": 0.0001852475313535617,
      "loss": 0.8525,
      "step": 54820
    },
    {
      "epoch": 0.962622237047701,
      "grad_norm": 0.07202715521465064,
      "learning_rate": 0.00018524167829847985,
      "loss": 0.8464,
      "step": 54830
    },
    {
      "epoch": 0.9627978019277024,
      "grad_norm": 0.06603802762527142,
      "learning_rate": 0.0001852358241760398,
      "loss": 0.8491,
      "step": 54840
    },
    {
      "epoch": 0.9629733668077037,
      "grad_norm": 0.06783559127794969,
      "learning_rate": 0.0001852299689863157,
      "loss": 0.8523,
      "step": 54850
    },
    {
      "epoch": 0.9631489316877052,
      "grad_norm": 0.05753180361026002,
      "learning_rate": 0.00018522411272938174,
      "loss": 0.8507,
      "step": 54860
    },
    {
      "epoch": 0.9633244965677066,
      "grad_norm": 0.08623594228687413,
      "learning_rate": 0.00018521825540531218,
      "loss": 0.8452,
      "step": 54870
    },
    {
      "epoch": 0.963500061447708,
      "grad_norm": 0.06923387287572728,
      "learning_rate": 0.00018521239701418113,
      "loss": 0.8472,
      "step": 54880
    },
    {
      "epoch": 0.9636756263277094,
      "grad_norm": 0.06073217248891654,
      "learning_rate": 0.0001852065375560629,
      "loss": 0.845,
      "step": 54890
    },
    {
      "epoch": 0.9638511912077108,
      "grad_norm": 0.05559230449661111,
      "learning_rate": 0.0001852006770310317,
      "loss": 0.85,
      "step": 54900
    },
    {
      "epoch": 0.9640267560877123,
      "grad_norm": 0.06243009981511851,
      "learning_rate": 0.00018519481543916177,
      "loss": 0.8555,
      "step": 54910
    },
    {
      "epoch": 0.9642023209677136,
      "grad_norm": 0.05890306446116319,
      "learning_rate": 0.0001851889527805274,
      "loss": 0.8508,
      "step": 54920
    },
    {
      "epoch": 0.964377885847715,
      "grad_norm": 0.05611702798284687,
      "learning_rate": 0.00018518308905520286,
      "loss": 0.8486,
      "step": 54930
    },
    {
      "epoch": 0.9645534507277165,
      "grad_norm": 0.05844910132407351,
      "learning_rate": 0.00018517722426326244,
      "loss": 0.8448,
      "step": 54940
    },
    {
      "epoch": 0.9647290156077178,
      "grad_norm": 0.059334716382845584,
      "learning_rate": 0.0001851713584047805,
      "loss": 0.8483,
      "step": 54950
    },
    {
      "epoch": 0.9649045804877192,
      "grad_norm": 0.0502403404457577,
      "learning_rate": 0.00018516549147983127,
      "loss": 0.8514,
      "step": 54960
    },
    {
      "epoch": 0.9650801453677207,
      "grad_norm": 0.0651734444949658,
      "learning_rate": 0.00018515962348848914,
      "loss": 0.8562,
      "step": 54970
    },
    {
      "epoch": 0.965255710247722,
      "grad_norm": 0.05078363686571217,
      "learning_rate": 0.00018515375443082849,
      "loss": 0.8537,
      "step": 54980
    },
    {
      "epoch": 0.9654312751277234,
      "grad_norm": 0.07005084369373878,
      "learning_rate": 0.0001851478843069236,
      "loss": 0.8538,
      "step": 54990
    },
    {
      "epoch": 0.9656068400077249,
      "grad_norm": 0.04930638134099371,
      "learning_rate": 0.0001851420131168489,
      "loss": 0.8525,
      "step": 55000
    },
    {
      "epoch": 0.9657824048877263,
      "grad_norm": 0.0571127276703054,
      "learning_rate": 0.00018513614086067878,
      "loss": 0.8436,
      "step": 55010
    },
    {
      "epoch": 0.9659579697677276,
      "grad_norm": 0.057161985037682224,
      "learning_rate": 0.0001851302675384876,
      "loss": 0.8509,
      "step": 55020
    },
    {
      "epoch": 0.9661335346477291,
      "grad_norm": 0.09423229017482862,
      "learning_rate": 0.00018512439315034984,
      "loss": 0.8579,
      "step": 55030
    },
    {
      "epoch": 0.9663090995277305,
      "grad_norm": 0.05799227052258098,
      "learning_rate": 0.0001851185176963399,
      "loss": 0.8504,
      "step": 55040
    },
    {
      "epoch": 0.9664846644077318,
      "grad_norm": 0.0631095850311377,
      "learning_rate": 0.0001851126411765322,
      "loss": 0.8493,
      "step": 55050
    },
    {
      "epoch": 0.9666602292877333,
      "grad_norm": 0.05709538869992791,
      "learning_rate": 0.0001851067635910012,
      "loss": 0.8514,
      "step": 55060
    },
    {
      "epoch": 0.9668357941677347,
      "grad_norm": 0.0674495336783126,
      "learning_rate": 0.0001851008849398214,
      "loss": 0.8505,
      "step": 55070
    },
    {
      "epoch": 0.967011359047736,
      "grad_norm": 0.05229014339336654,
      "learning_rate": 0.00018509500522306732,
      "loss": 0.8551,
      "step": 55080
    },
    {
      "epoch": 0.9671869239277375,
      "grad_norm": 0.07509144036773872,
      "learning_rate": 0.00018508912444081336,
      "loss": 0.8456,
      "step": 55090
    },
    {
      "epoch": 0.9673624888077389,
      "grad_norm": 0.0729085304413324,
      "learning_rate": 0.00018508324259313406,
      "loss": 0.8525,
      "step": 55100
    },
    {
      "epoch": 0.9675380536877403,
      "grad_norm": 0.047221179838645544,
      "learning_rate": 0.00018507735968010397,
      "loss": 0.8563,
      "step": 55110
    },
    {
      "epoch": 0.9677136185677417,
      "grad_norm": 0.07097445793519247,
      "learning_rate": 0.00018507147570179765,
      "loss": 0.8493,
      "step": 55120
    },
    {
      "epoch": 0.9678891834477431,
      "grad_norm": 0.0588745210348203,
      "learning_rate": 0.00018506559065828958,
      "loss": 0.8546,
      "step": 55130
    },
    {
      "epoch": 0.9680647483277445,
      "grad_norm": 0.08921540229976893,
      "learning_rate": 0.00018505970454965444,
      "loss": 0.8566,
      "step": 55140
    },
    {
      "epoch": 0.968240313207746,
      "grad_norm": 0.055123595532949335,
      "learning_rate": 0.00018505381737596665,
      "loss": 0.8492,
      "step": 55150
    },
    {
      "epoch": 0.9684158780877473,
      "grad_norm": 0.049647610382968284,
      "learning_rate": 0.0001850479291373009,
      "loss": 0.8516,
      "step": 55160
    },
    {
      "epoch": 0.9685914429677487,
      "grad_norm": 0.05680136393105547,
      "learning_rate": 0.00018504203983373183,
      "loss": 0.8532,
      "step": 55170
    },
    {
      "epoch": 0.9687670078477502,
      "grad_norm": 0.049758433550845256,
      "learning_rate": 0.00018503614946533396,
      "loss": 0.8514,
      "step": 55180
    },
    {
      "epoch": 0.9689425727277515,
      "grad_norm": 0.07338301561691404,
      "learning_rate": 0.000185030258032182,
      "loss": 0.8478,
      "step": 55190
    },
    {
      "epoch": 0.9691181376077529,
      "grad_norm": 0.0745504486611267,
      "learning_rate": 0.00018502436553435053,
      "loss": 0.844,
      "step": 55200
    },
    {
      "epoch": 0.9692937024877544,
      "grad_norm": 0.05432648325120253,
      "learning_rate": 0.00018501847197191428,
      "loss": 0.8458,
      "step": 55210
    },
    {
      "epoch": 0.9694692673677557,
      "grad_norm": 0.0600427835850341,
      "learning_rate": 0.0001850125773449479,
      "loss": 0.8453,
      "step": 55220
    },
    {
      "epoch": 0.9696448322477571,
      "grad_norm": 0.07028897274735327,
      "learning_rate": 0.000185006681653526,
      "loss": 0.8594,
      "step": 55230
    },
    {
      "epoch": 0.9698203971277586,
      "grad_norm": 0.070069624349586,
      "learning_rate": 0.00018500078489772342,
      "loss": 0.8567,
      "step": 55240
    },
    {
      "epoch": 0.96999596200776,
      "grad_norm": 0.07385128169616227,
      "learning_rate": 0.00018499488707761474,
      "loss": 0.8549,
      "step": 55250
    },
    {
      "epoch": 0.9701715268877614,
      "grad_norm": 0.055309723706483994,
      "learning_rate": 0.00018498898819327477,
      "loss": 0.8519,
      "step": 55260
    },
    {
      "epoch": 0.9703470917677628,
      "grad_norm": 0.05481935191645343,
      "learning_rate": 0.0001849830882447782,
      "loss": 0.8504,
      "step": 55270
    },
    {
      "epoch": 0.9705226566477642,
      "grad_norm": 0.05511184607440725,
      "learning_rate": 0.00018497718723219987,
      "loss": 0.856,
      "step": 55280
    },
    {
      "epoch": 0.9706982215277656,
      "grad_norm": 0.05043693779624383,
      "learning_rate": 0.00018497128515561443,
      "loss": 0.8526,
      "step": 55290
    },
    {
      "epoch": 0.970873786407767,
      "grad_norm": 0.05993918914221316,
      "learning_rate": 0.0001849653820150967,
      "loss": 0.8471,
      "step": 55300
    },
    {
      "epoch": 0.9710493512877684,
      "grad_norm": 0.06443026945290094,
      "learning_rate": 0.00018495947781072152,
      "loss": 0.8517,
      "step": 55310
    },
    {
      "epoch": 0.9712249161677698,
      "grad_norm": 0.056166968211571626,
      "learning_rate": 0.00018495357254256364,
      "loss": 0.8532,
      "step": 55320
    },
    {
      "epoch": 0.9714004810477712,
      "grad_norm": 0.06223028828059734,
      "learning_rate": 0.0001849476662106979,
      "loss": 0.8508,
      "step": 55330
    },
    {
      "epoch": 0.9715760459277726,
      "grad_norm": 0.06451481454764314,
      "learning_rate": 0.00018494175881519917,
      "loss": 0.8559,
      "step": 55340
    },
    {
      "epoch": 0.9717516108077741,
      "grad_norm": 0.07198287370170346,
      "learning_rate": 0.00018493585035614228,
      "loss": 0.8461,
      "step": 55350
    },
    {
      "epoch": 0.9719271756877754,
      "grad_norm": 0.06517862268917182,
      "learning_rate": 0.00018492994083360207,
      "loss": 0.851,
      "step": 55360
    },
    {
      "epoch": 0.9721027405677768,
      "grad_norm": 0.09044334137889143,
      "learning_rate": 0.0001849240302476534,
      "loss": 0.8492,
      "step": 55370
    },
    {
      "epoch": 0.9722783054477783,
      "grad_norm": 0.0791707695685372,
      "learning_rate": 0.00018491811859837123,
      "loss": 0.8461,
      "step": 55380
    },
    {
      "epoch": 0.9724538703277796,
      "grad_norm": 0.047997155773097606,
      "learning_rate": 0.00018491220588583035,
      "loss": 0.8539,
      "step": 55390
    },
    {
      "epoch": 0.972629435207781,
      "grad_norm": 0.05769530954061769,
      "learning_rate": 0.00018490629211010578,
      "loss": 0.8504,
      "step": 55400
    },
    {
      "epoch": 0.9728050000877825,
      "grad_norm": 0.03992966713341037,
      "learning_rate": 0.0001849003772712724,
      "loss": 0.8554,
      "step": 55410
    },
    {
      "epoch": 0.9729805649677838,
      "grad_norm": 0.07676516011113174,
      "learning_rate": 0.00018489446136940515,
      "loss": 0.8482,
      "step": 55420
    },
    {
      "epoch": 0.9731561298477852,
      "grad_norm": 0.057535122047676315,
      "learning_rate": 0.00018488854440457903,
      "loss": 0.8501,
      "step": 55430
    },
    {
      "epoch": 0.9733316947277867,
      "grad_norm": 0.06485475917239733,
      "learning_rate": 0.00018488262637686895,
      "loss": 0.8487,
      "step": 55440
    },
    {
      "epoch": 0.9735072596077881,
      "grad_norm": 0.05134617149605119,
      "learning_rate": 0.00018487670728634993,
      "loss": 0.8583,
      "step": 55450
    },
    {
      "epoch": 0.9736828244877894,
      "grad_norm": 0.06144904273514751,
      "learning_rate": 0.0001848707871330969,
      "loss": 0.8526,
      "step": 55460
    },
    {
      "epoch": 0.9738583893677909,
      "grad_norm": 0.057796119617351975,
      "learning_rate": 0.00018486486591718499,
      "loss": 0.8529,
      "step": 55470
    },
    {
      "epoch": 0.9740339542477923,
      "grad_norm": 0.08998231739668633,
      "learning_rate": 0.00018485894363868914,
      "loss": 0.8451,
      "step": 55480
    },
    {
      "epoch": 0.9742095191277936,
      "grad_norm": 0.06694759636276848,
      "learning_rate": 0.0001848530202976844,
      "loss": 0.8485,
      "step": 55490
    },
    {
      "epoch": 0.9743850840077951,
      "grad_norm": 0.06417931726286148,
      "learning_rate": 0.00018484709589424583,
      "loss": 0.8492,
      "step": 55500
    },
    {
      "epoch": 0.9745606488877965,
      "grad_norm": 0.06659707210848947,
      "learning_rate": 0.00018484117042844847,
      "loss": 0.8598,
      "step": 55510
    },
    {
      "epoch": 0.9747362137677978,
      "grad_norm": 0.07563730617461892,
      "learning_rate": 0.0001848352439003674,
      "loss": 0.8524,
      "step": 55520
    },
    {
      "epoch": 0.9749117786477993,
      "grad_norm": 0.07155945178052073,
      "learning_rate": 0.00018482931631007773,
      "loss": 0.8516,
      "step": 55530
    },
    {
      "epoch": 0.9750873435278007,
      "grad_norm": 0.049655900102784555,
      "learning_rate": 0.00018482338765765457,
      "loss": 0.856,
      "step": 55540
    },
    {
      "epoch": 0.975262908407802,
      "grad_norm": 0.0609501190736301,
      "learning_rate": 0.000184817457943173,
      "loss": 0.8488,
      "step": 55550
    },
    {
      "epoch": 0.9754384732878035,
      "grad_norm": 0.0735315568179351,
      "learning_rate": 0.0001848115271667082,
      "loss": 0.856,
      "step": 55560
    },
    {
      "epoch": 0.9756140381678049,
      "grad_norm": 0.05443047942757224,
      "learning_rate": 0.00018480559532833528,
      "loss": 0.8544,
      "step": 55570
    },
    {
      "epoch": 0.9757896030478063,
      "grad_norm": 0.06752363635914012,
      "learning_rate": 0.00018479966242812942,
      "loss": 0.8461,
      "step": 55580
    },
    {
      "epoch": 0.9759651679278077,
      "grad_norm": 0.05849240156022589,
      "learning_rate": 0.00018479372846616577,
      "loss": 0.8535,
      "step": 55590
    },
    {
      "epoch": 0.9761407328078091,
      "grad_norm": 0.06901549899903517,
      "learning_rate": 0.00018478779344251954,
      "loss": 0.845,
      "step": 55600
    },
    {
      "epoch": 0.9763162976878106,
      "grad_norm": 0.07548094580868875,
      "learning_rate": 0.00018478185735726585,
      "loss": 0.8494,
      "step": 55610
    },
    {
      "epoch": 0.976491862567812,
      "grad_norm": 0.09708198272709694,
      "learning_rate": 0.00018477592021048006,
      "loss": 0.8458,
      "step": 55620
    },
    {
      "epoch": 0.9766674274478133,
      "grad_norm": 0.0752112981656823,
      "learning_rate": 0.0001847699820022372,
      "loss": 0.8484,
      "step": 55630
    },
    {
      "epoch": 0.9768429923278148,
      "grad_norm": 0.056159828413710054,
      "learning_rate": 0.00018476404273261266,
      "loss": 0.8502,
      "step": 55640
    },
    {
      "epoch": 0.9770185572078162,
      "grad_norm": 0.068034387498023,
      "learning_rate": 0.00018475810240168168,
      "loss": 0.8497,
      "step": 55650
    },
    {
      "epoch": 0.9771941220878175,
      "grad_norm": 0.06044543016760586,
      "learning_rate": 0.00018475216100951946,
      "loss": 0.8509,
      "step": 55660
    },
    {
      "epoch": 0.977369686967819,
      "grad_norm": 0.07512595841475327,
      "learning_rate": 0.0001847462185562013,
      "loss": 0.8508,
      "step": 55670
    },
    {
      "epoch": 0.9775452518478204,
      "grad_norm": 0.06228754319453959,
      "learning_rate": 0.0001847402750418025,
      "loss": 0.8582,
      "step": 55680
    },
    {
      "epoch": 0.9777208167278217,
      "grad_norm": 0.05807298330666178,
      "learning_rate": 0.0001847343304663984,
      "loss": 0.8448,
      "step": 55690
    },
    {
      "epoch": 0.9778963816078232,
      "grad_norm": 0.05948352216811074,
      "learning_rate": 0.00018472838483006424,
      "loss": 0.8501,
      "step": 55700
    },
    {
      "epoch": 0.9780719464878246,
      "grad_norm": 0.050300660156148466,
      "learning_rate": 0.00018472243813287548,
      "loss": 0.8509,
      "step": 55710
    },
    {
      "epoch": 0.978247511367826,
      "grad_norm": 0.06682234113413782,
      "learning_rate": 0.0001847164903749073,
      "loss": 0.8596,
      "step": 55720
    },
    {
      "epoch": 0.9784230762478274,
      "grad_norm": 0.06597529322468294,
      "learning_rate": 0.00018471054155623512,
      "loss": 0.8538,
      "step": 55730
    },
    {
      "epoch": 0.9785986411278288,
      "grad_norm": 0.062095505587333275,
      "learning_rate": 0.00018470459167693442,
      "loss": 0.8544,
      "step": 55740
    },
    {
      "epoch": 0.9787742060078302,
      "grad_norm": 0.06281225941136828,
      "learning_rate": 0.00018469864073708046,
      "loss": 0.8445,
      "step": 55750
    },
    {
      "epoch": 0.9789497708878316,
      "grad_norm": 0.05208378120539585,
      "learning_rate": 0.00018469268873674868,
      "loss": 0.8525,
      "step": 55760
    },
    {
      "epoch": 0.979125335767833,
      "grad_norm": 0.08056171820162913,
      "learning_rate": 0.0001846867356760145,
      "loss": 0.8524,
      "step": 55770
    },
    {
      "epoch": 0.9793009006478344,
      "grad_norm": 0.06694171628558333,
      "learning_rate": 0.00018468078155495335,
      "loss": 0.8526,
      "step": 55780
    },
    {
      "epoch": 0.9794764655278358,
      "grad_norm": 0.06323177497721924,
      "learning_rate": 0.00018467482637364065,
      "loss": 0.8531,
      "step": 55790
    },
    {
      "epoch": 0.9796520304078372,
      "grad_norm": 0.05646302783777308,
      "learning_rate": 0.00018466887013215185,
      "loss": 0.8505,
      "step": 55800
    },
    {
      "epoch": 0.9798275952878386,
      "grad_norm": 0.0617915315689957,
      "learning_rate": 0.00018466291283056246,
      "loss": 0.8565,
      "step": 55810
    },
    {
      "epoch": 0.9800031601678401,
      "grad_norm": 0.06764278207616155,
      "learning_rate": 0.00018465695446894792,
      "loss": 0.8555,
      "step": 55820
    },
    {
      "epoch": 0.9801787250478414,
      "grad_norm": 0.0602906141468306,
      "learning_rate": 0.0001846509950473837,
      "loss": 0.8496,
      "step": 55830
    },
    {
      "epoch": 0.9803542899278428,
      "grad_norm": 0.06281601296602188,
      "learning_rate": 0.00018464503456594537,
      "loss": 0.8456,
      "step": 55840
    },
    {
      "epoch": 0.9805298548078443,
      "grad_norm": 0.06402301636234486,
      "learning_rate": 0.00018463907302470844,
      "loss": 0.857,
      "step": 55850
    },
    {
      "epoch": 0.9807054196878456,
      "grad_norm": 0.07276058733980836,
      "learning_rate": 0.0001846331104237484,
      "loss": 0.858,
      "step": 55860
    },
    {
      "epoch": 0.980880984567847,
      "grad_norm": 0.0579619233657369,
      "learning_rate": 0.00018462714676314082,
      "loss": 0.8562,
      "step": 55870
    },
    {
      "epoch": 0.9810565494478485,
      "grad_norm": 0.07557107832416061,
      "learning_rate": 0.0001846211820429613,
      "loss": 0.8475,
      "step": 55880
    },
    {
      "epoch": 0.9812321143278498,
      "grad_norm": 0.07079371601885488,
      "learning_rate": 0.00018461521626328534,
      "loss": 0.8467,
      "step": 55890
    },
    {
      "epoch": 0.9814076792078512,
      "grad_norm": 0.07008710048741984,
      "learning_rate": 0.00018460924942418855,
      "loss": 0.8543,
      "step": 55900
    },
    {
      "epoch": 0.9815832440878527,
      "grad_norm": 0.05294574699444911,
      "learning_rate": 0.0001846032815257466,
      "loss": 0.8559,
      "step": 55910
    },
    {
      "epoch": 0.9817588089678541,
      "grad_norm": 0.06899721980444608,
      "learning_rate": 0.000184597312568035,
      "loss": 0.8522,
      "step": 55920
    },
    {
      "epoch": 0.9819343738478554,
      "grad_norm": 0.06754631031773804,
      "learning_rate": 0.00018459134255112945,
      "loss": 0.8504,
      "step": 55930
    },
    {
      "epoch": 0.9821099387278569,
      "grad_norm": 0.057313839564732724,
      "learning_rate": 0.0001845853714751056,
      "loss": 0.8501,
      "step": 55940
    },
    {
      "epoch": 0.9822855036078583,
      "grad_norm": 0.07127588726819924,
      "learning_rate": 0.00018457939934003902,
      "loss": 0.8515,
      "step": 55950
    },
    {
      "epoch": 0.9824610684878596,
      "grad_norm": 0.06125392138724135,
      "learning_rate": 0.00018457342614600547,
      "loss": 0.8507,
      "step": 55960
    },
    {
      "epoch": 0.9826366333678611,
      "grad_norm": 0.07463119017923493,
      "learning_rate": 0.0001845674518930806,
      "loss": 0.8517,
      "step": 55970
    },
    {
      "epoch": 0.9828121982478625,
      "grad_norm": 0.07366189327701678,
      "learning_rate": 0.00018456147658134004,
      "loss": 0.8512,
      "step": 55980
    },
    {
      "epoch": 0.982987763127864,
      "grad_norm": 0.08153105381161743,
      "learning_rate": 0.00018455550021085964,
      "loss": 0.844,
      "step": 55990
    },
    {
      "epoch": 0.9831633280078653,
      "grad_norm": 0.11749778103322601,
      "learning_rate": 0.00018454952278171497,
      "loss": 0.8591,
      "step": 56000
    },
    {
      "epoch": 0.9833388928878667,
      "grad_norm": 0.08121817076314049,
      "learning_rate": 0.00018454354429398186,
      "loss": 0.8523,
      "step": 56010
    },
    {
      "epoch": 0.9835144577678682,
      "grad_norm": 0.07079463129232026,
      "learning_rate": 0.00018453756474773604,
      "loss": 0.8525,
      "step": 56020
    },
    {
      "epoch": 0.9836900226478695,
      "grad_norm": 0.08404419664192106,
      "learning_rate": 0.00018453158414305325,
      "loss": 0.8564,
      "step": 56030
    },
    {
      "epoch": 0.9838655875278709,
      "grad_norm": 0.07247140318245286,
      "learning_rate": 0.0001845256024800093,
      "loss": 0.8465,
      "step": 56040
    },
    {
      "epoch": 0.9840411524078724,
      "grad_norm": 0.08300618838905575,
      "learning_rate": 0.00018451961975867996,
      "loss": 0.8529,
      "step": 56050
    },
    {
      "epoch": 0.9842167172878737,
      "grad_norm": 0.06315732463736477,
      "learning_rate": 0.00018451363597914101,
      "loss": 0.8454,
      "step": 56060
    },
    {
      "epoch": 0.9843922821678751,
      "grad_norm": 0.05988911987197718,
      "learning_rate": 0.00018450765114146826,
      "loss": 0.8431,
      "step": 56070
    },
    {
      "epoch": 0.9845678470478766,
      "grad_norm": 0.06688925501363228,
      "learning_rate": 0.00018450166524573762,
      "loss": 0.845,
      "step": 56080
    },
    {
      "epoch": 0.984743411927878,
      "grad_norm": 0.049573475626486764,
      "learning_rate": 0.00018449567829202488,
      "loss": 0.8531,
      "step": 56090
    },
    {
      "epoch": 0.9849189768078793,
      "grad_norm": 0.05787847886658095,
      "learning_rate": 0.00018448969028040583,
      "loss": 0.8509,
      "step": 56100
    },
    {
      "epoch": 0.9850945416878808,
      "grad_norm": 0.0565327661723438,
      "learning_rate": 0.00018448370121095645,
      "loss": 0.8509,
      "step": 56110
    },
    {
      "epoch": 0.9852701065678822,
      "grad_norm": 0.05670153000843854,
      "learning_rate": 0.0001844777110837526,
      "loss": 0.8487,
      "step": 56120
    },
    {
      "epoch": 0.9854456714478835,
      "grad_norm": 0.0652228840376706,
      "learning_rate": 0.0001844717198988701,
      "loss": 0.847,
      "step": 56130
    },
    {
      "epoch": 0.985621236327885,
      "grad_norm": 0.05013247029384433,
      "learning_rate": 0.00018446572765638494,
      "loss": 0.858,
      "step": 56140
    },
    {
      "epoch": 0.9857968012078864,
      "grad_norm": 0.047333554475636315,
      "learning_rate": 0.00018445973435637304,
      "loss": 0.8431,
      "step": 56150
    },
    {
      "epoch": 0.9859723660878877,
      "grad_norm": 0.0494248327352652,
      "learning_rate": 0.00018445373999891027,
      "loss": 0.8546,
      "step": 56160
    },
    {
      "epoch": 0.9861479309678892,
      "grad_norm": 0.07182256827166052,
      "learning_rate": 0.00018444774458407264,
      "loss": 0.8477,
      "step": 56170
    },
    {
      "epoch": 0.9863234958478906,
      "grad_norm": 0.05194279666653261,
      "learning_rate": 0.00018444174811193607,
      "loss": 0.8547,
      "step": 56180
    },
    {
      "epoch": 0.986499060727892,
      "grad_norm": 0.04932393231451287,
      "learning_rate": 0.00018443575058257658,
      "loss": 0.8511,
      "step": 56190
    },
    {
      "epoch": 0.9866746256078934,
      "grad_norm": 0.060348620870103266,
      "learning_rate": 0.00018442975199607016,
      "loss": 0.8443,
      "step": 56200
    },
    {
      "epoch": 0.9868501904878948,
      "grad_norm": 0.06042712555335187,
      "learning_rate": 0.00018442375235249274,
      "loss": 0.8527,
      "step": 56210
    },
    {
      "epoch": 0.9870257553678962,
      "grad_norm": 0.056347009801086395,
      "learning_rate": 0.00018441775165192043,
      "loss": 0.8549,
      "step": 56220
    },
    {
      "epoch": 0.9872013202478976,
      "grad_norm": 0.06728413538086979,
      "learning_rate": 0.0001844117498944292,
      "loss": 0.8475,
      "step": 56230
    },
    {
      "epoch": 0.987376885127899,
      "grad_norm": 0.053558459664971955,
      "learning_rate": 0.00018440574708009511,
      "loss": 0.8576,
      "step": 56240
    },
    {
      "epoch": 0.9875524500079004,
      "grad_norm": 0.059613266674837434,
      "learning_rate": 0.00018439974320899423,
      "loss": 0.8533,
      "step": 56250
    },
    {
      "epoch": 0.9877280148879019,
      "grad_norm": 0.07330765755748743,
      "learning_rate": 0.00018439373828120262,
      "loss": 0.8455,
      "step": 56260
    },
    {
      "epoch": 0.9879035797679032,
      "grad_norm": 0.06545137923766547,
      "learning_rate": 0.00018438773229679636,
      "loss": 0.8521,
      "step": 56270
    },
    {
      "epoch": 0.9880791446479046,
      "grad_norm": 0.0681413508826837,
      "learning_rate": 0.00018438172525585159,
      "loss": 0.8476,
      "step": 56280
    },
    {
      "epoch": 0.9882547095279061,
      "grad_norm": 0.054582687974819806,
      "learning_rate": 0.00018437571715844433,
      "loss": 0.851,
      "step": 56290
    },
    {
      "epoch": 0.9884302744079074,
      "grad_norm": 0.0787039512134953,
      "learning_rate": 0.00018436970800465077,
      "loss": 0.8565,
      "step": 56300
    },
    {
      "epoch": 0.9886058392879088,
      "grad_norm": 0.06792270035908422,
      "learning_rate": 0.00018436369779454702,
      "loss": 0.8443,
      "step": 56310
    },
    {
      "epoch": 0.9887814041679103,
      "grad_norm": 0.056951868090402744,
      "learning_rate": 0.00018435768652820926,
      "loss": 0.8462,
      "step": 56320
    },
    {
      "epoch": 0.9889569690479116,
      "grad_norm": 0.07037454172714246,
      "learning_rate": 0.0001843516742057136,
      "loss": 0.8492,
      "step": 56330
    },
    {
      "epoch": 0.9891325339279131,
      "grad_norm": 0.06976826237475375,
      "learning_rate": 0.00018434566082713632,
      "loss": 0.8516,
      "step": 56340
    },
    {
      "epoch": 0.9893080988079145,
      "grad_norm": 0.10132524158005675,
      "learning_rate": 0.0001843396463925535,
      "loss": 0.8567,
      "step": 56350
    },
    {
      "epoch": 0.9894836636879158,
      "grad_norm": 0.06174736600354629,
      "learning_rate": 0.0001843336309020414,
      "loss": 0.8481,
      "step": 56360
    },
    {
      "epoch": 0.9896592285679173,
      "grad_norm": 0.08478511145093871,
      "learning_rate": 0.00018432761435567625,
      "loss": 0.8485,
      "step": 56370
    },
    {
      "epoch": 0.9898347934479187,
      "grad_norm": 0.05609521311932699,
      "learning_rate": 0.00018432159675353421,
      "loss": 0.8488,
      "step": 56380
    },
    {
      "epoch": 0.9900103583279201,
      "grad_norm": 0.06905812628860275,
      "learning_rate": 0.0001843155780956916,
      "loss": 0.8467,
      "step": 56390
    },
    {
      "epoch": 0.9901859232079215,
      "grad_norm": 0.06674349190671393,
      "learning_rate": 0.0001843095583822246,
      "loss": 0.8531,
      "step": 56400
    },
    {
      "epoch": 0.9903614880879229,
      "grad_norm": 0.06104107820700924,
      "learning_rate": 0.0001843035376132096,
      "loss": 0.8486,
      "step": 56410
    },
    {
      "epoch": 0.9905370529679243,
      "grad_norm": 0.05970292129141999,
      "learning_rate": 0.00018429751578872275,
      "loss": 0.8535,
      "step": 56420
    },
    {
      "epoch": 0.9907126178479257,
      "grad_norm": 0.058194744225966696,
      "learning_rate": 0.00018429149290884048,
      "loss": 0.8501,
      "step": 56430
    },
    {
      "epoch": 0.9908881827279271,
      "grad_norm": 0.06849019508211356,
      "learning_rate": 0.000184285468973639,
      "loss": 0.8464,
      "step": 56440
    },
    {
      "epoch": 0.9910637476079285,
      "grad_norm": 0.04606362680712893,
      "learning_rate": 0.00018427944398319464,
      "loss": 0.8452,
      "step": 56450
    },
    {
      "epoch": 0.99123931248793,
      "grad_norm": 0.06368143886102015,
      "learning_rate": 0.00018427341793758378,
      "loss": 0.851,
      "step": 56460
    },
    {
      "epoch": 0.9914148773679313,
      "grad_norm": 0.09604199952190666,
      "learning_rate": 0.00018426739083688277,
      "loss": 0.8481,
      "step": 56470
    },
    {
      "epoch": 0.9915904422479327,
      "grad_norm": 0.06421954434437761,
      "learning_rate": 0.00018426136268116797,
      "loss": 0.846,
      "step": 56480
    },
    {
      "epoch": 0.9917660071279342,
      "grad_norm": 0.08337476812986204,
      "learning_rate": 0.00018425533347051572,
      "loss": 0.8477,
      "step": 56490
    },
    {
      "epoch": 0.9919415720079355,
      "grad_norm": 0.05520254983394486,
      "learning_rate": 0.0001842493032050024,
      "loss": 0.8527,
      "step": 56500
    },
    {
      "epoch": 0.9921171368879369,
      "grad_norm": 0.09347147933649798,
      "learning_rate": 0.0001842432718847045,
      "loss": 0.8524,
      "step": 56510
    },
    {
      "epoch": 0.9922927017679384,
      "grad_norm": 0.07481419935521504,
      "learning_rate": 0.0001842372395096984,
      "loss": 0.8475,
      "step": 56520
    },
    {
      "epoch": 0.9924682666479397,
      "grad_norm": 0.06542498658361165,
      "learning_rate": 0.00018423120608006052,
      "loss": 0.8549,
      "step": 56530
    },
    {
      "epoch": 0.9926438315279411,
      "grad_norm": 0.05573432662120896,
      "learning_rate": 0.0001842251715958673,
      "loss": 0.8516,
      "step": 56540
    },
    {
      "epoch": 0.9928193964079426,
      "grad_norm": 0.07036756734125,
      "learning_rate": 0.00018421913605719522,
      "loss": 0.8438,
      "step": 56550
    },
    {
      "epoch": 0.992994961287944,
      "grad_norm": 0.05619597439454494,
      "learning_rate": 0.00018421309946412074,
      "loss": 0.8543,
      "step": 56560
    },
    {
      "epoch": 0.9931705261679453,
      "grad_norm": 0.0649945714224265,
      "learning_rate": 0.00018420706181672031,
      "loss": 0.8502,
      "step": 56570
    },
    {
      "epoch": 0.9933460910479468,
      "grad_norm": 0.07061338610675556,
      "learning_rate": 0.00018420102311507048,
      "loss": 0.8543,
      "step": 56580
    },
    {
      "epoch": 0.9935216559279482,
      "grad_norm": 0.05051706594064909,
      "learning_rate": 0.00018419498335924775,
      "loss": 0.8408,
      "step": 56590
    },
    {
      "epoch": 0.9936972208079495,
      "grad_norm": 0.09295975910918253,
      "learning_rate": 0.00018418894254932863,
      "loss": 0.8484,
      "step": 56600
    },
    {
      "epoch": 0.993872785687951,
      "grad_norm": 0.05899447647305215,
      "learning_rate": 0.00018418290068538965,
      "loss": 0.8572,
      "step": 56610
    },
    {
      "epoch": 0.9940483505679524,
      "grad_norm": 0.05268594729296781,
      "learning_rate": 0.00018417685776750742,
      "loss": 0.8443,
      "step": 56620
    },
    {
      "epoch": 0.9942239154479537,
      "grad_norm": 0.06315275451873188,
      "learning_rate": 0.0001841708137957584,
      "loss": 0.8519,
      "step": 56630
    },
    {
      "epoch": 0.9943994803279552,
      "grad_norm": 0.05647097474578115,
      "learning_rate": 0.00018416476877021926,
      "loss": 0.8461,
      "step": 56640
    },
    {
      "epoch": 0.9945750452079566,
      "grad_norm": 0.06620048972860083,
      "learning_rate": 0.00018415872269096658,
      "loss": 0.8536,
      "step": 56650
    },
    {
      "epoch": 0.994750610087958,
      "grad_norm": 0.04766645031854222,
      "learning_rate": 0.0001841526755580769,
      "loss": 0.8472,
      "step": 56660
    },
    {
      "epoch": 0.9949261749679594,
      "grad_norm": 0.07517769258811252,
      "learning_rate": 0.00018414662737162695,
      "loss": 0.8509,
      "step": 56670
    },
    {
      "epoch": 0.9951017398479608,
      "grad_norm": 0.06786891644699258,
      "learning_rate": 0.00018414057813169324,
      "loss": 0.8533,
      "step": 56680
    },
    {
      "epoch": 0.9952773047279622,
      "grad_norm": 0.05362721374942034,
      "learning_rate": 0.00018413452783835248,
      "loss": 0.8534,
      "step": 56690
    },
    {
      "epoch": 0.9954528696079636,
      "grad_norm": 0.042971255971688375,
      "learning_rate": 0.00018412847649168133,
      "loss": 0.8427,
      "step": 56700
    },
    {
      "epoch": 0.995628434487965,
      "grad_norm": 0.0652211044519968,
      "learning_rate": 0.00018412242409175647,
      "loss": 0.8417,
      "step": 56710
    },
    {
      "epoch": 0.9958039993679665,
      "grad_norm": 0.057365994117766286,
      "learning_rate": 0.0001841163706386545,
      "loss": 0.8581,
      "step": 56720
    },
    {
      "epoch": 0.9959795642479679,
      "grad_norm": 0.07947493003352432,
      "learning_rate": 0.00018411031613245227,
      "loss": 0.8607,
      "step": 56730
    },
    {
      "epoch": 0.9961551291279692,
      "grad_norm": 0.06411656465794277,
      "learning_rate": 0.00018410426057322634,
      "loss": 0.8538,
      "step": 56740
    },
    {
      "epoch": 0.9963306940079707,
      "grad_norm": 0.06188887999291985,
      "learning_rate": 0.00018409820396105353,
      "loss": 0.8576,
      "step": 56750
    },
    {
      "epoch": 0.9965062588879721,
      "grad_norm": 0.05357051268500942,
      "learning_rate": 0.00018409214629601056,
      "loss": 0.8506,
      "step": 56760
    },
    {
      "epoch": 0.9966818237679734,
      "grad_norm": 0.09782788521548237,
      "learning_rate": 0.00018408608757817415,
      "loss": 0.8592,
      "step": 56770
    },
    {
      "epoch": 0.9968573886479749,
      "grad_norm": 0.06770789578374656,
      "learning_rate": 0.0001840800278076211,
      "loss": 0.8542,
      "step": 56780
    },
    {
      "epoch": 0.9970329535279763,
      "grad_norm": 0.07003504006397199,
      "learning_rate": 0.00018407396698442817,
      "loss": 0.8452,
      "step": 56790
    },
    {
      "epoch": 0.9972085184079776,
      "grad_norm": 0.07361054241525321,
      "learning_rate": 0.00018406790510867217,
      "loss": 0.8472,
      "step": 56800
    },
    {
      "epoch": 0.9973840832879791,
      "grad_norm": 0.05888558212884699,
      "learning_rate": 0.00018406184218042987,
      "loss": 0.8499,
      "step": 56810
    },
    {
      "epoch": 0.9975596481679805,
      "grad_norm": 0.05470370472504471,
      "learning_rate": 0.00018405577819977814,
      "loss": 0.8466,
      "step": 56820
    },
    {
      "epoch": 0.9977352130479819,
      "grad_norm": 0.059073399945017376,
      "learning_rate": 0.00018404971316679375,
      "loss": 0.8539,
      "step": 56830
    },
    {
      "epoch": 0.9979107779279833,
      "grad_norm": 0.06191039384330902,
      "learning_rate": 0.0001840436470815536,
      "loss": 0.8511,
      "step": 56840
    },
    {
      "epoch": 0.9980863428079847,
      "grad_norm": 0.057650708398638235,
      "learning_rate": 0.00018403757994413453,
      "loss": 0.8495,
      "step": 56850
    },
    {
      "epoch": 0.9982619076879861,
      "grad_norm": 0.10941963656602653,
      "learning_rate": 0.00018403151175461338,
      "loss": 0.8454,
      "step": 56860
    },
    {
      "epoch": 0.9984374725679875,
      "grad_norm": 0.0520239432752115,
      "learning_rate": 0.0001840254425130671,
      "loss": 0.8553,
      "step": 56870
    },
    {
      "epoch": 0.9986130374479889,
      "grad_norm": 0.06854825422446295,
      "learning_rate": 0.00018401937221957252,
      "loss": 0.8448,
      "step": 56880
    },
    {
      "epoch": 0.9987886023279903,
      "grad_norm": 0.08650789307158079,
      "learning_rate": 0.00018401330087420659,
      "loss": 0.8452,
      "step": 56890
    },
    {
      "epoch": 0.9989641672079917,
      "grad_norm": 0.0737723513275301,
      "learning_rate": 0.00018400722847704626,
      "loss": 0.8511,
      "step": 56900
    },
    {
      "epoch": 0.9991397320879931,
      "grad_norm": 0.06481005578264762,
      "learning_rate": 0.0001840011550281684,
      "loss": 0.8506,
      "step": 56910
    },
    {
      "epoch": 0.9993152969679945,
      "grad_norm": 0.07010584482412711,
      "learning_rate": 0.00018399508052765,
      "loss": 0.8511,
      "step": 56920
    },
    {
      "epoch": 0.999490861847996,
      "grad_norm": 0.05847443267702205,
      "learning_rate": 0.00018398900497556804,
      "loss": 0.8477,
      "step": 56930
    },
    {
      "epoch": 0.9996664267279973,
      "grad_norm": 0.07484785510055697,
      "learning_rate": 0.00018398292837199946,
      "loss": 0.8526,
      "step": 56940
    },
    {
      "epoch": 0.9998419916079987,
      "grad_norm": 0.04767599237392813,
      "learning_rate": 0.00018397685071702133,
      "loss": 0.8498,
      "step": 56950
    },
    {
      "epoch": 1.000017556488,
      "grad_norm": 0.06918703166494092,
      "learning_rate": 0.00018397077201071056,
      "loss": 0.8581,
      "step": 56960
    },
    {
      "epoch": 1.0001931213680015,
      "grad_norm": 0.06473117854452935,
      "learning_rate": 0.0001839646922531442,
      "loss": 0.8505,
      "step": 56970
    },
    {
      "epoch": 1.000368686248003,
      "grad_norm": 0.05680574688835367,
      "learning_rate": 0.00018395861144439928,
      "loss": 0.8445,
      "step": 56980
    },
    {
      "epoch": 1.0005442511280043,
      "grad_norm": 0.06484087942590894,
      "learning_rate": 0.00018395252958455286,
      "loss": 0.851,
      "step": 56990
    },
    {
      "epoch": 1.0007198160080057,
      "grad_norm": 0.05203039329303289,
      "learning_rate": 0.00018394644667368203,
      "loss": 0.8556,
      "step": 57000
    },
    {
      "epoch": 1.0008953808880072,
      "grad_norm": 0.05403870907851163,
      "learning_rate": 0.00018394036271186378,
      "loss": 0.8478,
      "step": 57010
    },
    {
      "epoch": 1.0010709457680085,
      "grad_norm": 0.08053702986131214,
      "learning_rate": 0.00018393427769917525,
      "loss": 0.8586,
      "step": 57020
    },
    {
      "epoch": 1.00124651064801,
      "grad_norm": 0.05937897307065659,
      "learning_rate": 0.00018392819163569354,
      "loss": 0.859,
      "step": 57030
    },
    {
      "epoch": 1.0014220755280114,
      "grad_norm": 0.09016981125393919,
      "learning_rate": 0.00018392210452149575,
      "loss": 0.8473,
      "step": 57040
    },
    {
      "epoch": 1.0015976404080127,
      "grad_norm": 0.058537850534267896,
      "learning_rate": 0.00018391601635665898,
      "loss": 0.8606,
      "step": 57050
    },
    {
      "epoch": 1.0017732052880142,
      "grad_norm": 0.06643852484811973,
      "learning_rate": 0.00018390992714126042,
      "loss": 0.8476,
      "step": 57060
    },
    {
      "epoch": 1.0019487701680156,
      "grad_norm": 0.06052649080438298,
      "learning_rate": 0.0001839038368753772,
      "loss": 0.8558,
      "step": 57070
    },
    {
      "epoch": 1.002124335048017,
      "grad_norm": 0.052190441801842014,
      "learning_rate": 0.00018389774555908642,
      "loss": 0.8544,
      "step": 57080
    },
    {
      "epoch": 1.0022998999280184,
      "grad_norm": 0.08415895354995483,
      "learning_rate": 0.00018389165319246536,
      "loss": 0.8469,
      "step": 57090
    },
    {
      "epoch": 1.0024754648080199,
      "grad_norm": 0.09632799307428629,
      "learning_rate": 0.00018388555977559117,
      "loss": 0.8523,
      "step": 57100
    },
    {
      "epoch": 1.0026510296880211,
      "grad_norm": 0.06392317303066189,
      "learning_rate": 0.00018387946530854107,
      "loss": 0.846,
      "step": 57110
    },
    {
      "epoch": 1.0028265945680226,
      "grad_norm": 0.06347544817496636,
      "learning_rate": 0.0001838733697913922,
      "loss": 0.8507,
      "step": 57120
    },
    {
      "epoch": 1.003002159448024,
      "grad_norm": 0.06965695771566174,
      "learning_rate": 0.0001838672732242219,
      "loss": 0.8573,
      "step": 57130
    },
    {
      "epoch": 1.0031777243280253,
      "grad_norm": 0.05952840127679832,
      "learning_rate": 0.00018386117560710737,
      "loss": 0.851,
      "step": 57140
    },
    {
      "epoch": 1.0033532892080268,
      "grad_norm": 0.0633402594274687,
      "learning_rate": 0.00018385507694012584,
      "loss": 0.8536,
      "step": 57150
    },
    {
      "epoch": 1.0035288540880283,
      "grad_norm": 0.08668062767967304,
      "learning_rate": 0.0001838489772233546,
      "loss": 0.8536,
      "step": 57160
    },
    {
      "epoch": 1.0037044189680295,
      "grad_norm": 0.06663811168778713,
      "learning_rate": 0.00018384287645687098,
      "loss": 0.8539,
      "step": 57170
    },
    {
      "epoch": 1.003879983848031,
      "grad_norm": 0.06462644160106652,
      "learning_rate": 0.00018383677464075223,
      "loss": 0.8485,
      "step": 57180
    },
    {
      "epoch": 1.0040555487280325,
      "grad_norm": 0.07479265302812643,
      "learning_rate": 0.00018383067177507565,
      "loss": 0.8518,
      "step": 57190
    },
    {
      "epoch": 1.0042311136080337,
      "grad_norm": 0.06858342921845773,
      "learning_rate": 0.00018382456785991856,
      "loss": 0.8568,
      "step": 57200
    },
    {
      "epoch": 1.0044066784880352,
      "grad_norm": 0.046862929655702074,
      "learning_rate": 0.00018381846289535836,
      "loss": 0.8489,
      "step": 57210
    },
    {
      "epoch": 1.0045822433680367,
      "grad_norm": 0.071670336053507,
      "learning_rate": 0.00018381235688147237,
      "loss": 0.8503,
      "step": 57220
    },
    {
      "epoch": 1.004757808248038,
      "grad_norm": 0.0578126278571733,
      "learning_rate": 0.00018380624981833793,
      "loss": 0.8577,
      "step": 57230
    },
    {
      "epoch": 1.0049333731280394,
      "grad_norm": 0.061097077520016155,
      "learning_rate": 0.00018380014170603243,
      "loss": 0.8535,
      "step": 57240
    },
    {
      "epoch": 1.005108938008041,
      "grad_norm": 0.07323172040660607,
      "learning_rate": 0.00018379403254463328,
      "loss": 0.8488,
      "step": 57250
    },
    {
      "epoch": 1.0052845028880424,
      "grad_norm": 0.052290820354976446,
      "learning_rate": 0.00018378792233421788,
      "loss": 0.8452,
      "step": 57260
    },
    {
      "epoch": 1.0054600677680436,
      "grad_norm": 0.09005740821223741,
      "learning_rate": 0.0001837818110748636,
      "loss": 0.843,
      "step": 57270
    },
    {
      "epoch": 1.0056356326480451,
      "grad_norm": 0.06213299299029477,
      "learning_rate": 0.00018377569876664793,
      "loss": 0.8542,
      "step": 57280
    },
    {
      "epoch": 1.0058111975280466,
      "grad_norm": 0.05617461346883926,
      "learning_rate": 0.00018376958540964827,
      "loss": 0.8583,
      "step": 57290
    },
    {
      "epoch": 1.0059867624080479,
      "grad_norm": 0.06148942962679741,
      "learning_rate": 0.00018376347100394213,
      "loss": 0.8453,
      "step": 57300
    },
    {
      "epoch": 1.0061623272880493,
      "grad_norm": 0.05474982171720353,
      "learning_rate": 0.00018375735554960692,
      "loss": 0.8489,
      "step": 57310
    },
    {
      "epoch": 1.0063378921680508,
      "grad_norm": 0.056435434873019746,
      "learning_rate": 0.00018375123904672023,
      "loss": 0.844,
      "step": 57320
    },
    {
      "epoch": 1.006513457048052,
      "grad_norm": 0.05010497524289432,
      "learning_rate": 0.0001837451214953594,
      "loss": 0.8537,
      "step": 57330
    },
    {
      "epoch": 1.0066890219280535,
      "grad_norm": 0.04836249912443953,
      "learning_rate": 0.0001837390028956021,
      "loss": 0.8527,
      "step": 57340
    },
    {
      "epoch": 1.006864586808055,
      "grad_norm": 0.08644161719035882,
      "learning_rate": 0.00018373288324752572,
      "loss": 0.8539,
      "step": 57350
    },
    {
      "epoch": 1.0070401516880563,
      "grad_norm": 0.09109686916040854,
      "learning_rate": 0.00018372676255120786,
      "loss": 0.8473,
      "step": 57360
    },
    {
      "epoch": 1.0072157165680578,
      "grad_norm": 0.06927543306610721,
      "learning_rate": 0.00018372064080672605,
      "loss": 0.8467,
      "step": 57370
    },
    {
      "epoch": 1.0073912814480592,
      "grad_norm": 0.06497491409870639,
      "learning_rate": 0.00018371451801415788,
      "loss": 0.849,
      "step": 57380
    },
    {
      "epoch": 1.0075668463280605,
      "grad_norm": 0.08051949144149584,
      "learning_rate": 0.0001837083941735809,
      "loss": 0.8448,
      "step": 57390
    },
    {
      "epoch": 1.007742411208062,
      "grad_norm": 0.05035970355877847,
      "learning_rate": 0.00018370226928507275,
      "loss": 0.8484,
      "step": 57400
    },
    {
      "epoch": 1.0079179760880634,
      "grad_norm": 0.06262873140271641,
      "learning_rate": 0.000183696143348711,
      "loss": 0.8549,
      "step": 57410
    },
    {
      "epoch": 1.0080935409680647,
      "grad_norm": 0.07088847967570565,
      "learning_rate": 0.00018369001636457324,
      "loss": 0.8447,
      "step": 57420
    },
    {
      "epoch": 1.0082691058480662,
      "grad_norm": 0.06412971045479393,
      "learning_rate": 0.00018368388833273712,
      "loss": 0.8451,
      "step": 57430
    },
    {
      "epoch": 1.0084446707280676,
      "grad_norm": 0.04731865317324123,
      "learning_rate": 0.00018367775925328032,
      "loss": 0.8549,
      "step": 57440
    },
    {
      "epoch": 1.008620235608069,
      "grad_norm": 0.05165386632671862,
      "learning_rate": 0.00018367162912628045,
      "loss": 0.8495,
      "step": 57450
    },
    {
      "epoch": 1.0087958004880704,
      "grad_norm": 0.07100144485889673,
      "learning_rate": 0.0001836654979518152,
      "loss": 0.8523,
      "step": 57460
    },
    {
      "epoch": 1.0089713653680719,
      "grad_norm": 0.054626970270764,
      "learning_rate": 0.00018365936572996224,
      "loss": 0.8424,
      "step": 57470
    },
    {
      "epoch": 1.0091469302480731,
      "grad_norm": 0.05375977869279286,
      "learning_rate": 0.00018365323246079927,
      "loss": 0.8577,
      "step": 57480
    },
    {
      "epoch": 1.0093224951280746,
      "grad_norm": 0.0633282068624186,
      "learning_rate": 0.000183647098144404,
      "loss": 0.8534,
      "step": 57490
    },
    {
      "epoch": 1.009498060008076,
      "grad_norm": 0.05454254288494738,
      "learning_rate": 0.00018364096278085414,
      "loss": 0.8552,
      "step": 57500
    },
    {
      "epoch": 1.0096736248880773,
      "grad_norm": 0.0808417513824939,
      "learning_rate": 0.00018363482637022752,
      "loss": 0.849,
      "step": 57510
    },
    {
      "epoch": 1.0098491897680788,
      "grad_norm": 0.08147130000666776,
      "learning_rate": 0.00018362868891260174,
      "loss": 0.8585,
      "step": 57520
    },
    {
      "epoch": 1.0100247546480803,
      "grad_norm": 0.04893691527522899,
      "learning_rate": 0.00018362255040805468,
      "loss": 0.8511,
      "step": 57530
    },
    {
      "epoch": 1.0102003195280815,
      "grad_norm": 0.0567422245883815,
      "learning_rate": 0.000183616410856664,
      "loss": 0.8494,
      "step": 57540
    },
    {
      "epoch": 1.010375884408083,
      "grad_norm": 0.04703852926032065,
      "learning_rate": 0.00018361027025850762,
      "loss": 0.8543,
      "step": 57550
    },
    {
      "epoch": 1.0105514492880845,
      "grad_norm": 0.08210293026836615,
      "learning_rate": 0.00018360412861366328,
      "loss": 0.8496,
      "step": 57560
    },
    {
      "epoch": 1.0107270141680857,
      "grad_norm": 0.07525212557536587,
      "learning_rate": 0.0001835979859222088,
      "loss": 0.8539,
      "step": 57570
    },
    {
      "epoch": 1.0109025790480872,
      "grad_norm": 0.05547109797913806,
      "learning_rate": 0.000183591842184222,
      "loss": 0.8555,
      "step": 57580
    },
    {
      "epoch": 1.0110781439280887,
      "grad_norm": 0.07657341377584027,
      "learning_rate": 0.00018358569739978075,
      "loss": 0.8601,
      "step": 57590
    },
    {
      "epoch": 1.01125370880809,
      "grad_norm": 0.05401781220301768,
      "learning_rate": 0.00018357955156896285,
      "loss": 0.8501,
      "step": 57600
    },
    {
      "epoch": 1.0114292736880914,
      "grad_norm": 0.0541357695761068,
      "learning_rate": 0.00018357340469184625,
      "loss": 0.8524,
      "step": 57610
    },
    {
      "epoch": 1.011604838568093,
      "grad_norm": 0.061427691261375504,
      "learning_rate": 0.00018356725676850876,
      "loss": 0.8464,
      "step": 57620
    },
    {
      "epoch": 1.0117804034480942,
      "grad_norm": 0.05986837127715629,
      "learning_rate": 0.00018356110779902828,
      "loss": 0.8561,
      "step": 57630
    },
    {
      "epoch": 1.0119559683280956,
      "grad_norm": 0.0742637065059885,
      "learning_rate": 0.00018355495778348279,
      "loss": 0.8498,
      "step": 57640
    },
    {
      "epoch": 1.0121315332080971,
      "grad_norm": 0.06785985693396415,
      "learning_rate": 0.0001835488067219501,
      "loss": 0.855,
      "step": 57650
    },
    {
      "epoch": 1.0123070980880984,
      "grad_norm": 0.05278582111640081,
      "learning_rate": 0.00018354265461450827,
      "loss": 0.8528,
      "step": 57660
    },
    {
      "epoch": 1.0124826629680999,
      "grad_norm": 0.07342483253433364,
      "learning_rate": 0.00018353650146123515,
      "loss": 0.8467,
      "step": 57670
    },
    {
      "epoch": 1.0126582278481013,
      "grad_norm": 0.05807942433717392,
      "learning_rate": 0.0001835303472622088,
      "loss": 0.8594,
      "step": 57680
    },
    {
      "epoch": 1.0128337927281026,
      "grad_norm": 0.056591697621899806,
      "learning_rate": 0.00018352419201750708,
      "loss": 0.8501,
      "step": 57690
    },
    {
      "epoch": 1.013009357608104,
      "grad_norm": 0.05640676716357161,
      "learning_rate": 0.00018351803572720803,
      "loss": 0.8518,
      "step": 57700
    },
    {
      "epoch": 1.0131849224881055,
      "grad_norm": 0.0689348591332243,
      "learning_rate": 0.00018351187839138963,
      "loss": 0.8418,
      "step": 57710
    },
    {
      "epoch": 1.0133604873681068,
      "grad_norm": 0.076867691032374,
      "learning_rate": 0.00018350572001012998,
      "loss": 0.8519,
      "step": 57720
    },
    {
      "epoch": 1.0135360522481083,
      "grad_norm": 0.0729871281239425,
      "learning_rate": 0.00018349956058350702,
      "loss": 0.844,
      "step": 57730
    },
    {
      "epoch": 1.0137116171281098,
      "grad_norm": 0.05826216539829217,
      "learning_rate": 0.00018349340011159881,
      "loss": 0.8492,
      "step": 57740
    },
    {
      "epoch": 1.013887182008111,
      "grad_norm": 0.06688871895810904,
      "learning_rate": 0.0001834872385944834,
      "loss": 0.8413,
      "step": 57750
    },
    {
      "epoch": 1.0140627468881125,
      "grad_norm": 0.07285981851454083,
      "learning_rate": 0.0001834810760322389,
      "loss": 0.852,
      "step": 57760
    },
    {
      "epoch": 1.014238311768114,
      "grad_norm": 0.04928503133337006,
      "learning_rate": 0.0001834749124249433,
      "loss": 0.8533,
      "step": 57770
    },
    {
      "epoch": 1.0144138766481152,
      "grad_norm": 0.08917165412230857,
      "learning_rate": 0.0001834687477726748,
      "loss": 0.8502,
      "step": 57780
    },
    {
      "epoch": 1.0145894415281167,
      "grad_norm": 0.04826672034599844,
      "learning_rate": 0.00018346258207551145,
      "loss": 0.8467,
      "step": 57790
    },
    {
      "epoch": 1.0147650064081182,
      "grad_norm": 0.07794241527952483,
      "learning_rate": 0.0001834564153335314,
      "loss": 0.8546,
      "step": 57800
    },
    {
      "epoch": 1.0149405712881194,
      "grad_norm": 0.06603981556994042,
      "learning_rate": 0.00018345024754681276,
      "loss": 0.8502,
      "step": 57810
    },
    {
      "epoch": 1.015116136168121,
      "grad_norm": 0.050124003714769716,
      "learning_rate": 0.00018344407871543367,
      "loss": 0.8572,
      "step": 57820
    },
    {
      "epoch": 1.0152917010481224,
      "grad_norm": 0.07128740496516624,
      "learning_rate": 0.0001834379088394723,
      "loss": 0.8581,
      "step": 57830
    },
    {
      "epoch": 1.0154672659281236,
      "grad_norm": 0.059345822393300336,
      "learning_rate": 0.00018343173791900684,
      "loss": 0.844,
      "step": 57840
    },
    {
      "epoch": 1.0156428308081251,
      "grad_norm": 0.10336341097415111,
      "learning_rate": 0.0001834255659541154,
      "loss": 0.8474,
      "step": 57850
    },
    {
      "epoch": 1.0158183956881266,
      "grad_norm": 0.07664044718176762,
      "learning_rate": 0.00018341939294487633,
      "loss": 0.8511,
      "step": 57860
    },
    {
      "epoch": 1.0159939605681279,
      "grad_norm": 0.06834540617730213,
      "learning_rate": 0.0001834132188913677,
      "loss": 0.8602,
      "step": 57870
    },
    {
      "epoch": 1.0161695254481293,
      "grad_norm": 0.07175827046058275,
      "learning_rate": 0.00018340704379366784,
      "loss": 0.8413,
      "step": 57880
    },
    {
      "epoch": 1.0163450903281308,
      "grad_norm": 0.06567441791757853,
      "learning_rate": 0.00018340086765185492,
      "loss": 0.8501,
      "step": 57890
    },
    {
      "epoch": 1.016520655208132,
      "grad_norm": 0.04870018871827384,
      "learning_rate": 0.0001833946904660072,
      "loss": 0.8461,
      "step": 57900
    },
    {
      "epoch": 1.0166962200881335,
      "grad_norm": 0.07386237666885187,
      "learning_rate": 0.00018338851223620296,
      "loss": 0.8477,
      "step": 57910
    },
    {
      "epoch": 1.016871784968135,
      "grad_norm": 0.056876958540823225,
      "learning_rate": 0.0001833823329625205,
      "loss": 0.8499,
      "step": 57920
    },
    {
      "epoch": 1.0170473498481363,
      "grad_norm": 0.08787092546782102,
      "learning_rate": 0.0001833761526450381,
      "loss": 0.8583,
      "step": 57930
    },
    {
      "epoch": 1.0172229147281378,
      "grad_norm": 0.05715370023223837,
      "learning_rate": 0.00018336997128383403,
      "loss": 0.8541,
      "step": 57940
    },
    {
      "epoch": 1.0173984796081392,
      "grad_norm": 0.07200939597576614,
      "learning_rate": 0.00018336378887898666,
      "loss": 0.8453,
      "step": 57950
    },
    {
      "epoch": 1.0175740444881405,
      "grad_norm": 0.053764345622941495,
      "learning_rate": 0.00018335760543057429,
      "loss": 0.8404,
      "step": 57960
    },
    {
      "epoch": 1.017749609368142,
      "grad_norm": 0.06337248060573794,
      "learning_rate": 0.00018335142093867525,
      "loss": 0.8551,
      "step": 57970
    },
    {
      "epoch": 1.0179251742481434,
      "grad_norm": 0.06600124555816574,
      "learning_rate": 0.00018334523540336796,
      "loss": 0.8517,
      "step": 57980
    },
    {
      "epoch": 1.018100739128145,
      "grad_norm": 0.06953754689030639,
      "learning_rate": 0.00018333904882473074,
      "loss": 0.8548,
      "step": 57990
    },
    {
      "epoch": 1.0182763040081462,
      "grad_norm": 0.04981018855039167,
      "learning_rate": 0.000183332861202842,
      "loss": 0.8419,
      "step": 58000
    },
    {
      "epoch": 1.0184518688881476,
      "grad_norm": 0.0652896179346685,
      "learning_rate": 0.00018332667253778012,
      "loss": 0.8508,
      "step": 58010
    },
    {
      "epoch": 1.0186274337681491,
      "grad_norm": 0.05761211532589112,
      "learning_rate": 0.00018332048282962353,
      "loss": 0.8524,
      "step": 58020
    },
    {
      "epoch": 1.0188029986481504,
      "grad_norm": 0.08594992803423747,
      "learning_rate": 0.00018331429207845066,
      "loss": 0.8494,
      "step": 58030
    },
    {
      "epoch": 1.0189785635281519,
      "grad_norm": 0.059993933926691244,
      "learning_rate": 0.00018330810028433991,
      "loss": 0.8532,
      "step": 58040
    },
    {
      "epoch": 1.0191541284081533,
      "grad_norm": 0.09186723659652489,
      "learning_rate": 0.00018330190744736977,
      "loss": 0.8537,
      "step": 58050
    },
    {
      "epoch": 1.0193296932881546,
      "grad_norm": 0.07865287952263571,
      "learning_rate": 0.00018329571356761866,
      "loss": 0.8593,
      "step": 58060
    },
    {
      "epoch": 1.019505258168156,
      "grad_norm": 0.07047124366077023,
      "learning_rate": 0.00018328951864516512,
      "loss": 0.8456,
      "step": 58070
    },
    {
      "epoch": 1.0196808230481575,
      "grad_norm": 0.07841198988301461,
      "learning_rate": 0.0001832833226800876,
      "loss": 0.8514,
      "step": 58080
    },
    {
      "epoch": 1.0198563879281588,
      "grad_norm": 0.09499799047554278,
      "learning_rate": 0.00018327712567246462,
      "loss": 0.8577,
      "step": 58090
    },
    {
      "epoch": 1.0200319528081603,
      "grad_norm": 0.05308330059468653,
      "learning_rate": 0.00018327092762237466,
      "loss": 0.8549,
      "step": 58100
    },
    {
      "epoch": 1.0202075176881618,
      "grad_norm": 0.06660770225212095,
      "learning_rate": 0.0001832647285298963,
      "loss": 0.8545,
      "step": 58110
    },
    {
      "epoch": 1.020383082568163,
      "grad_norm": 0.062378153627281864,
      "learning_rate": 0.00018325852839510808,
      "loss": 0.8504,
      "step": 58120
    },
    {
      "epoch": 1.0205586474481645,
      "grad_norm": 0.04788546901018771,
      "learning_rate": 0.00018325232721808852,
      "loss": 0.8428,
      "step": 58130
    },
    {
      "epoch": 1.020734212328166,
      "grad_norm": 0.0705226793792809,
      "learning_rate": 0.0001832461249989162,
      "loss": 0.8547,
      "step": 58140
    },
    {
      "epoch": 1.0209097772081672,
      "grad_norm": 0.06192753789002412,
      "learning_rate": 0.00018323992173766975,
      "loss": 0.852,
      "step": 58150
    },
    {
      "epoch": 1.0210853420881687,
      "grad_norm": 0.06102591086115269,
      "learning_rate": 0.0001832337174344277,
      "loss": 0.8576,
      "step": 58160
    },
    {
      "epoch": 1.0212609069681702,
      "grad_norm": 0.06295166686828899,
      "learning_rate": 0.0001832275120892687,
      "loss": 0.8547,
      "step": 58170
    },
    {
      "epoch": 1.0214364718481714,
      "grad_norm": 0.07093651005780331,
      "learning_rate": 0.00018322130570227134,
      "loss": 0.8569,
      "step": 58180
    },
    {
      "epoch": 1.021612036728173,
      "grad_norm": 0.05713028662149447,
      "learning_rate": 0.00018321509827351427,
      "loss": 0.8484,
      "step": 58190
    },
    {
      "epoch": 1.0217876016081744,
      "grad_norm": 0.07693604598521016,
      "learning_rate": 0.00018320888980307622,
      "loss": 0.8544,
      "step": 58200
    },
    {
      "epoch": 1.0219631664881756,
      "grad_norm": 0.05192900159407114,
      "learning_rate": 0.00018320268029103573,
      "loss": 0.8525,
      "step": 58210
    },
    {
      "epoch": 1.0221387313681771,
      "grad_norm": 0.06363163802715563,
      "learning_rate": 0.00018319646973747152,
      "loss": 0.8564,
      "step": 58220
    },
    {
      "epoch": 1.0223142962481786,
      "grad_norm": 0.07238007669623532,
      "learning_rate": 0.0001831902581424623,
      "loss": 0.8525,
      "step": 58230
    },
    {
      "epoch": 1.0224898611281799,
      "grad_norm": 0.07321827542421633,
      "learning_rate": 0.00018318404550608677,
      "loss": 0.8515,
      "step": 58240
    },
    {
      "epoch": 1.0226654260081813,
      "grad_norm": 0.08254472669492502,
      "learning_rate": 0.00018317783182842363,
      "loss": 0.849,
      "step": 58250
    },
    {
      "epoch": 1.0228409908881828,
      "grad_norm": 0.0501080021301721,
      "learning_rate": 0.00018317161710955162,
      "loss": 0.8582,
      "step": 58260
    },
    {
      "epoch": 1.023016555768184,
      "grad_norm": 0.06675235332927723,
      "learning_rate": 0.00018316540134954946,
      "loss": 0.8512,
      "step": 58270
    },
    {
      "epoch": 1.0231921206481855,
      "grad_norm": 0.06738256244009651,
      "learning_rate": 0.0001831591845484959,
      "loss": 0.8415,
      "step": 58280
    },
    {
      "epoch": 1.023367685528187,
      "grad_norm": 0.09662761242424181,
      "learning_rate": 0.00018315296670646974,
      "loss": 0.8637,
      "step": 58290
    },
    {
      "epoch": 1.0235432504081883,
      "grad_norm": 0.06284790630972853,
      "learning_rate": 0.00018314674782354972,
      "loss": 0.8558,
      "step": 58300
    },
    {
      "epoch": 1.0237188152881898,
      "grad_norm": 0.06755791319114997,
      "learning_rate": 0.0001831405278998147,
      "loss": 0.8514,
      "step": 58310
    },
    {
      "epoch": 1.0238943801681912,
      "grad_norm": 0.06423521556295124,
      "learning_rate": 0.00018313430693534346,
      "loss": 0.8554,
      "step": 58320
    },
    {
      "epoch": 1.0240699450481925,
      "grad_norm": 0.05839798261831497,
      "learning_rate": 0.00018312808493021478,
      "loss": 0.8443,
      "step": 58330
    },
    {
      "epoch": 1.024245509928194,
      "grad_norm": 0.05185306284459377,
      "learning_rate": 0.00018312186188450754,
      "loss": 0.8576,
      "step": 58340
    },
    {
      "epoch": 1.0244210748081954,
      "grad_norm": 0.08794994255450836,
      "learning_rate": 0.00018311563779830055,
      "loss": 0.8487,
      "step": 58350
    },
    {
      "epoch": 1.0245966396881967,
      "grad_norm": 0.06678542137420172,
      "learning_rate": 0.0001831094126716727,
      "loss": 0.8557,
      "step": 58360
    },
    {
      "epoch": 1.0247722045681982,
      "grad_norm": 0.05752658197096915,
      "learning_rate": 0.00018310318650470287,
      "loss": 0.8478,
      "step": 58370
    },
    {
      "epoch": 1.0249477694481997,
      "grad_norm": 0.05336561239335987,
      "learning_rate": 0.00018309695929746993,
      "loss": 0.8504,
      "step": 58380
    },
    {
      "epoch": 1.025123334328201,
      "grad_norm": 0.0653095826699686,
      "learning_rate": 0.00018309073105005278,
      "loss": 0.853,
      "step": 58390
    },
    {
      "epoch": 1.0252988992082024,
      "grad_norm": 0.07088007532003532,
      "learning_rate": 0.0001830845017625303,
      "loss": 0.8511,
      "step": 58400
    },
    {
      "epoch": 1.0254744640882039,
      "grad_norm": 0.06225010916695357,
      "learning_rate": 0.00018307827143498152,
      "loss": 0.8503,
      "step": 58410
    },
    {
      "epoch": 1.0256500289682051,
      "grad_norm": 0.05851263842302583,
      "learning_rate": 0.00018307204006748527,
      "loss": 0.8538,
      "step": 58420
    },
    {
      "epoch": 1.0258255938482066,
      "grad_norm": 0.0455350449293496,
      "learning_rate": 0.0001830658076601205,
      "loss": 0.8486,
      "step": 58430
    },
    {
      "epoch": 1.026001158728208,
      "grad_norm": 0.06889666850906602,
      "learning_rate": 0.0001830595742129663,
      "loss": 0.858,
      "step": 58440
    },
    {
      "epoch": 1.0261767236082093,
      "grad_norm": 0.05406070504910497,
      "learning_rate": 0.00018305333972610152,
      "loss": 0.8433,
      "step": 58450
    },
    {
      "epoch": 1.0263522884882108,
      "grad_norm": 0.06646126789983098,
      "learning_rate": 0.00018304710419960523,
      "loss": 0.8499,
      "step": 58460
    },
    {
      "epoch": 1.0265278533682123,
      "grad_norm": 0.05558721030509597,
      "learning_rate": 0.00018304086763355637,
      "loss": 0.8524,
      "step": 58470
    },
    {
      "epoch": 1.0267034182482135,
      "grad_norm": 0.058733889263952684,
      "learning_rate": 0.00018303463002803403,
      "loss": 0.8553,
      "step": 58480
    },
    {
      "epoch": 1.026878983128215,
      "grad_norm": 0.06225521108194014,
      "learning_rate": 0.00018302839138311717,
      "loss": 0.8517,
      "step": 58490
    },
    {
      "epoch": 1.0270545480082165,
      "grad_norm": 0.05676633906632645,
      "learning_rate": 0.0001830221516988849,
      "loss": 0.8512,
      "step": 58500
    },
    {
      "epoch": 1.0272301128882178,
      "grad_norm": 0.06188892767658575,
      "learning_rate": 0.00018301591097541627,
      "loss": 0.8558,
      "step": 58510
    },
    {
      "epoch": 1.0274056777682192,
      "grad_norm": 0.09088045287313301,
      "learning_rate": 0.00018300966921279025,
      "loss": 0.8551,
      "step": 58520
    },
    {
      "epoch": 1.0275812426482207,
      "grad_norm": 0.06335220246267807,
      "learning_rate": 0.00018300342641108604,
      "loss": 0.8441,
      "step": 58530
    },
    {
      "epoch": 1.027756807528222,
      "grad_norm": 0.06428816447566754,
      "learning_rate": 0.00018299718257038276,
      "loss": 0.8421,
      "step": 58540
    },
    {
      "epoch": 1.0279323724082234,
      "grad_norm": 0.06925074819337303,
      "learning_rate": 0.0001829909376907594,
      "loss": 0.8453,
      "step": 58550
    },
    {
      "epoch": 1.028107937288225,
      "grad_norm": 0.06387005737583633,
      "learning_rate": 0.00018298469177229512,
      "loss": 0.8528,
      "step": 58560
    },
    {
      "epoch": 1.0282835021682262,
      "grad_norm": 0.10172365344309713,
      "learning_rate": 0.00018297844481506917,
      "loss": 0.8447,
      "step": 58570
    },
    {
      "epoch": 1.0284590670482276,
      "grad_norm": 0.07149841377044328,
      "learning_rate": 0.00018297219681916055,
      "loss": 0.8586,
      "step": 58580
    },
    {
      "epoch": 1.0286346319282291,
      "grad_norm": 0.07472584738402481,
      "learning_rate": 0.00018296594778464847,
      "loss": 0.8533,
      "step": 58590
    },
    {
      "epoch": 1.0288101968082304,
      "grad_norm": 0.052000393300650975,
      "learning_rate": 0.0001829596977116122,
      "loss": 0.8467,
      "step": 58600
    },
    {
      "epoch": 1.0289857616882319,
      "grad_norm": 0.06855359931435703,
      "learning_rate": 0.00018295344660013081,
      "loss": 0.8511,
      "step": 58610
    },
    {
      "epoch": 1.0291613265682333,
      "grad_norm": 0.05464095678388631,
      "learning_rate": 0.00018294719445028352,
      "loss": 0.8488,
      "step": 58620
    },
    {
      "epoch": 1.0293368914482346,
      "grad_norm": 0.0844166638275016,
      "learning_rate": 0.0001829409412621496,
      "loss": 0.8619,
      "step": 58630
    },
    {
      "epoch": 1.029512456328236,
      "grad_norm": 0.057903772648458855,
      "learning_rate": 0.00018293468703580823,
      "loss": 0.8572,
      "step": 58640
    },
    {
      "epoch": 1.0296880212082375,
      "grad_norm": 0.058665558485607686,
      "learning_rate": 0.00018292843177133872,
      "loss": 0.8457,
      "step": 58650
    },
    {
      "epoch": 1.0298635860882388,
      "grad_norm": 0.06144688219378745,
      "learning_rate": 0.00018292217546882021,
      "loss": 0.8555,
      "step": 58660
    },
    {
      "epoch": 1.0300391509682403,
      "grad_norm": 0.05957869877002116,
      "learning_rate": 0.0001829159181283321,
      "loss": 0.8424,
      "step": 58670
    },
    {
      "epoch": 1.0302147158482418,
      "grad_norm": 0.07160555239836215,
      "learning_rate": 0.00018290965974995354,
      "loss": 0.8511,
      "step": 58680
    },
    {
      "epoch": 1.030390280728243,
      "grad_norm": 0.07016293469240979,
      "learning_rate": 0.00018290340033376396,
      "loss": 0.8572,
      "step": 58690
    },
    {
      "epoch": 1.0305658456082445,
      "grad_norm": 0.06605535971517287,
      "learning_rate": 0.00018289713987984258,
      "loss": 0.8498,
      "step": 58700
    },
    {
      "epoch": 1.030741410488246,
      "grad_norm": 0.06075683839084408,
      "learning_rate": 0.00018289087838826876,
      "loss": 0.8544,
      "step": 58710
    },
    {
      "epoch": 1.0309169753682474,
      "grad_norm": 0.06142870691664489,
      "learning_rate": 0.00018288461585912178,
      "loss": 0.8516,
      "step": 58720
    },
    {
      "epoch": 1.0310925402482487,
      "grad_norm": 0.05119327211567094,
      "learning_rate": 0.00018287835229248103,
      "loss": 0.8616,
      "step": 58730
    },
    {
      "epoch": 1.0312681051282502,
      "grad_norm": 0.07024997157113695,
      "learning_rate": 0.00018287208768842585,
      "loss": 0.8553,
      "step": 58740
    },
    {
      "epoch": 1.0314436700082517,
      "grad_norm": 0.056671990618684664,
      "learning_rate": 0.00018286582204703566,
      "loss": 0.8502,
      "step": 58750
    },
    {
      "epoch": 1.031619234888253,
      "grad_norm": 0.06241984767141441,
      "learning_rate": 0.0001828595553683898,
      "loss": 0.8535,
      "step": 58760
    },
    {
      "epoch": 1.0317947997682544,
      "grad_norm": 0.048390680865320494,
      "learning_rate": 0.00018285328765256772,
      "loss": 0.8465,
      "step": 58770
    },
    {
      "epoch": 1.0319703646482559,
      "grad_norm": 0.04764545418986805,
      "learning_rate": 0.00018284701889964875,
      "loss": 0.8447,
      "step": 58780
    },
    {
      "epoch": 1.0321459295282571,
      "grad_norm": 0.06851520235169642,
      "learning_rate": 0.0001828407491097124,
      "loss": 0.8403,
      "step": 58790
    },
    {
      "epoch": 1.0323214944082586,
      "grad_norm": 0.0771587622264935,
      "learning_rate": 0.00018283447828283802,
      "loss": 0.8494,
      "step": 58800
    },
    {
      "epoch": 1.03249705928826,
      "grad_norm": 0.06858580505860112,
      "learning_rate": 0.00018282820641910518,
      "loss": 0.8433,
      "step": 58810
    },
    {
      "epoch": 1.0326726241682613,
      "grad_norm": 0.05424672555433141,
      "learning_rate": 0.00018282193351859327,
      "loss": 0.8473,
      "step": 58820
    },
    {
      "epoch": 1.0328481890482628,
      "grad_norm": 0.06443954198974967,
      "learning_rate": 0.00018281565958138177,
      "loss": 0.852,
      "step": 58830
    },
    {
      "epoch": 1.0330237539282643,
      "grad_norm": 0.05247772383166401,
      "learning_rate": 0.00018280938460755018,
      "loss": 0.8564,
      "step": 58840
    },
    {
      "epoch": 1.0331993188082655,
      "grad_norm": 0.07251464391914239,
      "learning_rate": 0.00018280310859717798,
      "loss": 0.8541,
      "step": 58850
    },
    {
      "epoch": 1.033374883688267,
      "grad_norm": 0.06354074757832422,
      "learning_rate": 0.00018279683155034474,
      "loss": 0.8452,
      "step": 58860
    },
    {
      "epoch": 1.0335504485682685,
      "grad_norm": 0.08892528048696376,
      "learning_rate": 0.00018279055346712996,
      "loss": 0.8537,
      "step": 58870
    },
    {
      "epoch": 1.0337260134482698,
      "grad_norm": 0.07738347543511388,
      "learning_rate": 0.00018278427434761325,
      "loss": 0.8546,
      "step": 58880
    },
    {
      "epoch": 1.0339015783282712,
      "grad_norm": 0.04352591445287215,
      "learning_rate": 0.00018277799419187408,
      "loss": 0.8568,
      "step": 58890
    },
    {
      "epoch": 1.0340771432082727,
      "grad_norm": 0.09674667311516869,
      "learning_rate": 0.000182771712999992,
      "loss": 0.8634,
      "step": 58900
    },
    {
      "epoch": 1.034252708088274,
      "grad_norm": 0.06694275351328151,
      "learning_rate": 0.00018276543077204668,
      "loss": 0.8557,
      "step": 58910
    },
    {
      "epoch": 1.0344282729682754,
      "grad_norm": 0.06078491227270534,
      "learning_rate": 0.0001827591475081177,
      "loss": 0.8532,
      "step": 58920
    },
    {
      "epoch": 1.034603837848277,
      "grad_norm": 0.05274242692029416,
      "learning_rate": 0.0001827528632082846,
      "loss": 0.8469,
      "step": 58930
    },
    {
      "epoch": 1.0347794027282782,
      "grad_norm": 0.06012145022910564,
      "learning_rate": 0.00018274657787262713,
      "loss": 0.8478,
      "step": 58940
    },
    {
      "epoch": 1.0349549676082797,
      "grad_norm": 0.0584502284113655,
      "learning_rate": 0.0001827402915012248,
      "loss": 0.8466,
      "step": 58950
    },
    {
      "epoch": 1.0351305324882811,
      "grad_norm": 0.05528668149244231,
      "learning_rate": 0.0001827340040941573,
      "loss": 0.856,
      "step": 58960
    },
    {
      "epoch": 1.0353060973682824,
      "grad_norm": 0.052555592115292475,
      "learning_rate": 0.00018272771565150435,
      "loss": 0.8534,
      "step": 58970
    },
    {
      "epoch": 1.0354816622482839,
      "grad_norm": 0.06579901303192404,
      "learning_rate": 0.00018272142617334556,
      "loss": 0.8522,
      "step": 58980
    },
    {
      "epoch": 1.0356572271282853,
      "grad_norm": 0.04569085828393595,
      "learning_rate": 0.00018271513565976062,
      "loss": 0.8581,
      "step": 58990
    },
    {
      "epoch": 1.0358327920082866,
      "grad_norm": 0.0635263080873014,
      "learning_rate": 0.00018270884411082926,
      "loss": 0.8504,
      "step": 59000
    },
    {
      "epoch": 1.036008356888288,
      "grad_norm": 0.06050521301764601,
      "learning_rate": 0.0001827025515266312,
      "loss": 0.8539,
      "step": 59010
    },
    {
      "epoch": 1.0361839217682896,
      "grad_norm": 0.06244376373378186,
      "learning_rate": 0.0001826962579072462,
      "loss": 0.8494,
      "step": 59020
    },
    {
      "epoch": 1.0363594866482908,
      "grad_norm": 0.06241958967076976,
      "learning_rate": 0.00018268996325275386,
      "loss": 0.8475,
      "step": 59030
    },
    {
      "epoch": 1.0365350515282923,
      "grad_norm": 0.05772991619075174,
      "learning_rate": 0.00018268366756323405,
      "loss": 0.844,
      "step": 59040
    },
    {
      "epoch": 1.0367106164082938,
      "grad_norm": 0.058457849006189,
      "learning_rate": 0.00018267737083876658,
      "loss": 0.8457,
      "step": 59050
    },
    {
      "epoch": 1.036886181288295,
      "grad_norm": 0.0979778695321663,
      "learning_rate": 0.0001826710730794311,
      "loss": 0.8418,
      "step": 59060
    },
    {
      "epoch": 1.0370617461682965,
      "grad_norm": 0.05744487451778929,
      "learning_rate": 0.0001826647742853075,
      "loss": 0.853,
      "step": 59070
    },
    {
      "epoch": 1.037237311048298,
      "grad_norm": 0.06819669534059113,
      "learning_rate": 0.00018265847445647556,
      "loss": 0.843,
      "step": 59080
    },
    {
      "epoch": 1.0374128759282992,
      "grad_norm": 0.07820539417656219,
      "learning_rate": 0.0001826521735930151,
      "loss": 0.8439,
      "step": 59090
    },
    {
      "epoch": 1.0375884408083007,
      "grad_norm": 0.06086369739861539,
      "learning_rate": 0.00018264587169500594,
      "loss": 0.8394,
      "step": 59100
    },
    {
      "epoch": 1.0377640056883022,
      "grad_norm": 0.064400296642715,
      "learning_rate": 0.00018263956876252794,
      "loss": 0.8506,
      "step": 59110
    },
    {
      "epoch": 1.0379395705683034,
      "grad_norm": 0.05313530874541342,
      "learning_rate": 0.00018263326479566095,
      "loss": 0.8572,
      "step": 59120
    },
    {
      "epoch": 1.038115135448305,
      "grad_norm": 0.05992661100311786,
      "learning_rate": 0.00018262695979448488,
      "loss": 0.8559,
      "step": 59130
    },
    {
      "epoch": 1.0382907003283064,
      "grad_norm": 0.08437269424706641,
      "learning_rate": 0.00018262065375907957,
      "loss": 0.8499,
      "step": 59140
    },
    {
      "epoch": 1.0384662652083076,
      "grad_norm": 0.05264483740787297,
      "learning_rate": 0.00018261434668952493,
      "loss": 0.8533,
      "step": 59150
    },
    {
      "epoch": 1.0386418300883091,
      "grad_norm": 0.05905083574324505,
      "learning_rate": 0.0001826080385859009,
      "loss": 0.8577,
      "step": 59160
    },
    {
      "epoch": 1.0388173949683106,
      "grad_norm": 0.054772455332478265,
      "learning_rate": 0.00018260172944828737,
      "loss": 0.8526,
      "step": 59170
    },
    {
      "epoch": 1.0389929598483119,
      "grad_norm": 0.06016118143089992,
      "learning_rate": 0.00018259541927676426,
      "loss": 0.8408,
      "step": 59180
    },
    {
      "epoch": 1.0391685247283133,
      "grad_norm": 0.0644477034965523,
      "learning_rate": 0.00018258910807141158,
      "loss": 0.8494,
      "step": 59190
    },
    {
      "epoch": 1.0393440896083148,
      "grad_norm": 0.06575079271226414,
      "learning_rate": 0.00018258279583230924,
      "loss": 0.8545,
      "step": 59200
    },
    {
      "epoch": 1.039519654488316,
      "grad_norm": 0.07524526199061304,
      "learning_rate": 0.00018257648255953725,
      "loss": 0.8505,
      "step": 59210
    },
    {
      "epoch": 1.0396952193683175,
      "grad_norm": 0.05251314010340045,
      "learning_rate": 0.00018257016825317562,
      "loss": 0.8511,
      "step": 59220
    },
    {
      "epoch": 1.039870784248319,
      "grad_norm": 0.06814930091776203,
      "learning_rate": 0.0001825638529133043,
      "loss": 0.8531,
      "step": 59230
    },
    {
      "epoch": 1.0400463491283203,
      "grad_norm": 0.06120418539745883,
      "learning_rate": 0.00018255753654000334,
      "loss": 0.8541,
      "step": 59240
    },
    {
      "epoch": 1.0402219140083218,
      "grad_norm": 0.06053758172941148,
      "learning_rate": 0.00018255121913335275,
      "loss": 0.8481,
      "step": 59250
    },
    {
      "epoch": 1.0403974788883232,
      "grad_norm": 0.10121029084845852,
      "learning_rate": 0.0001825449006934326,
      "loss": 0.8506,
      "step": 59260
    },
    {
      "epoch": 1.0405730437683245,
      "grad_norm": 0.05171912790477786,
      "learning_rate": 0.0001825385812203229,
      "loss": 0.8533,
      "step": 59270
    },
    {
      "epoch": 1.040748608648326,
      "grad_norm": 0.05097854868232172,
      "learning_rate": 0.0001825322607141038,
      "loss": 0.8496,
      "step": 59280
    },
    {
      "epoch": 1.0409241735283274,
      "grad_norm": 0.07669855659483728,
      "learning_rate": 0.0001825259391748553,
      "loss": 0.8543,
      "step": 59290
    },
    {
      "epoch": 1.0410997384083287,
      "grad_norm": 0.05323897996586778,
      "learning_rate": 0.00018251961660265752,
      "loss": 0.8519,
      "step": 59300
    },
    {
      "epoch": 1.0412753032883302,
      "grad_norm": 0.05443565017458132,
      "learning_rate": 0.00018251329299759058,
      "loss": 0.8507,
      "step": 59310
    },
    {
      "epoch": 1.0414508681683317,
      "grad_norm": 0.04839276308005289,
      "learning_rate": 0.00018250696835973458,
      "loss": 0.8435,
      "step": 59320
    },
    {
      "epoch": 1.041626433048333,
      "grad_norm": 0.05896201951421716,
      "learning_rate": 0.0001825006426891697,
      "loss": 0.851,
      "step": 59330
    },
    {
      "epoch": 1.0418019979283344,
      "grad_norm": 0.08124405091632159,
      "learning_rate": 0.00018249431598597605,
      "loss": 0.8568,
      "step": 59340
    },
    {
      "epoch": 1.0419775628083359,
      "grad_norm": 0.05242607236656098,
      "learning_rate": 0.0001824879882502338,
      "loss": 0.8483,
      "step": 59350
    },
    {
      "epoch": 1.0421531276883371,
      "grad_norm": 0.06871032736531749,
      "learning_rate": 0.00018248165948202313,
      "loss": 0.8507,
      "step": 59360
    },
    {
      "epoch": 1.0423286925683386,
      "grad_norm": 0.05385815706271565,
      "learning_rate": 0.00018247532968142417,
      "loss": 0.8528,
      "step": 59370
    },
    {
      "epoch": 1.04250425744834,
      "grad_norm": 0.05688091285725989,
      "learning_rate": 0.00018246899884851722,
      "loss": 0.8513,
      "step": 59380
    },
    {
      "epoch": 1.0426798223283413,
      "grad_norm": 0.049730424089249846,
      "learning_rate": 0.00018246266698338244,
      "loss": 0.8499,
      "step": 59390
    },
    {
      "epoch": 1.0428553872083428,
      "grad_norm": 0.055180171093295725,
      "learning_rate": 0.00018245633408610001,
      "loss": 0.851,
      "step": 59400
    },
    {
      "epoch": 1.0430309520883443,
      "grad_norm": 0.05913238731567488,
      "learning_rate": 0.00018245000015675025,
      "loss": 0.8534,
      "step": 59410
    },
    {
      "epoch": 1.0432065169683455,
      "grad_norm": 0.07159077583307875,
      "learning_rate": 0.00018244366519541336,
      "loss": 0.859,
      "step": 59420
    },
    {
      "epoch": 1.043382081848347,
      "grad_norm": 0.09137084928239794,
      "learning_rate": 0.00018243732920216965,
      "loss": 0.8372,
      "step": 59430
    },
    {
      "epoch": 1.0435576467283485,
      "grad_norm": 0.06852631703451934,
      "learning_rate": 0.00018243099217709935,
      "loss": 0.8494,
      "step": 59440
    },
    {
      "epoch": 1.04373321160835,
      "grad_norm": 0.05496300132348518,
      "learning_rate": 0.00018242465412028283,
      "loss": 0.8422,
      "step": 59450
    },
    {
      "epoch": 1.0439087764883512,
      "grad_norm": 0.05102458539377945,
      "learning_rate": 0.00018241831503180026,
      "loss": 0.8447,
      "step": 59460
    },
    {
      "epoch": 1.0440843413683527,
      "grad_norm": 0.07852256726401179,
      "learning_rate": 0.00018241197491173207,
      "loss": 0.8623,
      "step": 59470
    },
    {
      "epoch": 1.0442599062483542,
      "grad_norm": 0.04706346516004175,
      "learning_rate": 0.00018240563376015854,
      "loss": 0.8592,
      "step": 59480
    },
    {
      "epoch": 1.0444354711283554,
      "grad_norm": 0.061106763728070716,
      "learning_rate": 0.00018239929157716004,
      "loss": 0.8543,
      "step": 59490
    },
    {
      "epoch": 1.044611036008357,
      "grad_norm": 0.06823834680905128,
      "learning_rate": 0.00018239294836281693,
      "loss": 0.8451,
      "step": 59500
    },
    {
      "epoch": 1.0447866008883584,
      "grad_norm": 0.06370364436502943,
      "learning_rate": 0.00018238660411720953,
      "loss": 0.8547,
      "step": 59510
    },
    {
      "epoch": 1.0449621657683597,
      "grad_norm": 0.05359411386460144,
      "learning_rate": 0.00018238025884041828,
      "loss": 0.8539,
      "step": 59520
    },
    {
      "epoch": 1.0451377306483611,
      "grad_norm": 0.055879442258361625,
      "learning_rate": 0.00018237391253252355,
      "loss": 0.8493,
      "step": 59530
    },
    {
      "epoch": 1.0453132955283626,
      "grad_norm": 0.0789864512313926,
      "learning_rate": 0.00018236756519360577,
      "loss": 0.8499,
      "step": 59540
    },
    {
      "epoch": 1.0454888604083639,
      "grad_norm": 0.05817144788801788,
      "learning_rate": 0.00018236121682374532,
      "loss": 0.8494,
      "step": 59550
    },
    {
      "epoch": 1.0456644252883653,
      "grad_norm": 0.05033704422719863,
      "learning_rate": 0.00018235486742302267,
      "loss": 0.8497,
      "step": 59560
    },
    {
      "epoch": 1.0458399901683668,
      "grad_norm": 0.05537994639202765,
      "learning_rate": 0.00018234851699151824,
      "loss": 0.8533,
      "step": 59570
    },
    {
      "epoch": 1.046015555048368,
      "grad_norm": 0.057918234934014996,
      "learning_rate": 0.00018234216552931253,
      "loss": 0.8598,
      "step": 59580
    },
    {
      "epoch": 1.0461911199283696,
      "grad_norm": 0.07136444180872714,
      "learning_rate": 0.000182335813036486,
      "loss": 0.8551,
      "step": 59590
    },
    {
      "epoch": 1.046366684808371,
      "grad_norm": 0.05747745947659825,
      "learning_rate": 0.00018232945951311916,
      "loss": 0.8538,
      "step": 59600
    },
    {
      "epoch": 1.0465422496883723,
      "grad_norm": 0.049225298744631656,
      "learning_rate": 0.00018232310495929242,
      "loss": 0.8453,
      "step": 59610
    },
    {
      "epoch": 1.0467178145683738,
      "grad_norm": 0.05530997588518064,
      "learning_rate": 0.0001823167493750864,
      "loss": 0.8587,
      "step": 59620
    },
    {
      "epoch": 1.0468933794483752,
      "grad_norm": 0.07217187317462502,
      "learning_rate": 0.00018231039276058155,
      "loss": 0.8491,
      "step": 59630
    },
    {
      "epoch": 1.0470689443283765,
      "grad_norm": 0.0625253773556848,
      "learning_rate": 0.00018230403511585846,
      "loss": 0.8516,
      "step": 59640
    },
    {
      "epoch": 1.047244509208378,
      "grad_norm": 0.0822470694815677,
      "learning_rate": 0.00018229767644099764,
      "loss": 0.8528,
      "step": 59650
    },
    {
      "epoch": 1.0474200740883794,
      "grad_norm": 0.06510461067750206,
      "learning_rate": 0.0001822913167360797,
      "loss": 0.8466,
      "step": 59660
    },
    {
      "epoch": 1.0475956389683807,
      "grad_norm": 0.08985684172088924,
      "learning_rate": 0.00018228495600118518,
      "loss": 0.8589,
      "step": 59670
    },
    {
      "epoch": 1.0477712038483822,
      "grad_norm": 0.04783142537222484,
      "learning_rate": 0.0001822785942363947,
      "loss": 0.8591,
      "step": 59680
    },
    {
      "epoch": 1.0479467687283837,
      "grad_norm": 0.05270053707749643,
      "learning_rate": 0.00018227223144178886,
      "loss": 0.8612,
      "step": 59690
    },
    {
      "epoch": 1.048122333608385,
      "grad_norm": 0.05457911472318335,
      "learning_rate": 0.00018226586761744827,
      "loss": 0.8669,
      "step": 59700
    },
    {
      "epoch": 1.0482978984883864,
      "grad_norm": 0.05690696085843239,
      "learning_rate": 0.00018225950276345353,
      "loss": 0.8484,
      "step": 59710
    },
    {
      "epoch": 1.0484734633683879,
      "grad_norm": 0.0636856163696016,
      "learning_rate": 0.00018225313687988536,
      "loss": 0.844,
      "step": 59720
    },
    {
      "epoch": 1.0486490282483891,
      "grad_norm": 0.04963609633369366,
      "learning_rate": 0.00018224676996682435,
      "loss": 0.848,
      "step": 59730
    },
    {
      "epoch": 1.0488245931283906,
      "grad_norm": 0.07716499299858307,
      "learning_rate": 0.00018224040202435122,
      "loss": 0.846,
      "step": 59740
    },
    {
      "epoch": 1.049000158008392,
      "grad_norm": 0.057951502794932344,
      "learning_rate": 0.0001822340330525466,
      "loss": 0.8455,
      "step": 59750
    },
    {
      "epoch": 1.0491757228883933,
      "grad_norm": 0.05354112765900599,
      "learning_rate": 0.0001822276630514912,
      "loss": 0.855,
      "step": 59760
    },
    {
      "epoch": 1.0493512877683948,
      "grad_norm": 0.07293368382363283,
      "learning_rate": 0.0001822212920212658,
      "loss": 0.8482,
      "step": 59770
    },
    {
      "epoch": 1.0495268526483963,
      "grad_norm": 0.05882154391330544,
      "learning_rate": 0.000182214919961951,
      "loss": 0.856,
      "step": 59780
    },
    {
      "epoch": 1.0497024175283975,
      "grad_norm": 0.08228140145249668,
      "learning_rate": 0.00018220854687362763,
      "loss": 0.8603,
      "step": 59790
    },
    {
      "epoch": 1.049877982408399,
      "grad_norm": 0.05805524415204143,
      "learning_rate": 0.0001822021727563764,
      "loss": 0.8554,
      "step": 59800
    },
    {
      "epoch": 1.0500535472884005,
      "grad_norm": 0.06778988321763342,
      "learning_rate": 0.0001821957976102781,
      "loss": 0.8506,
      "step": 59810
    },
    {
      "epoch": 1.0502291121684018,
      "grad_norm": 0.05274368356212095,
      "learning_rate": 0.00018218942143541348,
      "loss": 0.8499,
      "step": 59820
    },
    {
      "epoch": 1.0504046770484032,
      "grad_norm": 0.08661603405370949,
      "learning_rate": 0.0001821830442318633,
      "loss": 0.8447,
      "step": 59830
    },
    {
      "epoch": 1.0505802419284047,
      "grad_norm": 0.06701909689130763,
      "learning_rate": 0.00018217666599970838,
      "loss": 0.8465,
      "step": 59840
    },
    {
      "epoch": 1.050755806808406,
      "grad_norm": 0.05921063450622248,
      "learning_rate": 0.0001821702867390296,
      "loss": 0.8531,
      "step": 59850
    },
    {
      "epoch": 1.0509313716884074,
      "grad_norm": 0.07232025496837932,
      "learning_rate": 0.00018216390644990767,
      "loss": 0.8553,
      "step": 59860
    },
    {
      "epoch": 1.051106936568409,
      "grad_norm": 0.07656223442616333,
      "learning_rate": 0.00018215752513242353,
      "loss": 0.8464,
      "step": 59870
    },
    {
      "epoch": 1.0512825014484102,
      "grad_norm": 0.05985959783039181,
      "learning_rate": 0.000182151142786658,
      "loss": 0.851,
      "step": 59880
    },
    {
      "epoch": 1.0514580663284117,
      "grad_norm": 0.08479674042037046,
      "learning_rate": 0.00018214475941269193,
      "loss": 0.8548,
      "step": 59890
    },
    {
      "epoch": 1.0516336312084131,
      "grad_norm": 0.0679422083515928,
      "learning_rate": 0.0001821383750106062,
      "loss": 0.8546,
      "step": 59900
    },
    {
      "epoch": 1.0518091960884144,
      "grad_norm": 0.07177584061917873,
      "learning_rate": 0.00018213198958048172,
      "loss": 0.8521,
      "step": 59910
    },
    {
      "epoch": 1.0519847609684159,
      "grad_norm": 0.05293992787714145,
      "learning_rate": 0.00018212560312239938,
      "loss": 0.8472,
      "step": 59920
    },
    {
      "epoch": 1.0521603258484173,
      "grad_norm": 0.06802688835810954,
      "learning_rate": 0.0001821192156364401,
      "loss": 0.8476,
      "step": 59930
    },
    {
      "epoch": 1.0523358907284186,
      "grad_norm": 0.07592673890468883,
      "learning_rate": 0.0001821128271226848,
      "loss": 0.8353,
      "step": 59940
    },
    {
      "epoch": 1.05251145560842,
      "grad_norm": 0.07781454336625276,
      "learning_rate": 0.00018210643758121442,
      "loss": 0.8432,
      "step": 59950
    },
    {
      "epoch": 1.0526870204884216,
      "grad_norm": 0.07209812090622028,
      "learning_rate": 0.00018210004701210995,
      "loss": 0.8534,
      "step": 59960
    },
    {
      "epoch": 1.0528625853684228,
      "grad_norm": 0.05027436975609625,
      "learning_rate": 0.00018209365541545233,
      "loss": 0.8492,
      "step": 59970
    },
    {
      "epoch": 1.0530381502484243,
      "grad_norm": 0.07153004868270849,
      "learning_rate": 0.00018208726279132258,
      "loss": 0.8548,
      "step": 59980
    },
    {
      "epoch": 1.0532137151284258,
      "grad_norm": 0.07598934099118468,
      "learning_rate": 0.00018208086913980168,
      "loss": 0.8508,
      "step": 59990
    },
    {
      "epoch": 1.053389280008427,
      "grad_norm": 0.050766138187549284,
      "learning_rate": 0.0001820744744609706,
      "loss": 0.8551,
      "step": 60000
    },
    {
      "epoch": 1.0535648448884285,
      "grad_norm": 0.05757190959896189,
      "learning_rate": 0.00018206807875491038,
      "loss": 0.8466,
      "step": 60010
    },
    {
      "epoch": 1.05374040976843,
      "grad_norm": 0.05908967816536562,
      "learning_rate": 0.0001820616820217021,
      "loss": 0.8529,
      "step": 60020
    },
    {
      "epoch": 1.0539159746484312,
      "grad_norm": 0.07039605655883681,
      "learning_rate": 0.00018205528426142674,
      "loss": 0.848,
      "step": 60030
    },
    {
      "epoch": 1.0540915395284327,
      "grad_norm": 0.0678504296544689,
      "learning_rate": 0.0001820488854741654,
      "loss": 0.8492,
      "step": 60040
    },
    {
      "epoch": 1.0542671044084342,
      "grad_norm": 0.055385797161915364,
      "learning_rate": 0.00018204248565999919,
      "loss": 0.8565,
      "step": 60050
    },
    {
      "epoch": 1.0544426692884354,
      "grad_norm": 0.06542702497832366,
      "learning_rate": 0.00018203608481900913,
      "loss": 0.845,
      "step": 60060
    },
    {
      "epoch": 1.054618234168437,
      "grad_norm": 0.05484369393478439,
      "learning_rate": 0.00018202968295127634,
      "loss": 0.8512,
      "step": 60070
    },
    {
      "epoch": 1.0547937990484384,
      "grad_norm": 0.07042963195526668,
      "learning_rate": 0.00018202328005688192,
      "loss": 0.8489,
      "step": 60080
    },
    {
      "epoch": 1.0549693639284397,
      "grad_norm": 0.07818137728547689,
      "learning_rate": 0.00018201687613590703,
      "loss": 0.8523,
      "step": 60090
    },
    {
      "epoch": 1.0551449288084411,
      "grad_norm": 0.04587432554177762,
      "learning_rate": 0.0001820104711884328,
      "loss": 0.8499,
      "step": 60100
    },
    {
      "epoch": 1.0553204936884426,
      "grad_norm": 0.08405606948938073,
      "learning_rate": 0.00018200406521454035,
      "loss": 0.8522,
      "step": 60110
    },
    {
      "epoch": 1.0554960585684439,
      "grad_norm": 0.06246531174915386,
      "learning_rate": 0.0001819976582143109,
      "loss": 0.8441,
      "step": 60120
    },
    {
      "epoch": 1.0556716234484453,
      "grad_norm": 0.08439884181635326,
      "learning_rate": 0.0001819912501878256,
      "loss": 0.8542,
      "step": 60130
    },
    {
      "epoch": 1.0558471883284468,
      "grad_norm": 0.055823392204144014,
      "learning_rate": 0.0001819848411351656,
      "loss": 0.8539,
      "step": 60140
    },
    {
      "epoch": 1.056022753208448,
      "grad_norm": 0.06475148965465664,
      "learning_rate": 0.00018197843105641217,
      "loss": 0.8605,
      "step": 60150
    },
    {
      "epoch": 1.0561983180884496,
      "grad_norm": 0.06015893172930975,
      "learning_rate": 0.0001819720199516465,
      "loss": 0.8425,
      "step": 60160
    },
    {
      "epoch": 1.056373882968451,
      "grad_norm": 0.06495310219567273,
      "learning_rate": 0.0001819656078209498,
      "loss": 0.848,
      "step": 60170
    },
    {
      "epoch": 1.0565494478484525,
      "grad_norm": 0.097599280458904,
      "learning_rate": 0.00018195919466440334,
      "loss": 0.8518,
      "step": 60180
    },
    {
      "epoch": 1.0567250127284538,
      "grad_norm": 0.049339583596514365,
      "learning_rate": 0.00018195278048208836,
      "loss": 0.8452,
      "step": 60190
    },
    {
      "epoch": 1.0569005776084552,
      "grad_norm": 0.07490693062339321,
      "learning_rate": 0.00018194636527408617,
      "loss": 0.8441,
      "step": 60200
    },
    {
      "epoch": 1.0570761424884567,
      "grad_norm": 0.05389972674789821,
      "learning_rate": 0.000181939949040478,
      "loss": 0.8593,
      "step": 60210
    },
    {
      "epoch": 1.057251707368458,
      "grad_norm": 0.06929403383954734,
      "learning_rate": 0.00018193353178134515,
      "loss": 0.855,
      "step": 60220
    },
    {
      "epoch": 1.0574272722484594,
      "grad_norm": 0.052455375451491634,
      "learning_rate": 0.00018192711349676897,
      "loss": 0.848,
      "step": 60230
    },
    {
      "epoch": 1.057602837128461,
      "grad_norm": 0.07986870692881229,
      "learning_rate": 0.00018192069418683073,
      "loss": 0.8558,
      "step": 60240
    },
    {
      "epoch": 1.0577784020084622,
      "grad_norm": 0.058719466134070884,
      "learning_rate": 0.0001819142738516118,
      "loss": 0.8537,
      "step": 60250
    },
    {
      "epoch": 1.0579539668884637,
      "grad_norm": 0.07400472975933123,
      "learning_rate": 0.0001819078524911935,
      "loss": 0.8457,
      "step": 60260
    },
    {
      "epoch": 1.0581295317684651,
      "grad_norm": 0.06306795058200172,
      "learning_rate": 0.00018190143010565722,
      "loss": 0.8485,
      "step": 60270
    },
    {
      "epoch": 1.0583050966484664,
      "grad_norm": 0.06687926133851442,
      "learning_rate": 0.00018189500669508427,
      "loss": 0.8581,
      "step": 60280
    },
    {
      "epoch": 1.0584806615284679,
      "grad_norm": 0.05938666668551222,
      "learning_rate": 0.00018188858225955614,
      "loss": 0.8552,
      "step": 60290
    },
    {
      "epoch": 1.0586562264084693,
      "grad_norm": 0.08407928570266666,
      "learning_rate": 0.00018188215679915414,
      "loss": 0.8562,
      "step": 60300
    },
    {
      "epoch": 1.0588317912884706,
      "grad_norm": 0.06378558947347295,
      "learning_rate": 0.00018187573031395971,
      "loss": 0.8515,
      "step": 60310
    },
    {
      "epoch": 1.059007356168472,
      "grad_norm": 0.05141027868010165,
      "learning_rate": 0.0001818693028040543,
      "loss": 0.8452,
      "step": 60320
    },
    {
      "epoch": 1.0591829210484736,
      "grad_norm": 0.06030524096928478,
      "learning_rate": 0.0001818628742695193,
      "loss": 0.8482,
      "step": 60330
    },
    {
      "epoch": 1.0593584859284748,
      "grad_norm": 0.06231427601159012,
      "learning_rate": 0.00018185644471043622,
      "loss": 0.8456,
      "step": 60340
    },
    {
      "epoch": 1.0595340508084763,
      "grad_norm": 0.05836957531744594,
      "learning_rate": 0.00018185001412688643,
      "loss": 0.8509,
      "step": 60350
    },
    {
      "epoch": 1.0597096156884778,
      "grad_norm": 0.055612231426766745,
      "learning_rate": 0.0001818435825189515,
      "loss": 0.8438,
      "step": 60360
    },
    {
      "epoch": 1.059885180568479,
      "grad_norm": 0.05880828060142469,
      "learning_rate": 0.0001818371498867129,
      "loss": 0.8482,
      "step": 60370
    },
    {
      "epoch": 1.0600607454484805,
      "grad_norm": 0.06617398352244167,
      "learning_rate": 0.0001818307162302521,
      "loss": 0.8544,
      "step": 60380
    },
    {
      "epoch": 1.060236310328482,
      "grad_norm": 0.07334782301858093,
      "learning_rate": 0.00018182428154965064,
      "loss": 0.8516,
      "step": 60390
    },
    {
      "epoch": 1.0604118752084832,
      "grad_norm": 0.06203087151513314,
      "learning_rate": 0.00018181784584499005,
      "loss": 0.8482,
      "step": 60400
    },
    {
      "epoch": 1.0605874400884847,
      "grad_norm": 0.07829769669299173,
      "learning_rate": 0.00018181140911635186,
      "loss": 0.8463,
      "step": 60410
    },
    {
      "epoch": 1.0607630049684862,
      "grad_norm": 0.060207453646319915,
      "learning_rate": 0.00018180497136381765,
      "loss": 0.8572,
      "step": 60420
    },
    {
      "epoch": 1.0609385698484874,
      "grad_norm": 0.045306627153866415,
      "learning_rate": 0.00018179853258746891,
      "loss": 0.8591,
      "step": 60430
    },
    {
      "epoch": 1.061114134728489,
      "grad_norm": 0.05306353879545023,
      "learning_rate": 0.00018179209278738732,
      "loss": 0.8462,
      "step": 60440
    },
    {
      "epoch": 1.0612896996084904,
      "grad_norm": 0.12197797229015668,
      "learning_rate": 0.00018178565196365443,
      "loss": 0.8494,
      "step": 60450
    },
    {
      "epoch": 1.0614652644884917,
      "grad_norm": 0.0914520840327145,
      "learning_rate": 0.00018177921011635183,
      "loss": 0.8502,
      "step": 60460
    },
    {
      "epoch": 1.0616408293684931,
      "grad_norm": 0.1084187280511267,
      "learning_rate": 0.00018177276724556116,
      "loss": 0.8493,
      "step": 60470
    },
    {
      "epoch": 1.0618163942484946,
      "grad_norm": 0.06660244043940805,
      "learning_rate": 0.00018176632335136406,
      "loss": 0.8535,
      "step": 60480
    },
    {
      "epoch": 1.0619919591284959,
      "grad_norm": 0.0662624074786878,
      "learning_rate": 0.00018175987843384217,
      "loss": 0.8416,
      "step": 60490
    },
    {
      "epoch": 1.0621675240084973,
      "grad_norm": 0.04919604887618097,
      "learning_rate": 0.00018175343249307712,
      "loss": 0.84,
      "step": 60500
    },
    {
      "epoch": 1.0623430888884988,
      "grad_norm": 0.06353012179673118,
      "learning_rate": 0.0001817469855291506,
      "loss": 0.8433,
      "step": 60510
    },
    {
      "epoch": 1.0625186537685,
      "grad_norm": 0.05388053053894124,
      "learning_rate": 0.00018174053754214432,
      "loss": 0.8544,
      "step": 60520
    },
    {
      "epoch": 1.0626942186485016,
      "grad_norm": 0.062444667774841034,
      "learning_rate": 0.00018173408853213994,
      "loss": 0.8441,
      "step": 60530
    },
    {
      "epoch": 1.062869783528503,
      "grad_norm": 0.0693002272237464,
      "learning_rate": 0.0001817276384992192,
      "loss": 0.8509,
      "step": 60540
    },
    {
      "epoch": 1.0630453484085043,
      "grad_norm": 0.052978278271207116,
      "learning_rate": 0.00018172118744346383,
      "loss": 0.8547,
      "step": 60550
    },
    {
      "epoch": 1.0632209132885058,
      "grad_norm": 0.06742162159863273,
      "learning_rate": 0.0001817147353649555,
      "loss": 0.8525,
      "step": 60560
    },
    {
      "epoch": 1.0633964781685072,
      "grad_norm": 0.05796068900843066,
      "learning_rate": 0.000181708282263776,
      "loss": 0.8478,
      "step": 60570
    },
    {
      "epoch": 1.0635720430485085,
      "grad_norm": 0.05027505077939828,
      "learning_rate": 0.00018170182814000713,
      "loss": 0.8542,
      "step": 60580
    },
    {
      "epoch": 1.06374760792851,
      "grad_norm": 0.05247471547047542,
      "learning_rate": 0.00018169537299373063,
      "loss": 0.8594,
      "step": 60590
    },
    {
      "epoch": 1.0639231728085115,
      "grad_norm": 0.053126297978310005,
      "learning_rate": 0.00018168891682502828,
      "loss": 0.8502,
      "step": 60600
    },
    {
      "epoch": 1.0640987376885127,
      "grad_norm": 0.07271335456704277,
      "learning_rate": 0.0001816824596339819,
      "loss": 0.8448,
      "step": 60610
    },
    {
      "epoch": 1.0642743025685142,
      "grad_norm": 0.07470848397232722,
      "learning_rate": 0.00018167600142067326,
      "loss": 0.8598,
      "step": 60620
    },
    {
      "epoch": 1.0644498674485157,
      "grad_norm": 0.07024577687439819,
      "learning_rate": 0.00018166954218518423,
      "loss": 0.8556,
      "step": 60630
    },
    {
      "epoch": 1.064625432328517,
      "grad_norm": 0.05031817691434465,
      "learning_rate": 0.00018166308192759666,
      "loss": 0.8551,
      "step": 60640
    },
    {
      "epoch": 1.0648009972085184,
      "grad_norm": 0.06874554214369889,
      "learning_rate": 0.00018165662064799237,
      "loss": 0.8569,
      "step": 60650
    },
    {
      "epoch": 1.0649765620885199,
      "grad_norm": 0.07168306585984462,
      "learning_rate": 0.00018165015834645327,
      "loss": 0.8527,
      "step": 60660
    },
    {
      "epoch": 1.0651521269685211,
      "grad_norm": 0.0508421155000683,
      "learning_rate": 0.00018164369502306115,
      "loss": 0.8468,
      "step": 60670
    },
    {
      "epoch": 1.0653276918485226,
      "grad_norm": 0.06819114306500294,
      "learning_rate": 0.000181637230677898,
      "loss": 0.8534,
      "step": 60680
    },
    {
      "epoch": 1.065503256728524,
      "grad_norm": 0.07608675414719157,
      "learning_rate": 0.0001816307653110457,
      "loss": 0.8506,
      "step": 60690
    },
    {
      "epoch": 1.0656788216085253,
      "grad_norm": 0.057272379684035715,
      "learning_rate": 0.0001816242989225861,
      "loss": 0.8477,
      "step": 60700
    },
    {
      "epoch": 1.0658543864885268,
      "grad_norm": 0.05149030788212542,
      "learning_rate": 0.00018161783151260123,
      "loss": 0.8423,
      "step": 60710
    },
    {
      "epoch": 1.0660299513685283,
      "grad_norm": 0.09792147442266744,
      "learning_rate": 0.00018161136308117294,
      "loss": 0.8464,
      "step": 60720
    },
    {
      "epoch": 1.0662055162485296,
      "grad_norm": 0.07470554623841476,
      "learning_rate": 0.00018160489362838324,
      "loss": 0.8494,
      "step": 60730
    },
    {
      "epoch": 1.066381081128531,
      "grad_norm": 0.0908576937628512,
      "learning_rate": 0.0001815984231543141,
      "loss": 0.8558,
      "step": 60740
    },
    {
      "epoch": 1.0665566460085325,
      "grad_norm": 0.08325156259659625,
      "learning_rate": 0.0001815919516590475,
      "loss": 0.8493,
      "step": 60750
    },
    {
      "epoch": 1.0667322108885338,
      "grad_norm": 0.05563149390691716,
      "learning_rate": 0.00018158547914266543,
      "loss": 0.8546,
      "step": 60760
    },
    {
      "epoch": 1.0669077757685352,
      "grad_norm": 0.06460796323081544,
      "learning_rate": 0.00018157900560524987,
      "loss": 0.8534,
      "step": 60770
    },
    {
      "epoch": 1.0670833406485367,
      "grad_norm": 0.06949839961214194,
      "learning_rate": 0.0001815725310468829,
      "loss": 0.8516,
      "step": 60780
    },
    {
      "epoch": 1.067258905528538,
      "grad_norm": 0.06850697540961595,
      "learning_rate": 0.00018156605546764648,
      "loss": 0.8412,
      "step": 60790
    },
    {
      "epoch": 1.0674344704085394,
      "grad_norm": 0.05781111048324669,
      "learning_rate": 0.00018155957886762273,
      "loss": 0.8491,
      "step": 60800
    },
    {
      "epoch": 1.067610035288541,
      "grad_norm": 0.04996215523687568,
      "learning_rate": 0.00018155310124689367,
      "loss": 0.8466,
      "step": 60810
    },
    {
      "epoch": 1.0677856001685422,
      "grad_norm": 0.05180375392416595,
      "learning_rate": 0.00018154662260554135,
      "loss": 0.8561,
      "step": 60820
    },
    {
      "epoch": 1.0679611650485437,
      "grad_norm": 0.095219420188232,
      "learning_rate": 0.0001815401429436479,
      "loss": 0.8498,
      "step": 60830
    },
    {
      "epoch": 1.0681367299285451,
      "grad_norm": 0.056150316628440125,
      "learning_rate": 0.00018153366226129544,
      "loss": 0.8512,
      "step": 60840
    },
    {
      "epoch": 1.0683122948085466,
      "grad_norm": 0.05139619700470689,
      "learning_rate": 0.00018152718055856602,
      "loss": 0.8548,
      "step": 60850
    },
    {
      "epoch": 1.0684878596885479,
      "grad_norm": 0.07333992728204673,
      "learning_rate": 0.00018152069783554178,
      "loss": 0.8526,
      "step": 60860
    },
    {
      "epoch": 1.0686634245685493,
      "grad_norm": 0.059614723417672585,
      "learning_rate": 0.00018151421409230486,
      "loss": 0.8619,
      "step": 60870
    },
    {
      "epoch": 1.0688389894485506,
      "grad_norm": 0.07012723867034898,
      "learning_rate": 0.00018150772932893744,
      "loss": 0.85,
      "step": 60880
    },
    {
      "epoch": 1.069014554328552,
      "grad_norm": 0.05975806011091313,
      "learning_rate": 0.00018150124354552162,
      "loss": 0.8515,
      "step": 60890
    },
    {
      "epoch": 1.0691901192085536,
      "grad_norm": 0.08850164659756968,
      "learning_rate": 0.00018149475674213966,
      "loss": 0.8488,
      "step": 60900
    },
    {
      "epoch": 1.069365684088555,
      "grad_norm": 0.08141962949462642,
      "learning_rate": 0.00018148826891887368,
      "loss": 0.8442,
      "step": 60910
    },
    {
      "epoch": 1.0695412489685563,
      "grad_norm": 0.09051061300875776,
      "learning_rate": 0.00018148178007580595,
      "loss": 0.8517,
      "step": 60920
    },
    {
      "epoch": 1.0697168138485578,
      "grad_norm": 0.06180742592581656,
      "learning_rate": 0.0001814752902130186,
      "loss": 0.8494,
      "step": 60930
    },
    {
      "epoch": 1.069892378728559,
      "grad_norm": 0.06460404353525807,
      "learning_rate": 0.00018146879933059388,
      "loss": 0.8498,
      "step": 60940
    },
    {
      "epoch": 1.0700679436085605,
      "grad_norm": 0.06476912171484074,
      "learning_rate": 0.00018146230742861413,
      "loss": 0.8542,
      "step": 60950
    },
    {
      "epoch": 1.070243508488562,
      "grad_norm": 0.05589654670773275,
      "learning_rate": 0.00018145581450716147,
      "loss": 0.861,
      "step": 60960
    },
    {
      "epoch": 1.0704190733685635,
      "grad_norm": 0.04819824577539295,
      "learning_rate": 0.0001814493205663182,
      "loss": 0.8538,
      "step": 60970
    },
    {
      "epoch": 1.0705946382485647,
      "grad_norm": 0.056856645006226385,
      "learning_rate": 0.00018144282560616666,
      "loss": 0.8546,
      "step": 60980
    },
    {
      "epoch": 1.0707702031285662,
      "grad_norm": 0.06434151242780092,
      "learning_rate": 0.0001814363296267891,
      "loss": 0.8511,
      "step": 60990
    },
    {
      "epoch": 1.0709457680085677,
      "grad_norm": 0.04564298270777429,
      "learning_rate": 0.0001814298326282678,
      "loss": 0.8543,
      "step": 61000
    },
    {
      "epoch": 1.071121332888569,
      "grad_norm": 0.08171134886969873,
      "learning_rate": 0.00018142333461068514,
      "loss": 0.8445,
      "step": 61010
    },
    {
      "epoch": 1.0712968977685704,
      "grad_norm": 0.07590989300190656,
      "learning_rate": 0.0001814168355741234,
      "loss": 0.8472,
      "step": 61020
    },
    {
      "epoch": 1.0714724626485719,
      "grad_norm": 0.08978821174043707,
      "learning_rate": 0.00018141033551866492,
      "loss": 0.8495,
      "step": 61030
    },
    {
      "epoch": 1.0716480275285731,
      "grad_norm": 0.05370060825402343,
      "learning_rate": 0.00018140383444439212,
      "loss": 0.8457,
      "step": 61040
    },
    {
      "epoch": 1.0718235924085746,
      "grad_norm": 0.055875171107074524,
      "learning_rate": 0.0001813973323513873,
      "loss": 0.858,
      "step": 61050
    },
    {
      "epoch": 1.071999157288576,
      "grad_norm": 0.06540885319956086,
      "learning_rate": 0.00018139082923973287,
      "loss": 0.8505,
      "step": 61060
    },
    {
      "epoch": 1.0721747221685773,
      "grad_norm": 0.07289620272540193,
      "learning_rate": 0.00018138432510951126,
      "loss": 0.8547,
      "step": 61070
    },
    {
      "epoch": 1.0723502870485788,
      "grad_norm": 0.13317645156450827,
      "learning_rate": 0.0001813778199608048,
      "loss": 0.842,
      "step": 61080
    },
    {
      "epoch": 1.0725258519285803,
      "grad_norm": 0.06865671313322506,
      "learning_rate": 0.000181371313793696,
      "loss": 0.8493,
      "step": 61090
    },
    {
      "epoch": 1.0727014168085816,
      "grad_norm": 0.06159295352173081,
      "learning_rate": 0.00018136480660826723,
      "loss": 0.8513,
      "step": 61100
    },
    {
      "epoch": 1.072876981688583,
      "grad_norm": 0.07209979180890977,
      "learning_rate": 0.00018135829840460095,
      "loss": 0.851,
      "step": 61110
    },
    {
      "epoch": 1.0730525465685845,
      "grad_norm": 0.06926485955596326,
      "learning_rate": 0.00018135178918277962,
      "loss": 0.8431,
      "step": 61120
    },
    {
      "epoch": 1.0732281114485858,
      "grad_norm": 0.074794521755562,
      "learning_rate": 0.00018134527894288576,
      "loss": 0.8469,
      "step": 61130
    },
    {
      "epoch": 1.0734036763285872,
      "grad_norm": 0.07153151045719304,
      "learning_rate": 0.0001813387676850018,
      "loss": 0.8484,
      "step": 61140
    },
    {
      "epoch": 1.0735792412085887,
      "grad_norm": 0.05743113435941426,
      "learning_rate": 0.00018133225540921025,
      "loss": 0.8489,
      "step": 61150
    },
    {
      "epoch": 1.07375480608859,
      "grad_norm": 0.047601002135514825,
      "learning_rate": 0.00018132574211559362,
      "loss": 0.8568,
      "step": 61160
    },
    {
      "epoch": 1.0739303709685915,
      "grad_norm": 0.05254497100347585,
      "learning_rate": 0.00018131922780423448,
      "loss": 0.8528,
      "step": 61170
    },
    {
      "epoch": 1.074105935848593,
      "grad_norm": 0.05298067376086963,
      "learning_rate": 0.0001813127124752153,
      "loss": 0.8438,
      "step": 61180
    },
    {
      "epoch": 1.0742815007285942,
      "grad_norm": 0.08144990072131607,
      "learning_rate": 0.00018130619612861866,
      "loss": 0.8447,
      "step": 61190
    },
    {
      "epoch": 1.0744570656085957,
      "grad_norm": 0.06164539233023013,
      "learning_rate": 0.00018129967876452711,
      "loss": 0.8487,
      "step": 61200
    },
    {
      "epoch": 1.0746326304885971,
      "grad_norm": 0.06384109247229344,
      "learning_rate": 0.00018129316038302328,
      "loss": 0.8485,
      "step": 61210
    },
    {
      "epoch": 1.0748081953685984,
      "grad_norm": 0.08737237174098515,
      "learning_rate": 0.00018128664098418967,
      "loss": 0.8518,
      "step": 61220
    },
    {
      "epoch": 1.0749837602485999,
      "grad_norm": 0.08074756456707065,
      "learning_rate": 0.000181280120568109,
      "loss": 0.8553,
      "step": 61230
    },
    {
      "epoch": 1.0751593251286014,
      "grad_norm": 0.05086431173438419,
      "learning_rate": 0.00018127359913486376,
      "loss": 0.856,
      "step": 61240
    },
    {
      "epoch": 1.0753348900086026,
      "grad_norm": 0.051295258405064326,
      "learning_rate": 0.00018126707668453666,
      "loss": 0.8477,
      "step": 61250
    },
    {
      "epoch": 1.075510454888604,
      "grad_norm": 0.07422356674181094,
      "learning_rate": 0.00018126055321721033,
      "loss": 0.8506,
      "step": 61260
    },
    {
      "epoch": 1.0756860197686056,
      "grad_norm": 0.05560415885161344,
      "learning_rate": 0.00018125402873296736,
      "loss": 0.847,
      "step": 61270
    },
    {
      "epoch": 1.0758615846486068,
      "grad_norm": 0.17953808678961924,
      "learning_rate": 0.00018124750323189053,
      "loss": 0.851,
      "step": 61280
    },
    {
      "epoch": 1.0760371495286083,
      "grad_norm": 0.07414710310545625,
      "learning_rate": 0.0001812409767140624,
      "loss": 0.8538,
      "step": 61290
    },
    {
      "epoch": 1.0762127144086098,
      "grad_norm": 0.04771236858301296,
      "learning_rate": 0.00018123444917956573,
      "loss": 0.8457,
      "step": 61300
    },
    {
      "epoch": 1.076388279288611,
      "grad_norm": 0.0660846303929062,
      "learning_rate": 0.00018122792062848323,
      "loss": 0.8463,
      "step": 61310
    },
    {
      "epoch": 1.0765638441686125,
      "grad_norm": 0.0657687237376536,
      "learning_rate": 0.00018122139106089758,
      "loss": 0.8491,
      "step": 61320
    },
    {
      "epoch": 1.076739409048614,
      "grad_norm": 0.07323678821287113,
      "learning_rate": 0.00018121486047689153,
      "loss": 0.8571,
      "step": 61330
    },
    {
      "epoch": 1.0769149739286152,
      "grad_norm": 0.05885156568943065,
      "learning_rate": 0.00018120832887654785,
      "loss": 0.8522,
      "step": 61340
    },
    {
      "epoch": 1.0770905388086167,
      "grad_norm": 0.05900419537821282,
      "learning_rate": 0.00018120179625994925,
      "loss": 0.8562,
      "step": 61350
    },
    {
      "epoch": 1.0772661036886182,
      "grad_norm": 0.07137643068363302,
      "learning_rate": 0.00018119526262717848,
      "loss": 0.8434,
      "step": 61360
    },
    {
      "epoch": 1.0774416685686194,
      "grad_norm": 0.07261975578836526,
      "learning_rate": 0.00018118872797831846,
      "loss": 0.8407,
      "step": 61370
    },
    {
      "epoch": 1.077617233448621,
      "grad_norm": 0.08410748329040518,
      "learning_rate": 0.00018118219231345184,
      "loss": 0.8531,
      "step": 61380
    },
    {
      "epoch": 1.0777927983286224,
      "grad_norm": 0.06814595942692064,
      "learning_rate": 0.00018117565563266147,
      "loss": 0.8511,
      "step": 61390
    },
    {
      "epoch": 1.0779683632086237,
      "grad_norm": 0.08661230790651667,
      "learning_rate": 0.00018116911793603018,
      "loss": 0.8489,
      "step": 61400
    },
    {
      "epoch": 1.0781439280886251,
      "grad_norm": 0.06295162123219887,
      "learning_rate": 0.00018116257922364077,
      "loss": 0.8475,
      "step": 61410
    },
    {
      "epoch": 1.0783194929686266,
      "grad_norm": 0.05999521759867772,
      "learning_rate": 0.0001811560394955762,
      "loss": 0.8464,
      "step": 61420
    },
    {
      "epoch": 1.0784950578486279,
      "grad_norm": 0.06390095949545617,
      "learning_rate": 0.00018114949875191918,
      "loss": 0.8546,
      "step": 61430
    },
    {
      "epoch": 1.0786706227286293,
      "grad_norm": 0.04882657467286585,
      "learning_rate": 0.00018114295699275264,
      "loss": 0.8494,
      "step": 61440
    },
    {
      "epoch": 1.0788461876086308,
      "grad_norm": 0.06789260078382853,
      "learning_rate": 0.0001811364142181595,
      "loss": 0.8501,
      "step": 61450
    },
    {
      "epoch": 1.079021752488632,
      "grad_norm": 0.056123297734853125,
      "learning_rate": 0.00018112987042822267,
      "loss": 0.8534,
      "step": 61460
    },
    {
      "epoch": 1.0791973173686336,
      "grad_norm": 0.06993204973128615,
      "learning_rate": 0.00018112332562302498,
      "loss": 0.8461,
      "step": 61470
    },
    {
      "epoch": 1.079372882248635,
      "grad_norm": 0.05418157517624619,
      "learning_rate": 0.00018111677980264942,
      "loss": 0.8522,
      "step": 61480
    },
    {
      "epoch": 1.0795484471286363,
      "grad_norm": 0.0625156686032595,
      "learning_rate": 0.0001811102329671789,
      "loss": 0.8546,
      "step": 61490
    },
    {
      "epoch": 1.0797240120086378,
      "grad_norm": 0.048455081786156186,
      "learning_rate": 0.00018110368511669637,
      "loss": 0.8528,
      "step": 61500
    },
    {
      "epoch": 1.0798995768886392,
      "grad_norm": 0.06729999027158527,
      "learning_rate": 0.0001810971362512848,
      "loss": 0.8474,
      "step": 61510
    },
    {
      "epoch": 1.0800751417686405,
      "grad_norm": 0.07429453016513118,
      "learning_rate": 0.00018109058637102716,
      "loss": 0.8444,
      "step": 61520
    },
    {
      "epoch": 1.080250706648642,
      "grad_norm": 0.04225325298497373,
      "learning_rate": 0.00018108403547600644,
      "loss": 0.8499,
      "step": 61530
    },
    {
      "epoch": 1.0804262715286435,
      "grad_norm": 0.07383898658660464,
      "learning_rate": 0.00018107748356630565,
      "loss": 0.8527,
      "step": 61540
    },
    {
      "epoch": 1.0806018364086447,
      "grad_norm": 0.059489934447039386,
      "learning_rate": 0.0001810709306420078,
      "loss": 0.855,
      "step": 61550
    },
    {
      "epoch": 1.0807774012886462,
      "grad_norm": 0.08308266130334194,
      "learning_rate": 0.0001810643767031959,
      "loss": 0.8507,
      "step": 61560
    },
    {
      "epoch": 1.0809529661686477,
      "grad_norm": 0.05140597948719829,
      "learning_rate": 0.00018105782174995303,
      "loss": 0.8458,
      "step": 61570
    },
    {
      "epoch": 1.0811285310486491,
      "grad_norm": 0.046821462169328885,
      "learning_rate": 0.0001810512657823622,
      "loss": 0.8484,
      "step": 61580
    },
    {
      "epoch": 1.0813040959286504,
      "grad_norm": 0.059194332170896154,
      "learning_rate": 0.00018104470880050646,
      "loss": 0.8594,
      "step": 61590
    },
    {
      "epoch": 1.0814796608086519,
      "grad_norm": 0.06596832387493809,
      "learning_rate": 0.00018103815080446897,
      "loss": 0.8501,
      "step": 61600
    },
    {
      "epoch": 1.0816552256886531,
      "grad_norm": 0.04712628563900404,
      "learning_rate": 0.00018103159179433273,
      "loss": 0.8501,
      "step": 61610
    },
    {
      "epoch": 1.0818307905686546,
      "grad_norm": 0.06692339181445993,
      "learning_rate": 0.0001810250317701809,
      "loss": 0.8538,
      "step": 61620
    },
    {
      "epoch": 1.082006355448656,
      "grad_norm": 0.05658241255977982,
      "learning_rate": 0.00018101847073209658,
      "loss": 0.8462,
      "step": 61630
    },
    {
      "epoch": 1.0821819203286576,
      "grad_norm": 0.06426585325539254,
      "learning_rate": 0.0001810119086801629,
      "loss": 0.8453,
      "step": 61640
    },
    {
      "epoch": 1.0823574852086588,
      "grad_norm": 0.05087924401149423,
      "learning_rate": 0.000181005345614463,
      "loss": 0.8498,
      "step": 61650
    },
    {
      "epoch": 1.0825330500886603,
      "grad_norm": 0.056728471334558384,
      "learning_rate": 0.00018099878153508004,
      "loss": 0.8482,
      "step": 61660
    },
    {
      "epoch": 1.0827086149686616,
      "grad_norm": 0.07496254226973188,
      "learning_rate": 0.00018099221644209717,
      "loss": 0.8518,
      "step": 61670
    },
    {
      "epoch": 1.082884179848663,
      "grad_norm": 0.052243609542536436,
      "learning_rate": 0.00018098565033559762,
      "loss": 0.8541,
      "step": 61680
    },
    {
      "epoch": 1.0830597447286645,
      "grad_norm": 0.06236640631687285,
      "learning_rate": 0.00018097908321566456,
      "loss": 0.853,
      "step": 61690
    },
    {
      "epoch": 1.083235309608666,
      "grad_norm": 0.07120128971789103,
      "learning_rate": 0.00018097251508238113,
      "loss": 0.8478,
      "step": 61700
    },
    {
      "epoch": 1.0834108744886672,
      "grad_norm": 0.06106208693982511,
      "learning_rate": 0.00018096594593583064,
      "loss": 0.8511,
      "step": 61710
    },
    {
      "epoch": 1.0835864393686687,
      "grad_norm": 0.07162533577556529,
      "learning_rate": 0.00018095937577609632,
      "loss": 0.8524,
      "step": 61720
    },
    {
      "epoch": 1.0837620042486702,
      "grad_norm": 0.053087796874725646,
      "learning_rate": 0.00018095280460326133,
      "loss": 0.8495,
      "step": 61730
    },
    {
      "epoch": 1.0839375691286715,
      "grad_norm": 0.06293402332507622,
      "learning_rate": 0.00018094623241740902,
      "loss": 0.8514,
      "step": 61740
    },
    {
      "epoch": 1.084113134008673,
      "grad_norm": 0.05531033034611246,
      "learning_rate": 0.0001809396592186226,
      "loss": 0.8482,
      "step": 61750
    },
    {
      "epoch": 1.0842886988886744,
      "grad_norm": 0.0613203833061528,
      "learning_rate": 0.0001809330850069854,
      "loss": 0.8466,
      "step": 61760
    },
    {
      "epoch": 1.0844642637686757,
      "grad_norm": 0.05916560995980617,
      "learning_rate": 0.0001809265097825807,
      "loss": 0.8539,
      "step": 61770
    },
    {
      "epoch": 1.0846398286486771,
      "grad_norm": 0.05469748895178347,
      "learning_rate": 0.00018091993354549176,
      "loss": 0.8503,
      "step": 61780
    },
    {
      "epoch": 1.0848153935286786,
      "grad_norm": 0.056909404242345925,
      "learning_rate": 0.000180913356295802,
      "loss": 0.8491,
      "step": 61790
    },
    {
      "epoch": 1.0849909584086799,
      "grad_norm": 0.08317196596843635,
      "learning_rate": 0.00018090677803359466,
      "loss": 0.8557,
      "step": 61800
    },
    {
      "epoch": 1.0851665232886814,
      "grad_norm": 0.05206790648214862,
      "learning_rate": 0.00018090019875895312,
      "loss": 0.8533,
      "step": 61810
    },
    {
      "epoch": 1.0853420881686828,
      "grad_norm": 0.05528122015136484,
      "learning_rate": 0.00018089361847196077,
      "loss": 0.8533,
      "step": 61820
    },
    {
      "epoch": 1.085517653048684,
      "grad_norm": 0.04752043415446419,
      "learning_rate": 0.00018088703717270097,
      "loss": 0.8578,
      "step": 61830
    },
    {
      "epoch": 1.0856932179286856,
      "grad_norm": 0.04724508712690143,
      "learning_rate": 0.00018088045486125707,
      "loss": 0.851,
      "step": 61840
    },
    {
      "epoch": 1.085868782808687,
      "grad_norm": 0.07630966103327565,
      "learning_rate": 0.00018087387153771254,
      "loss": 0.8585,
      "step": 61850
    },
    {
      "epoch": 1.0860443476886883,
      "grad_norm": 0.06499777699630122,
      "learning_rate": 0.00018086728720215074,
      "loss": 0.8524,
      "step": 61860
    },
    {
      "epoch": 1.0862199125686898,
      "grad_norm": 0.06458319472080983,
      "learning_rate": 0.00018086070185465507,
      "loss": 0.8491,
      "step": 61870
    },
    {
      "epoch": 1.0863954774486912,
      "grad_norm": 0.06638378603018875,
      "learning_rate": 0.00018085411549530902,
      "loss": 0.8514,
      "step": 61880
    },
    {
      "epoch": 1.0865710423286925,
      "grad_norm": 0.07080928967134459,
      "learning_rate": 0.00018084752812419603,
      "loss": 0.8459,
      "step": 61890
    },
    {
      "epoch": 1.086746607208694,
      "grad_norm": 0.07287026211078547,
      "learning_rate": 0.00018084093974139956,
      "loss": 0.8483,
      "step": 61900
    },
    {
      "epoch": 1.0869221720886955,
      "grad_norm": 0.08503901361811703,
      "learning_rate": 0.00018083435034700306,
      "loss": 0.8588,
      "step": 61910
    },
    {
      "epoch": 1.0870977369686967,
      "grad_norm": 0.07827749218119812,
      "learning_rate": 0.00018082775994109006,
      "loss": 0.8526,
      "step": 61920
    },
    {
      "epoch": 1.0872733018486982,
      "grad_norm": 0.05418022735514971,
      "learning_rate": 0.00018082116852374407,
      "loss": 0.8532,
      "step": 61930
    },
    {
      "epoch": 1.0874488667286997,
      "grad_norm": 0.0683115943138406,
      "learning_rate": 0.00018081457609504852,
      "loss": 0.8536,
      "step": 61940
    },
    {
      "epoch": 1.087624431608701,
      "grad_norm": 0.05843054249545686,
      "learning_rate": 0.00018080798265508705,
      "loss": 0.844,
      "step": 61950
    },
    {
      "epoch": 1.0877999964887024,
      "grad_norm": 0.059790847787108005,
      "learning_rate": 0.0001808013882039431,
      "loss": 0.8512,
      "step": 61960
    },
    {
      "epoch": 1.0879755613687039,
      "grad_norm": 0.05023299749625879,
      "learning_rate": 0.0001807947927417003,
      "loss": 0.8478,
      "step": 61970
    },
    {
      "epoch": 1.0881511262487051,
      "grad_norm": 0.08044081902243516,
      "learning_rate": 0.00018078819626844215,
      "loss": 0.8529,
      "step": 61980
    },
    {
      "epoch": 1.0883266911287066,
      "grad_norm": 0.06422885927349552,
      "learning_rate": 0.00018078159878425227,
      "loss": 0.8467,
      "step": 61990
    },
    {
      "epoch": 1.088502256008708,
      "grad_norm": 0.06155370304233618,
      "learning_rate": 0.00018077500028921428,
      "loss": 0.8414,
      "step": 62000
    },
    {
      "epoch": 1.0886778208887093,
      "grad_norm": 0.05410893706721108,
      "learning_rate": 0.00018076840078341168,
      "loss": 0.8482,
      "step": 62010
    },
    {
      "epoch": 1.0888533857687108,
      "grad_norm": 0.06082758702504676,
      "learning_rate": 0.0001807618002669282,
      "loss": 0.8522,
      "step": 62020
    },
    {
      "epoch": 1.0890289506487123,
      "grad_norm": 0.062640904931616,
      "learning_rate": 0.00018075519873984744,
      "loss": 0.8491,
      "step": 62030
    },
    {
      "epoch": 1.0892045155287136,
      "grad_norm": 0.06820114942737333,
      "learning_rate": 0.00018074859620225298,
      "loss": 0.8493,
      "step": 62040
    },
    {
      "epoch": 1.089380080408715,
      "grad_norm": 0.05791195640913787,
      "learning_rate": 0.0001807419926542286,
      "loss": 0.8514,
      "step": 62050
    },
    {
      "epoch": 1.0895556452887165,
      "grad_norm": 0.0615979206538511,
      "learning_rate": 0.00018073538809585776,
      "loss": 0.8507,
      "step": 62060
    },
    {
      "epoch": 1.0897312101687178,
      "grad_norm": 0.06396255850589375,
      "learning_rate": 0.0001807287825272244,
      "loss": 0.8516,
      "step": 62070
    },
    {
      "epoch": 1.0899067750487192,
      "grad_norm": 0.05060906550496095,
      "learning_rate": 0.00018072217594841203,
      "loss": 0.8527,
      "step": 62080
    },
    {
      "epoch": 1.0900823399287207,
      "grad_norm": 0.08157992451739296,
      "learning_rate": 0.00018071556835950442,
      "loss": 0.8493,
      "step": 62090
    },
    {
      "epoch": 1.090257904808722,
      "grad_norm": 0.056258443908712895,
      "learning_rate": 0.00018070895976058526,
      "loss": 0.8559,
      "step": 62100
    },
    {
      "epoch": 1.0904334696887235,
      "grad_norm": 0.04754357476679131,
      "learning_rate": 0.00018070235015173834,
      "loss": 0.8507,
      "step": 62110
    },
    {
      "epoch": 1.090609034568725,
      "grad_norm": 0.04978886466317822,
      "learning_rate": 0.00018069573953304735,
      "loss": 0.8497,
      "step": 62120
    },
    {
      "epoch": 1.0907845994487262,
      "grad_norm": 0.04939258716084147,
      "learning_rate": 0.00018068912790459606,
      "loss": 0.8545,
      "step": 62130
    },
    {
      "epoch": 1.0909601643287277,
      "grad_norm": 0.07184718545498965,
      "learning_rate": 0.00018068251526646826,
      "loss": 0.8546,
      "step": 62140
    },
    {
      "epoch": 1.0911357292087291,
      "grad_norm": 0.06334322503174526,
      "learning_rate": 0.0001806759016187477,
      "loss": 0.8466,
      "step": 62150
    },
    {
      "epoch": 1.0913112940887304,
      "grad_norm": 0.051934476821033526,
      "learning_rate": 0.00018066928696151822,
      "loss": 0.8503,
      "step": 62160
    },
    {
      "epoch": 1.0914868589687319,
      "grad_norm": 0.06583014769441684,
      "learning_rate": 0.0001806626712948636,
      "loss": 0.8435,
      "step": 62170
    },
    {
      "epoch": 1.0916624238487334,
      "grad_norm": 0.06287177525421733,
      "learning_rate": 0.00018065605461886766,
      "loss": 0.8501,
      "step": 62180
    },
    {
      "epoch": 1.0918379887287346,
      "grad_norm": 0.05545856932862779,
      "learning_rate": 0.00018064943693361425,
      "loss": 0.8512,
      "step": 62190
    },
    {
      "epoch": 1.092013553608736,
      "grad_norm": 0.06928479371450576,
      "learning_rate": 0.00018064281823918722,
      "loss": 0.8487,
      "step": 62200
    },
    {
      "epoch": 1.0921891184887376,
      "grad_norm": 0.06642067246293469,
      "learning_rate": 0.0001806361985356704,
      "loss": 0.8428,
      "step": 62210
    },
    {
      "epoch": 1.0923646833687388,
      "grad_norm": 0.06768775913397528,
      "learning_rate": 0.00018062957782314767,
      "loss": 0.8474,
      "step": 62220
    },
    {
      "epoch": 1.0925402482487403,
      "grad_norm": 0.059902596412086284,
      "learning_rate": 0.00018062295610170296,
      "loss": 0.8543,
      "step": 62230
    },
    {
      "epoch": 1.0927158131287418,
      "grad_norm": 0.07683787231397832,
      "learning_rate": 0.00018061633337142013,
      "loss": 0.8476,
      "step": 62240
    },
    {
      "epoch": 1.092891378008743,
      "grad_norm": 0.06383171922958475,
      "learning_rate": 0.00018060970963238305,
      "loss": 0.8485,
      "step": 62250
    },
    {
      "epoch": 1.0930669428887445,
      "grad_norm": 0.07517493995232265,
      "learning_rate": 0.00018060308488467578,
      "loss": 0.8524,
      "step": 62260
    },
    {
      "epoch": 1.093242507768746,
      "grad_norm": 0.04778258484979142,
      "learning_rate": 0.0001805964591283821,
      "loss": 0.8475,
      "step": 62270
    },
    {
      "epoch": 1.0934180726487472,
      "grad_norm": 0.05109455246665613,
      "learning_rate": 0.00018058983236358603,
      "loss": 0.8445,
      "step": 62280
    },
    {
      "epoch": 1.0935936375287487,
      "grad_norm": 0.06371268664659115,
      "learning_rate": 0.00018058320459037156,
      "loss": 0.8519,
      "step": 62290
    },
    {
      "epoch": 1.0937692024087502,
      "grad_norm": 0.0810717546056745,
      "learning_rate": 0.00018057657580882263,
      "loss": 0.8497,
      "step": 62300
    },
    {
      "epoch": 1.0939447672887517,
      "grad_norm": 0.0632514110472606,
      "learning_rate": 0.0001805699460190232,
      "loss": 0.8524,
      "step": 62310
    },
    {
      "epoch": 1.094120332168753,
      "grad_norm": 0.08294901567916205,
      "learning_rate": 0.00018056331522105736,
      "loss": 0.8478,
      "step": 62320
    },
    {
      "epoch": 1.0942958970487544,
      "grad_norm": 0.06059212974565299,
      "learning_rate": 0.00018055668341500903,
      "loss": 0.8444,
      "step": 62330
    },
    {
      "epoch": 1.0944714619287557,
      "grad_norm": 0.05243597581702847,
      "learning_rate": 0.0001805500506009623,
      "loss": 0.8486,
      "step": 62340
    },
    {
      "epoch": 1.0946470268087571,
      "grad_norm": 0.05784115910656347,
      "learning_rate": 0.00018054341677900117,
      "loss": 0.8493,
      "step": 62350
    },
    {
      "epoch": 1.0948225916887586,
      "grad_norm": 0.07022938987935388,
      "learning_rate": 0.0001805367819492097,
      "loss": 0.8465,
      "step": 62360
    },
    {
      "epoch": 1.09499815656876,
      "grad_norm": 0.06461527544835577,
      "learning_rate": 0.00018053014611167195,
      "loss": 0.8557,
      "step": 62370
    },
    {
      "epoch": 1.0951737214487614,
      "grad_norm": 0.06056698767609542,
      "learning_rate": 0.00018052350926647202,
      "loss": 0.8534,
      "step": 62380
    },
    {
      "epoch": 1.0953492863287628,
      "grad_norm": 0.08833395541729215,
      "learning_rate": 0.00018051687141369397,
      "loss": 0.8453,
      "step": 62390
    },
    {
      "epoch": 1.095524851208764,
      "grad_norm": 0.0504368964767674,
      "learning_rate": 0.00018051023255342193,
      "loss": 0.8544,
      "step": 62400
    },
    {
      "epoch": 1.0957004160887656,
      "grad_norm": 0.07123460281915148,
      "learning_rate": 0.00018050359268574002,
      "loss": 0.8613,
      "step": 62410
    },
    {
      "epoch": 1.095875980968767,
      "grad_norm": 0.06489886044733674,
      "learning_rate": 0.00018049695181073235,
      "loss": 0.8531,
      "step": 62420
    },
    {
      "epoch": 1.0960515458487685,
      "grad_norm": 0.048775178602669486,
      "learning_rate": 0.00018049030992848308,
      "loss": 0.8522,
      "step": 62430
    },
    {
      "epoch": 1.0962271107287698,
      "grad_norm": 0.06020943606535469,
      "learning_rate": 0.00018048366703907634,
      "loss": 0.8491,
      "step": 62440
    },
    {
      "epoch": 1.0964026756087712,
      "grad_norm": 0.09788900778500571,
      "learning_rate": 0.00018047702314259628,
      "loss": 0.8542,
      "step": 62450
    },
    {
      "epoch": 1.0965782404887727,
      "grad_norm": 0.05899009803508139,
      "learning_rate": 0.00018047037823912712,
      "loss": 0.8504,
      "step": 62460
    },
    {
      "epoch": 1.096753805368774,
      "grad_norm": 0.06939540620152164,
      "learning_rate": 0.00018046373232875304,
      "loss": 0.8484,
      "step": 62470
    },
    {
      "epoch": 1.0969293702487755,
      "grad_norm": 0.052364327132331204,
      "learning_rate": 0.00018045708541155825,
      "loss": 0.8471,
      "step": 62480
    },
    {
      "epoch": 1.097104935128777,
      "grad_norm": 0.05800470555926506,
      "learning_rate": 0.00018045043748762693,
      "loss": 0.8551,
      "step": 62490
    },
    {
      "epoch": 1.0972805000087782,
      "grad_norm": 0.04639319789121787,
      "learning_rate": 0.00018044378855704336,
      "loss": 0.8554,
      "step": 62500
    },
    {
      "epoch": 1.0974560648887797,
      "grad_norm": 0.0646342589031617,
      "learning_rate": 0.00018043713861989178,
      "loss": 0.848,
      "step": 62510
    },
    {
      "epoch": 1.0976316297687811,
      "grad_norm": 0.06315564122072045,
      "learning_rate": 0.00018043048767625643,
      "loss": 0.8565,
      "step": 62520
    },
    {
      "epoch": 1.0978071946487824,
      "grad_norm": 0.06466841618070558,
      "learning_rate": 0.00018042383572622157,
      "loss": 0.8541,
      "step": 62530
    },
    {
      "epoch": 1.0979827595287839,
      "grad_norm": 0.060684908154012326,
      "learning_rate": 0.00018041718276987144,
      "loss": 0.8527,
      "step": 62540
    },
    {
      "epoch": 1.0981583244087854,
      "grad_norm": 0.07587011901486988,
      "learning_rate": 0.00018041052880729043,
      "loss": 0.8506,
      "step": 62550
    },
    {
      "epoch": 1.0983338892887866,
      "grad_norm": 0.054362218018544815,
      "learning_rate": 0.0001804038738385628,
      "loss": 0.8547,
      "step": 62560
    },
    {
      "epoch": 1.098509454168788,
      "grad_norm": 0.06182183298292023,
      "learning_rate": 0.0001803972178637729,
      "loss": 0.8439,
      "step": 62570
    },
    {
      "epoch": 1.0986850190487896,
      "grad_norm": 0.07140853584081713,
      "learning_rate": 0.000180390560883005,
      "loss": 0.8539,
      "step": 62580
    },
    {
      "epoch": 1.0988605839287908,
      "grad_norm": 0.061352432974223846,
      "learning_rate": 0.00018038390289634346,
      "loss": 0.8534,
      "step": 62590
    },
    {
      "epoch": 1.0990361488087923,
      "grad_norm": 0.05323662902742413,
      "learning_rate": 0.00018037724390387267,
      "loss": 0.8447,
      "step": 62600
    },
    {
      "epoch": 1.0992117136887938,
      "grad_norm": 0.056477550406711326,
      "learning_rate": 0.000180370583905677,
      "loss": 0.8501,
      "step": 62610
    },
    {
      "epoch": 1.099387278568795,
      "grad_norm": 0.07379514660580348,
      "learning_rate": 0.0001803639229018408,
      "loss": 0.8488,
      "step": 62620
    },
    {
      "epoch": 1.0995628434487965,
      "grad_norm": 0.07947353526632331,
      "learning_rate": 0.00018035726089244854,
      "loss": 0.8506,
      "step": 62630
    },
    {
      "epoch": 1.099738408328798,
      "grad_norm": 0.06633609644949286,
      "learning_rate": 0.00018035059787758454,
      "loss": 0.8454,
      "step": 62640
    },
    {
      "epoch": 1.0999139732087992,
      "grad_norm": 0.05061086776793537,
      "learning_rate": 0.00018034393385733322,
      "loss": 0.8445,
      "step": 62650
    },
    {
      "epoch": 1.1000895380888007,
      "grad_norm": 0.0923237645197295,
      "learning_rate": 0.0001803372688317791,
      "loss": 0.854,
      "step": 62660
    },
    {
      "epoch": 1.1002651029688022,
      "grad_norm": 0.053915707627111,
      "learning_rate": 0.00018033060280100657,
      "loss": 0.851,
      "step": 62670
    },
    {
      "epoch": 1.1004406678488035,
      "grad_norm": 0.06730922242750162,
      "learning_rate": 0.00018032393576510007,
      "loss": 0.8442,
      "step": 62680
    },
    {
      "epoch": 1.100616232728805,
      "grad_norm": 0.05121199540074441,
      "learning_rate": 0.0001803172677241441,
      "loss": 0.8491,
      "step": 62690
    },
    {
      "epoch": 1.1007917976088064,
      "grad_norm": 0.08899140122181526,
      "learning_rate": 0.00018031059867822322,
      "loss": 0.8492,
      "step": 62700
    },
    {
      "epoch": 1.1009673624888077,
      "grad_norm": 0.09365323087330481,
      "learning_rate": 0.0001803039286274218,
      "loss": 0.8594,
      "step": 62710
    },
    {
      "epoch": 1.1011429273688091,
      "grad_norm": 0.05289444686322601,
      "learning_rate": 0.0001802972575718244,
      "loss": 0.8493,
      "step": 62720
    },
    {
      "epoch": 1.1013184922488106,
      "grad_norm": 0.04922873006672966,
      "learning_rate": 0.00018029058551151557,
      "loss": 0.8516,
      "step": 62730
    },
    {
      "epoch": 1.1014940571288119,
      "grad_norm": 0.06375934299782218,
      "learning_rate": 0.00018028391244657982,
      "loss": 0.8588,
      "step": 62740
    },
    {
      "epoch": 1.1016696220088134,
      "grad_norm": 0.06889876938072478,
      "learning_rate": 0.0001802772383771017,
      "loss": 0.8563,
      "step": 62750
    },
    {
      "epoch": 1.1018451868888148,
      "grad_norm": 0.05156762361187861,
      "learning_rate": 0.00018027056330316578,
      "loss": 0.8483,
      "step": 62760
    },
    {
      "epoch": 1.102020751768816,
      "grad_norm": 0.05789546086691783,
      "learning_rate": 0.0001802638872248566,
      "loss": 0.8546,
      "step": 62770
    },
    {
      "epoch": 1.1021963166488176,
      "grad_norm": 0.06991982227207343,
      "learning_rate": 0.00018025721014225884,
      "loss": 0.8612,
      "step": 62780
    },
    {
      "epoch": 1.102371881528819,
      "grad_norm": 0.06941891744316418,
      "learning_rate": 0.000180250532055457,
      "loss": 0.8594,
      "step": 62790
    },
    {
      "epoch": 1.1025474464088203,
      "grad_norm": 0.07064434819914521,
      "learning_rate": 0.00018024385296453574,
      "loss": 0.8487,
      "step": 62800
    },
    {
      "epoch": 1.1027230112888218,
      "grad_norm": 0.06406216162939506,
      "learning_rate": 0.00018023717286957966,
      "loss": 0.85,
      "step": 62810
    },
    {
      "epoch": 1.1028985761688233,
      "grad_norm": 0.06164652363047016,
      "learning_rate": 0.00018023049177067345,
      "loss": 0.8503,
      "step": 62820
    },
    {
      "epoch": 1.1030741410488245,
      "grad_norm": 0.05655805806531401,
      "learning_rate": 0.00018022380966790168,
      "loss": 0.8508,
      "step": 62830
    },
    {
      "epoch": 1.103249705928826,
      "grad_norm": 0.05893699772832099,
      "learning_rate": 0.0001802171265613491,
      "loss": 0.8492,
      "step": 62840
    },
    {
      "epoch": 1.1034252708088275,
      "grad_norm": 0.0704355296045491,
      "learning_rate": 0.0001802104424511003,
      "loss": 0.8475,
      "step": 62850
    },
    {
      "epoch": 1.1036008356888287,
      "grad_norm": 0.059920991490145264,
      "learning_rate": 0.00018020375733724005,
      "loss": 0.854,
      "step": 62860
    },
    {
      "epoch": 1.1037764005688302,
      "grad_norm": 0.08096592676760217,
      "learning_rate": 0.000180197071219853,
      "loss": 0.8514,
      "step": 62870
    },
    {
      "epoch": 1.1039519654488317,
      "grad_norm": 0.05561686236379571,
      "learning_rate": 0.00018019038409902388,
      "loss": 0.8548,
      "step": 62880
    },
    {
      "epoch": 1.104127530328833,
      "grad_norm": 0.04836728217699151,
      "learning_rate": 0.0001801836959748374,
      "loss": 0.8508,
      "step": 62890
    },
    {
      "epoch": 1.1043030952088344,
      "grad_norm": 0.06222220608742484,
      "learning_rate": 0.00018017700684737836,
      "loss": 0.8484,
      "step": 62900
    },
    {
      "epoch": 1.1044786600888359,
      "grad_norm": 0.08863599643461202,
      "learning_rate": 0.00018017031671673143,
      "loss": 0.8419,
      "step": 62910
    },
    {
      "epoch": 1.1046542249688371,
      "grad_norm": 0.06418956462334477,
      "learning_rate": 0.00018016362558298144,
      "loss": 0.8504,
      "step": 62920
    },
    {
      "epoch": 1.1048297898488386,
      "grad_norm": 0.04990723683174192,
      "learning_rate": 0.00018015693344621312,
      "loss": 0.8488,
      "step": 62930
    },
    {
      "epoch": 1.10500535472884,
      "grad_norm": 0.06584887878347451,
      "learning_rate": 0.00018015024030651128,
      "loss": 0.8428,
      "step": 62940
    },
    {
      "epoch": 1.1051809196088414,
      "grad_norm": 0.11972151271772306,
      "learning_rate": 0.00018014354616396075,
      "loss": 0.8554,
      "step": 62950
    },
    {
      "epoch": 1.1053564844888428,
      "grad_norm": 0.07607075459381887,
      "learning_rate": 0.00018013685101864635,
      "loss": 0.8464,
      "step": 62960
    },
    {
      "epoch": 1.1055320493688443,
      "grad_norm": 0.05442022113597486,
      "learning_rate": 0.00018013015487065284,
      "loss": 0.8485,
      "step": 62970
    },
    {
      "epoch": 1.1057076142488456,
      "grad_norm": 0.06839038394430898,
      "learning_rate": 0.00018012345772006508,
      "loss": 0.8453,
      "step": 62980
    },
    {
      "epoch": 1.105883179128847,
      "grad_norm": 0.06193393867733681,
      "learning_rate": 0.00018011675956696797,
      "loss": 0.8511,
      "step": 62990
    },
    {
      "epoch": 1.1060587440088485,
      "grad_norm": 0.11295017275503236,
      "learning_rate": 0.00018011006041144636,
      "loss": 0.848,
      "step": 63000
    },
    {
      "epoch": 1.1062343088888498,
      "grad_norm": 0.05425598863892647,
      "learning_rate": 0.0001801033602535851,
      "loss": 0.8473,
      "step": 63010
    },
    {
      "epoch": 1.1064098737688512,
      "grad_norm": 0.06241776887133908,
      "learning_rate": 0.00018009665909346915,
      "loss": 0.8473,
      "step": 63020
    },
    {
      "epoch": 1.1065854386488527,
      "grad_norm": 0.05766296909913649,
      "learning_rate": 0.00018008995693118333,
      "loss": 0.8461,
      "step": 63030
    },
    {
      "epoch": 1.1067610035288542,
      "grad_norm": 0.055452190402101274,
      "learning_rate": 0.00018008325376681262,
      "loss": 0.845,
      "step": 63040
    },
    {
      "epoch": 1.1069365684088555,
      "grad_norm": 0.06397577342526146,
      "learning_rate": 0.00018007654960044191,
      "loss": 0.8482,
      "step": 63050
    },
    {
      "epoch": 1.107112133288857,
      "grad_norm": 0.07827388631886845,
      "learning_rate": 0.0001800698444321562,
      "loss": 0.8485,
      "step": 63060
    },
    {
      "epoch": 1.1072876981688582,
      "grad_norm": 0.07395233042588756,
      "learning_rate": 0.00018006313826204038,
      "loss": 0.8454,
      "step": 63070
    },
    {
      "epoch": 1.1074632630488597,
      "grad_norm": 0.045638669454953415,
      "learning_rate": 0.00018005643109017942,
      "loss": 0.847,
      "step": 63080
    },
    {
      "epoch": 1.1076388279288611,
      "grad_norm": 0.061520712755089225,
      "learning_rate": 0.00018004972291665836,
      "loss": 0.845,
      "step": 63090
    },
    {
      "epoch": 1.1078143928088626,
      "grad_norm": 0.05332700926653336,
      "learning_rate": 0.00018004301374156215,
      "loss": 0.8544,
      "step": 63100
    },
    {
      "epoch": 1.1079899576888639,
      "grad_norm": 0.06567077990023226,
      "learning_rate": 0.00018003630356497582,
      "loss": 0.8551,
      "step": 63110
    },
    {
      "epoch": 1.1081655225688654,
      "grad_norm": 0.0880887561139268,
      "learning_rate": 0.00018002959238698436,
      "loss": 0.8451,
      "step": 63120
    },
    {
      "epoch": 1.1083410874488666,
      "grad_norm": 0.04606076345563506,
      "learning_rate": 0.00018002288020767284,
      "loss": 0.8504,
      "step": 63130
    },
    {
      "epoch": 1.108516652328868,
      "grad_norm": 0.052002597272976576,
      "learning_rate": 0.00018001616702712628,
      "loss": 0.8591,
      "step": 63140
    },
    {
      "epoch": 1.1086922172088696,
      "grad_norm": 0.06607111486300729,
      "learning_rate": 0.00018000945284542974,
      "loss": 0.8478,
      "step": 63150
    },
    {
      "epoch": 1.108867782088871,
      "grad_norm": 0.04730530309219683,
      "learning_rate": 0.00018000273766266827,
      "loss": 0.8485,
      "step": 63160
    },
    {
      "epoch": 1.1090433469688723,
      "grad_norm": 0.06268018671524156,
      "learning_rate": 0.00017999602147892698,
      "loss": 0.8542,
      "step": 63170
    },
    {
      "epoch": 1.1092189118488738,
      "grad_norm": 0.04987068547251352,
      "learning_rate": 0.00017998930429429094,
      "loss": 0.8442,
      "step": 63180
    },
    {
      "epoch": 1.1093944767288753,
      "grad_norm": 0.05065291536528936,
      "learning_rate": 0.00017998258610884532,
      "loss": 0.8528,
      "step": 63190
    },
    {
      "epoch": 1.1095700416088765,
      "grad_norm": 0.0632407934084691,
      "learning_rate": 0.0001799758669226752,
      "loss": 0.8531,
      "step": 63200
    },
    {
      "epoch": 1.109745606488878,
      "grad_norm": 0.06632721114776366,
      "learning_rate": 0.00017996914673586566,
      "loss": 0.8515,
      "step": 63210
    },
    {
      "epoch": 1.1099211713688795,
      "grad_norm": 0.07697551963593856,
      "learning_rate": 0.00017996242554850194,
      "loss": 0.8494,
      "step": 63220
    },
    {
      "epoch": 1.1100967362488807,
      "grad_norm": 0.060369546870639536,
      "learning_rate": 0.0001799557033606691,
      "loss": 0.8392,
      "step": 63230
    },
    {
      "epoch": 1.1102723011288822,
      "grad_norm": 0.070138534515946,
      "learning_rate": 0.00017994898017245243,
      "loss": 0.8455,
      "step": 63240
    },
    {
      "epoch": 1.1104478660088837,
      "grad_norm": 0.05315433463846012,
      "learning_rate": 0.000179942255983937,
      "loss": 0.8482,
      "step": 63250
    },
    {
      "epoch": 1.110623430888885,
      "grad_norm": 0.08259724321906353,
      "learning_rate": 0.0001799355307952081,
      "loss": 0.8497,
      "step": 63260
    },
    {
      "epoch": 1.1107989957688864,
      "grad_norm": 0.0494309506364185,
      "learning_rate": 0.00017992880460635085,
      "loss": 0.8581,
      "step": 63270
    },
    {
      "epoch": 1.1109745606488879,
      "grad_norm": 0.05489959474878671,
      "learning_rate": 0.00017992207741745057,
      "loss": 0.8499,
      "step": 63280
    },
    {
      "epoch": 1.1111501255288891,
      "grad_norm": 0.0620599687597512,
      "learning_rate": 0.00017991534922859244,
      "loss": 0.8471,
      "step": 63290
    },
    {
      "epoch": 1.1113256904088906,
      "grad_norm": 0.07629052351182475,
      "learning_rate": 0.00017990862003986168,
      "loss": 0.8466,
      "step": 63300
    },
    {
      "epoch": 1.111501255288892,
      "grad_norm": 0.10185522738588536,
      "learning_rate": 0.0001799018898513436,
      "loss": 0.8586,
      "step": 63310
    },
    {
      "epoch": 1.1116768201688934,
      "grad_norm": 0.061964786413389504,
      "learning_rate": 0.00017989515866312345,
      "loss": 0.8479,
      "step": 63320
    },
    {
      "epoch": 1.1118523850488948,
      "grad_norm": 0.05340941497823469,
      "learning_rate": 0.00017988842647528652,
      "loss": 0.8596,
      "step": 63330
    },
    {
      "epoch": 1.1120279499288963,
      "grad_norm": 0.06763126841374276,
      "learning_rate": 0.0001798816932879181,
      "loss": 0.8477,
      "step": 63340
    },
    {
      "epoch": 1.1122035148088976,
      "grad_norm": 0.06589109557840915,
      "learning_rate": 0.00017987495910110353,
      "loss": 0.8501,
      "step": 63350
    },
    {
      "epoch": 1.112379079688899,
      "grad_norm": 0.05755379930512501,
      "learning_rate": 0.0001798682239149281,
      "loss": 0.8558,
      "step": 63360
    },
    {
      "epoch": 1.1125546445689005,
      "grad_norm": 0.06497907940661354,
      "learning_rate": 0.00017986148772947716,
      "loss": 0.851,
      "step": 63370
    },
    {
      "epoch": 1.1127302094489018,
      "grad_norm": 0.06473328440362316,
      "learning_rate": 0.00017985475054483604,
      "loss": 0.8524,
      "step": 63380
    },
    {
      "epoch": 1.1129057743289033,
      "grad_norm": 0.08054536979728731,
      "learning_rate": 0.00017984801236109013,
      "loss": 0.8491,
      "step": 63390
    },
    {
      "epoch": 1.1130813392089047,
      "grad_norm": 0.05248842397515805,
      "learning_rate": 0.00017984127317832477,
      "loss": 0.8474,
      "step": 63400
    },
    {
      "epoch": 1.113256904088906,
      "grad_norm": 0.04922428781404128,
      "learning_rate": 0.0001798345329966254,
      "loss": 0.8457,
      "step": 63410
    },
    {
      "epoch": 1.1134324689689075,
      "grad_norm": 0.05988162075506045,
      "learning_rate": 0.00017982779181607738,
      "loss": 0.8497,
      "step": 63420
    },
    {
      "epoch": 1.113608033848909,
      "grad_norm": 0.06425647352251809,
      "learning_rate": 0.00017982104963676614,
      "loss": 0.8474,
      "step": 63430
    },
    {
      "epoch": 1.1137835987289102,
      "grad_norm": 0.09163917988138723,
      "learning_rate": 0.0001798143064587771,
      "loss": 0.8491,
      "step": 63440
    },
    {
      "epoch": 1.1139591636089117,
      "grad_norm": 0.07587613444955431,
      "learning_rate": 0.00017980756228219567,
      "loss": 0.8538,
      "step": 63450
    },
    {
      "epoch": 1.1141347284889132,
      "grad_norm": 0.055592415185579794,
      "learning_rate": 0.0001798008171071073,
      "loss": 0.8504,
      "step": 63460
    },
    {
      "epoch": 1.1143102933689144,
      "grad_norm": 0.057643432323901116,
      "learning_rate": 0.00017979407093359753,
      "loss": 0.8464,
      "step": 63470
    },
    {
      "epoch": 1.1144858582489159,
      "grad_norm": 0.05439464235901788,
      "learning_rate": 0.00017978732376175175,
      "loss": 0.8511,
      "step": 63480
    },
    {
      "epoch": 1.1146614231289174,
      "grad_norm": 0.05489035101905239,
      "learning_rate": 0.0001797805755916555,
      "loss": 0.8537,
      "step": 63490
    },
    {
      "epoch": 1.1148369880089186,
      "grad_norm": 0.045633988544810905,
      "learning_rate": 0.00017977382642339422,
      "loss": 0.849,
      "step": 63500
    },
    {
      "epoch": 1.11501255288892,
      "grad_norm": 0.06463972821484633,
      "learning_rate": 0.0001797670762570535,
      "loss": 0.854,
      "step": 63510
    },
    {
      "epoch": 1.1151881177689216,
      "grad_norm": 0.06529186837019142,
      "learning_rate": 0.00017976032509271878,
      "loss": 0.8526,
      "step": 63520
    },
    {
      "epoch": 1.1153636826489228,
      "grad_norm": 0.055228396535086356,
      "learning_rate": 0.0001797535729304757,
      "loss": 0.8477,
      "step": 63530
    },
    {
      "epoch": 1.1155392475289243,
      "grad_norm": 0.06977386670840079,
      "learning_rate": 0.00017974681977040973,
      "loss": 0.8547,
      "step": 63540
    },
    {
      "epoch": 1.1157148124089258,
      "grad_norm": 0.07482228447544345,
      "learning_rate": 0.00017974006561260648,
      "loss": 0.8511,
      "step": 63550
    },
    {
      "epoch": 1.115890377288927,
      "grad_norm": 0.058189156271562605,
      "learning_rate": 0.0001797333104571515,
      "loss": 0.8513,
      "step": 63560
    },
    {
      "epoch": 1.1160659421689285,
      "grad_norm": 0.0606953135572274,
      "learning_rate": 0.00017972655430413038,
      "loss": 0.8455,
      "step": 63570
    },
    {
      "epoch": 1.11624150704893,
      "grad_norm": 0.06540286732274568,
      "learning_rate": 0.00017971979715362873,
      "loss": 0.8471,
      "step": 63580
    },
    {
      "epoch": 1.1164170719289312,
      "grad_norm": 0.054966008055933484,
      "learning_rate": 0.00017971303900573218,
      "loss": 0.8492,
      "step": 63590
    },
    {
      "epoch": 1.1165926368089327,
      "grad_norm": 0.04253565300491529,
      "learning_rate": 0.00017970627986052632,
      "loss": 0.8479,
      "step": 63600
    },
    {
      "epoch": 1.1167682016889342,
      "grad_norm": 0.06370850176978006,
      "learning_rate": 0.0001796995197180968,
      "loss": 0.8538,
      "step": 63610
    },
    {
      "epoch": 1.1169437665689355,
      "grad_norm": 0.06748679661476702,
      "learning_rate": 0.00017969275857852932,
      "loss": 0.8535,
      "step": 63620
    },
    {
      "epoch": 1.117119331448937,
      "grad_norm": 0.07719412918408544,
      "learning_rate": 0.00017968599644190946,
      "loss": 0.8564,
      "step": 63630
    },
    {
      "epoch": 1.1172948963289384,
      "grad_norm": 0.052356051567401085,
      "learning_rate": 0.000179679233308323,
      "loss": 0.8521,
      "step": 63640
    },
    {
      "epoch": 1.1174704612089397,
      "grad_norm": 0.04828840376406467,
      "learning_rate": 0.00017967246917785555,
      "loss": 0.8505,
      "step": 63650
    },
    {
      "epoch": 1.1176460260889411,
      "grad_norm": 0.07468008035109112,
      "learning_rate": 0.00017966570405059283,
      "loss": 0.8496,
      "step": 63660
    },
    {
      "epoch": 1.1178215909689426,
      "grad_norm": 0.054770199149431674,
      "learning_rate": 0.00017965893792662056,
      "loss": 0.8543,
      "step": 63670
    },
    {
      "epoch": 1.1179971558489439,
      "grad_norm": 0.057438741626564996,
      "learning_rate": 0.0001796521708060245,
      "loss": 0.8632,
      "step": 63680
    },
    {
      "epoch": 1.1181727207289454,
      "grad_norm": 0.04785928792656027,
      "learning_rate": 0.00017964540268889035,
      "loss": 0.8522,
      "step": 63690
    },
    {
      "epoch": 1.1183482856089468,
      "grad_norm": 0.050807078002749646,
      "learning_rate": 0.00017963863357530385,
      "loss": 0.85,
      "step": 63700
    },
    {
      "epoch": 1.118523850488948,
      "grad_norm": 0.052668540826652545,
      "learning_rate": 0.00017963186346535082,
      "loss": 0.8511,
      "step": 63710
    },
    {
      "epoch": 1.1186994153689496,
      "grad_norm": 0.06840474290110996,
      "learning_rate": 0.000179625092359117,
      "loss": 0.8449,
      "step": 63720
    },
    {
      "epoch": 1.118874980248951,
      "grad_norm": 0.06643656486381799,
      "learning_rate": 0.00017961832025668818,
      "loss": 0.8494,
      "step": 63730
    },
    {
      "epoch": 1.1190505451289523,
      "grad_norm": 0.0647990380535818,
      "learning_rate": 0.0001796115471581502,
      "loss": 0.8455,
      "step": 63740
    },
    {
      "epoch": 1.1192261100089538,
      "grad_norm": 0.05093674943920201,
      "learning_rate": 0.0001796047730635888,
      "loss": 0.8602,
      "step": 63750
    },
    {
      "epoch": 1.1194016748889553,
      "grad_norm": 0.07167989582617355,
      "learning_rate": 0.00017959799797308994,
      "loss": 0.8467,
      "step": 63760
    },
    {
      "epoch": 1.1195772397689567,
      "grad_norm": 0.054552500018295834,
      "learning_rate": 0.00017959122188673936,
      "loss": 0.8454,
      "step": 63770
    },
    {
      "epoch": 1.119752804648958,
      "grad_norm": 0.06017898660213819,
      "learning_rate": 0.0001795844448046229,
      "loss": 0.8474,
      "step": 63780
    },
    {
      "epoch": 1.1199283695289595,
      "grad_norm": 0.06797131122057282,
      "learning_rate": 0.0001795776667268265,
      "loss": 0.8604,
      "step": 63790
    },
    {
      "epoch": 1.1201039344089607,
      "grad_norm": 0.058895838436704526,
      "learning_rate": 0.00017957088765343596,
      "loss": 0.8522,
      "step": 63800
    },
    {
      "epoch": 1.1202794992889622,
      "grad_norm": 0.06010545814142448,
      "learning_rate": 0.00017956410758453726,
      "loss": 0.8483,
      "step": 63810
    },
    {
      "epoch": 1.1204550641689637,
      "grad_norm": 0.056630815566001545,
      "learning_rate": 0.00017955732652021625,
      "loss": 0.8474,
      "step": 63820
    },
    {
      "epoch": 1.1206306290489652,
      "grad_norm": 0.05695779608103413,
      "learning_rate": 0.0001795505444605588,
      "loss": 0.8474,
      "step": 63830
    },
    {
      "epoch": 1.1208061939289664,
      "grad_norm": 0.05134219592439935,
      "learning_rate": 0.00017954376140565096,
      "loss": 0.851,
      "step": 63840
    },
    {
      "epoch": 1.1209817588089679,
      "grad_norm": 0.050805892462023106,
      "learning_rate": 0.0001795369773555786,
      "loss": 0.8516,
      "step": 63850
    },
    {
      "epoch": 1.1211573236889691,
      "grad_norm": 0.07672570583304174,
      "learning_rate": 0.00017953019231042766,
      "loss": 0.85,
      "step": 63860
    },
    {
      "epoch": 1.1213328885689706,
      "grad_norm": 0.06539542807973796,
      "learning_rate": 0.0001795234062702842,
      "loss": 0.8528,
      "step": 63870
    },
    {
      "epoch": 1.121508453448972,
      "grad_norm": 0.0462204765271002,
      "learning_rate": 0.00017951661923523406,
      "loss": 0.8484,
      "step": 63880
    },
    {
      "epoch": 1.1216840183289736,
      "grad_norm": 0.08899826063453344,
      "learning_rate": 0.00017950983120536333,
      "loss": 0.8508,
      "step": 63890
    },
    {
      "epoch": 1.1218595832089748,
      "grad_norm": 0.06570779841109815,
      "learning_rate": 0.00017950304218075798,
      "loss": 0.8479,
      "step": 63900
    },
    {
      "epoch": 1.1220351480889763,
      "grad_norm": 0.10201169511920036,
      "learning_rate": 0.0001794962521615041,
      "loss": 0.8526,
      "step": 63910
    },
    {
      "epoch": 1.1222107129689778,
      "grad_norm": 0.06344461443317778,
      "learning_rate": 0.0001794894611476876,
      "loss": 0.8488,
      "step": 63920
    },
    {
      "epoch": 1.122386277848979,
      "grad_norm": 0.06433479980017592,
      "learning_rate": 0.00017948266913939462,
      "loss": 0.8408,
      "step": 63930
    },
    {
      "epoch": 1.1225618427289805,
      "grad_norm": 0.07547090931739021,
      "learning_rate": 0.00017947587613671118,
      "loss": 0.852,
      "step": 63940
    },
    {
      "epoch": 1.122737407608982,
      "grad_norm": 0.06366747201900605,
      "learning_rate": 0.0001794690821397233,
      "loss": 0.8463,
      "step": 63950
    },
    {
      "epoch": 1.1229129724889833,
      "grad_norm": 0.05412937108067666,
      "learning_rate": 0.00017946228714851717,
      "loss": 0.8578,
      "step": 63960
    },
    {
      "epoch": 1.1230885373689847,
      "grad_norm": 0.05537034430084289,
      "learning_rate": 0.00017945549116317878,
      "loss": 0.8532,
      "step": 63970
    },
    {
      "epoch": 1.1232641022489862,
      "grad_norm": 0.08924424820753646,
      "learning_rate": 0.00017944869418379432,
      "loss": 0.848,
      "step": 63980
    },
    {
      "epoch": 1.1234396671289875,
      "grad_norm": 0.06083656503513563,
      "learning_rate": 0.00017944189621044984,
      "loss": 0.8476,
      "step": 63990
    },
    {
      "epoch": 1.123615232008989,
      "grad_norm": 0.057553205145431025,
      "learning_rate": 0.00017943509724323154,
      "loss": 0.8475,
      "step": 64000
    },
    {
      "epoch": 1.1237907968889904,
      "grad_norm": 0.05636739618677118,
      "learning_rate": 0.00017942829728222548,
      "loss": 0.8485,
      "step": 64010
    },
    {
      "epoch": 1.1239663617689917,
      "grad_norm": 0.04436540993238254,
      "learning_rate": 0.00017942149632751788,
      "loss": 0.8522,
      "step": 64020
    },
    {
      "epoch": 1.1241419266489932,
      "grad_norm": 0.06749653204007962,
      "learning_rate": 0.0001794146943791949,
      "loss": 0.8494,
      "step": 64030
    },
    {
      "epoch": 1.1243174915289946,
      "grad_norm": 0.059252310061947706,
      "learning_rate": 0.00017940789143734272,
      "loss": 0.8588,
      "step": 64040
    },
    {
      "epoch": 1.1244930564089959,
      "grad_norm": 0.0754921639203431,
      "learning_rate": 0.0001794010875020475,
      "loss": 0.8563,
      "step": 64050
    },
    {
      "epoch": 1.1246686212889974,
      "grad_norm": 0.07454543476261988,
      "learning_rate": 0.00017939428257339552,
      "loss": 0.8518,
      "step": 64060
    },
    {
      "epoch": 1.1248441861689988,
      "grad_norm": 0.06136797736512024,
      "learning_rate": 0.0001793874766514729,
      "loss": 0.8449,
      "step": 64070
    },
    {
      "epoch": 1.125019751049,
      "grad_norm": 0.05524850206837931,
      "learning_rate": 0.00017938066973636596,
      "loss": 0.8544,
      "step": 64080
    },
    {
      "epoch": 1.1251953159290016,
      "grad_norm": 0.07724745197306429,
      "learning_rate": 0.0001793738618281609,
      "loss": 0.85,
      "step": 64090
    },
    {
      "epoch": 1.125370880809003,
      "grad_norm": 0.07802334177946706,
      "learning_rate": 0.00017936705292694402,
      "loss": 0.8531,
      "step": 64100
    },
    {
      "epoch": 1.1255464456890043,
      "grad_norm": 0.0980805322732231,
      "learning_rate": 0.00017936024303280156,
      "loss": 0.8517,
      "step": 64110
    },
    {
      "epoch": 1.1257220105690058,
      "grad_norm": 0.06597614040998895,
      "learning_rate": 0.00017935343214581974,
      "loss": 0.8531,
      "step": 64120
    },
    {
      "epoch": 1.1258975754490073,
      "grad_norm": 0.07848159276408537,
      "learning_rate": 0.00017934662026608498,
      "loss": 0.8491,
      "step": 64130
    },
    {
      "epoch": 1.1260731403290085,
      "grad_norm": 0.0934104580981867,
      "learning_rate": 0.0001793398073936835,
      "loss": 0.85,
      "step": 64140
    },
    {
      "epoch": 1.12624870520901,
      "grad_norm": 0.05351224582750287,
      "learning_rate": 0.00017933299352870165,
      "loss": 0.8526,
      "step": 64150
    },
    {
      "epoch": 1.1264242700890115,
      "grad_norm": 0.07484598344893574,
      "learning_rate": 0.00017932617867122576,
      "loss": 0.855,
      "step": 64160
    },
    {
      "epoch": 1.1265998349690127,
      "grad_norm": 0.06095429146051287,
      "learning_rate": 0.0001793193628213422,
      "loss": 0.8511,
      "step": 64170
    },
    {
      "epoch": 1.1267753998490142,
      "grad_norm": 0.0814697978803053,
      "learning_rate": 0.00017931254597913727,
      "loss": 0.8522,
      "step": 64180
    },
    {
      "epoch": 1.1269509647290157,
      "grad_norm": 0.04642813783533168,
      "learning_rate": 0.00017930572814469736,
      "loss": 0.8486,
      "step": 64190
    },
    {
      "epoch": 1.127126529609017,
      "grad_norm": 0.07575311057387364,
      "learning_rate": 0.00017929890931810892,
      "loss": 0.8459,
      "step": 64200
    },
    {
      "epoch": 1.1273020944890184,
      "grad_norm": 0.058050937504048716,
      "learning_rate": 0.00017929208949945825,
      "loss": 0.8552,
      "step": 64210
    },
    {
      "epoch": 1.12747765936902,
      "grad_norm": 0.06337028818505674,
      "learning_rate": 0.00017928526868883185,
      "loss": 0.8488,
      "step": 64220
    },
    {
      "epoch": 1.1276532242490211,
      "grad_norm": 0.055030021595843075,
      "learning_rate": 0.00017927844688631604,
      "loss": 0.8491,
      "step": 64230
    },
    {
      "epoch": 1.1278287891290226,
      "grad_norm": 0.06613403585690923,
      "learning_rate": 0.00017927162409199734,
      "loss": 0.8589,
      "step": 64240
    },
    {
      "epoch": 1.128004354009024,
      "grad_norm": 0.07164951300964377,
      "learning_rate": 0.00017926480030596212,
      "loss": 0.8481,
      "step": 64250
    },
    {
      "epoch": 1.1281799188890254,
      "grad_norm": 0.07371632579140651,
      "learning_rate": 0.00017925797552829692,
      "loss": 0.8466,
      "step": 64260
    },
    {
      "epoch": 1.1283554837690268,
      "grad_norm": 0.08924661464969004,
      "learning_rate": 0.00017925114975908817,
      "loss": 0.8575,
      "step": 64270
    },
    {
      "epoch": 1.1285310486490283,
      "grad_norm": 0.09443891100734772,
      "learning_rate": 0.00017924432299842234,
      "loss": 0.854,
      "step": 64280
    },
    {
      "epoch": 1.1287066135290296,
      "grad_norm": 0.06742019625699383,
      "learning_rate": 0.00017923749524638593,
      "loss": 0.8515,
      "step": 64290
    },
    {
      "epoch": 1.128882178409031,
      "grad_norm": 0.06685700056430416,
      "learning_rate": 0.0001792306665030655,
      "loss": 0.8486,
      "step": 64300
    },
    {
      "epoch": 1.1290577432890325,
      "grad_norm": 0.0866266148296227,
      "learning_rate": 0.00017922383676854752,
      "loss": 0.8503,
      "step": 64310
    },
    {
      "epoch": 1.1292333081690338,
      "grad_norm": 0.06283351195830242,
      "learning_rate": 0.00017921700604291851,
      "loss": 0.852,
      "step": 64320
    },
    {
      "epoch": 1.1294088730490353,
      "grad_norm": 0.06361790580770263,
      "learning_rate": 0.00017921017432626508,
      "loss": 0.8451,
      "step": 64330
    },
    {
      "epoch": 1.1295844379290367,
      "grad_norm": 0.08370812389383761,
      "learning_rate": 0.0001792033416186737,
      "loss": 0.8428,
      "step": 64340
    },
    {
      "epoch": 1.129760002809038,
      "grad_norm": 0.04450841570506968,
      "learning_rate": 0.00017919650792023103,
      "loss": 0.8452,
      "step": 64350
    },
    {
      "epoch": 1.1299355676890395,
      "grad_norm": 0.07117174081116262,
      "learning_rate": 0.00017918967323102362,
      "loss": 0.853,
      "step": 64360
    },
    {
      "epoch": 1.130111132569041,
      "grad_norm": 0.060915954369232694,
      "learning_rate": 0.00017918283755113807,
      "loss": 0.851,
      "step": 64370
    },
    {
      "epoch": 1.1302866974490422,
      "grad_norm": 0.06518989406944396,
      "learning_rate": 0.000179176000880661,
      "loss": 0.8575,
      "step": 64380
    },
    {
      "epoch": 1.1304622623290437,
      "grad_norm": 0.07751870600300023,
      "learning_rate": 0.00017916916321967898,
      "loss": 0.854,
      "step": 64390
    },
    {
      "epoch": 1.1306378272090452,
      "grad_norm": 0.06345096356005113,
      "learning_rate": 0.0001791623245682787,
      "loss": 0.8577,
      "step": 64400
    },
    {
      "epoch": 1.1308133920890464,
      "grad_norm": 0.0689639174664518,
      "learning_rate": 0.0001791554849265468,
      "loss": 0.8507,
      "step": 64410
    },
    {
      "epoch": 1.1309889569690479,
      "grad_norm": 0.062027334642485284,
      "learning_rate": 0.00017914864429456988,
      "loss": 0.8569,
      "step": 64420
    },
    {
      "epoch": 1.1311645218490494,
      "grad_norm": 0.06967239199697572,
      "learning_rate": 0.0001791418026724347,
      "loss": 0.8493,
      "step": 64430
    },
    {
      "epoch": 1.1313400867290508,
      "grad_norm": 0.05711079731577256,
      "learning_rate": 0.0001791349600602279,
      "loss": 0.8437,
      "step": 64440
    },
    {
      "epoch": 1.131515651609052,
      "grad_norm": 0.07575012515068578,
      "learning_rate": 0.00017912811645803616,
      "loss": 0.8531,
      "step": 64450
    },
    {
      "epoch": 1.1316912164890536,
      "grad_norm": 0.07773840431068837,
      "learning_rate": 0.00017912127186594623,
      "loss": 0.8476,
      "step": 64460
    },
    {
      "epoch": 1.1318667813690548,
      "grad_norm": 0.12137045295994249,
      "learning_rate": 0.0001791144262840448,
      "loss": 0.8472,
      "step": 64470
    },
    {
      "epoch": 1.1320423462490563,
      "grad_norm": 0.08501963322417361,
      "learning_rate": 0.00017910757971241863,
      "loss": 0.8498,
      "step": 64480
    },
    {
      "epoch": 1.1322179111290578,
      "grad_norm": 0.08537962872228547,
      "learning_rate": 0.00017910073215115447,
      "loss": 0.8582,
      "step": 64490
    },
    {
      "epoch": 1.1323934760090593,
      "grad_norm": 0.04804774428462465,
      "learning_rate": 0.00017909388360033904,
      "loss": 0.8445,
      "step": 64500
    },
    {
      "epoch": 1.1325690408890605,
      "grad_norm": 0.06128329390485867,
      "learning_rate": 0.00017908703406005916,
      "loss": 0.8555,
      "step": 64510
    },
    {
      "epoch": 1.132744605769062,
      "grad_norm": 0.06303780461330723,
      "learning_rate": 0.0001790801835304016,
      "loss": 0.8609,
      "step": 64520
    },
    {
      "epoch": 1.1329201706490633,
      "grad_norm": 0.049871903247899224,
      "learning_rate": 0.00017907333201145308,
      "loss": 0.8476,
      "step": 64530
    },
    {
      "epoch": 1.1330957355290647,
      "grad_norm": 0.0509589994833353,
      "learning_rate": 0.00017906647950330053,
      "loss": 0.8551,
      "step": 64540
    },
    {
      "epoch": 1.1332713004090662,
      "grad_norm": 0.0702463784231623,
      "learning_rate": 0.00017905962600603078,
      "loss": 0.8466,
      "step": 64550
    },
    {
      "epoch": 1.1334468652890677,
      "grad_norm": 0.055926003659947626,
      "learning_rate": 0.0001790527715197305,
      "loss": 0.8453,
      "step": 64560
    },
    {
      "epoch": 1.133622430169069,
      "grad_norm": 0.05750843439463261,
      "learning_rate": 0.00017904591604448673,
      "loss": 0.8397,
      "step": 64570
    },
    {
      "epoch": 1.1337979950490704,
      "grad_norm": 0.056304781294947766,
      "learning_rate": 0.00017903905958038623,
      "loss": 0.8463,
      "step": 64580
    },
    {
      "epoch": 1.1339735599290717,
      "grad_norm": 0.07655366319004069,
      "learning_rate": 0.00017903220212751585,
      "loss": 0.8538,
      "step": 64590
    },
    {
      "epoch": 1.1341491248090732,
      "grad_norm": 0.056090017525195233,
      "learning_rate": 0.00017902534368596256,
      "loss": 0.8491,
      "step": 64600
    },
    {
      "epoch": 1.1343246896890746,
      "grad_norm": 0.05743415201211292,
      "learning_rate": 0.00017901848425581318,
      "loss": 0.8542,
      "step": 64610
    },
    {
      "epoch": 1.134500254569076,
      "grad_norm": 0.11929489585887838,
      "learning_rate": 0.00017901162383715467,
      "loss": 0.8505,
      "step": 64620
    },
    {
      "epoch": 1.1346758194490774,
      "grad_norm": 0.05365608373816396,
      "learning_rate": 0.00017900476243007393,
      "loss": 0.8497,
      "step": 64630
    },
    {
      "epoch": 1.1348513843290788,
      "grad_norm": 0.07708245504602781,
      "learning_rate": 0.00017899790003465788,
      "loss": 0.8491,
      "step": 64640
    },
    {
      "epoch": 1.13502694920908,
      "grad_norm": 0.05825661688318796,
      "learning_rate": 0.0001789910366509935,
      "loss": 0.8528,
      "step": 64650
    },
    {
      "epoch": 1.1352025140890816,
      "grad_norm": 0.07269608965664046,
      "learning_rate": 0.00017898417227916777,
      "loss": 0.8488,
      "step": 64660
    },
    {
      "epoch": 1.135378078969083,
      "grad_norm": 0.09014235189480956,
      "learning_rate": 0.0001789773069192676,
      "loss": 0.8472,
      "step": 64670
    },
    {
      "epoch": 1.1355536438490845,
      "grad_norm": 0.06881296719177808,
      "learning_rate": 0.00017897044057138003,
      "loss": 0.8503,
      "step": 64680
    },
    {
      "epoch": 1.1357292087290858,
      "grad_norm": 0.061919624039048694,
      "learning_rate": 0.000178963573235592,
      "loss": 0.858,
      "step": 64690
    },
    {
      "epoch": 1.1359047736090873,
      "grad_norm": 0.06065188644466135,
      "learning_rate": 0.0001789567049119906,
      "loss": 0.8448,
      "step": 64700
    },
    {
      "epoch": 1.1360803384890887,
      "grad_norm": 0.05618224909674693,
      "learning_rate": 0.00017894983560066277,
      "loss": 0.8551,
      "step": 64710
    },
    {
      "epoch": 1.13625590336909,
      "grad_norm": 0.0549106281157536,
      "learning_rate": 0.00017894296530169561,
      "loss": 0.8525,
      "step": 64720
    },
    {
      "epoch": 1.1364314682490915,
      "grad_norm": 0.04377383434017827,
      "learning_rate": 0.0001789360940151761,
      "loss": 0.852,
      "step": 64730
    },
    {
      "epoch": 1.136607033129093,
      "grad_norm": 0.04840613942029026,
      "learning_rate": 0.0001789292217411914,
      "loss": 0.8458,
      "step": 64740
    },
    {
      "epoch": 1.1367825980090942,
      "grad_norm": 0.09741409644743666,
      "learning_rate": 0.0001789223484798285,
      "loss": 0.8474,
      "step": 64750
    },
    {
      "epoch": 1.1369581628890957,
      "grad_norm": 0.0688803534816602,
      "learning_rate": 0.0001789154742311745,
      "loss": 0.8556,
      "step": 64760
    },
    {
      "epoch": 1.1371337277690972,
      "grad_norm": 0.06050206656579917,
      "learning_rate": 0.0001789085989953165,
      "loss": 0.8477,
      "step": 64770
    },
    {
      "epoch": 1.1373092926490984,
      "grad_norm": 0.06671176817032749,
      "learning_rate": 0.00017890172277234168,
      "loss": 0.8502,
      "step": 64780
    },
    {
      "epoch": 1.1374848575291,
      "grad_norm": 0.06443575927909653,
      "learning_rate": 0.00017889484556233705,
      "loss": 0.8599,
      "step": 64790
    },
    {
      "epoch": 1.1376604224091014,
      "grad_norm": 0.06873151186502408,
      "learning_rate": 0.0001788879673653898,
      "loss": 0.8426,
      "step": 64800
    },
    {
      "epoch": 1.1378359872891026,
      "grad_norm": 0.056816200813914045,
      "learning_rate": 0.0001788810881815871,
      "loss": 0.8533,
      "step": 64810
    },
    {
      "epoch": 1.138011552169104,
      "grad_norm": 0.06692795526328259,
      "learning_rate": 0.00017887420801101605,
      "loss": 0.8553,
      "step": 64820
    },
    {
      "epoch": 1.1381871170491056,
      "grad_norm": 0.07825063954329145,
      "learning_rate": 0.0001788673268537639,
      "loss": 0.8442,
      "step": 64830
    },
    {
      "epoch": 1.1383626819291068,
      "grad_norm": 0.08974884392498862,
      "learning_rate": 0.0001788604447099178,
      "loss": 0.8529,
      "step": 64840
    },
    {
      "epoch": 1.1385382468091083,
      "grad_norm": 0.05555451662457341,
      "learning_rate": 0.00017885356157956486,
      "loss": 0.8488,
      "step": 64850
    },
    {
      "epoch": 1.1387138116891098,
      "grad_norm": 0.07419935568109584,
      "learning_rate": 0.00017884667746279246,
      "loss": 0.8594,
      "step": 64860
    },
    {
      "epoch": 1.138889376569111,
      "grad_norm": 0.07241989876288825,
      "learning_rate": 0.0001788397923596877,
      "loss": 0.8414,
      "step": 64870
    },
    {
      "epoch": 1.1390649414491125,
      "grad_norm": 0.058514110104034324,
      "learning_rate": 0.00017883290627033788,
      "loss": 0.8528,
      "step": 64880
    },
    {
      "epoch": 1.139240506329114,
      "grad_norm": 0.04853061063301369,
      "learning_rate": 0.00017882601919483023,
      "loss": 0.8532,
      "step": 64890
    },
    {
      "epoch": 1.1394160712091153,
      "grad_norm": 0.08040436128898949,
      "learning_rate": 0.00017881913113325195,
      "loss": 0.8543,
      "step": 64900
    },
    {
      "epoch": 1.1395916360891167,
      "grad_norm": 0.09264796312076681,
      "learning_rate": 0.0001788122420856904,
      "loss": 0.8516,
      "step": 64910
    },
    {
      "epoch": 1.1397672009691182,
      "grad_norm": 0.059667100313423975,
      "learning_rate": 0.0001788053520522328,
      "loss": 0.847,
      "step": 64920
    },
    {
      "epoch": 1.1399427658491195,
      "grad_norm": 0.07524740511393918,
      "learning_rate": 0.0001787984610329665,
      "loss": 0.8503,
      "step": 64930
    },
    {
      "epoch": 1.140118330729121,
      "grad_norm": 0.06625663446651378,
      "learning_rate": 0.0001787915690279788,
      "loss": 0.8522,
      "step": 64940
    },
    {
      "epoch": 1.1402938956091224,
      "grad_norm": 0.061417080641106424,
      "learning_rate": 0.00017878467603735704,
      "loss": 0.8465,
      "step": 64950
    },
    {
      "epoch": 1.1404694604891237,
      "grad_norm": 0.06341511819080986,
      "learning_rate": 0.00017877778206118846,
      "loss": 0.8546,
      "step": 64960
    },
    {
      "epoch": 1.1406450253691252,
      "grad_norm": 0.049282135584748935,
      "learning_rate": 0.0001787708870995605,
      "loss": 0.85,
      "step": 64970
    },
    {
      "epoch": 1.1408205902491266,
      "grad_norm": 0.051013016729853325,
      "learning_rate": 0.0001787639911525605,
      "loss": 0.8461,
      "step": 64980
    },
    {
      "epoch": 1.1409961551291279,
      "grad_norm": 0.07233528856099104,
      "learning_rate": 0.00017875709422027585,
      "loss": 0.8549,
      "step": 64990
    },
    {
      "epoch": 1.1411717200091294,
      "grad_norm": 0.059517333281751836,
      "learning_rate": 0.00017875019630279385,
      "loss": 0.8575,
      "step": 65000
    },
    {
      "epoch": 1.1413472848891308,
      "grad_norm": 0.06874750970607965,
      "learning_rate": 0.00017874329740020205,
      "loss": 0.8455,
      "step": 65010
    },
    {
      "epoch": 1.141522849769132,
      "grad_norm": 0.07630248318196081,
      "learning_rate": 0.0001787363975125877,
      "loss": 0.8477,
      "step": 65020
    },
    {
      "epoch": 1.1416984146491336,
      "grad_norm": 0.04650036900284374,
      "learning_rate": 0.00017872949664003834,
      "loss": 0.8517,
      "step": 65030
    },
    {
      "epoch": 1.141873979529135,
      "grad_norm": 0.05443273047967385,
      "learning_rate": 0.0001787225947826413,
      "loss": 0.8459,
      "step": 65040
    },
    {
      "epoch": 1.1420495444091363,
      "grad_norm": 0.06422070023663035,
      "learning_rate": 0.00017871569194048416,
      "loss": 0.8517,
      "step": 65050
    },
    {
      "epoch": 1.1422251092891378,
      "grad_norm": 0.06780424140532242,
      "learning_rate": 0.00017870878811365428,
      "loss": 0.846,
      "step": 65060
    },
    {
      "epoch": 1.1424006741691393,
      "grad_norm": 0.07474889504490376,
      "learning_rate": 0.00017870188330223915,
      "loss": 0.8465,
      "step": 65070
    },
    {
      "epoch": 1.1425762390491405,
      "grad_norm": 0.05983562406483036,
      "learning_rate": 0.00017869497750632628,
      "loss": 0.8434,
      "step": 65080
    },
    {
      "epoch": 1.142751803929142,
      "grad_norm": 0.08961135539758446,
      "learning_rate": 0.0001786880707260032,
      "loss": 0.8513,
      "step": 65090
    },
    {
      "epoch": 1.1429273688091435,
      "grad_norm": 0.056932888646378466,
      "learning_rate": 0.00017868116296135728,
      "loss": 0.8506,
      "step": 65100
    },
    {
      "epoch": 1.1431029336891447,
      "grad_norm": 0.0657824312930575,
      "learning_rate": 0.00017867425421247621,
      "loss": 0.8452,
      "step": 65110
    },
    {
      "epoch": 1.1432784985691462,
      "grad_norm": 0.07962070235759343,
      "learning_rate": 0.0001786673444794474,
      "loss": 0.8508,
      "step": 65120
    },
    {
      "epoch": 1.1434540634491477,
      "grad_norm": 0.0586930425582206,
      "learning_rate": 0.0001786604337623585,
      "loss": 0.8583,
      "step": 65130
    },
    {
      "epoch": 1.143629628329149,
      "grad_norm": 0.08401509934169699,
      "learning_rate": 0.00017865352206129702,
      "loss": 0.8522,
      "step": 65140
    },
    {
      "epoch": 1.1438051932091504,
      "grad_norm": 0.0521168295431265,
      "learning_rate": 0.00017864660937635048,
      "loss": 0.8492,
      "step": 65150
    },
    {
      "epoch": 1.143980758089152,
      "grad_norm": 0.07806806849333076,
      "learning_rate": 0.00017863969570760654,
      "loss": 0.8542,
      "step": 65160
    },
    {
      "epoch": 1.1441563229691534,
      "grad_norm": 0.042353619799965166,
      "learning_rate": 0.0001786327810551528,
      "loss": 0.848,
      "step": 65170
    },
    {
      "epoch": 1.1443318878491546,
      "grad_norm": 0.07194816461530554,
      "learning_rate": 0.00017862586541907686,
      "loss": 0.8466,
      "step": 65180
    },
    {
      "epoch": 1.144507452729156,
      "grad_norm": 0.048046868198485825,
      "learning_rate": 0.0001786189487994663,
      "loss": 0.8564,
      "step": 65190
    },
    {
      "epoch": 1.1446830176091574,
      "grad_norm": 0.07978523613617172,
      "learning_rate": 0.00017861203119640882,
      "loss": 0.8541,
      "step": 65200
    },
    {
      "epoch": 1.1448585824891588,
      "grad_norm": 0.05346320159863859,
      "learning_rate": 0.00017860511260999194,
      "loss": 0.8486,
      "step": 65210
    },
    {
      "epoch": 1.1450341473691603,
      "grad_norm": 0.05782610333681152,
      "learning_rate": 0.00017859819304030347,
      "loss": 0.8474,
      "step": 65220
    },
    {
      "epoch": 1.1452097122491618,
      "grad_norm": 0.06660565521608956,
      "learning_rate": 0.00017859127248743103,
      "loss": 0.8477,
      "step": 65230
    },
    {
      "epoch": 1.145385277129163,
      "grad_norm": 0.06909388831509862,
      "learning_rate": 0.00017858435095146228,
      "loss": 0.8553,
      "step": 65240
    },
    {
      "epoch": 1.1455608420091645,
      "grad_norm": 0.056537693938721954,
      "learning_rate": 0.00017857742843248496,
      "loss": 0.8478,
      "step": 65250
    },
    {
      "epoch": 1.1457364068891658,
      "grad_norm": 0.046399326564750036,
      "learning_rate": 0.00017857050493058676,
      "loss": 0.8464,
      "step": 65260
    },
    {
      "epoch": 1.1459119717691673,
      "grad_norm": 0.06022187546369767,
      "learning_rate": 0.00017856358044585535,
      "loss": 0.8469,
      "step": 65270
    },
    {
      "epoch": 1.1460875366491687,
      "grad_norm": 0.05087639665262646,
      "learning_rate": 0.0001785566549783785,
      "loss": 0.8497,
      "step": 65280
    },
    {
      "epoch": 1.1462631015291702,
      "grad_norm": 0.059465141112899716,
      "learning_rate": 0.00017854972852824404,
      "loss": 0.8531,
      "step": 65290
    },
    {
      "epoch": 1.1464386664091715,
      "grad_norm": 0.0652573809714605,
      "learning_rate": 0.0001785428010955396,
      "loss": 0.8553,
      "step": 65300
    },
    {
      "epoch": 1.146614231289173,
      "grad_norm": 0.061393227535730246,
      "learning_rate": 0.00017853587268035304,
      "loss": 0.8566,
      "step": 65310
    },
    {
      "epoch": 1.1467897961691742,
      "grad_norm": 0.05826298473634454,
      "learning_rate": 0.00017852894328277208,
      "loss": 0.8496,
      "step": 65320
    },
    {
      "epoch": 1.1469653610491757,
      "grad_norm": 0.09054602092904923,
      "learning_rate": 0.0001785220129028846,
      "loss": 0.85,
      "step": 65330
    },
    {
      "epoch": 1.1471409259291772,
      "grad_norm": 0.04859043337884091,
      "learning_rate": 0.0001785150815407783,
      "loss": 0.8571,
      "step": 65340
    },
    {
      "epoch": 1.1473164908091786,
      "grad_norm": 0.08960439376850579,
      "learning_rate": 0.0001785081491965411,
      "loss": 0.8547,
      "step": 65350
    },
    {
      "epoch": 1.14749205568918,
      "grad_norm": 0.06025314241934363,
      "learning_rate": 0.00017850121587026076,
      "loss": 0.8527,
      "step": 65360
    },
    {
      "epoch": 1.1476676205691814,
      "grad_norm": 0.05595051232837018,
      "learning_rate": 0.0001784942815620252,
      "loss": 0.8517,
      "step": 65370
    },
    {
      "epoch": 1.1478431854491826,
      "grad_norm": 0.06191843871898107,
      "learning_rate": 0.00017848734627192223,
      "loss": 0.8514,
      "step": 65380
    },
    {
      "epoch": 1.148018750329184,
      "grad_norm": 0.06310461945938446,
      "learning_rate": 0.00017848041000003974,
      "loss": 0.8412,
      "step": 65390
    },
    {
      "epoch": 1.1481943152091856,
      "grad_norm": 0.053928681668453715,
      "learning_rate": 0.0001784734727464656,
      "loss": 0.8478,
      "step": 65400
    },
    {
      "epoch": 1.148369880089187,
      "grad_norm": 0.062439336317766186,
      "learning_rate": 0.00017846653451128772,
      "loss": 0.8537,
      "step": 65410
    },
    {
      "epoch": 1.1485454449691883,
      "grad_norm": 0.05598444988455744,
      "learning_rate": 0.00017845959529459398,
      "loss": 0.8475,
      "step": 65420
    },
    {
      "epoch": 1.1487210098491898,
      "grad_norm": 0.04602684452453043,
      "learning_rate": 0.00017845265509647233,
      "loss": 0.8487,
      "step": 65430
    },
    {
      "epoch": 1.1488965747291913,
      "grad_norm": 0.05518051956995834,
      "learning_rate": 0.0001784457139170107,
      "loss": 0.8559,
      "step": 65440
    },
    {
      "epoch": 1.1490721396091925,
      "grad_norm": 0.04828103501914329,
      "learning_rate": 0.00017843877175629703,
      "loss": 0.8491,
      "step": 65450
    },
    {
      "epoch": 1.149247704489194,
      "grad_norm": 0.053014525950073325,
      "learning_rate": 0.00017843182861441927,
      "loss": 0.846,
      "step": 65460
    },
    {
      "epoch": 1.1494232693691955,
      "grad_norm": 0.07219504654956571,
      "learning_rate": 0.00017842488449146545,
      "loss": 0.8502,
      "step": 65470
    },
    {
      "epoch": 1.1495988342491967,
      "grad_norm": 0.06352151320230581,
      "learning_rate": 0.00017841793938752344,
      "loss": 0.8492,
      "step": 65480
    },
    {
      "epoch": 1.1497743991291982,
      "grad_norm": 0.06248802473005766,
      "learning_rate": 0.00017841099330268134,
      "loss": 0.8513,
      "step": 65490
    },
    {
      "epoch": 1.1499499640091997,
      "grad_norm": 0.05080056984708842,
      "learning_rate": 0.0001784040462370271,
      "loss": 0.8516,
      "step": 65500
    },
    {
      "epoch": 1.150125528889201,
      "grad_norm": 0.06148073762187769,
      "learning_rate": 0.00017839709819064876,
      "loss": 0.8514,
      "step": 65510
    },
    {
      "epoch": 1.1503010937692024,
      "grad_norm": 0.10190667229300787,
      "learning_rate": 0.00017839014916363438,
      "loss": 0.8545,
      "step": 65520
    },
    {
      "epoch": 1.150476658649204,
      "grad_norm": 0.0594858067383562,
      "learning_rate": 0.00017838319915607195,
      "loss": 0.853,
      "step": 65530
    },
    {
      "epoch": 1.1506522235292052,
      "grad_norm": 0.08019502094252026,
      "learning_rate": 0.00017837624816804962,
      "loss": 0.8516,
      "step": 65540
    },
    {
      "epoch": 1.1508277884092066,
      "grad_norm": 0.053618058554678404,
      "learning_rate": 0.0001783692961996553,
      "loss": 0.8514,
      "step": 65550
    },
    {
      "epoch": 1.151003353289208,
      "grad_norm": 0.0822578792721255,
      "learning_rate": 0.0001783623432509772,
      "loss": 0.8556,
      "step": 65560
    },
    {
      "epoch": 1.1511789181692094,
      "grad_norm": 0.06767396284919129,
      "learning_rate": 0.00017835538932210341,
      "loss": 0.8487,
      "step": 65570
    },
    {
      "epoch": 1.1513544830492108,
      "grad_norm": 0.06978358682158159,
      "learning_rate": 0.000178348434413122,
      "loss": 0.8585,
      "step": 65580
    },
    {
      "epoch": 1.1515300479292123,
      "grad_norm": 0.047352717410492275,
      "learning_rate": 0.00017834147852412112,
      "loss": 0.856,
      "step": 65590
    },
    {
      "epoch": 1.1517056128092136,
      "grad_norm": 0.07250102885218605,
      "learning_rate": 0.00017833452165518884,
      "loss": 0.8493,
      "step": 65600
    },
    {
      "epoch": 1.151881177689215,
      "grad_norm": 0.047910922456476136,
      "learning_rate": 0.00017832756380641338,
      "loss": 0.8506,
      "step": 65610
    },
    {
      "epoch": 1.1520567425692165,
      "grad_norm": 0.0695945232496535,
      "learning_rate": 0.00017832060497788285,
      "loss": 0.8503,
      "step": 65620
    },
    {
      "epoch": 1.1522323074492178,
      "grad_norm": 0.1060707798693782,
      "learning_rate": 0.00017831364516968546,
      "loss": 0.8537,
      "step": 65630
    },
    {
      "epoch": 1.1524078723292193,
      "grad_norm": 0.07219199352828064,
      "learning_rate": 0.00017830668438190936,
      "loss": 0.8521,
      "step": 65640
    },
    {
      "epoch": 1.1525834372092207,
      "grad_norm": 0.09840239923815636,
      "learning_rate": 0.00017829972261464277,
      "loss": 0.8484,
      "step": 65650
    },
    {
      "epoch": 1.152759002089222,
      "grad_norm": 0.06936060986343986,
      "learning_rate": 0.00017829275986797383,
      "loss": 0.8454,
      "step": 65660
    },
    {
      "epoch": 1.1529345669692235,
      "grad_norm": 0.08478936690582792,
      "learning_rate": 0.00017828579614199083,
      "loss": 0.8451,
      "step": 65670
    },
    {
      "epoch": 1.153110131849225,
      "grad_norm": 0.07468908332686115,
      "learning_rate": 0.00017827883143678203,
      "loss": 0.8453,
      "step": 65680
    },
    {
      "epoch": 1.1532856967292262,
      "grad_norm": 0.04748669721207155,
      "learning_rate": 0.0001782718657524356,
      "loss": 0.8585,
      "step": 65690
    },
    {
      "epoch": 1.1534612616092277,
      "grad_norm": 0.050998189568138036,
      "learning_rate": 0.0001782648990890398,
      "loss": 0.8502,
      "step": 65700
    },
    {
      "epoch": 1.1536368264892292,
      "grad_norm": 0.049289143700438146,
      "learning_rate": 0.0001782579314466829,
      "loss": 0.8564,
      "step": 65710
    },
    {
      "epoch": 1.1538123913692304,
      "grad_norm": 0.05417514148235377,
      "learning_rate": 0.00017825096282545326,
      "loss": 0.8517,
      "step": 65720
    },
    {
      "epoch": 1.153987956249232,
      "grad_norm": 0.04615473636225306,
      "learning_rate": 0.0001782439932254391,
      "loss": 0.8402,
      "step": 65730
    },
    {
      "epoch": 1.1541635211292334,
      "grad_norm": 0.0691557497355119,
      "learning_rate": 0.0001782370226467287,
      "loss": 0.8465,
      "step": 65740
    },
    {
      "epoch": 1.1543390860092346,
      "grad_norm": 0.06496068070205695,
      "learning_rate": 0.00017823005108941043,
      "loss": 0.8564,
      "step": 65750
    },
    {
      "epoch": 1.154514650889236,
      "grad_norm": 0.05712487144101239,
      "learning_rate": 0.00017822307855357264,
      "loss": 0.8471,
      "step": 65760
    },
    {
      "epoch": 1.1546902157692376,
      "grad_norm": 0.06829034032628047,
      "learning_rate": 0.00017821610503930367,
      "loss": 0.8555,
      "step": 65770
    },
    {
      "epoch": 1.1548657806492388,
      "grad_norm": 0.055511230804936224,
      "learning_rate": 0.00017820913054669178,
      "loss": 0.8515,
      "step": 65780
    },
    {
      "epoch": 1.1550413455292403,
      "grad_norm": 0.050788449536395844,
      "learning_rate": 0.00017820215507582547,
      "loss": 0.8576,
      "step": 65790
    },
    {
      "epoch": 1.1552169104092418,
      "grad_norm": 0.0536047414480835,
      "learning_rate": 0.00017819517862679304,
      "loss": 0.8509,
      "step": 65800
    },
    {
      "epoch": 1.155392475289243,
      "grad_norm": 0.0709795705458288,
      "learning_rate": 0.00017818820119968289,
      "loss": 0.8533,
      "step": 65810
    },
    {
      "epoch": 1.1555680401692445,
      "grad_norm": 0.06281005076658071,
      "learning_rate": 0.00017818122279458348,
      "loss": 0.854,
      "step": 65820
    },
    {
      "epoch": 1.155743605049246,
      "grad_norm": 0.04555517320932309,
      "learning_rate": 0.00017817424341158312,
      "loss": 0.851,
      "step": 65830
    },
    {
      "epoch": 1.1559191699292473,
      "grad_norm": 0.0810383318268404,
      "learning_rate": 0.00017816726305077034,
      "loss": 0.8557,
      "step": 65840
    },
    {
      "epoch": 1.1560947348092487,
      "grad_norm": 0.08307556689303361,
      "learning_rate": 0.00017816028171223355,
      "loss": 0.84,
      "step": 65850
    },
    {
      "epoch": 1.1562702996892502,
      "grad_norm": 0.05552104203818568,
      "learning_rate": 0.00017815329939606118,
      "loss": 0.8517,
      "step": 65860
    },
    {
      "epoch": 1.1564458645692515,
      "grad_norm": 0.05480649367305328,
      "learning_rate": 0.00017814631610234174,
      "loss": 0.8509,
      "step": 65870
    },
    {
      "epoch": 1.156621429449253,
      "grad_norm": 0.05110836644689362,
      "learning_rate": 0.00017813933183116369,
      "loss": 0.8436,
      "step": 65880
    },
    {
      "epoch": 1.1567969943292544,
      "grad_norm": 0.08732043452682244,
      "learning_rate": 0.00017813234658261553,
      "loss": 0.8468,
      "step": 65890
    },
    {
      "epoch": 1.156972559209256,
      "grad_norm": 0.05578635056387005,
      "learning_rate": 0.00017812536035678573,
      "loss": 0.8477,
      "step": 65900
    },
    {
      "epoch": 1.1571481240892572,
      "grad_norm": 0.07419652654532265,
      "learning_rate": 0.00017811837315376289,
      "loss": 0.8459,
      "step": 65910
    },
    {
      "epoch": 1.1573236889692586,
      "grad_norm": 0.06657074637611524,
      "learning_rate": 0.0001781113849736354,
      "loss": 0.8524,
      "step": 65920
    },
    {
      "epoch": 1.15749925384926,
      "grad_norm": 0.05046268180693899,
      "learning_rate": 0.00017810439581649197,
      "loss": 0.8513,
      "step": 65930
    },
    {
      "epoch": 1.1576748187292614,
      "grad_norm": 0.05280512437828936,
      "learning_rate": 0.000178097405682421,
      "loss": 0.8467,
      "step": 65940
    },
    {
      "epoch": 1.1578503836092628,
      "grad_norm": 0.07973949213806192,
      "learning_rate": 0.00017809041457151118,
      "loss": 0.848,
      "step": 65950
    },
    {
      "epoch": 1.1580259484892643,
      "grad_norm": 0.05181168283969551,
      "learning_rate": 0.00017808342248385096,
      "loss": 0.8508,
      "step": 65960
    },
    {
      "epoch": 1.1582015133692656,
      "grad_norm": 0.058440511341605474,
      "learning_rate": 0.00017807642941952906,
      "loss": 0.8477,
      "step": 65970
    },
    {
      "epoch": 1.158377078249267,
      "grad_norm": 0.050259706533456225,
      "learning_rate": 0.00017806943537863404,
      "loss": 0.8508,
      "step": 65980
    },
    {
      "epoch": 1.1585526431292683,
      "grad_norm": 0.05099665276078354,
      "learning_rate": 0.00017806244036125442,
      "loss": 0.8544,
      "step": 65990
    },
    {
      "epoch": 1.1587282080092698,
      "grad_norm": 0.05699468543264203,
      "learning_rate": 0.000178055444367479,
      "loss": 0.8511,
      "step": 66000
    },
    {
      "epoch": 1.1589037728892713,
      "grad_norm": 0.0698138730295014,
      "learning_rate": 0.00017804844739739633,
      "loss": 0.8498,
      "step": 66010
    },
    {
      "epoch": 1.1590793377692727,
      "grad_norm": 0.05385141185150396,
      "learning_rate": 0.000178041449451095,
      "loss": 0.8435,
      "step": 66020
    },
    {
      "epoch": 1.159254902649274,
      "grad_norm": 0.07038408940865239,
      "learning_rate": 0.00017803445052866378,
      "loss": 0.8526,
      "step": 66030
    },
    {
      "epoch": 1.1594304675292755,
      "grad_norm": 0.08329939866489247,
      "learning_rate": 0.00017802745063019132,
      "loss": 0.8443,
      "step": 66040
    },
    {
      "epoch": 1.1596060324092767,
      "grad_norm": 0.058533176732972744,
      "learning_rate": 0.00017802044975576627,
      "loss": 0.8574,
      "step": 66050
    },
    {
      "epoch": 1.1597815972892782,
      "grad_norm": 0.07290921981180094,
      "learning_rate": 0.0001780134479054774,
      "loss": 0.8578,
      "step": 66060
    },
    {
      "epoch": 1.1599571621692797,
      "grad_norm": 0.05808508967732347,
      "learning_rate": 0.00017800644507941332,
      "loss": 0.848,
      "step": 66070
    },
    {
      "epoch": 1.1601327270492812,
      "grad_norm": 0.059870009436683705,
      "learning_rate": 0.00017799944127766288,
      "loss": 0.8526,
      "step": 66080
    },
    {
      "epoch": 1.1603082919292824,
      "grad_norm": 0.04698494323136607,
      "learning_rate": 0.00017799243650031476,
      "loss": 0.8501,
      "step": 66090
    },
    {
      "epoch": 1.160483856809284,
      "grad_norm": 0.05934899117823208,
      "learning_rate": 0.0001779854307474577,
      "loss": 0.8492,
      "step": 66100
    },
    {
      "epoch": 1.1606594216892852,
      "grad_norm": 0.07849186512066335,
      "learning_rate": 0.00017797842401918047,
      "loss": 0.841,
      "step": 66110
    },
    {
      "epoch": 1.1608349865692866,
      "grad_norm": 0.06258717135854691,
      "learning_rate": 0.00017797141631557188,
      "loss": 0.8546,
      "step": 66120
    },
    {
      "epoch": 1.161010551449288,
      "grad_norm": 0.06747446849566724,
      "learning_rate": 0.00017796440763672065,
      "loss": 0.8428,
      "step": 66130
    },
    {
      "epoch": 1.1611861163292896,
      "grad_norm": 0.051314013831164344,
      "learning_rate": 0.00017795739798271568,
      "loss": 0.8492,
      "step": 66140
    },
    {
      "epoch": 1.1613616812092908,
      "grad_norm": 0.058800644165707794,
      "learning_rate": 0.00017795038735364568,
      "loss": 0.8565,
      "step": 66150
    },
    {
      "epoch": 1.1615372460892923,
      "grad_norm": 0.058886693332433106,
      "learning_rate": 0.00017794337574959953,
      "loss": 0.8485,
      "step": 66160
    },
    {
      "epoch": 1.1617128109692938,
      "grad_norm": 0.07156118471696717,
      "learning_rate": 0.00017793636317066608,
      "loss": 0.8546,
      "step": 66170
    },
    {
      "epoch": 1.161888375849295,
      "grad_norm": 0.07887879684927326,
      "learning_rate": 0.00017792934961693418,
      "loss": 0.8449,
      "step": 66180
    },
    {
      "epoch": 1.1620639407292965,
      "grad_norm": 0.07085384397289655,
      "learning_rate": 0.0001779223350884927,
      "loss": 0.8529,
      "step": 66190
    },
    {
      "epoch": 1.162239505609298,
      "grad_norm": 0.06854932196505469,
      "learning_rate": 0.00017791531958543042,
      "loss": 0.8564,
      "step": 66200
    },
    {
      "epoch": 1.1624150704892993,
      "grad_norm": 0.054056520369489686,
      "learning_rate": 0.0001779083031078363,
      "loss": 0.8536,
      "step": 66210
    },
    {
      "epoch": 1.1625906353693007,
      "grad_norm": 0.08141182030914039,
      "learning_rate": 0.00017790128565579928,
      "loss": 0.8606,
      "step": 66220
    },
    {
      "epoch": 1.1627662002493022,
      "grad_norm": 0.08261423208465404,
      "learning_rate": 0.0001778942672294082,
      "loss": 0.8457,
      "step": 66230
    },
    {
      "epoch": 1.1629417651293035,
      "grad_norm": 0.0553588875672764,
      "learning_rate": 0.00017788724782875207,
      "loss": 0.8554,
      "step": 66240
    },
    {
      "epoch": 1.163117330009305,
      "grad_norm": 0.07902314068790685,
      "learning_rate": 0.00017788022745391974,
      "loss": 0.8574,
      "step": 66250
    },
    {
      "epoch": 1.1632928948893064,
      "grad_norm": 0.08599203113786288,
      "learning_rate": 0.0001778732061050002,
      "loss": 0.8519,
      "step": 66260
    },
    {
      "epoch": 1.1634684597693077,
      "grad_norm": 0.06829452050201011,
      "learning_rate": 0.0001778661837820824,
      "loss": 0.8512,
      "step": 66270
    },
    {
      "epoch": 1.1636440246493092,
      "grad_norm": 0.049757628693316056,
      "learning_rate": 0.00017785916048525534,
      "loss": 0.8506,
      "step": 66280
    },
    {
      "epoch": 1.1638195895293106,
      "grad_norm": 0.0623088587243796,
      "learning_rate": 0.00017785213621460794,
      "loss": 0.861,
      "step": 66290
    },
    {
      "epoch": 1.163995154409312,
      "grad_norm": 0.06929691263471661,
      "learning_rate": 0.00017784511097022928,
      "loss": 0.8415,
      "step": 66300
    },
    {
      "epoch": 1.1641707192893134,
      "grad_norm": 0.05268067304510227,
      "learning_rate": 0.00017783808475220833,
      "loss": 0.8448,
      "step": 66310
    },
    {
      "epoch": 1.1643462841693148,
      "grad_norm": 0.0522750847220847,
      "learning_rate": 0.00017783105756063414,
      "loss": 0.8472,
      "step": 66320
    },
    {
      "epoch": 1.164521849049316,
      "grad_norm": 0.05872613315890544,
      "learning_rate": 0.0001778240293955957,
      "loss": 0.8546,
      "step": 66330
    },
    {
      "epoch": 1.1646974139293176,
      "grad_norm": 0.07097497617153894,
      "learning_rate": 0.00017781700025718212,
      "loss": 0.8409,
      "step": 66340
    },
    {
      "epoch": 1.164872978809319,
      "grad_norm": 0.05617131045825325,
      "learning_rate": 0.00017780997014548239,
      "loss": 0.8528,
      "step": 66350
    },
    {
      "epoch": 1.1650485436893203,
      "grad_norm": 0.07083493870573336,
      "learning_rate": 0.00017780293906058563,
      "loss": 0.8517,
      "step": 66360
    },
    {
      "epoch": 1.1652241085693218,
      "grad_norm": 0.05346148718359194,
      "learning_rate": 0.0001777959070025809,
      "loss": 0.8524,
      "step": 66370
    },
    {
      "epoch": 1.1653996734493233,
      "grad_norm": 0.056057092996650654,
      "learning_rate": 0.00017778887397155733,
      "loss": 0.8517,
      "step": 66380
    },
    {
      "epoch": 1.1655752383293245,
      "grad_norm": 0.08400363844570798,
      "learning_rate": 0.000177781839967604,
      "loss": 0.8494,
      "step": 66390
    },
    {
      "epoch": 1.165750803209326,
      "grad_norm": 0.06467932096781472,
      "learning_rate": 0.00017777480499081002,
      "loss": 0.8523,
      "step": 66400
    },
    {
      "epoch": 1.1659263680893275,
      "grad_norm": 0.05664101239783248,
      "learning_rate": 0.00017776776904126462,
      "loss": 0.8458,
      "step": 66410
    },
    {
      "epoch": 1.1661019329693287,
      "grad_norm": 0.05131260601378193,
      "learning_rate": 0.00017776073211905682,
      "loss": 0.8483,
      "step": 66420
    },
    {
      "epoch": 1.1662774978493302,
      "grad_norm": 0.05501310571743309,
      "learning_rate": 0.0001777536942242758,
      "loss": 0.8517,
      "step": 66430
    },
    {
      "epoch": 1.1664530627293317,
      "grad_norm": 0.069582700508036,
      "learning_rate": 0.00017774665535701085,
      "loss": 0.8446,
      "step": 66440
    },
    {
      "epoch": 1.166628627609333,
      "grad_norm": 0.05685396782716212,
      "learning_rate": 0.00017773961551735098,
      "loss": 0.8402,
      "step": 66450
    },
    {
      "epoch": 1.1668041924893344,
      "grad_norm": 0.05284535576380621,
      "learning_rate": 0.00017773257470538552,
      "loss": 0.8564,
      "step": 66460
    },
    {
      "epoch": 1.166979757369336,
      "grad_norm": 0.061397146236154415,
      "learning_rate": 0.00017772553292120364,
      "loss": 0.854,
      "step": 66470
    },
    {
      "epoch": 1.1671553222493372,
      "grad_norm": 0.052443696613391946,
      "learning_rate": 0.00017771849016489454,
      "loss": 0.8501,
      "step": 66480
    },
    {
      "epoch": 1.1673308871293386,
      "grad_norm": 0.059688754470641175,
      "learning_rate": 0.00017771144643654747,
      "loss": 0.8445,
      "step": 66490
    },
    {
      "epoch": 1.1675064520093401,
      "grad_norm": 0.08220822033179061,
      "learning_rate": 0.00017770440173625165,
      "loss": 0.8458,
      "step": 66500
    },
    {
      "epoch": 1.1676820168893414,
      "grad_norm": 0.0519389050332011,
      "learning_rate": 0.00017769735606409638,
      "loss": 0.8458,
      "step": 66510
    },
    {
      "epoch": 1.1678575817693428,
      "grad_norm": 0.05648143183694809,
      "learning_rate": 0.00017769030942017092,
      "loss": 0.8464,
      "step": 66520
    },
    {
      "epoch": 1.1680331466493443,
      "grad_norm": 0.05325645438707554,
      "learning_rate": 0.0001776832618045645,
      "loss": 0.8505,
      "step": 66530
    },
    {
      "epoch": 1.1682087115293456,
      "grad_norm": 0.056350257247781615,
      "learning_rate": 0.00017767621321736652,
      "loss": 0.8452,
      "step": 66540
    },
    {
      "epoch": 1.168384276409347,
      "grad_norm": 0.05845730886864941,
      "learning_rate": 0.0001776691636586662,
      "loss": 0.8533,
      "step": 66550
    },
    {
      "epoch": 1.1685598412893485,
      "grad_norm": 0.060069070624068265,
      "learning_rate": 0.00017766211312855287,
      "loss": 0.8505,
      "step": 66560
    },
    {
      "epoch": 1.1687354061693498,
      "grad_norm": 0.06386736850337466,
      "learning_rate": 0.0001776550616271159,
      "loss": 0.848,
      "step": 66570
    },
    {
      "epoch": 1.1689109710493513,
      "grad_norm": 0.0720697295772668,
      "learning_rate": 0.00017764800915444457,
      "loss": 0.8531,
      "step": 66580
    },
    {
      "epoch": 1.1690865359293527,
      "grad_norm": 0.04392984298939825,
      "learning_rate": 0.00017764095571062827,
      "loss": 0.8564,
      "step": 66590
    },
    {
      "epoch": 1.169262100809354,
      "grad_norm": 0.05599968607601634,
      "learning_rate": 0.00017763390129575638,
      "loss": 0.8495,
      "step": 66600
    },
    {
      "epoch": 1.1694376656893555,
      "grad_norm": 0.0772421089477914,
      "learning_rate": 0.0001776268459099183,
      "loss": 0.8518,
      "step": 66610
    },
    {
      "epoch": 1.169613230569357,
      "grad_norm": 0.0629693221886483,
      "learning_rate": 0.0001776197895532034,
      "loss": 0.8516,
      "step": 66620
    },
    {
      "epoch": 1.1697887954493584,
      "grad_norm": 0.06585383553311876,
      "learning_rate": 0.00017761273222570106,
      "loss": 0.8506,
      "step": 66630
    },
    {
      "epoch": 1.1699643603293597,
      "grad_norm": 0.0710670454714907,
      "learning_rate": 0.00017760567392750075,
      "loss": 0.8467,
      "step": 66640
    },
    {
      "epoch": 1.1701399252093612,
      "grad_norm": 0.0638130130135592,
      "learning_rate": 0.0001775986146586918,
      "loss": 0.8528,
      "step": 66650
    },
    {
      "epoch": 1.1703154900893624,
      "grad_norm": 0.09690107760646788,
      "learning_rate": 0.0001775915544193638,
      "loss": 0.8562,
      "step": 66660
    },
    {
      "epoch": 1.170491054969364,
      "grad_norm": 0.05865057129415335,
      "learning_rate": 0.0001775844932096061,
      "loss": 0.8518,
      "step": 66670
    },
    {
      "epoch": 1.1706666198493654,
      "grad_norm": 0.04545844160917943,
      "learning_rate": 0.0001775774310295082,
      "loss": 0.8567,
      "step": 66680
    },
    {
      "epoch": 1.1708421847293669,
      "grad_norm": 0.05590429581796855,
      "learning_rate": 0.00017757036787915953,
      "loss": 0.8587,
      "step": 66690
    },
    {
      "epoch": 1.171017749609368,
      "grad_norm": 0.06109372377954633,
      "learning_rate": 0.00017756330375864966,
      "loss": 0.8435,
      "step": 66700
    },
    {
      "epoch": 1.1711933144893696,
      "grad_norm": 0.05507628020747419,
      "learning_rate": 0.00017755623866806805,
      "loss": 0.8513,
      "step": 66710
    },
    {
      "epoch": 1.1713688793693708,
      "grad_norm": 0.05216008454888361,
      "learning_rate": 0.00017754917260750423,
      "loss": 0.8536,
      "step": 66720
    },
    {
      "epoch": 1.1715444442493723,
      "grad_norm": 0.0562614898103783,
      "learning_rate": 0.00017754210557704774,
      "loss": 0.852,
      "step": 66730
    },
    {
      "epoch": 1.1717200091293738,
      "grad_norm": 0.0549727426136307,
      "learning_rate": 0.00017753503757678808,
      "loss": 0.8528,
      "step": 66740
    },
    {
      "epoch": 1.1718955740093753,
      "grad_norm": 0.0505680109617179,
      "learning_rate": 0.00017752796860681481,
      "loss": 0.8394,
      "step": 66750
    },
    {
      "epoch": 1.1720711388893765,
      "grad_norm": 0.0548838783419539,
      "learning_rate": 0.00017752089866721754,
      "loss": 0.8573,
      "step": 66760
    },
    {
      "epoch": 1.172246703769378,
      "grad_norm": 0.053355479053723814,
      "learning_rate": 0.00017751382775808584,
      "loss": 0.8556,
      "step": 66770
    },
    {
      "epoch": 1.1724222686493793,
      "grad_norm": 0.07008800785900905,
      "learning_rate": 0.00017750675587950926,
      "loss": 0.8492,
      "step": 66780
    },
    {
      "epoch": 1.1725978335293807,
      "grad_norm": 0.048221391784391415,
      "learning_rate": 0.00017749968303157739,
      "loss": 0.8544,
      "step": 66790
    },
    {
      "epoch": 1.1727733984093822,
      "grad_norm": 0.06091881945858441,
      "learning_rate": 0.00017749260921437989,
      "loss": 0.8494,
      "step": 66800
    },
    {
      "epoch": 1.1729489632893837,
      "grad_norm": 0.05747206685628641,
      "learning_rate": 0.00017748553442800636,
      "loss": 0.8454,
      "step": 66810
    },
    {
      "epoch": 1.173124528169385,
      "grad_norm": 0.045378190164994954,
      "learning_rate": 0.00017747845867254646,
      "loss": 0.8483,
      "step": 66820
    },
    {
      "epoch": 1.1733000930493864,
      "grad_norm": 0.06517101850911525,
      "learning_rate": 0.00017747138194808986,
      "loss": 0.8499,
      "step": 66830
    },
    {
      "epoch": 1.1734756579293877,
      "grad_norm": 0.05041206547529954,
      "learning_rate": 0.00017746430425472618,
      "loss": 0.8582,
      "step": 66840
    },
    {
      "epoch": 1.1736512228093892,
      "grad_norm": 0.05722215125248116,
      "learning_rate": 0.00017745722559254508,
      "loss": 0.8435,
      "step": 66850
    },
    {
      "epoch": 1.1738267876893906,
      "grad_norm": 0.0733420529107375,
      "learning_rate": 0.00017745014596163628,
      "loss": 0.8459,
      "step": 66860
    },
    {
      "epoch": 1.1740023525693921,
      "grad_norm": 0.06847855518168443,
      "learning_rate": 0.00017744306536208952,
      "loss": 0.8486,
      "step": 66870
    },
    {
      "epoch": 1.1741779174493934,
      "grad_norm": 0.07691223758555013,
      "learning_rate": 0.00017743598379399447,
      "loss": 0.8544,
      "step": 66880
    },
    {
      "epoch": 1.1743534823293948,
      "grad_norm": 0.05823793730715337,
      "learning_rate": 0.00017742890125744084,
      "loss": 0.8554,
      "step": 66890
    },
    {
      "epoch": 1.1745290472093963,
      "grad_norm": 0.048486578242918715,
      "learning_rate": 0.00017742181775251838,
      "loss": 0.8518,
      "step": 66900
    },
    {
      "epoch": 1.1747046120893976,
      "grad_norm": 0.057293387698345175,
      "learning_rate": 0.00017741473327931686,
      "loss": 0.8545,
      "step": 66910
    },
    {
      "epoch": 1.174880176969399,
      "grad_norm": 0.07070590777651231,
      "learning_rate": 0.000177407647837926,
      "loss": 0.8462,
      "step": 66920
    },
    {
      "epoch": 1.1750557418494005,
      "grad_norm": 0.06005307381297477,
      "learning_rate": 0.0001774005614284356,
      "loss": 0.8552,
      "step": 66930
    },
    {
      "epoch": 1.1752313067294018,
      "grad_norm": 0.07663227961891576,
      "learning_rate": 0.00017739347405093547,
      "loss": 0.843,
      "step": 66940
    },
    {
      "epoch": 1.1754068716094033,
      "grad_norm": 0.05906012681473801,
      "learning_rate": 0.00017738638570551532,
      "loss": 0.8481,
      "step": 66950
    },
    {
      "epoch": 1.1755824364894047,
      "grad_norm": 0.05788251094674904,
      "learning_rate": 0.00017737929639226508,
      "loss": 0.8465,
      "step": 66960
    },
    {
      "epoch": 1.175758001369406,
      "grad_norm": 0.051338629802013265,
      "learning_rate": 0.00017737220611127453,
      "loss": 0.8567,
      "step": 66970
    },
    {
      "epoch": 1.1759335662494075,
      "grad_norm": 0.06972162911564192,
      "learning_rate": 0.00017736511486263348,
      "loss": 0.8495,
      "step": 66980
    },
    {
      "epoch": 1.176109131129409,
      "grad_norm": 0.05857337274284438,
      "learning_rate": 0.00017735802264643179,
      "loss": 0.8431,
      "step": 66990
    },
    {
      "epoch": 1.1762846960094102,
      "grad_norm": 0.06967364828865967,
      "learning_rate": 0.00017735092946275928,
      "loss": 0.8449,
      "step": 67000
    },
    {
      "epoch": 1.1764602608894117,
      "grad_norm": 0.06054159508014688,
      "learning_rate": 0.0001773438353117059,
      "loss": 0.8476,
      "step": 67010
    },
    {
      "epoch": 1.1766358257694132,
      "grad_norm": 0.06445257492235898,
      "learning_rate": 0.0001773367401933615,
      "loss": 0.841,
      "step": 67020
    },
    {
      "epoch": 1.1768113906494144,
      "grad_norm": 0.060582631884227245,
      "learning_rate": 0.00017732964410781593,
      "loss": 0.8527,
      "step": 67030
    },
    {
      "epoch": 1.176986955529416,
      "grad_norm": 0.06988296285013493,
      "learning_rate": 0.0001773225470551592,
      "loss": 0.8605,
      "step": 67040
    },
    {
      "epoch": 1.1771625204094174,
      "grad_norm": 0.07944978705916177,
      "learning_rate": 0.00017731544903548113,
      "loss": 0.8529,
      "step": 67050
    },
    {
      "epoch": 1.1773380852894186,
      "grad_norm": 0.05961790251007727,
      "learning_rate": 0.00017730835004887166,
      "loss": 0.8579,
      "step": 67060
    },
    {
      "epoch": 1.1775136501694201,
      "grad_norm": 0.06467193746131072,
      "learning_rate": 0.00017730125009542084,
      "loss": 0.8475,
      "step": 67070
    },
    {
      "epoch": 1.1776892150494216,
      "grad_norm": 0.06944216356391178,
      "learning_rate": 0.00017729414917521852,
      "loss": 0.8485,
      "step": 67080
    },
    {
      "epoch": 1.1778647799294228,
      "grad_norm": 0.07832731944921628,
      "learning_rate": 0.00017728704728835472,
      "loss": 0.8502,
      "step": 67090
    },
    {
      "epoch": 1.1780403448094243,
      "grad_norm": 0.07516385358907642,
      "learning_rate": 0.00017727994443491942,
      "loss": 0.8444,
      "step": 67100
    },
    {
      "epoch": 1.1782159096894258,
      "grad_norm": 0.06383906357490506,
      "learning_rate": 0.0001772728406150026,
      "loss": 0.854,
      "step": 67110
    },
    {
      "epoch": 1.178391474569427,
      "grad_norm": 0.08534104672068801,
      "learning_rate": 0.00017726573582869425,
      "loss": 0.8538,
      "step": 67120
    },
    {
      "epoch": 1.1785670394494285,
      "grad_norm": 0.06341679099461608,
      "learning_rate": 0.00017725863007608445,
      "loss": 0.8542,
      "step": 67130
    },
    {
      "epoch": 1.17874260432943,
      "grad_norm": 0.06538292195167111,
      "learning_rate": 0.00017725152335726314,
      "loss": 0.8563,
      "step": 67140
    },
    {
      "epoch": 1.1789181692094313,
      "grad_norm": 0.0628232792343453,
      "learning_rate": 0.00017724441567232047,
      "loss": 0.8534,
      "step": 67150
    },
    {
      "epoch": 1.1790937340894327,
      "grad_norm": 0.05421107862893153,
      "learning_rate": 0.00017723730702134643,
      "loss": 0.8523,
      "step": 67160
    },
    {
      "epoch": 1.1792692989694342,
      "grad_norm": 0.06375443848076065,
      "learning_rate": 0.00017723019740443105,
      "loss": 0.8529,
      "step": 67170
    },
    {
      "epoch": 1.1794448638494355,
      "grad_norm": 0.04716519736196937,
      "learning_rate": 0.0001772230868216645,
      "loss": 0.8453,
      "step": 67180
    },
    {
      "epoch": 1.179620428729437,
      "grad_norm": 0.07236200986764063,
      "learning_rate": 0.00017721597527313684,
      "loss": 0.8496,
      "step": 67190
    },
    {
      "epoch": 1.1797959936094384,
      "grad_norm": 0.06877261101755047,
      "learning_rate": 0.00017720886275893815,
      "loss": 0.8573,
      "step": 67200
    },
    {
      "epoch": 1.1799715584894397,
      "grad_norm": 0.06251214223254317,
      "learning_rate": 0.0001772017492791586,
      "loss": 0.8512,
      "step": 67210
    },
    {
      "epoch": 1.1801471233694412,
      "grad_norm": 0.06175735608481265,
      "learning_rate": 0.00017719463483388824,
      "loss": 0.8437,
      "step": 67220
    },
    {
      "epoch": 1.1803226882494426,
      "grad_norm": 0.05800718277563114,
      "learning_rate": 0.00017718751942321726,
      "loss": 0.8479,
      "step": 67230
    },
    {
      "epoch": 1.180498253129444,
      "grad_norm": 0.058780250262052934,
      "learning_rate": 0.0001771804030472358,
      "loss": 0.8454,
      "step": 67240
    },
    {
      "epoch": 1.1806738180094454,
      "grad_norm": 0.08157186702704701,
      "learning_rate": 0.00017717328570603403,
      "loss": 0.8569,
      "step": 67250
    },
    {
      "epoch": 1.1808493828894469,
      "grad_norm": 0.07622885696536895,
      "learning_rate": 0.00017716616739970215,
      "loss": 0.8564,
      "step": 67260
    },
    {
      "epoch": 1.181024947769448,
      "grad_norm": 0.05343993142616904,
      "learning_rate": 0.00017715904812833034,
      "loss": 0.8492,
      "step": 67270
    },
    {
      "epoch": 1.1812005126494496,
      "grad_norm": 0.08713444112906572,
      "learning_rate": 0.0001771519278920088,
      "loss": 0.8509,
      "step": 67280
    },
    {
      "epoch": 1.181376077529451,
      "grad_norm": 0.06361466937338843,
      "learning_rate": 0.00017714480669082773,
      "loss": 0.8476,
      "step": 67290
    },
    {
      "epoch": 1.1815516424094523,
      "grad_norm": 0.053540267851679105,
      "learning_rate": 0.00017713768452487733,
      "loss": 0.8449,
      "step": 67300
    },
    {
      "epoch": 1.1817272072894538,
      "grad_norm": 0.06492100982754472,
      "learning_rate": 0.00017713056139424788,
      "loss": 0.8417,
      "step": 67310
    },
    {
      "epoch": 1.1819027721694553,
      "grad_norm": 0.04865320494819983,
      "learning_rate": 0.00017712343729902967,
      "loss": 0.8558,
      "step": 67320
    },
    {
      "epoch": 1.1820783370494565,
      "grad_norm": 0.07936294082476478,
      "learning_rate": 0.0001771163122393129,
      "loss": 0.8531,
      "step": 67330
    },
    {
      "epoch": 1.182253901929458,
      "grad_norm": 0.04717877527097928,
      "learning_rate": 0.00017710918621518788,
      "loss": 0.8596,
      "step": 67340
    },
    {
      "epoch": 1.1824294668094595,
      "grad_norm": 0.10725122468949852,
      "learning_rate": 0.00017710205922674485,
      "loss": 0.8558,
      "step": 67350
    },
    {
      "epoch": 1.182605031689461,
      "grad_norm": 0.0642563111647284,
      "learning_rate": 0.00017709493127407417,
      "loss": 0.8367,
      "step": 67360
    },
    {
      "epoch": 1.1827805965694622,
      "grad_norm": 0.07379139757525217,
      "learning_rate": 0.0001770878023572661,
      "loss": 0.8501,
      "step": 67370
    },
    {
      "epoch": 1.1829561614494637,
      "grad_norm": 0.06992606703006038,
      "learning_rate": 0.00017708067247641104,
      "loss": 0.8514,
      "step": 67380
    },
    {
      "epoch": 1.183131726329465,
      "grad_norm": 0.052770859167486577,
      "learning_rate": 0.00017707354163159925,
      "loss": 0.8468,
      "step": 67390
    },
    {
      "epoch": 1.1833072912094664,
      "grad_norm": 0.0808574473202574,
      "learning_rate": 0.00017706640982292113,
      "loss": 0.855,
      "step": 67400
    },
    {
      "epoch": 1.183482856089468,
      "grad_norm": 0.05162558743641875,
      "learning_rate": 0.000177059277050467,
      "loss": 0.8522,
      "step": 67410
    },
    {
      "epoch": 1.1836584209694694,
      "grad_norm": 0.06706585968071539,
      "learning_rate": 0.00017705214331432725,
      "loss": 0.8469,
      "step": 67420
    },
    {
      "epoch": 1.1838339858494706,
      "grad_norm": 0.06167762678608576,
      "learning_rate": 0.00017704500861459224,
      "loss": 0.8522,
      "step": 67430
    },
    {
      "epoch": 1.1840095507294721,
      "grad_norm": 0.05797151475000293,
      "learning_rate": 0.00017703787295135249,
      "loss": 0.8524,
      "step": 67440
    },
    {
      "epoch": 1.1841851156094734,
      "grad_norm": 0.05374671171746175,
      "learning_rate": 0.00017703073632469824,
      "loss": 0.8442,
      "step": 67450
    },
    {
      "epoch": 1.1843606804894748,
      "grad_norm": 0.07224151683310609,
      "learning_rate": 0.00017702359873472,
      "loss": 0.8572,
      "step": 67460
    },
    {
      "epoch": 1.1845362453694763,
      "grad_norm": 0.06220874026731289,
      "learning_rate": 0.00017701646018150822,
      "loss": 0.8638,
      "step": 67470
    },
    {
      "epoch": 1.1847118102494778,
      "grad_norm": 0.05621311471394867,
      "learning_rate": 0.00017700932066515328,
      "loss": 0.8573,
      "step": 67480
    },
    {
      "epoch": 1.184887375129479,
      "grad_norm": 0.07036596814669407,
      "learning_rate": 0.00017700218018574572,
      "loss": 0.845,
      "step": 67490
    },
    {
      "epoch": 1.1850629400094805,
      "grad_norm": 0.0595722105541128,
      "learning_rate": 0.00017699503874337596,
      "loss": 0.8456,
      "step": 67500
    },
    {
      "epoch": 1.1852385048894818,
      "grad_norm": 0.05647392717426963,
      "learning_rate": 0.0001769878963381345,
      "loss": 0.8511,
      "step": 67510
    },
    {
      "epoch": 1.1854140697694833,
      "grad_norm": 0.059567274184094056,
      "learning_rate": 0.0001769807529701118,
      "loss": 0.8528,
      "step": 67520
    },
    {
      "epoch": 1.1855896346494847,
      "grad_norm": 0.08766537573539245,
      "learning_rate": 0.0001769736086393984,
      "loss": 0.8515,
      "step": 67530
    },
    {
      "epoch": 1.1857651995294862,
      "grad_norm": 0.0698475038501547,
      "learning_rate": 0.00017696646334608486,
      "loss": 0.8486,
      "step": 67540
    },
    {
      "epoch": 1.1859407644094875,
      "grad_norm": 0.05508670525679419,
      "learning_rate": 0.00017695931709026165,
      "loss": 0.8506,
      "step": 67550
    },
    {
      "epoch": 1.186116329289489,
      "grad_norm": 0.06242413879525068,
      "learning_rate": 0.00017695216987201933,
      "loss": 0.8509,
      "step": 67560
    },
    {
      "epoch": 1.1862918941694902,
      "grad_norm": 0.0712810759802949,
      "learning_rate": 0.00017694502169144846,
      "loss": 0.8461,
      "step": 67570
    },
    {
      "epoch": 1.1864674590494917,
      "grad_norm": 0.06520369575885583,
      "learning_rate": 0.00017693787254863963,
      "loss": 0.8538,
      "step": 67580
    },
    {
      "epoch": 1.1866430239294932,
      "grad_norm": 0.05920091389444902,
      "learning_rate": 0.00017693072244368336,
      "loss": 0.847,
      "step": 67590
    },
    {
      "epoch": 1.1868185888094946,
      "grad_norm": 0.07630954809771702,
      "learning_rate": 0.00017692357137667032,
      "loss": 0.8507,
      "step": 67600
    },
    {
      "epoch": 1.186994153689496,
      "grad_norm": 0.07928604110414293,
      "learning_rate": 0.00017691641934769106,
      "loss": 0.8475,
      "step": 67610
    },
    {
      "epoch": 1.1871697185694974,
      "grad_norm": 0.059202416191917816,
      "learning_rate": 0.00017690926635683621,
      "loss": 0.8429,
      "step": 67620
    },
    {
      "epoch": 1.1873452834494989,
      "grad_norm": 0.06616237685175655,
      "learning_rate": 0.00017690211240419643,
      "loss": 0.8494,
      "step": 67630
    },
    {
      "epoch": 1.1875208483295001,
      "grad_norm": 0.052689868497912094,
      "learning_rate": 0.00017689495748986233,
      "loss": 0.8595,
      "step": 67640
    },
    {
      "epoch": 1.1876964132095016,
      "grad_norm": 0.06444928198949626,
      "learning_rate": 0.00017688780161392457,
      "loss": 0.8604,
      "step": 67650
    },
    {
      "epoch": 1.187871978089503,
      "grad_norm": 0.08135219749830415,
      "learning_rate": 0.00017688064477647383,
      "loss": 0.846,
      "step": 67660
    },
    {
      "epoch": 1.1880475429695043,
      "grad_norm": 0.05757779110802372,
      "learning_rate": 0.00017687348697760077,
      "loss": 0.85,
      "step": 67670
    },
    {
      "epoch": 1.1882231078495058,
      "grad_norm": 0.0595017749324334,
      "learning_rate": 0.0001768663282173961,
      "loss": 0.8488,
      "step": 67680
    },
    {
      "epoch": 1.1883986727295073,
      "grad_norm": 0.05113542070396218,
      "learning_rate": 0.0001768591684959505,
      "loss": 0.85,
      "step": 67690
    },
    {
      "epoch": 1.1885742376095085,
      "grad_norm": 0.0620066009334156,
      "learning_rate": 0.00017685200781335472,
      "loss": 0.8503,
      "step": 67700
    },
    {
      "epoch": 1.18874980248951,
      "grad_norm": 0.05219265283804171,
      "learning_rate": 0.00017684484616969944,
      "loss": 0.8526,
      "step": 67710
    },
    {
      "epoch": 1.1889253673695115,
      "grad_norm": 0.05424207610501709,
      "learning_rate": 0.00017683768356507545,
      "loss": 0.8484,
      "step": 67720
    },
    {
      "epoch": 1.1891009322495127,
      "grad_norm": 0.05959350588437233,
      "learning_rate": 0.0001768305199995734,
      "loss": 0.856,
      "step": 67730
    },
    {
      "epoch": 1.1892764971295142,
      "grad_norm": 0.05064134188045007,
      "learning_rate": 0.00017682335547328424,
      "loss": 0.8556,
      "step": 67740
    },
    {
      "epoch": 1.1894520620095157,
      "grad_norm": 0.08141825406092594,
      "learning_rate": 0.00017681618998629855,
      "loss": 0.8574,
      "step": 67750
    },
    {
      "epoch": 1.189627626889517,
      "grad_norm": 0.052501959134430266,
      "learning_rate": 0.00017680902353870725,
      "loss": 0.8582,
      "step": 67760
    },
    {
      "epoch": 1.1898031917695184,
      "grad_norm": 0.050729500140934435,
      "learning_rate": 0.00017680185613060107,
      "loss": 0.8509,
      "step": 67770
    },
    {
      "epoch": 1.18997875664952,
      "grad_norm": 0.05037801347633384,
      "learning_rate": 0.00017679468776207082,
      "loss": 0.8525,
      "step": 67780
    },
    {
      "epoch": 1.1901543215295212,
      "grad_norm": 0.05183115904981923,
      "learning_rate": 0.0001767875184332074,
      "loss": 0.8556,
      "step": 67790
    },
    {
      "epoch": 1.1903298864095226,
      "grad_norm": 0.060697644816557816,
      "learning_rate": 0.00017678034814410157,
      "loss": 0.8542,
      "step": 67800
    },
    {
      "epoch": 1.1905054512895241,
      "grad_norm": 0.10894634938825079,
      "learning_rate": 0.0001767731768948442,
      "loss": 0.8509,
      "step": 67810
    },
    {
      "epoch": 1.1906810161695254,
      "grad_norm": 0.075088742111408,
      "learning_rate": 0.00017676600468552619,
      "loss": 0.8471,
      "step": 67820
    },
    {
      "epoch": 1.1908565810495269,
      "grad_norm": 0.08502479283124258,
      "learning_rate": 0.00017675883151623836,
      "loss": 0.8471,
      "step": 67830
    },
    {
      "epoch": 1.1910321459295283,
      "grad_norm": 0.05642416363633012,
      "learning_rate": 0.0001767516573870716,
      "loss": 0.8507,
      "step": 67840
    },
    {
      "epoch": 1.1912077108095296,
      "grad_norm": 0.05797392175268148,
      "learning_rate": 0.00017674448229811684,
      "loss": 0.8531,
      "step": 67850
    },
    {
      "epoch": 1.191383275689531,
      "grad_norm": 0.061552172028314235,
      "learning_rate": 0.00017673730624946497,
      "loss": 0.8558,
      "step": 67860
    },
    {
      "epoch": 1.1915588405695325,
      "grad_norm": 0.07336547511095161,
      "learning_rate": 0.0001767301292412069,
      "loss": 0.8495,
      "step": 67870
    },
    {
      "epoch": 1.1917344054495338,
      "grad_norm": 0.061323290069446285,
      "learning_rate": 0.0001767229512734336,
      "loss": 0.8523,
      "step": 67880
    },
    {
      "epoch": 1.1919099703295353,
      "grad_norm": 0.05896961088571825,
      "learning_rate": 0.000176715772346236,
      "loss": 0.8512,
      "step": 67890
    },
    {
      "epoch": 1.1920855352095368,
      "grad_norm": 0.04801453116647186,
      "learning_rate": 0.00017670859245970502,
      "loss": 0.8515,
      "step": 67900
    },
    {
      "epoch": 1.192261100089538,
      "grad_norm": 0.059949185107009195,
      "learning_rate": 0.0001767014116139317,
      "loss": 0.8504,
      "step": 67910
    },
    {
      "epoch": 1.1924366649695395,
      "grad_norm": 0.08926758502281133,
      "learning_rate": 0.000176694229809007,
      "loss": 0.8542,
      "step": 67920
    },
    {
      "epoch": 1.192612229849541,
      "grad_norm": 0.08101465792457696,
      "learning_rate": 0.00017668704704502185,
      "loss": 0.8479,
      "step": 67930
    },
    {
      "epoch": 1.1927877947295422,
      "grad_norm": 0.05654519803962655,
      "learning_rate": 0.00017667986332206732,
      "loss": 0.8502,
      "step": 67940
    },
    {
      "epoch": 1.1929633596095437,
      "grad_norm": 0.052927380340810015,
      "learning_rate": 0.00017667267864023445,
      "loss": 0.8575,
      "step": 67950
    },
    {
      "epoch": 1.1931389244895452,
      "grad_norm": 0.05991060120884993,
      "learning_rate": 0.00017666549299961423,
      "loss": 0.8505,
      "step": 67960
    },
    {
      "epoch": 1.1933144893695464,
      "grad_norm": 0.06226335979452559,
      "learning_rate": 0.0001766583064002977,
      "loss": 0.8414,
      "step": 67970
    },
    {
      "epoch": 1.193490054249548,
      "grad_norm": 0.05328483430086237,
      "learning_rate": 0.00017665111884237596,
      "loss": 0.8521,
      "step": 67980
    },
    {
      "epoch": 1.1936656191295494,
      "grad_norm": 0.05492369408633701,
      "learning_rate": 0.00017664393032593997,
      "loss": 0.8517,
      "step": 67990
    },
    {
      "epoch": 1.1938411840095506,
      "grad_norm": 0.060211674836663225,
      "learning_rate": 0.00017663674085108098,
      "loss": 0.8509,
      "step": 68000
    },
    {
      "epoch": 1.1940167488895521,
      "grad_norm": 0.06713325739527343,
      "learning_rate": 0.0001766295504178899,
      "loss": 0.8513,
      "step": 68010
    },
    {
      "epoch": 1.1941923137695536,
      "grad_norm": 0.07126774735205454,
      "learning_rate": 0.00017662235902645795,
      "loss": 0.8445,
      "step": 68020
    },
    {
      "epoch": 1.1943678786495548,
      "grad_norm": 0.06084211317499684,
      "learning_rate": 0.00017661516667687624,
      "loss": 0.8454,
      "step": 68030
    },
    {
      "epoch": 1.1945434435295563,
      "grad_norm": 0.04738165679386999,
      "learning_rate": 0.0001766079733692359,
      "loss": 0.8551,
      "step": 68040
    },
    {
      "epoch": 1.1947190084095578,
      "grad_norm": 0.08652389995039579,
      "learning_rate": 0.00017660077910362798,
      "loss": 0.8497,
      "step": 68050
    },
    {
      "epoch": 1.194894573289559,
      "grad_norm": 0.06427436917784148,
      "learning_rate": 0.00017659358388014376,
      "loss": 0.8506,
      "step": 68060
    },
    {
      "epoch": 1.1950701381695605,
      "grad_norm": 0.0581965061977424,
      "learning_rate": 0.0001765863876988743,
      "loss": 0.8536,
      "step": 68070
    },
    {
      "epoch": 1.195245703049562,
      "grad_norm": 0.06126236484107329,
      "learning_rate": 0.00017657919055991085,
      "loss": 0.8511,
      "step": 68080
    },
    {
      "epoch": 1.1954212679295635,
      "grad_norm": 0.07970029542741533,
      "learning_rate": 0.00017657199246334457,
      "loss": 0.849,
      "step": 68090
    },
    {
      "epoch": 1.1955968328095647,
      "grad_norm": 0.05417556330680199,
      "learning_rate": 0.00017656479340926663,
      "loss": 0.8584,
      "step": 68100
    },
    {
      "epoch": 1.1957723976895662,
      "grad_norm": 0.05900429878959442,
      "learning_rate": 0.00017655759339776832,
      "loss": 0.8542,
      "step": 68110
    },
    {
      "epoch": 1.1959479625695675,
      "grad_norm": 0.07054165013147308,
      "learning_rate": 0.0001765503924289408,
      "loss": 0.8494,
      "step": 68120
    },
    {
      "epoch": 1.196123527449569,
      "grad_norm": 0.06200845161570574,
      "learning_rate": 0.00017654319050287534,
      "loss": 0.862,
      "step": 68130
    },
    {
      "epoch": 1.1962990923295704,
      "grad_norm": 0.05370971021206481,
      "learning_rate": 0.00017653598761966318,
      "loss": 0.8454,
      "step": 68140
    },
    {
      "epoch": 1.196474657209572,
      "grad_norm": 0.05323392921959561,
      "learning_rate": 0.00017652878377939554,
      "loss": 0.8552,
      "step": 68150
    },
    {
      "epoch": 1.1966502220895732,
      "grad_norm": 0.054002722692070605,
      "learning_rate": 0.00017652157898216374,
      "loss": 0.8515,
      "step": 68160
    },
    {
      "epoch": 1.1968257869695746,
      "grad_norm": 0.07038399533168309,
      "learning_rate": 0.0001765143732280591,
      "loss": 0.8513,
      "step": 68170
    },
    {
      "epoch": 1.197001351849576,
      "grad_norm": 0.04416428959312981,
      "learning_rate": 0.00017650716651717286,
      "loss": 0.8424,
      "step": 68180
    },
    {
      "epoch": 1.1971769167295774,
      "grad_norm": 0.05559538231312577,
      "learning_rate": 0.00017649995884959637,
      "loss": 0.853,
      "step": 68190
    },
    {
      "epoch": 1.1973524816095789,
      "grad_norm": 0.05145914048051736,
      "learning_rate": 0.0001764927502254209,
      "loss": 0.847,
      "step": 68200
    },
    {
      "epoch": 1.1975280464895803,
      "grad_norm": 0.06211014738785306,
      "learning_rate": 0.00017648554064473784,
      "loss": 0.8545,
      "step": 68210
    },
    {
      "epoch": 1.1977036113695816,
      "grad_norm": 0.07459084554515544,
      "learning_rate": 0.0001764783301076385,
      "loss": 0.8536,
      "step": 68220
    },
    {
      "epoch": 1.197879176249583,
      "grad_norm": 0.0537059431786568,
      "learning_rate": 0.00017647111861421425,
      "loss": 0.8518,
      "step": 68230
    },
    {
      "epoch": 1.1980547411295843,
      "grad_norm": 0.05785505990875088,
      "learning_rate": 0.00017646390616455647,
      "loss": 0.8499,
      "step": 68240
    },
    {
      "epoch": 1.1982303060095858,
      "grad_norm": 0.06246533023340328,
      "learning_rate": 0.00017645669275875655,
      "loss": 0.8557,
      "step": 68250
    },
    {
      "epoch": 1.1984058708895873,
      "grad_norm": 0.05958125993282195,
      "learning_rate": 0.00017644947839690585,
      "loss": 0.8528,
      "step": 68260
    },
    {
      "epoch": 1.1985814357695888,
      "grad_norm": 0.05098095487199019,
      "learning_rate": 0.0001764422630790958,
      "loss": 0.8482,
      "step": 68270
    },
    {
      "epoch": 1.19875700064959,
      "grad_norm": 0.05291755069266746,
      "learning_rate": 0.0001764350468054178,
      "loss": 0.8481,
      "step": 68280
    },
    {
      "epoch": 1.1989325655295915,
      "grad_norm": 0.060165467055452324,
      "learning_rate": 0.00017642782957596337,
      "loss": 0.8498,
      "step": 68290
    },
    {
      "epoch": 1.1991081304095927,
      "grad_norm": 0.05686762719275186,
      "learning_rate": 0.00017642061139082381,
      "loss": 0.8518,
      "step": 68300
    },
    {
      "epoch": 1.1992836952895942,
      "grad_norm": 0.06483531609868058,
      "learning_rate": 0.00017641339225009066,
      "loss": 0.8541,
      "step": 68310
    },
    {
      "epoch": 1.1994592601695957,
      "grad_norm": 0.10581028368664525,
      "learning_rate": 0.00017640617215385543,
      "loss": 0.8498,
      "step": 68320
    },
    {
      "epoch": 1.1996348250495972,
      "grad_norm": 0.05621892980224658,
      "learning_rate": 0.00017639895110220947,
      "loss": 0.8554,
      "step": 68330
    },
    {
      "epoch": 1.1998103899295984,
      "grad_norm": 0.06533123213008375,
      "learning_rate": 0.0001763917290952444,
      "loss": 0.8538,
      "step": 68340
    },
    {
      "epoch": 1.1999859548096,
      "grad_norm": 0.05093549499353779,
      "learning_rate": 0.00017638450613305167,
      "loss": 0.8484,
      "step": 68350
    },
    {
      "epoch": 1.2001615196896014,
      "grad_norm": 0.06330282841375084,
      "learning_rate": 0.00017637728221572282,
      "loss": 0.8545,
      "step": 68360
    },
    {
      "epoch": 1.2003370845696026,
      "grad_norm": 0.05240208463101434,
      "learning_rate": 0.0001763700573433493,
      "loss": 0.8497,
      "step": 68370
    },
    {
      "epoch": 1.2005126494496041,
      "grad_norm": 0.06765163942644778,
      "learning_rate": 0.00017636283151602276,
      "loss": 0.8551,
      "step": 68380
    },
    {
      "epoch": 1.2006882143296056,
      "grad_norm": 0.057820337366809846,
      "learning_rate": 0.00017635560473383468,
      "loss": 0.853,
      "step": 68390
    },
    {
      "epoch": 1.2008637792096069,
      "grad_norm": 0.05864721140919555,
      "learning_rate": 0.00017634837699687664,
      "loss": 0.8397,
      "step": 68400
    },
    {
      "epoch": 1.2010393440896083,
      "grad_norm": 0.049423121354390184,
      "learning_rate": 0.00017634114830524024,
      "loss": 0.8525,
      "step": 68410
    },
    {
      "epoch": 1.2012149089696098,
      "grad_norm": 0.052339606757918246,
      "learning_rate": 0.00017633391865901703,
      "loss": 0.8503,
      "step": 68420
    },
    {
      "epoch": 1.201390473849611,
      "grad_norm": 0.08167167609397699,
      "learning_rate": 0.00017632668805829866,
      "loss": 0.8649,
      "step": 68430
    },
    {
      "epoch": 1.2015660387296125,
      "grad_norm": 0.08643592158088705,
      "learning_rate": 0.00017631945650317668,
      "loss": 0.8562,
      "step": 68440
    },
    {
      "epoch": 1.201741603609614,
      "grad_norm": 0.06307181493103463,
      "learning_rate": 0.0001763122239937428,
      "loss": 0.8508,
      "step": 68450
    },
    {
      "epoch": 1.2019171684896153,
      "grad_norm": 0.06508992987185001,
      "learning_rate": 0.00017630499053008857,
      "loss": 0.8502,
      "step": 68460
    },
    {
      "epoch": 1.2020927333696168,
      "grad_norm": 0.05472667063061287,
      "learning_rate": 0.0001762977561123057,
      "loss": 0.8463,
      "step": 68470
    },
    {
      "epoch": 1.2022682982496182,
      "grad_norm": 0.11361418310163382,
      "learning_rate": 0.00017629052074048581,
      "loss": 0.8527,
      "step": 68480
    },
    {
      "epoch": 1.2024438631296195,
      "grad_norm": 0.08261288138813609,
      "learning_rate": 0.00017628328441472062,
      "loss": 0.8524,
      "step": 68490
    },
    {
      "epoch": 1.202619428009621,
      "grad_norm": 0.06572393442417926,
      "learning_rate": 0.00017627604713510176,
      "loss": 0.8468,
      "step": 68500
    },
    {
      "epoch": 1.2027949928896224,
      "grad_norm": 0.0515639038572139,
      "learning_rate": 0.00017626880890172098,
      "loss": 0.848,
      "step": 68510
    },
    {
      "epoch": 1.2029705577696237,
      "grad_norm": 0.05899910414264789,
      "learning_rate": 0.00017626156971466993,
      "loss": 0.8432,
      "step": 68520
    },
    {
      "epoch": 1.2031461226496252,
      "grad_norm": 0.050252182457455385,
      "learning_rate": 0.0001762543295740404,
      "loss": 0.8416,
      "step": 68530
    },
    {
      "epoch": 1.2033216875296266,
      "grad_norm": 0.05361427005047982,
      "learning_rate": 0.0001762470884799241,
      "loss": 0.8432,
      "step": 68540
    },
    {
      "epoch": 1.203497252409628,
      "grad_norm": 0.06739472266624336,
      "learning_rate": 0.00017623984643241276,
      "loss": 0.8478,
      "step": 68550
    },
    {
      "epoch": 1.2036728172896294,
      "grad_norm": 0.07434965384391823,
      "learning_rate": 0.0001762326034315981,
      "loss": 0.8467,
      "step": 68560
    },
    {
      "epoch": 1.2038483821696309,
      "grad_norm": 0.0563125276162476,
      "learning_rate": 0.00017622535947757196,
      "loss": 0.8505,
      "step": 68570
    },
    {
      "epoch": 1.2040239470496321,
      "grad_norm": 0.07870919864403823,
      "learning_rate": 0.00017621811457042613,
      "loss": 0.8486,
      "step": 68580
    },
    {
      "epoch": 1.2041995119296336,
      "grad_norm": 0.055055760193395524,
      "learning_rate": 0.00017621086871025236,
      "loss": 0.8355,
      "step": 68590
    },
    {
      "epoch": 1.204375076809635,
      "grad_norm": 0.057501993248541355,
      "learning_rate": 0.00017620362189714247,
      "loss": 0.8515,
      "step": 68600
    },
    {
      "epoch": 1.2045506416896363,
      "grad_norm": 0.05400975171499392,
      "learning_rate": 0.0001761963741311883,
      "loss": 0.859,
      "step": 68610
    },
    {
      "epoch": 1.2047262065696378,
      "grad_norm": 0.06067898401756758,
      "learning_rate": 0.0001761891254124816,
      "loss": 0.8535,
      "step": 68620
    },
    {
      "epoch": 1.2049017714496393,
      "grad_norm": 0.06677549879473428,
      "learning_rate": 0.0001761818757411143,
      "loss": 0.8538,
      "step": 68630
    },
    {
      "epoch": 1.2050773363296405,
      "grad_norm": 0.04881873623776551,
      "learning_rate": 0.00017617462511717823,
      "loss": 0.8581,
      "step": 68640
    },
    {
      "epoch": 1.205252901209642,
      "grad_norm": 0.10137700518661401,
      "learning_rate": 0.00017616737354076524,
      "loss": 0.8499,
      "step": 68650
    },
    {
      "epoch": 1.2054284660896435,
      "grad_norm": 0.06607282623756078,
      "learning_rate": 0.00017616012101196723,
      "loss": 0.8408,
      "step": 68660
    },
    {
      "epoch": 1.2056040309696447,
      "grad_norm": 0.06373843028143356,
      "learning_rate": 0.0001761528675308761,
      "loss": 0.8477,
      "step": 68670
    },
    {
      "epoch": 1.2057795958496462,
      "grad_norm": 0.06663963043686887,
      "learning_rate": 0.00017614561309758372,
      "loss": 0.851,
      "step": 68680
    },
    {
      "epoch": 1.2059551607296477,
      "grad_norm": 0.06720741866042469,
      "learning_rate": 0.00017613835771218203,
      "loss": 0.8478,
      "step": 68690
    },
    {
      "epoch": 1.206130725609649,
      "grad_norm": 0.05400735399818228,
      "learning_rate": 0.0001761311013747629,
      "loss": 0.8536,
      "step": 68700
    },
    {
      "epoch": 1.2063062904896504,
      "grad_norm": 0.0798636377120103,
      "learning_rate": 0.00017612384408541838,
      "loss": 0.8545,
      "step": 68710
    },
    {
      "epoch": 1.206481855369652,
      "grad_norm": 0.05484729287176015,
      "learning_rate": 0.00017611658584424034,
      "loss": 0.8531,
      "step": 68720
    },
    {
      "epoch": 1.2066574202496532,
      "grad_norm": 0.06922675059744755,
      "learning_rate": 0.00017610932665132074,
      "loss": 0.8532,
      "step": 68730
    },
    {
      "epoch": 1.2068329851296546,
      "grad_norm": 0.07365080119479793,
      "learning_rate": 0.00017610206650675158,
      "loss": 0.8447,
      "step": 68740
    },
    {
      "epoch": 1.2070085500096561,
      "grad_norm": 0.06717944857332152,
      "learning_rate": 0.0001760948054106249,
      "loss": 0.8522,
      "step": 68750
    },
    {
      "epoch": 1.2071841148896574,
      "grad_norm": 0.062201782440058154,
      "learning_rate": 0.00017608754336303256,
      "loss": 0.8475,
      "step": 68760
    },
    {
      "epoch": 1.2073596797696589,
      "grad_norm": 0.07834124628873017,
      "learning_rate": 0.0001760802803640667,
      "loss": 0.8547,
      "step": 68770
    },
    {
      "epoch": 1.2075352446496603,
      "grad_norm": 0.07671479470001409,
      "learning_rate": 0.00017607301641381928,
      "loss": 0.8561,
      "step": 68780
    },
    {
      "epoch": 1.2077108095296616,
      "grad_norm": 0.05986118210793201,
      "learning_rate": 0.00017606575151238237,
      "loss": 0.8415,
      "step": 68790
    },
    {
      "epoch": 1.207886374409663,
      "grad_norm": 0.07079499477664508,
      "learning_rate": 0.00017605848565984802,
      "loss": 0.848,
      "step": 68800
    },
    {
      "epoch": 1.2080619392896645,
      "grad_norm": 0.06423413948055286,
      "learning_rate": 0.00017605121885630822,
      "loss": 0.8522,
      "step": 68810
    },
    {
      "epoch": 1.208237504169666,
      "grad_norm": 0.0720516966260325,
      "learning_rate": 0.00017604395110185512,
      "loss": 0.8489,
      "step": 68820
    },
    {
      "epoch": 1.2084130690496673,
      "grad_norm": 0.06757284254182908,
      "learning_rate": 0.0001760366823965808,
      "loss": 0.8534,
      "step": 68830
    },
    {
      "epoch": 1.2085886339296688,
      "grad_norm": 0.046862398851652526,
      "learning_rate": 0.0001760294127405773,
      "loss": 0.8519,
      "step": 68840
    },
    {
      "epoch": 1.20876419880967,
      "grad_norm": 0.0718801849989231,
      "learning_rate": 0.00017602214213393677,
      "loss": 0.8476,
      "step": 68850
    },
    {
      "epoch": 1.2089397636896715,
      "grad_norm": 0.05236892597359392,
      "learning_rate": 0.00017601487057675134,
      "loss": 0.8485,
      "step": 68860
    },
    {
      "epoch": 1.209115328569673,
      "grad_norm": 0.06187647303692835,
      "learning_rate": 0.0001760075980691131,
      "loss": 0.8477,
      "step": 68870
    },
    {
      "epoch": 1.2092908934496744,
      "grad_norm": 0.059152757544697315,
      "learning_rate": 0.00017600032461111422,
      "loss": 0.8472,
      "step": 68880
    },
    {
      "epoch": 1.2094664583296757,
      "grad_norm": 0.04876773528117157,
      "learning_rate": 0.00017599305020284686,
      "loss": 0.8509,
      "step": 68890
    },
    {
      "epoch": 1.2096420232096772,
      "grad_norm": 0.055939515251194664,
      "learning_rate": 0.0001759857748444032,
      "loss": 0.8492,
      "step": 68900
    },
    {
      "epoch": 1.2098175880896784,
      "grad_norm": 0.06287391137675026,
      "learning_rate": 0.00017597849853587537,
      "loss": 0.8541,
      "step": 68910
    },
    {
      "epoch": 1.20999315296968,
      "grad_norm": 0.0771239458067529,
      "learning_rate": 0.00017597122127735562,
      "loss": 0.8438,
      "step": 68920
    },
    {
      "epoch": 1.2101687178496814,
      "grad_norm": 0.052783618244214986,
      "learning_rate": 0.0001759639430689361,
      "loss": 0.8473,
      "step": 68930
    },
    {
      "epoch": 1.2103442827296829,
      "grad_norm": 0.060946031843426084,
      "learning_rate": 0.00017595666391070908,
      "loss": 0.8465,
      "step": 68940
    },
    {
      "epoch": 1.2105198476096841,
      "grad_norm": 0.04753286715365804,
      "learning_rate": 0.00017594938380276675,
      "loss": 0.8478,
      "step": 68950
    },
    {
      "epoch": 1.2106954124896856,
      "grad_norm": 0.077065519500785,
      "learning_rate": 0.00017594210274520135,
      "loss": 0.8447,
      "step": 68960
    },
    {
      "epoch": 1.2108709773696869,
      "grad_norm": 0.0998974496994921,
      "learning_rate": 0.00017593482073810518,
      "loss": 0.8486,
      "step": 68970
    },
    {
      "epoch": 1.2110465422496883,
      "grad_norm": 0.0517369365782822,
      "learning_rate": 0.00017592753778157048,
      "loss": 0.8468,
      "step": 68980
    },
    {
      "epoch": 1.2112221071296898,
      "grad_norm": 0.06963274750687529,
      "learning_rate": 0.00017592025387568947,
      "loss": 0.8481,
      "step": 68990
    },
    {
      "epoch": 1.2113976720096913,
      "grad_norm": 0.05734830437294415,
      "learning_rate": 0.00017591296902055452,
      "loss": 0.8462,
      "step": 69000
    },
    {
      "epoch": 1.2115732368896925,
      "grad_norm": 0.07209924523274788,
      "learning_rate": 0.00017590568321625786,
      "loss": 0.8473,
      "step": 69010
    },
    {
      "epoch": 1.211748801769694,
      "grad_norm": 0.05715778388323867,
      "learning_rate": 0.00017589839646289186,
      "loss": 0.8461,
      "step": 69020
    },
    {
      "epoch": 1.2119243666496953,
      "grad_norm": 0.06657130156472842,
      "learning_rate": 0.0001758911087605488,
      "loss": 0.8537,
      "step": 69030
    },
    {
      "epoch": 1.2120999315296968,
      "grad_norm": 0.06537961072635133,
      "learning_rate": 0.0001758838201093211,
      "loss": 0.8456,
      "step": 69040
    },
    {
      "epoch": 1.2122754964096982,
      "grad_norm": 0.05477249875765379,
      "learning_rate": 0.000175876530509301,
      "loss": 0.8433,
      "step": 69050
    },
    {
      "epoch": 1.2124510612896997,
      "grad_norm": 0.07339596471454082,
      "learning_rate": 0.00017586923996058094,
      "loss": 0.8431,
      "step": 69060
    },
    {
      "epoch": 1.212626626169701,
      "grad_norm": 0.09291578307297488,
      "learning_rate": 0.0001758619484632532,
      "loss": 0.8498,
      "step": 69070
    },
    {
      "epoch": 1.2128021910497024,
      "grad_norm": 0.06807497414678475,
      "learning_rate": 0.00017585465601741026,
      "loss": 0.8514,
      "step": 69080
    },
    {
      "epoch": 1.212977755929704,
      "grad_norm": 0.06670845153107383,
      "learning_rate": 0.00017584736262314448,
      "loss": 0.8439,
      "step": 69090
    },
    {
      "epoch": 1.2131533208097052,
      "grad_norm": 0.06457680334282899,
      "learning_rate": 0.00017584006828054829,
      "loss": 0.8436,
      "step": 69100
    },
    {
      "epoch": 1.2133288856897066,
      "grad_norm": 0.05035505281002478,
      "learning_rate": 0.00017583277298971405,
      "loss": 0.8492,
      "step": 69110
    },
    {
      "epoch": 1.2135044505697081,
      "grad_norm": 0.0447472579279613,
      "learning_rate": 0.00017582547675073427,
      "loss": 0.8523,
      "step": 69120
    },
    {
      "epoch": 1.2136800154497094,
      "grad_norm": 0.1687829848096704,
      "learning_rate": 0.00017581817956370137,
      "loss": 0.849,
      "step": 69130
    },
    {
      "epoch": 1.2138555803297109,
      "grad_norm": 0.053600931685913074,
      "learning_rate": 0.00017581088142870778,
      "loss": 0.8474,
      "step": 69140
    },
    {
      "epoch": 1.2140311452097123,
      "grad_norm": 0.04916754977571986,
      "learning_rate": 0.000175803582345846,
      "loss": 0.8465,
      "step": 69150
    },
    {
      "epoch": 1.2142067100897136,
      "grad_norm": 0.061195991225114654,
      "learning_rate": 0.00017579628231520846,
      "loss": 0.8466,
      "step": 69160
    },
    {
      "epoch": 1.214382274969715,
      "grad_norm": 0.06193755223709708,
      "learning_rate": 0.0001757889813368877,
      "loss": 0.8545,
      "step": 69170
    },
    {
      "epoch": 1.2145578398497165,
      "grad_norm": 0.06638544815263384,
      "learning_rate": 0.00017578167941097623,
      "loss": 0.8494,
      "step": 69180
    },
    {
      "epoch": 1.2147334047297178,
      "grad_norm": 0.07429519521728069,
      "learning_rate": 0.00017577437653756655,
      "loss": 0.8562,
      "step": 69190
    },
    {
      "epoch": 1.2149089696097193,
      "grad_norm": 0.054440028036854424,
      "learning_rate": 0.00017576707271675114,
      "loss": 0.8418,
      "step": 69200
    },
    {
      "epoch": 1.2150845344897208,
      "grad_norm": 0.06826029525942919,
      "learning_rate": 0.00017575976794862263,
      "loss": 0.8516,
      "step": 69210
    },
    {
      "epoch": 1.215260099369722,
      "grad_norm": 0.05298518108961945,
      "learning_rate": 0.00017575246223327354,
      "loss": 0.8511,
      "step": 69220
    },
    {
      "epoch": 1.2154356642497235,
      "grad_norm": 0.05732340920822943,
      "learning_rate": 0.00017574515557079642,
      "loss": 0.8494,
      "step": 69230
    },
    {
      "epoch": 1.215611229129725,
      "grad_norm": 0.05176755370599254,
      "learning_rate": 0.00017573784796128385,
      "loss": 0.8537,
      "step": 69240
    },
    {
      "epoch": 1.2157867940097262,
      "grad_norm": 0.050565333995817925,
      "learning_rate": 0.00017573053940482845,
      "loss": 0.8552,
      "step": 69250
    },
    {
      "epoch": 1.2159623588897277,
      "grad_norm": 0.06253849639103926,
      "learning_rate": 0.00017572322990152276,
      "loss": 0.8467,
      "step": 69260
    },
    {
      "epoch": 1.2161379237697292,
      "grad_norm": 0.07611928713742057,
      "learning_rate": 0.00017571591945145942,
      "loss": 0.8574,
      "step": 69270
    },
    {
      "epoch": 1.2163134886497304,
      "grad_norm": 0.06111267593519427,
      "learning_rate": 0.0001757086080547311,
      "loss": 0.8506,
      "step": 69280
    },
    {
      "epoch": 1.216489053529732,
      "grad_norm": 0.056705651497911,
      "learning_rate": 0.00017570129571143036,
      "loss": 0.8519,
      "step": 69290
    },
    {
      "epoch": 1.2166646184097334,
      "grad_norm": 0.0697326440456881,
      "learning_rate": 0.0001756939824216499,
      "loss": 0.8399,
      "step": 69300
    },
    {
      "epoch": 1.2168401832897346,
      "grad_norm": 0.07259754321131326,
      "learning_rate": 0.00017568666818548235,
      "loss": 0.8575,
      "step": 69310
    },
    {
      "epoch": 1.2170157481697361,
      "grad_norm": 0.06044859516082529,
      "learning_rate": 0.00017567935300302046,
      "loss": 0.8515,
      "step": 69320
    },
    {
      "epoch": 1.2171913130497376,
      "grad_norm": 0.06466935770307206,
      "learning_rate": 0.0001756720368743568,
      "loss": 0.8556,
      "step": 69330
    },
    {
      "epoch": 1.2173668779297389,
      "grad_norm": 0.06720824574142885,
      "learning_rate": 0.00017566471979958413,
      "loss": 0.8546,
      "step": 69340
    },
    {
      "epoch": 1.2175424428097403,
      "grad_norm": 0.06204501997183284,
      "learning_rate": 0.00017565740177879517,
      "loss": 0.8502,
      "step": 69350
    },
    {
      "epoch": 1.2177180076897418,
      "grad_norm": 0.0663108875589936,
      "learning_rate": 0.0001756500828120826,
      "loss": 0.8475,
      "step": 69360
    },
    {
      "epoch": 1.217893572569743,
      "grad_norm": 0.06643947748694637,
      "learning_rate": 0.00017564276289953917,
      "loss": 0.8408,
      "step": 69370
    },
    {
      "epoch": 1.2180691374497445,
      "grad_norm": 0.09005544439371192,
      "learning_rate": 0.00017563544204125765,
      "loss": 0.8488,
      "step": 69380
    },
    {
      "epoch": 1.218244702329746,
      "grad_norm": 0.08460806843913546,
      "learning_rate": 0.00017562812023733077,
      "loss": 0.856,
      "step": 69390
    },
    {
      "epoch": 1.2184202672097473,
      "grad_norm": 0.05844904347034901,
      "learning_rate": 0.0001756207974878513,
      "loss": 0.8487,
      "step": 69400
    },
    {
      "epoch": 1.2185958320897488,
      "grad_norm": 0.07833982090934612,
      "learning_rate": 0.00017561347379291202,
      "loss": 0.8518,
      "step": 69410
    },
    {
      "epoch": 1.2187713969697502,
      "grad_norm": 0.0659666374915996,
      "learning_rate": 0.00017560614915260577,
      "loss": 0.853,
      "step": 69420
    },
    {
      "epoch": 1.2189469618497515,
      "grad_norm": 0.060769665505354176,
      "learning_rate": 0.00017559882356702527,
      "loss": 0.8513,
      "step": 69430
    },
    {
      "epoch": 1.219122526729753,
      "grad_norm": 0.06731081089827386,
      "learning_rate": 0.00017559149703626336,
      "loss": 0.8482,
      "step": 69440
    },
    {
      "epoch": 1.2192980916097544,
      "grad_norm": 0.05637729025646567,
      "learning_rate": 0.00017558416956041292,
      "loss": 0.8438,
      "step": 69450
    },
    {
      "epoch": 1.2194736564897557,
      "grad_norm": 0.07932381785353959,
      "learning_rate": 0.00017557684113956676,
      "loss": 0.8427,
      "step": 69460
    },
    {
      "epoch": 1.2196492213697572,
      "grad_norm": 0.0564685907245687,
      "learning_rate": 0.00017556951177381767,
      "loss": 0.8448,
      "step": 69470
    },
    {
      "epoch": 1.2198247862497587,
      "grad_norm": 0.08613592688437455,
      "learning_rate": 0.00017556218146325866,
      "loss": 0.8507,
      "step": 69480
    },
    {
      "epoch": 1.22000035112976,
      "grad_norm": 0.056753718626496384,
      "learning_rate": 0.00017555485020798251,
      "loss": 0.846,
      "step": 69490
    },
    {
      "epoch": 1.2201759160097614,
      "grad_norm": 0.0542743191347832,
      "learning_rate": 0.00017554751800808206,
      "loss": 0.852,
      "step": 69500
    },
    {
      "epoch": 1.2203514808897629,
      "grad_norm": 0.06101160838009717,
      "learning_rate": 0.00017554018486365034,
      "loss": 0.8438,
      "step": 69510
    },
    {
      "epoch": 1.2205270457697641,
      "grad_norm": 0.06982711578743546,
      "learning_rate": 0.00017553285077478013,
      "loss": 0.848,
      "step": 69520
    },
    {
      "epoch": 1.2207026106497656,
      "grad_norm": 0.07448514184516164,
      "learning_rate": 0.00017552551574156444,
      "loss": 0.8495,
      "step": 69530
    },
    {
      "epoch": 1.220878175529767,
      "grad_norm": 0.05802176041991153,
      "learning_rate": 0.00017551817976409618,
      "loss": 0.8561,
      "step": 69540
    },
    {
      "epoch": 1.2210537404097686,
      "grad_norm": 0.06969236988038642,
      "learning_rate": 0.00017551084284246832,
      "loss": 0.8518,
      "step": 69550
    },
    {
      "epoch": 1.2212293052897698,
      "grad_norm": 0.059985478164959964,
      "learning_rate": 0.00017550350497677378,
      "loss": 0.8422,
      "step": 69560
    },
    {
      "epoch": 1.2214048701697713,
      "grad_norm": 0.07222582335168742,
      "learning_rate": 0.00017549616616710558,
      "loss": 0.8545,
      "step": 69570
    },
    {
      "epoch": 1.2215804350497725,
      "grad_norm": 0.08377697449244088,
      "learning_rate": 0.00017548882641355663,
      "loss": 0.8449,
      "step": 69580
    },
    {
      "epoch": 1.221755999929774,
      "grad_norm": 0.050569961934486665,
      "learning_rate": 0.00017548148571622002,
      "loss": 0.8484,
      "step": 69590
    },
    {
      "epoch": 1.2219315648097755,
      "grad_norm": 0.07039380942405787,
      "learning_rate": 0.0001754741440751887,
      "loss": 0.8484,
      "step": 69600
    },
    {
      "epoch": 1.222107129689777,
      "grad_norm": 0.06662121390827146,
      "learning_rate": 0.0001754668014905557,
      "loss": 0.8535,
      "step": 69610
    },
    {
      "epoch": 1.2222826945697782,
      "grad_norm": 0.055404857439450596,
      "learning_rate": 0.00017545945796241405,
      "loss": 0.8522,
      "step": 69620
    },
    {
      "epoch": 1.2224582594497797,
      "grad_norm": 0.08054613793592483,
      "learning_rate": 0.0001754521134908568,
      "loss": 0.861,
      "step": 69630
    },
    {
      "epoch": 1.222633824329781,
      "grad_norm": 0.06569557798653189,
      "learning_rate": 0.00017544476807597698,
      "loss": 0.8545,
      "step": 69640
    },
    {
      "epoch": 1.2228093892097824,
      "grad_norm": 0.05668004703130368,
      "learning_rate": 0.00017543742171786773,
      "loss": 0.8506,
      "step": 69650
    },
    {
      "epoch": 1.222984954089784,
      "grad_norm": 0.04982552782777401,
      "learning_rate": 0.00017543007441662205,
      "loss": 0.8486,
      "step": 69660
    },
    {
      "epoch": 1.2231605189697854,
      "grad_norm": 0.07903628729005291,
      "learning_rate": 0.00017542272617233307,
      "loss": 0.8539,
      "step": 69670
    },
    {
      "epoch": 1.2233360838497866,
      "grad_norm": 0.06854061156159683,
      "learning_rate": 0.00017541537698509389,
      "loss": 0.8523,
      "step": 69680
    },
    {
      "epoch": 1.2235116487297881,
      "grad_norm": 0.06921236768269293,
      "learning_rate": 0.0001754080268549976,
      "loss": 0.8457,
      "step": 69690
    },
    {
      "epoch": 1.2236872136097894,
      "grad_norm": 0.07467007263462531,
      "learning_rate": 0.00017540067578213734,
      "loss": 0.8401,
      "step": 69700
    },
    {
      "epoch": 1.2238627784897909,
      "grad_norm": 0.06442352935390228,
      "learning_rate": 0.00017539332376660628,
      "loss": 0.8613,
      "step": 69710
    },
    {
      "epoch": 1.2240383433697923,
      "grad_norm": 0.05448942160776433,
      "learning_rate": 0.00017538597080849757,
      "loss": 0.8552,
      "step": 69720
    },
    {
      "epoch": 1.2242139082497938,
      "grad_norm": 0.07500490877314263,
      "learning_rate": 0.00017537861690790433,
      "loss": 0.8509,
      "step": 69730
    },
    {
      "epoch": 1.224389473129795,
      "grad_norm": 0.05658754517262798,
      "learning_rate": 0.00017537126206491977,
      "loss": 0.8424,
      "step": 69740
    },
    {
      "epoch": 1.2245650380097965,
      "grad_norm": 0.062475016919624676,
      "learning_rate": 0.00017536390627963708,
      "loss": 0.8457,
      "step": 69750
    },
    {
      "epoch": 1.2247406028897978,
      "grad_norm": 0.049412015644022936,
      "learning_rate": 0.0001753565495521494,
      "loss": 0.854,
      "step": 69760
    },
    {
      "epoch": 1.2249161677697993,
      "grad_norm": 0.0547648962993172,
      "learning_rate": 0.00017534919188255004,
      "loss": 0.8436,
      "step": 69770
    },
    {
      "epoch": 1.2250917326498008,
      "grad_norm": 0.20710294204698404,
      "learning_rate": 0.00017534183327093215,
      "loss": 0.8446,
      "step": 69780
    },
    {
      "epoch": 1.2252672975298022,
      "grad_norm": 0.06538102298317117,
      "learning_rate": 0.000175334473717389,
      "loss": 0.8506,
      "step": 69790
    },
    {
      "epoch": 1.2254428624098035,
      "grad_norm": 0.05908137729345147,
      "learning_rate": 0.00017532711322201377,
      "loss": 0.8393,
      "step": 69800
    },
    {
      "epoch": 1.225618427289805,
      "grad_norm": 0.07649036914146527,
      "learning_rate": 0.0001753197517848998,
      "loss": 0.8561,
      "step": 69810
    },
    {
      "epoch": 1.2257939921698064,
      "grad_norm": 0.06403974574395713,
      "learning_rate": 0.00017531238940614034,
      "loss": 0.8492,
      "step": 69820
    },
    {
      "epoch": 1.2259695570498077,
      "grad_norm": 0.04567697363957614,
      "learning_rate": 0.00017530502608582865,
      "loss": 0.8512,
      "step": 69830
    },
    {
      "epoch": 1.2261451219298092,
      "grad_norm": 0.06557015337303851,
      "learning_rate": 0.00017529766182405808,
      "loss": 0.8474,
      "step": 69840
    },
    {
      "epoch": 1.2263206868098107,
      "grad_norm": 0.05434518599917705,
      "learning_rate": 0.00017529029662092185,
      "loss": 0.8484,
      "step": 69850
    },
    {
      "epoch": 1.226496251689812,
      "grad_norm": 0.05130546265250548,
      "learning_rate": 0.00017528293047651335,
      "loss": 0.8531,
      "step": 69860
    },
    {
      "epoch": 1.2266718165698134,
      "grad_norm": 0.08313390214601013,
      "learning_rate": 0.00017527556339092586,
      "loss": 0.8476,
      "step": 69870
    },
    {
      "epoch": 1.2268473814498149,
      "grad_norm": 0.11818272508901662,
      "learning_rate": 0.00017526819536425273,
      "loss": 0.8527,
      "step": 69880
    },
    {
      "epoch": 1.2270229463298161,
      "grad_norm": 0.05937280520432867,
      "learning_rate": 0.00017526082639658736,
      "loss": 0.8529,
      "step": 69890
    },
    {
      "epoch": 1.2271985112098176,
      "grad_norm": 0.07258374091354232,
      "learning_rate": 0.0001752534564880231,
      "loss": 0.8453,
      "step": 69900
    },
    {
      "epoch": 1.227374076089819,
      "grad_norm": 0.07756297199555673,
      "learning_rate": 0.00017524608563865334,
      "loss": 0.8513,
      "step": 69910
    },
    {
      "epoch": 1.2275496409698203,
      "grad_norm": 0.05194453079252272,
      "learning_rate": 0.00017523871384857138,
      "loss": 0.8499,
      "step": 69920
    },
    {
      "epoch": 1.2277252058498218,
      "grad_norm": 0.08650559834841534,
      "learning_rate": 0.00017523134111787072,
      "loss": 0.8575,
      "step": 69930
    },
    {
      "epoch": 1.2279007707298233,
      "grad_norm": 0.05712853446211042,
      "learning_rate": 0.00017522396744664473,
      "loss": 0.8557,
      "step": 69940
    },
    {
      "epoch": 1.2280763356098245,
      "grad_norm": 0.08010143545451982,
      "learning_rate": 0.00017521659283498687,
      "loss": 0.842,
      "step": 69950
    },
    {
      "epoch": 1.228251900489826,
      "grad_norm": 0.06748136514028745,
      "learning_rate": 0.00017520921728299055,
      "loss": 0.8607,
      "step": 69960
    },
    {
      "epoch": 1.2284274653698275,
      "grad_norm": 0.06069663903030926,
      "learning_rate": 0.00017520184079074922,
      "loss": 0.8476,
      "step": 69970
    },
    {
      "epoch": 1.2286030302498288,
      "grad_norm": 0.06191603647885196,
      "learning_rate": 0.00017519446335835629,
      "loss": 0.8526,
      "step": 69980
    },
    {
      "epoch": 1.2287785951298302,
      "grad_norm": 0.04624608149921816,
      "learning_rate": 0.00017518708498590537,
      "loss": 0.8577,
      "step": 69990
    },
    {
      "epoch": 1.2289541600098317,
      "grad_norm": 0.053365300795504236,
      "learning_rate": 0.00017517970567348983,
      "loss": 0.8421,
      "step": 70000
    },
    {
      "epoch": 1.229129724889833,
      "grad_norm": 0.06375302158917526,
      "learning_rate": 0.00017517232542120322,
      "loss": 0.8464,
      "step": 70010
    },
    {
      "epoch": 1.2293052897698344,
      "grad_norm": 0.05350302558558072,
      "learning_rate": 0.00017516494422913903,
      "loss": 0.8554,
      "step": 70020
    },
    {
      "epoch": 1.229480854649836,
      "grad_norm": 0.09020340353536971,
      "learning_rate": 0.00017515756209739077,
      "loss": 0.8387,
      "step": 70030
    },
    {
      "epoch": 1.2296564195298372,
      "grad_norm": 0.05714395117296419,
      "learning_rate": 0.000175150179026052,
      "loss": 0.8455,
      "step": 70040
    },
    {
      "epoch": 1.2298319844098387,
      "grad_norm": 0.05666711208781458,
      "learning_rate": 0.00017514279501521624,
      "loss": 0.846,
      "step": 70050
    },
    {
      "epoch": 1.2300075492898401,
      "grad_norm": 0.04725293871285169,
      "learning_rate": 0.00017513541006497705,
      "loss": 0.8445,
      "step": 70060
    },
    {
      "epoch": 1.2301831141698414,
      "grad_norm": 0.057045320790698496,
      "learning_rate": 0.00017512802417542802,
      "loss": 0.8548,
      "step": 70070
    },
    {
      "epoch": 1.2303586790498429,
      "grad_norm": 0.0640004154525448,
      "learning_rate": 0.00017512063734666271,
      "loss": 0.8493,
      "step": 70080
    },
    {
      "epoch": 1.2305342439298443,
      "grad_norm": 0.05146079331483483,
      "learning_rate": 0.0001751132495787747,
      "loss": 0.8488,
      "step": 70090
    },
    {
      "epoch": 1.2307098088098456,
      "grad_norm": 0.057560968828308585,
      "learning_rate": 0.00017510586087185768,
      "loss": 0.845,
      "step": 70100
    },
    {
      "epoch": 1.230885373689847,
      "grad_norm": 0.06845261589932354,
      "learning_rate": 0.0001750984712260051,
      "loss": 0.8479,
      "step": 70110
    },
    {
      "epoch": 1.2310609385698486,
      "grad_norm": 0.056816968951460045,
      "learning_rate": 0.0001750910806413108,
      "loss": 0.8479,
      "step": 70120
    },
    {
      "epoch": 1.2312365034498498,
      "grad_norm": 0.054837835115689196,
      "learning_rate": 0.00017508368911786824,
      "loss": 0.8488,
      "step": 70130
    },
    {
      "epoch": 1.2314120683298513,
      "grad_norm": 0.056291010481640674,
      "learning_rate": 0.00017507629665577113,
      "loss": 0.8482,
      "step": 70140
    },
    {
      "epoch": 1.2315876332098528,
      "grad_norm": 0.05845934635395875,
      "learning_rate": 0.00017506890325511319,
      "loss": 0.8614,
      "step": 70150
    },
    {
      "epoch": 1.231763198089854,
      "grad_norm": 0.05408887999555541,
      "learning_rate": 0.00017506150891598803,
      "loss": 0.8498,
      "step": 70160
    },
    {
      "epoch": 1.2319387629698555,
      "grad_norm": 0.05021042109400439,
      "learning_rate": 0.0001750541136384894,
      "loss": 0.8497,
      "step": 70170
    },
    {
      "epoch": 1.232114327849857,
      "grad_norm": 0.04948423853027652,
      "learning_rate": 0.00017504671742271088,
      "loss": 0.8525,
      "step": 70180
    },
    {
      "epoch": 1.2322898927298582,
      "grad_norm": 0.07534161840780414,
      "learning_rate": 0.0001750393202687463,
      "loss": 0.854,
      "step": 70190
    },
    {
      "epoch": 1.2324654576098597,
      "grad_norm": 0.060948614981680804,
      "learning_rate": 0.00017503192217668937,
      "loss": 0.8484,
      "step": 70200
    },
    {
      "epoch": 1.2326410224898612,
      "grad_norm": 0.07597228474402204,
      "learning_rate": 0.00017502452314663373,
      "loss": 0.8422,
      "step": 70210
    },
    {
      "epoch": 1.2328165873698624,
      "grad_norm": 0.049638931513817354,
      "learning_rate": 0.00017501712317867323,
      "loss": 0.8431,
      "step": 70220
    },
    {
      "epoch": 1.232992152249864,
      "grad_norm": 0.07627831424753798,
      "learning_rate": 0.0001750097222729016,
      "loss": 0.8492,
      "step": 70230
    },
    {
      "epoch": 1.2331677171298654,
      "grad_norm": 0.057685309147513776,
      "learning_rate": 0.00017500232042941258,
      "loss": 0.843,
      "step": 70240
    },
    {
      "epoch": 1.2333432820098666,
      "grad_norm": 0.049723141874490596,
      "learning_rate": 0.00017499491764830002,
      "loss": 0.8439,
      "step": 70250
    },
    {
      "epoch": 1.2335188468898681,
      "grad_norm": 0.08772333335035615,
      "learning_rate": 0.00017498751392965763,
      "loss": 0.8447,
      "step": 70260
    },
    {
      "epoch": 1.2336944117698696,
      "grad_norm": 0.06887623269254212,
      "learning_rate": 0.00017498010927357932,
      "loss": 0.8548,
      "step": 70270
    },
    {
      "epoch": 1.233869976649871,
      "grad_norm": 0.06025074210650839,
      "learning_rate": 0.00017497270368015876,
      "loss": 0.8519,
      "step": 70280
    },
    {
      "epoch": 1.2340455415298723,
      "grad_norm": 0.05841854756976415,
      "learning_rate": 0.00017496529714948995,
      "loss": 0.8442,
      "step": 70290
    },
    {
      "epoch": 1.2342211064098738,
      "grad_norm": 0.07394819924179563,
      "learning_rate": 0.0001749578896816666,
      "loss": 0.8427,
      "step": 70300
    },
    {
      "epoch": 1.234396671289875,
      "grad_norm": 0.05066939957284587,
      "learning_rate": 0.00017495048127678264,
      "loss": 0.8478,
      "step": 70310
    },
    {
      "epoch": 1.2345722361698765,
      "grad_norm": 0.08073720996547826,
      "learning_rate": 0.00017494307193493188,
      "loss": 0.8503,
      "step": 70320
    },
    {
      "epoch": 1.234747801049878,
      "grad_norm": 0.07042337854482139,
      "learning_rate": 0.00017493566165620826,
      "loss": 0.859,
      "step": 70330
    },
    {
      "epoch": 1.2349233659298795,
      "grad_norm": 0.06837233541505913,
      "learning_rate": 0.00017492825044070564,
      "loss": 0.8505,
      "step": 70340
    },
    {
      "epoch": 1.2350989308098808,
      "grad_norm": 0.06698661653085462,
      "learning_rate": 0.00017492083828851788,
      "loss": 0.8567,
      "step": 70350
    },
    {
      "epoch": 1.2352744956898822,
      "grad_norm": 0.06614409532117663,
      "learning_rate": 0.00017491342519973895,
      "loss": 0.8515,
      "step": 70360
    },
    {
      "epoch": 1.2354500605698835,
      "grad_norm": 0.06471652033636478,
      "learning_rate": 0.00017490601117446276,
      "loss": 0.8477,
      "step": 70370
    },
    {
      "epoch": 1.235625625449885,
      "grad_norm": 0.09065022812924818,
      "learning_rate": 0.00017489859621278324,
      "loss": 0.8493,
      "step": 70380
    },
    {
      "epoch": 1.2358011903298864,
      "grad_norm": 0.06407011216079457,
      "learning_rate": 0.00017489118031479436,
      "loss": 0.8488,
      "step": 70390
    },
    {
      "epoch": 1.235976755209888,
      "grad_norm": 0.0771491240113959,
      "learning_rate": 0.00017488376348059005,
      "loss": 0.8506,
      "step": 70400
    },
    {
      "epoch": 1.2361523200898892,
      "grad_norm": 0.06372160846360474,
      "learning_rate": 0.0001748763457102643,
      "loss": 0.8549,
      "step": 70410
    },
    {
      "epoch": 1.2363278849698907,
      "grad_norm": 0.06745477697592123,
      "learning_rate": 0.0001748689270039111,
      "loss": 0.8537,
      "step": 70420
    },
    {
      "epoch": 1.236503449849892,
      "grad_norm": 0.061698903589906214,
      "learning_rate": 0.00017486150736162443,
      "loss": 0.8472,
      "step": 70430
    },
    {
      "epoch": 1.2366790147298934,
      "grad_norm": 0.0572284764718278,
      "learning_rate": 0.00017485408678349828,
      "loss": 0.8528,
      "step": 70440
    },
    {
      "epoch": 1.2368545796098949,
      "grad_norm": 0.05135653843910542,
      "learning_rate": 0.00017484666526962675,
      "loss": 0.8467,
      "step": 70450
    },
    {
      "epoch": 1.2370301444898963,
      "grad_norm": 0.0664857801551169,
      "learning_rate": 0.0001748392428201038,
      "loss": 0.857,
      "step": 70460
    },
    {
      "epoch": 1.2372057093698976,
      "grad_norm": 0.06218381747685543,
      "learning_rate": 0.0001748318194350235,
      "loss": 0.852,
      "step": 70470
    },
    {
      "epoch": 1.237381274249899,
      "grad_norm": 0.05046606942055646,
      "learning_rate": 0.00017482439511447984,
      "loss": 0.8487,
      "step": 70480
    },
    {
      "epoch": 1.2375568391299003,
      "grad_norm": 0.07261251853068028,
      "learning_rate": 0.000174816969858567,
      "loss": 0.8475,
      "step": 70490
    },
    {
      "epoch": 1.2377324040099018,
      "grad_norm": 0.05705039149602674,
      "learning_rate": 0.000174809543667379,
      "loss": 0.8483,
      "step": 70500
    },
    {
      "epoch": 1.2379079688899033,
      "grad_norm": 0.05744975802254922,
      "learning_rate": 0.0001748021165410099,
      "loss": 0.8524,
      "step": 70510
    },
    {
      "epoch": 1.2380835337699048,
      "grad_norm": 0.04942967934115302,
      "learning_rate": 0.00017479468847955387,
      "loss": 0.8438,
      "step": 70520
    },
    {
      "epoch": 1.238259098649906,
      "grad_norm": 0.0529200285852686,
      "learning_rate": 0.000174787259483105,
      "loss": 0.8443,
      "step": 70530
    },
    {
      "epoch": 1.2384346635299075,
      "grad_norm": 0.07003674977585658,
      "learning_rate": 0.0001747798295517574,
      "loss": 0.8584,
      "step": 70540
    },
    {
      "epoch": 1.238610228409909,
      "grad_norm": 0.0659436832989422,
      "learning_rate": 0.0001747723986856052,
      "loss": 0.8509,
      "step": 70550
    },
    {
      "epoch": 1.2387857932899102,
      "grad_norm": 0.06758444623126689,
      "learning_rate": 0.0001747649668847426,
      "loss": 0.8394,
      "step": 70560
    },
    {
      "epoch": 1.2389613581699117,
      "grad_norm": 0.05759868103867673,
      "learning_rate": 0.00017475753414926371,
      "loss": 0.8494,
      "step": 70570
    },
    {
      "epoch": 1.2391369230499132,
      "grad_norm": 0.06909016873809708,
      "learning_rate": 0.00017475010047926276,
      "loss": 0.8481,
      "step": 70580
    },
    {
      "epoch": 1.2393124879299144,
      "grad_norm": 0.062322926438455516,
      "learning_rate": 0.00017474266587483387,
      "loss": 0.8484,
      "step": 70590
    },
    {
      "epoch": 1.239488052809916,
      "grad_norm": 0.05745948533557679,
      "learning_rate": 0.0001747352303360713,
      "loss": 0.8507,
      "step": 70600
    },
    {
      "epoch": 1.2396636176899174,
      "grad_norm": 0.0738878811538337,
      "learning_rate": 0.0001747277938630692,
      "loss": 0.8534,
      "step": 70610
    },
    {
      "epoch": 1.2398391825699187,
      "grad_norm": 0.055996679486077755,
      "learning_rate": 0.00017472035645592183,
      "loss": 0.8524,
      "step": 70620
    },
    {
      "epoch": 1.2400147474499201,
      "grad_norm": 0.0632894600364455,
      "learning_rate": 0.00017471291811472342,
      "loss": 0.852,
      "step": 70630
    },
    {
      "epoch": 1.2401903123299216,
      "grad_norm": 0.06862150495351448,
      "learning_rate": 0.00017470547883956825,
      "loss": 0.8519,
      "step": 70640
    },
    {
      "epoch": 1.2403658772099229,
      "grad_norm": 0.047142137257782664,
      "learning_rate": 0.0001746980386305505,
      "loss": 0.8553,
      "step": 70650
    },
    {
      "epoch": 1.2405414420899243,
      "grad_norm": 0.06174881842513459,
      "learning_rate": 0.00017469059748776445,
      "loss": 0.853,
      "step": 70660
    },
    {
      "epoch": 1.2407170069699258,
      "grad_norm": 0.06577101742433711,
      "learning_rate": 0.00017468315541130445,
      "loss": 0.847,
      "step": 70670
    },
    {
      "epoch": 1.240892571849927,
      "grad_norm": 0.07198566263399268,
      "learning_rate": 0.00017467571240126472,
      "loss": 0.8499,
      "step": 70680
    },
    {
      "epoch": 1.2410681367299286,
      "grad_norm": 0.061125696759248116,
      "learning_rate": 0.0001746682684577396,
      "loss": 0.848,
      "step": 70690
    },
    {
      "epoch": 1.24124370160993,
      "grad_norm": 0.07351507050691881,
      "learning_rate": 0.00017466082358082342,
      "loss": 0.8482,
      "step": 70700
    },
    {
      "epoch": 1.2414192664899313,
      "grad_norm": 0.061026096355788685,
      "learning_rate": 0.00017465337777061048,
      "loss": 0.8482,
      "step": 70710
    },
    {
      "epoch": 1.2415948313699328,
      "grad_norm": 0.04879252720602575,
      "learning_rate": 0.00017464593102719514,
      "loss": 0.8514,
      "step": 70720
    },
    {
      "epoch": 1.2417703962499342,
      "grad_norm": 0.05941037682008558,
      "learning_rate": 0.0001746384833506717,
      "loss": 0.8482,
      "step": 70730
    },
    {
      "epoch": 1.2419459611299355,
      "grad_norm": 0.06936732893192446,
      "learning_rate": 0.00017463103474113457,
      "loss": 0.8514,
      "step": 70740
    },
    {
      "epoch": 1.242121526009937,
      "grad_norm": 0.05680575683579412,
      "learning_rate": 0.00017462358519867817,
      "loss": 0.8472,
      "step": 70750
    },
    {
      "epoch": 1.2422970908899384,
      "grad_norm": 0.05068548345993255,
      "learning_rate": 0.00017461613472339677,
      "loss": 0.8483,
      "step": 70760
    },
    {
      "epoch": 1.2424726557699397,
      "grad_norm": 0.05453043289697494,
      "learning_rate": 0.00017460868331538489,
      "loss": 0.8456,
      "step": 70770
    },
    {
      "epoch": 1.2426482206499412,
      "grad_norm": 0.0863670922665748,
      "learning_rate": 0.00017460123097473685,
      "loss": 0.8442,
      "step": 70780
    },
    {
      "epoch": 1.2428237855299427,
      "grad_norm": 0.05510072298296477,
      "learning_rate": 0.00017459377770154714,
      "loss": 0.8557,
      "step": 70790
    },
    {
      "epoch": 1.242999350409944,
      "grad_norm": 0.05307542057323678,
      "learning_rate": 0.0001745863234959101,
      "loss": 0.8485,
      "step": 70800
    },
    {
      "epoch": 1.2431749152899454,
      "grad_norm": 0.08164823524092528,
      "learning_rate": 0.00017457886835792029,
      "loss": 0.8493,
      "step": 70810
    },
    {
      "epoch": 1.2433504801699469,
      "grad_norm": 0.0660909094560127,
      "learning_rate": 0.00017457141228767212,
      "loss": 0.8473,
      "step": 70820
    },
    {
      "epoch": 1.2435260450499481,
      "grad_norm": 0.054728226637645476,
      "learning_rate": 0.00017456395528526002,
      "loss": 0.848,
      "step": 70830
    },
    {
      "epoch": 1.2437016099299496,
      "grad_norm": 0.05591165377019224,
      "learning_rate": 0.00017455649735077854,
      "loss": 0.8453,
      "step": 70840
    },
    {
      "epoch": 1.243877174809951,
      "grad_norm": 0.047877847652184814,
      "learning_rate": 0.00017454903848432215,
      "loss": 0.8461,
      "step": 70850
    },
    {
      "epoch": 1.2440527396899523,
      "grad_norm": 0.05323197383776144,
      "learning_rate": 0.00017454157868598532,
      "loss": 0.8486,
      "step": 70860
    },
    {
      "epoch": 1.2442283045699538,
      "grad_norm": 0.05526286192030539,
      "learning_rate": 0.0001745341179558626,
      "loss": 0.855,
      "step": 70870
    },
    {
      "epoch": 1.2444038694499553,
      "grad_norm": 0.06282451069475004,
      "learning_rate": 0.0001745266562940485,
      "loss": 0.8532,
      "step": 70880
    },
    {
      "epoch": 1.2445794343299565,
      "grad_norm": 0.05436100878952081,
      "learning_rate": 0.0001745191937006376,
      "loss": 0.8503,
      "step": 70890
    },
    {
      "epoch": 1.244754999209958,
      "grad_norm": 0.05525344173821724,
      "learning_rate": 0.0001745117301757244,
      "loss": 0.8547,
      "step": 70900
    },
    {
      "epoch": 1.2449305640899595,
      "grad_norm": 0.06844423838826798,
      "learning_rate": 0.0001745042657194035,
      "loss": 0.8491,
      "step": 70910
    },
    {
      "epoch": 1.2451061289699608,
      "grad_norm": 0.05967991118706802,
      "learning_rate": 0.00017449680033176948,
      "loss": 0.8525,
      "step": 70920
    },
    {
      "epoch": 1.2452816938499622,
      "grad_norm": 0.05788535414473954,
      "learning_rate": 0.0001744893340129169,
      "loss": 0.8565,
      "step": 70930
    },
    {
      "epoch": 1.2454572587299637,
      "grad_norm": 0.09561551243364096,
      "learning_rate": 0.00017448186676294038,
      "loss": 0.8458,
      "step": 70940
    },
    {
      "epoch": 1.245632823609965,
      "grad_norm": 0.062307128708685156,
      "learning_rate": 0.00017447439858193453,
      "loss": 0.8526,
      "step": 70950
    },
    {
      "epoch": 1.2458083884899664,
      "grad_norm": 0.1266215381305119,
      "learning_rate": 0.00017446692946999397,
      "loss": 0.8426,
      "step": 70960
    },
    {
      "epoch": 1.245983953369968,
      "grad_norm": 0.04754469088505023,
      "learning_rate": 0.0001744594594272133,
      "loss": 0.8473,
      "step": 70970
    },
    {
      "epoch": 1.2461595182499692,
      "grad_norm": 0.05318588050604438,
      "learning_rate": 0.00017445198845368726,
      "loss": 0.8507,
      "step": 70980
    },
    {
      "epoch": 1.2463350831299707,
      "grad_norm": 0.047214477480335686,
      "learning_rate": 0.00017444451654951039,
      "loss": 0.8556,
      "step": 70990
    },
    {
      "epoch": 1.2465106480099721,
      "grad_norm": 0.06652059786883814,
      "learning_rate": 0.00017443704371477748,
      "loss": 0.8499,
      "step": 71000
    },
    {
      "epoch": 1.2466862128899736,
      "grad_norm": 0.06874009624408782,
      "learning_rate": 0.00017442956994958311,
      "loss": 0.8439,
      "step": 71010
    },
    {
      "epoch": 1.2468617777699749,
      "grad_norm": 0.06496633284639548,
      "learning_rate": 0.00017442209525402201,
      "loss": 0.8604,
      "step": 71020
    },
    {
      "epoch": 1.2470373426499763,
      "grad_norm": 0.057381389808009074,
      "learning_rate": 0.00017441461962818891,
      "loss": 0.8476,
      "step": 71030
    },
    {
      "epoch": 1.2472129075299776,
      "grad_norm": 0.060711137281949976,
      "learning_rate": 0.00017440714307217852,
      "loss": 0.8564,
      "step": 71040
    },
    {
      "epoch": 1.247388472409979,
      "grad_norm": 0.07107096087488948,
      "learning_rate": 0.00017439966558608554,
      "loss": 0.8441,
      "step": 71050
    },
    {
      "epoch": 1.2475640372899806,
      "grad_norm": 0.06098899304150364,
      "learning_rate": 0.00017439218717000472,
      "loss": 0.8451,
      "step": 71060
    },
    {
      "epoch": 1.247739602169982,
      "grad_norm": 0.07390376797116668,
      "learning_rate": 0.00017438470782403084,
      "loss": 0.8491,
      "step": 71070
    },
    {
      "epoch": 1.2479151670499833,
      "grad_norm": 0.06814123298171698,
      "learning_rate": 0.00017437722754825864,
      "loss": 0.8445,
      "step": 71080
    },
    {
      "epoch": 1.2480907319299848,
      "grad_norm": 0.059562274164934516,
      "learning_rate": 0.00017436974634278292,
      "loss": 0.8493,
      "step": 71090
    },
    {
      "epoch": 1.248266296809986,
      "grad_norm": 0.049702607298005186,
      "learning_rate": 0.00017436226420769843,
      "loss": 0.8545,
      "step": 71100
    },
    {
      "epoch": 1.2484418616899875,
      "grad_norm": 0.05255299711406666,
      "learning_rate": 0.0001743547811431,
      "loss": 0.8416,
      "step": 71110
    },
    {
      "epoch": 1.248617426569989,
      "grad_norm": 0.053848288834945614,
      "learning_rate": 0.00017434729714908246,
      "loss": 0.8458,
      "step": 71120
    },
    {
      "epoch": 1.2487929914499905,
      "grad_norm": 0.06590928651277758,
      "learning_rate": 0.00017433981222574055,
      "loss": 0.8501,
      "step": 71130
    },
    {
      "epoch": 1.2489685563299917,
      "grad_norm": 0.05043799682951517,
      "learning_rate": 0.00017433232637316924,
      "loss": 0.8393,
      "step": 71140
    },
    {
      "epoch": 1.2491441212099932,
      "grad_norm": 0.07898123945906525,
      "learning_rate": 0.00017432483959146325,
      "loss": 0.8528,
      "step": 71150
    },
    {
      "epoch": 1.2493196860899944,
      "grad_norm": 0.061491943679739314,
      "learning_rate": 0.00017431735188071752,
      "loss": 0.8512,
      "step": 71160
    },
    {
      "epoch": 1.249495250969996,
      "grad_norm": 0.051860288119397295,
      "learning_rate": 0.00017430986324102688,
      "loss": 0.8544,
      "step": 71170
    },
    {
      "epoch": 1.2496708158499974,
      "grad_norm": 0.053404323196867345,
      "learning_rate": 0.0001743023736724862,
      "loss": 0.8493,
      "step": 71180
    },
    {
      "epoch": 1.2498463807299989,
      "grad_norm": 0.07860801462238394,
      "learning_rate": 0.0001742948831751904,
      "loss": 0.8401,
      "step": 71190
    },
    {
      "epoch": 1.2500219456100001,
      "grad_norm": 0.06833293188677632,
      "learning_rate": 0.00017428739174923442,
      "loss": 0.8563,
      "step": 71200
    },
    {
      "epoch": 1.2501975104900016,
      "grad_norm": 0.053137995488233856,
      "learning_rate": 0.0001742798993947131,
      "loss": 0.8509,
      "step": 71210
    },
    {
      "epoch": 1.2503730753700029,
      "grad_norm": 0.070047305010806,
      "learning_rate": 0.00017427240611172143,
      "loss": 0.8512,
      "step": 71220
    },
    {
      "epoch": 1.2505486402500043,
      "grad_norm": 0.06353102330969142,
      "learning_rate": 0.0001742649119003543,
      "loss": 0.8533,
      "step": 71230
    },
    {
      "epoch": 1.2507242051300058,
      "grad_norm": 0.07352829558695258,
      "learning_rate": 0.00017425741676070674,
      "loss": 0.8489,
      "step": 71240
    },
    {
      "epoch": 1.2508997700100073,
      "grad_norm": 0.06248178545920373,
      "learning_rate": 0.00017424992069287366,
      "loss": 0.8482,
      "step": 71250
    },
    {
      "epoch": 1.2510753348900086,
      "grad_norm": 0.07637779754864395,
      "learning_rate": 0.00017424242369695005,
      "loss": 0.8565,
      "step": 71260
    },
    {
      "epoch": 1.25125089977001,
      "grad_norm": 0.07937740898637931,
      "learning_rate": 0.00017423492577303088,
      "loss": 0.8469,
      "step": 71270
    },
    {
      "epoch": 1.2514264646500113,
      "grad_norm": 0.07784489406771485,
      "learning_rate": 0.00017422742692121116,
      "loss": 0.8467,
      "step": 71280
    },
    {
      "epoch": 1.2516020295300128,
      "grad_norm": 0.08986918358172007,
      "learning_rate": 0.00017421992714158592,
      "loss": 0.8418,
      "step": 71290
    },
    {
      "epoch": 1.2517775944100142,
      "grad_norm": 0.08553028143359326,
      "learning_rate": 0.00017421242643425017,
      "loss": 0.8516,
      "step": 71300
    },
    {
      "epoch": 1.2519531592900157,
      "grad_norm": 0.055998971431877764,
      "learning_rate": 0.00017420492479929895,
      "loss": 0.8491,
      "step": 71310
    },
    {
      "epoch": 1.252128724170017,
      "grad_norm": 0.051181080075843675,
      "learning_rate": 0.0001741974222368273,
      "loss": 0.8509,
      "step": 71320
    },
    {
      "epoch": 1.2523042890500184,
      "grad_norm": 0.06671159670498754,
      "learning_rate": 0.00017418991874693028,
      "loss": 0.8529,
      "step": 71330
    },
    {
      "epoch": 1.2524798539300197,
      "grad_norm": 0.0906317826475553,
      "learning_rate": 0.00017418241432970295,
      "loss": 0.8478,
      "step": 71340
    },
    {
      "epoch": 1.2526554188100212,
      "grad_norm": 0.0493578496941245,
      "learning_rate": 0.0001741749089852404,
      "loss": 0.8441,
      "step": 71350
    },
    {
      "epoch": 1.2528309836900227,
      "grad_norm": 0.07040135865410757,
      "learning_rate": 0.00017416740271363777,
      "loss": 0.852,
      "step": 71360
    },
    {
      "epoch": 1.2530065485700241,
      "grad_norm": 0.0587252911181794,
      "learning_rate": 0.0001741598955149901,
      "loss": 0.8509,
      "step": 71370
    },
    {
      "epoch": 1.2531821134500254,
      "grad_norm": 0.08036325179727981,
      "learning_rate": 0.00017415238738939255,
      "loss": 0.8481,
      "step": 71380
    },
    {
      "epoch": 1.2533576783300269,
      "grad_norm": 0.04560209005454114,
      "learning_rate": 0.00017414487833694023,
      "loss": 0.8492,
      "step": 71390
    },
    {
      "epoch": 1.2535332432100283,
      "grad_norm": 0.05682358485516191,
      "learning_rate": 0.00017413736835772823,
      "loss": 0.8485,
      "step": 71400
    },
    {
      "epoch": 1.2537088080900296,
      "grad_norm": 0.11139356243448319,
      "learning_rate": 0.00017412985745185182,
      "loss": 0.8523,
      "step": 71410
    },
    {
      "epoch": 1.253884372970031,
      "grad_norm": 0.08922348385964067,
      "learning_rate": 0.00017412234561940608,
      "loss": 0.8448,
      "step": 71420
    },
    {
      "epoch": 1.2540599378500326,
      "grad_norm": 0.06361566463303313,
      "learning_rate": 0.00017411483286048618,
      "loss": 0.8543,
      "step": 71430
    },
    {
      "epoch": 1.2542355027300338,
      "grad_norm": 0.06943448399206519,
      "learning_rate": 0.00017410731917518737,
      "loss": 0.8493,
      "step": 71440
    },
    {
      "epoch": 1.2544110676100353,
      "grad_norm": 0.0658537153105197,
      "learning_rate": 0.00017409980456360483,
      "loss": 0.8514,
      "step": 71450
    },
    {
      "epoch": 1.2545866324900368,
      "grad_norm": 0.05314740074453562,
      "learning_rate": 0.0001740922890258337,
      "loss": 0.8399,
      "step": 71460
    },
    {
      "epoch": 1.254762197370038,
      "grad_norm": 0.08451334726423906,
      "learning_rate": 0.0001740847725619693,
      "loss": 0.8476,
      "step": 71470
    },
    {
      "epoch": 1.2549377622500395,
      "grad_norm": 0.07429359906106023,
      "learning_rate": 0.00017407725517210683,
      "loss": 0.8506,
      "step": 71480
    },
    {
      "epoch": 1.255113327130041,
      "grad_norm": 0.053321225304388195,
      "learning_rate": 0.00017406973685634152,
      "loss": 0.8419,
      "step": 71490
    },
    {
      "epoch": 1.2552888920100422,
      "grad_norm": 0.05090160019938407,
      "learning_rate": 0.00017406221761476863,
      "loss": 0.839,
      "step": 71500
    },
    {
      "epoch": 1.2554644568900437,
      "grad_norm": 0.09109227816624284,
      "learning_rate": 0.00017405469744748343,
      "loss": 0.8428,
      "step": 71510
    },
    {
      "epoch": 1.2556400217700452,
      "grad_norm": 0.06279216973962057,
      "learning_rate": 0.00017404717635458123,
      "loss": 0.8466,
      "step": 71520
    },
    {
      "epoch": 1.2558155866500464,
      "grad_norm": 0.07770183779370252,
      "learning_rate": 0.00017403965433615732,
      "loss": 0.8579,
      "step": 71530
    },
    {
      "epoch": 1.255991151530048,
      "grad_norm": 0.04916857738109744,
      "learning_rate": 0.000174032131392307,
      "loss": 0.8524,
      "step": 71540
    },
    {
      "epoch": 1.2561667164100494,
      "grad_norm": 0.05409344758576986,
      "learning_rate": 0.00017402460752312553,
      "loss": 0.8446,
      "step": 71550
    },
    {
      "epoch": 1.2563422812900507,
      "grad_norm": 0.0726233153289712,
      "learning_rate": 0.00017401708272870832,
      "loss": 0.851,
      "step": 71560
    },
    {
      "epoch": 1.2565178461700521,
      "grad_norm": 0.057366658486299925,
      "learning_rate": 0.0001740095570091507,
      "loss": 0.8524,
      "step": 71570
    },
    {
      "epoch": 1.2566934110500536,
      "grad_norm": 0.0632896811212927,
      "learning_rate": 0.00017400203036454793,
      "loss": 0.8499,
      "step": 71580
    },
    {
      "epoch": 1.2568689759300549,
      "grad_norm": 0.06916350002442036,
      "learning_rate": 0.0001739945027949955,
      "loss": 0.8533,
      "step": 71590
    },
    {
      "epoch": 1.2570445408100563,
      "grad_norm": 0.06488734058309144,
      "learning_rate": 0.00017398697430058872,
      "loss": 0.8535,
      "step": 71600
    },
    {
      "epoch": 1.2572201056900578,
      "grad_norm": 0.062070905997827025,
      "learning_rate": 0.00017397944488142298,
      "loss": 0.8532,
      "step": 71610
    },
    {
      "epoch": 1.2573956705700593,
      "grad_norm": 0.05837587774647428,
      "learning_rate": 0.0001739719145375937,
      "loss": 0.8473,
      "step": 71620
    },
    {
      "epoch": 1.2575712354500606,
      "grad_norm": 0.06953716318400378,
      "learning_rate": 0.00017396438326919625,
      "loss": 0.8412,
      "step": 71630
    },
    {
      "epoch": 1.257746800330062,
      "grad_norm": 0.0527386776939109,
      "learning_rate": 0.00017395685107632613,
      "loss": 0.8583,
      "step": 71640
    },
    {
      "epoch": 1.2579223652100633,
      "grad_norm": 0.08197377651957635,
      "learning_rate": 0.0001739493179590787,
      "loss": 0.8509,
      "step": 71650
    },
    {
      "epoch": 1.2580979300900648,
      "grad_norm": 0.05812068512939487,
      "learning_rate": 0.00017394178391754944,
      "loss": 0.8508,
      "step": 71660
    },
    {
      "epoch": 1.2582734949700662,
      "grad_norm": 0.05803741818164181,
      "learning_rate": 0.00017393424895183378,
      "loss": 0.8572,
      "step": 71670
    },
    {
      "epoch": 1.2584490598500677,
      "grad_norm": 0.0610304702355124,
      "learning_rate": 0.0001739267130620272,
      "loss": 0.8531,
      "step": 71680
    },
    {
      "epoch": 1.258624624730069,
      "grad_norm": 0.08607748874491718,
      "learning_rate": 0.00017391917624822522,
      "loss": 0.8521,
      "step": 71690
    },
    {
      "epoch": 1.2588001896100705,
      "grad_norm": 0.05136774917703949,
      "learning_rate": 0.00017391163851052324,
      "loss": 0.8516,
      "step": 71700
    },
    {
      "epoch": 1.2589757544900717,
      "grad_norm": 0.05310295725214159,
      "learning_rate": 0.0001739040998490169,
      "loss": 0.8582,
      "step": 71710
    },
    {
      "epoch": 1.2591513193700732,
      "grad_norm": 0.05909427703800031,
      "learning_rate": 0.00017389656026380162,
      "loss": 0.8442,
      "step": 71720
    },
    {
      "epoch": 1.2593268842500747,
      "grad_norm": 0.06920109910118952,
      "learning_rate": 0.00017388901975497293,
      "loss": 0.8495,
      "step": 71730
    },
    {
      "epoch": 1.2595024491300761,
      "grad_norm": 0.08031423994713356,
      "learning_rate": 0.0001738814783226264,
      "loss": 0.8515,
      "step": 71740
    },
    {
      "epoch": 1.2596780140100774,
      "grad_norm": 0.0780928151247025,
      "learning_rate": 0.00017387393596685763,
      "loss": 0.8615,
      "step": 71750
    },
    {
      "epoch": 1.2598535788900789,
      "grad_norm": 0.10481713102630366,
      "learning_rate": 0.00017386639268776203,
      "loss": 0.8472,
      "step": 71760
    },
    {
      "epoch": 1.2600291437700801,
      "grad_norm": 0.05770982968373696,
      "learning_rate": 0.00017385884848543533,
      "loss": 0.8476,
      "step": 71770
    },
    {
      "epoch": 1.2602047086500816,
      "grad_norm": 0.07262518321790301,
      "learning_rate": 0.00017385130335997303,
      "loss": 0.8484,
      "step": 71780
    },
    {
      "epoch": 1.260380273530083,
      "grad_norm": 0.05512697141255702,
      "learning_rate": 0.00017384375731147077,
      "loss": 0.854,
      "step": 71790
    },
    {
      "epoch": 1.2605558384100846,
      "grad_norm": 0.05680316446851685,
      "learning_rate": 0.00017383621034002412,
      "loss": 0.8455,
      "step": 71800
    },
    {
      "epoch": 1.2607314032900858,
      "grad_norm": 0.13879500673206,
      "learning_rate": 0.00017382866244572875,
      "loss": 0.8396,
      "step": 71810
    },
    {
      "epoch": 1.2609069681700873,
      "grad_norm": 0.056798179717928804,
      "learning_rate": 0.00017382111362868025,
      "loss": 0.8507,
      "step": 71820
    },
    {
      "epoch": 1.2610825330500885,
      "grad_norm": 0.08153001435661585,
      "learning_rate": 0.0001738135638889743,
      "loss": 0.8489,
      "step": 71830
    },
    {
      "epoch": 1.26125809793009,
      "grad_norm": 0.06099042661420578,
      "learning_rate": 0.00017380601322670655,
      "loss": 0.8541,
      "step": 71840
    },
    {
      "epoch": 1.2614336628100915,
      "grad_norm": 0.05206430758727015,
      "learning_rate": 0.00017379846164197263,
      "loss": 0.8541,
      "step": 71850
    },
    {
      "epoch": 1.261609227690093,
      "grad_norm": 0.050678718383597866,
      "learning_rate": 0.00017379090913486827,
      "loss": 0.8462,
      "step": 71860
    },
    {
      "epoch": 1.2617847925700942,
      "grad_norm": 0.05956329717122795,
      "learning_rate": 0.00017378335570548915,
      "loss": 0.8479,
      "step": 71870
    },
    {
      "epoch": 1.2619603574500957,
      "grad_norm": 0.06724492144055447,
      "learning_rate": 0.00017377580135393095,
      "loss": 0.8499,
      "step": 71880
    },
    {
      "epoch": 1.262135922330097,
      "grad_norm": 0.06495932140413674,
      "learning_rate": 0.00017376824608028943,
      "loss": 0.8495,
      "step": 71890
    },
    {
      "epoch": 1.2623114872100984,
      "grad_norm": 0.05800867894715242,
      "learning_rate": 0.00017376068988466024,
      "loss": 0.8459,
      "step": 71900
    },
    {
      "epoch": 1.2624870520901,
      "grad_norm": 0.08027348081651836,
      "learning_rate": 0.00017375313276713918,
      "loss": 0.8547,
      "step": 71910
    },
    {
      "epoch": 1.2626626169701014,
      "grad_norm": 0.057944606714618686,
      "learning_rate": 0.000173745574727822,
      "loss": 0.8401,
      "step": 71920
    },
    {
      "epoch": 1.2628381818501027,
      "grad_norm": 0.07110636409592908,
      "learning_rate": 0.00017373801576680442,
      "loss": 0.8442,
      "step": 71930
    },
    {
      "epoch": 1.2630137467301041,
      "grad_norm": 0.0557716149161617,
      "learning_rate": 0.00017373045588418227,
      "loss": 0.8531,
      "step": 71940
    },
    {
      "epoch": 1.2631893116101054,
      "grad_norm": 0.06008205288788792,
      "learning_rate": 0.0001737228950800513,
      "loss": 0.8437,
      "step": 71950
    },
    {
      "epoch": 1.2633648764901069,
      "grad_norm": 0.07183681043828141,
      "learning_rate": 0.00017371533335450726,
      "loss": 0.8505,
      "step": 71960
    },
    {
      "epoch": 1.2635404413701083,
      "grad_norm": 0.05257656848204088,
      "learning_rate": 0.00017370777070764608,
      "loss": 0.8572,
      "step": 71970
    },
    {
      "epoch": 1.2637160062501098,
      "grad_norm": 0.04864732401223535,
      "learning_rate": 0.00017370020713956347,
      "loss": 0.8533,
      "step": 71980
    },
    {
      "epoch": 1.263891571130111,
      "grad_norm": 0.05519761310507869,
      "learning_rate": 0.0001736926426503553,
      "loss": 0.8559,
      "step": 71990
    },
    {
      "epoch": 1.2640671360101126,
      "grad_norm": 0.0579481539704185,
      "learning_rate": 0.00017368507724011743,
      "loss": 0.8517,
      "step": 72000
    },
    {
      "epoch": 1.2642427008901138,
      "grad_norm": 0.06939207835880319,
      "learning_rate": 0.0001736775109089457,
      "loss": 0.8465,
      "step": 72010
    },
    {
      "epoch": 1.2644182657701153,
      "grad_norm": 0.0553914543329916,
      "learning_rate": 0.00017366994365693598,
      "loss": 0.8485,
      "step": 72020
    },
    {
      "epoch": 1.2645938306501168,
      "grad_norm": 0.07266915344246179,
      "learning_rate": 0.00017366237548418416,
      "loss": 0.8493,
      "step": 72030
    },
    {
      "epoch": 1.2647693955301182,
      "grad_norm": 0.0637203162671689,
      "learning_rate": 0.0001736548063907861,
      "loss": 0.8486,
      "step": 72040
    },
    {
      "epoch": 1.2649449604101195,
      "grad_norm": 0.06736655417740198,
      "learning_rate": 0.00017364723637683773,
      "loss": 0.8448,
      "step": 72050
    },
    {
      "epoch": 1.265120525290121,
      "grad_norm": 0.06406376781908135,
      "learning_rate": 0.00017363966544243494,
      "loss": 0.8466,
      "step": 72060
    },
    {
      "epoch": 1.2652960901701222,
      "grad_norm": 0.06831457300789981,
      "learning_rate": 0.0001736320935876737,
      "loss": 0.853,
      "step": 72070
    },
    {
      "epoch": 1.2654716550501237,
      "grad_norm": 0.1024071017555008,
      "learning_rate": 0.00017362452081264986,
      "loss": 0.8516,
      "step": 72080
    },
    {
      "epoch": 1.2656472199301252,
      "grad_norm": 0.05422877947563077,
      "learning_rate": 0.00017361694711745944,
      "loss": 0.8532,
      "step": 72090
    },
    {
      "epoch": 1.2658227848101267,
      "grad_norm": 0.047274158847977185,
      "learning_rate": 0.0001736093725021984,
      "loss": 0.8397,
      "step": 72100
    },
    {
      "epoch": 1.265998349690128,
      "grad_norm": 0.05971160516134048,
      "learning_rate": 0.0001736017969669627,
      "loss": 0.8516,
      "step": 72110
    },
    {
      "epoch": 1.2661739145701294,
      "grad_norm": 0.0576977159343568,
      "learning_rate": 0.00017359422051184832,
      "loss": 0.8541,
      "step": 72120
    },
    {
      "epoch": 1.2663494794501309,
      "grad_norm": 0.06110890120661097,
      "learning_rate": 0.00017358664313695123,
      "loss": 0.8505,
      "step": 72130
    },
    {
      "epoch": 1.2665250443301321,
      "grad_norm": 0.06651129925774067,
      "learning_rate": 0.00017357906484236746,
      "loss": 0.8618,
      "step": 72140
    },
    {
      "epoch": 1.2667006092101336,
      "grad_norm": 0.1012973574493652,
      "learning_rate": 0.00017357148562819306,
      "loss": 0.8614,
      "step": 72150
    },
    {
      "epoch": 1.266876174090135,
      "grad_norm": 0.054266394096495855,
      "learning_rate": 0.00017356390549452398,
      "loss": 0.8544,
      "step": 72160
    },
    {
      "epoch": 1.2670517389701363,
      "grad_norm": 0.0718110344774396,
      "learning_rate": 0.00017355632444145634,
      "loss": 0.8517,
      "step": 72170
    },
    {
      "epoch": 1.2672273038501378,
      "grad_norm": 0.048974919094991305,
      "learning_rate": 0.00017354874246908617,
      "loss": 0.8526,
      "step": 72180
    },
    {
      "epoch": 1.2674028687301393,
      "grad_norm": 0.04972756332151006,
      "learning_rate": 0.00017354115957750953,
      "loss": 0.8538,
      "step": 72190
    },
    {
      "epoch": 1.2675784336101406,
      "grad_norm": 0.05420287367649407,
      "learning_rate": 0.00017353357576682248,
      "loss": 0.8441,
      "step": 72200
    },
    {
      "epoch": 1.267753998490142,
      "grad_norm": 0.06134042765217292,
      "learning_rate": 0.00017352599103712113,
      "loss": 0.8608,
      "step": 72210
    },
    {
      "epoch": 1.2679295633701435,
      "grad_norm": 0.06547837576705963,
      "learning_rate": 0.00017351840538850158,
      "loss": 0.8449,
      "step": 72220
    },
    {
      "epoch": 1.2681051282501448,
      "grad_norm": 0.08289446739796197,
      "learning_rate": 0.0001735108188210599,
      "loss": 0.8487,
      "step": 72230
    },
    {
      "epoch": 1.2682806931301462,
      "grad_norm": 0.06848268757865146,
      "learning_rate": 0.00017350323133489233,
      "loss": 0.8549,
      "step": 72240
    },
    {
      "epoch": 1.2684562580101477,
      "grad_norm": 0.05988075332982882,
      "learning_rate": 0.00017349564293009487,
      "loss": 0.8522,
      "step": 72250
    },
    {
      "epoch": 1.268631822890149,
      "grad_norm": 0.05230366962040503,
      "learning_rate": 0.00017348805360676368,
      "loss": 0.8466,
      "step": 72260
    },
    {
      "epoch": 1.2688073877701505,
      "grad_norm": 0.05397959649639676,
      "learning_rate": 0.00017348046336499503,
      "loss": 0.8452,
      "step": 72270
    },
    {
      "epoch": 1.268982952650152,
      "grad_norm": 0.06766406516841464,
      "learning_rate": 0.00017347287220488499,
      "loss": 0.8501,
      "step": 72280
    },
    {
      "epoch": 1.2691585175301532,
      "grad_norm": 0.061307581258155656,
      "learning_rate": 0.00017346528012652975,
      "loss": 0.8496,
      "step": 72290
    },
    {
      "epoch": 1.2693340824101547,
      "grad_norm": 0.06586352797348763,
      "learning_rate": 0.00017345768713002556,
      "loss": 0.8476,
      "step": 72300
    },
    {
      "epoch": 1.2695096472901561,
      "grad_norm": 0.06585264956194647,
      "learning_rate": 0.00017345009321546856,
      "loss": 0.8406,
      "step": 72310
    },
    {
      "epoch": 1.2696852121701574,
      "grad_norm": 0.05605543023658029,
      "learning_rate": 0.00017344249838295502,
      "loss": 0.857,
      "step": 72320
    },
    {
      "epoch": 1.2698607770501589,
      "grad_norm": 0.04833524288304587,
      "learning_rate": 0.0001734349026325811,
      "loss": 0.8467,
      "step": 72330
    },
    {
      "epoch": 1.2700363419301604,
      "grad_norm": 0.08902168362225377,
      "learning_rate": 0.00017342730596444313,
      "loss": 0.8561,
      "step": 72340
    },
    {
      "epoch": 1.2702119068101618,
      "grad_norm": 0.06151489913496178,
      "learning_rate": 0.0001734197083786373,
      "loss": 0.8496,
      "step": 72350
    },
    {
      "epoch": 1.270387471690163,
      "grad_norm": 0.06622829686583584,
      "learning_rate": 0.00017341210987525986,
      "loss": 0.8468,
      "step": 72360
    },
    {
      "epoch": 1.2705630365701646,
      "grad_norm": 0.052106701269316213,
      "learning_rate": 0.00017340451045440712,
      "loss": 0.8518,
      "step": 72370
    },
    {
      "epoch": 1.2707386014501658,
      "grad_norm": 0.06304563616026228,
      "learning_rate": 0.00017339691011617538,
      "loss": 0.8534,
      "step": 72380
    },
    {
      "epoch": 1.2709141663301673,
      "grad_norm": 0.05243629016220857,
      "learning_rate": 0.0001733893088606609,
      "loss": 0.8546,
      "step": 72390
    },
    {
      "epoch": 1.2710897312101688,
      "grad_norm": 0.0657653151245965,
      "learning_rate": 0.00017338170668796,
      "loss": 0.8526,
      "step": 72400
    },
    {
      "epoch": 1.2712652960901702,
      "grad_norm": 0.062392395702821984,
      "learning_rate": 0.000173374103598169,
      "loss": 0.8552,
      "step": 72410
    },
    {
      "epoch": 1.2714408609701715,
      "grad_norm": 0.06393423153858042,
      "learning_rate": 0.00017336649959138423,
      "loss": 0.8512,
      "step": 72420
    },
    {
      "epoch": 1.271616425850173,
      "grad_norm": 0.05689527049104964,
      "learning_rate": 0.00017335889466770207,
      "loss": 0.8505,
      "step": 72430
    },
    {
      "epoch": 1.2717919907301742,
      "grad_norm": 0.0646077600188929,
      "learning_rate": 0.00017335128882721882,
      "loss": 0.8542,
      "step": 72440
    },
    {
      "epoch": 1.2719675556101757,
      "grad_norm": 0.07986456395887101,
      "learning_rate": 0.0001733436820700309,
      "loss": 0.8555,
      "step": 72450
    },
    {
      "epoch": 1.2721431204901772,
      "grad_norm": 0.08026660209438408,
      "learning_rate": 0.00017333607439623465,
      "loss": 0.852,
      "step": 72460
    },
    {
      "epoch": 1.2723186853701787,
      "grad_norm": 0.05046487354773926,
      "learning_rate": 0.00017332846580592648,
      "loss": 0.8533,
      "step": 72470
    },
    {
      "epoch": 1.27249425025018,
      "grad_norm": 0.07364286914334957,
      "learning_rate": 0.00017332085629920277,
      "loss": 0.8497,
      "step": 72480
    },
    {
      "epoch": 1.2726698151301814,
      "grad_norm": 0.06971278341374511,
      "learning_rate": 0.00017331324587615995,
      "loss": 0.8424,
      "step": 72490
    },
    {
      "epoch": 1.2728453800101827,
      "grad_norm": 0.06711466986083427,
      "learning_rate": 0.00017330563453689445,
      "loss": 0.8435,
      "step": 72500
    },
    {
      "epoch": 1.2730209448901841,
      "grad_norm": 0.0641308819366934,
      "learning_rate": 0.0001732980222815027,
      "loss": 0.8539,
      "step": 72510
    },
    {
      "epoch": 1.2731965097701856,
      "grad_norm": 0.045907248472457224,
      "learning_rate": 0.00017329040911008117,
      "loss": 0.8453,
      "step": 72520
    },
    {
      "epoch": 1.273372074650187,
      "grad_norm": 0.05283710967053061,
      "learning_rate": 0.00017328279502272628,
      "loss": 0.8482,
      "step": 72530
    },
    {
      "epoch": 1.2735476395301883,
      "grad_norm": 0.05916214260815508,
      "learning_rate": 0.00017327518001953457,
      "loss": 0.8518,
      "step": 72540
    },
    {
      "epoch": 1.2737232044101898,
      "grad_norm": 0.06615755680430736,
      "learning_rate": 0.00017326756410060242,
      "loss": 0.8415,
      "step": 72550
    },
    {
      "epoch": 1.273898769290191,
      "grad_norm": 0.07144374967754195,
      "learning_rate": 0.0001732599472660264,
      "loss": 0.8471,
      "step": 72560
    },
    {
      "epoch": 1.2740743341701926,
      "grad_norm": 0.046337313769100455,
      "learning_rate": 0.00017325232951590302,
      "loss": 0.8459,
      "step": 72570
    },
    {
      "epoch": 1.274249899050194,
      "grad_norm": 0.055423655857742615,
      "learning_rate": 0.00017324471085032876,
      "loss": 0.8594,
      "step": 72580
    },
    {
      "epoch": 1.2744254639301955,
      "grad_norm": 0.05130429995609372,
      "learning_rate": 0.00017323709126940018,
      "loss": 0.8552,
      "step": 72590
    },
    {
      "epoch": 1.2746010288101968,
      "grad_norm": 0.05594185927898379,
      "learning_rate": 0.00017322947077321382,
      "loss": 0.8591,
      "step": 72600
    },
    {
      "epoch": 1.2747765936901982,
      "grad_norm": 0.06823069283042688,
      "learning_rate": 0.00017322184936186617,
      "loss": 0.8543,
      "step": 72610
    },
    {
      "epoch": 1.2749521585701995,
      "grad_norm": 0.06239265888770375,
      "learning_rate": 0.0001732142270354539,
      "loss": 0.8454,
      "step": 72620
    },
    {
      "epoch": 1.275127723450201,
      "grad_norm": 0.05646378767194004,
      "learning_rate": 0.00017320660379407352,
      "loss": 0.8497,
      "step": 72630
    },
    {
      "epoch": 1.2753032883302025,
      "grad_norm": 0.07257632681849639,
      "learning_rate": 0.00017319897963782163,
      "loss": 0.8573,
      "step": 72640
    },
    {
      "epoch": 1.275478853210204,
      "grad_norm": 0.06949373634403548,
      "learning_rate": 0.00017319135456679482,
      "loss": 0.8564,
      "step": 72650
    },
    {
      "epoch": 1.2756544180902052,
      "grad_norm": 0.058246926931712614,
      "learning_rate": 0.0001731837285810897,
      "loss": 0.8509,
      "step": 72660
    },
    {
      "epoch": 1.2758299829702067,
      "grad_norm": 0.048165019902746284,
      "learning_rate": 0.00017317610168080295,
      "loss": 0.8526,
      "step": 72670
    },
    {
      "epoch": 1.276005547850208,
      "grad_norm": 0.05413299504504936,
      "learning_rate": 0.00017316847386603112,
      "loss": 0.8467,
      "step": 72680
    },
    {
      "epoch": 1.2761811127302094,
      "grad_norm": 0.053186351151287344,
      "learning_rate": 0.00017316084513687092,
      "loss": 0.8463,
      "step": 72690
    },
    {
      "epoch": 1.2763566776102109,
      "grad_norm": 0.06358401881461759,
      "learning_rate": 0.00017315321549341898,
      "loss": 0.8434,
      "step": 72700
    },
    {
      "epoch": 1.2765322424902124,
      "grad_norm": 0.04697959023258742,
      "learning_rate": 0.00017314558493577195,
      "loss": 0.8504,
      "step": 72710
    },
    {
      "epoch": 1.2767078073702136,
      "grad_norm": 0.051475151524720854,
      "learning_rate": 0.00017313795346402658,
      "loss": 0.8483,
      "step": 72720
    },
    {
      "epoch": 1.276883372250215,
      "grad_norm": 0.05742523074432333,
      "learning_rate": 0.00017313032107827943,
      "loss": 0.8504,
      "step": 72730
    },
    {
      "epoch": 1.2770589371302163,
      "grad_norm": 0.059494853202269486,
      "learning_rate": 0.0001731226877786274,
      "loss": 0.8404,
      "step": 72740
    },
    {
      "epoch": 1.2772345020102178,
      "grad_norm": 0.05955973719935818,
      "learning_rate": 0.00017311505356516703,
      "loss": 0.8465,
      "step": 72750
    },
    {
      "epoch": 1.2774100668902193,
      "grad_norm": 0.04999269014143875,
      "learning_rate": 0.00017310741843799506,
      "loss": 0.844,
      "step": 72760
    },
    {
      "epoch": 1.2775856317702208,
      "grad_norm": 0.08725427444858735,
      "learning_rate": 0.00017309978239720834,
      "loss": 0.8426,
      "step": 72770
    },
    {
      "epoch": 1.277761196650222,
      "grad_norm": 0.05707390072735316,
      "learning_rate": 0.00017309214544290359,
      "loss": 0.8459,
      "step": 72780
    },
    {
      "epoch": 1.2779367615302235,
      "grad_norm": 0.08681256765146149,
      "learning_rate": 0.0001730845075751775,
      "loss": 0.8452,
      "step": 72790
    },
    {
      "epoch": 1.2781123264102248,
      "grad_norm": 0.046286860785253243,
      "learning_rate": 0.00017307686879412687,
      "loss": 0.8398,
      "step": 72800
    },
    {
      "epoch": 1.2782878912902262,
      "grad_norm": 0.07048158332825585,
      "learning_rate": 0.0001730692290998485,
      "loss": 0.8554,
      "step": 72810
    },
    {
      "epoch": 1.2784634561702277,
      "grad_norm": 0.056448096353753065,
      "learning_rate": 0.0001730615884924392,
      "loss": 0.8426,
      "step": 72820
    },
    {
      "epoch": 1.2786390210502292,
      "grad_norm": 0.07271774448539563,
      "learning_rate": 0.0001730539469719958,
      "loss": 0.8472,
      "step": 72830
    },
    {
      "epoch": 1.2788145859302305,
      "grad_norm": 0.05857636678119472,
      "learning_rate": 0.00017304630453861503,
      "loss": 0.8471,
      "step": 72840
    },
    {
      "epoch": 1.278990150810232,
      "grad_norm": 0.060207472891722076,
      "learning_rate": 0.00017303866119239378,
      "loss": 0.8497,
      "step": 72850
    },
    {
      "epoch": 1.2791657156902334,
      "grad_norm": 0.05218615906913941,
      "learning_rate": 0.00017303101693342892,
      "loss": 0.852,
      "step": 72860
    },
    {
      "epoch": 1.2793412805702347,
      "grad_norm": 0.06048029677913812,
      "learning_rate": 0.00017302337176181725,
      "loss": 0.8475,
      "step": 72870
    },
    {
      "epoch": 1.2795168454502361,
      "grad_norm": 0.053816530674992105,
      "learning_rate": 0.00017301572567765566,
      "loss": 0.8484,
      "step": 72880
    },
    {
      "epoch": 1.2796924103302376,
      "grad_norm": 0.056669346164721905,
      "learning_rate": 0.00017300807868104098,
      "loss": 0.8592,
      "step": 72890
    },
    {
      "epoch": 1.2798679752102389,
      "grad_norm": 0.052883452869660376,
      "learning_rate": 0.0001730004307720702,
      "loss": 0.8472,
      "step": 72900
    },
    {
      "epoch": 1.2800435400902404,
      "grad_norm": 0.06702091291346095,
      "learning_rate": 0.00017299278195084014,
      "loss": 0.8503,
      "step": 72910
    },
    {
      "epoch": 1.2802191049702418,
      "grad_norm": 0.08140161519254802,
      "learning_rate": 0.00017298513221744775,
      "loss": 0.8521,
      "step": 72920
    },
    {
      "epoch": 1.280394669850243,
      "grad_norm": 0.08533559243539675,
      "learning_rate": 0.0001729774815719899,
      "loss": 0.8432,
      "step": 72930
    },
    {
      "epoch": 1.2805702347302446,
      "grad_norm": 0.10022293730734241,
      "learning_rate": 0.00017296983001456362,
      "loss": 0.8525,
      "step": 72940
    },
    {
      "epoch": 1.280745799610246,
      "grad_norm": 0.05012401481222619,
      "learning_rate": 0.0001729621775452658,
      "loss": 0.8546,
      "step": 72950
    },
    {
      "epoch": 1.2809213644902473,
      "grad_norm": 0.06816378768973584,
      "learning_rate": 0.00017295452416419334,
      "loss": 0.8538,
      "step": 72960
    },
    {
      "epoch": 1.2810969293702488,
      "grad_norm": 0.07413291488588591,
      "learning_rate": 0.00017294686987144332,
      "loss": 0.8462,
      "step": 72970
    },
    {
      "epoch": 1.2812724942502502,
      "grad_norm": 0.05150579050445229,
      "learning_rate": 0.00017293921466711264,
      "loss": 0.8515,
      "step": 72980
    },
    {
      "epoch": 1.2814480591302515,
      "grad_norm": 0.04975258485050058,
      "learning_rate": 0.00017293155855129831,
      "loss": 0.8571,
      "step": 72990
    },
    {
      "epoch": 1.281623624010253,
      "grad_norm": 0.06778766738961775,
      "learning_rate": 0.00017292390152409738,
      "loss": 0.8522,
      "step": 73000
    },
    {
      "epoch": 1.2817991888902545,
      "grad_norm": 0.05877493633006253,
      "learning_rate": 0.00017291624358560683,
      "loss": 0.8567,
      "step": 73010
    },
    {
      "epoch": 1.2819747537702557,
      "grad_norm": 0.05002161831591634,
      "learning_rate": 0.0001729085847359237,
      "loss": 0.8493,
      "step": 73020
    },
    {
      "epoch": 1.2821503186502572,
      "grad_norm": 0.08775193488987867,
      "learning_rate": 0.000172900924975145,
      "loss": 0.8501,
      "step": 73030
    },
    {
      "epoch": 1.2823258835302587,
      "grad_norm": 0.06104925594190535,
      "learning_rate": 0.00017289326430336782,
      "loss": 0.8476,
      "step": 73040
    },
    {
      "epoch": 1.28250144841026,
      "grad_norm": 0.061826210746164234,
      "learning_rate": 0.00017288560272068918,
      "loss": 0.8497,
      "step": 73050
    },
    {
      "epoch": 1.2826770132902614,
      "grad_norm": 0.08607715501946567,
      "learning_rate": 0.00017287794022720617,
      "loss": 0.848,
      "step": 73060
    },
    {
      "epoch": 1.2828525781702629,
      "grad_norm": 0.04693847877221865,
      "learning_rate": 0.00017287027682301592,
      "loss": 0.8453,
      "step": 73070
    },
    {
      "epoch": 1.2830281430502644,
      "grad_norm": 0.06252067545791616,
      "learning_rate": 0.00017286261250821548,
      "loss": 0.8455,
      "step": 73080
    },
    {
      "epoch": 1.2832037079302656,
      "grad_norm": 0.05886150603115208,
      "learning_rate": 0.00017285494728290192,
      "loss": 0.8519,
      "step": 73090
    },
    {
      "epoch": 1.283379272810267,
      "grad_norm": 0.055847853286046585,
      "learning_rate": 0.00017284728114717244,
      "loss": 0.8486,
      "step": 73100
    },
    {
      "epoch": 1.2835548376902683,
      "grad_norm": 0.07091610610890987,
      "learning_rate": 0.00017283961410112416,
      "loss": 0.8506,
      "step": 73110
    },
    {
      "epoch": 1.2837304025702698,
      "grad_norm": 0.045546295898004685,
      "learning_rate": 0.0001728319461448542,
      "loss": 0.8487,
      "step": 73120
    },
    {
      "epoch": 1.2839059674502713,
      "grad_norm": 0.07422576191386472,
      "learning_rate": 0.00017282427727845968,
      "loss": 0.8482,
      "step": 73130
    },
    {
      "epoch": 1.2840815323302728,
      "grad_norm": 0.08061921439310213,
      "learning_rate": 0.00017281660750203783,
      "loss": 0.8581,
      "step": 73140
    },
    {
      "epoch": 1.284257097210274,
      "grad_norm": 0.08552275451379841,
      "learning_rate": 0.00017280893681568578,
      "loss": 0.8428,
      "step": 73150
    },
    {
      "epoch": 1.2844326620902755,
      "grad_norm": 0.07054857493480682,
      "learning_rate": 0.00017280126521950075,
      "loss": 0.8555,
      "step": 73160
    },
    {
      "epoch": 1.2846082269702768,
      "grad_norm": 0.07914060809130374,
      "learning_rate": 0.00017279359271357993,
      "loss": 0.8462,
      "step": 73170
    },
    {
      "epoch": 1.2847837918502782,
      "grad_norm": 0.09986566742970417,
      "learning_rate": 0.00017278591929802053,
      "loss": 0.8462,
      "step": 73180
    },
    {
      "epoch": 1.2849593567302797,
      "grad_norm": 0.06263158823671339,
      "learning_rate": 0.00017277824497291978,
      "loss": 0.8507,
      "step": 73190
    },
    {
      "epoch": 1.2851349216102812,
      "grad_norm": 0.06134670374982775,
      "learning_rate": 0.0001727705697383749,
      "loss": 0.847,
      "step": 73200
    },
    {
      "epoch": 1.2853104864902825,
      "grad_norm": 0.05246838822047935,
      "learning_rate": 0.00017276289359448316,
      "loss": 0.8515,
      "step": 73210
    },
    {
      "epoch": 1.285486051370284,
      "grad_norm": 0.06058361048257272,
      "learning_rate": 0.00017275521654134176,
      "loss": 0.8544,
      "step": 73220
    },
    {
      "epoch": 1.2856616162502852,
      "grad_norm": 0.06304913951836562,
      "learning_rate": 0.00017274753857904806,
      "loss": 0.8473,
      "step": 73230
    },
    {
      "epoch": 1.2858371811302867,
      "grad_norm": 0.05657239081423363,
      "learning_rate": 0.00017273985970769928,
      "loss": 0.849,
      "step": 73240
    },
    {
      "epoch": 1.2860127460102881,
      "grad_norm": 0.06913965697659578,
      "learning_rate": 0.00017273217992739274,
      "loss": 0.8409,
      "step": 73250
    },
    {
      "epoch": 1.2861883108902896,
      "grad_norm": 0.062354743205830156,
      "learning_rate": 0.00017272449923822573,
      "loss": 0.8473,
      "step": 73260
    },
    {
      "epoch": 1.2863638757702909,
      "grad_norm": 0.04967282649331241,
      "learning_rate": 0.00017271681764029553,
      "loss": 0.848,
      "step": 73270
    },
    {
      "epoch": 1.2865394406502924,
      "grad_norm": 0.06033087017913372,
      "learning_rate": 0.00017270913513369956,
      "loss": 0.848,
      "step": 73280
    },
    {
      "epoch": 1.2867150055302936,
      "grad_norm": 0.05398420280585345,
      "learning_rate": 0.00017270145171853512,
      "loss": 0.8481,
      "step": 73290
    },
    {
      "epoch": 1.286890570410295,
      "grad_norm": 0.07126507686212222,
      "learning_rate": 0.0001726937673948995,
      "loss": 0.859,
      "step": 73300
    },
    {
      "epoch": 1.2870661352902966,
      "grad_norm": 0.05981011155717568,
      "learning_rate": 0.00017268608216289012,
      "loss": 0.8506,
      "step": 73310
    },
    {
      "epoch": 1.287241700170298,
      "grad_norm": 0.07538819519276833,
      "learning_rate": 0.0001726783960226043,
      "loss": 0.8496,
      "step": 73320
    },
    {
      "epoch": 1.2874172650502993,
      "grad_norm": 0.09359454331340025,
      "learning_rate": 0.00017267070897413953,
      "loss": 0.845,
      "step": 73330
    },
    {
      "epoch": 1.2875928299303008,
      "grad_norm": 0.07023328232191793,
      "learning_rate": 0.0001726630210175931,
      "loss": 0.8534,
      "step": 73340
    },
    {
      "epoch": 1.287768394810302,
      "grad_norm": 0.07183785923844149,
      "learning_rate": 0.00017265533215306248,
      "loss": 0.8474,
      "step": 73350
    },
    {
      "epoch": 1.2879439596903035,
      "grad_norm": 0.07464511369042902,
      "learning_rate": 0.00017264764238064507,
      "loss": 0.8501,
      "step": 73360
    },
    {
      "epoch": 1.288119524570305,
      "grad_norm": 0.06811377391890289,
      "learning_rate": 0.00017263995170043828,
      "loss": 0.8507,
      "step": 73370
    },
    {
      "epoch": 1.2882950894503065,
      "grad_norm": 0.05223193210588684,
      "learning_rate": 0.0001726322601125396,
      "loss": 0.8456,
      "step": 73380
    },
    {
      "epoch": 1.2884706543303077,
      "grad_norm": 0.05160386821395264,
      "learning_rate": 0.00017262456761704643,
      "loss": 0.8556,
      "step": 73390
    },
    {
      "epoch": 1.2886462192103092,
      "grad_norm": 0.0742316599104114,
      "learning_rate": 0.0001726168742140563,
      "loss": 0.8446,
      "step": 73400
    },
    {
      "epoch": 1.2888217840903105,
      "grad_norm": 0.060016081976680624,
      "learning_rate": 0.0001726091799036666,
      "loss": 0.8523,
      "step": 73410
    },
    {
      "epoch": 1.288997348970312,
      "grad_norm": 0.05805920348578582,
      "learning_rate": 0.0001726014846859749,
      "loss": 0.8575,
      "step": 73420
    },
    {
      "epoch": 1.2891729138503134,
      "grad_norm": 0.06521551254226224,
      "learning_rate": 0.00017259378856107865,
      "loss": 0.8463,
      "step": 73430
    },
    {
      "epoch": 1.2893484787303149,
      "grad_norm": 0.05966878519005171,
      "learning_rate": 0.00017258609152907538,
      "loss": 0.8482,
      "step": 73440
    },
    {
      "epoch": 1.2895240436103161,
      "grad_norm": 0.07252794059649711,
      "learning_rate": 0.00017257839359006261,
      "loss": 0.8512,
      "step": 73450
    },
    {
      "epoch": 1.2896996084903176,
      "grad_norm": 0.07472958201270627,
      "learning_rate": 0.00017257069474413788,
      "loss": 0.8449,
      "step": 73460
    },
    {
      "epoch": 1.2898751733703189,
      "grad_norm": 0.09011592848803353,
      "learning_rate": 0.00017256299499139873,
      "loss": 0.8432,
      "step": 73470
    },
    {
      "epoch": 1.2900507382503204,
      "grad_norm": 0.061540605353765376,
      "learning_rate": 0.00017255529433194273,
      "loss": 0.849,
      "step": 73480
    },
    {
      "epoch": 1.2902263031303218,
      "grad_norm": 0.07314905093762057,
      "learning_rate": 0.00017254759276586742,
      "loss": 0.8566,
      "step": 73490
    },
    {
      "epoch": 1.2904018680103233,
      "grad_norm": 0.09060770964520681,
      "learning_rate": 0.0001725398902932704,
      "loss": 0.8539,
      "step": 73500
    },
    {
      "epoch": 1.2905774328903246,
      "grad_norm": 0.0558176225167753,
      "learning_rate": 0.00017253218691424925,
      "loss": 0.8527,
      "step": 73510
    },
    {
      "epoch": 1.290752997770326,
      "grad_norm": 0.05762935425478981,
      "learning_rate": 0.00017252448262890162,
      "loss": 0.8569,
      "step": 73520
    },
    {
      "epoch": 1.2909285626503273,
      "grad_norm": 0.05911664024271116,
      "learning_rate": 0.00017251677743732505,
      "loss": 0.8374,
      "step": 73530
    },
    {
      "epoch": 1.2911041275303288,
      "grad_norm": 0.05571685001882873,
      "learning_rate": 0.00017250907133961724,
      "loss": 0.845,
      "step": 73540
    },
    {
      "epoch": 1.2912796924103302,
      "grad_norm": 0.05876544596606894,
      "learning_rate": 0.00017250136433587577,
      "loss": 0.854,
      "step": 73550
    },
    {
      "epoch": 1.2914552572903317,
      "grad_norm": 0.05886615259475377,
      "learning_rate": 0.0001724936564261983,
      "loss": 0.842,
      "step": 73560
    },
    {
      "epoch": 1.291630822170333,
      "grad_norm": 0.06155967664031375,
      "learning_rate": 0.0001724859476106825,
      "loss": 0.8504,
      "step": 73570
    },
    {
      "epoch": 1.2918063870503345,
      "grad_norm": 0.058761895654702606,
      "learning_rate": 0.00017247823788942604,
      "loss": 0.8483,
      "step": 73580
    },
    {
      "epoch": 1.291981951930336,
      "grad_norm": 0.06555737423044639,
      "learning_rate": 0.00017247052726252664,
      "loss": 0.8494,
      "step": 73590
    },
    {
      "epoch": 1.2921575168103372,
      "grad_norm": 0.04532774699752811,
      "learning_rate": 0.00017246281573008194,
      "loss": 0.8515,
      "step": 73600
    },
    {
      "epoch": 1.2923330816903387,
      "grad_norm": 0.059993960623318146,
      "learning_rate": 0.00017245510329218968,
      "loss": 0.8563,
      "step": 73610
    },
    {
      "epoch": 1.2925086465703401,
      "grad_norm": 0.06709992219917356,
      "learning_rate": 0.00017244738994894756,
      "loss": 0.8506,
      "step": 73620
    },
    {
      "epoch": 1.2926842114503414,
      "grad_norm": 0.04893265864291791,
      "learning_rate": 0.0001724396757004533,
      "loss": 0.8487,
      "step": 73630
    },
    {
      "epoch": 1.2928597763303429,
      "grad_norm": 0.06381086854309087,
      "learning_rate": 0.00017243196054680468,
      "loss": 0.8514,
      "step": 73640
    },
    {
      "epoch": 1.2930353412103444,
      "grad_norm": 0.08792255914605342,
      "learning_rate": 0.00017242424448809943,
      "loss": 0.8538,
      "step": 73650
    },
    {
      "epoch": 1.2932109060903456,
      "grad_norm": 0.06509480969464801,
      "learning_rate": 0.00017241652752443532,
      "loss": 0.8652,
      "step": 73660
    },
    {
      "epoch": 1.293386470970347,
      "grad_norm": 0.05452320234984717,
      "learning_rate": 0.00017240880965591009,
      "loss": 0.8486,
      "step": 73670
    },
    {
      "epoch": 1.2935620358503486,
      "grad_norm": 0.06175166784135403,
      "learning_rate": 0.00017240109088262156,
      "loss": 0.8506,
      "step": 73680
    },
    {
      "epoch": 1.2937376007303498,
      "grad_norm": 0.06977605276281898,
      "learning_rate": 0.00017239337120466758,
      "loss": 0.8449,
      "step": 73690
    },
    {
      "epoch": 1.2939131656103513,
      "grad_norm": 0.05296309382784827,
      "learning_rate": 0.00017238565062214584,
      "loss": 0.8501,
      "step": 73700
    },
    {
      "epoch": 1.2940887304903528,
      "grad_norm": 0.05579145797069548,
      "learning_rate": 0.00017237792913515424,
      "loss": 0.8534,
      "step": 73710
    },
    {
      "epoch": 1.294264295370354,
      "grad_norm": 0.12507311210278219,
      "learning_rate": 0.0001723702067437906,
      "loss": 0.8451,
      "step": 73720
    },
    {
      "epoch": 1.2944398602503555,
      "grad_norm": 0.060054124646705107,
      "learning_rate": 0.0001723624834481528,
      "loss": 0.8496,
      "step": 73730
    },
    {
      "epoch": 1.294615425130357,
      "grad_norm": 0.05347420571166263,
      "learning_rate": 0.00017235475924833862,
      "loss": 0.8537,
      "step": 73740
    },
    {
      "epoch": 1.2947909900103582,
      "grad_norm": 0.07621497645519654,
      "learning_rate": 0.00017234703414444595,
      "loss": 0.8493,
      "step": 73750
    },
    {
      "epoch": 1.2949665548903597,
      "grad_norm": 0.058149868082597075,
      "learning_rate": 0.00017233930813657274,
      "loss": 0.8531,
      "step": 73760
    },
    {
      "epoch": 1.2951421197703612,
      "grad_norm": 0.05304157838711208,
      "learning_rate": 0.0001723315812248168,
      "loss": 0.8537,
      "step": 73770
    },
    {
      "epoch": 1.2953176846503625,
      "grad_norm": 0.06409574173616833,
      "learning_rate": 0.00017232385340927602,
      "loss": 0.8457,
      "step": 73780
    },
    {
      "epoch": 1.295493249530364,
      "grad_norm": 0.05647404112808053,
      "learning_rate": 0.00017231612469004837,
      "loss": 0.8486,
      "step": 73790
    },
    {
      "epoch": 1.2956688144103654,
      "grad_norm": 0.08565247546534585,
      "learning_rate": 0.00017230839506723175,
      "loss": 0.8596,
      "step": 73800
    },
    {
      "epoch": 1.2958443792903669,
      "grad_norm": 0.07005167272926219,
      "learning_rate": 0.0001723006645409241,
      "loss": 0.8479,
      "step": 73810
    },
    {
      "epoch": 1.2960199441703681,
      "grad_norm": 0.0635759074902479,
      "learning_rate": 0.00017229293311122338,
      "loss": 0.8507,
      "step": 73820
    },
    {
      "epoch": 1.2961955090503696,
      "grad_norm": 0.06347821631926169,
      "learning_rate": 0.00017228520077822753,
      "loss": 0.8498,
      "step": 73830
    },
    {
      "epoch": 1.2963710739303709,
      "grad_norm": 0.05369291177329647,
      "learning_rate": 0.0001722774675420345,
      "loss": 0.8425,
      "step": 73840
    },
    {
      "epoch": 1.2965466388103724,
      "grad_norm": 0.06211198560294258,
      "learning_rate": 0.00017226973340274233,
      "loss": 0.8464,
      "step": 73850
    },
    {
      "epoch": 1.2967222036903738,
      "grad_norm": 0.048219313549484964,
      "learning_rate": 0.00017226199836044897,
      "loss": 0.8525,
      "step": 73860
    },
    {
      "epoch": 1.2968977685703753,
      "grad_norm": 0.048478708878928314,
      "learning_rate": 0.00017225426241525242,
      "loss": 0.8536,
      "step": 73870
    },
    {
      "epoch": 1.2970733334503766,
      "grad_norm": 0.07027740719019251,
      "learning_rate": 0.0001722465255672507,
      "loss": 0.8493,
      "step": 73880
    },
    {
      "epoch": 1.297248898330378,
      "grad_norm": 0.055289382373716216,
      "learning_rate": 0.0001722387878165419,
      "loss": 0.8527,
      "step": 73890
    },
    {
      "epoch": 1.2974244632103793,
      "grad_norm": 0.05887788820449988,
      "learning_rate": 0.00017223104916322393,
      "loss": 0.8527,
      "step": 73900
    },
    {
      "epoch": 1.2976000280903808,
      "grad_norm": 0.05288512733886304,
      "learning_rate": 0.00017222330960739497,
      "loss": 0.85,
      "step": 73910
    },
    {
      "epoch": 1.2977755929703823,
      "grad_norm": 0.06513423317048611,
      "learning_rate": 0.00017221556914915302,
      "loss": 0.8559,
      "step": 73920
    },
    {
      "epoch": 1.2979511578503837,
      "grad_norm": 0.07767494054580486,
      "learning_rate": 0.00017220782778859617,
      "loss": 0.8469,
      "step": 73930
    },
    {
      "epoch": 1.298126722730385,
      "grad_norm": 0.05318072931353284,
      "learning_rate": 0.00017220008552582244,
      "loss": 0.8468,
      "step": 73940
    },
    {
      "epoch": 1.2983022876103865,
      "grad_norm": 0.07121327122755636,
      "learning_rate": 0.00017219234236093002,
      "loss": 0.8536,
      "step": 73950
    },
    {
      "epoch": 1.2984778524903877,
      "grad_norm": 0.05510307045987649,
      "learning_rate": 0.00017218459829401695,
      "loss": 0.8507,
      "step": 73960
    },
    {
      "epoch": 1.2986534173703892,
      "grad_norm": 0.05341214117251008,
      "learning_rate": 0.00017217685332518144,
      "loss": 0.8429,
      "step": 73970
    },
    {
      "epoch": 1.2988289822503907,
      "grad_norm": 0.05512429632890281,
      "learning_rate": 0.00017216910745452153,
      "loss": 0.845,
      "step": 73980
    },
    {
      "epoch": 1.2990045471303922,
      "grad_norm": 0.05751227820346895,
      "learning_rate": 0.00017216136068213538,
      "loss": 0.8421,
      "step": 73990
    },
    {
      "epoch": 1.2991801120103934,
      "grad_norm": 0.05181629059815639,
      "learning_rate": 0.00017215361300812113,
      "loss": 0.8501,
      "step": 74000
    },
    {
      "epoch": 1.2993556768903949,
      "grad_norm": 0.06440408512565507,
      "learning_rate": 0.000172145864432577,
      "loss": 0.8518,
      "step": 74010
    },
    {
      "epoch": 1.2995312417703961,
      "grad_norm": 0.07565814868340986,
      "learning_rate": 0.0001721381149556011,
      "loss": 0.8451,
      "step": 74020
    },
    {
      "epoch": 1.2997068066503976,
      "grad_norm": 0.0555412818417386,
      "learning_rate": 0.00017213036457729166,
      "loss": 0.8531,
      "step": 74030
    },
    {
      "epoch": 1.299882371530399,
      "grad_norm": 0.051146012978933425,
      "learning_rate": 0.0001721226132977469,
      "loss": 0.8467,
      "step": 74040
    },
    {
      "epoch": 1.3000579364104006,
      "grad_norm": 0.05056515490857162,
      "learning_rate": 0.00017211486111706498,
      "loss": 0.8526,
      "step": 74050
    },
    {
      "epoch": 1.3002335012904018,
      "grad_norm": 0.05333324766522711,
      "learning_rate": 0.00017210710803534414,
      "loss": 0.8451,
      "step": 74060
    },
    {
      "epoch": 1.3004090661704033,
      "grad_norm": 0.06572787055006905,
      "learning_rate": 0.00017209935405268262,
      "loss": 0.8521,
      "step": 74070
    },
    {
      "epoch": 1.3005846310504046,
      "grad_norm": 0.07445121323941153,
      "learning_rate": 0.00017209159916917866,
      "loss": 0.853,
      "step": 74080
    },
    {
      "epoch": 1.300760195930406,
      "grad_norm": 0.0515126761945955,
      "learning_rate": 0.00017208384338493048,
      "loss": 0.8476,
      "step": 74090
    },
    {
      "epoch": 1.3009357608104075,
      "grad_norm": 0.06760225586881861,
      "learning_rate": 0.0001720760867000364,
      "loss": 0.8496,
      "step": 74100
    },
    {
      "epoch": 1.301111325690409,
      "grad_norm": 0.06060267836063854,
      "learning_rate": 0.00017206832911459473,
      "loss": 0.8423,
      "step": 74110
    },
    {
      "epoch": 1.3012868905704102,
      "grad_norm": 0.06290533002531122,
      "learning_rate": 0.00017206057062870363,
      "loss": 0.8495,
      "step": 74120
    },
    {
      "epoch": 1.3014624554504117,
      "grad_norm": 0.0557818686777932,
      "learning_rate": 0.00017205281124246154,
      "loss": 0.845,
      "step": 74130
    },
    {
      "epoch": 1.301638020330413,
      "grad_norm": 0.05243812395221059,
      "learning_rate": 0.00017204505095596667,
      "loss": 0.8472,
      "step": 74140
    },
    {
      "epoch": 1.3018135852104145,
      "grad_norm": 0.07226005132807674,
      "learning_rate": 0.00017203728976931742,
      "loss": 0.8559,
      "step": 74150
    },
    {
      "epoch": 1.301989150090416,
      "grad_norm": 0.049779600983327704,
      "learning_rate": 0.0001720295276826121,
      "loss": 0.853,
      "step": 74160
    },
    {
      "epoch": 1.3021647149704174,
      "grad_norm": 0.05018129822018999,
      "learning_rate": 0.00017202176469594902,
      "loss": 0.8415,
      "step": 74170
    },
    {
      "epoch": 1.3023402798504187,
      "grad_norm": 0.05006961274925179,
      "learning_rate": 0.0001720140008094266,
      "loss": 0.8442,
      "step": 74180
    },
    {
      "epoch": 1.3025158447304201,
      "grad_norm": 0.050518046628059234,
      "learning_rate": 0.00017200623602314316,
      "loss": 0.852,
      "step": 74190
    },
    {
      "epoch": 1.3026914096104214,
      "grad_norm": 0.053831005675684065,
      "learning_rate": 0.00017199847033719706,
      "loss": 0.8462,
      "step": 74200
    },
    {
      "epoch": 1.3028669744904229,
      "grad_norm": 0.05328959497845417,
      "learning_rate": 0.00017199070375168682,
      "loss": 0.8482,
      "step": 74210
    },
    {
      "epoch": 1.3030425393704244,
      "grad_norm": 0.0543331971644189,
      "learning_rate": 0.00017198293626671072,
      "loss": 0.852,
      "step": 74220
    },
    {
      "epoch": 1.3032181042504258,
      "grad_norm": 0.06351644917449928,
      "learning_rate": 0.0001719751678823672,
      "loss": 0.8529,
      "step": 74230
    },
    {
      "epoch": 1.303393669130427,
      "grad_norm": 0.0739957701330899,
      "learning_rate": 0.00017196739859875473,
      "loss": 0.8517,
      "step": 74240
    },
    {
      "epoch": 1.3035692340104286,
      "grad_norm": 0.11600927742609123,
      "learning_rate": 0.0001719596284159717,
      "loss": 0.8477,
      "step": 74250
    },
    {
      "epoch": 1.3037447988904298,
      "grad_norm": 0.05647655982025251,
      "learning_rate": 0.0001719518573341166,
      "loss": 0.8483,
      "step": 74260
    },
    {
      "epoch": 1.3039203637704313,
      "grad_norm": 0.0875380367113941,
      "learning_rate": 0.00017194408535328785,
      "loss": 0.8563,
      "step": 74270
    },
    {
      "epoch": 1.3040959286504328,
      "grad_norm": 0.05259583727774517,
      "learning_rate": 0.00017193631247358393,
      "loss": 0.8393,
      "step": 74280
    },
    {
      "epoch": 1.3042714935304343,
      "grad_norm": 0.06474860800886133,
      "learning_rate": 0.00017192853869510335,
      "loss": 0.847,
      "step": 74290
    },
    {
      "epoch": 1.3044470584104355,
      "grad_norm": 0.05376025196625303,
      "learning_rate": 0.00017192076401794458,
      "loss": 0.853,
      "step": 74300
    },
    {
      "epoch": 1.304622623290437,
      "grad_norm": 0.05812213251064769,
      "learning_rate": 0.00017191298844220613,
      "loss": 0.85,
      "step": 74310
    },
    {
      "epoch": 1.3047981881704385,
      "grad_norm": 0.05316708782884518,
      "learning_rate": 0.00017190521196798654,
      "loss": 0.852,
      "step": 74320
    },
    {
      "epoch": 1.3049737530504397,
      "grad_norm": 0.051911696992523716,
      "learning_rate": 0.00017189743459538438,
      "loss": 0.8536,
      "step": 74330
    },
    {
      "epoch": 1.3051493179304412,
      "grad_norm": 0.06109475599935442,
      "learning_rate": 0.00017188965632449808,
      "loss": 0.8442,
      "step": 74340
    },
    {
      "epoch": 1.3053248828104427,
      "grad_norm": 0.0532595383462383,
      "learning_rate": 0.00017188187715542626,
      "loss": 0.8441,
      "step": 74350
    },
    {
      "epoch": 1.305500447690444,
      "grad_norm": 0.04807762527855715,
      "learning_rate": 0.00017187409708826744,
      "loss": 0.8413,
      "step": 74360
    },
    {
      "epoch": 1.3056760125704454,
      "grad_norm": 0.07000787920486874,
      "learning_rate": 0.00017186631612312026,
      "loss": 0.8475,
      "step": 74370
    },
    {
      "epoch": 1.3058515774504469,
      "grad_norm": 0.08317030827402234,
      "learning_rate": 0.0001718585342600833,
      "loss": 0.8492,
      "step": 74380
    },
    {
      "epoch": 1.3060271423304481,
      "grad_norm": 0.06819193958522235,
      "learning_rate": 0.00017185075149925508,
      "loss": 0.8478,
      "step": 74390
    },
    {
      "epoch": 1.3062027072104496,
      "grad_norm": 0.06045399599317072,
      "learning_rate": 0.00017184296784073427,
      "loss": 0.853,
      "step": 74400
    },
    {
      "epoch": 1.306378272090451,
      "grad_norm": 0.0735309259029005,
      "learning_rate": 0.0001718351832846195,
      "loss": 0.8504,
      "step": 74410
    },
    {
      "epoch": 1.3065538369704524,
      "grad_norm": 0.059913019340722976,
      "learning_rate": 0.00017182739783100937,
      "loss": 0.8545,
      "step": 74420
    },
    {
      "epoch": 1.3067294018504538,
      "grad_norm": 0.059487900281193995,
      "learning_rate": 0.0001718196114800025,
      "loss": 0.8512,
      "step": 74430
    },
    {
      "epoch": 1.3069049667304553,
      "grad_norm": 0.0852715109939951,
      "learning_rate": 0.00017181182423169764,
      "loss": 0.8528,
      "step": 74440
    },
    {
      "epoch": 1.3070805316104566,
      "grad_norm": 0.056606057857806334,
      "learning_rate": 0.00017180403608619337,
      "loss": 0.8564,
      "step": 74450
    },
    {
      "epoch": 1.307256096490458,
      "grad_norm": 0.04634162129282192,
      "learning_rate": 0.00017179624704358838,
      "loss": 0.8525,
      "step": 74460
    },
    {
      "epoch": 1.3074316613704595,
      "grad_norm": 0.0719560642879591,
      "learning_rate": 0.0001717884571039814,
      "loss": 0.8596,
      "step": 74470
    },
    {
      "epoch": 1.3076072262504608,
      "grad_norm": 0.054642753056798464,
      "learning_rate": 0.00017178066626747108,
      "loss": 0.8462,
      "step": 74480
    },
    {
      "epoch": 1.3077827911304623,
      "grad_norm": 0.0637323641291427,
      "learning_rate": 0.0001717728745341562,
      "loss": 0.8486,
      "step": 74490
    },
    {
      "epoch": 1.3079583560104637,
      "grad_norm": 0.06443681402166967,
      "learning_rate": 0.00017176508190413538,
      "loss": 0.8508,
      "step": 74500
    },
    {
      "epoch": 1.308133920890465,
      "grad_norm": 0.06674804073158158,
      "learning_rate": 0.00017175728837750745,
      "loss": 0.8526,
      "step": 74510
    },
    {
      "epoch": 1.3083094857704665,
      "grad_norm": 0.07923480061522463,
      "learning_rate": 0.00017174949395437104,
      "loss": 0.85,
      "step": 74520
    },
    {
      "epoch": 1.308485050650468,
      "grad_norm": 0.04246419123334665,
      "learning_rate": 0.00017174169863482506,
      "loss": 0.8491,
      "step": 74530
    },
    {
      "epoch": 1.3086606155304694,
      "grad_norm": 0.057107736471411374,
      "learning_rate": 0.00017173390241896817,
      "loss": 0.8392,
      "step": 74540
    },
    {
      "epoch": 1.3088361804104707,
      "grad_norm": 0.06622786972117783,
      "learning_rate": 0.0001717261053068992,
      "loss": 0.8439,
      "step": 74550
    },
    {
      "epoch": 1.3090117452904722,
      "grad_norm": 0.06759733393305513,
      "learning_rate": 0.00017171830729871692,
      "loss": 0.8546,
      "step": 74560
    },
    {
      "epoch": 1.3091873101704734,
      "grad_norm": 0.06972073996836038,
      "learning_rate": 0.0001717105083945201,
      "loss": 0.8518,
      "step": 74570
    },
    {
      "epoch": 1.3093628750504749,
      "grad_norm": 0.051185568305431146,
      "learning_rate": 0.00017170270859440763,
      "loss": 0.8496,
      "step": 74580
    },
    {
      "epoch": 1.3095384399304764,
      "grad_norm": 0.04640943364806615,
      "learning_rate": 0.00017169490789847825,
      "loss": 0.8475,
      "step": 74590
    },
    {
      "epoch": 1.3097140048104778,
      "grad_norm": 0.05578177524568846,
      "learning_rate": 0.0001716871063068309,
      "loss": 0.8591,
      "step": 74600
    },
    {
      "epoch": 1.309889569690479,
      "grad_norm": 0.05531884254570576,
      "learning_rate": 0.0001716793038195643,
      "loss": 0.853,
      "step": 74610
    },
    {
      "epoch": 1.3100651345704806,
      "grad_norm": 0.052473297452198794,
      "learning_rate": 0.0001716715004367774,
      "loss": 0.8519,
      "step": 74620
    },
    {
      "epoch": 1.3102406994504818,
      "grad_norm": 0.05217579122796836,
      "learning_rate": 0.00017166369615856904,
      "loss": 0.8473,
      "step": 74630
    },
    {
      "epoch": 1.3104162643304833,
      "grad_norm": 0.07882802352372614,
      "learning_rate": 0.0001716558909850381,
      "loss": 0.8511,
      "step": 74640
    },
    {
      "epoch": 1.3105918292104848,
      "grad_norm": 0.08054906828775832,
      "learning_rate": 0.0001716480849162835,
      "loss": 0.8565,
      "step": 74650
    },
    {
      "epoch": 1.3107673940904863,
      "grad_norm": 0.06554469989297952,
      "learning_rate": 0.00017164027795240412,
      "loss": 0.8484,
      "step": 74660
    },
    {
      "epoch": 1.3109429589704875,
      "grad_norm": 0.07383733306434374,
      "learning_rate": 0.0001716324700934989,
      "loss": 0.8552,
      "step": 74670
    },
    {
      "epoch": 1.311118523850489,
      "grad_norm": 0.04464345561945777,
      "learning_rate": 0.00017162466133966673,
      "loss": 0.8493,
      "step": 74680
    },
    {
      "epoch": 1.3112940887304902,
      "grad_norm": 0.06419453283806215,
      "learning_rate": 0.00017161685169100654,
      "loss": 0.8524,
      "step": 74690
    },
    {
      "epoch": 1.3114696536104917,
      "grad_norm": 0.06694940559983877,
      "learning_rate": 0.00017160904114761733,
      "loss": 0.8476,
      "step": 74700
    },
    {
      "epoch": 1.3116452184904932,
      "grad_norm": 0.061020155870225586,
      "learning_rate": 0.00017160122970959804,
      "loss": 0.8439,
      "step": 74710
    },
    {
      "epoch": 1.3118207833704947,
      "grad_norm": 0.07065104924649951,
      "learning_rate": 0.0001715934173770476,
      "loss": 0.8518,
      "step": 74720
    },
    {
      "epoch": 1.311996348250496,
      "grad_norm": 0.06635263438364587,
      "learning_rate": 0.0001715856041500651,
      "loss": 0.8382,
      "step": 74730
    },
    {
      "epoch": 1.3121719131304974,
      "grad_norm": 0.05733698818132237,
      "learning_rate": 0.00017157779002874947,
      "loss": 0.8475,
      "step": 74740
    },
    {
      "epoch": 1.3123474780104987,
      "grad_norm": 0.05668802372944086,
      "learning_rate": 0.00017156997501319962,
      "loss": 0.8468,
      "step": 74750
    },
    {
      "epoch": 1.3125230428905001,
      "grad_norm": 0.05880841725375793,
      "learning_rate": 0.00017156215910351473,
      "loss": 0.8488,
      "step": 74760
    },
    {
      "epoch": 1.3126986077705016,
      "grad_norm": 0.05541042146455401,
      "learning_rate": 0.00017155434229979374,
      "loss": 0.8497,
      "step": 74770
    },
    {
      "epoch": 1.312874172650503,
      "grad_norm": 0.07463297570049932,
      "learning_rate": 0.00017154652460213573,
      "loss": 0.8469,
      "step": 74780
    },
    {
      "epoch": 1.3130497375305044,
      "grad_norm": 0.059560277062893206,
      "learning_rate": 0.00017153870601063973,
      "loss": 0.8482,
      "step": 74790
    },
    {
      "epoch": 1.3132253024105058,
      "grad_norm": 0.0664243643520026,
      "learning_rate": 0.0001715308865254048,
      "loss": 0.8455,
      "step": 74800
    },
    {
      "epoch": 1.313400867290507,
      "grad_norm": 0.06735038849521446,
      "learning_rate": 0.00017152306614653002,
      "loss": 0.8541,
      "step": 74810
    },
    {
      "epoch": 1.3135764321705086,
      "grad_norm": 0.059154348481150545,
      "learning_rate": 0.0001715152448741145,
      "loss": 0.8442,
      "step": 74820
    },
    {
      "epoch": 1.31375199705051,
      "grad_norm": 0.06541746199554971,
      "learning_rate": 0.00017150742270825728,
      "loss": 0.8446,
      "step": 74830
    },
    {
      "epoch": 1.3139275619305115,
      "grad_norm": 0.07729713643211511,
      "learning_rate": 0.00017149959964905752,
      "loss": 0.8511,
      "step": 74840
    },
    {
      "epoch": 1.3141031268105128,
      "grad_norm": 0.05765986082939206,
      "learning_rate": 0.0001714917756966143,
      "loss": 0.8454,
      "step": 74850
    },
    {
      "epoch": 1.3142786916905143,
      "grad_norm": 0.0461168926784104,
      "learning_rate": 0.00017148395085102677,
      "loss": 0.8576,
      "step": 74860
    },
    {
      "epoch": 1.3144542565705155,
      "grad_norm": 0.06452079813430821,
      "learning_rate": 0.0001714761251123941,
      "loss": 0.8522,
      "step": 74870
    },
    {
      "epoch": 1.314629821450517,
      "grad_norm": 0.06020677007407519,
      "learning_rate": 0.00017146829848081537,
      "loss": 0.8531,
      "step": 74880
    },
    {
      "epoch": 1.3148053863305185,
      "grad_norm": 0.06988162760331364,
      "learning_rate": 0.00017146047095638987,
      "loss": 0.8482,
      "step": 74890
    },
    {
      "epoch": 1.31498095121052,
      "grad_norm": 0.052769930881010174,
      "learning_rate": 0.00017145264253921665,
      "loss": 0.859,
      "step": 74900
    },
    {
      "epoch": 1.3151565160905212,
      "grad_norm": 0.059686027340659864,
      "learning_rate": 0.00017144481322939496,
      "loss": 0.8516,
      "step": 74910
    },
    {
      "epoch": 1.3153320809705227,
      "grad_norm": 0.0569589783238258,
      "learning_rate": 0.00017143698302702398,
      "loss": 0.855,
      "step": 74920
    },
    {
      "epoch": 1.315507645850524,
      "grad_norm": 0.051733064706925035,
      "learning_rate": 0.00017142915193220292,
      "loss": 0.8518,
      "step": 74930
    },
    {
      "epoch": 1.3156832107305254,
      "grad_norm": 0.06082229339635138,
      "learning_rate": 0.00017142131994503102,
      "loss": 0.8503,
      "step": 74940
    },
    {
      "epoch": 1.3158587756105269,
      "grad_norm": 0.05468611888989276,
      "learning_rate": 0.00017141348706560753,
      "loss": 0.8449,
      "step": 74950
    },
    {
      "epoch": 1.3160343404905284,
      "grad_norm": 0.06370364292548603,
      "learning_rate": 0.0001714056532940316,
      "loss": 0.8546,
      "step": 74960
    },
    {
      "epoch": 1.3162099053705296,
      "grad_norm": 0.04507859516052771,
      "learning_rate": 0.0001713978186304026,
      "loss": 0.8576,
      "step": 74970
    },
    {
      "epoch": 1.316385470250531,
      "grad_norm": 0.06583453293104122,
      "learning_rate": 0.0001713899830748197,
      "loss": 0.8513,
      "step": 74980
    },
    {
      "epoch": 1.3165610351305324,
      "grad_norm": 0.08481209795720843,
      "learning_rate": 0.00017138214662738225,
      "loss": 0.8483,
      "step": 74990
    },
    {
      "epoch": 1.3167366000105338,
      "grad_norm": 0.047362788274666195,
      "learning_rate": 0.0001713743092881895,
      "loss": 0.8577,
      "step": 75000
    },
    {
      "epoch": 1.3169121648905353,
      "grad_norm": 0.06892597456723117,
      "learning_rate": 0.0001713664710573408,
      "loss": 0.8482,
      "step": 75010
    },
    {
      "epoch": 1.3170877297705368,
      "grad_norm": 0.06139764551286314,
      "learning_rate": 0.00017135863193493543,
      "loss": 0.8475,
      "step": 75020
    },
    {
      "epoch": 1.317263294650538,
      "grad_norm": 0.06137213407156039,
      "learning_rate": 0.00017135079192107265,
      "loss": 0.8531,
      "step": 75030
    },
    {
      "epoch": 1.3174388595305395,
      "grad_norm": 0.060681317134714405,
      "learning_rate": 0.00017134295101585193,
      "loss": 0.8551,
      "step": 75040
    },
    {
      "epoch": 1.317614424410541,
      "grad_norm": 0.06958993944619012,
      "learning_rate": 0.00017133510921937245,
      "loss": 0.8512,
      "step": 75050
    },
    {
      "epoch": 1.3177899892905423,
      "grad_norm": 0.07822673505143969,
      "learning_rate": 0.00017132726653173373,
      "loss": 0.8472,
      "step": 75060
    },
    {
      "epoch": 1.3179655541705437,
      "grad_norm": 0.05362696573438668,
      "learning_rate": 0.00017131942295303503,
      "loss": 0.8389,
      "step": 75070
    },
    {
      "epoch": 1.3181411190505452,
      "grad_norm": 0.055315832950675686,
      "learning_rate": 0.00017131157848337576,
      "loss": 0.8456,
      "step": 75080
    },
    {
      "epoch": 1.3183166839305465,
      "grad_norm": 0.08263795549808108,
      "learning_rate": 0.00017130373312285535,
      "loss": 0.8508,
      "step": 75090
    },
    {
      "epoch": 1.318492248810548,
      "grad_norm": 0.0895446594918832,
      "learning_rate": 0.0001712958868715731,
      "loss": 0.8472,
      "step": 75100
    },
    {
      "epoch": 1.3186678136905494,
      "grad_norm": 0.06606916190707574,
      "learning_rate": 0.00017128803972962857,
      "loss": 0.8459,
      "step": 75110
    },
    {
      "epoch": 1.3188433785705507,
      "grad_norm": 0.0848749110091261,
      "learning_rate": 0.00017128019169712107,
      "loss": 0.8511,
      "step": 75120
    },
    {
      "epoch": 1.3190189434505522,
      "grad_norm": 0.08260538880269408,
      "learning_rate": 0.00017127234277415008,
      "loss": 0.8494,
      "step": 75130
    },
    {
      "epoch": 1.3191945083305536,
      "grad_norm": 0.07969226823167924,
      "learning_rate": 0.00017126449296081504,
      "loss": 0.8458,
      "step": 75140
    },
    {
      "epoch": 1.3193700732105549,
      "grad_norm": 0.06702989113783846,
      "learning_rate": 0.0001712566422572154,
      "loss": 0.8447,
      "step": 75150
    },
    {
      "epoch": 1.3195456380905564,
      "grad_norm": 0.055160250417652934,
      "learning_rate": 0.00017124879066345063,
      "loss": 0.8485,
      "step": 75160
    },
    {
      "epoch": 1.3197212029705578,
      "grad_norm": 0.04815133197641721,
      "learning_rate": 0.00017124093817962025,
      "loss": 0.8431,
      "step": 75170
    },
    {
      "epoch": 1.319896767850559,
      "grad_norm": 0.050148633970874046,
      "learning_rate": 0.00017123308480582368,
      "loss": 0.8567,
      "step": 75180
    },
    {
      "epoch": 1.3200723327305606,
      "grad_norm": 0.05325793859302122,
      "learning_rate": 0.00017122523054216048,
      "loss": 0.8415,
      "step": 75190
    },
    {
      "epoch": 1.320247897610562,
      "grad_norm": 0.05880171259868474,
      "learning_rate": 0.00017121737538873018,
      "loss": 0.8428,
      "step": 75200
    },
    {
      "epoch": 1.3204234624905635,
      "grad_norm": 0.056787487794800445,
      "learning_rate": 0.00017120951934563222,
      "loss": 0.8486,
      "step": 75210
    },
    {
      "epoch": 1.3205990273705648,
      "grad_norm": 0.0840854038656484,
      "learning_rate": 0.00017120166241296625,
      "loss": 0.8499,
      "step": 75220
    },
    {
      "epoch": 1.3207745922505663,
      "grad_norm": 0.05880948558264276,
      "learning_rate": 0.00017119380459083174,
      "loss": 0.8421,
      "step": 75230
    },
    {
      "epoch": 1.3209501571305675,
      "grad_norm": 0.062192642466828416,
      "learning_rate": 0.00017118594587932828,
      "loss": 0.8529,
      "step": 75240
    },
    {
      "epoch": 1.321125722010569,
      "grad_norm": 0.0837300327757396,
      "learning_rate": 0.00017117808627855542,
      "loss": 0.8553,
      "step": 75250
    },
    {
      "epoch": 1.3213012868905705,
      "grad_norm": 0.05001919923252674,
      "learning_rate": 0.00017117022578861277,
      "loss": 0.8481,
      "step": 75260
    },
    {
      "epoch": 1.321476851770572,
      "grad_norm": 0.05389523525669501,
      "learning_rate": 0.0001711623644095999,
      "loss": 0.8526,
      "step": 75270
    },
    {
      "epoch": 1.3216524166505732,
      "grad_norm": 0.05312733765982867,
      "learning_rate": 0.00017115450214161645,
      "loss": 0.8541,
      "step": 75280
    },
    {
      "epoch": 1.3218279815305747,
      "grad_norm": 0.06685575979572611,
      "learning_rate": 0.00017114663898476197,
      "loss": 0.8379,
      "step": 75290
    },
    {
      "epoch": 1.322003546410576,
      "grad_norm": 0.08327092701884586,
      "learning_rate": 0.0001711387749391362,
      "loss": 0.8511,
      "step": 75300
    },
    {
      "epoch": 1.3221791112905774,
      "grad_norm": 0.050788046450746485,
      "learning_rate": 0.00017113091000483865,
      "loss": 0.8426,
      "step": 75310
    },
    {
      "epoch": 1.322354676170579,
      "grad_norm": 0.05635260165337142,
      "learning_rate": 0.00017112304418196906,
      "loss": 0.8463,
      "step": 75320
    },
    {
      "epoch": 1.3225302410505804,
      "grad_norm": 0.07633224465194655,
      "learning_rate": 0.00017111517747062704,
      "loss": 0.844,
      "step": 75330
    },
    {
      "epoch": 1.3227058059305816,
      "grad_norm": 0.051350956252422836,
      "learning_rate": 0.00017110730987091228,
      "loss": 0.8432,
      "step": 75340
    },
    {
      "epoch": 1.322881370810583,
      "grad_norm": 0.06620301750755238,
      "learning_rate": 0.00017109944138292448,
      "loss": 0.8494,
      "step": 75350
    },
    {
      "epoch": 1.3230569356905844,
      "grad_norm": 0.048625935971782004,
      "learning_rate": 0.0001710915720067633,
      "loss": 0.8541,
      "step": 75360
    },
    {
      "epoch": 1.3232325005705858,
      "grad_norm": 0.052302443800912896,
      "learning_rate": 0.0001710837017425285,
      "loss": 0.8492,
      "step": 75370
    },
    {
      "epoch": 1.3234080654505873,
      "grad_norm": 0.05221721603430902,
      "learning_rate": 0.00017107583059031975,
      "loss": 0.853,
      "step": 75380
    },
    {
      "epoch": 1.3235836303305888,
      "grad_norm": 0.05744516061685842,
      "learning_rate": 0.00017106795855023683,
      "loss": 0.8523,
      "step": 75390
    },
    {
      "epoch": 1.32375919521059,
      "grad_norm": 0.06223374103100518,
      "learning_rate": 0.00017106008562237942,
      "loss": 0.8433,
      "step": 75400
    },
    {
      "epoch": 1.3239347600905915,
      "grad_norm": 0.0452022893483144,
      "learning_rate": 0.0001710522118068473,
      "loss": 0.8495,
      "step": 75410
    },
    {
      "epoch": 1.3241103249705928,
      "grad_norm": 0.06349290288444069,
      "learning_rate": 0.0001710443371037402,
      "loss": 0.8539,
      "step": 75420
    },
    {
      "epoch": 1.3242858898505943,
      "grad_norm": 0.06189853596805175,
      "learning_rate": 0.000171036461513158,
      "loss": 0.8586,
      "step": 75430
    },
    {
      "epoch": 1.3244614547305957,
      "grad_norm": 0.06463421676971393,
      "learning_rate": 0.00017102858503520037,
      "loss": 0.8507,
      "step": 75440
    },
    {
      "epoch": 1.3246370196105972,
      "grad_norm": 0.07225162696460125,
      "learning_rate": 0.00017102070766996714,
      "loss": 0.8546,
      "step": 75450
    },
    {
      "epoch": 1.3248125844905985,
      "grad_norm": 0.05373625263840432,
      "learning_rate": 0.0001710128294175581,
      "loss": 0.846,
      "step": 75460
    },
    {
      "epoch": 1.3249881493706,
      "grad_norm": 0.08647898200511674,
      "learning_rate": 0.00017100495027807313,
      "loss": 0.8524,
      "step": 75470
    },
    {
      "epoch": 1.3251637142506012,
      "grad_norm": 0.07449623536984011,
      "learning_rate": 0.00017099707025161203,
      "loss": 0.8567,
      "step": 75480
    },
    {
      "epoch": 1.3253392791306027,
      "grad_norm": 0.047321317332670074,
      "learning_rate": 0.0001709891893382746,
      "loss": 0.8454,
      "step": 75490
    },
    {
      "epoch": 1.3255148440106042,
      "grad_norm": 0.07446498584258145,
      "learning_rate": 0.00017098130753816074,
      "loss": 0.8476,
      "step": 75500
    },
    {
      "epoch": 1.3256904088906056,
      "grad_norm": 0.07777551511468009,
      "learning_rate": 0.00017097342485137033,
      "loss": 0.8497,
      "step": 75510
    },
    {
      "epoch": 1.3258659737706069,
      "grad_norm": 0.06325480480226522,
      "learning_rate": 0.00017096554127800319,
      "loss": 0.8462,
      "step": 75520
    },
    {
      "epoch": 1.3260415386506084,
      "grad_norm": 0.058233305011368905,
      "learning_rate": 0.00017095765681815925,
      "loss": 0.8536,
      "step": 75530
    },
    {
      "epoch": 1.3262171035306096,
      "grad_norm": 0.06733164301119036,
      "learning_rate": 0.00017094977147193838,
      "loss": 0.8516,
      "step": 75540
    },
    {
      "epoch": 1.326392668410611,
      "grad_norm": 0.054694509559854367,
      "learning_rate": 0.00017094188523944054,
      "loss": 0.8554,
      "step": 75550
    },
    {
      "epoch": 1.3265682332906126,
      "grad_norm": 0.06916118651118602,
      "learning_rate": 0.00017093399812076555,
      "loss": 0.8455,
      "step": 75560
    },
    {
      "epoch": 1.326743798170614,
      "grad_norm": 0.06442436998713348,
      "learning_rate": 0.0001709261101160134,
      "loss": 0.8509,
      "step": 75570
    },
    {
      "epoch": 1.3269193630506153,
      "grad_norm": 0.059234086415499675,
      "learning_rate": 0.00017091822122528404,
      "loss": 0.8469,
      "step": 75580
    },
    {
      "epoch": 1.3270949279306168,
      "grad_norm": 0.06093072134988412,
      "learning_rate": 0.00017091033144867745,
      "loss": 0.8518,
      "step": 75590
    },
    {
      "epoch": 1.327270492810618,
      "grad_norm": 0.08349800293961679,
      "learning_rate": 0.00017090244078629357,
      "loss": 0.8504,
      "step": 75600
    },
    {
      "epoch": 1.3274460576906195,
      "grad_norm": 0.07547420938459602,
      "learning_rate": 0.0001708945492382323,
      "loss": 0.8456,
      "step": 75610
    },
    {
      "epoch": 1.327621622570621,
      "grad_norm": 0.06560657899751278,
      "learning_rate": 0.00017088665680459376,
      "loss": 0.8436,
      "step": 75620
    },
    {
      "epoch": 1.3277971874506225,
      "grad_norm": 0.05321532587251813,
      "learning_rate": 0.00017087876348547787,
      "loss": 0.8514,
      "step": 75630
    },
    {
      "epoch": 1.3279727523306237,
      "grad_norm": 0.05100014425663975,
      "learning_rate": 0.00017087086928098462,
      "loss": 0.8519,
      "step": 75640
    },
    {
      "epoch": 1.3281483172106252,
      "grad_norm": 0.05090774867351129,
      "learning_rate": 0.0001708629741912141,
      "loss": 0.8481,
      "step": 75650
    },
    {
      "epoch": 1.3283238820906265,
      "grad_norm": 0.06856145094219757,
      "learning_rate": 0.0001708550782162663,
      "loss": 0.8471,
      "step": 75660
    },
    {
      "epoch": 1.328499446970628,
      "grad_norm": 0.06877078546870233,
      "learning_rate": 0.0001708471813562413,
      "loss": 0.8523,
      "step": 75670
    },
    {
      "epoch": 1.3286750118506294,
      "grad_norm": 0.05707509016034957,
      "learning_rate": 0.0001708392836112391,
      "loss": 0.8493,
      "step": 75680
    },
    {
      "epoch": 1.328850576730631,
      "grad_norm": 0.06363413630117916,
      "learning_rate": 0.00017083138498135976,
      "loss": 0.8437,
      "step": 75690
    },
    {
      "epoch": 1.3290261416106322,
      "grad_norm": 0.07496775475136136,
      "learning_rate": 0.00017082348546670345,
      "loss": 0.8468,
      "step": 75700
    },
    {
      "epoch": 1.3292017064906336,
      "grad_norm": 0.062188008819777936,
      "learning_rate": 0.00017081558506737014,
      "loss": 0.8507,
      "step": 75710
    },
    {
      "epoch": 1.3293772713706349,
      "grad_norm": 0.053570335550356896,
      "learning_rate": 0.00017080768378346,
      "loss": 0.8438,
      "step": 75720
    },
    {
      "epoch": 1.3295528362506364,
      "grad_norm": 0.0562176269983682,
      "learning_rate": 0.00017079978161507316,
      "loss": 0.8563,
      "step": 75730
    },
    {
      "epoch": 1.3297284011306378,
      "grad_norm": 0.06544673077535408,
      "learning_rate": 0.0001707918785623097,
      "loss": 0.8517,
      "step": 75740
    },
    {
      "epoch": 1.3299039660106393,
      "grad_norm": 0.04711214661359434,
      "learning_rate": 0.00017078397462526974,
      "loss": 0.8537,
      "step": 75750
    },
    {
      "epoch": 1.3300795308906406,
      "grad_norm": 0.07359694691080831,
      "learning_rate": 0.00017077606980405344,
      "loss": 0.851,
      "step": 75760
    },
    {
      "epoch": 1.330255095770642,
      "grad_norm": 0.06915889132716112,
      "learning_rate": 0.00017076816409876102,
      "loss": 0.8495,
      "step": 75770
    },
    {
      "epoch": 1.3304306606506435,
      "grad_norm": 0.052207577610965276,
      "learning_rate": 0.00017076025750949257,
      "loss": 0.8478,
      "step": 75780
    },
    {
      "epoch": 1.3306062255306448,
      "grad_norm": 0.07479050941318063,
      "learning_rate": 0.00017075235003634828,
      "loss": 0.8476,
      "step": 75790
    },
    {
      "epoch": 1.3307817904106463,
      "grad_norm": 0.08587781839816346,
      "learning_rate": 0.00017074444167942832,
      "loss": 0.8444,
      "step": 75800
    },
    {
      "epoch": 1.3309573552906477,
      "grad_norm": 0.05309213701256835,
      "learning_rate": 0.00017073653243883293,
      "loss": 0.8486,
      "step": 75810
    },
    {
      "epoch": 1.331132920170649,
      "grad_norm": 0.07615311936783091,
      "learning_rate": 0.00017072862231466232,
      "loss": 0.8437,
      "step": 75820
    },
    {
      "epoch": 1.3313084850506505,
      "grad_norm": 0.045941045711390203,
      "learning_rate": 0.0001707207113070167,
      "loss": 0.8433,
      "step": 75830
    },
    {
      "epoch": 1.331484049930652,
      "grad_norm": 0.04574417788801244,
      "learning_rate": 0.00017071279941599632,
      "loss": 0.8503,
      "step": 75840
    },
    {
      "epoch": 1.3316596148106532,
      "grad_norm": 0.07005790519832959,
      "learning_rate": 0.0001707048866417014,
      "loss": 0.8543,
      "step": 75850
    },
    {
      "epoch": 1.3318351796906547,
      "grad_norm": 0.07006032718967621,
      "learning_rate": 0.00017069697298423219,
      "loss": 0.851,
      "step": 75860
    },
    {
      "epoch": 1.3320107445706562,
      "grad_norm": 0.046020134988948413,
      "learning_rate": 0.00017068905844368895,
      "loss": 0.841,
      "step": 75870
    },
    {
      "epoch": 1.3321863094506574,
      "grad_norm": 0.05012177538714212,
      "learning_rate": 0.00017068114302017204,
      "loss": 0.851,
      "step": 75880
    },
    {
      "epoch": 1.332361874330659,
      "grad_norm": 0.05983242716382999,
      "learning_rate": 0.0001706732267137817,
      "loss": 0.8534,
      "step": 75890
    },
    {
      "epoch": 1.3325374392106604,
      "grad_norm": 0.06127554639818103,
      "learning_rate": 0.00017066530952461817,
      "loss": 0.8505,
      "step": 75900
    },
    {
      "epoch": 1.3327130040906616,
      "grad_norm": 0.08113342718931067,
      "learning_rate": 0.00017065739145278187,
      "loss": 0.8499,
      "step": 75910
    },
    {
      "epoch": 1.332888568970663,
      "grad_norm": 0.07558165263718662,
      "learning_rate": 0.00017064947249837304,
      "loss": 0.8444,
      "step": 75920
    },
    {
      "epoch": 1.3330641338506646,
      "grad_norm": 0.08325143110054631,
      "learning_rate": 0.00017064155266149203,
      "loss": 0.8467,
      "step": 75930
    },
    {
      "epoch": 1.333239698730666,
      "grad_norm": 0.07038032390775617,
      "learning_rate": 0.00017063363194223924,
      "loss": 0.8571,
      "step": 75940
    },
    {
      "epoch": 1.3334152636106673,
      "grad_norm": 0.05445567527150817,
      "learning_rate": 0.00017062571034071497,
      "loss": 0.8475,
      "step": 75950
    },
    {
      "epoch": 1.3335908284906688,
      "grad_norm": 0.055884008247775695,
      "learning_rate": 0.0001706177878570196,
      "loss": 0.8545,
      "step": 75960
    },
    {
      "epoch": 1.33376639337067,
      "grad_norm": 0.06862368011146021,
      "learning_rate": 0.00017060986449125352,
      "loss": 0.8534,
      "step": 75970
    },
    {
      "epoch": 1.3339419582506715,
      "grad_norm": 0.0601197366141486,
      "learning_rate": 0.00017060194024351712,
      "loss": 0.8465,
      "step": 75980
    },
    {
      "epoch": 1.334117523130673,
      "grad_norm": 0.08184159845634237,
      "learning_rate": 0.00017059401511391081,
      "loss": 0.8517,
      "step": 75990
    },
    {
      "epoch": 1.3342930880106745,
      "grad_norm": 0.0466538337877709,
      "learning_rate": 0.00017058608910253497,
      "loss": 0.854,
      "step": 76000
    },
    {
      "epoch": 1.3344686528906757,
      "grad_norm": 0.06019853392624011,
      "learning_rate": 0.00017057816220949007,
      "loss": 0.8513,
      "step": 76010
    },
    {
      "epoch": 1.3346442177706772,
      "grad_norm": 0.07320400159744017,
      "learning_rate": 0.0001705702344348765,
      "loss": 0.8393,
      "step": 76020
    },
    {
      "epoch": 1.3348197826506785,
      "grad_norm": 0.05348825942682483,
      "learning_rate": 0.00017056230577879472,
      "loss": 0.8528,
      "step": 76030
    },
    {
      "epoch": 1.33499534753068,
      "grad_norm": 0.08115482382279772,
      "learning_rate": 0.00017055437624134522,
      "loss": 0.8513,
      "step": 76040
    },
    {
      "epoch": 1.3351709124106814,
      "grad_norm": 0.049519905566244936,
      "learning_rate": 0.0001705464458226284,
      "loss": 0.8481,
      "step": 76050
    },
    {
      "epoch": 1.335346477290683,
      "grad_norm": 0.0648811274252808,
      "learning_rate": 0.00017053851452274483,
      "loss": 0.8534,
      "step": 76060
    },
    {
      "epoch": 1.3355220421706842,
      "grad_norm": 0.05307638407914913,
      "learning_rate": 0.00017053058234179494,
      "loss": 0.8474,
      "step": 76070
    },
    {
      "epoch": 1.3356976070506856,
      "grad_norm": 0.04967201707323457,
      "learning_rate": 0.00017052264927987923,
      "loss": 0.8473,
      "step": 76080
    },
    {
      "epoch": 1.3358731719306869,
      "grad_norm": 0.058447173382868076,
      "learning_rate": 0.00017051471533709823,
      "loss": 0.8485,
      "step": 76090
    },
    {
      "epoch": 1.3360487368106884,
      "grad_norm": 0.04825286360093728,
      "learning_rate": 0.00017050678051355248,
      "loss": 0.8506,
      "step": 76100
    },
    {
      "epoch": 1.3362243016906898,
      "grad_norm": 0.0647530000961546,
      "learning_rate": 0.00017049884480934254,
      "loss": 0.8497,
      "step": 76110
    },
    {
      "epoch": 1.3363998665706913,
      "grad_norm": 0.06337974076826894,
      "learning_rate": 0.00017049090822456882,
      "loss": 0.8419,
      "step": 76120
    },
    {
      "epoch": 1.3365754314506926,
      "grad_norm": 0.06425865116558935,
      "learning_rate": 0.00017048297075933206,
      "loss": 0.8467,
      "step": 76130
    },
    {
      "epoch": 1.336750996330694,
      "grad_norm": 0.06962871000513726,
      "learning_rate": 0.0001704750324137327,
      "loss": 0.8437,
      "step": 76140
    },
    {
      "epoch": 1.3369265612106953,
      "grad_norm": 0.04534623110728847,
      "learning_rate": 0.00017046709318787137,
      "loss": 0.856,
      "step": 76150
    },
    {
      "epoch": 1.3371021260906968,
      "grad_norm": 0.07571862872538243,
      "learning_rate": 0.00017045915308184866,
      "loss": 0.8434,
      "step": 76160
    },
    {
      "epoch": 1.3372776909706983,
      "grad_norm": 0.07328025247303306,
      "learning_rate": 0.00017045121209576517,
      "loss": 0.8558,
      "step": 76170
    },
    {
      "epoch": 1.3374532558506997,
      "grad_norm": 0.07297489631849996,
      "learning_rate": 0.00017044327022972151,
      "loss": 0.8459,
      "step": 76180
    },
    {
      "epoch": 1.337628820730701,
      "grad_norm": 0.05811021033922465,
      "learning_rate": 0.0001704353274838183,
      "loss": 0.8572,
      "step": 76190
    },
    {
      "epoch": 1.3378043856107025,
      "grad_norm": 0.08842322468962704,
      "learning_rate": 0.00017042738385815618,
      "loss": 0.8466,
      "step": 76200
    },
    {
      "epoch": 1.3379799504907037,
      "grad_norm": 0.04836864938520365,
      "learning_rate": 0.00017041943935283585,
      "loss": 0.8539,
      "step": 76210
    },
    {
      "epoch": 1.3381555153707052,
      "grad_norm": 0.04419167066462444,
      "learning_rate": 0.00017041149396795788,
      "loss": 0.8431,
      "step": 76220
    },
    {
      "epoch": 1.3383310802507067,
      "grad_norm": 0.05199425936771737,
      "learning_rate": 0.000170403547703623,
      "loss": 0.8484,
      "step": 76230
    },
    {
      "epoch": 1.3385066451307082,
      "grad_norm": 0.06446656781100452,
      "learning_rate": 0.00017039560055993183,
      "loss": 0.8612,
      "step": 76240
    },
    {
      "epoch": 1.3386822100107094,
      "grad_norm": 0.05556019184811688,
      "learning_rate": 0.00017038765253698512,
      "loss": 0.8442,
      "step": 76250
    },
    {
      "epoch": 1.338857774890711,
      "grad_norm": 0.05739764597300805,
      "learning_rate": 0.00017037970363488353,
      "loss": 0.8486,
      "step": 76260
    },
    {
      "epoch": 1.3390333397707121,
      "grad_norm": 0.076713731443272,
      "learning_rate": 0.00017037175385372787,
      "loss": 0.8604,
      "step": 76270
    },
    {
      "epoch": 1.3392089046507136,
      "grad_norm": 0.06518968457612513,
      "learning_rate": 0.00017036380319361875,
      "loss": 0.8434,
      "step": 76280
    },
    {
      "epoch": 1.339384469530715,
      "grad_norm": 0.06619447323858677,
      "learning_rate": 0.00017035585165465692,
      "loss": 0.8555,
      "step": 76290
    },
    {
      "epoch": 1.3395600344107166,
      "grad_norm": 0.0703916621682637,
      "learning_rate": 0.00017034789923694318,
      "loss": 0.8362,
      "step": 76300
    },
    {
      "epoch": 1.3397355992907178,
      "grad_norm": 0.08242011474274821,
      "learning_rate": 0.00017033994594057826,
      "loss": 0.8523,
      "step": 76310
    },
    {
      "epoch": 1.3399111641707193,
      "grad_norm": 0.04911882165450667,
      "learning_rate": 0.00017033199176566297,
      "loss": 0.8432,
      "step": 76320
    },
    {
      "epoch": 1.3400867290507206,
      "grad_norm": 0.07464336878599503,
      "learning_rate": 0.00017032403671229801,
      "loss": 0.8481,
      "step": 76330
    },
    {
      "epoch": 1.340262293930722,
      "grad_norm": 0.05820566205058757,
      "learning_rate": 0.00017031608078058426,
      "loss": 0.8501,
      "step": 76340
    },
    {
      "epoch": 1.3404378588107235,
      "grad_norm": 0.053707217514421195,
      "learning_rate": 0.00017030812397062251,
      "loss": 0.8466,
      "step": 76350
    },
    {
      "epoch": 1.340613423690725,
      "grad_norm": 0.05837142204913718,
      "learning_rate": 0.0001703001662825135,
      "loss": 0.8491,
      "step": 76360
    },
    {
      "epoch": 1.3407889885707263,
      "grad_norm": 0.04841375908626427,
      "learning_rate": 0.00017029220771635809,
      "loss": 0.8416,
      "step": 76370
    },
    {
      "epoch": 1.3409645534507277,
      "grad_norm": 0.05223501913056539,
      "learning_rate": 0.00017028424827225712,
      "loss": 0.8463,
      "step": 76380
    },
    {
      "epoch": 1.341140118330729,
      "grad_norm": 0.061536248868039635,
      "learning_rate": 0.00017027628795031148,
      "loss": 0.8488,
      "step": 76390
    },
    {
      "epoch": 1.3413156832107305,
      "grad_norm": 0.05147271100695557,
      "learning_rate": 0.00017026832675062198,
      "loss": 0.8509,
      "step": 76400
    },
    {
      "epoch": 1.341491248090732,
      "grad_norm": 0.061417981128669426,
      "learning_rate": 0.0001702603646732895,
      "loss": 0.8498,
      "step": 76410
    },
    {
      "epoch": 1.3416668129707334,
      "grad_norm": 0.05057510122620781,
      "learning_rate": 0.00017025240171841496,
      "loss": 0.8509,
      "step": 76420
    },
    {
      "epoch": 1.3418423778507347,
      "grad_norm": 0.06339593600690849,
      "learning_rate": 0.00017024443788609918,
      "loss": 0.846,
      "step": 76430
    },
    {
      "epoch": 1.3420179427307362,
      "grad_norm": 0.05580376138723276,
      "learning_rate": 0.0001702364731764431,
      "loss": 0.8422,
      "step": 76440
    },
    {
      "epoch": 1.3421935076107374,
      "grad_norm": 0.05375265090117153,
      "learning_rate": 0.00017022850758954767,
      "loss": 0.8519,
      "step": 76450
    },
    {
      "epoch": 1.342369072490739,
      "grad_norm": 0.09189003927237982,
      "learning_rate": 0.0001702205411255138,
      "loss": 0.85,
      "step": 76460
    },
    {
      "epoch": 1.3425446373707404,
      "grad_norm": 0.058164689467708613,
      "learning_rate": 0.00017021257378444237,
      "loss": 0.8461,
      "step": 76470
    },
    {
      "epoch": 1.3427202022507418,
      "grad_norm": 0.05274709850988405,
      "learning_rate": 0.00017020460556643432,
      "loss": 0.8459,
      "step": 76480
    },
    {
      "epoch": 1.342895767130743,
      "grad_norm": 0.0932106435330047,
      "learning_rate": 0.00017019663647159072,
      "loss": 0.8537,
      "step": 76490
    },
    {
      "epoch": 1.3430713320107446,
      "grad_norm": 0.06748810469488133,
      "learning_rate": 0.00017018866650001245,
      "loss": 0.8567,
      "step": 76500
    },
    {
      "epoch": 1.343246896890746,
      "grad_norm": 0.06573077822024427,
      "learning_rate": 0.00017018069565180051,
      "loss": 0.8522,
      "step": 76510
    },
    {
      "epoch": 1.3434224617707473,
      "grad_norm": 0.05828187177767365,
      "learning_rate": 0.0001701727239270559,
      "loss": 0.8544,
      "step": 76520
    },
    {
      "epoch": 1.3435980266507488,
      "grad_norm": 0.05428087472777464,
      "learning_rate": 0.00017016475132587965,
      "loss": 0.8576,
      "step": 76530
    },
    {
      "epoch": 1.3437735915307503,
      "grad_norm": 0.062060826918511655,
      "learning_rate": 0.0001701567778483727,
      "loss": 0.8516,
      "step": 76540
    },
    {
      "epoch": 1.3439491564107515,
      "grad_norm": 0.058556008923173905,
      "learning_rate": 0.00017014880349463615,
      "loss": 0.8566,
      "step": 76550
    },
    {
      "epoch": 1.344124721290753,
      "grad_norm": 0.08347779485169887,
      "learning_rate": 0.000170140828264771,
      "loss": 0.8527,
      "step": 76560
    },
    {
      "epoch": 1.3443002861707545,
      "grad_norm": 0.054339431134023335,
      "learning_rate": 0.0001701328521588783,
      "loss": 0.8497,
      "step": 76570
    },
    {
      "epoch": 1.3444758510507557,
      "grad_norm": 0.053783094953709996,
      "learning_rate": 0.00017012487517705912,
      "loss": 0.8544,
      "step": 76580
    },
    {
      "epoch": 1.3446514159307572,
      "grad_norm": 0.06633683960824416,
      "learning_rate": 0.00017011689731941454,
      "loss": 0.845,
      "step": 76590
    },
    {
      "epoch": 1.3448269808107587,
      "grad_norm": 0.09120141373478864,
      "learning_rate": 0.00017010891858604564,
      "loss": 0.8556,
      "step": 76600
    },
    {
      "epoch": 1.34500254569076,
      "grad_norm": 0.08675100934244405,
      "learning_rate": 0.00017010093897705343,
      "loss": 0.8544,
      "step": 76610
    },
    {
      "epoch": 1.3451781105707614,
      "grad_norm": 0.059662188467441195,
      "learning_rate": 0.00017009295849253913,
      "loss": 0.8487,
      "step": 76620
    },
    {
      "epoch": 1.345353675450763,
      "grad_norm": 0.05913960394102192,
      "learning_rate": 0.00017008497713260379,
      "loss": 0.8443,
      "step": 76630
    },
    {
      "epoch": 1.3455292403307642,
      "grad_norm": 0.0652673010964782,
      "learning_rate": 0.00017007699489734855,
      "loss": 0.8508,
      "step": 76640
    },
    {
      "epoch": 1.3457048052107656,
      "grad_norm": 0.0540638926478725,
      "learning_rate": 0.00017006901178687455,
      "loss": 0.8484,
      "step": 76650
    },
    {
      "epoch": 1.345880370090767,
      "grad_norm": 0.061786724790206055,
      "learning_rate": 0.00017006102780128292,
      "loss": 0.8521,
      "step": 76660
    },
    {
      "epoch": 1.3460559349707686,
      "grad_norm": 0.06185661148687561,
      "learning_rate": 0.00017005304294067484,
      "loss": 0.8521,
      "step": 76670
    },
    {
      "epoch": 1.3462314998507698,
      "grad_norm": 0.05986047037967302,
      "learning_rate": 0.00017004505720515147,
      "loss": 0.8517,
      "step": 76680
    },
    {
      "epoch": 1.3464070647307713,
      "grad_norm": 0.07367067556060511,
      "learning_rate": 0.000170037070594814,
      "loss": 0.8488,
      "step": 76690
    },
    {
      "epoch": 1.3465826296107726,
      "grad_norm": 0.05538555228845806,
      "learning_rate": 0.00017002908310976363,
      "loss": 0.8531,
      "step": 76700
    },
    {
      "epoch": 1.346758194490774,
      "grad_norm": 0.08513335301801978,
      "learning_rate": 0.0001700210947501015,
      "loss": 0.8465,
      "step": 76710
    },
    {
      "epoch": 1.3469337593707755,
      "grad_norm": 0.07167744426166522,
      "learning_rate": 0.00017001310551592892,
      "loss": 0.8431,
      "step": 76720
    },
    {
      "epoch": 1.347109324250777,
      "grad_norm": 0.06990942488104501,
      "learning_rate": 0.00017000511540734705,
      "loss": 0.8449,
      "step": 76730
    },
    {
      "epoch": 1.3472848891307783,
      "grad_norm": 0.06523161773889964,
      "learning_rate": 0.00016999712442445714,
      "loss": 0.849,
      "step": 76740
    },
    {
      "epoch": 1.3474604540107797,
      "grad_norm": 0.06150755775330128,
      "learning_rate": 0.0001699891325673604,
      "loss": 0.8494,
      "step": 76750
    },
    {
      "epoch": 1.347636018890781,
      "grad_norm": 0.04897897083963353,
      "learning_rate": 0.0001699811398361582,
      "loss": 0.8453,
      "step": 76760
    },
    {
      "epoch": 1.3478115837707825,
      "grad_norm": 0.060374938222906314,
      "learning_rate": 0.0001699731462309517,
      "loss": 0.8457,
      "step": 76770
    },
    {
      "epoch": 1.347987148650784,
      "grad_norm": 0.04244892279966634,
      "learning_rate": 0.00016996515175184222,
      "loss": 0.8408,
      "step": 76780
    },
    {
      "epoch": 1.3481627135307854,
      "grad_norm": 0.04124124520783707,
      "learning_rate": 0.0001699571563989311,
      "loss": 0.844,
      "step": 76790
    },
    {
      "epoch": 1.3483382784107867,
      "grad_norm": 0.05397345545417103,
      "learning_rate": 0.00016994916017231952,
      "loss": 0.8478,
      "step": 76800
    },
    {
      "epoch": 1.3485138432907882,
      "grad_norm": 0.055716279968680754,
      "learning_rate": 0.0001699411630721089,
      "loss": 0.8558,
      "step": 76810
    },
    {
      "epoch": 1.3486894081707894,
      "grad_norm": 0.04534451655700431,
      "learning_rate": 0.00016993316509840054,
      "loss": 0.8563,
      "step": 76820
    },
    {
      "epoch": 1.348864973050791,
      "grad_norm": 0.04965604611393762,
      "learning_rate": 0.00016992516625129574,
      "loss": 0.8428,
      "step": 76830
    },
    {
      "epoch": 1.3490405379307924,
      "grad_norm": 0.06390435864421967,
      "learning_rate": 0.00016991716653089588,
      "loss": 0.8467,
      "step": 76840
    },
    {
      "epoch": 1.3492161028107938,
      "grad_norm": 0.07046953863498778,
      "learning_rate": 0.00016990916593730233,
      "loss": 0.8493,
      "step": 76850
    },
    {
      "epoch": 1.349391667690795,
      "grad_norm": 0.07536902552960514,
      "learning_rate": 0.00016990116447061646,
      "loss": 0.8448,
      "step": 76860
    },
    {
      "epoch": 1.3495672325707966,
      "grad_norm": 0.09092531567056232,
      "learning_rate": 0.0001698931621309396,
      "loss": 0.8498,
      "step": 76870
    },
    {
      "epoch": 1.3497427974507978,
      "grad_norm": 0.07832628141652447,
      "learning_rate": 0.00016988515891837316,
      "loss": 0.8502,
      "step": 76880
    },
    {
      "epoch": 1.3499183623307993,
      "grad_norm": 0.0708606066921496,
      "learning_rate": 0.00016987715483301859,
      "loss": 0.8545,
      "step": 76890
    },
    {
      "epoch": 1.3500939272108008,
      "grad_norm": 0.06673619377781896,
      "learning_rate": 0.0001698691498749772,
      "loss": 0.8548,
      "step": 76900
    },
    {
      "epoch": 1.3502694920908023,
      "grad_norm": 0.06747472718020812,
      "learning_rate": 0.00016986114404435058,
      "loss": 0.8512,
      "step": 76910
    },
    {
      "epoch": 1.3504450569708035,
      "grad_norm": 0.07651785789599742,
      "learning_rate": 0.00016985313734124,
      "loss": 0.8458,
      "step": 76920
    },
    {
      "epoch": 1.350620621850805,
      "grad_norm": 0.055338728097640316,
      "learning_rate": 0.000169845129765747,
      "loss": 0.8463,
      "step": 76930
    },
    {
      "epoch": 1.3507961867308063,
      "grad_norm": 0.053294112399295705,
      "learning_rate": 0.000169837121317973,
      "loss": 0.8455,
      "step": 76940
    },
    {
      "epoch": 1.3509717516108077,
      "grad_norm": 0.08854519767401693,
      "learning_rate": 0.00016982911199801948,
      "loss": 0.8436,
      "step": 76950
    },
    {
      "epoch": 1.3511473164908092,
      "grad_norm": 0.05125014610352081,
      "learning_rate": 0.00016982110180598791,
      "loss": 0.8439,
      "step": 76960
    },
    {
      "epoch": 1.3513228813708107,
      "grad_norm": 0.06225603316685289,
      "learning_rate": 0.00016981309074197984,
      "loss": 0.8512,
      "step": 76970
    },
    {
      "epoch": 1.351498446250812,
      "grad_norm": 0.055202896452329814,
      "learning_rate": 0.0001698050788060967,
      "loss": 0.8444,
      "step": 76980
    },
    {
      "epoch": 1.3516740111308134,
      "grad_norm": 0.06599033296567273,
      "learning_rate": 0.00016979706599844,
      "loss": 0.8534,
      "step": 76990
    },
    {
      "epoch": 1.3518495760108147,
      "grad_norm": 0.07064819781330606,
      "learning_rate": 0.00016978905231911125,
      "loss": 0.8524,
      "step": 77000
    },
    {
      "epoch": 1.3520251408908162,
      "grad_norm": 0.0956589708059054,
      "learning_rate": 0.0001697810377682121,
      "loss": 0.853,
      "step": 77010
    },
    {
      "epoch": 1.3522007057708176,
      "grad_norm": 0.05510510695801866,
      "learning_rate": 0.00016977302234584402,
      "loss": 0.8455,
      "step": 77020
    },
    {
      "epoch": 1.3523762706508191,
      "grad_norm": 0.07306430700051864,
      "learning_rate": 0.00016976500605210855,
      "loss": 0.8453,
      "step": 77030
    },
    {
      "epoch": 1.3525518355308204,
      "grad_norm": 0.07095901523344574,
      "learning_rate": 0.00016975698888710726,
      "loss": 0.85,
      "step": 77040
    },
    {
      "epoch": 1.3527274004108218,
      "grad_norm": 0.04826415536486713,
      "learning_rate": 0.00016974897085094174,
      "loss": 0.8409,
      "step": 77050
    },
    {
      "epoch": 1.352902965290823,
      "grad_norm": 0.04802517430766489,
      "learning_rate": 0.00016974095194371356,
      "loss": 0.8453,
      "step": 77060
    },
    {
      "epoch": 1.3530785301708246,
      "grad_norm": 0.05923635639300022,
      "learning_rate": 0.00016973293216552442,
      "loss": 0.8486,
      "step": 77070
    },
    {
      "epoch": 1.353254095050826,
      "grad_norm": 0.05342970221156565,
      "learning_rate": 0.0001697249115164758,
      "loss": 0.8414,
      "step": 77080
    },
    {
      "epoch": 1.3534296599308275,
      "grad_norm": 0.06680001977592652,
      "learning_rate": 0.0001697168899966694,
      "loss": 0.8519,
      "step": 77090
    },
    {
      "epoch": 1.3536052248108288,
      "grad_norm": 0.05140791246912955,
      "learning_rate": 0.00016970886760620684,
      "loss": 0.852,
      "step": 77100
    },
    {
      "epoch": 1.3537807896908303,
      "grad_norm": 0.06006985797212989,
      "learning_rate": 0.00016970084434518974,
      "loss": 0.8497,
      "step": 77110
    },
    {
      "epoch": 1.3539563545708315,
      "grad_norm": 0.05360580503725353,
      "learning_rate": 0.00016969282021371977,
      "loss": 0.8502,
      "step": 77120
    },
    {
      "epoch": 1.354131919450833,
      "grad_norm": 0.05295761543253292,
      "learning_rate": 0.00016968479521189863,
      "loss": 0.8512,
      "step": 77130
    },
    {
      "epoch": 1.3543074843308345,
      "grad_norm": 0.06180507383384494,
      "learning_rate": 0.00016967676933982793,
      "loss": 0.8529,
      "step": 77140
    },
    {
      "epoch": 1.354483049210836,
      "grad_norm": 0.06834841656389164,
      "learning_rate": 0.00016966874259760948,
      "loss": 0.8515,
      "step": 77150
    },
    {
      "epoch": 1.3546586140908372,
      "grad_norm": 0.05407450520433601,
      "learning_rate": 0.00016966071498534483,
      "loss": 0.8448,
      "step": 77160
    },
    {
      "epoch": 1.3548341789708387,
      "grad_norm": 0.061715643650045636,
      "learning_rate": 0.00016965268650313579,
      "loss": 0.8486,
      "step": 77170
    },
    {
      "epoch": 1.35500974385084,
      "grad_norm": 0.06352405207264442,
      "learning_rate": 0.00016964465715108403,
      "loss": 0.8441,
      "step": 77180
    },
    {
      "epoch": 1.3551853087308414,
      "grad_norm": 0.06811848508400585,
      "learning_rate": 0.00016963662692929133,
      "loss": 0.8472,
      "step": 77190
    },
    {
      "epoch": 1.355360873610843,
      "grad_norm": 0.060795810569747075,
      "learning_rate": 0.00016962859583785942,
      "loss": 0.851,
      "step": 77200
    },
    {
      "epoch": 1.3555364384908444,
      "grad_norm": 0.05265867131524301,
      "learning_rate": 0.00016962056387689005,
      "loss": 0.8485,
      "step": 77210
    },
    {
      "epoch": 1.3557120033708456,
      "grad_norm": 0.053652829295225364,
      "learning_rate": 0.00016961253104648495,
      "loss": 0.8505,
      "step": 77220
    },
    {
      "epoch": 1.355887568250847,
      "grad_norm": 0.05556272225996789,
      "learning_rate": 0.00016960449734674596,
      "loss": 0.8475,
      "step": 77230
    },
    {
      "epoch": 1.3560631331308486,
      "grad_norm": 0.07485061696827737,
      "learning_rate": 0.00016959646277777484,
      "loss": 0.8563,
      "step": 77240
    },
    {
      "epoch": 1.3562386980108498,
      "grad_norm": 0.06761099736457528,
      "learning_rate": 0.00016958842733967336,
      "loss": 0.8507,
      "step": 77250
    },
    {
      "epoch": 1.3564142628908513,
      "grad_norm": 0.051963337824891855,
      "learning_rate": 0.0001695803910325434,
      "loss": 0.8478,
      "step": 77260
    },
    {
      "epoch": 1.3565898277708528,
      "grad_norm": 0.10217879464544954,
      "learning_rate": 0.00016957235385648672,
      "loss": 0.8619,
      "step": 77270
    },
    {
      "epoch": 1.356765392650854,
      "grad_norm": 0.06463708975984911,
      "learning_rate": 0.0001695643158116052,
      "loss": 0.8501,
      "step": 77280
    },
    {
      "epoch": 1.3569409575308555,
      "grad_norm": 0.08990834349000446,
      "learning_rate": 0.00016955627689800064,
      "loss": 0.8363,
      "step": 77290
    },
    {
      "epoch": 1.357116522410857,
      "grad_norm": 0.05534726517735422,
      "learning_rate": 0.00016954823711577493,
      "loss": 0.8519,
      "step": 77300
    },
    {
      "epoch": 1.3572920872908583,
      "grad_norm": 0.05613175695742679,
      "learning_rate": 0.00016954019646502988,
      "loss": 0.849,
      "step": 77310
    },
    {
      "epoch": 1.3574676521708597,
      "grad_norm": 0.05925591893250529,
      "learning_rate": 0.00016953215494586743,
      "loss": 0.851,
      "step": 77320
    },
    {
      "epoch": 1.3576432170508612,
      "grad_norm": 0.06729646428060317,
      "learning_rate": 0.00016952411255838942,
      "loss": 0.8575,
      "step": 77330
    },
    {
      "epoch": 1.3578187819308625,
      "grad_norm": 0.05078430710988355,
      "learning_rate": 0.00016951606930269778,
      "loss": 0.8511,
      "step": 77340
    },
    {
      "epoch": 1.357994346810864,
      "grad_norm": 0.07373589279231717,
      "learning_rate": 0.00016950802517889444,
      "loss": 0.8377,
      "step": 77350
    },
    {
      "epoch": 1.3581699116908654,
      "grad_norm": 0.06603925198537992,
      "learning_rate": 0.00016949998018708128,
      "loss": 0.8584,
      "step": 77360
    },
    {
      "epoch": 1.3583454765708667,
      "grad_norm": 0.08102534341847584,
      "learning_rate": 0.00016949193432736027,
      "loss": 0.85,
      "step": 77370
    },
    {
      "epoch": 1.3585210414508682,
      "grad_norm": 0.0785996467269745,
      "learning_rate": 0.00016948388759983326,
      "loss": 0.8411,
      "step": 77380
    },
    {
      "epoch": 1.3586966063308696,
      "grad_norm": 0.06094132329848468,
      "learning_rate": 0.0001694758400046023,
      "loss": 0.8522,
      "step": 77390
    },
    {
      "epoch": 1.3588721712108711,
      "grad_norm": 0.0748571035400303,
      "learning_rate": 0.00016946779154176933,
      "loss": 0.8527,
      "step": 77400
    },
    {
      "epoch": 1.3590477360908724,
      "grad_norm": 0.06438524377058397,
      "learning_rate": 0.00016945974221143633,
      "loss": 0.8455,
      "step": 77410
    },
    {
      "epoch": 1.3592233009708738,
      "grad_norm": 0.06643832692718347,
      "learning_rate": 0.00016945169201370525,
      "loss": 0.8494,
      "step": 77420
    },
    {
      "epoch": 1.359398865850875,
      "grad_norm": 0.047677903334280375,
      "learning_rate": 0.0001694436409486781,
      "loss": 0.842,
      "step": 77430
    },
    {
      "epoch": 1.3595744307308766,
      "grad_norm": 0.08023700700260031,
      "learning_rate": 0.00016943558901645693,
      "loss": 0.8549,
      "step": 77440
    },
    {
      "epoch": 1.359749995610878,
      "grad_norm": 0.059473165097926946,
      "learning_rate": 0.00016942753621714373,
      "loss": 0.8559,
      "step": 77450
    },
    {
      "epoch": 1.3599255604908795,
      "grad_norm": 0.052929993658826606,
      "learning_rate": 0.0001694194825508405,
      "loss": 0.8446,
      "step": 77460
    },
    {
      "epoch": 1.3601011253708808,
      "grad_norm": 0.057286133003299204,
      "learning_rate": 0.0001694114280176494,
      "loss": 0.8513,
      "step": 77470
    },
    {
      "epoch": 1.3602766902508823,
      "grad_norm": 0.055338717447643675,
      "learning_rate": 0.00016940337261767233,
      "loss": 0.8504,
      "step": 77480
    },
    {
      "epoch": 1.3604522551308835,
      "grad_norm": 0.05214022713582364,
      "learning_rate": 0.00016939531635101144,
      "loss": 0.8475,
      "step": 77490
    },
    {
      "epoch": 1.360627820010885,
      "grad_norm": 0.058114927216248245,
      "learning_rate": 0.00016938725921776873,
      "loss": 0.8438,
      "step": 77500
    },
    {
      "epoch": 1.3608033848908865,
      "grad_norm": 0.0796204862589523,
      "learning_rate": 0.0001693792012180464,
      "loss": 0.8532,
      "step": 77510
    },
    {
      "epoch": 1.360978949770888,
      "grad_norm": 0.07219669887346077,
      "learning_rate": 0.00016937114235194648,
      "loss": 0.8552,
      "step": 77520
    },
    {
      "epoch": 1.3611545146508892,
      "grad_norm": 0.06239055390052521,
      "learning_rate": 0.00016936308261957105,
      "loss": 0.8476,
      "step": 77530
    },
    {
      "epoch": 1.3613300795308907,
      "grad_norm": 0.07264147397367343,
      "learning_rate": 0.0001693550220210223,
      "loss": 0.8512,
      "step": 77540
    },
    {
      "epoch": 1.361505644410892,
      "grad_norm": 0.056825252765985046,
      "learning_rate": 0.0001693469605564023,
      "loss": 0.8491,
      "step": 77550
    },
    {
      "epoch": 1.3616812092908934,
      "grad_norm": 0.06613718278915062,
      "learning_rate": 0.00016933889822581322,
      "loss": 0.8527,
      "step": 77560
    },
    {
      "epoch": 1.361856774170895,
      "grad_norm": 0.07038071868797745,
      "learning_rate": 0.0001693308350293572,
      "loss": 0.8523,
      "step": 77570
    },
    {
      "epoch": 1.3620323390508964,
      "grad_norm": 0.04839875297650354,
      "learning_rate": 0.0001693227709671364,
      "loss": 0.8437,
      "step": 77580
    },
    {
      "epoch": 1.3622079039308976,
      "grad_norm": 0.06047923139230421,
      "learning_rate": 0.000169314706039253,
      "loss": 0.8576,
      "step": 77590
    },
    {
      "epoch": 1.3623834688108991,
      "grad_norm": 0.050889296224315955,
      "learning_rate": 0.00016930664024580918,
      "loss": 0.8499,
      "step": 77600
    },
    {
      "epoch": 1.3625590336909004,
      "grad_norm": 0.06252619914740223,
      "learning_rate": 0.00016929857358690716,
      "loss": 0.8558,
      "step": 77610
    },
    {
      "epoch": 1.3627345985709018,
      "grad_norm": 0.057443839018566936,
      "learning_rate": 0.00016929050606264906,
      "loss": 0.842,
      "step": 77620
    },
    {
      "epoch": 1.3629101634509033,
      "grad_norm": 0.05245679647704677,
      "learning_rate": 0.00016928243767313722,
      "loss": 0.8497,
      "step": 77630
    },
    {
      "epoch": 1.3630857283309048,
      "grad_norm": 0.06112343864472478,
      "learning_rate": 0.0001692743684184738,
      "loss": 0.8422,
      "step": 77640
    },
    {
      "epoch": 1.363261293210906,
      "grad_norm": 0.05505272390493824,
      "learning_rate": 0.00016926629829876102,
      "loss": 0.8505,
      "step": 77650
    },
    {
      "epoch": 1.3634368580909075,
      "grad_norm": 0.06408644970896067,
      "learning_rate": 0.0001692582273141012,
      "loss": 0.8559,
      "step": 77660
    },
    {
      "epoch": 1.3636124229709088,
      "grad_norm": 0.07729432633629606,
      "learning_rate": 0.0001692501554645965,
      "loss": 0.8413,
      "step": 77670
    },
    {
      "epoch": 1.3637879878509103,
      "grad_norm": 0.05923501133100925,
      "learning_rate": 0.0001692420827503493,
      "loss": 0.8503,
      "step": 77680
    },
    {
      "epoch": 1.3639635527309117,
      "grad_norm": 0.05254781564202942,
      "learning_rate": 0.0001692340091714618,
      "loss": 0.8569,
      "step": 77690
    },
    {
      "epoch": 1.3641391176109132,
      "grad_norm": 0.057203043334860544,
      "learning_rate": 0.00016922593472803635,
      "loss": 0.8534,
      "step": 77700
    },
    {
      "epoch": 1.3643146824909145,
      "grad_norm": 0.06154618807076067,
      "learning_rate": 0.00016921785942017523,
      "loss": 0.8557,
      "step": 77710
    },
    {
      "epoch": 1.364490247370916,
      "grad_norm": 0.06322550942302864,
      "learning_rate": 0.00016920978324798077,
      "loss": 0.8476,
      "step": 77720
    },
    {
      "epoch": 1.3646658122509172,
      "grad_norm": 0.07080704219863468,
      "learning_rate": 0.00016920170621155525,
      "loss": 0.8476,
      "step": 77730
    },
    {
      "epoch": 1.3648413771309187,
      "grad_norm": 0.048027057295499005,
      "learning_rate": 0.00016919362831100106,
      "loss": 0.8446,
      "step": 77740
    },
    {
      "epoch": 1.3650169420109202,
      "grad_norm": 0.059745663028626675,
      "learning_rate": 0.00016918554954642052,
      "loss": 0.8475,
      "step": 77750
    },
    {
      "epoch": 1.3651925068909216,
      "grad_norm": 0.07279628264267145,
      "learning_rate": 0.00016917746991791603,
      "loss": 0.8425,
      "step": 77760
    },
    {
      "epoch": 1.365368071770923,
      "grad_norm": 0.06011711753051195,
      "learning_rate": 0.00016916938942558989,
      "loss": 0.8437,
      "step": 77770
    },
    {
      "epoch": 1.3655436366509244,
      "grad_norm": 0.046114164270699774,
      "learning_rate": 0.00016916130806954457,
      "loss": 0.8472,
      "step": 77780
    },
    {
      "epoch": 1.3657192015309256,
      "grad_norm": 0.04727456792351673,
      "learning_rate": 0.00016915322584988235,
      "loss": 0.8506,
      "step": 77790
    },
    {
      "epoch": 1.365894766410927,
      "grad_norm": 0.053380966402414826,
      "learning_rate": 0.00016914514276670573,
      "loss": 0.848,
      "step": 77800
    },
    {
      "epoch": 1.3660703312909286,
      "grad_norm": 0.04401024504474506,
      "learning_rate": 0.0001691370588201171,
      "loss": 0.8504,
      "step": 77810
    },
    {
      "epoch": 1.36624589617093,
      "grad_norm": 0.061258762389955375,
      "learning_rate": 0.00016912897401021886,
      "loss": 0.8605,
      "step": 77820
    },
    {
      "epoch": 1.3664214610509313,
      "grad_norm": 0.0801380915407228,
      "learning_rate": 0.00016912088833711348,
      "loss": 0.8492,
      "step": 77830
    },
    {
      "epoch": 1.3665970259309328,
      "grad_norm": 0.08717321426198449,
      "learning_rate": 0.00016911280180090336,
      "loss": 0.8471,
      "step": 77840
    },
    {
      "epoch": 1.366772590810934,
      "grad_norm": 0.0702749143402834,
      "learning_rate": 0.000169104714401691,
      "loss": 0.8498,
      "step": 77850
    },
    {
      "epoch": 1.3669481556909355,
      "grad_norm": 0.05121893247806179,
      "learning_rate": 0.0001690966261395789,
      "loss": 0.8426,
      "step": 77860
    },
    {
      "epoch": 1.367123720570937,
      "grad_norm": 0.058579309775073814,
      "learning_rate": 0.00016908853701466946,
      "loss": 0.8515,
      "step": 77870
    },
    {
      "epoch": 1.3672992854509385,
      "grad_norm": 0.07211005277037046,
      "learning_rate": 0.00016908044702706519,
      "loss": 0.8533,
      "step": 77880
    },
    {
      "epoch": 1.3674748503309397,
      "grad_norm": 0.05607432962212349,
      "learning_rate": 0.00016907235617686864,
      "loss": 0.8525,
      "step": 77890
    },
    {
      "epoch": 1.3676504152109412,
      "grad_norm": 0.03983366930430906,
      "learning_rate": 0.00016906426446418226,
      "loss": 0.865,
      "step": 77900
    },
    {
      "epoch": 1.3678259800909425,
      "grad_norm": 0.03989745499269955,
      "learning_rate": 0.0001690561718891086,
      "loss": 0.8474,
      "step": 77910
    },
    {
      "epoch": 1.368001544970944,
      "grad_norm": 0.05726013084809209,
      "learning_rate": 0.00016904807845175023,
      "loss": 0.8484,
      "step": 77920
    },
    {
      "epoch": 1.3681771098509454,
      "grad_norm": 0.06414630025558843,
      "learning_rate": 0.00016903998415220966,
      "loss": 0.8485,
      "step": 77930
    },
    {
      "epoch": 1.368352674730947,
      "grad_norm": 0.06340501838985042,
      "learning_rate": 0.0001690318889905894,
      "loss": 0.8609,
      "step": 77940
    },
    {
      "epoch": 1.3685282396109482,
      "grad_norm": 0.04502514093642912,
      "learning_rate": 0.00016902379296699214,
      "loss": 0.8593,
      "step": 77950
    },
    {
      "epoch": 1.3687038044909496,
      "grad_norm": 0.05270780997437195,
      "learning_rate": 0.0001690156960815204,
      "loss": 0.8452,
      "step": 77960
    },
    {
      "epoch": 1.3688793693709511,
      "grad_norm": 0.07734057258630019,
      "learning_rate": 0.00016900759833427667,
      "loss": 0.8523,
      "step": 77970
    },
    {
      "epoch": 1.3690549342509524,
      "grad_norm": 0.06193493007177415,
      "learning_rate": 0.00016899949972536368,
      "loss": 0.8425,
      "step": 77980
    },
    {
      "epoch": 1.3692304991309538,
      "grad_norm": 0.06543834801697972,
      "learning_rate": 0.00016899140025488401,
      "loss": 0.8493,
      "step": 77990
    },
    {
      "epoch": 1.3694060640109553,
      "grad_norm": 0.055738484044190864,
      "learning_rate": 0.00016898329992294026,
      "loss": 0.857,
      "step": 78000
    },
    {
      "epoch": 1.3695816288909566,
      "grad_norm": 0.06086867350241575,
      "learning_rate": 0.00016897519872963507,
      "loss": 0.8393,
      "step": 78010
    },
    {
      "epoch": 1.369757193770958,
      "grad_norm": 0.06379661553479445,
      "learning_rate": 0.00016896709667507106,
      "loss": 0.8549,
      "step": 78020
    },
    {
      "epoch": 1.3699327586509595,
      "grad_norm": 0.05916035511416921,
      "learning_rate": 0.00016895899375935094,
      "loss": 0.8481,
      "step": 78030
    },
    {
      "epoch": 1.3701083235309608,
      "grad_norm": 0.050570802817114256,
      "learning_rate": 0.00016895088998257737,
      "loss": 0.8455,
      "step": 78040
    },
    {
      "epoch": 1.3702838884109623,
      "grad_norm": 0.04760793778088413,
      "learning_rate": 0.00016894278534485295,
      "loss": 0.8553,
      "step": 78050
    },
    {
      "epoch": 1.3704594532909637,
      "grad_norm": 0.057781949836057285,
      "learning_rate": 0.00016893467984628043,
      "loss": 0.843,
      "step": 78060
    },
    {
      "epoch": 1.370635018170965,
      "grad_norm": 0.054975585063359124,
      "learning_rate": 0.00016892657348696252,
      "loss": 0.842,
      "step": 78070
    },
    {
      "epoch": 1.3708105830509665,
      "grad_norm": 0.06868483194039612,
      "learning_rate": 0.00016891846626700187,
      "loss": 0.8416,
      "step": 78080
    },
    {
      "epoch": 1.370986147930968,
      "grad_norm": 0.058961101175575036,
      "learning_rate": 0.00016891035818650125,
      "loss": 0.854,
      "step": 78090
    },
    {
      "epoch": 1.3711617128109692,
      "grad_norm": 0.051174857190877714,
      "learning_rate": 0.0001689022492455634,
      "loss": 0.8546,
      "step": 78100
    },
    {
      "epoch": 1.3713372776909707,
      "grad_norm": 0.0449259419604081,
      "learning_rate": 0.00016889413944429097,
      "loss": 0.8496,
      "step": 78110
    },
    {
      "epoch": 1.3715128425709722,
      "grad_norm": 0.06023783209176246,
      "learning_rate": 0.00016888602878278687,
      "loss": 0.8432,
      "step": 78120
    },
    {
      "epoch": 1.3716884074509736,
      "grad_norm": 0.10721629460883414,
      "learning_rate": 0.00016887791726115373,
      "loss": 0.8478,
      "step": 78130
    },
    {
      "epoch": 1.371863972330975,
      "grad_norm": 0.05366890699207747,
      "learning_rate": 0.00016886980487949437,
      "loss": 0.8482,
      "step": 78140
    },
    {
      "epoch": 1.3720395372109764,
      "grad_norm": 0.07949020266171807,
      "learning_rate": 0.0001688616916379116,
      "loss": 0.8515,
      "step": 78150
    },
    {
      "epoch": 1.3722151020909776,
      "grad_norm": 0.051404069276899556,
      "learning_rate": 0.00016885357753650818,
      "loss": 0.8468,
      "step": 78160
    },
    {
      "epoch": 1.3723906669709791,
      "grad_norm": 0.06708239869208785,
      "learning_rate": 0.0001688454625753869,
      "loss": 0.8508,
      "step": 78170
    },
    {
      "epoch": 1.3725662318509806,
      "grad_norm": 0.06432576374776744,
      "learning_rate": 0.00016883734675465062,
      "loss": 0.8493,
      "step": 78180
    },
    {
      "epoch": 1.372741796730982,
      "grad_norm": 0.06073231608675784,
      "learning_rate": 0.00016882923007440216,
      "loss": 0.8518,
      "step": 78190
    },
    {
      "epoch": 1.3729173616109833,
      "grad_norm": 0.06037276622624786,
      "learning_rate": 0.00016882111253474434,
      "loss": 0.8518,
      "step": 78200
    },
    {
      "epoch": 1.3730929264909848,
      "grad_norm": 0.07632626935142091,
      "learning_rate": 0.00016881299413578005,
      "loss": 0.8504,
      "step": 78210
    },
    {
      "epoch": 1.373268491370986,
      "grad_norm": 0.06086222655743199,
      "learning_rate": 0.00016880487487761213,
      "loss": 0.8518,
      "step": 78220
    },
    {
      "epoch": 1.3734440562509875,
      "grad_norm": 0.05813675036862018,
      "learning_rate": 0.00016879675476034347,
      "loss": 0.8603,
      "step": 78230
    },
    {
      "epoch": 1.373619621130989,
      "grad_norm": 0.06269187032370047,
      "learning_rate": 0.0001687886337840769,
      "loss": 0.8489,
      "step": 78240
    },
    {
      "epoch": 1.3737951860109905,
      "grad_norm": 0.05974458192877722,
      "learning_rate": 0.00016878051194891533,
      "loss": 0.8491,
      "step": 78250
    },
    {
      "epoch": 1.3739707508909917,
      "grad_norm": 0.05404477027612948,
      "learning_rate": 0.00016877238925496174,
      "loss": 0.8513,
      "step": 78260
    },
    {
      "epoch": 1.3741463157709932,
      "grad_norm": 0.06894611829684023,
      "learning_rate": 0.00016876426570231894,
      "loss": 0.8436,
      "step": 78270
    },
    {
      "epoch": 1.3743218806509945,
      "grad_norm": 0.05591530488017604,
      "learning_rate": 0.0001687561412910899,
      "loss": 0.8459,
      "step": 78280
    },
    {
      "epoch": 1.374497445530996,
      "grad_norm": 0.059524053249916935,
      "learning_rate": 0.0001687480160213776,
      "loss": 0.8462,
      "step": 78290
    },
    {
      "epoch": 1.3746730104109974,
      "grad_norm": 0.0535092167454499,
      "learning_rate": 0.00016873988989328495,
      "loss": 0.8475,
      "step": 78300
    },
    {
      "epoch": 1.374848575290999,
      "grad_norm": 0.05926627046483781,
      "learning_rate": 0.00016873176290691487,
      "loss": 0.8472,
      "step": 78310
    },
    {
      "epoch": 1.3750241401710002,
      "grad_norm": 0.10782514148048612,
      "learning_rate": 0.00016872363506237044,
      "loss": 0.8489,
      "step": 78320
    },
    {
      "epoch": 1.3751997050510016,
      "grad_norm": 0.0591120204111127,
      "learning_rate": 0.00016871550635975455,
      "loss": 0.8491,
      "step": 78330
    },
    {
      "epoch": 1.375375269931003,
      "grad_norm": 0.0630334560042117,
      "learning_rate": 0.00016870737679917022,
      "loss": 0.8501,
      "step": 78340
    },
    {
      "epoch": 1.3755508348110044,
      "grad_norm": 0.04310383344508578,
      "learning_rate": 0.00016869924638072042,
      "loss": 0.8556,
      "step": 78350
    },
    {
      "epoch": 1.3757263996910059,
      "grad_norm": 0.056562212913292505,
      "learning_rate": 0.0001686911151045082,
      "loss": 0.8519,
      "step": 78360
    },
    {
      "epoch": 1.3759019645710073,
      "grad_norm": 0.04815072165149406,
      "learning_rate": 0.0001686829829706366,
      "loss": 0.8479,
      "step": 78370
    },
    {
      "epoch": 1.3760775294510086,
      "grad_norm": 0.07613192532276995,
      "learning_rate": 0.00016867484997920863,
      "loss": 0.845,
      "step": 78380
    },
    {
      "epoch": 1.37625309433101,
      "grad_norm": 0.06662045749042166,
      "learning_rate": 0.00016866671613032736,
      "loss": 0.845,
      "step": 78390
    },
    {
      "epoch": 1.3764286592110113,
      "grad_norm": 0.06787488412626587,
      "learning_rate": 0.0001686585814240958,
      "loss": 0.8422,
      "step": 78400
    },
    {
      "epoch": 1.3766042240910128,
      "grad_norm": 0.07269254431209486,
      "learning_rate": 0.00016865044586061704,
      "loss": 0.8539,
      "step": 78410
    },
    {
      "epoch": 1.3767797889710143,
      "grad_norm": 0.065960595136813,
      "learning_rate": 0.00016864230943999418,
      "loss": 0.8515,
      "step": 78420
    },
    {
      "epoch": 1.3769553538510158,
      "grad_norm": 0.053199605071870756,
      "learning_rate": 0.0001686341721623303,
      "loss": 0.8478,
      "step": 78430
    },
    {
      "epoch": 1.377130918731017,
      "grad_norm": 0.09903780380541713,
      "learning_rate": 0.00016862603402772848,
      "loss": 0.8479,
      "step": 78440
    },
    {
      "epoch": 1.3773064836110185,
      "grad_norm": 0.0964481765377408,
      "learning_rate": 0.00016861789503629186,
      "loss": 0.8427,
      "step": 78450
    },
    {
      "epoch": 1.3774820484910197,
      "grad_norm": 0.07210780299256254,
      "learning_rate": 0.00016860975518812355,
      "loss": 0.8513,
      "step": 78460
    },
    {
      "epoch": 1.3776576133710212,
      "grad_norm": 0.06887212360926338,
      "learning_rate": 0.00016860161448332668,
      "loss": 0.8608,
      "step": 78470
    },
    {
      "epoch": 1.3778331782510227,
      "grad_norm": 0.07959448982723166,
      "learning_rate": 0.0001685934729220044,
      "loss": 0.8528,
      "step": 78480
    },
    {
      "epoch": 1.3780087431310242,
      "grad_norm": 0.06151981045777631,
      "learning_rate": 0.00016858533050425988,
      "loss": 0.8566,
      "step": 78490
    },
    {
      "epoch": 1.3781843080110254,
      "grad_norm": 0.05873830592665453,
      "learning_rate": 0.00016857718723019627,
      "loss": 0.8384,
      "step": 78500
    },
    {
      "epoch": 1.378359872891027,
      "grad_norm": 0.05019969525747996,
      "learning_rate": 0.00016856904309991673,
      "loss": 0.8426,
      "step": 78510
    },
    {
      "epoch": 1.3785354377710282,
      "grad_norm": 0.05101375431376813,
      "learning_rate": 0.00016856089811352447,
      "loss": 0.8453,
      "step": 78520
    },
    {
      "epoch": 1.3787110026510296,
      "grad_norm": 0.08879430056455936,
      "learning_rate": 0.00016855275227112265,
      "loss": 0.85,
      "step": 78530
    },
    {
      "epoch": 1.3788865675310311,
      "grad_norm": 0.05383868882782366,
      "learning_rate": 0.00016854460557281454,
      "loss": 0.8468,
      "step": 78540
    },
    {
      "epoch": 1.3790621324110326,
      "grad_norm": 0.05509925725492399,
      "learning_rate": 0.00016853645801870337,
      "loss": 0.851,
      "step": 78550
    },
    {
      "epoch": 1.3792376972910338,
      "grad_norm": 0.07711127583894853,
      "learning_rate": 0.00016852830960889227,
      "loss": 0.8537,
      "step": 78560
    },
    {
      "epoch": 1.3794132621710353,
      "grad_norm": 0.06651443226645253,
      "learning_rate": 0.0001685201603434846,
      "loss": 0.8476,
      "step": 78570
    },
    {
      "epoch": 1.3795888270510366,
      "grad_norm": 0.05730325500196916,
      "learning_rate": 0.00016851201022258355,
      "loss": 0.8349,
      "step": 78580
    },
    {
      "epoch": 1.379764391931038,
      "grad_norm": 0.05116722733395741,
      "learning_rate": 0.00016850385924629236,
      "loss": 0.8481,
      "step": 78590
    },
    {
      "epoch": 1.3799399568110395,
      "grad_norm": 0.10581807727785406,
      "learning_rate": 0.00016849570741471432,
      "loss": 0.8398,
      "step": 78600
    },
    {
      "epoch": 1.380115521691041,
      "grad_norm": 0.06209771165941736,
      "learning_rate": 0.0001684875547279528,
      "loss": 0.8492,
      "step": 78610
    },
    {
      "epoch": 1.3802910865710423,
      "grad_norm": 0.09430060902109223,
      "learning_rate": 0.00016847940118611097,
      "loss": 0.8515,
      "step": 78620
    },
    {
      "epoch": 1.3804666514510437,
      "grad_norm": 0.06276210684879888,
      "learning_rate": 0.00016847124678929222,
      "loss": 0.8491,
      "step": 78630
    },
    {
      "epoch": 1.380642216331045,
      "grad_norm": 0.04489185989044163,
      "learning_rate": 0.00016846309153759984,
      "loss": 0.8469,
      "step": 78640
    },
    {
      "epoch": 1.3808177812110465,
      "grad_norm": 0.050925052807485806,
      "learning_rate": 0.00016845493543113716,
      "loss": 0.8488,
      "step": 78650
    },
    {
      "epoch": 1.380993346091048,
      "grad_norm": 0.05870777422964705,
      "learning_rate": 0.00016844677847000752,
      "loss": 0.8551,
      "step": 78660
    },
    {
      "epoch": 1.3811689109710494,
      "grad_norm": 0.0784375103814371,
      "learning_rate": 0.00016843862065431424,
      "loss": 0.8499,
      "step": 78670
    },
    {
      "epoch": 1.3813444758510507,
      "grad_norm": 0.07951408471598821,
      "learning_rate": 0.00016843046198416076,
      "loss": 0.8525,
      "step": 78680
    },
    {
      "epoch": 1.3815200407310522,
      "grad_norm": 0.08768248555788793,
      "learning_rate": 0.0001684223024596504,
      "loss": 0.8498,
      "step": 78690
    },
    {
      "epoch": 1.3816956056110536,
      "grad_norm": 0.05783587781970763,
      "learning_rate": 0.00016841414208088653,
      "loss": 0.8397,
      "step": 78700
    },
    {
      "epoch": 1.381871170491055,
      "grad_norm": 0.05034933125557133,
      "learning_rate": 0.00016840598084797256,
      "loss": 0.8532,
      "step": 78710
    },
    {
      "epoch": 1.3820467353710564,
      "grad_norm": 0.05626402911246426,
      "learning_rate": 0.00016839781876101188,
      "loss": 0.849,
      "step": 78720
    },
    {
      "epoch": 1.3822223002510579,
      "grad_norm": 0.07261129807699103,
      "learning_rate": 0.00016838965582010795,
      "loss": 0.847,
      "step": 78730
    },
    {
      "epoch": 1.3823978651310591,
      "grad_norm": 0.05552718678147833,
      "learning_rate": 0.00016838149202536416,
      "loss": 0.8464,
      "step": 78740
    },
    {
      "epoch": 1.3825734300110606,
      "grad_norm": 0.06347399527994904,
      "learning_rate": 0.00016837332737688394,
      "loss": 0.8512,
      "step": 78750
    },
    {
      "epoch": 1.382748994891062,
      "grad_norm": 0.048535972907131866,
      "learning_rate": 0.00016836516187477077,
      "loss": 0.8522,
      "step": 78760
    },
    {
      "epoch": 1.3829245597710633,
      "grad_norm": 0.06647986635039914,
      "learning_rate": 0.00016835699551912806,
      "loss": 0.8476,
      "step": 78770
    },
    {
      "epoch": 1.3831001246510648,
      "grad_norm": 0.06734490849933253,
      "learning_rate": 0.00016834882831005932,
      "loss": 0.8446,
      "step": 78780
    },
    {
      "epoch": 1.3832756895310663,
      "grad_norm": 0.06817396040382352,
      "learning_rate": 0.000168340660247668,
      "loss": 0.8457,
      "step": 78790
    },
    {
      "epoch": 1.3834512544110675,
      "grad_norm": 0.06780886205730813,
      "learning_rate": 0.00016833249133205763,
      "loss": 0.8525,
      "step": 78800
    },
    {
      "epoch": 1.383626819291069,
      "grad_norm": 0.06817856393989492,
      "learning_rate": 0.00016832432156333168,
      "loss": 0.8399,
      "step": 78810
    },
    {
      "epoch": 1.3838023841710705,
      "grad_norm": 0.06415108096199527,
      "learning_rate": 0.00016831615094159365,
      "loss": 0.8499,
      "step": 78820
    },
    {
      "epoch": 1.3839779490510717,
      "grad_norm": 0.055133587207536625,
      "learning_rate": 0.00016830797946694714,
      "loss": 0.8454,
      "step": 78830
    },
    {
      "epoch": 1.3841535139310732,
      "grad_norm": 0.06782507770795224,
      "learning_rate": 0.0001682998071394956,
      "loss": 0.8494,
      "step": 78840
    },
    {
      "epoch": 1.3843290788110747,
      "grad_norm": 0.053408358220809735,
      "learning_rate": 0.00016829163395934264,
      "loss": 0.8511,
      "step": 78850
    },
    {
      "epoch": 1.3845046436910762,
      "grad_norm": 0.07303523900365039,
      "learning_rate": 0.00016828345992659177,
      "loss": 0.843,
      "step": 78860
    },
    {
      "epoch": 1.3846802085710774,
      "grad_norm": 0.05252409863798492,
      "learning_rate": 0.00016827528504134652,
      "loss": 0.8466,
      "step": 78870
    },
    {
      "epoch": 1.384855773451079,
      "grad_norm": 0.047975256649109815,
      "learning_rate": 0.00016826710930371055,
      "loss": 0.8491,
      "step": 78880
    },
    {
      "epoch": 1.3850313383310802,
      "grad_norm": 0.05555473402600641,
      "learning_rate": 0.0001682589327137874,
      "loss": 0.8528,
      "step": 78890
    },
    {
      "epoch": 1.3852069032110816,
      "grad_norm": 0.060204000317047986,
      "learning_rate": 0.0001682507552716807,
      "loss": 0.8439,
      "step": 78900
    },
    {
      "epoch": 1.3853824680910831,
      "grad_norm": 0.09480414066769242,
      "learning_rate": 0.00016824257697749403,
      "loss": 0.8583,
      "step": 78910
    },
    {
      "epoch": 1.3855580329710846,
      "grad_norm": 0.05706154653125506,
      "learning_rate": 0.00016823439783133105,
      "loss": 0.8498,
      "step": 78920
    },
    {
      "epoch": 1.3857335978510859,
      "grad_norm": 0.049794745443520534,
      "learning_rate": 0.00016822621783329532,
      "loss": 0.8453,
      "step": 78930
    },
    {
      "epoch": 1.3859091627310873,
      "grad_norm": 0.06835132828434913,
      "learning_rate": 0.00016821803698349058,
      "loss": 0.8542,
      "step": 78940
    },
    {
      "epoch": 1.3860847276110886,
      "grad_norm": 0.06423733828825791,
      "learning_rate": 0.00016820985528202037,
      "loss": 0.8419,
      "step": 78950
    },
    {
      "epoch": 1.38626029249109,
      "grad_norm": 0.05642716228492373,
      "learning_rate": 0.00016820167272898844,
      "loss": 0.8507,
      "step": 78960
    },
    {
      "epoch": 1.3864358573710915,
      "grad_norm": 0.0571297055788433,
      "learning_rate": 0.00016819348932449845,
      "loss": 0.853,
      "step": 78970
    },
    {
      "epoch": 1.386611422251093,
      "grad_norm": 0.0711808622890846,
      "learning_rate": 0.00016818530506865407,
      "loss": 0.8419,
      "step": 78980
    },
    {
      "epoch": 1.3867869871310943,
      "grad_norm": 0.057332633229737556,
      "learning_rate": 0.00016817711996155898,
      "loss": 0.8459,
      "step": 78990
    },
    {
      "epoch": 1.3869625520110958,
      "grad_norm": 0.05670767767981809,
      "learning_rate": 0.00016816893400331691,
      "loss": 0.8589,
      "step": 79000
    },
    {
      "epoch": 1.387138116891097,
      "grad_norm": 0.0602102817251728,
      "learning_rate": 0.00016816074719403157,
      "loss": 0.8494,
      "step": 79010
    },
    {
      "epoch": 1.3873136817710985,
      "grad_norm": 0.05201930073393287,
      "learning_rate": 0.00016815255953380674,
      "loss": 0.8443,
      "step": 79020
    },
    {
      "epoch": 1.3874892466511,
      "grad_norm": 0.05091621738300832,
      "learning_rate": 0.00016814437102274606,
      "loss": 0.8437,
      "step": 79030
    },
    {
      "epoch": 1.3876648115311014,
      "grad_norm": 0.07472487847754303,
      "learning_rate": 0.0001681361816609533,
      "loss": 0.8465,
      "step": 79040
    },
    {
      "epoch": 1.3878403764111027,
      "grad_norm": 0.08201736894127766,
      "learning_rate": 0.0001681279914485323,
      "loss": 0.8453,
      "step": 79050
    },
    {
      "epoch": 1.3880159412911042,
      "grad_norm": 0.08938824555846861,
      "learning_rate": 0.0001681198003855868,
      "loss": 0.8454,
      "step": 79060
    },
    {
      "epoch": 1.3881915061711054,
      "grad_norm": 0.07611819174169084,
      "learning_rate": 0.00016811160847222053,
      "loss": 0.8492,
      "step": 79070
    },
    {
      "epoch": 1.388367071051107,
      "grad_norm": 0.056792043707831155,
      "learning_rate": 0.00016810341570853735,
      "loss": 0.8523,
      "step": 79080
    },
    {
      "epoch": 1.3885426359311084,
      "grad_norm": 0.05615429547536988,
      "learning_rate": 0.00016809522209464101,
      "loss": 0.8508,
      "step": 79090
    },
    {
      "epoch": 1.3887182008111099,
      "grad_norm": 0.06199155904748937,
      "learning_rate": 0.00016808702763063533,
      "loss": 0.8517,
      "step": 79100
    },
    {
      "epoch": 1.3888937656911111,
      "grad_norm": 0.06347151570954157,
      "learning_rate": 0.00016807883231662418,
      "loss": 0.8445,
      "step": 79110
    },
    {
      "epoch": 1.3890693305711126,
      "grad_norm": 0.07267975854471338,
      "learning_rate": 0.00016807063615271138,
      "loss": 0.8433,
      "step": 79120
    },
    {
      "epoch": 1.3892448954511138,
      "grad_norm": 0.0763009007026998,
      "learning_rate": 0.00016806243913900076,
      "loss": 0.8507,
      "step": 79130
    },
    {
      "epoch": 1.3894204603311153,
      "grad_norm": 0.05692341870416724,
      "learning_rate": 0.00016805424127559616,
      "loss": 0.8531,
      "step": 79140
    },
    {
      "epoch": 1.3895960252111168,
      "grad_norm": 0.0448652377971292,
      "learning_rate": 0.00016804604256260148,
      "loss": 0.848,
      "step": 79150
    },
    {
      "epoch": 1.3897715900911183,
      "grad_norm": 0.057473964021836285,
      "learning_rate": 0.0001680378430001206,
      "loss": 0.8453,
      "step": 79160
    },
    {
      "epoch": 1.3899471549711195,
      "grad_norm": 0.07820397258791767,
      "learning_rate": 0.00016802964258825738,
      "loss": 0.8512,
      "step": 79170
    },
    {
      "epoch": 1.390122719851121,
      "grad_norm": 0.07023807756215208,
      "learning_rate": 0.0001680214413271158,
      "loss": 0.8522,
      "step": 79180
    },
    {
      "epoch": 1.3902982847311223,
      "grad_norm": 0.05790518508918679,
      "learning_rate": 0.00016801323921679967,
      "loss": 0.8444,
      "step": 79190
    },
    {
      "epoch": 1.3904738496111237,
      "grad_norm": 0.04515198520493097,
      "learning_rate": 0.00016800503625741294,
      "loss": 0.8515,
      "step": 79200
    },
    {
      "epoch": 1.3906494144911252,
      "grad_norm": 0.04918464115044383,
      "learning_rate": 0.00016799683244905956,
      "loss": 0.8501,
      "step": 79210
    },
    {
      "epoch": 1.3908249793711267,
      "grad_norm": 0.04418892667215201,
      "learning_rate": 0.0001679886277918435,
      "loss": 0.8527,
      "step": 79220
    },
    {
      "epoch": 1.391000544251128,
      "grad_norm": 0.04702985202102055,
      "learning_rate": 0.0001679804222858687,
      "loss": 0.8438,
      "step": 79230
    },
    {
      "epoch": 1.3911761091311294,
      "grad_norm": 0.08660730790653737,
      "learning_rate": 0.00016797221593123907,
      "loss": 0.838,
      "step": 79240
    },
    {
      "epoch": 1.3913516740111307,
      "grad_norm": 0.056649686906684905,
      "learning_rate": 0.00016796400872805866,
      "loss": 0.8463,
      "step": 79250
    },
    {
      "epoch": 1.3915272388911322,
      "grad_norm": 0.06088877967753432,
      "learning_rate": 0.00016795580067643145,
      "loss": 0.8448,
      "step": 79260
    },
    {
      "epoch": 1.3917028037711336,
      "grad_norm": 0.06011206953590539,
      "learning_rate": 0.00016794759177646135,
      "loss": 0.845,
      "step": 79270
    },
    {
      "epoch": 1.3918783686511351,
      "grad_norm": 0.05800630958411314,
      "learning_rate": 0.00016793938202825246,
      "loss": 0.8529,
      "step": 79280
    },
    {
      "epoch": 1.3920539335311364,
      "grad_norm": 0.08527961574147586,
      "learning_rate": 0.00016793117143190878,
      "loss": 0.848,
      "step": 79290
    },
    {
      "epoch": 1.3922294984111379,
      "grad_norm": 0.05845465183381883,
      "learning_rate": 0.00016792295998753433,
      "loss": 0.8519,
      "step": 79300
    },
    {
      "epoch": 1.3924050632911391,
      "grad_norm": 0.05410498877602398,
      "learning_rate": 0.00016791474769523314,
      "loss": 0.8512,
      "step": 79310
    },
    {
      "epoch": 1.3925806281711406,
      "grad_norm": 0.047412129093375895,
      "learning_rate": 0.0001679065345551093,
      "loss": 0.8522,
      "step": 79320
    },
    {
      "epoch": 1.392756193051142,
      "grad_norm": 0.06430566949215867,
      "learning_rate": 0.0001678983205672668,
      "loss": 0.834,
      "step": 79330
    },
    {
      "epoch": 1.3929317579311435,
      "grad_norm": 0.04293639514967941,
      "learning_rate": 0.00016789010573180978,
      "loss": 0.8481,
      "step": 79340
    },
    {
      "epoch": 1.3931073228111448,
      "grad_norm": 0.08284641414215144,
      "learning_rate": 0.0001678818900488423,
      "loss": 0.8414,
      "step": 79350
    },
    {
      "epoch": 1.3932828876911463,
      "grad_norm": 0.05781833092937468,
      "learning_rate": 0.00016787367351846845,
      "loss": 0.8531,
      "step": 79360
    },
    {
      "epoch": 1.3934584525711475,
      "grad_norm": 0.053493605929655096,
      "learning_rate": 0.00016786545614079235,
      "loss": 0.8408,
      "step": 79370
    },
    {
      "epoch": 1.393634017451149,
      "grad_norm": 0.05499692847789034,
      "learning_rate": 0.0001678572379159181,
      "loss": 0.8512,
      "step": 79380
    },
    {
      "epoch": 1.3938095823311505,
      "grad_norm": 0.052279235834738966,
      "learning_rate": 0.00016784901884394984,
      "loss": 0.8556,
      "step": 79390
    },
    {
      "epoch": 1.393985147211152,
      "grad_norm": 0.07096700037632646,
      "learning_rate": 0.0001678407989249917,
      "loss": 0.8454,
      "step": 79400
    },
    {
      "epoch": 1.3941607120911532,
      "grad_norm": 0.060188456721607865,
      "learning_rate": 0.0001678325781591478,
      "loss": 0.8462,
      "step": 79410
    },
    {
      "epoch": 1.3943362769711547,
      "grad_norm": 0.07786647584791796,
      "learning_rate": 0.00016782435654652235,
      "loss": 0.8578,
      "step": 79420
    },
    {
      "epoch": 1.3945118418511562,
      "grad_norm": 0.05805863295628635,
      "learning_rate": 0.0001678161340872195,
      "loss": 0.844,
      "step": 79430
    },
    {
      "epoch": 1.3946874067311574,
      "grad_norm": 0.08469596498162564,
      "learning_rate": 0.00016780791078134343,
      "loss": 0.8585,
      "step": 79440
    },
    {
      "epoch": 1.394862971611159,
      "grad_norm": 0.06616028960039877,
      "learning_rate": 0.0001677996866289983,
      "loss": 0.8506,
      "step": 79450
    },
    {
      "epoch": 1.3950385364911604,
      "grad_norm": 0.07286813177656401,
      "learning_rate": 0.0001677914616302884,
      "loss": 0.8489,
      "step": 79460
    },
    {
      "epoch": 1.3952141013711616,
      "grad_norm": 0.05687366805607219,
      "learning_rate": 0.0001677832357853178,
      "loss": 0.8467,
      "step": 79470
    },
    {
      "epoch": 1.3953896662511631,
      "grad_norm": 0.05323284452853973,
      "learning_rate": 0.00016777500909419084,
      "loss": 0.8538,
      "step": 79480
    },
    {
      "epoch": 1.3955652311311646,
      "grad_norm": 0.0443920152877011,
      "learning_rate": 0.00016776678155701173,
      "loss": 0.8438,
      "step": 79490
    },
    {
      "epoch": 1.3957407960111659,
      "grad_norm": 0.07235982419387883,
      "learning_rate": 0.00016775855317388466,
      "loss": 0.8501,
      "step": 79500
    },
    {
      "epoch": 1.3959163608911673,
      "grad_norm": 0.07115561093165047,
      "learning_rate": 0.00016775032394491396,
      "loss": 0.8505,
      "step": 79510
    },
    {
      "epoch": 1.3960919257711688,
      "grad_norm": 0.0620121954572445,
      "learning_rate": 0.00016774209387020387,
      "loss": 0.8493,
      "step": 79520
    },
    {
      "epoch": 1.39626749065117,
      "grad_norm": 0.08450451470492404,
      "learning_rate": 0.00016773386294985867,
      "loss": 0.86,
      "step": 79530
    },
    {
      "epoch": 1.3964430555311715,
      "grad_norm": 0.05924780078822191,
      "learning_rate": 0.0001677256311839826,
      "loss": 0.8532,
      "step": 79540
    },
    {
      "epoch": 1.396618620411173,
      "grad_norm": 0.04885497982240299,
      "learning_rate": 0.00016771739857268,
      "loss": 0.852,
      "step": 79550
    },
    {
      "epoch": 1.3967941852911743,
      "grad_norm": 0.0681660737509734,
      "learning_rate": 0.00016770916511605518,
      "loss": 0.8498,
      "step": 79560
    },
    {
      "epoch": 1.3969697501711758,
      "grad_norm": 0.058692000075455227,
      "learning_rate": 0.00016770093081421247,
      "loss": 0.85,
      "step": 79570
    },
    {
      "epoch": 1.3971453150511772,
      "grad_norm": 0.05988779590226452,
      "learning_rate": 0.00016769269566725616,
      "loss": 0.8502,
      "step": 79580
    },
    {
      "epoch": 1.3973208799311787,
      "grad_norm": 0.06154904200556931,
      "learning_rate": 0.00016768445967529062,
      "loss": 0.8551,
      "step": 79590
    },
    {
      "epoch": 1.39749644481118,
      "grad_norm": 0.048274612357811435,
      "learning_rate": 0.00016767622283842023,
      "loss": 0.8434,
      "step": 79600
    },
    {
      "epoch": 1.3976720096911814,
      "grad_norm": 0.05840426567030655,
      "learning_rate": 0.0001676679851567493,
      "loss": 0.8482,
      "step": 79610
    },
    {
      "epoch": 1.3978475745711827,
      "grad_norm": 0.060119999238901864,
      "learning_rate": 0.0001676597466303822,
      "loss": 0.8455,
      "step": 79620
    },
    {
      "epoch": 1.3980231394511842,
      "grad_norm": 0.05604172794776091,
      "learning_rate": 0.00016765150725942335,
      "loss": 0.8503,
      "step": 79630
    },
    {
      "epoch": 1.3981987043311856,
      "grad_norm": 0.07233379678350892,
      "learning_rate": 0.0001676432670439771,
      "loss": 0.8432,
      "step": 79640
    },
    {
      "epoch": 1.3983742692111871,
      "grad_norm": 0.05574474654796039,
      "learning_rate": 0.0001676350259841479,
      "loss": 0.8504,
      "step": 79650
    },
    {
      "epoch": 1.3985498340911884,
      "grad_norm": 0.06798980652102544,
      "learning_rate": 0.0001676267840800402,
      "loss": 0.8466,
      "step": 79660
    },
    {
      "epoch": 1.3987253989711899,
      "grad_norm": 0.06351787894438427,
      "learning_rate": 0.0001676185413317583,
      "loss": 0.8441,
      "step": 79670
    },
    {
      "epoch": 1.3989009638511911,
      "grad_norm": 0.05562847013160359,
      "learning_rate": 0.00016761029773940675,
      "loss": 0.8542,
      "step": 79680
    },
    {
      "epoch": 1.3990765287311926,
      "grad_norm": 0.06826896947532406,
      "learning_rate": 0.00016760205330308995,
      "loss": 0.843,
      "step": 79690
    },
    {
      "epoch": 1.399252093611194,
      "grad_norm": 0.0499224413680815,
      "learning_rate": 0.0001675938080229124,
      "loss": 0.8541,
      "step": 79700
    },
    {
      "epoch": 1.3994276584911955,
      "grad_norm": 0.06070634697106178,
      "learning_rate": 0.00016758556189897853,
      "loss": 0.8559,
      "step": 79710
    },
    {
      "epoch": 1.3996032233711968,
      "grad_norm": 0.06482726683235214,
      "learning_rate": 0.0001675773149313928,
      "loss": 0.8473,
      "step": 79720
    },
    {
      "epoch": 1.3997787882511983,
      "grad_norm": 0.06776975545604932,
      "learning_rate": 0.00016756906712025977,
      "loss": 0.8452,
      "step": 79730
    },
    {
      "epoch": 1.3999543531311995,
      "grad_norm": 0.06299876291278901,
      "learning_rate": 0.00016756081846568387,
      "loss": 0.8427,
      "step": 79740
    },
    {
      "epoch": 1.400129918011201,
      "grad_norm": 0.05141373341853048,
      "learning_rate": 0.00016755256896776967,
      "loss": 0.8441,
      "step": 79750
    },
    {
      "epoch": 1.4003054828912025,
      "grad_norm": 0.06792807011831067,
      "learning_rate": 0.00016754431862662166,
      "loss": 0.8472,
      "step": 79760
    },
    {
      "epoch": 1.400481047771204,
      "grad_norm": 0.08005923894112484,
      "learning_rate": 0.00016753606744234438,
      "loss": 0.8468,
      "step": 79770
    },
    {
      "epoch": 1.4006566126512052,
      "grad_norm": 0.060029051320345306,
      "learning_rate": 0.00016752781541504237,
      "loss": 0.8414,
      "step": 79780
    },
    {
      "epoch": 1.4008321775312067,
      "grad_norm": 0.04883403367973953,
      "learning_rate": 0.00016751956254482014,
      "loss": 0.8507,
      "step": 79790
    },
    {
      "epoch": 1.401007742411208,
      "grad_norm": 0.0583673545144961,
      "learning_rate": 0.00016751130883178238,
      "loss": 0.8452,
      "step": 79800
    },
    {
      "epoch": 1.4011833072912094,
      "grad_norm": 0.07885467103676112,
      "learning_rate": 0.00016750305427603354,
      "loss": 0.8484,
      "step": 79810
    },
    {
      "epoch": 1.401358872171211,
      "grad_norm": 0.06455164855737235,
      "learning_rate": 0.0001674947988776783,
      "loss": 0.851,
      "step": 79820
    },
    {
      "epoch": 1.4015344370512124,
      "grad_norm": 0.05687922599526649,
      "learning_rate": 0.0001674865426368212,
      "loss": 0.8458,
      "step": 79830
    },
    {
      "epoch": 1.4017100019312136,
      "grad_norm": 0.05590273517928278,
      "learning_rate": 0.00016747828555356683,
      "loss": 0.8412,
      "step": 79840
    },
    {
      "epoch": 1.4018855668112151,
      "grad_norm": 0.06222962085003173,
      "learning_rate": 0.0001674700276280199,
      "loss": 0.8508,
      "step": 79850
    },
    {
      "epoch": 1.4020611316912164,
      "grad_norm": 0.09571362319427813,
      "learning_rate": 0.00016746176886028492,
      "loss": 0.8494,
      "step": 79860
    },
    {
      "epoch": 1.4022366965712179,
      "grad_norm": 0.0666946793427891,
      "learning_rate": 0.00016745350925046664,
      "loss": 0.8561,
      "step": 79870
    },
    {
      "epoch": 1.4024122614512193,
      "grad_norm": 0.06079641401660231,
      "learning_rate": 0.00016744524879866963,
      "loss": 0.8454,
      "step": 79880
    },
    {
      "epoch": 1.4025878263312208,
      "grad_norm": 0.052625939384957854,
      "learning_rate": 0.0001674369875049986,
      "loss": 0.8529,
      "step": 79890
    },
    {
      "epoch": 1.402763391211222,
      "grad_norm": 0.06265815583750872,
      "learning_rate": 0.00016742872536955818,
      "loss": 0.8554,
      "step": 79900
    },
    {
      "epoch": 1.4029389560912235,
      "grad_norm": 0.08353979269445572,
      "learning_rate": 0.0001674204623924531,
      "loss": 0.8531,
      "step": 79910
    },
    {
      "epoch": 1.4031145209712248,
      "grad_norm": 0.05034377679050616,
      "learning_rate": 0.00016741219857378806,
      "loss": 0.8517,
      "step": 79920
    },
    {
      "epoch": 1.4032900858512263,
      "grad_norm": 0.053511877541355696,
      "learning_rate": 0.00016740393391366772,
      "loss": 0.8484,
      "step": 79930
    },
    {
      "epoch": 1.4034656507312278,
      "grad_norm": 0.06506930898824914,
      "learning_rate": 0.00016739566841219677,
      "loss": 0.8537,
      "step": 79940
    },
    {
      "epoch": 1.4036412156112292,
      "grad_norm": 0.07447349163683194,
      "learning_rate": 0.00016738740206948,
      "loss": 0.8471,
      "step": 79950
    },
    {
      "epoch": 1.4038167804912305,
      "grad_norm": 0.07901490792668296,
      "learning_rate": 0.00016737913488562215,
      "loss": 0.852,
      "step": 79960
    },
    {
      "epoch": 1.403992345371232,
      "grad_norm": 0.06084866498166804,
      "learning_rate": 0.0001673708668607279,
      "loss": 0.8413,
      "step": 79970
    },
    {
      "epoch": 1.4041679102512332,
      "grad_norm": 0.048140856116035476,
      "learning_rate": 0.00016736259799490202,
      "loss": 0.8491,
      "step": 79980
    },
    {
      "epoch": 1.4043434751312347,
      "grad_norm": 0.04329567645123289,
      "learning_rate": 0.00016735432828824938,
      "loss": 0.8485,
      "step": 79990
    },
    {
      "epoch": 1.4045190400112362,
      "grad_norm": 0.07200779262494765,
      "learning_rate": 0.00016734605774087466,
      "loss": 0.8555,
      "step": 80000
    },
    {
      "epoch": 1.4046946048912377,
      "grad_norm": 0.05315096174957423,
      "learning_rate": 0.00016733778635288267,
      "loss": 0.8486,
      "step": 80010
    },
    {
      "epoch": 1.404870169771239,
      "grad_norm": 0.047884821586913665,
      "learning_rate": 0.0001673295141243782,
      "loss": 0.8521,
      "step": 80020
    },
    {
      "epoch": 1.4050457346512404,
      "grad_norm": 0.04808177917761078,
      "learning_rate": 0.00016732124105546607,
      "loss": 0.8456,
      "step": 80030
    },
    {
      "epoch": 1.4052212995312416,
      "grad_norm": 0.052769744127138855,
      "learning_rate": 0.00016731296714625113,
      "loss": 0.8515,
      "step": 80040
    },
    {
      "epoch": 1.4053968644112431,
      "grad_norm": 0.07892395478675614,
      "learning_rate": 0.0001673046923968382,
      "loss": 0.8518,
      "step": 80050
    },
    {
      "epoch": 1.4055724292912446,
      "grad_norm": 0.06924384266543553,
      "learning_rate": 0.00016729641680733208,
      "loss": 0.852,
      "step": 80060
    },
    {
      "epoch": 1.405747994171246,
      "grad_norm": 0.07000768263983016,
      "learning_rate": 0.00016728814037783764,
      "loss": 0.8518,
      "step": 80070
    },
    {
      "epoch": 1.4059235590512473,
      "grad_norm": 0.06736220886048667,
      "learning_rate": 0.00016727986310845978,
      "loss": 0.8423,
      "step": 80080
    },
    {
      "epoch": 1.4060991239312488,
      "grad_norm": 0.06820437390111074,
      "learning_rate": 0.00016727158499930334,
      "loss": 0.8461,
      "step": 80090
    },
    {
      "epoch": 1.40627468881125,
      "grad_norm": 0.05043150609854019,
      "learning_rate": 0.00016726330605047325,
      "loss": 0.8488,
      "step": 80100
    },
    {
      "epoch": 1.4064502536912515,
      "grad_norm": 0.06001908911003726,
      "learning_rate": 0.00016725502626207435,
      "loss": 0.8457,
      "step": 80110
    },
    {
      "epoch": 1.406625818571253,
      "grad_norm": 0.05811075227051948,
      "learning_rate": 0.00016724674563421154,
      "loss": 0.8532,
      "step": 80120
    },
    {
      "epoch": 1.4068013834512545,
      "grad_norm": 0.06108370094350564,
      "learning_rate": 0.00016723846416698984,
      "loss": 0.8511,
      "step": 80130
    },
    {
      "epoch": 1.4069769483312558,
      "grad_norm": 0.055129669973650486,
      "learning_rate": 0.00016723018186051402,
      "loss": 0.8455,
      "step": 80140
    },
    {
      "epoch": 1.4071525132112572,
      "grad_norm": 0.10033422290674208,
      "learning_rate": 0.00016722189871488916,
      "loss": 0.8489,
      "step": 80150
    },
    {
      "epoch": 1.4073280780912587,
      "grad_norm": 0.045254052155617445,
      "learning_rate": 0.00016721361473022014,
      "loss": 0.8468,
      "step": 80160
    },
    {
      "epoch": 1.40750364297126,
      "grad_norm": 0.06505215589383978,
      "learning_rate": 0.0001672053299066119,
      "loss": 0.8449,
      "step": 80170
    },
    {
      "epoch": 1.4076792078512614,
      "grad_norm": 0.052360306240289506,
      "learning_rate": 0.00016719704424416948,
      "loss": 0.8479,
      "step": 80180
    },
    {
      "epoch": 1.407854772731263,
      "grad_norm": 0.045069864589437104,
      "learning_rate": 0.0001671887577429978,
      "loss": 0.8509,
      "step": 80190
    },
    {
      "epoch": 1.4080303376112642,
      "grad_norm": 0.061195897270228186,
      "learning_rate": 0.0001671804704032019,
      "loss": 0.8496,
      "step": 80200
    },
    {
      "epoch": 1.4082059024912656,
      "grad_norm": 0.06136509338373765,
      "learning_rate": 0.00016717218222488673,
      "loss": 0.852,
      "step": 80210
    },
    {
      "epoch": 1.4083814673712671,
      "grad_norm": 0.07817534219840108,
      "learning_rate": 0.00016716389320815732,
      "loss": 0.8523,
      "step": 80220
    },
    {
      "epoch": 1.4085570322512684,
      "grad_norm": 0.049996458606025404,
      "learning_rate": 0.00016715560335311876,
      "loss": 0.8533,
      "step": 80230
    },
    {
      "epoch": 1.4087325971312699,
      "grad_norm": 0.06790224884776236,
      "learning_rate": 0.00016714731265987597,
      "loss": 0.8475,
      "step": 80240
    },
    {
      "epoch": 1.4089081620112713,
      "grad_norm": 0.06206258548093151,
      "learning_rate": 0.00016713902112853408,
      "loss": 0.8462,
      "step": 80250
    },
    {
      "epoch": 1.4090837268912726,
      "grad_norm": 0.07243583759455179,
      "learning_rate": 0.0001671307287591981,
      "loss": 0.8477,
      "step": 80260
    },
    {
      "epoch": 1.409259291771274,
      "grad_norm": 0.05807279551745136,
      "learning_rate": 0.00016712243555197314,
      "loss": 0.8493,
      "step": 80270
    },
    {
      "epoch": 1.4094348566512755,
      "grad_norm": 0.05773946869174366,
      "learning_rate": 0.00016711414150696424,
      "loss": 0.8501,
      "step": 80280
    },
    {
      "epoch": 1.4096104215312768,
      "grad_norm": 0.07008765154988973,
      "learning_rate": 0.00016710584662427646,
      "loss": 0.8488,
      "step": 80290
    },
    {
      "epoch": 1.4097859864112783,
      "grad_norm": 0.05332802770332509,
      "learning_rate": 0.00016709755090401498,
      "loss": 0.8539,
      "step": 80300
    },
    {
      "epoch": 1.4099615512912798,
      "grad_norm": 0.06237016481335205,
      "learning_rate": 0.00016708925434628485,
      "loss": 0.8467,
      "step": 80310
    },
    {
      "epoch": 1.4101371161712812,
      "grad_norm": 0.04611994129465062,
      "learning_rate": 0.00016708095695119117,
      "loss": 0.8464,
      "step": 80320
    },
    {
      "epoch": 1.4103126810512825,
      "grad_norm": 0.043454836907558504,
      "learning_rate": 0.00016707265871883912,
      "loss": 0.8462,
      "step": 80330
    },
    {
      "epoch": 1.410488245931284,
      "grad_norm": 0.07143960305866896,
      "learning_rate": 0.00016706435964933386,
      "loss": 0.8475,
      "step": 80340
    },
    {
      "epoch": 1.4106638108112852,
      "grad_norm": 0.06157879204717294,
      "learning_rate": 0.00016705605974278045,
      "loss": 0.8466,
      "step": 80350
    },
    {
      "epoch": 1.4108393756912867,
      "grad_norm": 0.06683636727926681,
      "learning_rate": 0.00016704775899928413,
      "loss": 0.8479,
      "step": 80360
    },
    {
      "epoch": 1.4110149405712882,
      "grad_norm": 0.05986512598024288,
      "learning_rate": 0.00016703945741895003,
      "loss": 0.8409,
      "step": 80370
    },
    {
      "epoch": 1.4111905054512897,
      "grad_norm": 0.06726621301965512,
      "learning_rate": 0.00016703115500188337,
      "loss": 0.8475,
      "step": 80380
    },
    {
      "epoch": 1.411366070331291,
      "grad_norm": 0.04549994900267229,
      "learning_rate": 0.0001670228517481893,
      "loss": 0.8543,
      "step": 80390
    },
    {
      "epoch": 1.4115416352112924,
      "grad_norm": 0.06584618799734351,
      "learning_rate": 0.00016701454765797306,
      "loss": 0.8425,
      "step": 80400
    },
    {
      "epoch": 1.4117172000912936,
      "grad_norm": 0.06301732972848835,
      "learning_rate": 0.00016700624273133984,
      "loss": 0.8452,
      "step": 80410
    },
    {
      "epoch": 1.4118927649712951,
      "grad_norm": 0.063154019401425,
      "learning_rate": 0.00016699793696839485,
      "loss": 0.8514,
      "step": 80420
    },
    {
      "epoch": 1.4120683298512966,
      "grad_norm": 0.058541584384326155,
      "learning_rate": 0.00016698963036924339,
      "loss": 0.8486,
      "step": 80430
    },
    {
      "epoch": 1.412243894731298,
      "grad_norm": 0.0672825827331428,
      "learning_rate": 0.00016698132293399066,
      "loss": 0.8515,
      "step": 80440
    },
    {
      "epoch": 1.4124194596112993,
      "grad_norm": 0.06656440865103634,
      "learning_rate": 0.00016697301466274194,
      "loss": 0.85,
      "step": 80450
    },
    {
      "epoch": 1.4125950244913008,
      "grad_norm": 0.06086301546677466,
      "learning_rate": 0.0001669647055556024,
      "loss": 0.8514,
      "step": 80460
    },
    {
      "epoch": 1.412770589371302,
      "grad_norm": 0.09377049866750063,
      "learning_rate": 0.0001669563956126775,
      "loss": 0.8586,
      "step": 80470
    },
    {
      "epoch": 1.4129461542513035,
      "grad_norm": 0.052940420111098105,
      "learning_rate": 0.00016694808483407237,
      "loss": 0.8465,
      "step": 80480
    },
    {
      "epoch": 1.413121719131305,
      "grad_norm": 0.06721437119269852,
      "learning_rate": 0.0001669397732198924,
      "loss": 0.8509,
      "step": 80490
    },
    {
      "epoch": 1.4132972840113065,
      "grad_norm": 0.053955980155329465,
      "learning_rate": 0.00016693146077024283,
      "loss": 0.8467,
      "step": 80500
    },
    {
      "epoch": 1.4134728488913078,
      "grad_norm": 0.05140887591316693,
      "learning_rate": 0.00016692314748522903,
      "loss": 0.852,
      "step": 80510
    },
    {
      "epoch": 1.4136484137713092,
      "grad_norm": 0.07117886362256355,
      "learning_rate": 0.00016691483336495636,
      "loss": 0.854,
      "step": 80520
    },
    {
      "epoch": 1.4138239786513105,
      "grad_norm": 0.07141302732152804,
      "learning_rate": 0.00016690651840953004,
      "loss": 0.8476,
      "step": 80530
    },
    {
      "epoch": 1.413999543531312,
      "grad_norm": 0.05327403536966666,
      "learning_rate": 0.00016689820261905553,
      "loss": 0.8542,
      "step": 80540
    },
    {
      "epoch": 1.4141751084113134,
      "grad_norm": 0.08608376675795494,
      "learning_rate": 0.00016688988599363816,
      "loss": 0.8414,
      "step": 80550
    },
    {
      "epoch": 1.414350673291315,
      "grad_norm": 0.06770721603177154,
      "learning_rate": 0.00016688156853338334,
      "loss": 0.8548,
      "step": 80560
    },
    {
      "epoch": 1.4145262381713162,
      "grad_norm": 0.04831665315232489,
      "learning_rate": 0.00016687325023839638,
      "loss": 0.8499,
      "step": 80570
    },
    {
      "epoch": 1.4147018030513177,
      "grad_norm": 0.04993581057461329,
      "learning_rate": 0.00016686493110878276,
      "loss": 0.8519,
      "step": 80580
    },
    {
      "epoch": 1.414877367931319,
      "grad_norm": 0.04811251431682981,
      "learning_rate": 0.00016685661114464778,
      "loss": 0.8473,
      "step": 80590
    },
    {
      "epoch": 1.4150529328113204,
      "grad_norm": 0.07105266551668855,
      "learning_rate": 0.000166848290346097,
      "loss": 0.8512,
      "step": 80600
    },
    {
      "epoch": 1.4152284976913219,
      "grad_norm": 0.05435750663485612,
      "learning_rate": 0.0001668399687132357,
      "loss": 0.8494,
      "step": 80610
    },
    {
      "epoch": 1.4154040625713233,
      "grad_norm": 0.05739231996224823,
      "learning_rate": 0.0001668316462461694,
      "loss": 0.8523,
      "step": 80620
    },
    {
      "epoch": 1.4155796274513246,
      "grad_norm": 0.0632155244916602,
      "learning_rate": 0.0001668233229450035,
      "loss": 0.8482,
      "step": 80630
    },
    {
      "epoch": 1.415755192331326,
      "grad_norm": 0.06876150753964218,
      "learning_rate": 0.0001668149988098435,
      "loss": 0.8509,
      "step": 80640
    },
    {
      "epoch": 1.4159307572113273,
      "grad_norm": 0.0715239860968698,
      "learning_rate": 0.00016680667384079485,
      "loss": 0.8524,
      "step": 80650
    },
    {
      "epoch": 1.4161063220913288,
      "grad_norm": 0.07304957695481351,
      "learning_rate": 0.00016679834803796302,
      "loss": 0.8474,
      "step": 80660
    },
    {
      "epoch": 1.4162818869713303,
      "grad_norm": 0.055250242734909825,
      "learning_rate": 0.00016679002140145352,
      "loss": 0.8483,
      "step": 80670
    },
    {
      "epoch": 1.4164574518513318,
      "grad_norm": 0.06151511790253977,
      "learning_rate": 0.00016678169393137185,
      "loss": 0.8481,
      "step": 80680
    },
    {
      "epoch": 1.416633016731333,
      "grad_norm": 0.08563207875738589,
      "learning_rate": 0.00016677336562782348,
      "loss": 0.8445,
      "step": 80690
    },
    {
      "epoch": 1.4168085816113345,
      "grad_norm": 0.11257590688993666,
      "learning_rate": 0.00016676503649091403,
      "loss": 0.8494,
      "step": 80700
    },
    {
      "epoch": 1.4169841464913357,
      "grad_norm": 0.07547966629663137,
      "learning_rate": 0.0001667567065207489,
      "loss": 0.8533,
      "step": 80710
    },
    {
      "epoch": 1.4171597113713372,
      "grad_norm": 0.0771635021143166,
      "learning_rate": 0.0001667483757174337,
      "loss": 0.8474,
      "step": 80720
    },
    {
      "epoch": 1.4173352762513387,
      "grad_norm": 0.05811437035027275,
      "learning_rate": 0.000166740044081074,
      "loss": 0.844,
      "step": 80730
    },
    {
      "epoch": 1.4175108411313402,
      "grad_norm": 0.050026539368022156,
      "learning_rate": 0.00016673171161177535,
      "loss": 0.842,
      "step": 80740
    },
    {
      "epoch": 1.4176864060113414,
      "grad_norm": 0.05374637996156687,
      "learning_rate": 0.0001667233783096433,
      "loss": 0.849,
      "step": 80750
    },
    {
      "epoch": 1.417861970891343,
      "grad_norm": 0.0708850722045445,
      "learning_rate": 0.00016671504417478347,
      "loss": 0.8501,
      "step": 80760
    },
    {
      "epoch": 1.4180375357713442,
      "grad_norm": 0.06733848193230633,
      "learning_rate": 0.0001667067092073014,
      "loss": 0.8493,
      "step": 80770
    },
    {
      "epoch": 1.4182131006513456,
      "grad_norm": 0.05788749315730294,
      "learning_rate": 0.00016669837340730275,
      "loss": 0.8515,
      "step": 80780
    },
    {
      "epoch": 1.4183886655313471,
      "grad_norm": 0.053075988905549425,
      "learning_rate": 0.0001666900367748931,
      "loss": 0.8477,
      "step": 80790
    },
    {
      "epoch": 1.4185642304113486,
      "grad_norm": 0.0663407373452303,
      "learning_rate": 0.0001666816993101781,
      "loss": 0.8526,
      "step": 80800
    },
    {
      "epoch": 1.4187397952913499,
      "grad_norm": 0.05510470585519964,
      "learning_rate": 0.0001666733610132634,
      "loss": 0.8583,
      "step": 80810
    },
    {
      "epoch": 1.4189153601713513,
      "grad_norm": 0.06724430881653688,
      "learning_rate": 0.00016666502188425464,
      "loss": 0.856,
      "step": 80820
    },
    {
      "epoch": 1.4190909250513526,
      "grad_norm": 0.06843771075307524,
      "learning_rate": 0.00016665668192325744,
      "loss": 0.8488,
      "step": 80830
    },
    {
      "epoch": 1.419266489931354,
      "grad_norm": 0.07061594249580785,
      "learning_rate": 0.00016664834113037753,
      "loss": 0.8469,
      "step": 80840
    },
    {
      "epoch": 1.4194420548113555,
      "grad_norm": 0.05690046946858986,
      "learning_rate": 0.00016663999950572051,
      "loss": 0.8503,
      "step": 80850
    },
    {
      "epoch": 1.419617619691357,
      "grad_norm": 0.0715613504113208,
      "learning_rate": 0.00016663165704939215,
      "loss": 0.8509,
      "step": 80860
    },
    {
      "epoch": 1.4197931845713583,
      "grad_norm": 0.06258772483823959,
      "learning_rate": 0.00016662331376149814,
      "loss": 0.8517,
      "step": 80870
    },
    {
      "epoch": 1.4199687494513598,
      "grad_norm": 0.06653892892486479,
      "learning_rate": 0.00016661496964214413,
      "loss": 0.8458,
      "step": 80880
    },
    {
      "epoch": 1.4201443143313612,
      "grad_norm": 0.07056399848911848,
      "learning_rate": 0.00016660662469143587,
      "loss": 0.8496,
      "step": 80890
    },
    {
      "epoch": 1.4203198792113625,
      "grad_norm": 0.05267893447498468,
      "learning_rate": 0.00016659827890947912,
      "loss": 0.8473,
      "step": 80900
    },
    {
      "epoch": 1.420495444091364,
      "grad_norm": 0.0922316067623594,
      "learning_rate": 0.00016658993229637964,
      "loss": 0.8387,
      "step": 80910
    },
    {
      "epoch": 1.4206710089713654,
      "grad_norm": 0.06444592424418351,
      "learning_rate": 0.00016658158485224306,
      "loss": 0.8522,
      "step": 80920
    },
    {
      "epoch": 1.4208465738513667,
      "grad_norm": 0.05676799245717015,
      "learning_rate": 0.0001665732365771753,
      "loss": 0.8457,
      "step": 80930
    },
    {
      "epoch": 1.4210221387313682,
      "grad_norm": 0.050412879734346175,
      "learning_rate": 0.00016656488747128208,
      "loss": 0.8535,
      "step": 80940
    },
    {
      "epoch": 1.4211977036113697,
      "grad_norm": 0.046524226771478217,
      "learning_rate": 0.00016655653753466912,
      "loss": 0.843,
      "step": 80950
    },
    {
      "epoch": 1.421373268491371,
      "grad_norm": 0.07127400184191716,
      "learning_rate": 0.00016654818676744226,
      "loss": 0.8492,
      "step": 80960
    },
    {
      "epoch": 1.4215488333713724,
      "grad_norm": 0.0629920590720666,
      "learning_rate": 0.00016653983516970734,
      "loss": 0.8565,
      "step": 80970
    },
    {
      "epoch": 1.4217243982513739,
      "grad_norm": 0.06498636695386655,
      "learning_rate": 0.00016653148274157014,
      "loss": 0.8523,
      "step": 80980
    },
    {
      "epoch": 1.4218999631313751,
      "grad_norm": 0.0882525169240322,
      "learning_rate": 0.0001665231294831365,
      "loss": 0.8537,
      "step": 80990
    },
    {
      "epoch": 1.4220755280113766,
      "grad_norm": 0.053451584159286775,
      "learning_rate": 0.00016651477539451223,
      "loss": 0.8476,
      "step": 81000
    },
    {
      "epoch": 1.422251092891378,
      "grad_norm": 0.0575202652742174,
      "learning_rate": 0.00016650642047580323,
      "loss": 0.8575,
      "step": 81010
    },
    {
      "epoch": 1.4224266577713793,
      "grad_norm": 0.039878215272903306,
      "learning_rate": 0.0001664980647271153,
      "loss": 0.8599,
      "step": 81020
    },
    {
      "epoch": 1.4226022226513808,
      "grad_norm": 0.061318468466322634,
      "learning_rate": 0.0001664897081485544,
      "loss": 0.8509,
      "step": 81030
    },
    {
      "epoch": 1.4227777875313823,
      "grad_norm": 0.061726354119980544,
      "learning_rate": 0.00016648135074022629,
      "loss": 0.8467,
      "step": 81040
    },
    {
      "epoch": 1.4229533524113838,
      "grad_norm": 0.06559577890286558,
      "learning_rate": 0.00016647299250223692,
      "loss": 0.851,
      "step": 81050
    },
    {
      "epoch": 1.423128917291385,
      "grad_norm": 0.05893476402516462,
      "learning_rate": 0.00016646463343469222,
      "loss": 0.8495,
      "step": 81060
    },
    {
      "epoch": 1.4233044821713865,
      "grad_norm": 0.05391029353417367,
      "learning_rate": 0.00016645627353769802,
      "loss": 0.8467,
      "step": 81070
    },
    {
      "epoch": 1.4234800470513878,
      "grad_norm": 0.06387303672039503,
      "learning_rate": 0.0001664479128113603,
      "loss": 0.8533,
      "step": 81080
    },
    {
      "epoch": 1.4236556119313892,
      "grad_norm": 0.061969887956094656,
      "learning_rate": 0.00016643955125578502,
      "loss": 0.8508,
      "step": 81090
    },
    {
      "epoch": 1.4238311768113907,
      "grad_norm": 0.04680135502872836,
      "learning_rate": 0.00016643118887107804,
      "loss": 0.8519,
      "step": 81100
    },
    {
      "epoch": 1.4240067416913922,
      "grad_norm": 0.05429387205143117,
      "learning_rate": 0.00016642282565734536,
      "loss": 0.8513,
      "step": 81110
    },
    {
      "epoch": 1.4241823065713934,
      "grad_norm": 0.0529076558652055,
      "learning_rate": 0.000166414461614693,
      "loss": 0.8474,
      "step": 81120
    },
    {
      "epoch": 1.424357871451395,
      "grad_norm": 0.061439474769756044,
      "learning_rate": 0.0001664060967432268,
      "loss": 0.8468,
      "step": 81130
    },
    {
      "epoch": 1.4245334363313962,
      "grad_norm": 0.06226914967962695,
      "learning_rate": 0.00016639773104305285,
      "loss": 0.8498,
      "step": 81140
    },
    {
      "epoch": 1.4247090012113977,
      "grad_norm": 0.04020420233785825,
      "learning_rate": 0.00016638936451427713,
      "loss": 0.8544,
      "step": 81150
    },
    {
      "epoch": 1.4248845660913991,
      "grad_norm": 0.04986726675399071,
      "learning_rate": 0.00016638099715700564,
      "loss": 0.8546,
      "step": 81160
    },
    {
      "epoch": 1.4250601309714006,
      "grad_norm": 0.04357940597462885,
      "learning_rate": 0.00016637262897134433,
      "loss": 0.8494,
      "step": 81170
    },
    {
      "epoch": 1.4252356958514019,
      "grad_norm": 0.055826780929132534,
      "learning_rate": 0.00016636425995739934,
      "loss": 0.8507,
      "step": 81180
    },
    {
      "epoch": 1.4254112607314033,
      "grad_norm": 0.06555626165159685,
      "learning_rate": 0.0001663558901152766,
      "loss": 0.8539,
      "step": 81190
    },
    {
      "epoch": 1.4255868256114046,
      "grad_norm": 0.06535223110610641,
      "learning_rate": 0.00016634751944508223,
      "loss": 0.8472,
      "step": 81200
    },
    {
      "epoch": 1.425762390491406,
      "grad_norm": 0.051787666827150726,
      "learning_rate": 0.00016633914794692226,
      "loss": 0.8466,
      "step": 81210
    },
    {
      "epoch": 1.4259379553714076,
      "grad_norm": 0.06212584807396637,
      "learning_rate": 0.00016633077562090275,
      "loss": 0.853,
      "step": 81220
    },
    {
      "epoch": 1.426113520251409,
      "grad_norm": 0.05084729890019404,
      "learning_rate": 0.00016632240246712985,
      "loss": 0.8491,
      "step": 81230
    },
    {
      "epoch": 1.4262890851314103,
      "grad_norm": 0.06273268135393979,
      "learning_rate": 0.00016631402848570954,
      "loss": 0.8516,
      "step": 81240
    },
    {
      "epoch": 1.4264646500114118,
      "grad_norm": 0.06267495906146463,
      "learning_rate": 0.00016630565367674799,
      "loss": 0.8584,
      "step": 81250
    },
    {
      "epoch": 1.426640214891413,
      "grad_norm": 0.057761233457685385,
      "learning_rate": 0.00016629727804035128,
      "loss": 0.8503,
      "step": 81260
    },
    {
      "epoch": 1.4268157797714145,
      "grad_norm": 0.059981895558747785,
      "learning_rate": 0.00016628890157662556,
      "loss": 0.85,
      "step": 81270
    },
    {
      "epoch": 1.426991344651416,
      "grad_norm": 0.05558639966814963,
      "learning_rate": 0.00016628052428567693,
      "loss": 0.8526,
      "step": 81280
    },
    {
      "epoch": 1.4271669095314174,
      "grad_norm": 0.05811373842918574,
      "learning_rate": 0.00016627214616761154,
      "loss": 0.8478,
      "step": 81290
    },
    {
      "epoch": 1.4273424744114187,
      "grad_norm": 0.0719687935629689,
      "learning_rate": 0.00016626376722253555,
      "loss": 0.8439,
      "step": 81300
    },
    {
      "epoch": 1.4275180392914202,
      "grad_norm": 0.058390661752836595,
      "learning_rate": 0.00016625538745055512,
      "loss": 0.8507,
      "step": 81310
    },
    {
      "epoch": 1.4276936041714214,
      "grad_norm": 0.0555816186649208,
      "learning_rate": 0.00016624700685177645,
      "loss": 0.8501,
      "step": 81320
    },
    {
      "epoch": 1.427869169051423,
      "grad_norm": 0.0674769552386713,
      "learning_rate": 0.0001662386254263057,
      "loss": 0.8503,
      "step": 81330
    },
    {
      "epoch": 1.4280447339314244,
      "grad_norm": 0.06113339159514674,
      "learning_rate": 0.00016623024317424903,
      "loss": 0.8529,
      "step": 81340
    },
    {
      "epoch": 1.4282202988114259,
      "grad_norm": 0.060408720216776186,
      "learning_rate": 0.00016622186009571268,
      "loss": 0.8517,
      "step": 81350
    },
    {
      "epoch": 1.4283958636914271,
      "grad_norm": 0.061103079774586455,
      "learning_rate": 0.00016621347619080287,
      "loss": 0.8488,
      "step": 81360
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 0.0620235064494806,
      "learning_rate": 0.00016620509145962584,
      "loss": 0.8467,
      "step": 81370
    },
    {
      "epoch": 1.4287469934514299,
      "grad_norm": 0.06323893700047466,
      "learning_rate": 0.0001661967059022878,
      "loss": 0.8508,
      "step": 81380
    },
    {
      "epoch": 1.4289225583314313,
      "grad_norm": 0.05795277704850642,
      "learning_rate": 0.000166188319518895,
      "loss": 0.8478,
      "step": 81390
    },
    {
      "epoch": 1.4290981232114328,
      "grad_norm": 0.08837848881221459,
      "learning_rate": 0.0001661799323095537,
      "loss": 0.8515,
      "step": 81400
    },
    {
      "epoch": 1.4292736880914343,
      "grad_norm": 0.06375950244576993,
      "learning_rate": 0.00016617154427437013,
      "loss": 0.8473,
      "step": 81410
    },
    {
      "epoch": 1.4294492529714355,
      "grad_norm": 0.06698571555624108,
      "learning_rate": 0.00016616315541345065,
      "loss": 0.8428,
      "step": 81420
    },
    {
      "epoch": 1.429624817851437,
      "grad_norm": 0.05157190999690618,
      "learning_rate": 0.00016615476572690146,
      "loss": 0.8438,
      "step": 81430
    },
    {
      "epoch": 1.4298003827314383,
      "grad_norm": 0.03986045202689041,
      "learning_rate": 0.0001661463752148289,
      "loss": 0.8477,
      "step": 81440
    },
    {
      "epoch": 1.4299759476114398,
      "grad_norm": 0.05881115207930584,
      "learning_rate": 0.0001661379838773393,
      "loss": 0.8532,
      "step": 81450
    },
    {
      "epoch": 1.4301515124914412,
      "grad_norm": 0.07434109866335734,
      "learning_rate": 0.000166129591714539,
      "loss": 0.8434,
      "step": 81460
    },
    {
      "epoch": 1.4303270773714427,
      "grad_norm": 0.05399343587026232,
      "learning_rate": 0.00016612119872653426,
      "loss": 0.8528,
      "step": 81470
    },
    {
      "epoch": 1.430502642251444,
      "grad_norm": 0.07188516252819399,
      "learning_rate": 0.00016611280491343142,
      "loss": 0.8463,
      "step": 81480
    },
    {
      "epoch": 1.4306782071314454,
      "grad_norm": 0.06324962723208821,
      "learning_rate": 0.0001661044102753369,
      "loss": 0.8529,
      "step": 81490
    },
    {
      "epoch": 1.4308537720114467,
      "grad_norm": 0.09136438715595296,
      "learning_rate": 0.000166096014812357,
      "loss": 0.8486,
      "step": 81500
    },
    {
      "epoch": 1.4310293368914482,
      "grad_norm": 0.09771243497291682,
      "learning_rate": 0.00016608761852459814,
      "loss": 0.8398,
      "step": 81510
    },
    {
      "epoch": 1.4312049017714497,
      "grad_norm": 0.07331110550954995,
      "learning_rate": 0.00016607922141216667,
      "loss": 0.8473,
      "step": 81520
    },
    {
      "epoch": 1.4313804666514511,
      "grad_norm": 0.06708247996388103,
      "learning_rate": 0.000166070823475169,
      "loss": 0.8424,
      "step": 81530
    },
    {
      "epoch": 1.4315560315314524,
      "grad_norm": 0.12329089348265725,
      "learning_rate": 0.00016606242471371152,
      "loss": 0.8487,
      "step": 81540
    },
    {
      "epoch": 1.4317315964114539,
      "grad_norm": 0.05183586870098595,
      "learning_rate": 0.00016605402512790063,
      "loss": 0.8485,
      "step": 81550
    },
    {
      "epoch": 1.4319071612914551,
      "grad_norm": 0.0646207466466142,
      "learning_rate": 0.0001660456247178428,
      "loss": 0.8497,
      "step": 81560
    },
    {
      "epoch": 1.4320827261714566,
      "grad_norm": 0.06697027095790936,
      "learning_rate": 0.00016603722348364444,
      "loss": 0.8632,
      "step": 81570
    },
    {
      "epoch": 1.432258291051458,
      "grad_norm": 0.07446657797285865,
      "learning_rate": 0.00016602882142541196,
      "loss": 0.8451,
      "step": 81580
    },
    {
      "epoch": 1.4324338559314596,
      "grad_norm": 0.06607769473796238,
      "learning_rate": 0.0001660204185432519,
      "loss": 0.8513,
      "step": 81590
    },
    {
      "epoch": 1.4326094208114608,
      "grad_norm": 0.06242651524695068,
      "learning_rate": 0.00016601201483727065,
      "loss": 0.8479,
      "step": 81600
    },
    {
      "epoch": 1.4327849856914623,
      "grad_norm": 0.055657406672754275,
      "learning_rate": 0.00016600361030757473,
      "loss": 0.8542,
      "step": 81610
    },
    {
      "epoch": 1.4329605505714638,
      "grad_norm": 0.05284858111903712,
      "learning_rate": 0.0001659952049542706,
      "loss": 0.8506,
      "step": 81620
    },
    {
      "epoch": 1.433136115451465,
      "grad_norm": 0.04955510222906755,
      "learning_rate": 0.00016598679877746474,
      "loss": 0.8508,
      "step": 81630
    },
    {
      "epoch": 1.4333116803314665,
      "grad_norm": 0.07368767858491977,
      "learning_rate": 0.0001659783917772637,
      "loss": 0.8556,
      "step": 81640
    },
    {
      "epoch": 1.433487245211468,
      "grad_norm": 0.05556705233878034,
      "learning_rate": 0.00016596998395377398,
      "loss": 0.8486,
      "step": 81650
    },
    {
      "epoch": 1.4336628100914692,
      "grad_norm": 0.07948674157324008,
      "learning_rate": 0.00016596157530710212,
      "loss": 0.8501,
      "step": 81660
    },
    {
      "epoch": 1.4338383749714707,
      "grad_norm": 0.06648287269144297,
      "learning_rate": 0.00016595316583735466,
      "loss": 0.8414,
      "step": 81670
    },
    {
      "epoch": 1.4340139398514722,
      "grad_norm": 0.07041830993190713,
      "learning_rate": 0.00016594475554463812,
      "loss": 0.8492,
      "step": 81680
    },
    {
      "epoch": 1.4341895047314734,
      "grad_norm": 0.06885257675439539,
      "learning_rate": 0.0001659363444290591,
      "loss": 0.8503,
      "step": 81690
    },
    {
      "epoch": 1.434365069611475,
      "grad_norm": 0.051308979577393665,
      "learning_rate": 0.00016592793249072416,
      "loss": 0.8464,
      "step": 81700
    },
    {
      "epoch": 1.4345406344914764,
      "grad_norm": 0.04913027586347958,
      "learning_rate": 0.00016591951972973984,
      "loss": 0.8547,
      "step": 81710
    },
    {
      "epoch": 1.4347161993714777,
      "grad_norm": 0.05417085238021662,
      "learning_rate": 0.00016591110614621278,
      "loss": 0.8492,
      "step": 81720
    },
    {
      "epoch": 1.4348917642514791,
      "grad_norm": 0.05587054149041163,
      "learning_rate": 0.00016590269174024958,
      "loss": 0.8478,
      "step": 81730
    },
    {
      "epoch": 1.4350673291314806,
      "grad_norm": 0.048485853618943024,
      "learning_rate": 0.00016589427651195684,
      "loss": 0.848,
      "step": 81740
    },
    {
      "epoch": 1.4352428940114819,
      "grad_norm": 0.06471525365516678,
      "learning_rate": 0.00016588586046144114,
      "loss": 0.8413,
      "step": 81750
    },
    {
      "epoch": 1.4354184588914833,
      "grad_norm": 0.052107167711944624,
      "learning_rate": 0.00016587744358880917,
      "loss": 0.8508,
      "step": 81760
    },
    {
      "epoch": 1.4355940237714848,
      "grad_norm": 0.05682243999027412,
      "learning_rate": 0.00016586902589416755,
      "loss": 0.8512,
      "step": 81770
    },
    {
      "epoch": 1.4357695886514863,
      "grad_norm": 0.05891079161437684,
      "learning_rate": 0.00016586060737762296,
      "loss": 0.844,
      "step": 81780
    },
    {
      "epoch": 1.4359451535314876,
      "grad_norm": 0.06307454563568696,
      "learning_rate": 0.00016585218803928203,
      "loss": 0.8467,
      "step": 81790
    },
    {
      "epoch": 1.436120718411489,
      "grad_norm": 0.051401804872663034,
      "learning_rate": 0.00016584376787925144,
      "loss": 0.8426,
      "step": 81800
    },
    {
      "epoch": 1.4362962832914903,
      "grad_norm": 0.05362755633021735,
      "learning_rate": 0.00016583534689763792,
      "loss": 0.8457,
      "step": 81810
    },
    {
      "epoch": 1.4364718481714918,
      "grad_norm": 0.07928538027659685,
      "learning_rate": 0.00016582692509454808,
      "loss": 0.8518,
      "step": 81820
    },
    {
      "epoch": 1.4366474130514932,
      "grad_norm": 0.053000829459285345,
      "learning_rate": 0.0001658185024700887,
      "loss": 0.8527,
      "step": 81830
    },
    {
      "epoch": 1.4368229779314947,
      "grad_norm": 0.05562059742252245,
      "learning_rate": 0.00016581007902436646,
      "loss": 0.8547,
      "step": 81840
    },
    {
      "epoch": 1.436998542811496,
      "grad_norm": 0.053263224761995305,
      "learning_rate": 0.00016580165475748812,
      "loss": 0.8486,
      "step": 81850
    },
    {
      "epoch": 1.4371741076914974,
      "grad_norm": 0.06257231971634099,
      "learning_rate": 0.00016579322966956038,
      "loss": 0.8464,
      "step": 81860
    },
    {
      "epoch": 1.4373496725714987,
      "grad_norm": 0.0586297690994776,
      "learning_rate": 0.00016578480376069001,
      "loss": 0.8507,
      "step": 81870
    },
    {
      "epoch": 1.4375252374515002,
      "grad_norm": 0.06171643193441123,
      "learning_rate": 0.00016577637703098376,
      "loss": 0.8532,
      "step": 81880
    },
    {
      "epoch": 1.4377008023315017,
      "grad_norm": 0.05528167056162864,
      "learning_rate": 0.00016576794948054842,
      "loss": 0.8444,
      "step": 81890
    },
    {
      "epoch": 1.4378763672115031,
      "grad_norm": 0.06580397591747844,
      "learning_rate": 0.00016575952110949068,
      "loss": 0.8503,
      "step": 81900
    },
    {
      "epoch": 1.4380519320915044,
      "grad_norm": 0.0608182540680753,
      "learning_rate": 0.00016575109191791746,
      "loss": 0.8553,
      "step": 81910
    },
    {
      "epoch": 1.4382274969715059,
      "grad_norm": 0.05277243169276416,
      "learning_rate": 0.00016574266190593547,
      "loss": 0.8525,
      "step": 81920
    },
    {
      "epoch": 1.4384030618515071,
      "grad_norm": 0.11068214863161625,
      "learning_rate": 0.00016573423107365157,
      "loss": 0.8451,
      "step": 81930
    },
    {
      "epoch": 1.4385786267315086,
      "grad_norm": 0.06849025260690104,
      "learning_rate": 0.00016572579942117255,
      "loss": 0.8501,
      "step": 81940
    },
    {
      "epoch": 1.43875419161151,
      "grad_norm": 0.04822011983380511,
      "learning_rate": 0.0001657173669486052,
      "loss": 0.85,
      "step": 81950
    },
    {
      "epoch": 1.4389297564915116,
      "grad_norm": 0.06489428690518545,
      "learning_rate": 0.00016570893365605647,
      "loss": 0.8444,
      "step": 81960
    },
    {
      "epoch": 1.4391053213715128,
      "grad_norm": 0.06402113775475579,
      "learning_rate": 0.00016570049954363312,
      "loss": 0.8529,
      "step": 81970
    },
    {
      "epoch": 1.4392808862515143,
      "grad_norm": 0.04553408421155728,
      "learning_rate": 0.00016569206461144204,
      "loss": 0.8465,
      "step": 81980
    },
    {
      "epoch": 1.4394564511315155,
      "grad_norm": 0.05634814547259967,
      "learning_rate": 0.0001656836288595901,
      "loss": 0.8497,
      "step": 81990
    },
    {
      "epoch": 1.439632016011517,
      "grad_norm": 0.06299148352068196,
      "learning_rate": 0.00016567519228818418,
      "loss": 0.8451,
      "step": 82000
    },
    {
      "epoch": 1.4398075808915185,
      "grad_norm": 0.05339113141962948,
      "learning_rate": 0.0001656667548973312,
      "loss": 0.8486,
      "step": 82010
    },
    {
      "epoch": 1.43998314577152,
      "grad_norm": 0.06741599750643325,
      "learning_rate": 0.00016565831668713803,
      "loss": 0.8467,
      "step": 82020
    },
    {
      "epoch": 1.4401587106515212,
      "grad_norm": 0.06613869107629984,
      "learning_rate": 0.0001656498776577116,
      "loss": 0.8427,
      "step": 82030
    },
    {
      "epoch": 1.4403342755315227,
      "grad_norm": 0.05106372330010516,
      "learning_rate": 0.0001656414378091588,
      "loss": 0.8454,
      "step": 82040
    },
    {
      "epoch": 1.440509840411524,
      "grad_norm": 0.04480645203991791,
      "learning_rate": 0.00016563299714158664,
      "loss": 0.8476,
      "step": 82050
    },
    {
      "epoch": 1.4406854052915254,
      "grad_norm": 0.0497163642930676,
      "learning_rate": 0.000165624555655102,
      "loss": 0.8427,
      "step": 82060
    },
    {
      "epoch": 1.440860970171527,
      "grad_norm": 0.04155836953612483,
      "learning_rate": 0.00016561611334981184,
      "loss": 0.8457,
      "step": 82070
    },
    {
      "epoch": 1.4410365350515284,
      "grad_norm": 0.051953023074655436,
      "learning_rate": 0.00016560767022582318,
      "loss": 0.8427,
      "step": 82080
    },
    {
      "epoch": 1.4412120999315297,
      "grad_norm": 0.05988964207114093,
      "learning_rate": 0.00016559922628324293,
      "loss": 0.8527,
      "step": 82090
    },
    {
      "epoch": 1.4413876648115311,
      "grad_norm": 0.0632986658250819,
      "learning_rate": 0.00016559078152217813,
      "loss": 0.8499,
      "step": 82100
    },
    {
      "epoch": 1.4415632296915324,
      "grad_norm": 0.05574400948195976,
      "learning_rate": 0.00016558233594273568,
      "loss": 0.8461,
      "step": 82110
    },
    {
      "epoch": 1.4417387945715339,
      "grad_norm": 0.06335068320433018,
      "learning_rate": 0.0001655738895450227,
      "loss": 0.8537,
      "step": 82120
    },
    {
      "epoch": 1.4419143594515353,
      "grad_norm": 0.052641206780308286,
      "learning_rate": 0.0001655654423291462,
      "loss": 0.8559,
      "step": 82130
    },
    {
      "epoch": 1.4420899243315368,
      "grad_norm": 0.07566180787884327,
      "learning_rate": 0.0001655569942952131,
      "loss": 0.8503,
      "step": 82140
    },
    {
      "epoch": 1.442265489211538,
      "grad_norm": 0.05021101734129216,
      "learning_rate": 0.0001655485454433306,
      "loss": 0.8512,
      "step": 82150
    },
    {
      "epoch": 1.4424410540915396,
      "grad_norm": 0.06673553609808773,
      "learning_rate": 0.00016554009577360556,
      "loss": 0.8487,
      "step": 82160
    },
    {
      "epoch": 1.4426166189715408,
      "grad_norm": 0.05936777009905232,
      "learning_rate": 0.0001655316452861452,
      "loss": 0.8523,
      "step": 82170
    },
    {
      "epoch": 1.4427921838515423,
      "grad_norm": 0.049148136786596096,
      "learning_rate": 0.0001655231939810565,
      "loss": 0.8485,
      "step": 82180
    },
    {
      "epoch": 1.4429677487315438,
      "grad_norm": 0.056237701206464846,
      "learning_rate": 0.00016551474185844657,
      "loss": 0.8498,
      "step": 82190
    },
    {
      "epoch": 1.4431433136115452,
      "grad_norm": 0.053333551322937586,
      "learning_rate": 0.00016550628891842252,
      "loss": 0.8442,
      "step": 82200
    },
    {
      "epoch": 1.4433188784915465,
      "grad_norm": 0.05332003937787498,
      "learning_rate": 0.0001654978351610914,
      "loss": 0.8586,
      "step": 82210
    },
    {
      "epoch": 1.443494443371548,
      "grad_norm": 0.0622895199246092,
      "learning_rate": 0.00016548938058656037,
      "loss": 0.8484,
      "step": 82220
    },
    {
      "epoch": 1.4436700082515492,
      "grad_norm": 0.05646201757485554,
      "learning_rate": 0.0001654809251949365,
      "loss": 0.848,
      "step": 82230
    },
    {
      "epoch": 1.4438455731315507,
      "grad_norm": 0.0586135558777663,
      "learning_rate": 0.00016547246898632696,
      "loss": 0.8499,
      "step": 82240
    },
    {
      "epoch": 1.4440211380115522,
      "grad_norm": 0.05858635713115686,
      "learning_rate": 0.00016546401196083888,
      "loss": 0.851,
      "step": 82250
    },
    {
      "epoch": 1.4441967028915537,
      "grad_norm": 0.05669222807484035,
      "learning_rate": 0.00016545555411857945,
      "loss": 0.8527,
      "step": 82260
    },
    {
      "epoch": 1.444372267771555,
      "grad_norm": 0.05294480643399446,
      "learning_rate": 0.00016544709545965577,
      "loss": 0.847,
      "step": 82270
    },
    {
      "epoch": 1.4445478326515564,
      "grad_norm": 0.0701926658759587,
      "learning_rate": 0.00016543863598417503,
      "loss": 0.8379,
      "step": 82280
    },
    {
      "epoch": 1.4447233975315577,
      "grad_norm": 0.05516968519372344,
      "learning_rate": 0.00016543017569224444,
      "loss": 0.8517,
      "step": 82290
    },
    {
      "epoch": 1.4448989624115591,
      "grad_norm": 0.049788854389740754,
      "learning_rate": 0.00016542171458397117,
      "loss": 0.8547,
      "step": 82300
    },
    {
      "epoch": 1.4450745272915606,
      "grad_norm": 0.08829046259936403,
      "learning_rate": 0.0001654132526594624,
      "loss": 0.8536,
      "step": 82310
    },
    {
      "epoch": 1.445250092171562,
      "grad_norm": 0.05105319996191013,
      "learning_rate": 0.0001654047899188254,
      "loss": 0.8539,
      "step": 82320
    },
    {
      "epoch": 1.4454256570515633,
      "grad_norm": 0.06952533071417957,
      "learning_rate": 0.00016539632636216738,
      "loss": 0.847,
      "step": 82330
    },
    {
      "epoch": 1.4456012219315648,
      "grad_norm": 0.06139622014626932,
      "learning_rate": 0.00016538786198959555,
      "loss": 0.8522,
      "step": 82340
    },
    {
      "epoch": 1.4457767868115663,
      "grad_norm": 0.0751115200031959,
      "learning_rate": 0.00016537939680121716,
      "loss": 0.8499,
      "step": 82350
    },
    {
      "epoch": 1.4459523516915676,
      "grad_norm": 0.051252937127982914,
      "learning_rate": 0.00016537093079713948,
      "loss": 0.8476,
      "step": 82360
    },
    {
      "epoch": 1.446127916571569,
      "grad_norm": 0.09184272665107596,
      "learning_rate": 0.0001653624639774698,
      "loss": 0.8494,
      "step": 82370
    },
    {
      "epoch": 1.4463034814515705,
      "grad_norm": 0.09774901348923852,
      "learning_rate": 0.00016535399634231532,
      "loss": 0.8535,
      "step": 82380
    },
    {
      "epoch": 1.4464790463315718,
      "grad_norm": 0.09097648279859763,
      "learning_rate": 0.00016534552789178338,
      "loss": 0.8435,
      "step": 82390
    },
    {
      "epoch": 1.4466546112115732,
      "grad_norm": 0.05205209323808941,
      "learning_rate": 0.00016533705862598131,
      "loss": 0.8482,
      "step": 82400
    },
    {
      "epoch": 1.4468301760915747,
      "grad_norm": 0.06821978544268174,
      "learning_rate": 0.00016532858854501634,
      "loss": 0.8443,
      "step": 82410
    },
    {
      "epoch": 1.447005740971576,
      "grad_norm": 0.05651481654830118,
      "learning_rate": 0.00016532011764899585,
      "loss": 0.8462,
      "step": 82420
    },
    {
      "epoch": 1.4471813058515774,
      "grad_norm": 0.059524699556812076,
      "learning_rate": 0.0001653116459380271,
      "loss": 0.8511,
      "step": 82430
    },
    {
      "epoch": 1.447356870731579,
      "grad_norm": 0.05181151595439115,
      "learning_rate": 0.00016530317341221753,
      "loss": 0.8556,
      "step": 82440
    },
    {
      "epoch": 1.4475324356115802,
      "grad_norm": 0.05962709418244785,
      "learning_rate": 0.00016529470007167437,
      "loss": 0.8528,
      "step": 82450
    },
    {
      "epoch": 1.4477080004915817,
      "grad_norm": 0.055199076211060974,
      "learning_rate": 0.0001652862259165051,
      "loss": 0.8418,
      "step": 82460
    },
    {
      "epoch": 1.4478835653715831,
      "grad_norm": 0.06690756759506102,
      "learning_rate": 0.000165277750946817,
      "loss": 0.8435,
      "step": 82470
    },
    {
      "epoch": 1.4480591302515844,
      "grad_norm": 0.07657314676329585,
      "learning_rate": 0.00016526927516271743,
      "loss": 0.8501,
      "step": 82480
    },
    {
      "epoch": 1.4482346951315859,
      "grad_norm": 0.04764099759795592,
      "learning_rate": 0.00016526079856431388,
      "loss": 0.8561,
      "step": 82490
    },
    {
      "epoch": 1.4484102600115873,
      "grad_norm": 0.05729134826122519,
      "learning_rate": 0.00016525232115171368,
      "loss": 0.8503,
      "step": 82500
    },
    {
      "epoch": 1.4485858248915888,
      "grad_norm": 0.058505868935257097,
      "learning_rate": 0.00016524384292502427,
      "loss": 0.841,
      "step": 82510
    },
    {
      "epoch": 1.44876138977159,
      "grad_norm": 0.05474617460730219,
      "learning_rate": 0.00016523536388435303,
      "loss": 0.8469,
      "step": 82520
    },
    {
      "epoch": 1.4489369546515916,
      "grad_norm": 0.0572850499550895,
      "learning_rate": 0.00016522688402980743,
      "loss": 0.853,
      "step": 82530
    },
    {
      "epoch": 1.4491125195315928,
      "grad_norm": 0.07818363016703796,
      "learning_rate": 0.00016521840336149494,
      "loss": 0.8446,
      "step": 82540
    },
    {
      "epoch": 1.4492880844115943,
      "grad_norm": 0.05875160174067679,
      "learning_rate": 0.0001652099218795229,
      "loss": 0.8478,
      "step": 82550
    },
    {
      "epoch": 1.4494636492915958,
      "grad_norm": 0.06557796067218928,
      "learning_rate": 0.0001652014395839989,
      "loss": 0.8449,
      "step": 82560
    },
    {
      "epoch": 1.4496392141715972,
      "grad_norm": 0.0712333135236993,
      "learning_rate": 0.00016519295647503038,
      "loss": 0.8523,
      "step": 82570
    },
    {
      "epoch": 1.4498147790515985,
      "grad_norm": 0.06038061081455513,
      "learning_rate": 0.0001651844725527247,
      "loss": 0.857,
      "step": 82580
    },
    {
      "epoch": 1.4499903439316,
      "grad_norm": 0.08669381286246113,
      "learning_rate": 0.00016517598781718953,
      "loss": 0.8498,
      "step": 82590
    },
    {
      "epoch": 1.4501659088116012,
      "grad_norm": 0.0728078319195362,
      "learning_rate": 0.0001651675022685323,
      "loss": 0.854,
      "step": 82600
    },
    {
      "epoch": 1.4503414736916027,
      "grad_norm": 0.046159052408452084,
      "learning_rate": 0.00016515901590686047,
      "loss": 0.8519,
      "step": 82610
    },
    {
      "epoch": 1.4505170385716042,
      "grad_norm": 0.06888371081698937,
      "learning_rate": 0.00016515052873228162,
      "loss": 0.8483,
      "step": 82620
    },
    {
      "epoch": 1.4506926034516057,
      "grad_norm": 0.07636135616082246,
      "learning_rate": 0.00016514204074490329,
      "loss": 0.8507,
      "step": 82630
    },
    {
      "epoch": 1.450868168331607,
      "grad_norm": 0.07229532162430803,
      "learning_rate": 0.00016513355194483305,
      "loss": 0.848,
      "step": 82640
    },
    {
      "epoch": 1.4510437332116084,
      "grad_norm": 0.10631567242050408,
      "learning_rate": 0.00016512506233217836,
      "loss": 0.8419,
      "step": 82650
    },
    {
      "epoch": 1.4512192980916097,
      "grad_norm": 0.0545622194481474,
      "learning_rate": 0.0001651165719070469,
      "loss": 0.8461,
      "step": 82660
    },
    {
      "epoch": 1.4513948629716111,
      "grad_norm": 0.0435817354886127,
      "learning_rate": 0.0001651080806695461,
      "loss": 0.8475,
      "step": 82670
    },
    {
      "epoch": 1.4515704278516126,
      "grad_norm": 0.05249490232378303,
      "learning_rate": 0.0001650995886197837,
      "loss": 0.8462,
      "step": 82680
    },
    {
      "epoch": 1.451745992731614,
      "grad_norm": 0.05195188826090613,
      "learning_rate": 0.0001650910957578672,
      "loss": 0.8448,
      "step": 82690
    },
    {
      "epoch": 1.4519215576116153,
      "grad_norm": 0.06809873419284677,
      "learning_rate": 0.00016508260208390423,
      "loss": 0.8518,
      "step": 82700
    },
    {
      "epoch": 1.4520971224916168,
      "grad_norm": 0.058655186888511575,
      "learning_rate": 0.00016507410759800245,
      "loss": 0.8466,
      "step": 82710
    },
    {
      "epoch": 1.452272687371618,
      "grad_norm": 0.07326411862899015,
      "learning_rate": 0.0001650656123002694,
      "loss": 0.8538,
      "step": 82720
    },
    {
      "epoch": 1.4524482522516196,
      "grad_norm": 0.05380323163386629,
      "learning_rate": 0.00016505711619081278,
      "loss": 0.8532,
      "step": 82730
    },
    {
      "epoch": 1.452623817131621,
      "grad_norm": 0.08144974454351225,
      "learning_rate": 0.00016504861926974022,
      "loss": 0.8559,
      "step": 82740
    },
    {
      "epoch": 1.4527993820116225,
      "grad_norm": 0.047155717114269864,
      "learning_rate": 0.00016504012153715938,
      "loss": 0.8515,
      "step": 82750
    },
    {
      "epoch": 1.4529749468916238,
      "grad_norm": 0.0737277422789445,
      "learning_rate": 0.00016503162299317793,
      "loss": 0.8467,
      "step": 82760
    },
    {
      "epoch": 1.4531505117716252,
      "grad_norm": 0.05078842601942875,
      "learning_rate": 0.00016502312363790353,
      "loss": 0.8471,
      "step": 82770
    },
    {
      "epoch": 1.4533260766516265,
      "grad_norm": 0.05518370818999657,
      "learning_rate": 0.0001650146234714439,
      "loss": 0.8456,
      "step": 82780
    },
    {
      "epoch": 1.453501641531628,
      "grad_norm": 0.06089900147026913,
      "learning_rate": 0.0001650061224939067,
      "loss": 0.857,
      "step": 82790
    },
    {
      "epoch": 1.4536772064116295,
      "grad_norm": 0.07948722631877808,
      "learning_rate": 0.0001649976207053997,
      "loss": 0.8549,
      "step": 82800
    },
    {
      "epoch": 1.453852771291631,
      "grad_norm": 0.049523388940105544,
      "learning_rate": 0.00016498911810603059,
      "loss": 0.8482,
      "step": 82810
    },
    {
      "epoch": 1.4540283361716322,
      "grad_norm": 0.08741808458519698,
      "learning_rate": 0.00016498061469590703,
      "loss": 0.8396,
      "step": 82820
    },
    {
      "epoch": 1.4542039010516337,
      "grad_norm": 0.056388813095355005,
      "learning_rate": 0.00016497211047513687,
      "loss": 0.8462,
      "step": 82830
    },
    {
      "epoch": 1.454379465931635,
      "grad_norm": 0.04713287934280101,
      "learning_rate": 0.00016496360544382782,
      "loss": 0.845,
      "step": 82840
    },
    {
      "epoch": 1.4545550308116364,
      "grad_norm": 0.06428904949149719,
      "learning_rate": 0.00016495509960208756,
      "loss": 0.8456,
      "step": 82850
    },
    {
      "epoch": 1.4547305956916379,
      "grad_norm": 0.053069493654574175,
      "learning_rate": 0.000164946592950024,
      "loss": 0.8465,
      "step": 82860
    },
    {
      "epoch": 1.4549061605716394,
      "grad_norm": 0.05758706400883537,
      "learning_rate": 0.0001649380854877448,
      "loss": 0.8516,
      "step": 82870
    },
    {
      "epoch": 1.4550817254516406,
      "grad_norm": 0.08819547233180827,
      "learning_rate": 0.00016492957721535783,
      "loss": 0.8464,
      "step": 82880
    },
    {
      "epoch": 1.455257290331642,
      "grad_norm": 0.05788796384935592,
      "learning_rate": 0.00016492106813297087,
      "loss": 0.8433,
      "step": 82890
    },
    {
      "epoch": 1.4554328552116433,
      "grad_norm": 0.06512481571733301,
      "learning_rate": 0.0001649125582406917,
      "loss": 0.8545,
      "step": 82900
    },
    {
      "epoch": 1.4556084200916448,
      "grad_norm": 0.07420216935785456,
      "learning_rate": 0.0001649040475386282,
      "loss": 0.8407,
      "step": 82910
    },
    {
      "epoch": 1.4557839849716463,
      "grad_norm": 0.06678653818312046,
      "learning_rate": 0.00016489553602688813,
      "loss": 0.8525,
      "step": 82920
    },
    {
      "epoch": 1.4559595498516478,
      "grad_norm": 0.05470701464128382,
      "learning_rate": 0.00016488702370557942,
      "loss": 0.8531,
      "step": 82930
    },
    {
      "epoch": 1.456135114731649,
      "grad_norm": 0.07499048019279897,
      "learning_rate": 0.00016487851057480986,
      "loss": 0.8586,
      "step": 82940
    },
    {
      "epoch": 1.4563106796116505,
      "grad_norm": 0.06814406292430976,
      "learning_rate": 0.00016486999663468732,
      "loss": 0.8497,
      "step": 82950
    },
    {
      "epoch": 1.4564862444916518,
      "grad_norm": 0.044869614866475845,
      "learning_rate": 0.00016486148188531969,
      "loss": 0.8446,
      "step": 82960
    },
    {
      "epoch": 1.4566618093716532,
      "grad_norm": 0.056202092827752285,
      "learning_rate": 0.00016485296632681487,
      "loss": 0.8447,
      "step": 82970
    },
    {
      "epoch": 1.4568373742516547,
      "grad_norm": 0.059034832560779016,
      "learning_rate": 0.0001648444499592807,
      "loss": 0.8502,
      "step": 82980
    },
    {
      "epoch": 1.4570129391316562,
      "grad_norm": 0.059779592427453405,
      "learning_rate": 0.0001648359327828251,
      "loss": 0.8494,
      "step": 82990
    },
    {
      "epoch": 1.4571885040116574,
      "grad_norm": 0.06370818222399202,
      "learning_rate": 0.00016482741479755602,
      "loss": 0.8525,
      "step": 83000
    },
    {
      "epoch": 1.457364068891659,
      "grad_norm": 0.06487183681948328,
      "learning_rate": 0.00016481889600358134,
      "loss": 0.8522,
      "step": 83010
    },
    {
      "epoch": 1.4575396337716602,
      "grad_norm": 0.04940061312054276,
      "learning_rate": 0.00016481037640100904,
      "loss": 0.8608,
      "step": 83020
    },
    {
      "epoch": 1.4577151986516617,
      "grad_norm": 0.0693936519404394,
      "learning_rate": 0.000164801855989947,
      "loss": 0.854,
      "step": 83030
    },
    {
      "epoch": 1.4578907635316631,
      "grad_norm": 0.05501299495831504,
      "learning_rate": 0.00016479333477050328,
      "loss": 0.8432,
      "step": 83040
    },
    {
      "epoch": 1.4580663284116646,
      "grad_norm": 0.05228364267367111,
      "learning_rate": 0.00016478481274278574,
      "loss": 0.8465,
      "step": 83050
    },
    {
      "epoch": 1.4582418932916659,
      "grad_norm": 0.06473960932345034,
      "learning_rate": 0.0001647762899069024,
      "loss": 0.8513,
      "step": 83060
    },
    {
      "epoch": 1.4584174581716673,
      "grad_norm": 0.07384947836707849,
      "learning_rate": 0.0001647677662629613,
      "loss": 0.8414,
      "step": 83070
    },
    {
      "epoch": 1.4585930230516688,
      "grad_norm": 0.06715855929886091,
      "learning_rate": 0.00016475924181107033,
      "loss": 0.8422,
      "step": 83080
    },
    {
      "epoch": 1.45876858793167,
      "grad_norm": 0.0756824368167195,
      "learning_rate": 0.00016475071655133755,
      "loss": 0.8562,
      "step": 83090
    },
    {
      "epoch": 1.4589441528116716,
      "grad_norm": 0.061943030296957184,
      "learning_rate": 0.00016474219048387098,
      "loss": 0.8417,
      "step": 83100
    },
    {
      "epoch": 1.459119717691673,
      "grad_norm": 0.07601170316511748,
      "learning_rate": 0.00016473366360877863,
      "loss": 0.8438,
      "step": 83110
    },
    {
      "epoch": 1.4592952825716743,
      "grad_norm": 0.05867038056409399,
      "learning_rate": 0.00016472513592616857,
      "loss": 0.851,
      "step": 83120
    },
    {
      "epoch": 1.4594708474516758,
      "grad_norm": 0.07655701596646221,
      "learning_rate": 0.00016471660743614877,
      "loss": 0.8446,
      "step": 83130
    },
    {
      "epoch": 1.4596464123316772,
      "grad_norm": 0.0695438356911637,
      "learning_rate": 0.0001647080781388274,
      "loss": 0.8454,
      "step": 83140
    },
    {
      "epoch": 1.4598219772116785,
      "grad_norm": 0.058979625055680766,
      "learning_rate": 0.00016469954803431243,
      "loss": 0.8379,
      "step": 83150
    },
    {
      "epoch": 1.45999754209168,
      "grad_norm": 0.06308580621163681,
      "learning_rate": 0.00016469101712271203,
      "loss": 0.8465,
      "step": 83160
    },
    {
      "epoch": 1.4601731069716815,
      "grad_norm": 0.07146374327914055,
      "learning_rate": 0.0001646824854041342,
      "loss": 0.8454,
      "step": 83170
    },
    {
      "epoch": 1.4603486718516827,
      "grad_norm": 0.05678535993778328,
      "learning_rate": 0.0001646739528786871,
      "loss": 0.8499,
      "step": 83180
    },
    {
      "epoch": 1.4605242367316842,
      "grad_norm": 0.05258110863697477,
      "learning_rate": 0.00016466541954647876,
      "loss": 0.8541,
      "step": 83190
    },
    {
      "epoch": 1.4606998016116857,
      "grad_norm": 0.05367740198889225,
      "learning_rate": 0.0001646568854076174,
      "loss": 0.854,
      "step": 83200
    },
    {
      "epoch": 1.460875366491687,
      "grad_norm": 0.08222962544632216,
      "learning_rate": 0.0001646483504622111,
      "loss": 0.8505,
      "step": 83210
    },
    {
      "epoch": 1.4610509313716884,
      "grad_norm": 0.056274129687850644,
      "learning_rate": 0.00016463981471036797,
      "loss": 0.8462,
      "step": 83220
    },
    {
      "epoch": 1.4612264962516899,
      "grad_norm": 0.06012879764926438,
      "learning_rate": 0.0001646312781521962,
      "loss": 0.8455,
      "step": 83230
    },
    {
      "epoch": 1.4614020611316914,
      "grad_norm": 0.059232792482478355,
      "learning_rate": 0.00016462274078780402,
      "loss": 0.8485,
      "step": 83240
    },
    {
      "epoch": 1.4615776260116926,
      "grad_norm": 0.06122545405959567,
      "learning_rate": 0.0001646142026172994,
      "loss": 0.8442,
      "step": 83250
    },
    {
      "epoch": 1.461753190891694,
      "grad_norm": 0.0705622437425045,
      "learning_rate": 0.00016460566364079072,
      "loss": 0.8513,
      "step": 83260
    },
    {
      "epoch": 1.4619287557716953,
      "grad_norm": 0.05179115137758072,
      "learning_rate": 0.00016459712385838608,
      "loss": 0.8502,
      "step": 83270
    },
    {
      "epoch": 1.4621043206516968,
      "grad_norm": 0.05473910442649749,
      "learning_rate": 0.0001645885832701937,
      "loss": 0.854,
      "step": 83280
    },
    {
      "epoch": 1.4622798855316983,
      "grad_norm": 0.055243334565992716,
      "learning_rate": 0.00016458004187632177,
      "loss": 0.8448,
      "step": 83290
    },
    {
      "epoch": 1.4624554504116998,
      "grad_norm": 0.09460163290911132,
      "learning_rate": 0.00016457149967687848,
      "loss": 0.8504,
      "step": 83300
    },
    {
      "epoch": 1.462631015291701,
      "grad_norm": 0.0633446499842625,
      "learning_rate": 0.00016456295667197215,
      "loss": 0.8471,
      "step": 83310
    },
    {
      "epoch": 1.4628065801717025,
      "grad_norm": 0.06357857543201693,
      "learning_rate": 0.000164554412861711,
      "loss": 0.845,
      "step": 83320
    },
    {
      "epoch": 1.4629821450517038,
      "grad_norm": 0.07747755809653474,
      "learning_rate": 0.00016454586824620322,
      "loss": 0.8483,
      "step": 83330
    },
    {
      "epoch": 1.4631577099317052,
      "grad_norm": 0.07020983443058701,
      "learning_rate": 0.0001645373228255571,
      "loss": 0.852,
      "step": 83340
    },
    {
      "epoch": 1.4633332748117067,
      "grad_norm": 0.05860229914707375,
      "learning_rate": 0.00016452877659988094,
      "loss": 0.8515,
      "step": 83350
    },
    {
      "epoch": 1.4635088396917082,
      "grad_norm": 0.06203618615569894,
      "learning_rate": 0.00016452022956928302,
      "loss": 0.8469,
      "step": 83360
    },
    {
      "epoch": 1.4636844045717095,
      "grad_norm": 0.06011692296326116,
      "learning_rate": 0.0001645116817338716,
      "loss": 0.8496,
      "step": 83370
    },
    {
      "epoch": 1.463859969451711,
      "grad_norm": 0.06084773890408523,
      "learning_rate": 0.00016450313309375502,
      "loss": 0.8456,
      "step": 83380
    },
    {
      "epoch": 1.4640355343317122,
      "grad_norm": 0.0498935081040404,
      "learning_rate": 0.0001644945836490416,
      "loss": 0.8518,
      "step": 83390
    },
    {
      "epoch": 1.4642110992117137,
      "grad_norm": 0.06991196831058599,
      "learning_rate": 0.00016448603339983955,
      "loss": 0.8595,
      "step": 83400
    },
    {
      "epoch": 1.4643866640917151,
      "grad_norm": 0.06999009353745347,
      "learning_rate": 0.00016447748234625734,
      "loss": 0.8508,
      "step": 83410
    },
    {
      "epoch": 1.4645622289717166,
      "grad_norm": 0.06670348182021099,
      "learning_rate": 0.00016446893048840329,
      "loss": 0.8389,
      "step": 83420
    },
    {
      "epoch": 1.4647377938517179,
      "grad_norm": 0.05261730885533947,
      "learning_rate": 0.00016446037782638566,
      "loss": 0.8468,
      "step": 83430
    },
    {
      "epoch": 1.4649133587317194,
      "grad_norm": 0.10406067017673917,
      "learning_rate": 0.00016445182436031295,
      "loss": 0.8428,
      "step": 83440
    },
    {
      "epoch": 1.4650889236117206,
      "grad_norm": 0.08351982728278491,
      "learning_rate": 0.00016444327009029343,
      "loss": 0.8501,
      "step": 83450
    },
    {
      "epoch": 1.465264488491722,
      "grad_norm": 0.0600436719476642,
      "learning_rate": 0.0001644347150164355,
      "loss": 0.8441,
      "step": 83460
    },
    {
      "epoch": 1.4654400533717236,
      "grad_norm": 0.04789150846823069,
      "learning_rate": 0.0001644261591388476,
      "loss": 0.848,
      "step": 83470
    },
    {
      "epoch": 1.465615618251725,
      "grad_norm": 0.04636475912563549,
      "learning_rate": 0.00016441760245763812,
      "loss": 0.8451,
      "step": 83480
    },
    {
      "epoch": 1.4657911831317263,
      "grad_norm": 0.05750500373571481,
      "learning_rate": 0.00016440904497291544,
      "loss": 0.8528,
      "step": 83490
    },
    {
      "epoch": 1.4659667480117278,
      "grad_norm": 0.05366277473859499,
      "learning_rate": 0.00016440048668478802,
      "loss": 0.8489,
      "step": 83500
    },
    {
      "epoch": 1.466142312891729,
      "grad_norm": 0.0624007484429861,
      "learning_rate": 0.0001643919275933643,
      "loss": 0.8443,
      "step": 83510
    },
    {
      "epoch": 1.4663178777717305,
      "grad_norm": 0.05469634577681651,
      "learning_rate": 0.00016438336769875267,
      "loss": 0.8476,
      "step": 83520
    },
    {
      "epoch": 1.466493442651732,
      "grad_norm": 0.05363039040125383,
      "learning_rate": 0.00016437480700106163,
      "loss": 0.8457,
      "step": 83530
    },
    {
      "epoch": 1.4666690075317335,
      "grad_norm": 0.05134204952727971,
      "learning_rate": 0.00016436624550039965,
      "loss": 0.8475,
      "step": 83540
    },
    {
      "epoch": 1.4668445724117347,
      "grad_norm": 0.08867293772960044,
      "learning_rate": 0.0001643576831968752,
      "loss": 0.8498,
      "step": 83550
    },
    {
      "epoch": 1.4670201372917362,
      "grad_norm": 0.050583944348932146,
      "learning_rate": 0.00016434912009059675,
      "loss": 0.8548,
      "step": 83560
    },
    {
      "epoch": 1.4671957021717374,
      "grad_norm": 0.05833836105327453,
      "learning_rate": 0.0001643405561816728,
      "loss": 0.8536,
      "step": 83570
    },
    {
      "epoch": 1.467371267051739,
      "grad_norm": 0.08186630579267011,
      "learning_rate": 0.0001643319914702119,
      "loss": 0.8474,
      "step": 83580
    },
    {
      "epoch": 1.4675468319317404,
      "grad_norm": 0.07892507111381213,
      "learning_rate": 0.0001643234259563225,
      "loss": 0.8421,
      "step": 83590
    },
    {
      "epoch": 1.4677223968117419,
      "grad_norm": 0.06979043911993592,
      "learning_rate": 0.00016431485964011318,
      "loss": 0.8487,
      "step": 83600
    },
    {
      "epoch": 1.4678979616917431,
      "grad_norm": 0.05480735524762101,
      "learning_rate": 0.00016430629252169244,
      "loss": 0.847,
      "step": 83610
    },
    {
      "epoch": 1.4680735265717446,
      "grad_norm": 0.09894346199036219,
      "learning_rate": 0.00016429772460116882,
      "loss": 0.8524,
      "step": 83620
    },
    {
      "epoch": 1.4682490914517459,
      "grad_norm": 0.056376703250638475,
      "learning_rate": 0.00016428915587865091,
      "loss": 0.8511,
      "step": 83630
    },
    {
      "epoch": 1.4684246563317473,
      "grad_norm": 0.07990472456030612,
      "learning_rate": 0.0001642805863542473,
      "loss": 0.8536,
      "step": 83640
    },
    {
      "epoch": 1.4686002212117488,
      "grad_norm": 0.06913545581843061,
      "learning_rate": 0.0001642720160280665,
      "loss": 0.847,
      "step": 83650
    },
    {
      "epoch": 1.4687757860917503,
      "grad_norm": 0.0579034662810748,
      "learning_rate": 0.00016426344490021714,
      "loss": 0.8586,
      "step": 83660
    },
    {
      "epoch": 1.4689513509717516,
      "grad_norm": 0.09994456265800757,
      "learning_rate": 0.00016425487297080783,
      "loss": 0.8463,
      "step": 83670
    },
    {
      "epoch": 1.469126915851753,
      "grad_norm": 0.07106796846528536,
      "learning_rate": 0.0001642463002399471,
      "loss": 0.8495,
      "step": 83680
    },
    {
      "epoch": 1.4693024807317543,
      "grad_norm": 0.07993415696982449,
      "learning_rate": 0.0001642377267077437,
      "loss": 0.8547,
      "step": 83690
    },
    {
      "epoch": 1.4694780456117558,
      "grad_norm": 0.061182466852569464,
      "learning_rate": 0.00016422915237430613,
      "loss": 0.8485,
      "step": 83700
    },
    {
      "epoch": 1.4696536104917572,
      "grad_norm": 0.05323573383754816,
      "learning_rate": 0.0001642205772397431,
      "loss": 0.8568,
      "step": 83710
    },
    {
      "epoch": 1.4698291753717587,
      "grad_norm": 0.07028395777526043,
      "learning_rate": 0.00016421200130416322,
      "loss": 0.8525,
      "step": 83720
    },
    {
      "epoch": 1.47000474025176,
      "grad_norm": 0.04755331155984356,
      "learning_rate": 0.00016420342456767518,
      "loss": 0.8558,
      "step": 83730
    },
    {
      "epoch": 1.4701803051317615,
      "grad_norm": 0.06387308413249157,
      "learning_rate": 0.00016419484703038766,
      "loss": 0.8487,
      "step": 83740
    },
    {
      "epoch": 1.4703558700117627,
      "grad_norm": 0.06563129726165447,
      "learning_rate": 0.0001641862686924093,
      "loss": 0.8415,
      "step": 83750
    },
    {
      "epoch": 1.4705314348917642,
      "grad_norm": 0.05237930419296945,
      "learning_rate": 0.0001641776895538488,
      "loss": 0.8482,
      "step": 83760
    },
    {
      "epoch": 1.4707069997717657,
      "grad_norm": 0.0489651462929584,
      "learning_rate": 0.00016416910961481487,
      "loss": 0.8493,
      "step": 83770
    },
    {
      "epoch": 1.4708825646517671,
      "grad_norm": 0.046624895007500146,
      "learning_rate": 0.00016416052887541623,
      "loss": 0.8442,
      "step": 83780
    },
    {
      "epoch": 1.4710581295317684,
      "grad_norm": 0.05797369646396325,
      "learning_rate": 0.00016415194733576156,
      "loss": 0.8447,
      "step": 83790
    },
    {
      "epoch": 1.4712336944117699,
      "grad_norm": 0.05295624759712933,
      "learning_rate": 0.00016414336499595965,
      "loss": 0.8472,
      "step": 83800
    },
    {
      "epoch": 1.4714092592917714,
      "grad_norm": 0.08413271679331502,
      "learning_rate": 0.00016413478185611915,
      "loss": 0.8482,
      "step": 83810
    },
    {
      "epoch": 1.4715848241717726,
      "grad_norm": 0.07818568793907567,
      "learning_rate": 0.00016412619791634887,
      "loss": 0.8395,
      "step": 83820
    },
    {
      "epoch": 1.471760389051774,
      "grad_norm": 0.059634263072868406,
      "learning_rate": 0.0001641176131767576,
      "loss": 0.8433,
      "step": 83830
    },
    {
      "epoch": 1.4719359539317756,
      "grad_norm": 0.05589605812794592,
      "learning_rate": 0.00016410902763745405,
      "loss": 0.8559,
      "step": 83840
    },
    {
      "epoch": 1.4721115188117768,
      "grad_norm": 0.061234815576302074,
      "learning_rate": 0.00016410044129854708,
      "loss": 0.8465,
      "step": 83850
    },
    {
      "epoch": 1.4722870836917783,
      "grad_norm": 0.07052593468086003,
      "learning_rate": 0.00016409185416014534,
      "loss": 0.8444,
      "step": 83860
    },
    {
      "epoch": 1.4724626485717798,
      "grad_norm": 0.0522579041376801,
      "learning_rate": 0.00016408326622235778,
      "loss": 0.8549,
      "step": 83870
    },
    {
      "epoch": 1.472638213451781,
      "grad_norm": 0.06310910903587647,
      "learning_rate": 0.00016407467748529313,
      "loss": 0.8511,
      "step": 83880
    },
    {
      "epoch": 1.4728137783317825,
      "grad_norm": 0.0832827627986753,
      "learning_rate": 0.0001640660879490602,
      "loss": 0.8445,
      "step": 83890
    },
    {
      "epoch": 1.472989343211784,
      "grad_norm": 0.0677492593520059,
      "learning_rate": 0.00016405749761376788,
      "loss": 0.8544,
      "step": 83900
    },
    {
      "epoch": 1.4731649080917852,
      "grad_norm": 0.07750607024301977,
      "learning_rate": 0.00016404890647952496,
      "loss": 0.8475,
      "step": 83910
    },
    {
      "epoch": 1.4733404729717867,
      "grad_norm": 0.05787392720430598,
      "learning_rate": 0.00016404031454644036,
      "loss": 0.8527,
      "step": 83920
    },
    {
      "epoch": 1.4735160378517882,
      "grad_norm": 0.053864694781824596,
      "learning_rate": 0.00016403172181462286,
      "loss": 0.8405,
      "step": 83930
    },
    {
      "epoch": 1.4736916027317895,
      "grad_norm": 0.05432627195765173,
      "learning_rate": 0.0001640231282841813,
      "loss": 0.8486,
      "step": 83940
    },
    {
      "epoch": 1.473867167611791,
      "grad_norm": 0.07757548115447277,
      "learning_rate": 0.0001640145339552247,
      "loss": 0.8471,
      "step": 83950
    },
    {
      "epoch": 1.4740427324917924,
      "grad_norm": 0.05999409516123081,
      "learning_rate": 0.00016400593882786189,
      "loss": 0.8434,
      "step": 83960
    },
    {
      "epoch": 1.4742182973717939,
      "grad_norm": 0.08409206617593677,
      "learning_rate": 0.00016399734290220176,
      "loss": 0.8476,
      "step": 83970
    },
    {
      "epoch": 1.4743938622517951,
      "grad_norm": 0.05432163849708818,
      "learning_rate": 0.00016398874617835319,
      "loss": 0.845,
      "step": 83980
    },
    {
      "epoch": 1.4745694271317966,
      "grad_norm": 0.05949145528716372,
      "learning_rate": 0.00016398014865642513,
      "loss": 0.8517,
      "step": 83990
    },
    {
      "epoch": 1.4747449920117979,
      "grad_norm": 0.05583554594522306,
      "learning_rate": 0.00016397155033652652,
      "loss": 0.8527,
      "step": 84000
    },
    {
      "epoch": 1.4749205568917994,
      "grad_norm": 0.06418454773620946,
      "learning_rate": 0.00016396295121876632,
      "loss": 0.8446,
      "step": 84010
    },
    {
      "epoch": 1.4750961217718008,
      "grad_norm": 0.06976343582211729,
      "learning_rate": 0.00016395435130325348,
      "loss": 0.8568,
      "step": 84020
    },
    {
      "epoch": 1.4752716866518023,
      "grad_norm": 0.05664942856251392,
      "learning_rate": 0.00016394575059009686,
      "loss": 0.8449,
      "step": 84030
    },
    {
      "epoch": 1.4754472515318036,
      "grad_norm": 0.054188841288438074,
      "learning_rate": 0.00016393714907940557,
      "loss": 0.845,
      "step": 84040
    },
    {
      "epoch": 1.475622816411805,
      "grad_norm": 0.0809605667133745,
      "learning_rate": 0.00016392854677128857,
      "loss": 0.8454,
      "step": 84050
    },
    {
      "epoch": 1.4757983812918063,
      "grad_norm": 0.0611854658646288,
      "learning_rate": 0.00016391994366585477,
      "loss": 0.8489,
      "step": 84060
    },
    {
      "epoch": 1.4759739461718078,
      "grad_norm": 0.07558577034626296,
      "learning_rate": 0.00016391133976321326,
      "loss": 0.849,
      "step": 84070
    },
    {
      "epoch": 1.4761495110518092,
      "grad_norm": 0.05818857331538572,
      "learning_rate": 0.000163902735063473,
      "loss": 0.8486,
      "step": 84080
    },
    {
      "epoch": 1.4763250759318107,
      "grad_norm": 0.0704530965842298,
      "learning_rate": 0.00016389412956674304,
      "loss": 0.8499,
      "step": 84090
    },
    {
      "epoch": 1.476500640811812,
      "grad_norm": 0.06953434139112732,
      "learning_rate": 0.0001638855232731324,
      "loss": 0.8502,
      "step": 84100
    },
    {
      "epoch": 1.4766762056918135,
      "grad_norm": 0.06513762206350118,
      "learning_rate": 0.00016387691618275014,
      "loss": 0.8445,
      "step": 84110
    },
    {
      "epoch": 1.4768517705718147,
      "grad_norm": 0.06906031369162255,
      "learning_rate": 0.00016386830829570527,
      "loss": 0.8527,
      "step": 84120
    },
    {
      "epoch": 1.4770273354518162,
      "grad_norm": 0.0640613637784772,
      "learning_rate": 0.0001638596996121069,
      "loss": 0.839,
      "step": 84130
    },
    {
      "epoch": 1.4772029003318177,
      "grad_norm": 0.06869763897199568,
      "learning_rate": 0.0001638510901320641,
      "loss": 0.8441,
      "step": 84140
    },
    {
      "epoch": 1.4773784652118191,
      "grad_norm": 0.04911610700239731,
      "learning_rate": 0.00016384247985568596,
      "loss": 0.8398,
      "step": 84150
    },
    {
      "epoch": 1.4775540300918204,
      "grad_norm": 0.04870044223991355,
      "learning_rate": 0.0001638338687830815,
      "loss": 0.8443,
      "step": 84160
    },
    {
      "epoch": 1.4777295949718219,
      "grad_norm": 0.06476066285295365,
      "learning_rate": 0.00016382525691435991,
      "loss": 0.8521,
      "step": 84170
    },
    {
      "epoch": 1.4779051598518231,
      "grad_norm": 0.05703805239549158,
      "learning_rate": 0.0001638166442496303,
      "loss": 0.8503,
      "step": 84180
    },
    {
      "epoch": 1.4780807247318246,
      "grad_norm": 0.04973424567291954,
      "learning_rate": 0.00016380803078900176,
      "loss": 0.8445,
      "step": 84190
    },
    {
      "epoch": 1.478256289611826,
      "grad_norm": 0.04907243412242144,
      "learning_rate": 0.00016379941653258342,
      "loss": 0.8564,
      "step": 84200
    },
    {
      "epoch": 1.4784318544918276,
      "grad_norm": 0.0654915003154788,
      "learning_rate": 0.0001637908014804844,
      "loss": 0.856,
      "step": 84210
    },
    {
      "epoch": 1.4786074193718288,
      "grad_norm": 0.06813303181921693,
      "learning_rate": 0.00016378218563281396,
      "loss": 0.8474,
      "step": 84220
    },
    {
      "epoch": 1.4787829842518303,
      "grad_norm": 0.06940234956850423,
      "learning_rate": 0.00016377356898968114,
      "loss": 0.8457,
      "step": 84230
    },
    {
      "epoch": 1.4789585491318316,
      "grad_norm": 0.05506172187636441,
      "learning_rate": 0.00016376495155119523,
      "loss": 0.851,
      "step": 84240
    },
    {
      "epoch": 1.479134114011833,
      "grad_norm": 0.04765432518736365,
      "learning_rate": 0.0001637563333174653,
      "loss": 0.8479,
      "step": 84250
    },
    {
      "epoch": 1.4793096788918345,
      "grad_norm": 0.05123380366558768,
      "learning_rate": 0.00016374771428860063,
      "loss": 0.8467,
      "step": 84260
    },
    {
      "epoch": 1.479485243771836,
      "grad_norm": 0.061235502163980896,
      "learning_rate": 0.00016373909446471037,
      "loss": 0.8531,
      "step": 84270
    },
    {
      "epoch": 1.4796608086518372,
      "grad_norm": 0.07220142004407613,
      "learning_rate": 0.0001637304738459038,
      "loss": 0.8458,
      "step": 84280
    },
    {
      "epoch": 1.4798363735318387,
      "grad_norm": 0.054361561597827505,
      "learning_rate": 0.0001637218524322901,
      "loss": 0.8625,
      "step": 84290
    },
    {
      "epoch": 1.48001193841184,
      "grad_norm": 0.07592914107142865,
      "learning_rate": 0.00016371323022397845,
      "loss": 0.8459,
      "step": 84300
    },
    {
      "epoch": 1.4801875032918415,
      "grad_norm": 0.0774039235809577,
      "learning_rate": 0.0001637046072210782,
      "loss": 0.8469,
      "step": 84310
    },
    {
      "epoch": 1.480363068171843,
      "grad_norm": 0.0779929985513404,
      "learning_rate": 0.00016369598342369856,
      "loss": 0.8433,
      "step": 84320
    },
    {
      "epoch": 1.4805386330518444,
      "grad_norm": 0.06173047184112846,
      "learning_rate": 0.00016368735883194875,
      "loss": 0.8537,
      "step": 84330
    },
    {
      "epoch": 1.4807141979318457,
      "grad_norm": 0.059739375694670416,
      "learning_rate": 0.00016367873344593813,
      "loss": 0.8565,
      "step": 84340
    },
    {
      "epoch": 1.4808897628118471,
      "grad_norm": 0.056878831431495074,
      "learning_rate": 0.00016367010726577594,
      "loss": 0.8537,
      "step": 84350
    },
    {
      "epoch": 1.4810653276918484,
      "grad_norm": 0.06957267565009297,
      "learning_rate": 0.0001636614802915715,
      "loss": 0.839,
      "step": 84360
    },
    {
      "epoch": 1.4812408925718499,
      "grad_norm": 0.05472944689439533,
      "learning_rate": 0.0001636528525234341,
      "loss": 0.8511,
      "step": 84370
    },
    {
      "epoch": 1.4814164574518514,
      "grad_norm": 0.068115991335367,
      "learning_rate": 0.000163644223961473,
      "loss": 0.848,
      "step": 84380
    },
    {
      "epoch": 1.4815920223318528,
      "grad_norm": 0.0830855554778225,
      "learning_rate": 0.0001636355946057976,
      "loss": 0.8452,
      "step": 84390
    },
    {
      "epoch": 1.481767587211854,
      "grad_norm": 0.07414770887781472,
      "learning_rate": 0.00016362696445651722,
      "loss": 0.8539,
      "step": 84400
    },
    {
      "epoch": 1.4819431520918556,
      "grad_norm": 0.07106414004090908,
      "learning_rate": 0.00016361833351374118,
      "loss": 0.841,
      "step": 84410
    },
    {
      "epoch": 1.4821187169718568,
      "grad_norm": 0.046989133966727666,
      "learning_rate": 0.00016360970177757886,
      "loss": 0.8501,
      "step": 84420
    },
    {
      "epoch": 1.4822942818518583,
      "grad_norm": 0.09925399749235998,
      "learning_rate": 0.00016360106924813965,
      "loss": 0.8456,
      "step": 84430
    },
    {
      "epoch": 1.4824698467318598,
      "grad_norm": 0.05512724387462179,
      "learning_rate": 0.0001635924359255329,
      "loss": 0.8495,
      "step": 84440
    },
    {
      "epoch": 1.4826454116118613,
      "grad_norm": 0.056717313275357804,
      "learning_rate": 0.0001635838018098679,
      "loss": 0.8394,
      "step": 84450
    },
    {
      "epoch": 1.4828209764918625,
      "grad_norm": 0.06828312676328038,
      "learning_rate": 0.00016357516690125422,
      "loss": 0.8453,
      "step": 84460
    },
    {
      "epoch": 1.482996541371864,
      "grad_norm": 0.05449147759917187,
      "learning_rate": 0.00016356653119980116,
      "loss": 0.8439,
      "step": 84470
    },
    {
      "epoch": 1.4831721062518652,
      "grad_norm": 0.06374790294417038,
      "learning_rate": 0.00016355789470561818,
      "loss": 0.8482,
      "step": 84480
    },
    {
      "epoch": 1.4833476711318667,
      "grad_norm": 0.06231291889742601,
      "learning_rate": 0.00016354925741881465,
      "loss": 0.8474,
      "step": 84490
    },
    {
      "epoch": 1.4835232360118682,
      "grad_norm": 0.05718750085956236,
      "learning_rate": 0.00016354061933950004,
      "loss": 0.8464,
      "step": 84500
    },
    {
      "epoch": 1.4836988008918697,
      "grad_norm": 0.056667141868149105,
      "learning_rate": 0.00016353198046778383,
      "loss": 0.862,
      "step": 84510
    },
    {
      "epoch": 1.483874365771871,
      "grad_norm": 0.05767194199882116,
      "learning_rate": 0.00016352334080377542,
      "loss": 0.8506,
      "step": 84520
    },
    {
      "epoch": 1.4840499306518724,
      "grad_norm": 0.06741602243429867,
      "learning_rate": 0.0001635147003475843,
      "loss": 0.8533,
      "step": 84530
    },
    {
      "epoch": 1.4842254955318739,
      "grad_norm": 0.0719872704396953,
      "learning_rate": 0.00016350605909931995,
      "loss": 0.8474,
      "step": 84540
    },
    {
      "epoch": 1.4844010604118751,
      "grad_norm": 0.059957767920791064,
      "learning_rate": 0.00016349741705909188,
      "loss": 0.8474,
      "step": 84550
    },
    {
      "epoch": 1.4845766252918766,
      "grad_norm": 0.05238885083326305,
      "learning_rate": 0.00016348877422700953,
      "loss": 0.8471,
      "step": 84560
    },
    {
      "epoch": 1.484752190171878,
      "grad_norm": 0.052659975055194784,
      "learning_rate": 0.0001634801306031824,
      "loss": 0.8466,
      "step": 84570
    },
    {
      "epoch": 1.4849277550518794,
      "grad_norm": 0.06367777247355702,
      "learning_rate": 0.0001634714861877201,
      "loss": 0.8494,
      "step": 84580
    },
    {
      "epoch": 1.4851033199318808,
      "grad_norm": 0.06797520074457872,
      "learning_rate": 0.0001634628409807321,
      "loss": 0.8495,
      "step": 84590
    },
    {
      "epoch": 1.4852788848118823,
      "grad_norm": 0.06489518239529428,
      "learning_rate": 0.0001634541949823279,
      "loss": 0.8539,
      "step": 84600
    },
    {
      "epoch": 1.4854544496918836,
      "grad_norm": 0.059280217142603016,
      "learning_rate": 0.00016344554819261708,
      "loss": 0.8478,
      "step": 84610
    },
    {
      "epoch": 1.485630014571885,
      "grad_norm": 0.054951523024527306,
      "learning_rate": 0.0001634369006117092,
      "loss": 0.8494,
      "step": 84620
    },
    {
      "epoch": 1.4858055794518865,
      "grad_norm": 0.0509198352793718,
      "learning_rate": 0.00016342825223971384,
      "loss": 0.8461,
      "step": 84630
    },
    {
      "epoch": 1.4859811443318878,
      "grad_norm": 0.08097559214355181,
      "learning_rate": 0.00016341960307674055,
      "loss": 0.853,
      "step": 84640
    },
    {
      "epoch": 1.4861567092118892,
      "grad_norm": 0.09073657388524083,
      "learning_rate": 0.00016341095312289895,
      "loss": 0.8465,
      "step": 84650
    },
    {
      "epoch": 1.4863322740918907,
      "grad_norm": 0.05597060869306052,
      "learning_rate": 0.00016340230237829855,
      "loss": 0.8441,
      "step": 84660
    },
    {
      "epoch": 1.486507838971892,
      "grad_norm": 0.06405056900285544,
      "learning_rate": 0.00016339365084304906,
      "loss": 0.8489,
      "step": 84670
    },
    {
      "epoch": 1.4866834038518935,
      "grad_norm": 0.07507183327286537,
      "learning_rate": 0.00016338499851726007,
      "loss": 0.8515,
      "step": 84680
    },
    {
      "epoch": 1.486858968731895,
      "grad_norm": 0.06186822543453404,
      "learning_rate": 0.0001633763454010412,
      "loss": 0.8492,
      "step": 84690
    },
    {
      "epoch": 1.4870345336118964,
      "grad_norm": 0.06728424360050769,
      "learning_rate": 0.00016336769149450205,
      "loss": 0.8516,
      "step": 84700
    },
    {
      "epoch": 1.4872100984918977,
      "grad_norm": 0.06313058988266618,
      "learning_rate": 0.00016335903679775228,
      "loss": 0.8546,
      "step": 84710
    },
    {
      "epoch": 1.4873856633718991,
      "grad_norm": 0.07012106372395695,
      "learning_rate": 0.0001633503813109016,
      "loss": 0.8389,
      "step": 84720
    },
    {
      "epoch": 1.4875612282519004,
      "grad_norm": 0.04965338273055245,
      "learning_rate": 0.00016334172503405963,
      "loss": 0.8503,
      "step": 84730
    },
    {
      "epoch": 1.4877367931319019,
      "grad_norm": 0.06594092120578991,
      "learning_rate": 0.00016333306796733604,
      "loss": 0.8502,
      "step": 84740
    },
    {
      "epoch": 1.4879123580119034,
      "grad_norm": 0.07733255419851398,
      "learning_rate": 0.00016332441011084053,
      "loss": 0.8471,
      "step": 84750
    },
    {
      "epoch": 1.4880879228919048,
      "grad_norm": 0.07144134479436538,
      "learning_rate": 0.00016331575146468283,
      "loss": 0.8485,
      "step": 84760
    },
    {
      "epoch": 1.488263487771906,
      "grad_norm": 0.061377113130857266,
      "learning_rate": 0.0001633070920289726,
      "loss": 0.8477,
      "step": 84770
    },
    {
      "epoch": 1.4884390526519076,
      "grad_norm": 0.058492717189140495,
      "learning_rate": 0.0001632984318038196,
      "loss": 0.8426,
      "step": 84780
    },
    {
      "epoch": 1.4886146175319088,
      "grad_norm": 0.06249317094658534,
      "learning_rate": 0.00016328977078933348,
      "loss": 0.8537,
      "step": 84790
    },
    {
      "epoch": 1.4887901824119103,
      "grad_norm": 0.08350555665413405,
      "learning_rate": 0.00016328110898562403,
      "loss": 0.846,
      "step": 84800
    },
    {
      "epoch": 1.4889657472919118,
      "grad_norm": 0.06239562239804185,
      "learning_rate": 0.000163272446392801,
      "loss": 0.8494,
      "step": 84810
    },
    {
      "epoch": 1.4891413121719133,
      "grad_norm": 0.07620392853966715,
      "learning_rate": 0.00016326378301097417,
      "loss": 0.8525,
      "step": 84820
    },
    {
      "epoch": 1.4893168770519145,
      "grad_norm": 0.10732709929486454,
      "learning_rate": 0.00016325511884025327,
      "loss": 0.8441,
      "step": 84830
    },
    {
      "epoch": 1.489492441931916,
      "grad_norm": 0.06668229758336931,
      "learning_rate": 0.00016324645388074804,
      "loss": 0.8427,
      "step": 84840
    },
    {
      "epoch": 1.4896680068119172,
      "grad_norm": 0.049112831668743404,
      "learning_rate": 0.00016323778813256835,
      "loss": 0.8565,
      "step": 84850
    },
    {
      "epoch": 1.4898435716919187,
      "grad_norm": 0.06384265954601573,
      "learning_rate": 0.00016322912159582396,
      "loss": 0.8481,
      "step": 84860
    },
    {
      "epoch": 1.4900191365719202,
      "grad_norm": 0.06779440427829297,
      "learning_rate": 0.0001632204542706247,
      "loss": 0.8531,
      "step": 84870
    },
    {
      "epoch": 1.4901947014519217,
      "grad_norm": 0.06094112196354802,
      "learning_rate": 0.0001632117861570803,
      "loss": 0.8518,
      "step": 84880
    },
    {
      "epoch": 1.490370266331923,
      "grad_norm": 0.06988054835496674,
      "learning_rate": 0.00016320311725530063,
      "loss": 0.8492,
      "step": 84890
    },
    {
      "epoch": 1.4905458312119244,
      "grad_norm": 0.050223239474612005,
      "learning_rate": 0.0001631944475653956,
      "loss": 0.8597,
      "step": 84900
    },
    {
      "epoch": 1.4907213960919257,
      "grad_norm": 0.05936625026763667,
      "learning_rate": 0.00016318577708747495,
      "loss": 0.8474,
      "step": 84910
    },
    {
      "epoch": 1.4908969609719271,
      "grad_norm": 0.0718880486520171,
      "learning_rate": 0.00016317710582164864,
      "loss": 0.8583,
      "step": 84920
    },
    {
      "epoch": 1.4910725258519286,
      "grad_norm": 0.05436058980299165,
      "learning_rate": 0.00016316843376802643,
      "loss": 0.8502,
      "step": 84930
    },
    {
      "epoch": 1.49124809073193,
      "grad_norm": 0.04439330237700216,
      "learning_rate": 0.00016315976092671832,
      "loss": 0.8492,
      "step": 84940
    },
    {
      "epoch": 1.4914236556119314,
      "grad_norm": 0.06551722637677,
      "learning_rate": 0.00016315108729783408,
      "loss": 0.8541,
      "step": 84950
    },
    {
      "epoch": 1.4915992204919328,
      "grad_norm": 0.07391160409925793,
      "learning_rate": 0.00016314241288148366,
      "loss": 0.8503,
      "step": 84960
    },
    {
      "epoch": 1.491774785371934,
      "grad_norm": 0.051161447234192026,
      "learning_rate": 0.00016313373767777695,
      "loss": 0.849,
      "step": 84970
    },
    {
      "epoch": 1.4919503502519356,
      "grad_norm": 0.048684300740799435,
      "learning_rate": 0.00016312506168682386,
      "loss": 0.8508,
      "step": 84980
    },
    {
      "epoch": 1.492125915131937,
      "grad_norm": 0.051970921177277994,
      "learning_rate": 0.00016311638490873438,
      "loss": 0.8459,
      "step": 84990
    },
    {
      "epoch": 1.4923014800119385,
      "grad_norm": 0.06270977106599583,
      "learning_rate": 0.00016310770734361837,
      "loss": 0.8394,
      "step": 85000
    },
    {
      "epoch": 1.4924770448919398,
      "grad_norm": 0.0446748467797246,
      "learning_rate": 0.0001630990289915858,
      "loss": 0.8476,
      "step": 85010
    },
    {
      "epoch": 1.4926526097719413,
      "grad_norm": 0.059960838755978234,
      "learning_rate": 0.00016309034985274667,
      "loss": 0.8416,
      "step": 85020
    },
    {
      "epoch": 1.4928281746519425,
      "grad_norm": 0.0576482748976259,
      "learning_rate": 0.00016308166992721088,
      "loss": 0.8465,
      "step": 85030
    },
    {
      "epoch": 1.493003739531944,
      "grad_norm": 0.06330609572854133,
      "learning_rate": 0.00016307298921508842,
      "loss": 0.8537,
      "step": 85040
    },
    {
      "epoch": 1.4931793044119455,
      "grad_norm": 0.058125203207492544,
      "learning_rate": 0.00016306430771648932,
      "loss": 0.8512,
      "step": 85050
    },
    {
      "epoch": 1.493354869291947,
      "grad_norm": 0.056121056513863066,
      "learning_rate": 0.0001630556254315235,
      "loss": 0.8488,
      "step": 85060
    },
    {
      "epoch": 1.4935304341719482,
      "grad_norm": 0.07989557970538212,
      "learning_rate": 0.00016304694236030104,
      "loss": 0.8519,
      "step": 85070
    },
    {
      "epoch": 1.4937059990519497,
      "grad_norm": 0.05415180253555022,
      "learning_rate": 0.00016303825850293193,
      "loss": 0.848,
      "step": 85080
    },
    {
      "epoch": 1.493881563931951,
      "grad_norm": 0.05996008738061254,
      "learning_rate": 0.0001630295738595262,
      "loss": 0.8514,
      "step": 85090
    },
    {
      "epoch": 1.4940571288119524,
      "grad_norm": 0.056911680979165186,
      "learning_rate": 0.00016302088843019388,
      "loss": 0.8481,
      "step": 85100
    },
    {
      "epoch": 1.4942326936919539,
      "grad_norm": 0.04945714197517774,
      "learning_rate": 0.00016301220221504503,
      "loss": 0.8485,
      "step": 85110
    },
    {
      "epoch": 1.4944082585719554,
      "grad_norm": 0.04714749303159526,
      "learning_rate": 0.0001630035152141897,
      "loss": 0.8516,
      "step": 85120
    },
    {
      "epoch": 1.4945838234519566,
      "grad_norm": 0.052296920888769305,
      "learning_rate": 0.00016299482742773786,
      "loss": 0.8538,
      "step": 85130
    },
    {
      "epoch": 1.494759388331958,
      "grad_norm": 0.05999899599214665,
      "learning_rate": 0.0001629861388557998,
      "loss": 0.8538,
      "step": 85140
    },
    {
      "epoch": 1.4949349532119593,
      "grad_norm": 0.06745093610033026,
      "learning_rate": 0.00016297744949848538,
      "loss": 0.8539,
      "step": 85150
    },
    {
      "epoch": 1.4951105180919608,
      "grad_norm": 0.048370962664320345,
      "learning_rate": 0.00016296875935590489,
      "loss": 0.8484,
      "step": 85160
    },
    {
      "epoch": 1.4952860829719623,
      "grad_norm": 0.06551952167915291,
      "learning_rate": 0.00016296006842816824,
      "loss": 0.8452,
      "step": 85170
    },
    {
      "epoch": 1.4954616478519638,
      "grad_norm": 0.07825809564141771,
      "learning_rate": 0.00016295137671538568,
      "loss": 0.8478,
      "step": 85180
    },
    {
      "epoch": 1.495637212731965,
      "grad_norm": 0.0695792884829635,
      "learning_rate": 0.00016294268421766737,
      "loss": 0.8501,
      "step": 85190
    },
    {
      "epoch": 1.4958127776119665,
      "grad_norm": 0.048016825503609646,
      "learning_rate": 0.00016293399093512333,
      "loss": 0.8514,
      "step": 85200
    },
    {
      "epoch": 1.4959883424919678,
      "grad_norm": 0.05875486305176089,
      "learning_rate": 0.00016292529686786378,
      "loss": 0.8484,
      "step": 85210
    },
    {
      "epoch": 1.4961639073719692,
      "grad_norm": 0.09267098532789439,
      "learning_rate": 0.0001629166020159988,
      "loss": 0.8477,
      "step": 85220
    },
    {
      "epoch": 1.4963394722519707,
      "grad_norm": 0.06733592745807193,
      "learning_rate": 0.00016290790637963864,
      "loss": 0.8546,
      "step": 85230
    },
    {
      "epoch": 1.4965150371319722,
      "grad_norm": 0.04724825306223196,
      "learning_rate": 0.00016289920995889344,
      "loss": 0.8445,
      "step": 85240
    },
    {
      "epoch": 1.4966906020119735,
      "grad_norm": 0.08620257709659636,
      "learning_rate": 0.00016289051275387339,
      "loss": 0.8515,
      "step": 85250
    },
    {
      "epoch": 1.496866166891975,
      "grad_norm": 0.0554718147362301,
      "learning_rate": 0.0001628818147646887,
      "loss": 0.8378,
      "step": 85260
    },
    {
      "epoch": 1.4970417317719764,
      "grad_norm": 0.07694900641142244,
      "learning_rate": 0.00016287311599144953,
      "loss": 0.8486,
      "step": 85270
    },
    {
      "epoch": 1.4972172966519777,
      "grad_norm": 0.05007222178657658,
      "learning_rate": 0.00016286441643426614,
      "loss": 0.8437,
      "step": 85280
    },
    {
      "epoch": 1.4973928615319791,
      "grad_norm": 0.07378223069001816,
      "learning_rate": 0.00016285571609324872,
      "loss": 0.8514,
      "step": 85290
    },
    {
      "epoch": 1.4975684264119806,
      "grad_norm": 0.05704071768024681,
      "learning_rate": 0.0001628470149685075,
      "loss": 0.849,
      "step": 85300
    },
    {
      "epoch": 1.4977439912919819,
      "grad_norm": 0.046743409754910474,
      "learning_rate": 0.0001628383130601528,
      "loss": 0.8459,
      "step": 85310
    },
    {
      "epoch": 1.4979195561719834,
      "grad_norm": 0.06704876370401013,
      "learning_rate": 0.0001628296103682948,
      "loss": 0.8519,
      "step": 85320
    },
    {
      "epoch": 1.4980951210519848,
      "grad_norm": 0.044754194361982656,
      "learning_rate": 0.00016282090689304378,
      "loss": 0.8463,
      "step": 85330
    },
    {
      "epoch": 1.498270685931986,
      "grad_norm": 0.05755186379176119,
      "learning_rate": 0.00016281220263451003,
      "loss": 0.8381,
      "step": 85340
    },
    {
      "epoch": 1.4984462508119876,
      "grad_norm": 0.08271502509191732,
      "learning_rate": 0.00016280349759280381,
      "loss": 0.8431,
      "step": 85350
    },
    {
      "epoch": 1.498621815691989,
      "grad_norm": 0.07955894807724696,
      "learning_rate": 0.00016279479176803545,
      "loss": 0.8499,
      "step": 85360
    },
    {
      "epoch": 1.4987973805719903,
      "grad_norm": 0.05073220190457187,
      "learning_rate": 0.00016278608516031524,
      "loss": 0.8396,
      "step": 85370
    },
    {
      "epoch": 1.4989729454519918,
      "grad_norm": 0.055461504334839436,
      "learning_rate": 0.0001627773777697535,
      "loss": 0.8519,
      "step": 85380
    },
    {
      "epoch": 1.4991485103319933,
      "grad_norm": 0.051768127213665756,
      "learning_rate": 0.00016276866959646048,
      "loss": 0.8521,
      "step": 85390
    },
    {
      "epoch": 1.4993240752119945,
      "grad_norm": 0.06294595064398639,
      "learning_rate": 0.00016275996064054663,
      "loss": 0.8415,
      "step": 85400
    },
    {
      "epoch": 1.499499640091996,
      "grad_norm": 0.05901077406401167,
      "learning_rate": 0.00016275125090212225,
      "loss": 0.851,
      "step": 85410
    },
    {
      "epoch": 1.4996752049719975,
      "grad_norm": 0.06391734006111374,
      "learning_rate": 0.00016274254038129766,
      "loss": 0.8512,
      "step": 85420
    },
    {
      "epoch": 1.499850769851999,
      "grad_norm": 0.05510738750434037,
      "learning_rate": 0.00016273382907818325,
      "loss": 0.8502,
      "step": 85430
    },
    {
      "epoch": 1.5000263347320002,
      "grad_norm": 0.050577122247998936,
      "learning_rate": 0.00016272511699288943,
      "loss": 0.8502,
      "step": 85440
    },
    {
      "epoch": 1.5002018996120015,
      "grad_norm": 0.0471238099165864,
      "learning_rate": 0.00016271640412552653,
      "loss": 0.8498,
      "step": 85450
    },
    {
      "epoch": 1.500377464492003,
      "grad_norm": 0.06307486157706155,
      "learning_rate": 0.00016270769047620497,
      "loss": 0.8532,
      "step": 85460
    },
    {
      "epoch": 1.5005530293720044,
      "grad_norm": 0.0635443781931037,
      "learning_rate": 0.0001626989760450351,
      "loss": 0.848,
      "step": 85470
    },
    {
      "epoch": 1.5007285942520059,
      "grad_norm": 0.09380020895070419,
      "learning_rate": 0.0001626902608321274,
      "loss": 0.846,
      "step": 85480
    },
    {
      "epoch": 1.5009041591320074,
      "grad_norm": 0.0647225702036555,
      "learning_rate": 0.00016268154483759228,
      "loss": 0.849,
      "step": 85490
    },
    {
      "epoch": 1.5010797240120086,
      "grad_norm": 0.0682100856311433,
      "learning_rate": 0.0001626728280615402,
      "loss": 0.8537,
      "step": 85500
    },
    {
      "epoch": 1.5012552888920099,
      "grad_norm": 0.052472152480540414,
      "learning_rate": 0.0001626641105040815,
      "loss": 0.8517,
      "step": 85510
    },
    {
      "epoch": 1.5014308537720114,
      "grad_norm": 0.06232311727949744,
      "learning_rate": 0.00016265539216532674,
      "loss": 0.8485,
      "step": 85520
    },
    {
      "epoch": 1.5016064186520128,
      "grad_norm": 0.061603883928408106,
      "learning_rate": 0.00016264667304538637,
      "loss": 0.8445,
      "step": 85530
    },
    {
      "epoch": 1.5017819835320143,
      "grad_norm": 0.05745857855712221,
      "learning_rate": 0.0001626379531443708,
      "loss": 0.8476,
      "step": 85540
    },
    {
      "epoch": 1.5019575484120158,
      "grad_norm": 0.079129672779203,
      "learning_rate": 0.00016262923246239058,
      "loss": 0.8466,
      "step": 85550
    },
    {
      "epoch": 1.502133113292017,
      "grad_norm": 0.0501990732208953,
      "learning_rate": 0.00016262051099955616,
      "loss": 0.8465,
      "step": 85560
    },
    {
      "epoch": 1.5023086781720185,
      "grad_norm": 0.07440203587616415,
      "learning_rate": 0.00016261178875597804,
      "loss": 0.8441,
      "step": 85570
    },
    {
      "epoch": 1.5024842430520198,
      "grad_norm": 0.07068421870261067,
      "learning_rate": 0.00016260306573176678,
      "loss": 0.8483,
      "step": 85580
    },
    {
      "epoch": 1.5026598079320213,
      "grad_norm": 0.06749126607644819,
      "learning_rate": 0.0001625943419270328,
      "loss": 0.8481,
      "step": 85590
    },
    {
      "epoch": 1.5028353728120227,
      "grad_norm": 0.05967033929803527,
      "learning_rate": 0.0001625856173418868,
      "loss": 0.8493,
      "step": 85600
    },
    {
      "epoch": 1.5030109376920242,
      "grad_norm": 0.05058458007127856,
      "learning_rate": 0.00016257689197643916,
      "loss": 0.8459,
      "step": 85610
    },
    {
      "epoch": 1.5031865025720255,
      "grad_norm": 0.0523622548065938,
      "learning_rate": 0.00016256816583080054,
      "loss": 0.8401,
      "step": 85620
    },
    {
      "epoch": 1.503362067452027,
      "grad_norm": 0.07579631732221498,
      "learning_rate": 0.00016255943890508144,
      "loss": 0.849,
      "step": 85630
    },
    {
      "epoch": 1.5035376323320282,
      "grad_norm": 0.06118679451780421,
      "learning_rate": 0.00016255071119939242,
      "loss": 0.848,
      "step": 85640
    },
    {
      "epoch": 1.5037131972120297,
      "grad_norm": 0.06358814472091377,
      "learning_rate": 0.00016254198271384413,
      "loss": 0.8462,
      "step": 85650
    },
    {
      "epoch": 1.5038887620920312,
      "grad_norm": 0.05235680101125393,
      "learning_rate": 0.0001625332534485471,
      "loss": 0.8483,
      "step": 85660
    },
    {
      "epoch": 1.5040643269720326,
      "grad_norm": 0.051116107672444254,
      "learning_rate": 0.000162524523403612,
      "loss": 0.8493,
      "step": 85670
    },
    {
      "epoch": 1.5042398918520339,
      "grad_norm": 0.05706348774239943,
      "learning_rate": 0.00016251579257914935,
      "loss": 0.8504,
      "step": 85680
    },
    {
      "epoch": 1.5044154567320354,
      "grad_norm": 0.06283099113551988,
      "learning_rate": 0.00016250706097526983,
      "loss": 0.8468,
      "step": 85690
    },
    {
      "epoch": 1.5045910216120366,
      "grad_norm": 0.06785363780458141,
      "learning_rate": 0.00016249832859208409,
      "loss": 0.8433,
      "step": 85700
    },
    {
      "epoch": 1.504766586492038,
      "grad_norm": 0.05313805933061661,
      "learning_rate": 0.0001624895954297027,
      "loss": 0.8543,
      "step": 85710
    },
    {
      "epoch": 1.5049421513720396,
      "grad_norm": 0.07005367376119512,
      "learning_rate": 0.0001624808614882364,
      "loss": 0.8519,
      "step": 85720
    },
    {
      "epoch": 1.505117716252041,
      "grad_norm": 0.06856601024398878,
      "learning_rate": 0.00016247212676779576,
      "loss": 0.8459,
      "step": 85730
    },
    {
      "epoch": 1.5052932811320423,
      "grad_norm": 0.07788595400484398,
      "learning_rate": 0.00016246339126849152,
      "loss": 0.8515,
      "step": 85740
    },
    {
      "epoch": 1.5054688460120438,
      "grad_norm": 0.06295773281039757,
      "learning_rate": 0.0001624546549904343,
      "loss": 0.8484,
      "step": 85750
    },
    {
      "epoch": 1.505644410892045,
      "grad_norm": 0.05527858201053732,
      "learning_rate": 0.00016244591793373482,
      "loss": 0.8466,
      "step": 85760
    },
    {
      "epoch": 1.5058199757720465,
      "grad_norm": 0.060878254649846024,
      "learning_rate": 0.00016243718009850385,
      "loss": 0.8487,
      "step": 85770
    },
    {
      "epoch": 1.505995540652048,
      "grad_norm": 0.07290879917452907,
      "learning_rate": 0.00016242844148485196,
      "loss": 0.8526,
      "step": 85780
    },
    {
      "epoch": 1.5061711055320495,
      "grad_norm": 0.05875182632784148,
      "learning_rate": 0.00016241970209288997,
      "loss": 0.8595,
      "step": 85790
    },
    {
      "epoch": 1.5063466704120507,
      "grad_norm": 0.06309584002727202,
      "learning_rate": 0.00016241096192272863,
      "loss": 0.8475,
      "step": 85800
    },
    {
      "epoch": 1.5065222352920522,
      "grad_norm": 0.0548415701823207,
      "learning_rate": 0.00016240222097447854,
      "loss": 0.8532,
      "step": 85810
    },
    {
      "epoch": 1.5066978001720535,
      "grad_norm": 0.056276370027825796,
      "learning_rate": 0.0001623934792482506,
      "loss": 0.8416,
      "step": 85820
    },
    {
      "epoch": 1.506873365052055,
      "grad_norm": 0.04577840504585291,
      "learning_rate": 0.00016238473674415552,
      "loss": 0.8614,
      "step": 85830
    },
    {
      "epoch": 1.5070489299320564,
      "grad_norm": 0.06231267807697012,
      "learning_rate": 0.00016237599346230404,
      "loss": 0.8556,
      "step": 85840
    },
    {
      "epoch": 1.507224494812058,
      "grad_norm": 0.048476374692895806,
      "learning_rate": 0.00016236724940280696,
      "loss": 0.8488,
      "step": 85850
    },
    {
      "epoch": 1.5074000596920591,
      "grad_norm": 0.059678883040279615,
      "learning_rate": 0.00016235850456577507,
      "loss": 0.847,
      "step": 85860
    },
    {
      "epoch": 1.5075756245720606,
      "grad_norm": 0.06944737800292897,
      "learning_rate": 0.0001623497589513192,
      "loss": 0.8587,
      "step": 85870
    },
    {
      "epoch": 1.5077511894520619,
      "grad_norm": 0.08266394824651445,
      "learning_rate": 0.0001623410125595501,
      "loss": 0.8474,
      "step": 85880
    },
    {
      "epoch": 1.5079267543320634,
      "grad_norm": 0.04732012313363443,
      "learning_rate": 0.0001623322653905786,
      "loss": 0.8508,
      "step": 85890
    },
    {
      "epoch": 1.5081023192120648,
      "grad_norm": 0.058003046854311645,
      "learning_rate": 0.00016232351744451554,
      "loss": 0.8496,
      "step": 85900
    },
    {
      "epoch": 1.5082778840920663,
      "grad_norm": 0.06727420650216087,
      "learning_rate": 0.00016231476872147183,
      "loss": 0.8465,
      "step": 85910
    },
    {
      "epoch": 1.5084534489720678,
      "grad_norm": 0.048054441113197455,
      "learning_rate": 0.00016230601922155816,
      "loss": 0.847,
      "step": 85920
    },
    {
      "epoch": 1.508629013852069,
      "grad_norm": 0.05688285350103343,
      "learning_rate": 0.0001622972689448855,
      "loss": 0.8478,
      "step": 85930
    },
    {
      "epoch": 1.5088045787320703,
      "grad_norm": 0.048554361023743664,
      "learning_rate": 0.00016228851789156472,
      "loss": 0.8463,
      "step": 85940
    },
    {
      "epoch": 1.5089801436120718,
      "grad_norm": 0.05489045583231937,
      "learning_rate": 0.00016227976606170667,
      "loss": 0.8473,
      "step": 85950
    },
    {
      "epoch": 1.5091557084920733,
      "grad_norm": 0.05296501137175214,
      "learning_rate": 0.00016227101345542223,
      "loss": 0.8544,
      "step": 85960
    },
    {
      "epoch": 1.5093312733720747,
      "grad_norm": 0.055868744061980405,
      "learning_rate": 0.0001622622600728223,
      "loss": 0.847,
      "step": 85970
    },
    {
      "epoch": 1.5095068382520762,
      "grad_norm": 0.0805085521443216,
      "learning_rate": 0.0001622535059140178,
      "loss": 0.8497,
      "step": 85980
    },
    {
      "epoch": 1.5096824031320775,
      "grad_norm": 0.06327830308869412,
      "learning_rate": 0.00016224475097911966,
      "loss": 0.8441,
      "step": 85990
    },
    {
      "epoch": 1.5098579680120787,
      "grad_norm": 0.053776983724978696,
      "learning_rate": 0.00016223599526823875,
      "loss": 0.841,
      "step": 86000
    },
    {
      "epoch": 1.5100335328920802,
      "grad_norm": 0.06253812142781577,
      "learning_rate": 0.0001622272387814861,
      "loss": 0.8491,
      "step": 86010
    },
    {
      "epoch": 1.5102090977720817,
      "grad_norm": 0.05938730128778984,
      "learning_rate": 0.00016221848151897253,
      "loss": 0.8485,
      "step": 86020
    },
    {
      "epoch": 1.5103846626520832,
      "grad_norm": 0.06419680886680385,
      "learning_rate": 0.0001622097234808091,
      "loss": 0.8423,
      "step": 86030
    },
    {
      "epoch": 1.5105602275320846,
      "grad_norm": 0.06477009006894974,
      "learning_rate": 0.00016220096466710676,
      "loss": 0.8442,
      "step": 86040
    },
    {
      "epoch": 1.5107357924120859,
      "grad_norm": 0.05908773950606371,
      "learning_rate": 0.00016219220507797646,
      "loss": 0.8497,
      "step": 86050
    },
    {
      "epoch": 1.5109113572920871,
      "grad_norm": 0.057562583469793295,
      "learning_rate": 0.0001621834447135292,
      "loss": 0.8466,
      "step": 86060
    },
    {
      "epoch": 1.5110869221720886,
      "grad_norm": 0.053319958319703015,
      "learning_rate": 0.00016217468357387595,
      "loss": 0.8524,
      "step": 86070
    },
    {
      "epoch": 1.51126248705209,
      "grad_norm": 0.06138819568067083,
      "learning_rate": 0.00016216592165912776,
      "loss": 0.8533,
      "step": 86080
    },
    {
      "epoch": 1.5114380519320916,
      "grad_norm": 0.06378648442746798,
      "learning_rate": 0.00016215715896939558,
      "loss": 0.8519,
      "step": 86090
    },
    {
      "epoch": 1.511613616812093,
      "grad_norm": 0.0528837104798008,
      "learning_rate": 0.00016214839550479052,
      "loss": 0.8464,
      "step": 86100
    },
    {
      "epoch": 1.5117891816920943,
      "grad_norm": 0.05824481337113094,
      "learning_rate": 0.00016213963126542353,
      "loss": 0.8455,
      "step": 86110
    },
    {
      "epoch": 1.5119647465720956,
      "grad_norm": 0.0677577243028919,
      "learning_rate": 0.0001621308662514057,
      "loss": 0.8465,
      "step": 86120
    },
    {
      "epoch": 1.512140311452097,
      "grad_norm": 0.057872580437550405,
      "learning_rate": 0.00016212210046284811,
      "loss": 0.8498,
      "step": 86130
    },
    {
      "epoch": 1.5123158763320985,
      "grad_norm": 0.057625977492178126,
      "learning_rate": 0.00016211333389986178,
      "loss": 0.8566,
      "step": 86140
    },
    {
      "epoch": 1.5124914412121,
      "grad_norm": 0.04671276197760504,
      "learning_rate": 0.0001621045665625578,
      "loss": 0.8464,
      "step": 86150
    },
    {
      "epoch": 1.5126670060921015,
      "grad_norm": 0.06009961833816527,
      "learning_rate": 0.00016209579845104724,
      "loss": 0.8486,
      "step": 86160
    },
    {
      "epoch": 1.5128425709721027,
      "grad_norm": 0.04964831102771233,
      "learning_rate": 0.00016208702956544118,
      "loss": 0.85,
      "step": 86170
    },
    {
      "epoch": 1.513018135852104,
      "grad_norm": 0.053023440932321524,
      "learning_rate": 0.00016207825990585082,
      "loss": 0.8481,
      "step": 86180
    },
    {
      "epoch": 1.5131937007321055,
      "grad_norm": 0.05392135612593896,
      "learning_rate": 0.00016206948947238714,
      "loss": 0.8466,
      "step": 86190
    },
    {
      "epoch": 1.513369265612107,
      "grad_norm": 0.05058214363273187,
      "learning_rate": 0.00016206071826516133,
      "loss": 0.8582,
      "step": 86200
    },
    {
      "epoch": 1.5135448304921084,
      "grad_norm": 0.07926169676809618,
      "learning_rate": 0.00016205194628428452,
      "loss": 0.8525,
      "step": 86210
    },
    {
      "epoch": 1.51372039537211,
      "grad_norm": 0.05083987334013135,
      "learning_rate": 0.0001620431735298679,
      "loss": 0.8497,
      "step": 86220
    },
    {
      "epoch": 1.5138959602521112,
      "grad_norm": 0.057070676618394156,
      "learning_rate": 0.0001620344000020225,
      "loss": 0.8538,
      "step": 86230
    },
    {
      "epoch": 1.5140715251321124,
      "grad_norm": 0.0481101888779182,
      "learning_rate": 0.0001620256257008596,
      "loss": 0.8414,
      "step": 86240
    },
    {
      "epoch": 1.5142470900121139,
      "grad_norm": 0.0854041821862504,
      "learning_rate": 0.00016201685062649032,
      "loss": 0.8374,
      "step": 86250
    },
    {
      "epoch": 1.5144226548921154,
      "grad_norm": 0.05122869848677946,
      "learning_rate": 0.00016200807477902582,
      "loss": 0.8494,
      "step": 86260
    },
    {
      "epoch": 1.5145982197721168,
      "grad_norm": 0.06875814154082803,
      "learning_rate": 0.00016199929815857735,
      "loss": 0.8431,
      "step": 86270
    },
    {
      "epoch": 1.5147737846521183,
      "grad_norm": 0.05216343431174146,
      "learning_rate": 0.00016199052076525604,
      "loss": 0.8507,
      "step": 86280
    },
    {
      "epoch": 1.5149493495321196,
      "grad_norm": 0.06286092979801777,
      "learning_rate": 0.00016198174259917321,
      "loss": 0.845,
      "step": 86290
    },
    {
      "epoch": 1.515124914412121,
      "grad_norm": 0.058617679396980364,
      "learning_rate": 0.00016197296366044,
      "loss": 0.8457,
      "step": 86300
    },
    {
      "epoch": 1.5153004792921223,
      "grad_norm": 0.0629213830452067,
      "learning_rate": 0.00016196418394916762,
      "loss": 0.8512,
      "step": 86310
    },
    {
      "epoch": 1.5154760441721238,
      "grad_norm": 0.063616070033483,
      "learning_rate": 0.00016195540346546737,
      "loss": 0.8512,
      "step": 86320
    },
    {
      "epoch": 1.5156516090521253,
      "grad_norm": 0.062021907951642544,
      "learning_rate": 0.00016194662220945044,
      "loss": 0.8456,
      "step": 86330
    },
    {
      "epoch": 1.5158271739321267,
      "grad_norm": 0.06825838205973485,
      "learning_rate": 0.00016193784018122818,
      "loss": 0.8521,
      "step": 86340
    },
    {
      "epoch": 1.516002738812128,
      "grad_norm": 0.08942697390639488,
      "learning_rate": 0.00016192905738091176,
      "loss": 0.8443,
      "step": 86350
    },
    {
      "epoch": 1.5161783036921295,
      "grad_norm": 0.06263521244464022,
      "learning_rate": 0.00016192027380861255,
      "loss": 0.8564,
      "step": 86360
    },
    {
      "epoch": 1.5163538685721307,
      "grad_norm": 0.0745167347032606,
      "learning_rate": 0.00016191148946444176,
      "loss": 0.8444,
      "step": 86370
    },
    {
      "epoch": 1.5165294334521322,
      "grad_norm": 0.05716968926478812,
      "learning_rate": 0.00016190270434851073,
      "loss": 0.8442,
      "step": 86380
    },
    {
      "epoch": 1.5167049983321337,
      "grad_norm": 0.06705683538516093,
      "learning_rate": 0.0001618939184609308,
      "loss": 0.8497,
      "step": 86390
    },
    {
      "epoch": 1.5168805632121352,
      "grad_norm": 0.06766030002074785,
      "learning_rate": 0.0001618851318018132,
      "loss": 0.8455,
      "step": 86400
    },
    {
      "epoch": 1.5170561280921364,
      "grad_norm": 0.08243371146166309,
      "learning_rate": 0.00016187634437126935,
      "loss": 0.8453,
      "step": 86410
    },
    {
      "epoch": 1.517231692972138,
      "grad_norm": 0.05807804443216681,
      "learning_rate": 0.00016186755616941055,
      "loss": 0.8487,
      "step": 86420
    },
    {
      "epoch": 1.5174072578521391,
      "grad_norm": 0.06329721458983907,
      "learning_rate": 0.0001618587671963481,
      "loss": 0.8523,
      "step": 86430
    },
    {
      "epoch": 1.5175828227321406,
      "grad_norm": 0.061765758744956274,
      "learning_rate": 0.00016184997745219345,
      "loss": 0.8524,
      "step": 86440
    },
    {
      "epoch": 1.517758387612142,
      "grad_norm": 0.050386214792974164,
      "learning_rate": 0.00016184118693705793,
      "loss": 0.8483,
      "step": 86450
    },
    {
      "epoch": 1.5179339524921436,
      "grad_norm": 0.07514813983635174,
      "learning_rate": 0.00016183239565105287,
      "loss": 0.8524,
      "step": 86460
    },
    {
      "epoch": 1.5181095173721448,
      "grad_norm": 0.05221487742288582,
      "learning_rate": 0.00016182360359428972,
      "loss": 0.8468,
      "step": 86470
    },
    {
      "epoch": 1.5182850822521463,
      "grad_norm": 0.05974335394998764,
      "learning_rate": 0.00016181481076687988,
      "loss": 0.8488,
      "step": 86480
    },
    {
      "epoch": 1.5184606471321476,
      "grad_norm": 0.054529064795244074,
      "learning_rate": 0.0001618060171689347,
      "loss": 0.8509,
      "step": 86490
    },
    {
      "epoch": 1.518636212012149,
      "grad_norm": 0.06346294034341483,
      "learning_rate": 0.00016179722280056566,
      "loss": 0.8437,
      "step": 86500
    },
    {
      "epoch": 1.5188117768921505,
      "grad_norm": 0.05214411557545104,
      "learning_rate": 0.0001617884276618841,
      "loss": 0.843,
      "step": 86510
    },
    {
      "epoch": 1.518987341772152,
      "grad_norm": 0.047725496393188675,
      "learning_rate": 0.00016177963175300158,
      "loss": 0.8448,
      "step": 86520
    },
    {
      "epoch": 1.5191629066521533,
      "grad_norm": 0.05004892542516842,
      "learning_rate": 0.00016177083507402942,
      "loss": 0.8485,
      "step": 86530
    },
    {
      "epoch": 1.5193384715321547,
      "grad_norm": 0.05691287859981427,
      "learning_rate": 0.00016176203762507916,
      "loss": 0.8471,
      "step": 86540
    },
    {
      "epoch": 1.519514036412156,
      "grad_norm": 0.07395482278608699,
      "learning_rate": 0.00016175323940626222,
      "loss": 0.8511,
      "step": 86550
    },
    {
      "epoch": 1.5196896012921575,
      "grad_norm": 0.0564465428307387,
      "learning_rate": 0.00016174444041769015,
      "loss": 0.8486,
      "step": 86560
    },
    {
      "epoch": 1.519865166172159,
      "grad_norm": 0.08348240411103278,
      "learning_rate": 0.00016173564065947432,
      "loss": 0.8418,
      "step": 86570
    },
    {
      "epoch": 1.5200407310521604,
      "grad_norm": 0.05349076752897858,
      "learning_rate": 0.0001617268401317263,
      "loss": 0.8503,
      "step": 86580
    },
    {
      "epoch": 1.5202162959321617,
      "grad_norm": 0.08011798428548436,
      "learning_rate": 0.00016171803883455757,
      "loss": 0.8517,
      "step": 86590
    },
    {
      "epoch": 1.5203918608121632,
      "grad_norm": 0.06193740060523879,
      "learning_rate": 0.00016170923676807968,
      "loss": 0.8467,
      "step": 86600
    },
    {
      "epoch": 1.5205674256921644,
      "grad_norm": 0.05833356462329386,
      "learning_rate": 0.0001617004339324041,
      "loss": 0.8398,
      "step": 86610
    },
    {
      "epoch": 1.5207429905721659,
      "grad_norm": 0.05208778570037501,
      "learning_rate": 0.0001616916303276424,
      "loss": 0.8489,
      "step": 86620
    },
    {
      "epoch": 1.5209185554521674,
      "grad_norm": 0.08481433090438993,
      "learning_rate": 0.00016168282595390611,
      "loss": 0.8504,
      "step": 86630
    },
    {
      "epoch": 1.5210941203321688,
      "grad_norm": 0.05310315895826084,
      "learning_rate": 0.00016167402081130675,
      "loss": 0.8497,
      "step": 86640
    },
    {
      "epoch": 1.5212696852121703,
      "grad_norm": 0.05912947420059262,
      "learning_rate": 0.00016166521489995598,
      "loss": 0.8467,
      "step": 86650
    },
    {
      "epoch": 1.5214452500921716,
      "grad_norm": 0.08268700658489052,
      "learning_rate": 0.00016165640821996525,
      "loss": 0.857,
      "step": 86660
    },
    {
      "epoch": 1.5216208149721728,
      "grad_norm": 0.07699342378675848,
      "learning_rate": 0.00016164760077144624,
      "loss": 0.8507,
      "step": 86670
    },
    {
      "epoch": 1.5217963798521743,
      "grad_norm": 0.06191423168575263,
      "learning_rate": 0.00016163879255451052,
      "loss": 0.8458,
      "step": 86680
    },
    {
      "epoch": 1.5219719447321758,
      "grad_norm": 0.05799929760574704,
      "learning_rate": 0.00016162998356926967,
      "loss": 0.8483,
      "step": 86690
    },
    {
      "epoch": 1.5221475096121773,
      "grad_norm": 0.061434988788709156,
      "learning_rate": 0.0001616211738158353,
      "loss": 0.8488,
      "step": 86700
    },
    {
      "epoch": 1.5223230744921787,
      "grad_norm": 0.05771688749038098,
      "learning_rate": 0.000161612363294319,
      "loss": 0.8468,
      "step": 86710
    },
    {
      "epoch": 1.52249863937218,
      "grad_norm": 0.0734842233832023,
      "learning_rate": 0.0001616035520048325,
      "loss": 0.8509,
      "step": 86720
    },
    {
      "epoch": 1.5226742042521813,
      "grad_norm": 0.06116457811116448,
      "learning_rate": 0.00016159473994748733,
      "loss": 0.8491,
      "step": 86730
    },
    {
      "epoch": 1.5228497691321827,
      "grad_norm": 0.09351058420269774,
      "learning_rate": 0.00016158592712239522,
      "loss": 0.8464,
      "step": 86740
    },
    {
      "epoch": 1.5230253340121842,
      "grad_norm": 0.06388689818887336,
      "learning_rate": 0.0001615771135296678,
      "loss": 0.8505,
      "step": 86750
    },
    {
      "epoch": 1.5232008988921857,
      "grad_norm": 0.06550855651603729,
      "learning_rate": 0.00016156829916941676,
      "loss": 0.8534,
      "step": 86760
    },
    {
      "epoch": 1.5233764637721872,
      "grad_norm": 0.06343404774389007,
      "learning_rate": 0.00016155948404175373,
      "loss": 0.8461,
      "step": 86770
    },
    {
      "epoch": 1.5235520286521884,
      "grad_norm": 0.04935512151307696,
      "learning_rate": 0.00016155066814679044,
      "loss": 0.8517,
      "step": 86780
    },
    {
      "epoch": 1.5237275935321897,
      "grad_norm": 0.050634705054338904,
      "learning_rate": 0.0001615418514846386,
      "loss": 0.8459,
      "step": 86790
    },
    {
      "epoch": 1.5239031584121912,
      "grad_norm": 0.055821547945348306,
      "learning_rate": 0.00016153303405540986,
      "loss": 0.8535,
      "step": 86800
    },
    {
      "epoch": 1.5240787232921926,
      "grad_norm": 0.06255330923742852,
      "learning_rate": 0.000161524215859216,
      "loss": 0.8575,
      "step": 86810
    },
    {
      "epoch": 1.524254288172194,
      "grad_norm": 0.06429462059806693,
      "learning_rate": 0.00016151539689616872,
      "loss": 0.8485,
      "step": 86820
    },
    {
      "epoch": 1.5244298530521956,
      "grad_norm": 0.05726636350707398,
      "learning_rate": 0.00016150657716637977,
      "loss": 0.8523,
      "step": 86830
    },
    {
      "epoch": 1.5246054179321968,
      "grad_norm": 0.05251033215479348,
      "learning_rate": 0.00016149775666996093,
      "loss": 0.8474,
      "step": 86840
    },
    {
      "epoch": 1.524780982812198,
      "grad_norm": 0.06147636205708735,
      "learning_rate": 0.00016148893540702386,
      "loss": 0.847,
      "step": 86850
    },
    {
      "epoch": 1.5249565476921996,
      "grad_norm": 0.05120487190724601,
      "learning_rate": 0.0001614801133776804,
      "loss": 0.8428,
      "step": 86860
    },
    {
      "epoch": 1.525132112572201,
      "grad_norm": 0.04901220770076652,
      "learning_rate": 0.00016147129058204231,
      "loss": 0.8573,
      "step": 86870
    },
    {
      "epoch": 1.5253076774522025,
      "grad_norm": 0.06861281662469745,
      "learning_rate": 0.0001614624670202214,
      "loss": 0.8484,
      "step": 86880
    },
    {
      "epoch": 1.525483242332204,
      "grad_norm": 0.08977434293422712,
      "learning_rate": 0.00016145364269232944,
      "loss": 0.8447,
      "step": 86890
    },
    {
      "epoch": 1.5256588072122053,
      "grad_norm": 0.05456233004690548,
      "learning_rate": 0.0001614448175984782,
      "loss": 0.849,
      "step": 86900
    },
    {
      "epoch": 1.5258343720922065,
      "grad_norm": 0.08800706328168684,
      "learning_rate": 0.0001614359917387796,
      "loss": 0.8451,
      "step": 86910
    },
    {
      "epoch": 1.526009936972208,
      "grad_norm": 0.051605202513334204,
      "learning_rate": 0.00016142716511334534,
      "loss": 0.8517,
      "step": 86920
    },
    {
      "epoch": 1.5261855018522095,
      "grad_norm": 0.07941939359691608,
      "learning_rate": 0.00016141833772228737,
      "loss": 0.8493,
      "step": 86930
    },
    {
      "epoch": 1.526361066732211,
      "grad_norm": 0.04309566701474835,
      "learning_rate": 0.00016140950956571747,
      "loss": 0.8459,
      "step": 86940
    },
    {
      "epoch": 1.5265366316122124,
      "grad_norm": 0.08269895612275076,
      "learning_rate": 0.00016140068064374747,
      "loss": 0.8459,
      "step": 86950
    },
    {
      "epoch": 1.5267121964922137,
      "grad_norm": 0.04610401384235539,
      "learning_rate": 0.00016139185095648934,
      "loss": 0.8532,
      "step": 86960
    },
    {
      "epoch": 1.526887761372215,
      "grad_norm": 0.06401566033960888,
      "learning_rate": 0.00016138302050405483,
      "loss": 0.8451,
      "step": 86970
    },
    {
      "epoch": 1.5270633262522164,
      "grad_norm": 0.06639276073344114,
      "learning_rate": 0.00016137418928655587,
      "loss": 0.8474,
      "step": 86980
    },
    {
      "epoch": 1.527238891132218,
      "grad_norm": 0.06387142394216752,
      "learning_rate": 0.0001613653573041044,
      "loss": 0.854,
      "step": 86990
    },
    {
      "epoch": 1.5274144560122194,
      "grad_norm": 0.05232708542064106,
      "learning_rate": 0.00016135652455681224,
      "loss": 0.8553,
      "step": 87000
    },
    {
      "epoch": 1.5275900208922208,
      "grad_norm": 0.04840983355797537,
      "learning_rate": 0.0001613476910447914,
      "loss": 0.849,
      "step": 87010
    },
    {
      "epoch": 1.527765585772222,
      "grad_norm": 0.06647213265026873,
      "learning_rate": 0.00016133885676815373,
      "loss": 0.8504,
      "step": 87020
    },
    {
      "epoch": 1.5279411506522236,
      "grad_norm": 0.0580104884005136,
      "learning_rate": 0.00016133002172701117,
      "loss": 0.8544,
      "step": 87030
    },
    {
      "epoch": 1.5281167155322248,
      "grad_norm": 0.05483272549226063,
      "learning_rate": 0.00016132118592147567,
      "loss": 0.8479,
      "step": 87040
    },
    {
      "epoch": 1.5282922804122263,
      "grad_norm": 0.04904221453829947,
      "learning_rate": 0.00016131234935165916,
      "loss": 0.8387,
      "step": 87050
    },
    {
      "epoch": 1.5284678452922278,
      "grad_norm": 0.06126613812093932,
      "learning_rate": 0.00016130351201767369,
      "loss": 0.845,
      "step": 87060
    },
    {
      "epoch": 1.5286434101722293,
      "grad_norm": 0.06595228859161271,
      "learning_rate": 0.00016129467391963112,
      "loss": 0.8605,
      "step": 87070
    },
    {
      "epoch": 1.5288189750522305,
      "grad_norm": 0.05187293609306512,
      "learning_rate": 0.0001612858350576435,
      "loss": 0.855,
      "step": 87080
    },
    {
      "epoch": 1.528994539932232,
      "grad_norm": 0.08092952635345907,
      "learning_rate": 0.00016127699543182276,
      "loss": 0.856,
      "step": 87090
    },
    {
      "epoch": 1.5291701048122333,
      "grad_norm": 0.07301361950208915,
      "learning_rate": 0.000161268155042281,
      "loss": 0.8433,
      "step": 87100
    },
    {
      "epoch": 1.5293456696922347,
      "grad_norm": 0.04697799891793817,
      "learning_rate": 0.0001612593138891301,
      "loss": 0.8467,
      "step": 87110
    },
    {
      "epoch": 1.5295212345722362,
      "grad_norm": 0.0796054616909619,
      "learning_rate": 0.00016125047197248216,
      "loss": 0.849,
      "step": 87120
    },
    {
      "epoch": 1.5296967994522377,
      "grad_norm": 0.06399707729951509,
      "learning_rate": 0.00016124162929244922,
      "loss": 0.8447,
      "step": 87130
    },
    {
      "epoch": 1.529872364332239,
      "grad_norm": 0.07740700001596557,
      "learning_rate": 0.00016123278584914325,
      "loss": 0.8562,
      "step": 87140
    },
    {
      "epoch": 1.5300479292122404,
      "grad_norm": 0.0908341519632411,
      "learning_rate": 0.00016122394164267637,
      "loss": 0.8486,
      "step": 87150
    },
    {
      "epoch": 1.5302234940922417,
      "grad_norm": 0.10163251479421867,
      "learning_rate": 0.0001612150966731606,
      "loss": 0.8477,
      "step": 87160
    },
    {
      "epoch": 1.5303990589722432,
      "grad_norm": 0.09429663817832137,
      "learning_rate": 0.000161206250940708,
      "loss": 0.85,
      "step": 87170
    },
    {
      "epoch": 1.5305746238522446,
      "grad_norm": 0.06658048182586411,
      "learning_rate": 0.00016119740444543067,
      "loss": 0.8443,
      "step": 87180
    },
    {
      "epoch": 1.530750188732246,
      "grad_norm": 0.05406209280113798,
      "learning_rate": 0.0001611885571874407,
      "loss": 0.8562,
      "step": 87190
    },
    {
      "epoch": 1.5309257536122474,
      "grad_norm": 0.059237322677251916,
      "learning_rate": 0.0001611797091668502,
      "loss": 0.8458,
      "step": 87200
    },
    {
      "epoch": 1.5311013184922488,
      "grad_norm": 0.05309648243252786,
      "learning_rate": 0.00016117086038377118,
      "loss": 0.8449,
      "step": 87210
    },
    {
      "epoch": 1.53127688337225,
      "grad_norm": 0.06457631692850127,
      "learning_rate": 0.00016116201083831588,
      "loss": 0.8515,
      "step": 87220
    },
    {
      "epoch": 1.5314524482522516,
      "grad_norm": 0.09518947025961881,
      "learning_rate": 0.00016115316053059636,
      "loss": 0.8471,
      "step": 87230
    },
    {
      "epoch": 1.531628013132253,
      "grad_norm": 0.10593933165237389,
      "learning_rate": 0.0001611443094607248,
      "loss": 0.8511,
      "step": 87240
    },
    {
      "epoch": 1.5318035780122545,
      "grad_norm": 0.09908597014959801,
      "learning_rate": 0.00016113545762881328,
      "loss": 0.8418,
      "step": 87250
    },
    {
      "epoch": 1.5319791428922558,
      "grad_norm": 0.050837745246552817,
      "learning_rate": 0.00016112660503497396,
      "loss": 0.8493,
      "step": 87260
    },
    {
      "epoch": 1.5321547077722573,
      "grad_norm": 0.05955484770390563,
      "learning_rate": 0.00016111775167931906,
      "loss": 0.8596,
      "step": 87270
    },
    {
      "epoch": 1.5323302726522585,
      "grad_norm": 0.05704589416456655,
      "learning_rate": 0.0001611088975619607,
      "loss": 0.8454,
      "step": 87280
    },
    {
      "epoch": 1.53250583753226,
      "grad_norm": 0.05798373116837513,
      "learning_rate": 0.00016110004268301114,
      "loss": 0.8528,
      "step": 87290
    },
    {
      "epoch": 1.5326814024122615,
      "grad_norm": 0.05668659137455504,
      "learning_rate": 0.0001610911870425825,
      "loss": 0.8437,
      "step": 87300
    },
    {
      "epoch": 1.532856967292263,
      "grad_norm": 0.06896407700024566,
      "learning_rate": 0.000161082330640787,
      "loss": 0.8466,
      "step": 87310
    },
    {
      "epoch": 1.5330325321722642,
      "grad_norm": 0.07653411216899161,
      "learning_rate": 0.00016107347347773686,
      "loss": 0.8505,
      "step": 87320
    },
    {
      "epoch": 1.5332080970522657,
      "grad_norm": 0.07065416197429104,
      "learning_rate": 0.0001610646155535443,
      "loss": 0.8461,
      "step": 87330
    },
    {
      "epoch": 1.533383661932267,
      "grad_norm": 0.04906838958087765,
      "learning_rate": 0.00016105575686832152,
      "loss": 0.8596,
      "step": 87340
    },
    {
      "epoch": 1.5335592268122684,
      "grad_norm": 0.0896259078849688,
      "learning_rate": 0.00016104689742218081,
      "loss": 0.8552,
      "step": 87350
    },
    {
      "epoch": 1.53373479169227,
      "grad_norm": 0.0652538895072593,
      "learning_rate": 0.0001610380372152344,
      "loss": 0.8498,
      "step": 87360
    },
    {
      "epoch": 1.5339103565722714,
      "grad_norm": 0.04449124333082787,
      "learning_rate": 0.0001610291762475946,
      "loss": 0.8502,
      "step": 87370
    },
    {
      "epoch": 1.5340859214522728,
      "grad_norm": 0.07342958265951748,
      "learning_rate": 0.00016102031451937357,
      "loss": 0.8453,
      "step": 87380
    },
    {
      "epoch": 1.534261486332274,
      "grad_norm": 0.07231260951152439,
      "learning_rate": 0.00016101145203068366,
      "loss": 0.8495,
      "step": 87390
    },
    {
      "epoch": 1.5344370512122754,
      "grad_norm": 0.06410511641247607,
      "learning_rate": 0.00016100258878163715,
      "loss": 0.8488,
      "step": 87400
    },
    {
      "epoch": 1.5346126160922768,
      "grad_norm": 0.05746891951317036,
      "learning_rate": 0.00016099372477234637,
      "loss": 0.8424,
      "step": 87410
    },
    {
      "epoch": 1.5347881809722783,
      "grad_norm": 0.07055229471665345,
      "learning_rate": 0.0001609848600029236,
      "loss": 0.8421,
      "step": 87420
    },
    {
      "epoch": 1.5349637458522798,
      "grad_norm": 0.0641875069671382,
      "learning_rate": 0.00016097599447348113,
      "loss": 0.8461,
      "step": 87430
    },
    {
      "epoch": 1.5351393107322813,
      "grad_norm": 0.0672263308602931,
      "learning_rate": 0.0001609671281841313,
      "loss": 0.8498,
      "step": 87440
    },
    {
      "epoch": 1.5353148756122825,
      "grad_norm": 0.0540187387652721,
      "learning_rate": 0.00016095826113498652,
      "loss": 0.8411,
      "step": 87450
    },
    {
      "epoch": 1.5354904404922838,
      "grad_norm": 0.07837583629926312,
      "learning_rate": 0.00016094939332615904,
      "loss": 0.8471,
      "step": 87460
    },
    {
      "epoch": 1.5356660053722853,
      "grad_norm": 0.05997264869302118,
      "learning_rate": 0.00016094052475776127,
      "loss": 0.8516,
      "step": 87470
    },
    {
      "epoch": 1.5358415702522867,
      "grad_norm": 0.051542115722653346,
      "learning_rate": 0.00016093165542990553,
      "loss": 0.8514,
      "step": 87480
    },
    {
      "epoch": 1.5360171351322882,
      "grad_norm": 0.05936927701813411,
      "learning_rate": 0.00016092278534270428,
      "loss": 0.8498,
      "step": 87490
    },
    {
      "epoch": 1.5361927000122897,
      "grad_norm": 0.053781518579854526,
      "learning_rate": 0.00016091391449626983,
      "loss": 0.8404,
      "step": 87500
    },
    {
      "epoch": 1.536368264892291,
      "grad_norm": 0.07623883831972471,
      "learning_rate": 0.00016090504289071459,
      "loss": 0.8488,
      "step": 87510
    },
    {
      "epoch": 1.5365438297722922,
      "grad_norm": 0.05255936214948905,
      "learning_rate": 0.000160896170526151,
      "loss": 0.8524,
      "step": 87520
    },
    {
      "epoch": 1.5367193946522937,
      "grad_norm": 0.08313722369180819,
      "learning_rate": 0.00016088729740269147,
      "loss": 0.8522,
      "step": 87530
    },
    {
      "epoch": 1.5368949595322952,
      "grad_norm": 0.052804189343949155,
      "learning_rate": 0.00016087842352044838,
      "loss": 0.8387,
      "step": 87540
    },
    {
      "epoch": 1.5370705244122966,
      "grad_norm": 0.059734540653162936,
      "learning_rate": 0.00016086954887953417,
      "loss": 0.8458,
      "step": 87550
    },
    {
      "epoch": 1.5372460892922981,
      "grad_norm": 0.06979985560434786,
      "learning_rate": 0.0001608606734800613,
      "loss": 0.8459,
      "step": 87560
    },
    {
      "epoch": 1.5374216541722994,
      "grad_norm": 0.06211247914637952,
      "learning_rate": 0.00016085179732214225,
      "loss": 0.8446,
      "step": 87570
    },
    {
      "epoch": 1.5375972190523006,
      "grad_norm": 0.059060646566745235,
      "learning_rate": 0.00016084292040588946,
      "loss": 0.8548,
      "step": 87580
    },
    {
      "epoch": 1.537772783932302,
      "grad_norm": 0.07040032833723661,
      "learning_rate": 0.0001608340427314154,
      "loss": 0.8546,
      "step": 87590
    },
    {
      "epoch": 1.5379483488123036,
      "grad_norm": 0.04967446449371587,
      "learning_rate": 0.00016082516429883253,
      "loss": 0.8519,
      "step": 87600
    },
    {
      "epoch": 1.538123913692305,
      "grad_norm": 0.0620176248655262,
      "learning_rate": 0.0001608162851082534,
      "loss": 0.8491,
      "step": 87610
    },
    {
      "epoch": 1.5382994785723065,
      "grad_norm": 0.06416613123765057,
      "learning_rate": 0.00016080740515979044,
      "loss": 0.8481,
      "step": 87620
    },
    {
      "epoch": 1.5384750434523078,
      "grad_norm": 0.0683890551241037,
      "learning_rate": 0.0001607985244535562,
      "loss": 0.8476,
      "step": 87630
    },
    {
      "epoch": 1.538650608332309,
      "grad_norm": 0.05795003118731202,
      "learning_rate": 0.0001607896429896632,
      "loss": 0.8475,
      "step": 87640
    },
    {
      "epoch": 1.5388261732123105,
      "grad_norm": 0.059713221295219596,
      "learning_rate": 0.00016078076076822396,
      "loss": 0.8475,
      "step": 87650
    },
    {
      "epoch": 1.539001738092312,
      "grad_norm": 0.054975066890178434,
      "learning_rate": 0.000160771877789351,
      "loss": 0.8543,
      "step": 87660
    },
    {
      "epoch": 1.5391773029723135,
      "grad_norm": 0.05927370398092108,
      "learning_rate": 0.00016076299405315693,
      "loss": 0.8499,
      "step": 87670
    },
    {
      "epoch": 1.539352867852315,
      "grad_norm": 0.056411845635283515,
      "learning_rate": 0.00016075410955975424,
      "loss": 0.857,
      "step": 87680
    },
    {
      "epoch": 1.5395284327323162,
      "grad_norm": 0.05807270826615168,
      "learning_rate": 0.00016074522430925558,
      "loss": 0.8454,
      "step": 87690
    },
    {
      "epoch": 1.5397039976123177,
      "grad_norm": 0.0457204141384161,
      "learning_rate": 0.00016073633830177345,
      "loss": 0.8428,
      "step": 87700
    },
    {
      "epoch": 1.539879562492319,
      "grad_norm": 0.04756695590700125,
      "learning_rate": 0.00016072745153742046,
      "loss": 0.8553,
      "step": 87710
    },
    {
      "epoch": 1.5400551273723204,
      "grad_norm": 0.07280462075296015,
      "learning_rate": 0.0001607185640163092,
      "loss": 0.8487,
      "step": 87720
    },
    {
      "epoch": 1.540230692252322,
      "grad_norm": 0.07437366113979114,
      "learning_rate": 0.00016070967573855234,
      "loss": 0.8509,
      "step": 87730
    },
    {
      "epoch": 1.5404062571323234,
      "grad_norm": 0.04774820171415767,
      "learning_rate": 0.0001607007867042624,
      "loss": 0.8526,
      "step": 87740
    },
    {
      "epoch": 1.5405818220123246,
      "grad_norm": 0.05420414238923236,
      "learning_rate": 0.00016069189691355205,
      "loss": 0.8525,
      "step": 87750
    },
    {
      "epoch": 1.540757386892326,
      "grad_norm": 0.05015751928265845,
      "learning_rate": 0.00016068300636653396,
      "loss": 0.8408,
      "step": 87760
    },
    {
      "epoch": 1.5409329517723274,
      "grad_norm": 0.05923379846443744,
      "learning_rate": 0.00016067411506332072,
      "loss": 0.8376,
      "step": 87770
    },
    {
      "epoch": 1.5411085166523288,
      "grad_norm": 0.05461698614414783,
      "learning_rate": 0.000160665223004025,
      "loss": 0.8451,
      "step": 87780
    },
    {
      "epoch": 1.5412840815323303,
      "grad_norm": 0.056349955354787234,
      "learning_rate": 0.0001606563301887595,
      "loss": 0.8392,
      "step": 87790
    },
    {
      "epoch": 1.5414596464123318,
      "grad_norm": 0.08569945676805348,
      "learning_rate": 0.0001606474366176368,
      "loss": 0.8524,
      "step": 87800
    },
    {
      "epoch": 1.541635211292333,
      "grad_norm": 0.05197683106540259,
      "learning_rate": 0.0001606385422907697,
      "loss": 0.8498,
      "step": 87810
    },
    {
      "epoch": 1.5418107761723345,
      "grad_norm": 0.05393253961275397,
      "learning_rate": 0.00016062964720827078,
      "loss": 0.8518,
      "step": 87820
    },
    {
      "epoch": 1.5419863410523358,
      "grad_norm": 0.07610877284370929,
      "learning_rate": 0.00016062075137025287,
      "loss": 0.8511,
      "step": 87830
    },
    {
      "epoch": 1.5421619059323373,
      "grad_norm": 0.050628250674311905,
      "learning_rate": 0.0001606118547768286,
      "loss": 0.8456,
      "step": 87840
    },
    {
      "epoch": 1.5423374708123387,
      "grad_norm": 0.13170136400283008,
      "learning_rate": 0.00016060295742811067,
      "loss": 0.8499,
      "step": 87850
    },
    {
      "epoch": 1.5425130356923402,
      "grad_norm": 0.04651088554056199,
      "learning_rate": 0.0001605940593242119,
      "loss": 0.8522,
      "step": 87860
    },
    {
      "epoch": 1.5426886005723415,
      "grad_norm": 0.05349352984971621,
      "learning_rate": 0.00016058516046524494,
      "loss": 0.8372,
      "step": 87870
    },
    {
      "epoch": 1.542864165452343,
      "grad_norm": 0.07119577885276931,
      "learning_rate": 0.0001605762608513226,
      "loss": 0.8492,
      "step": 87880
    },
    {
      "epoch": 1.5430397303323442,
      "grad_norm": 0.04957065353699747,
      "learning_rate": 0.0001605673604825576,
      "loss": 0.8546,
      "step": 87890
    },
    {
      "epoch": 1.5432152952123457,
      "grad_norm": 0.06799939258613708,
      "learning_rate": 0.0001605584593590627,
      "loss": 0.849,
      "step": 87900
    },
    {
      "epoch": 1.5433908600923472,
      "grad_norm": 0.06489945452482677,
      "learning_rate": 0.00016054955748095076,
      "loss": 0.8538,
      "step": 87910
    },
    {
      "epoch": 1.5435664249723486,
      "grad_norm": 0.06237649670453726,
      "learning_rate": 0.0001605406548483345,
      "loss": 0.8497,
      "step": 87920
    },
    {
      "epoch": 1.54374198985235,
      "grad_norm": 0.0788560675158259,
      "learning_rate": 0.00016053175146132673,
      "loss": 0.8488,
      "step": 87930
    },
    {
      "epoch": 1.5439175547323514,
      "grad_norm": 0.07864765688486974,
      "learning_rate": 0.00016052284732004024,
      "loss": 0.8551,
      "step": 87940
    },
    {
      "epoch": 1.5440931196123526,
      "grad_norm": 0.06374191006307688,
      "learning_rate": 0.00016051394242458787,
      "loss": 0.8443,
      "step": 87950
    },
    {
      "epoch": 1.544268684492354,
      "grad_norm": 0.05335728777712646,
      "learning_rate": 0.00016050503677508246,
      "loss": 0.848,
      "step": 87960
    },
    {
      "epoch": 1.5444442493723556,
      "grad_norm": 0.07922264242369605,
      "learning_rate": 0.0001604961303716368,
      "loss": 0.8404,
      "step": 87970
    },
    {
      "epoch": 1.544619814252357,
      "grad_norm": 0.05412170908488372,
      "learning_rate": 0.00016048722321436383,
      "loss": 0.8492,
      "step": 87980
    },
    {
      "epoch": 1.5447953791323583,
      "grad_norm": 0.05647199280411568,
      "learning_rate": 0.00016047831530337627,
      "loss": 0.8514,
      "step": 87990
    },
    {
      "epoch": 1.5449709440123598,
      "grad_norm": 0.061543795564335896,
      "learning_rate": 0.0001604694066387871,
      "loss": 0.8488,
      "step": 88000
    },
    {
      "epoch": 1.545146508892361,
      "grad_norm": 0.050694011367287774,
      "learning_rate": 0.00016046049722070915,
      "loss": 0.8501,
      "step": 88010
    },
    {
      "epoch": 1.5453220737723625,
      "grad_norm": 0.051124258563554285,
      "learning_rate": 0.00016045158704925525,
      "loss": 0.8494,
      "step": 88020
    },
    {
      "epoch": 1.545497638652364,
      "grad_norm": 0.06302862810433006,
      "learning_rate": 0.00016044267612453842,
      "loss": 0.8473,
      "step": 88030
    },
    {
      "epoch": 1.5456732035323655,
      "grad_norm": 0.06096943740106845,
      "learning_rate": 0.00016043376444667146,
      "loss": 0.8454,
      "step": 88040
    },
    {
      "epoch": 1.5458487684123667,
      "grad_norm": 0.0645368290324632,
      "learning_rate": 0.00016042485201576729,
      "loss": 0.8383,
      "step": 88050
    },
    {
      "epoch": 1.5460243332923682,
      "grad_norm": 0.049624653515132525,
      "learning_rate": 0.00016041593883193888,
      "loss": 0.8462,
      "step": 88060
    },
    {
      "epoch": 1.5461998981723695,
      "grad_norm": 0.05055230813281649,
      "learning_rate": 0.00016040702489529912,
      "loss": 0.8472,
      "step": 88070
    },
    {
      "epoch": 1.546375463052371,
      "grad_norm": 0.05367393950291271,
      "learning_rate": 0.000160398110205961,
      "loss": 0.8503,
      "step": 88080
    },
    {
      "epoch": 1.5465510279323724,
      "grad_norm": 0.07735134052896653,
      "learning_rate": 0.00016038919476403743,
      "loss": 0.8496,
      "step": 88090
    },
    {
      "epoch": 1.546726592812374,
      "grad_norm": 0.051543715312734266,
      "learning_rate": 0.00016038027856964143,
      "loss": 0.852,
      "step": 88100
    },
    {
      "epoch": 1.5469021576923754,
      "grad_norm": 0.06713413382240226,
      "learning_rate": 0.00016037136162288582,
      "loss": 0.8573,
      "step": 88110
    },
    {
      "epoch": 1.5470777225723766,
      "grad_norm": 0.06739465327836253,
      "learning_rate": 0.00016036244392388376,
      "loss": 0.844,
      "step": 88120
    },
    {
      "epoch": 1.547253287452378,
      "grad_norm": 0.0668324989970712,
      "learning_rate": 0.00016035352547274812,
      "loss": 0.8467,
      "step": 88130
    },
    {
      "epoch": 1.5474288523323794,
      "grad_norm": 0.09528290421779649,
      "learning_rate": 0.00016034460626959194,
      "loss": 0.8426,
      "step": 88140
    },
    {
      "epoch": 1.5476044172123808,
      "grad_norm": 0.08148505263461221,
      "learning_rate": 0.00016033568631452821,
      "loss": 0.845,
      "step": 88150
    },
    {
      "epoch": 1.5477799820923823,
      "grad_norm": 0.08306487545775168,
      "learning_rate": 0.00016032676560767,
      "loss": 0.8387,
      "step": 88160
    },
    {
      "epoch": 1.5479555469723838,
      "grad_norm": 0.0631228546814343,
      "learning_rate": 0.00016031784414913028,
      "loss": 0.8491,
      "step": 88170
    },
    {
      "epoch": 1.548131111852385,
      "grad_norm": 0.06170747194712506,
      "learning_rate": 0.00016030892193902212,
      "loss": 0.8579,
      "step": 88180
    },
    {
      "epoch": 1.5483066767323863,
      "grad_norm": 0.053697332745529736,
      "learning_rate": 0.0001602999989774585,
      "loss": 0.8477,
      "step": 88190
    },
    {
      "epoch": 1.5484822416123878,
      "grad_norm": 0.0553295765604524,
      "learning_rate": 0.0001602910752645526,
      "loss": 0.8519,
      "step": 88200
    },
    {
      "epoch": 1.5486578064923893,
      "grad_norm": 0.0647998212387906,
      "learning_rate": 0.00016028215080041738,
      "loss": 0.846,
      "step": 88210
    },
    {
      "epoch": 1.5488333713723907,
      "grad_norm": 0.05476234694836397,
      "learning_rate": 0.00016027322558516597,
      "loss": 0.8514,
      "step": 88220
    },
    {
      "epoch": 1.5490089362523922,
      "grad_norm": 0.06282222476014848,
      "learning_rate": 0.00016026429961891138,
      "loss": 0.855,
      "step": 88230
    },
    {
      "epoch": 1.5491845011323935,
      "grad_norm": 0.06701701338471586,
      "learning_rate": 0.0001602553729017668,
      "loss": 0.8494,
      "step": 88240
    },
    {
      "epoch": 1.5493600660123947,
      "grad_norm": 0.08155078073142208,
      "learning_rate": 0.00016024644543384528,
      "loss": 0.8481,
      "step": 88250
    },
    {
      "epoch": 1.5495356308923962,
      "grad_norm": 0.05729407581065136,
      "learning_rate": 0.00016023751721525994,
      "loss": 0.8444,
      "step": 88260
    },
    {
      "epoch": 1.5497111957723977,
      "grad_norm": 0.05783369762170022,
      "learning_rate": 0.00016022858824612393,
      "loss": 0.8434,
      "step": 88270
    },
    {
      "epoch": 1.5498867606523992,
      "grad_norm": 0.051016863323899306,
      "learning_rate": 0.00016021965852655028,
      "loss": 0.8476,
      "step": 88280
    },
    {
      "epoch": 1.5500623255324006,
      "grad_norm": 0.07662158958791614,
      "learning_rate": 0.0001602107280566523,
      "loss": 0.8476,
      "step": 88290
    },
    {
      "epoch": 1.550237890412402,
      "grad_norm": 0.0691013778827576,
      "learning_rate": 0.000160201796836543,
      "loss": 0.8516,
      "step": 88300
    },
    {
      "epoch": 1.5504134552924032,
      "grad_norm": 0.05963990022038693,
      "learning_rate": 0.00016019286486633556,
      "loss": 0.8574,
      "step": 88310
    },
    {
      "epoch": 1.5505890201724046,
      "grad_norm": 0.05430415160057141,
      "learning_rate": 0.00016018393214614323,
      "loss": 0.8429,
      "step": 88320
    },
    {
      "epoch": 1.550764585052406,
      "grad_norm": 0.05996548764599677,
      "learning_rate": 0.0001601749986760791,
      "loss": 0.8515,
      "step": 88330
    },
    {
      "epoch": 1.5509401499324076,
      "grad_norm": 0.06526205407218191,
      "learning_rate": 0.00016016606445625644,
      "loss": 0.8466,
      "step": 88340
    },
    {
      "epoch": 1.551115714812409,
      "grad_norm": 0.0712808859374712,
      "learning_rate": 0.0001601571294867884,
      "loss": 0.8466,
      "step": 88350
    },
    {
      "epoch": 1.5512912796924103,
      "grad_norm": 0.05384173280871978,
      "learning_rate": 0.00016014819376778812,
      "loss": 0.8496,
      "step": 88360
    },
    {
      "epoch": 1.5514668445724116,
      "grad_norm": 0.08317107856886348,
      "learning_rate": 0.00016013925729936896,
      "loss": 0.8514,
      "step": 88370
    },
    {
      "epoch": 1.551642409452413,
      "grad_norm": 0.05111917805532528,
      "learning_rate": 0.00016013032008164402,
      "loss": 0.8432,
      "step": 88380
    },
    {
      "epoch": 1.5518179743324145,
      "grad_norm": 0.08576891342094982,
      "learning_rate": 0.00016012138211472668,
      "loss": 0.8459,
      "step": 88390
    },
    {
      "epoch": 1.551993539212416,
      "grad_norm": 0.06495403405657989,
      "learning_rate": 0.00016011244339873002,
      "loss": 0.8493,
      "step": 88400
    },
    {
      "epoch": 1.5521691040924175,
      "grad_norm": 0.07032652576154196,
      "learning_rate": 0.00016010350393376742,
      "loss": 0.8525,
      "step": 88410
    },
    {
      "epoch": 1.5523446689724187,
      "grad_norm": 0.04925381706446803,
      "learning_rate": 0.0001600945637199521,
      "loss": 0.8388,
      "step": 88420
    },
    {
      "epoch": 1.5525202338524202,
      "grad_norm": 0.07071921778151775,
      "learning_rate": 0.00016008562275739727,
      "loss": 0.8532,
      "step": 88430
    },
    {
      "epoch": 1.5526957987324215,
      "grad_norm": 0.04921774009760984,
      "learning_rate": 0.00016007668104621635,
      "loss": 0.8489,
      "step": 88440
    },
    {
      "epoch": 1.552871363612423,
      "grad_norm": 0.05282641106717509,
      "learning_rate": 0.0001600677385865225,
      "loss": 0.852,
      "step": 88450
    },
    {
      "epoch": 1.5530469284924244,
      "grad_norm": 0.054797419822645896,
      "learning_rate": 0.0001600587953784291,
      "loss": 0.8455,
      "step": 88460
    },
    {
      "epoch": 1.553222493372426,
      "grad_norm": 0.04378570430651425,
      "learning_rate": 0.00016004985142204945,
      "loss": 0.8483,
      "step": 88470
    },
    {
      "epoch": 1.5533980582524272,
      "grad_norm": 0.05204091314603641,
      "learning_rate": 0.00016004090671749684,
      "loss": 0.8548,
      "step": 88480
    },
    {
      "epoch": 1.5535736231324286,
      "grad_norm": 0.05254989047127055,
      "learning_rate": 0.00016003196126488468,
      "loss": 0.8469,
      "step": 88490
    },
    {
      "epoch": 1.55374918801243,
      "grad_norm": 0.05012565057882266,
      "learning_rate": 0.00016002301506432622,
      "loss": 0.8443,
      "step": 88500
    },
    {
      "epoch": 1.5539247528924314,
      "grad_norm": 0.048538592089351934,
      "learning_rate": 0.0001600140681159349,
      "loss": 0.843,
      "step": 88510
    },
    {
      "epoch": 1.5541003177724328,
      "grad_norm": 0.0457394050187913,
      "learning_rate": 0.00016000512041982394,
      "loss": 0.8564,
      "step": 88520
    },
    {
      "epoch": 1.5542758826524343,
      "grad_norm": 0.06649201881095054,
      "learning_rate": 0.00015999617197610684,
      "loss": 0.852,
      "step": 88530
    },
    {
      "epoch": 1.5544514475324356,
      "grad_norm": 0.04672136897033637,
      "learning_rate": 0.00015998722278489695,
      "loss": 0.8473,
      "step": 88540
    },
    {
      "epoch": 1.554627012412437,
      "grad_norm": 0.06294784327979482,
      "learning_rate": 0.0001599782728463076,
      "loss": 0.8506,
      "step": 88550
    },
    {
      "epoch": 1.5548025772924383,
      "grad_norm": 0.051452873912756825,
      "learning_rate": 0.00015996932216045227,
      "loss": 0.8463,
      "step": 88560
    },
    {
      "epoch": 1.5549781421724398,
      "grad_norm": 0.049189441520006205,
      "learning_rate": 0.00015996037072744433,
      "loss": 0.8497,
      "step": 88570
    },
    {
      "epoch": 1.5551537070524413,
      "grad_norm": 0.0724836231786538,
      "learning_rate": 0.00015995141854739718,
      "loss": 0.8474,
      "step": 88580
    },
    {
      "epoch": 1.5553292719324427,
      "grad_norm": 0.05804144157832132,
      "learning_rate": 0.00015994246562042424,
      "loss": 0.8434,
      "step": 88590
    },
    {
      "epoch": 1.555504836812444,
      "grad_norm": 0.05003232552640343,
      "learning_rate": 0.00015993351194663896,
      "loss": 0.8426,
      "step": 88600
    },
    {
      "epoch": 1.5556804016924455,
      "grad_norm": 0.06031231487671292,
      "learning_rate": 0.00015992455752615484,
      "loss": 0.854,
      "step": 88610
    },
    {
      "epoch": 1.5558559665724467,
      "grad_norm": 0.04576508548163368,
      "learning_rate": 0.00015991560235908525,
      "loss": 0.8486,
      "step": 88620
    },
    {
      "epoch": 1.5560315314524482,
      "grad_norm": 0.06051559688133626,
      "learning_rate": 0.00015990664644554368,
      "loss": 0.8506,
      "step": 88630
    },
    {
      "epoch": 1.5562070963324497,
      "grad_norm": 0.054291799131279586,
      "learning_rate": 0.00015989768978564363,
      "loss": 0.8509,
      "step": 88640
    },
    {
      "epoch": 1.5563826612124512,
      "grad_norm": 0.08643777691916793,
      "learning_rate": 0.00015988873237949856,
      "loss": 0.848,
      "step": 88650
    },
    {
      "epoch": 1.5565582260924524,
      "grad_norm": 0.05036279806804781,
      "learning_rate": 0.00015987977422722194,
      "loss": 0.8549,
      "step": 88660
    },
    {
      "epoch": 1.556733790972454,
      "grad_norm": 0.05762000529333138,
      "learning_rate": 0.00015987081532892734,
      "loss": 0.8501,
      "step": 88670
    },
    {
      "epoch": 1.5569093558524552,
      "grad_norm": 0.06219722378045775,
      "learning_rate": 0.0001598618556847282,
      "loss": 0.8512,
      "step": 88680
    },
    {
      "epoch": 1.5570849207324566,
      "grad_norm": 0.05751065250900086,
      "learning_rate": 0.00015985289529473807,
      "loss": 0.8504,
      "step": 88690
    },
    {
      "epoch": 1.5572604856124581,
      "grad_norm": 0.07027597801459065,
      "learning_rate": 0.00015984393415907045,
      "loss": 0.8565,
      "step": 88700
    },
    {
      "epoch": 1.5574360504924596,
      "grad_norm": 0.052995549901803926,
      "learning_rate": 0.00015983497227783892,
      "loss": 0.8506,
      "step": 88710
    },
    {
      "epoch": 1.5576116153724608,
      "grad_norm": 0.062092029269798184,
      "learning_rate": 0.00015982600965115705,
      "loss": 0.8498,
      "step": 88720
    },
    {
      "epoch": 1.5577871802524623,
      "grad_norm": 0.07814759280050043,
      "learning_rate": 0.00015981704627913835,
      "loss": 0.86,
      "step": 88730
    },
    {
      "epoch": 1.5579627451324636,
      "grad_norm": 0.05685220999653638,
      "learning_rate": 0.00015980808216189637,
      "loss": 0.851,
      "step": 88740
    },
    {
      "epoch": 1.558138310012465,
      "grad_norm": 0.09110692982604841,
      "learning_rate": 0.00015979911729954477,
      "loss": 0.8511,
      "step": 88750
    },
    {
      "epoch": 1.5583138748924665,
      "grad_norm": 0.05996319216297568,
      "learning_rate": 0.00015979015169219704,
      "loss": 0.8452,
      "step": 88760
    },
    {
      "epoch": 1.558489439772468,
      "grad_norm": 0.0604021298173742,
      "learning_rate": 0.00015978118533996685,
      "loss": 0.8536,
      "step": 88770
    },
    {
      "epoch": 1.5586650046524693,
      "grad_norm": 0.06390410238208531,
      "learning_rate": 0.00015977221824296778,
      "loss": 0.8506,
      "step": 88780
    },
    {
      "epoch": 1.5588405695324707,
      "grad_norm": 0.061472288992131446,
      "learning_rate": 0.0001597632504013134,
      "loss": 0.8434,
      "step": 88790
    },
    {
      "epoch": 1.559016134412472,
      "grad_norm": 0.07439558690993833,
      "learning_rate": 0.0001597542818151174,
      "loss": 0.8446,
      "step": 88800
    },
    {
      "epoch": 1.5591916992924735,
      "grad_norm": 0.05360402589715738,
      "learning_rate": 0.0001597453124844934,
      "loss": 0.845,
      "step": 88810
    },
    {
      "epoch": 1.559367264172475,
      "grad_norm": 0.0598744398739961,
      "learning_rate": 0.000159736342409555,
      "loss": 0.8514,
      "step": 88820
    },
    {
      "epoch": 1.5595428290524764,
      "grad_norm": 0.0661249309422794,
      "learning_rate": 0.00015972737159041593,
      "loss": 0.8487,
      "step": 88830
    },
    {
      "epoch": 1.559718393932478,
      "grad_norm": 0.07636917032945023,
      "learning_rate": 0.00015971840002718978,
      "loss": 0.8436,
      "step": 88840
    },
    {
      "epoch": 1.5598939588124792,
      "grad_norm": 0.060277134665484365,
      "learning_rate": 0.00015970942771999027,
      "loss": 0.8427,
      "step": 88850
    },
    {
      "epoch": 1.5600695236924804,
      "grad_norm": 0.046329278382515875,
      "learning_rate": 0.00015970045466893104,
      "loss": 0.848,
      "step": 88860
    },
    {
      "epoch": 1.560245088572482,
      "grad_norm": 0.05969738909978531,
      "learning_rate": 0.00015969148087412578,
      "loss": 0.8429,
      "step": 88870
    },
    {
      "epoch": 1.5604206534524834,
      "grad_norm": 0.04783374187444261,
      "learning_rate": 0.00015968250633568821,
      "loss": 0.857,
      "step": 88880
    },
    {
      "epoch": 1.5605962183324849,
      "grad_norm": 0.05374600030585941,
      "learning_rate": 0.0001596735310537321,
      "loss": 0.8426,
      "step": 88890
    },
    {
      "epoch": 1.5607717832124863,
      "grad_norm": 0.05059625071216359,
      "learning_rate": 0.00015966455502837106,
      "loss": 0.8595,
      "step": 88900
    },
    {
      "epoch": 1.5609473480924876,
      "grad_norm": 0.059065205009966426,
      "learning_rate": 0.00015965557825971886,
      "loss": 0.844,
      "step": 88910
    },
    {
      "epoch": 1.5611229129724888,
      "grad_norm": 0.054870494587684636,
      "learning_rate": 0.00015964660074788926,
      "loss": 0.8436,
      "step": 88920
    },
    {
      "epoch": 1.5612984778524903,
      "grad_norm": 0.04675668622036516,
      "learning_rate": 0.00015963762249299597,
      "loss": 0.8498,
      "step": 88930
    },
    {
      "epoch": 1.5614740427324918,
      "grad_norm": 0.052069446860781625,
      "learning_rate": 0.0001596286434951528,
      "loss": 0.8507,
      "step": 88940
    },
    {
      "epoch": 1.5616496076124933,
      "grad_norm": 0.0752571607433736,
      "learning_rate": 0.00015961966375447345,
      "loss": 0.8507,
      "step": 88950
    },
    {
      "epoch": 1.5618251724924948,
      "grad_norm": 0.06489792951138974,
      "learning_rate": 0.00015961068327107175,
      "loss": 0.8514,
      "step": 88960
    },
    {
      "epoch": 1.562000737372496,
      "grad_norm": 0.09207941834341338,
      "learning_rate": 0.00015960170204506143,
      "loss": 0.8466,
      "step": 88970
    },
    {
      "epoch": 1.5621763022524973,
      "grad_norm": 0.06858110588479889,
      "learning_rate": 0.0001595927200765563,
      "loss": 0.8516,
      "step": 88980
    },
    {
      "epoch": 1.5623518671324987,
      "grad_norm": 0.07442538658884262,
      "learning_rate": 0.0001595837373656702,
      "loss": 0.8462,
      "step": 88990
    },
    {
      "epoch": 1.5625274320125002,
      "grad_norm": 0.047590865544009577,
      "learning_rate": 0.00015957475391251694,
      "loss": 0.8532,
      "step": 89000
    },
    {
      "epoch": 1.5627029968925017,
      "grad_norm": 0.04527021988690423,
      "learning_rate": 0.00015956576971721032,
      "loss": 0.849,
      "step": 89010
    },
    {
      "epoch": 1.5628785617725032,
      "grad_norm": 0.05727803923957486,
      "learning_rate": 0.00015955678477986418,
      "loss": 0.848,
      "step": 89020
    },
    {
      "epoch": 1.5630541266525044,
      "grad_norm": 0.07024344774194537,
      "learning_rate": 0.00015954779910059232,
      "loss": 0.847,
      "step": 89030
    },
    {
      "epoch": 1.5632296915325057,
      "grad_norm": 0.05419523783918599,
      "learning_rate": 0.00015953881267950863,
      "loss": 0.8458,
      "step": 89040
    },
    {
      "epoch": 1.5634052564125072,
      "grad_norm": 0.05893936786791296,
      "learning_rate": 0.00015952982551672697,
      "loss": 0.8529,
      "step": 89050
    },
    {
      "epoch": 1.5635808212925086,
      "grad_norm": 0.061816623961772434,
      "learning_rate": 0.00015952083761236122,
      "loss": 0.8474,
      "step": 89060
    },
    {
      "epoch": 1.5637563861725101,
      "grad_norm": 0.047065375943145873,
      "learning_rate": 0.00015951184896652524,
      "loss": 0.8439,
      "step": 89070
    },
    {
      "epoch": 1.5639319510525116,
      "grad_norm": 0.059529795079337505,
      "learning_rate": 0.0001595028595793329,
      "loss": 0.8474,
      "step": 89080
    },
    {
      "epoch": 1.5641075159325128,
      "grad_norm": 0.05539493935948797,
      "learning_rate": 0.00015949386945089813,
      "loss": 0.8422,
      "step": 89090
    },
    {
      "epoch": 1.564283080812514,
      "grad_norm": 0.048387218981702375,
      "learning_rate": 0.0001594848785813348,
      "loss": 0.8483,
      "step": 89100
    },
    {
      "epoch": 1.5644586456925156,
      "grad_norm": 0.05972449231083537,
      "learning_rate": 0.00015947588697075687,
      "loss": 0.8467,
      "step": 89110
    },
    {
      "epoch": 1.564634210572517,
      "grad_norm": 0.08242853930650401,
      "learning_rate": 0.00015946689461927827,
      "loss": 0.8506,
      "step": 89120
    },
    {
      "epoch": 1.5648097754525185,
      "grad_norm": 0.0691499123426812,
      "learning_rate": 0.00015945790152701285,
      "loss": 0.8512,
      "step": 89130
    },
    {
      "epoch": 1.56498534033252,
      "grad_norm": 0.06379789932670288,
      "learning_rate": 0.00015944890769407468,
      "loss": 0.85,
      "step": 89140
    },
    {
      "epoch": 1.5651609052125213,
      "grad_norm": 0.05564746835500519,
      "learning_rate": 0.00015943991312057762,
      "loss": 0.8464,
      "step": 89150
    },
    {
      "epoch": 1.5653364700925227,
      "grad_norm": 0.09475315615288553,
      "learning_rate": 0.00015943091780663563,
      "loss": 0.8497,
      "step": 89160
    },
    {
      "epoch": 1.565512034972524,
      "grad_norm": 0.08547003845546587,
      "learning_rate": 0.00015942192175236275,
      "loss": 0.8434,
      "step": 89170
    },
    {
      "epoch": 1.5656875998525255,
      "grad_norm": 0.06677338418707977,
      "learning_rate": 0.00015941292495787293,
      "loss": 0.849,
      "step": 89180
    },
    {
      "epoch": 1.565863164732527,
      "grad_norm": 0.06306045779049665,
      "learning_rate": 0.00015940392742328015,
      "loss": 0.8485,
      "step": 89190
    },
    {
      "epoch": 1.5660387296125284,
      "grad_norm": 0.05095060983567082,
      "learning_rate": 0.0001593949291486984,
      "loss": 0.8597,
      "step": 89200
    },
    {
      "epoch": 1.5662142944925297,
      "grad_norm": 0.05247095994731161,
      "learning_rate": 0.0001593859301342417,
      "loss": 0.8445,
      "step": 89210
    },
    {
      "epoch": 1.5663898593725312,
      "grad_norm": 0.05761501744974937,
      "learning_rate": 0.0001593769303800241,
      "loss": 0.8476,
      "step": 89220
    },
    {
      "epoch": 1.5665654242525324,
      "grad_norm": 0.07615885218608445,
      "learning_rate": 0.0001593679298861596,
      "loss": 0.8518,
      "step": 89230
    },
    {
      "epoch": 1.566740989132534,
      "grad_norm": 0.04910385453818074,
      "learning_rate": 0.00015935892865276223,
      "loss": 0.8526,
      "step": 89240
    },
    {
      "epoch": 1.5669165540125354,
      "grad_norm": 0.05883131024198244,
      "learning_rate": 0.00015934992667994608,
      "loss": 0.8539,
      "step": 89250
    },
    {
      "epoch": 1.5670921188925369,
      "grad_norm": 0.05811371220460152,
      "learning_rate": 0.00015934092396782516,
      "loss": 0.8498,
      "step": 89260
    },
    {
      "epoch": 1.5672676837725381,
      "grad_norm": 0.07172066859847419,
      "learning_rate": 0.00015933192051651354,
      "loss": 0.8402,
      "step": 89270
    },
    {
      "epoch": 1.5674432486525396,
      "grad_norm": 0.04694687412383926,
      "learning_rate": 0.0001593229163261253,
      "loss": 0.8441,
      "step": 89280
    },
    {
      "epoch": 1.5676188135325408,
      "grad_norm": 0.055858899026244004,
      "learning_rate": 0.00015931391139677455,
      "loss": 0.8447,
      "step": 89290
    },
    {
      "epoch": 1.5677943784125423,
      "grad_norm": 0.04296509976650514,
      "learning_rate": 0.00015930490572857534,
      "loss": 0.8465,
      "step": 89300
    },
    {
      "epoch": 1.5679699432925438,
      "grad_norm": 0.06049585235252407,
      "learning_rate": 0.00015929589932164184,
      "loss": 0.8488,
      "step": 89310
    },
    {
      "epoch": 1.5681455081725453,
      "grad_norm": 0.06827125783006885,
      "learning_rate": 0.00015928689217608813,
      "loss": 0.8499,
      "step": 89320
    },
    {
      "epoch": 1.5683210730525465,
      "grad_norm": 0.060336242316247964,
      "learning_rate": 0.0001592778842920283,
      "loss": 0.844,
      "step": 89330
    },
    {
      "epoch": 1.568496637932548,
      "grad_norm": 0.06377148726042152,
      "learning_rate": 0.0001592688756695765,
      "loss": 0.8518,
      "step": 89340
    },
    {
      "epoch": 1.5686722028125493,
      "grad_norm": 0.07547871726697879,
      "learning_rate": 0.00015925986630884687,
      "loss": 0.8492,
      "step": 89350
    },
    {
      "epoch": 1.5688477676925507,
      "grad_norm": 0.06967033526518446,
      "learning_rate": 0.00015925085620995363,
      "loss": 0.8531,
      "step": 89360
    },
    {
      "epoch": 1.5690233325725522,
      "grad_norm": 0.09818772962951894,
      "learning_rate": 0.0001592418453730108,
      "loss": 0.8557,
      "step": 89370
    },
    {
      "epoch": 1.5691988974525537,
      "grad_norm": 0.05753860856423985,
      "learning_rate": 0.00015923283379813265,
      "loss": 0.8457,
      "step": 89380
    },
    {
      "epoch": 1.569374462332555,
      "grad_norm": 0.07231264138816644,
      "learning_rate": 0.00015922382148543336,
      "loss": 0.8465,
      "step": 89390
    },
    {
      "epoch": 1.5695500272125564,
      "grad_norm": 0.05732115554831139,
      "learning_rate": 0.00015921480843502706,
      "loss": 0.8516,
      "step": 89400
    },
    {
      "epoch": 1.5697255920925577,
      "grad_norm": 0.08177058162724586,
      "learning_rate": 0.000159205794647028,
      "loss": 0.8497,
      "step": 89410
    },
    {
      "epoch": 1.5699011569725592,
      "grad_norm": 0.058071733135123246,
      "learning_rate": 0.0001591967801215504,
      "loss": 0.8424,
      "step": 89420
    },
    {
      "epoch": 1.5700767218525606,
      "grad_norm": 0.08192167399127567,
      "learning_rate": 0.0001591877648587084,
      "loss": 0.8479,
      "step": 89430
    },
    {
      "epoch": 1.5702522867325621,
      "grad_norm": 0.08388286292140544,
      "learning_rate": 0.00015917874885861626,
      "loss": 0.8501,
      "step": 89440
    },
    {
      "epoch": 1.5704278516125634,
      "grad_norm": 0.07889809089431103,
      "learning_rate": 0.00015916973212138825,
      "loss": 0.8483,
      "step": 89450
    },
    {
      "epoch": 1.5706034164925649,
      "grad_norm": 0.050777020569458375,
      "learning_rate": 0.00015916071464713857,
      "loss": 0.8538,
      "step": 89460
    },
    {
      "epoch": 1.570778981372566,
      "grad_norm": 0.06529639602846891,
      "learning_rate": 0.00015915169643598148,
      "loss": 0.8427,
      "step": 89470
    },
    {
      "epoch": 1.5709545462525676,
      "grad_norm": 0.08184648453335215,
      "learning_rate": 0.00015914267748803125,
      "loss": 0.8537,
      "step": 89480
    },
    {
      "epoch": 1.571130111132569,
      "grad_norm": 0.05959528169391161,
      "learning_rate": 0.0001591336578034022,
      "loss": 0.8445,
      "step": 89490
    },
    {
      "epoch": 1.5713056760125705,
      "grad_norm": 0.04853878681723957,
      "learning_rate": 0.0001591246373822085,
      "loss": 0.8429,
      "step": 89500
    },
    {
      "epoch": 1.5714812408925718,
      "grad_norm": 0.06152299262628816,
      "learning_rate": 0.00015911561622456458,
      "loss": 0.8494,
      "step": 89510
    },
    {
      "epoch": 1.5716568057725733,
      "grad_norm": 0.05247759037694654,
      "learning_rate": 0.00015910659433058462,
      "loss": 0.8495,
      "step": 89520
    },
    {
      "epoch": 1.5718323706525745,
      "grad_norm": 0.05628279628958935,
      "learning_rate": 0.00015909757170038297,
      "loss": 0.8478,
      "step": 89530
    },
    {
      "epoch": 1.572007935532576,
      "grad_norm": 0.04531924732510581,
      "learning_rate": 0.00015908854833407394,
      "loss": 0.8597,
      "step": 89540
    },
    {
      "epoch": 1.5721835004125775,
      "grad_norm": 0.08984047390467315,
      "learning_rate": 0.00015907952423177188,
      "loss": 0.849,
      "step": 89550
    },
    {
      "epoch": 1.572359065292579,
      "grad_norm": 0.06390749096286903,
      "learning_rate": 0.00015907049939359116,
      "loss": 0.8351,
      "step": 89560
    },
    {
      "epoch": 1.5725346301725804,
      "grad_norm": 0.11485183125526226,
      "learning_rate": 0.00015906147381964604,
      "loss": 0.8496,
      "step": 89570
    },
    {
      "epoch": 1.5727101950525817,
      "grad_norm": 0.04914515352020735,
      "learning_rate": 0.00015905244751005093,
      "loss": 0.8442,
      "step": 89580
    },
    {
      "epoch": 1.572885759932583,
      "grad_norm": 0.053491241413736325,
      "learning_rate": 0.00015904342046492017,
      "loss": 0.8492,
      "step": 89590
    },
    {
      "epoch": 1.5730613248125844,
      "grad_norm": 0.08577397729741201,
      "learning_rate": 0.00015903439268436816,
      "loss": 0.8486,
      "step": 89600
    },
    {
      "epoch": 1.573236889692586,
      "grad_norm": 0.06977827278470844,
      "learning_rate": 0.0001590253641685093,
      "loss": 0.8612,
      "step": 89610
    },
    {
      "epoch": 1.5734124545725874,
      "grad_norm": 0.047024881107853266,
      "learning_rate": 0.0001590163349174579,
      "loss": 0.8557,
      "step": 89620
    },
    {
      "epoch": 1.5735880194525889,
      "grad_norm": 0.05163372702844441,
      "learning_rate": 0.00015900730493132847,
      "loss": 0.8457,
      "step": 89630
    },
    {
      "epoch": 1.5737635843325901,
      "grad_norm": 0.05126481068248318,
      "learning_rate": 0.00015899827421023534,
      "loss": 0.8478,
      "step": 89640
    },
    {
      "epoch": 1.5739391492125914,
      "grad_norm": 0.11860950350092861,
      "learning_rate": 0.00015898924275429296,
      "loss": 0.8385,
      "step": 89650
    },
    {
      "epoch": 1.5741147140925928,
      "grad_norm": 0.06080400724719035,
      "learning_rate": 0.00015898021056361578,
      "loss": 0.8465,
      "step": 89660
    },
    {
      "epoch": 1.5742902789725943,
      "grad_norm": 0.05252602091441076,
      "learning_rate": 0.0001589711776383182,
      "loss": 0.8539,
      "step": 89670
    },
    {
      "epoch": 1.5744658438525958,
      "grad_norm": 0.055558391321736136,
      "learning_rate": 0.0001589621439785147,
      "loss": 0.8344,
      "step": 89680
    },
    {
      "epoch": 1.5746414087325973,
      "grad_norm": 0.048554017455228554,
      "learning_rate": 0.0001589531095843197,
      "loss": 0.8643,
      "step": 89690
    },
    {
      "epoch": 1.5748169736125985,
      "grad_norm": 0.07786323951809927,
      "learning_rate": 0.00015894407445584777,
      "loss": 0.8608,
      "step": 89700
    },
    {
      "epoch": 1.5749925384925998,
      "grad_norm": 0.052666706499331396,
      "learning_rate": 0.0001589350385932132,
      "loss": 0.8483,
      "step": 89710
    },
    {
      "epoch": 1.5751681033726013,
      "grad_norm": 0.060375617508938,
      "learning_rate": 0.00015892600199653067,
      "loss": 0.8514,
      "step": 89720
    },
    {
      "epoch": 1.5753436682526027,
      "grad_norm": 0.07132466310747458,
      "learning_rate": 0.00015891696466591458,
      "loss": 0.8456,
      "step": 89730
    },
    {
      "epoch": 1.5755192331326042,
      "grad_norm": 0.07568586055292299,
      "learning_rate": 0.00015890792660147945,
      "loss": 0.8564,
      "step": 89740
    },
    {
      "epoch": 1.5756947980126057,
      "grad_norm": 0.05961020734504238,
      "learning_rate": 0.00015889888780333978,
      "loss": 0.844,
      "step": 89750
    },
    {
      "epoch": 1.575870362892607,
      "grad_norm": 0.049964481026431744,
      "learning_rate": 0.0001588898482716101,
      "loss": 0.8391,
      "step": 89760
    },
    {
      "epoch": 1.5760459277726082,
      "grad_norm": 0.05407445579332524,
      "learning_rate": 0.00015888080800640497,
      "loss": 0.8507,
      "step": 89770
    },
    {
      "epoch": 1.5762214926526097,
      "grad_norm": 0.06506645932523047,
      "learning_rate": 0.00015887176700783888,
      "loss": 0.8466,
      "step": 89780
    },
    {
      "epoch": 1.5763970575326112,
      "grad_norm": 0.05758450640730029,
      "learning_rate": 0.00015886272527602642,
      "loss": 0.8512,
      "step": 89790
    },
    {
      "epoch": 1.5765726224126126,
      "grad_norm": 0.052506611765551124,
      "learning_rate": 0.00015885368281108215,
      "loss": 0.8523,
      "step": 89800
    },
    {
      "epoch": 1.5767481872926141,
      "grad_norm": 0.061253303751228574,
      "learning_rate": 0.00015884463961312063,
      "loss": 0.8407,
      "step": 89810
    },
    {
      "epoch": 1.5769237521726154,
      "grad_norm": 0.052101287067195114,
      "learning_rate": 0.00015883559568225644,
      "loss": 0.8495,
      "step": 89820
    },
    {
      "epoch": 1.5770993170526166,
      "grad_norm": 0.04721081385674925,
      "learning_rate": 0.00015882655101860416,
      "loss": 0.8491,
      "step": 89830
    },
    {
      "epoch": 1.5772748819326181,
      "grad_norm": 0.04598883507952143,
      "learning_rate": 0.0001588175056222784,
      "loss": 0.845,
      "step": 89840
    },
    {
      "epoch": 1.5774504468126196,
      "grad_norm": 0.0523946461932713,
      "learning_rate": 0.00015880845949339377,
      "loss": 0.8433,
      "step": 89850
    },
    {
      "epoch": 1.577626011692621,
      "grad_norm": 0.07913755532499649,
      "learning_rate": 0.00015879941263206488,
      "loss": 0.8518,
      "step": 89860
    },
    {
      "epoch": 1.5778015765726225,
      "grad_norm": 0.057784471720608965,
      "learning_rate": 0.00015879036503840634,
      "loss": 0.8454,
      "step": 89870
    },
    {
      "epoch": 1.5779771414526238,
      "grad_norm": 0.0498389468070698,
      "learning_rate": 0.00015878131671253284,
      "loss": 0.8469,
      "step": 89880
    },
    {
      "epoch": 1.5781527063326253,
      "grad_norm": 0.09276174894668411,
      "learning_rate": 0.00015877226765455892,
      "loss": 0.8475,
      "step": 89890
    },
    {
      "epoch": 1.5783282712126265,
      "grad_norm": 0.0777544995685139,
      "learning_rate": 0.00015876321786459932,
      "loss": 0.8507,
      "step": 89900
    },
    {
      "epoch": 1.578503836092628,
      "grad_norm": 0.0710354551816941,
      "learning_rate": 0.00015875416734276873,
      "loss": 0.8457,
      "step": 89910
    },
    {
      "epoch": 1.5786794009726295,
      "grad_norm": 0.06011961498007995,
      "learning_rate": 0.00015874511608918174,
      "loss": 0.8512,
      "step": 89920
    },
    {
      "epoch": 1.578854965852631,
      "grad_norm": 0.0519436394024867,
      "learning_rate": 0.00015873606410395303,
      "loss": 0.8412,
      "step": 89930
    },
    {
      "epoch": 1.5790305307326322,
      "grad_norm": 0.07501284688683946,
      "learning_rate": 0.00015872701138719736,
      "loss": 0.8539,
      "step": 89940
    },
    {
      "epoch": 1.5792060956126337,
      "grad_norm": 0.08460171056173171,
      "learning_rate": 0.00015871795793902935,
      "loss": 0.8504,
      "step": 89950
    },
    {
      "epoch": 1.579381660492635,
      "grad_norm": 0.06234735586767833,
      "learning_rate": 0.0001587089037595638,
      "loss": 0.8506,
      "step": 89960
    },
    {
      "epoch": 1.5795572253726364,
      "grad_norm": 0.08231478675029717,
      "learning_rate": 0.00015869984884891536,
      "loss": 0.8435,
      "step": 89970
    },
    {
      "epoch": 1.579732790252638,
      "grad_norm": 0.04497837180579472,
      "learning_rate": 0.00015869079320719875,
      "loss": 0.8516,
      "step": 89980
    },
    {
      "epoch": 1.5799083551326394,
      "grad_norm": 0.05788115214336489,
      "learning_rate": 0.00015868173683452878,
      "loss": 0.8416,
      "step": 89990
    },
    {
      "epoch": 1.5800839200126406,
      "grad_norm": 0.053367191804467425,
      "learning_rate": 0.00015867267973102014,
      "loss": 0.8475,
      "step": 90000
    },
    {
      "epoch": 1.5802594848926421,
      "grad_norm": 0.0817991724129242,
      "learning_rate": 0.00015866362189678756,
      "loss": 0.8437,
      "step": 90010
    },
    {
      "epoch": 1.5804350497726434,
      "grad_norm": 0.0592551582884363,
      "learning_rate": 0.0001586545633319459,
      "loss": 0.8413,
      "step": 90020
    },
    {
      "epoch": 1.5806106146526449,
      "grad_norm": 0.0734989714498931,
      "learning_rate": 0.0001586455040366098,
      "loss": 0.858,
      "step": 90030
    },
    {
      "epoch": 1.5807861795326463,
      "grad_norm": 0.04987343159925907,
      "learning_rate": 0.00015863644401089414,
      "loss": 0.8474,
      "step": 90040
    },
    {
      "epoch": 1.5809617444126478,
      "grad_norm": 0.07396840234814855,
      "learning_rate": 0.00015862738325491372,
      "loss": 0.8472,
      "step": 90050
    },
    {
      "epoch": 1.581137309292649,
      "grad_norm": 0.044668026838924145,
      "learning_rate": 0.0001586183217687833,
      "loss": 0.8492,
      "step": 90060
    },
    {
      "epoch": 1.5813128741726505,
      "grad_norm": 0.06952996641361733,
      "learning_rate": 0.0001586092595526177,
      "loss": 0.8398,
      "step": 90070
    },
    {
      "epoch": 1.5814884390526518,
      "grad_norm": 0.08469776901996329,
      "learning_rate": 0.0001586001966065317,
      "loss": 0.8413,
      "step": 90080
    },
    {
      "epoch": 1.5816640039326533,
      "grad_norm": 0.05476052027144533,
      "learning_rate": 0.00015859113293064024,
      "loss": 0.8417,
      "step": 90090
    },
    {
      "epoch": 1.5818395688126548,
      "grad_norm": 0.06509463728255599,
      "learning_rate": 0.00015858206852505802,
      "loss": 0.8495,
      "step": 90100
    },
    {
      "epoch": 1.5820151336926562,
      "grad_norm": 0.08604584871852632,
      "learning_rate": 0.00015857300338990003,
      "loss": 0.847,
      "step": 90110
    },
    {
      "epoch": 1.5821906985726575,
      "grad_norm": 0.06739688704903073,
      "learning_rate": 0.000158563937525281,
      "loss": 0.8461,
      "step": 90120
    },
    {
      "epoch": 1.582366263452659,
      "grad_norm": 0.09777909220835573,
      "learning_rate": 0.00015855487093131588,
      "loss": 0.8481,
      "step": 90130
    },
    {
      "epoch": 1.5825418283326602,
      "grad_norm": 0.05025014392632189,
      "learning_rate": 0.00015854580360811948,
      "loss": 0.8436,
      "step": 90140
    },
    {
      "epoch": 1.5827173932126617,
      "grad_norm": 0.0729243499720481,
      "learning_rate": 0.0001585367355558067,
      "loss": 0.8455,
      "step": 90150
    },
    {
      "epoch": 1.5828929580926632,
      "grad_norm": 0.0482957020443582,
      "learning_rate": 0.00015852766677449252,
      "loss": 0.8526,
      "step": 90160
    },
    {
      "epoch": 1.5830685229726646,
      "grad_norm": 0.054153812546168946,
      "learning_rate": 0.00015851859726429172,
      "loss": 0.8479,
      "step": 90170
    },
    {
      "epoch": 1.583244087852666,
      "grad_norm": 0.08461447668551615,
      "learning_rate": 0.00015850952702531928,
      "loss": 0.8451,
      "step": 90180
    },
    {
      "epoch": 1.5834196527326674,
      "grad_norm": 0.07154162677638357,
      "learning_rate": 0.00015850045605769015,
      "loss": 0.8487,
      "step": 90190
    },
    {
      "epoch": 1.5835952176126686,
      "grad_norm": 0.07594048337357331,
      "learning_rate": 0.0001584913843615192,
      "loss": 0.849,
      "step": 90200
    },
    {
      "epoch": 1.5837707824926701,
      "grad_norm": 0.06999016281372322,
      "learning_rate": 0.0001584823119369214,
      "loss": 0.8568,
      "step": 90210
    },
    {
      "epoch": 1.5839463473726716,
      "grad_norm": 0.050394061084736666,
      "learning_rate": 0.00015847323878401167,
      "loss": 0.849,
      "step": 90220
    },
    {
      "epoch": 1.584121912252673,
      "grad_norm": 0.060138169105124595,
      "learning_rate": 0.000158464164902905,
      "loss": 0.8462,
      "step": 90230
    },
    {
      "epoch": 1.5842974771326743,
      "grad_norm": 0.07027868292364067,
      "learning_rate": 0.00015845509029371638,
      "loss": 0.8513,
      "step": 90240
    },
    {
      "epoch": 1.5844730420126758,
      "grad_norm": 0.04541132588069494,
      "learning_rate": 0.0001584460149565607,
      "loss": 0.8526,
      "step": 90250
    },
    {
      "epoch": 1.584648606892677,
      "grad_norm": 0.06084610774428921,
      "learning_rate": 0.00015843693889155302,
      "loss": 0.8468,
      "step": 90260
    },
    {
      "epoch": 1.5848241717726785,
      "grad_norm": 0.06127498141467736,
      "learning_rate": 0.00015842786209880832,
      "loss": 0.8508,
      "step": 90270
    },
    {
      "epoch": 1.58499973665268,
      "grad_norm": 0.058022499459925124,
      "learning_rate": 0.00015841878457844162,
      "loss": 0.8481,
      "step": 90280
    },
    {
      "epoch": 1.5851753015326815,
      "grad_norm": 0.041820053314453244,
      "learning_rate": 0.00015840970633056788,
      "loss": 0.848,
      "step": 90290
    },
    {
      "epoch": 1.585350866412683,
      "grad_norm": 0.05442291501709164,
      "learning_rate": 0.00015840062735530218,
      "loss": 0.8456,
      "step": 90300
    },
    {
      "epoch": 1.5855264312926842,
      "grad_norm": 0.05165539523610851,
      "learning_rate": 0.0001583915476527595,
      "loss": 0.8428,
      "step": 90310
    },
    {
      "epoch": 1.5857019961726855,
      "grad_norm": 0.0636537012941317,
      "learning_rate": 0.00015838246722305497,
      "loss": 0.8536,
      "step": 90320
    },
    {
      "epoch": 1.585877561052687,
      "grad_norm": 0.05476275584323656,
      "learning_rate": 0.00015837338606630353,
      "loss": 0.8473,
      "step": 90330
    },
    {
      "epoch": 1.5860531259326884,
      "grad_norm": 0.06901978495077143,
      "learning_rate": 0.0001583643041826203,
      "loss": 0.858,
      "step": 90340
    },
    {
      "epoch": 1.58622869081269,
      "grad_norm": 0.05222246542501554,
      "learning_rate": 0.00015835522157212035,
      "loss": 0.8578,
      "step": 90350
    },
    {
      "epoch": 1.5864042556926914,
      "grad_norm": 0.06391788906812292,
      "learning_rate": 0.00015834613823491878,
      "loss": 0.8503,
      "step": 90360
    },
    {
      "epoch": 1.5865798205726926,
      "grad_norm": 0.06548491471328449,
      "learning_rate": 0.00015833705417113062,
      "loss": 0.8554,
      "step": 90370
    },
    {
      "epoch": 1.586755385452694,
      "grad_norm": 0.051357414083936376,
      "learning_rate": 0.00015832796938087103,
      "loss": 0.8527,
      "step": 90380
    },
    {
      "epoch": 1.5869309503326954,
      "grad_norm": 0.05231001878214001,
      "learning_rate": 0.00015831888386425504,
      "loss": 0.8503,
      "step": 90390
    },
    {
      "epoch": 1.5871065152126969,
      "grad_norm": 0.053460475763059005,
      "learning_rate": 0.00015830979762139778,
      "loss": 0.8486,
      "step": 90400
    },
    {
      "epoch": 1.5872820800926983,
      "grad_norm": 0.044879595600270016,
      "learning_rate": 0.00015830071065241446,
      "loss": 0.8373,
      "step": 90410
    },
    {
      "epoch": 1.5874576449726998,
      "grad_norm": 0.07645300618312376,
      "learning_rate": 0.00015829162295742015,
      "loss": 0.8407,
      "step": 90420
    },
    {
      "epoch": 1.587633209852701,
      "grad_norm": 0.04259223793089963,
      "learning_rate": 0.00015828253453653,
      "loss": 0.8498,
      "step": 90430
    },
    {
      "epoch": 1.5878087747327023,
      "grad_norm": 0.0609911649909262,
      "learning_rate": 0.00015827344538985912,
      "loss": 0.8511,
      "step": 90440
    },
    {
      "epoch": 1.5879843396127038,
      "grad_norm": 0.07819172725809735,
      "learning_rate": 0.00015826435551752273,
      "loss": 0.8415,
      "step": 90450
    },
    {
      "epoch": 1.5881599044927053,
      "grad_norm": 0.058382885605504156,
      "learning_rate": 0.00015825526491963597,
      "loss": 0.8425,
      "step": 90460
    },
    {
      "epoch": 1.5883354693727068,
      "grad_norm": 0.052316011791255106,
      "learning_rate": 0.00015824617359631402,
      "loss": 0.8493,
      "step": 90470
    },
    {
      "epoch": 1.5885110342527082,
      "grad_norm": 0.07705183126547047,
      "learning_rate": 0.0001582370815476721,
      "loss": 0.851,
      "step": 90480
    },
    {
      "epoch": 1.5886865991327095,
      "grad_norm": 0.06972954839739358,
      "learning_rate": 0.0001582279887738254,
      "loss": 0.8548,
      "step": 90490
    },
    {
      "epoch": 1.5888621640127107,
      "grad_norm": 0.08106907011150026,
      "learning_rate": 0.0001582188952748891,
      "loss": 0.8479,
      "step": 90500
    },
    {
      "epoch": 1.5890377288927122,
      "grad_norm": 0.05155107819420603,
      "learning_rate": 0.00015820980105097845,
      "loss": 0.8603,
      "step": 90510
    },
    {
      "epoch": 1.5892132937727137,
      "grad_norm": 0.051467485906838036,
      "learning_rate": 0.0001582007061022086,
      "loss": 0.85,
      "step": 90520
    },
    {
      "epoch": 1.5893888586527152,
      "grad_norm": 0.05002601541806635,
      "learning_rate": 0.00015819161042869491,
      "loss": 0.8492,
      "step": 90530
    },
    {
      "epoch": 1.5895644235327167,
      "grad_norm": 0.07646383146857179,
      "learning_rate": 0.00015818251403055254,
      "loss": 0.8531,
      "step": 90540
    },
    {
      "epoch": 1.589739988412718,
      "grad_norm": 0.07916278862989352,
      "learning_rate": 0.00015817341690789672,
      "loss": 0.8509,
      "step": 90550
    },
    {
      "epoch": 1.5899155532927192,
      "grad_norm": 0.0685491624118804,
      "learning_rate": 0.00015816431906084277,
      "loss": 0.8508,
      "step": 90560
    },
    {
      "epoch": 1.5900911181727206,
      "grad_norm": 0.05355526453782794,
      "learning_rate": 0.00015815522048950596,
      "loss": 0.8481,
      "step": 90570
    },
    {
      "epoch": 1.5902666830527221,
      "grad_norm": 0.0517058665443025,
      "learning_rate": 0.00015814612119400154,
      "loss": 0.8491,
      "step": 90580
    },
    {
      "epoch": 1.5904422479327236,
      "grad_norm": 0.05248230055126385,
      "learning_rate": 0.00015813702117444483,
      "loss": 0.8346,
      "step": 90590
    },
    {
      "epoch": 1.590617812812725,
      "grad_norm": 0.05603836946452892,
      "learning_rate": 0.00015812792043095111,
      "loss": 0.8497,
      "step": 90600
    },
    {
      "epoch": 1.5907933776927263,
      "grad_norm": 0.07966594058521918,
      "learning_rate": 0.00015811881896363563,
      "loss": 0.8551,
      "step": 90610
    },
    {
      "epoch": 1.5909689425727278,
      "grad_norm": 0.05456411351032947,
      "learning_rate": 0.00015810971677261384,
      "loss": 0.856,
      "step": 90620
    },
    {
      "epoch": 1.591144507452729,
      "grad_norm": 0.06167278567269861,
      "learning_rate": 0.0001581006138580009,
      "loss": 0.8545,
      "step": 90630
    },
    {
      "epoch": 1.5913200723327305,
      "grad_norm": 0.051146027285322485,
      "learning_rate": 0.00015809151021991234,
      "loss": 0.8506,
      "step": 90640
    },
    {
      "epoch": 1.591495637212732,
      "grad_norm": 0.061096621395542554,
      "learning_rate": 0.00015808240585846338,
      "loss": 0.8477,
      "step": 90650
    },
    {
      "epoch": 1.5916712020927335,
      "grad_norm": 0.05167251227375376,
      "learning_rate": 0.00015807330077376936,
      "loss": 0.8392,
      "step": 90660
    },
    {
      "epoch": 1.5918467669727348,
      "grad_norm": 0.06989966893018146,
      "learning_rate": 0.00015806419496594573,
      "loss": 0.8486,
      "step": 90670
    },
    {
      "epoch": 1.5920223318527362,
      "grad_norm": 0.05048650369548772,
      "learning_rate": 0.00015805508843510778,
      "loss": 0.8423,
      "step": 90680
    },
    {
      "epoch": 1.5921978967327375,
      "grad_norm": 0.05847404539567389,
      "learning_rate": 0.00015804598118137093,
      "loss": 0.8514,
      "step": 90690
    },
    {
      "epoch": 1.592373461612739,
      "grad_norm": 0.05732545183119213,
      "learning_rate": 0.0001580368732048506,
      "loss": 0.8448,
      "step": 90700
    },
    {
      "epoch": 1.5925490264927404,
      "grad_norm": 0.08368328745926819,
      "learning_rate": 0.00015802776450566207,
      "loss": 0.8486,
      "step": 90710
    },
    {
      "epoch": 1.592724591372742,
      "grad_norm": 0.05724524333472676,
      "learning_rate": 0.00015801865508392092,
      "loss": 0.8438,
      "step": 90720
    },
    {
      "epoch": 1.5929001562527432,
      "grad_norm": 0.051032603445990345,
      "learning_rate": 0.00015800954493974247,
      "loss": 0.8526,
      "step": 90730
    },
    {
      "epoch": 1.5930757211327446,
      "grad_norm": 0.05779852063069735,
      "learning_rate": 0.00015800043407324213,
      "loss": 0.849,
      "step": 90740
    },
    {
      "epoch": 1.593251286012746,
      "grad_norm": 0.06966093090000702,
      "learning_rate": 0.00015799132248453537,
      "loss": 0.846,
      "step": 90750
    },
    {
      "epoch": 1.5934268508927474,
      "grad_norm": 0.0637014725410287,
      "learning_rate": 0.00015798221017373763,
      "loss": 0.8439,
      "step": 90760
    },
    {
      "epoch": 1.5936024157727489,
      "grad_norm": 0.05897383169779125,
      "learning_rate": 0.0001579730971409644,
      "loss": 0.8459,
      "step": 90770
    },
    {
      "epoch": 1.5937779806527503,
      "grad_norm": 0.06346822325337578,
      "learning_rate": 0.00015796398338633108,
      "loss": 0.8579,
      "step": 90780
    },
    {
      "epoch": 1.5939535455327516,
      "grad_norm": 0.06190303178759484,
      "learning_rate": 0.00015795486890995318,
      "loss": 0.851,
      "step": 90790
    },
    {
      "epoch": 1.594129110412753,
      "grad_norm": 0.05339750947779269,
      "learning_rate": 0.00015794575371194617,
      "loss": 0.8509,
      "step": 90800
    },
    {
      "epoch": 1.5943046752927543,
      "grad_norm": 0.0622686181858218,
      "learning_rate": 0.00015793663779242552,
      "loss": 0.8503,
      "step": 90810
    },
    {
      "epoch": 1.5944802401727558,
      "grad_norm": 0.05135955920579832,
      "learning_rate": 0.00015792752115150683,
      "loss": 0.8417,
      "step": 90820
    },
    {
      "epoch": 1.5946558050527573,
      "grad_norm": 0.060281370214628725,
      "learning_rate": 0.0001579184037893055,
      "loss": 0.8471,
      "step": 90830
    },
    {
      "epoch": 1.5948313699327588,
      "grad_norm": 0.06177625638115615,
      "learning_rate": 0.0001579092857059371,
      "loss": 0.8488,
      "step": 90840
    },
    {
      "epoch": 1.59500693481276,
      "grad_norm": 0.06402918011686869,
      "learning_rate": 0.0001579001669015171,
      "loss": 0.8467,
      "step": 90850
    },
    {
      "epoch": 1.5951824996927615,
      "grad_norm": 0.07926249988595808,
      "learning_rate": 0.00015789104737616108,
      "loss": 0.8411,
      "step": 90860
    },
    {
      "epoch": 1.5953580645727627,
      "grad_norm": 0.05404522707959022,
      "learning_rate": 0.00015788192712998467,
      "loss": 0.8512,
      "step": 90870
    },
    {
      "epoch": 1.5955336294527642,
      "grad_norm": 0.06804284931719541,
      "learning_rate": 0.00015787280616310327,
      "loss": 0.8461,
      "step": 90880
    },
    {
      "epoch": 1.5957091943327657,
      "grad_norm": 0.04455488311985065,
      "learning_rate": 0.0001578636844756325,
      "loss": 0.8558,
      "step": 90890
    },
    {
      "epoch": 1.5958847592127672,
      "grad_norm": 0.05747085437095737,
      "learning_rate": 0.00015785456206768802,
      "loss": 0.856,
      "step": 90900
    },
    {
      "epoch": 1.5960603240927684,
      "grad_norm": 0.07255860330326162,
      "learning_rate": 0.00015784543893938526,
      "loss": 0.8441,
      "step": 90910
    },
    {
      "epoch": 1.59623588897277,
      "grad_norm": 0.0582782475689912,
      "learning_rate": 0.00015783631509083996,
      "loss": 0.8515,
      "step": 90920
    },
    {
      "epoch": 1.5964114538527712,
      "grad_norm": 0.07167422844264902,
      "learning_rate": 0.00015782719052216763,
      "loss": 0.8474,
      "step": 90930
    },
    {
      "epoch": 1.5965870187327726,
      "grad_norm": 0.07363454923855306,
      "learning_rate": 0.00015781806523348392,
      "loss": 0.8499,
      "step": 90940
    },
    {
      "epoch": 1.5967625836127741,
      "grad_norm": 0.0694143857759646,
      "learning_rate": 0.0001578089392249044,
      "loss": 0.8544,
      "step": 90950
    },
    {
      "epoch": 1.5969381484927756,
      "grad_norm": 0.04751511555202065,
      "learning_rate": 0.0001577998124965448,
      "loss": 0.8439,
      "step": 90960
    },
    {
      "epoch": 1.5971137133727769,
      "grad_norm": 0.06452048517569504,
      "learning_rate": 0.00015779068504852061,
      "loss": 0.843,
      "step": 90970
    },
    {
      "epoch": 1.5972892782527783,
      "grad_norm": 0.0677086073498115,
      "learning_rate": 0.0001577815568809476,
      "loss": 0.8557,
      "step": 90980
    },
    {
      "epoch": 1.5974648431327796,
      "grad_norm": 0.0947987057938556,
      "learning_rate": 0.00015777242799394137,
      "loss": 0.8424,
      "step": 90990
    },
    {
      "epoch": 1.597640408012781,
      "grad_norm": 0.060467064045625164,
      "learning_rate": 0.00015776329838761762,
      "loss": 0.8485,
      "step": 91000
    },
    {
      "epoch": 1.5978159728927825,
      "grad_norm": 0.05240289682728228,
      "learning_rate": 0.000157754168062092,
      "loss": 0.8554,
      "step": 91010
    },
    {
      "epoch": 1.597991537772784,
      "grad_norm": 0.06183856245702646,
      "learning_rate": 0.00015774503701748013,
      "loss": 0.8509,
      "step": 91020
    },
    {
      "epoch": 1.5981671026527855,
      "grad_norm": 0.057817412017159266,
      "learning_rate": 0.0001577359052538978,
      "loss": 0.8439,
      "step": 91030
    },
    {
      "epoch": 1.5983426675327868,
      "grad_norm": 0.07869308131096858,
      "learning_rate": 0.00015772677277146074,
      "loss": 0.8538,
      "step": 91040
    },
    {
      "epoch": 1.598518232412788,
      "grad_norm": 0.06734035881065863,
      "learning_rate": 0.0001577176395702845,
      "loss": 0.8497,
      "step": 91050
    },
    {
      "epoch": 1.5986937972927895,
      "grad_norm": 0.046347515199865214,
      "learning_rate": 0.00015770850565048497,
      "loss": 0.8543,
      "step": 91060
    },
    {
      "epoch": 1.598869362172791,
      "grad_norm": 0.06802572241749714,
      "learning_rate": 0.00015769937101217776,
      "loss": 0.8474,
      "step": 91070
    },
    {
      "epoch": 1.5990449270527924,
      "grad_norm": 0.05088480739225804,
      "learning_rate": 0.00015769023565547868,
      "loss": 0.8522,
      "step": 91080
    },
    {
      "epoch": 1.599220491932794,
      "grad_norm": 0.05682588099678468,
      "learning_rate": 0.00015768109958050344,
      "loss": 0.8509,
      "step": 91090
    },
    {
      "epoch": 1.5993960568127952,
      "grad_norm": 0.07077703487044108,
      "learning_rate": 0.0001576719627873678,
      "loss": 0.849,
      "step": 91100
    },
    {
      "epoch": 1.5995716216927964,
      "grad_norm": 0.04634585797972743,
      "learning_rate": 0.0001576628252761875,
      "loss": 0.848,
      "step": 91110
    },
    {
      "epoch": 1.599747186572798,
      "grad_norm": 0.06994160622375067,
      "learning_rate": 0.0001576536870470784,
      "loss": 0.8537,
      "step": 91120
    },
    {
      "epoch": 1.5999227514527994,
      "grad_norm": 0.05000514765939688,
      "learning_rate": 0.00015764454810015624,
      "loss": 0.843,
      "step": 91130
    },
    {
      "epoch": 1.6000983163328009,
      "grad_norm": 0.06971450206658995,
      "learning_rate": 0.00015763540843553674,
      "loss": 0.8522,
      "step": 91140
    },
    {
      "epoch": 1.6002738812128023,
      "grad_norm": 0.055795072894269146,
      "learning_rate": 0.0001576262680533358,
      "loss": 0.8531,
      "step": 91150
    },
    {
      "epoch": 1.6004494460928036,
      "grad_norm": 0.053088558520548085,
      "learning_rate": 0.00015761712695366918,
      "loss": 0.8439,
      "step": 91160
    },
    {
      "epoch": 1.6006250109728049,
      "grad_norm": 0.05264633628137883,
      "learning_rate": 0.0001576079851366527,
      "loss": 0.85,
      "step": 91170
    },
    {
      "epoch": 1.6008005758528063,
      "grad_norm": 0.04918563515554702,
      "learning_rate": 0.00015759884260240222,
      "loss": 0.8369,
      "step": 91180
    },
    {
      "epoch": 1.6009761407328078,
      "grad_norm": 0.09549794202954581,
      "learning_rate": 0.00015758969935103357,
      "loss": 0.8474,
      "step": 91190
    },
    {
      "epoch": 1.6011517056128093,
      "grad_norm": 0.08867892240997448,
      "learning_rate": 0.0001575805553826625,
      "loss": 0.8573,
      "step": 91200
    },
    {
      "epoch": 1.6013272704928108,
      "grad_norm": 0.06349763283582022,
      "learning_rate": 0.00015757141069740505,
      "loss": 0.853,
      "step": 91210
    },
    {
      "epoch": 1.601502835372812,
      "grad_norm": 0.06303406060857004,
      "learning_rate": 0.00015756226529537696,
      "loss": 0.8535,
      "step": 91220
    },
    {
      "epoch": 1.6016784002528133,
      "grad_norm": 0.04368385586802328,
      "learning_rate": 0.00015755311917669412,
      "loss": 0.8448,
      "step": 91230
    },
    {
      "epoch": 1.6018539651328147,
      "grad_norm": 0.06494193775489887,
      "learning_rate": 0.0001575439723414724,
      "loss": 0.85,
      "step": 91240
    },
    {
      "epoch": 1.6020295300128162,
      "grad_norm": 0.07372031348155193,
      "learning_rate": 0.00015753482478982772,
      "loss": 0.8377,
      "step": 91250
    },
    {
      "epoch": 1.6022050948928177,
      "grad_norm": 0.04910391632707975,
      "learning_rate": 0.000157525676521876,
      "loss": 0.8403,
      "step": 91260
    },
    {
      "epoch": 1.6023806597728192,
      "grad_norm": 0.07275912593923371,
      "learning_rate": 0.0001575165275377331,
      "loss": 0.844,
      "step": 91270
    },
    {
      "epoch": 1.6025562246528204,
      "grad_norm": 0.07815117869557738,
      "learning_rate": 0.00015750737783751497,
      "loss": 0.844,
      "step": 91280
    },
    {
      "epoch": 1.6027317895328217,
      "grad_norm": 0.045732170865893214,
      "learning_rate": 0.00015749822742133754,
      "loss": 0.8458,
      "step": 91290
    },
    {
      "epoch": 1.6029073544128232,
      "grad_norm": 0.05627871712950593,
      "learning_rate": 0.00015748907628931674,
      "loss": 0.8455,
      "step": 91300
    },
    {
      "epoch": 1.6030829192928246,
      "grad_norm": 0.09796676722758486,
      "learning_rate": 0.00015747992444156852,
      "loss": 0.848,
      "step": 91310
    },
    {
      "epoch": 1.6032584841728261,
      "grad_norm": 0.05365628939681921,
      "learning_rate": 0.00015747077187820883,
      "loss": 0.8481,
      "step": 91320
    },
    {
      "epoch": 1.6034340490528276,
      "grad_norm": 0.058556066450472864,
      "learning_rate": 0.00015746161859935361,
      "loss": 0.8493,
      "step": 91330
    },
    {
      "epoch": 1.6036096139328289,
      "grad_norm": 0.06287460081476147,
      "learning_rate": 0.0001574524646051189,
      "loss": 0.8474,
      "step": 91340
    },
    {
      "epoch": 1.6037851788128303,
      "grad_norm": 0.05012901716026252,
      "learning_rate": 0.00015744330989562064,
      "loss": 0.8462,
      "step": 91350
    },
    {
      "epoch": 1.6039607436928316,
      "grad_norm": 0.05166106862436587,
      "learning_rate": 0.00015743415447097483,
      "loss": 0.8523,
      "step": 91360
    },
    {
      "epoch": 1.604136308572833,
      "grad_norm": 0.060191452760553654,
      "learning_rate": 0.00015742499833129743,
      "loss": 0.8552,
      "step": 91370
    },
    {
      "epoch": 1.6043118734528345,
      "grad_norm": 0.06475776157527671,
      "learning_rate": 0.00015741584147670453,
      "loss": 0.8433,
      "step": 91380
    },
    {
      "epoch": 1.604487438332836,
      "grad_norm": 0.06525077659441123,
      "learning_rate": 0.0001574066839073121,
      "loss": 0.846,
      "step": 91390
    },
    {
      "epoch": 1.6046630032128373,
      "grad_norm": 0.05821473694129956,
      "learning_rate": 0.00015739752562323616,
      "loss": 0.8609,
      "step": 91400
    },
    {
      "epoch": 1.6048385680928388,
      "grad_norm": 0.05945963092466336,
      "learning_rate": 0.00015738836662459278,
      "loss": 0.8446,
      "step": 91410
    },
    {
      "epoch": 1.60501413297284,
      "grad_norm": 0.05246507726981029,
      "learning_rate": 0.00015737920691149797,
      "loss": 0.8499,
      "step": 91420
    },
    {
      "epoch": 1.6051896978528415,
      "grad_norm": 0.05555310303098996,
      "learning_rate": 0.0001573700464840678,
      "loss": 0.8549,
      "step": 91430
    },
    {
      "epoch": 1.605365262732843,
      "grad_norm": 0.049754416506118025,
      "learning_rate": 0.00015736088534241835,
      "loss": 0.8493,
      "step": 91440
    },
    {
      "epoch": 1.6055408276128444,
      "grad_norm": 0.0669976484130464,
      "learning_rate": 0.00015735172348666567,
      "loss": 0.8455,
      "step": 91450
    },
    {
      "epoch": 1.6057163924928457,
      "grad_norm": 0.0519056548940216,
      "learning_rate": 0.00015734256091692585,
      "loss": 0.8525,
      "step": 91460
    },
    {
      "epoch": 1.6058919573728472,
      "grad_norm": 0.06012237694817399,
      "learning_rate": 0.000157333397633315,
      "loss": 0.8507,
      "step": 91470
    },
    {
      "epoch": 1.6060675222528484,
      "grad_norm": 0.05523918885921821,
      "learning_rate": 0.0001573242336359492,
      "loss": 0.8527,
      "step": 91480
    },
    {
      "epoch": 1.60624308713285,
      "grad_norm": 0.06273448362501817,
      "learning_rate": 0.00015731506892494458,
      "loss": 0.8479,
      "step": 91490
    },
    {
      "epoch": 1.6064186520128514,
      "grad_norm": 0.07332672344257368,
      "learning_rate": 0.00015730590350041722,
      "loss": 0.8472,
      "step": 91500
    },
    {
      "epoch": 1.6065942168928529,
      "grad_norm": 0.0711517682984081,
      "learning_rate": 0.00015729673736248327,
      "loss": 0.8485,
      "step": 91510
    },
    {
      "epoch": 1.6067697817728541,
      "grad_norm": 0.06142207533967349,
      "learning_rate": 0.0001572875705112589,
      "loss": 0.8475,
      "step": 91520
    },
    {
      "epoch": 1.6069453466528556,
      "grad_norm": 0.09750607802192332,
      "learning_rate": 0.0001572784029468602,
      "loss": 0.8488,
      "step": 91530
    },
    {
      "epoch": 1.6071209115328569,
      "grad_norm": 0.058634920511593684,
      "learning_rate": 0.00015726923466940332,
      "loss": 0.8513,
      "step": 91540
    },
    {
      "epoch": 1.6072964764128583,
      "grad_norm": 0.06017798755646878,
      "learning_rate": 0.00015726006567900452,
      "loss": 0.8473,
      "step": 91550
    },
    {
      "epoch": 1.6074720412928598,
      "grad_norm": 0.08858183466123853,
      "learning_rate": 0.00015725089597577987,
      "loss": 0.844,
      "step": 91560
    },
    {
      "epoch": 1.6076476061728613,
      "grad_norm": 0.07433450639938674,
      "learning_rate": 0.00015724172555984562,
      "loss": 0.8519,
      "step": 91570
    },
    {
      "epoch": 1.6078231710528625,
      "grad_norm": 0.04953709744656424,
      "learning_rate": 0.0001572325544313179,
      "loss": 0.8491,
      "step": 91580
    },
    {
      "epoch": 1.607998735932864,
      "grad_norm": 0.052256940816319195,
      "learning_rate": 0.00015722338259031293,
      "loss": 0.8528,
      "step": 91590
    },
    {
      "epoch": 1.6081743008128653,
      "grad_norm": 0.05201197726534871,
      "learning_rate": 0.00015721421003694693,
      "loss": 0.8433,
      "step": 91600
    },
    {
      "epoch": 1.6083498656928668,
      "grad_norm": 0.05651357186059328,
      "learning_rate": 0.0001572050367713361,
      "loss": 0.8452,
      "step": 91610
    },
    {
      "epoch": 1.6085254305728682,
      "grad_norm": 0.0632188573427134,
      "learning_rate": 0.00015719586279359671,
      "loss": 0.8532,
      "step": 91620
    },
    {
      "epoch": 1.6087009954528697,
      "grad_norm": 0.050272765852413456,
      "learning_rate": 0.00015718668810384495,
      "loss": 0.8443,
      "step": 91630
    },
    {
      "epoch": 1.608876560332871,
      "grad_norm": 0.04797464557409146,
      "learning_rate": 0.00015717751270219712,
      "loss": 0.8453,
      "step": 91640
    },
    {
      "epoch": 1.6090521252128724,
      "grad_norm": 0.058121839063618716,
      "learning_rate": 0.0001571683365887694,
      "loss": 0.8502,
      "step": 91650
    },
    {
      "epoch": 1.6092276900928737,
      "grad_norm": 0.061265995566192576,
      "learning_rate": 0.00015715915976367812,
      "loss": 0.8491,
      "step": 91660
    },
    {
      "epoch": 1.6094032549728752,
      "grad_norm": 0.07880892483378525,
      "learning_rate": 0.00015714998222703949,
      "loss": 0.8484,
      "step": 91670
    },
    {
      "epoch": 1.6095788198528767,
      "grad_norm": 0.05354571666830261,
      "learning_rate": 0.00015714080397896985,
      "loss": 0.8568,
      "step": 91680
    },
    {
      "epoch": 1.6097543847328781,
      "grad_norm": 0.07556497106677107,
      "learning_rate": 0.00015713162501958546,
      "loss": 0.8459,
      "step": 91690
    },
    {
      "epoch": 1.6099299496128794,
      "grad_norm": 0.057510117989226485,
      "learning_rate": 0.0001571224453490026,
      "loss": 0.8546,
      "step": 91700
    },
    {
      "epoch": 1.6101055144928809,
      "grad_norm": 0.0742856290567842,
      "learning_rate": 0.00015711326496733763,
      "loss": 0.8449,
      "step": 91710
    },
    {
      "epoch": 1.6102810793728821,
      "grad_norm": 0.058729546233274026,
      "learning_rate": 0.00015710408387470683,
      "loss": 0.8437,
      "step": 91720
    },
    {
      "epoch": 1.6104566442528836,
      "grad_norm": 0.06802876162784813,
      "learning_rate": 0.0001570949020712265,
      "loss": 0.8444,
      "step": 91730
    },
    {
      "epoch": 1.610632209132885,
      "grad_norm": 0.04834947761171741,
      "learning_rate": 0.00015708571955701307,
      "loss": 0.8508,
      "step": 91740
    },
    {
      "epoch": 1.6108077740128866,
      "grad_norm": 0.1018302778677276,
      "learning_rate": 0.00015707653633218277,
      "loss": 0.8499,
      "step": 91750
    },
    {
      "epoch": 1.610983338892888,
      "grad_norm": 0.05523449964262309,
      "learning_rate": 0.0001570673523968521,
      "loss": 0.848,
      "step": 91760
    },
    {
      "epoch": 1.6111589037728893,
      "grad_norm": 0.05754321803079446,
      "learning_rate": 0.0001570581677511372,
      "loss": 0.8516,
      "step": 91770
    },
    {
      "epoch": 1.6113344686528905,
      "grad_norm": 0.07653230022556887,
      "learning_rate": 0.00015704898239515464,
      "loss": 0.8434,
      "step": 91780
    },
    {
      "epoch": 1.611510033532892,
      "grad_norm": 0.05225267788296726,
      "learning_rate": 0.00015703979632902078,
      "loss": 0.8528,
      "step": 91790
    },
    {
      "epoch": 1.6116855984128935,
      "grad_norm": 0.06205477766080048,
      "learning_rate": 0.00015703060955285186,
      "loss": 0.8348,
      "step": 91800
    },
    {
      "epoch": 1.611861163292895,
      "grad_norm": 0.0888610816851985,
      "learning_rate": 0.00015702142206676443,
      "loss": 0.8474,
      "step": 91810
    },
    {
      "epoch": 1.6120367281728964,
      "grad_norm": 0.052633991622013314,
      "learning_rate": 0.00015701223387087486,
      "loss": 0.8443,
      "step": 91820
    },
    {
      "epoch": 1.6122122930528977,
      "grad_norm": 0.06485789066747355,
      "learning_rate": 0.0001570030449652995,
      "loss": 0.8514,
      "step": 91830
    },
    {
      "epoch": 1.612387857932899,
      "grad_norm": 0.06827918884383587,
      "learning_rate": 0.0001569938553501549,
      "loss": 0.8514,
      "step": 91840
    },
    {
      "epoch": 1.6125634228129004,
      "grad_norm": 0.062248641692512714,
      "learning_rate": 0.00015698466502555737,
      "loss": 0.8532,
      "step": 91850
    },
    {
      "epoch": 1.612738987692902,
      "grad_norm": 0.0733468429482594,
      "learning_rate": 0.0001569754739916234,
      "loss": 0.8394,
      "step": 91860
    },
    {
      "epoch": 1.6129145525729034,
      "grad_norm": 0.04796396540110186,
      "learning_rate": 0.00015696628224846947,
      "loss": 0.8477,
      "step": 91870
    },
    {
      "epoch": 1.6130901174529049,
      "grad_norm": 0.04677555675843478,
      "learning_rate": 0.00015695708979621205,
      "loss": 0.8538,
      "step": 91880
    },
    {
      "epoch": 1.6132656823329061,
      "grad_norm": 0.05745246204054577,
      "learning_rate": 0.00015694789663496753,
      "loss": 0.85,
      "step": 91890
    },
    {
      "epoch": 1.6134412472129074,
      "grad_norm": 0.07501941437791308,
      "learning_rate": 0.00015693870276485244,
      "loss": 0.8503,
      "step": 91900
    },
    {
      "epoch": 1.6136168120929089,
      "grad_norm": 0.04504932900697843,
      "learning_rate": 0.0001569295081859833,
      "loss": 0.8494,
      "step": 91910
    },
    {
      "epoch": 1.6137923769729103,
      "grad_norm": 0.07867292998033575,
      "learning_rate": 0.00015692031289847653,
      "loss": 0.8493,
      "step": 91920
    },
    {
      "epoch": 1.6139679418529118,
      "grad_norm": 0.05457835176155886,
      "learning_rate": 0.00015691111690244872,
      "loss": 0.8443,
      "step": 91930
    },
    {
      "epoch": 1.6141435067329133,
      "grad_norm": 0.06951346885478155,
      "learning_rate": 0.0001569019201980163,
      "loss": 0.8505,
      "step": 91940
    },
    {
      "epoch": 1.6143190716129145,
      "grad_norm": 0.05311360481657236,
      "learning_rate": 0.00015689272278529586,
      "loss": 0.8478,
      "step": 91950
    },
    {
      "epoch": 1.6144946364929158,
      "grad_norm": 0.06862041612558054,
      "learning_rate": 0.0001568835246644039,
      "loss": 0.8364,
      "step": 91960
    },
    {
      "epoch": 1.6146702013729173,
      "grad_norm": 0.05171109398779789,
      "learning_rate": 0.000156874325835457,
      "loss": 0.8528,
      "step": 91970
    },
    {
      "epoch": 1.6148457662529188,
      "grad_norm": 0.050311046416699666,
      "learning_rate": 0.00015686512629857165,
      "loss": 0.8468,
      "step": 91980
    },
    {
      "epoch": 1.6150213311329202,
      "grad_norm": 0.05678179343155231,
      "learning_rate": 0.00015685592605386447,
      "loss": 0.8504,
      "step": 91990
    },
    {
      "epoch": 1.6151968960129217,
      "grad_norm": 0.08147487310892643,
      "learning_rate": 0.00015684672510145198,
      "loss": 0.8537,
      "step": 92000
    },
    {
      "epoch": 1.615372460892923,
      "grad_norm": 0.059483420563188075,
      "learning_rate": 0.00015683752344145078,
      "loss": 0.8438,
      "step": 92010
    },
    {
      "epoch": 1.6155480257729242,
      "grad_norm": 0.0709812267798307,
      "learning_rate": 0.00015682832107397743,
      "loss": 0.8445,
      "step": 92020
    },
    {
      "epoch": 1.6157235906529257,
      "grad_norm": 0.0491997444644071,
      "learning_rate": 0.0001568191179991486,
      "loss": 0.8549,
      "step": 92030
    },
    {
      "epoch": 1.6158991555329272,
      "grad_norm": 0.050884178669810674,
      "learning_rate": 0.00015680991421708083,
      "loss": 0.8444,
      "step": 92040
    },
    {
      "epoch": 1.6160747204129287,
      "grad_norm": 0.05110720954694618,
      "learning_rate": 0.00015680070972789076,
      "loss": 0.8531,
      "step": 92050
    },
    {
      "epoch": 1.6162502852929301,
      "grad_norm": 0.06770730895974163,
      "learning_rate": 0.00015679150453169502,
      "loss": 0.842,
      "step": 92060
    },
    {
      "epoch": 1.6164258501729314,
      "grad_norm": 0.07627702140078982,
      "learning_rate": 0.0001567822986286102,
      "loss": 0.8462,
      "step": 92070
    },
    {
      "epoch": 1.6166014150529329,
      "grad_norm": 0.05620232437980508,
      "learning_rate": 0.000156773092018753,
      "loss": 0.8609,
      "step": 92080
    },
    {
      "epoch": 1.6167769799329341,
      "grad_norm": 0.04468474342703498,
      "learning_rate": 0.00015676388470224,
      "loss": 0.8482,
      "step": 92090
    },
    {
      "epoch": 1.6169525448129356,
      "grad_norm": 0.06187641558994192,
      "learning_rate": 0.00015675467667918789,
      "loss": 0.8522,
      "step": 92100
    },
    {
      "epoch": 1.617128109692937,
      "grad_norm": 0.08808014540767041,
      "learning_rate": 0.00015674546794971334,
      "loss": 0.8448,
      "step": 92110
    },
    {
      "epoch": 1.6173036745729386,
      "grad_norm": 0.051657050214011006,
      "learning_rate": 0.00015673625851393304,
      "loss": 0.8415,
      "step": 92120
    },
    {
      "epoch": 1.6174792394529398,
      "grad_norm": 0.0653644526622678,
      "learning_rate": 0.00015672704837196367,
      "loss": 0.8507,
      "step": 92130
    },
    {
      "epoch": 1.6176548043329413,
      "grad_norm": 0.05206445223771713,
      "learning_rate": 0.0001567178375239219,
      "loss": 0.8435,
      "step": 92140
    },
    {
      "epoch": 1.6178303692129425,
      "grad_norm": 0.08734013950763092,
      "learning_rate": 0.00015670862596992447,
      "loss": 0.8491,
      "step": 92150
    },
    {
      "epoch": 1.618005934092944,
      "grad_norm": 0.05811441579998736,
      "learning_rate": 0.00015669941371008805,
      "loss": 0.8477,
      "step": 92160
    },
    {
      "epoch": 1.6181814989729455,
      "grad_norm": 0.04488625629155584,
      "learning_rate": 0.0001566902007445294,
      "loss": 0.8479,
      "step": 92170
    },
    {
      "epoch": 1.618357063852947,
      "grad_norm": 0.06018926909245393,
      "learning_rate": 0.0001566809870733652,
      "loss": 0.8414,
      "step": 92180
    },
    {
      "epoch": 1.6185326287329482,
      "grad_norm": 0.07502576798546222,
      "learning_rate": 0.00015667177269671224,
      "loss": 0.8491,
      "step": 92190
    },
    {
      "epoch": 1.6187081936129497,
      "grad_norm": 0.09269385118498787,
      "learning_rate": 0.0001566625576146873,
      "loss": 0.8454,
      "step": 92200
    },
    {
      "epoch": 1.618883758492951,
      "grad_norm": 0.05277418204238713,
      "learning_rate": 0.00015665334182740702,
      "loss": 0.8457,
      "step": 92210
    },
    {
      "epoch": 1.6190593233729524,
      "grad_norm": 0.057901920981282426,
      "learning_rate": 0.00015664412533498826,
      "loss": 0.8372,
      "step": 92220
    },
    {
      "epoch": 1.619234888252954,
      "grad_norm": 0.05391818838765353,
      "learning_rate": 0.0001566349081375478,
      "loss": 0.8534,
      "step": 92230
    },
    {
      "epoch": 1.6194104531329554,
      "grad_norm": 0.05655681662066759,
      "learning_rate": 0.00015662569023520233,
      "loss": 0.8482,
      "step": 92240
    },
    {
      "epoch": 1.6195860180129567,
      "grad_norm": 0.054086050835373135,
      "learning_rate": 0.00015661647162806873,
      "loss": 0.846,
      "step": 92250
    },
    {
      "epoch": 1.6197615828929581,
      "grad_norm": 0.05205043246701837,
      "learning_rate": 0.0001566072523162638,
      "loss": 0.8568,
      "step": 92260
    },
    {
      "epoch": 1.6199371477729594,
      "grad_norm": 0.10623413832666874,
      "learning_rate": 0.0001565980322999043,
      "loss": 0.8468,
      "step": 92270
    },
    {
      "epoch": 1.6201127126529609,
      "grad_norm": 0.06000227901884539,
      "learning_rate": 0.0001565888115791071,
      "loss": 0.8464,
      "step": 92280
    },
    {
      "epoch": 1.6202882775329623,
      "grad_norm": 0.04775154507919116,
      "learning_rate": 0.00015657959015398897,
      "loss": 0.8459,
      "step": 92290
    },
    {
      "epoch": 1.6204638424129638,
      "grad_norm": 0.06069640780448736,
      "learning_rate": 0.0001565703680246668,
      "loss": 0.8469,
      "step": 92300
    },
    {
      "epoch": 1.620639407292965,
      "grad_norm": 0.051700307545915654,
      "learning_rate": 0.0001565611451912574,
      "loss": 0.8503,
      "step": 92310
    },
    {
      "epoch": 1.6208149721729666,
      "grad_norm": 0.051811252949011004,
      "learning_rate": 0.00015655192165387767,
      "loss": 0.8448,
      "step": 92320
    },
    {
      "epoch": 1.6209905370529678,
      "grad_norm": 0.05988035092879162,
      "learning_rate": 0.00015654269741264442,
      "loss": 0.8467,
      "step": 92330
    },
    {
      "epoch": 1.6211661019329693,
      "grad_norm": 0.052679319105170645,
      "learning_rate": 0.0001565334724676746,
      "loss": 0.8516,
      "step": 92340
    },
    {
      "epoch": 1.6213416668129708,
      "grad_norm": 0.05981028271487582,
      "learning_rate": 0.00015652424681908498,
      "loss": 0.8418,
      "step": 92350
    },
    {
      "epoch": 1.6215172316929722,
      "grad_norm": 0.04999591846751715,
      "learning_rate": 0.00015651502046699252,
      "loss": 0.8503,
      "step": 92360
    },
    {
      "epoch": 1.6216927965729735,
      "grad_norm": 0.054951380721945775,
      "learning_rate": 0.00015650579341151415,
      "loss": 0.8429,
      "step": 92370
    },
    {
      "epoch": 1.621868361452975,
      "grad_norm": 0.062003664449568646,
      "learning_rate": 0.0001564965656527667,
      "loss": 0.8426,
      "step": 92380
    },
    {
      "epoch": 1.6220439263329762,
      "grad_norm": 0.04788974130819218,
      "learning_rate": 0.00015648733719086713,
      "loss": 0.8401,
      "step": 92390
    },
    {
      "epoch": 1.6222194912129777,
      "grad_norm": 0.05832114099005633,
      "learning_rate": 0.0001564781080259324,
      "loss": 0.8502,
      "step": 92400
    },
    {
      "epoch": 1.6223950560929792,
      "grad_norm": 0.05592410725133814,
      "learning_rate": 0.00015646887815807937,
      "loss": 0.8557,
      "step": 92410
    },
    {
      "epoch": 1.6225706209729807,
      "grad_norm": 0.07455191797722206,
      "learning_rate": 0.00015645964758742505,
      "loss": 0.8497,
      "step": 92420
    },
    {
      "epoch": 1.622746185852982,
      "grad_norm": 0.06064521646808161,
      "learning_rate": 0.00015645041631408638,
      "loss": 0.8466,
      "step": 92430
    },
    {
      "epoch": 1.6229217507329834,
      "grad_norm": 0.08176374048337834,
      "learning_rate": 0.00015644118433818025,
      "loss": 0.8439,
      "step": 92440
    },
    {
      "epoch": 1.6230973156129846,
      "grad_norm": 0.05716780033480381,
      "learning_rate": 0.0001564319516598237,
      "loss": 0.8537,
      "step": 92450
    },
    {
      "epoch": 1.6232728804929861,
      "grad_norm": 0.059353820442681365,
      "learning_rate": 0.00015642271827913374,
      "loss": 0.8402,
      "step": 92460
    },
    {
      "epoch": 1.6234484453729876,
      "grad_norm": 0.0950327231947554,
      "learning_rate": 0.0001564134841962273,
      "loss": 0.849,
      "step": 92470
    },
    {
      "epoch": 1.623624010252989,
      "grad_norm": 0.06730761709434536,
      "learning_rate": 0.00015640424941122137,
      "loss": 0.8488,
      "step": 92480
    },
    {
      "epoch": 1.6237995751329906,
      "grad_norm": 0.05946700461846626,
      "learning_rate": 0.00015639501392423303,
      "loss": 0.8522,
      "step": 92490
    },
    {
      "epoch": 1.6239751400129918,
      "grad_norm": 0.07683308542570587,
      "learning_rate": 0.0001563857777353792,
      "loss": 0.8437,
      "step": 92500
    },
    {
      "epoch": 1.624150704892993,
      "grad_norm": 0.05460451362865213,
      "learning_rate": 0.00015637654084477699,
      "loss": 0.8509,
      "step": 92510
    },
    {
      "epoch": 1.6243262697729945,
      "grad_norm": 0.06098668570077405,
      "learning_rate": 0.00015636730325254339,
      "loss": 0.8461,
      "step": 92520
    },
    {
      "epoch": 1.624501834652996,
      "grad_norm": 0.04020228689400136,
      "learning_rate": 0.00015635806495879542,
      "loss": 0.8501,
      "step": 92530
    },
    {
      "epoch": 1.6246773995329975,
      "grad_norm": 0.06338296593068658,
      "learning_rate": 0.0001563488259636502,
      "loss": 0.8474,
      "step": 92540
    },
    {
      "epoch": 1.624852964412999,
      "grad_norm": 0.06315611513719134,
      "learning_rate": 0.0001563395862672247,
      "loss": 0.8652,
      "step": 92550
    },
    {
      "epoch": 1.6250285292930002,
      "grad_norm": 0.07746296093124516,
      "learning_rate": 0.00015633034586963603,
      "loss": 0.8461,
      "step": 92560
    },
    {
      "epoch": 1.6252040941730015,
      "grad_norm": 0.061099130381150356,
      "learning_rate": 0.00015632110477100132,
      "loss": 0.8535,
      "step": 92570
    },
    {
      "epoch": 1.625379659053003,
      "grad_norm": 0.044475952703707945,
      "learning_rate": 0.0001563118629714376,
      "loss": 0.8471,
      "step": 92580
    },
    {
      "epoch": 1.6255552239330044,
      "grad_norm": 0.05392082821257424,
      "learning_rate": 0.00015630262047106196,
      "loss": 0.8482,
      "step": 92590
    },
    {
      "epoch": 1.625730788813006,
      "grad_norm": 0.047080834990325074,
      "learning_rate": 0.00015629337726999154,
      "loss": 0.8563,
      "step": 92600
    },
    {
      "epoch": 1.6259063536930074,
      "grad_norm": 0.04985045811453559,
      "learning_rate": 0.0001562841333683434,
      "loss": 0.8487,
      "step": 92610
    },
    {
      "epoch": 1.6260819185730087,
      "grad_norm": 0.043029943669385715,
      "learning_rate": 0.00015627488876623474,
      "loss": 0.8445,
      "step": 92620
    },
    {
      "epoch": 1.62625748345301,
      "grad_norm": 0.08720968253128636,
      "learning_rate": 0.00015626564346378258,
      "loss": 0.8501,
      "step": 92630
    },
    {
      "epoch": 1.6264330483330114,
      "grad_norm": 0.05658301616063183,
      "learning_rate": 0.0001562563974611042,
      "loss": 0.8462,
      "step": 92640
    },
    {
      "epoch": 1.6266086132130129,
      "grad_norm": 0.09706837569948921,
      "learning_rate": 0.00015624715075831662,
      "loss": 0.8428,
      "step": 92650
    },
    {
      "epoch": 1.6267841780930143,
      "grad_norm": 0.057333635734844596,
      "learning_rate": 0.0001562379033555371,
      "loss": 0.8496,
      "step": 92660
    },
    {
      "epoch": 1.6269597429730158,
      "grad_norm": 0.07594648278293901,
      "learning_rate": 0.00015622865525288272,
      "loss": 0.8452,
      "step": 92670
    },
    {
      "epoch": 1.627135307853017,
      "grad_norm": 0.06458515158404665,
      "learning_rate": 0.0001562194064504707,
      "loss": 0.8465,
      "step": 92680
    },
    {
      "epoch": 1.6273108727330183,
      "grad_norm": 0.06427933638924332,
      "learning_rate": 0.00015621015694841823,
      "loss": 0.8425,
      "step": 92690
    },
    {
      "epoch": 1.6274864376130198,
      "grad_norm": 0.07018146426506693,
      "learning_rate": 0.00015620090674684246,
      "loss": 0.8494,
      "step": 92700
    },
    {
      "epoch": 1.6276620024930213,
      "grad_norm": 0.047801818249598976,
      "learning_rate": 0.00015619165584586066,
      "loss": 0.8466,
      "step": 92710
    },
    {
      "epoch": 1.6278375673730228,
      "grad_norm": 0.08840509540208691,
      "learning_rate": 0.00015618240424558995,
      "loss": 0.8465,
      "step": 92720
    },
    {
      "epoch": 1.6280131322530242,
      "grad_norm": 0.056129550395446495,
      "learning_rate": 0.00015617315194614768,
      "loss": 0.8491,
      "step": 92730
    },
    {
      "epoch": 1.6281886971330255,
      "grad_norm": 0.05991943635206426,
      "learning_rate": 0.00015616389894765095,
      "loss": 0.8441,
      "step": 92740
    },
    {
      "epoch": 1.6283642620130268,
      "grad_norm": 0.06768700756135608,
      "learning_rate": 0.00015615464525021706,
      "loss": 0.8542,
      "step": 92750
    },
    {
      "epoch": 1.6285398268930282,
      "grad_norm": 0.07001517560391825,
      "learning_rate": 0.0001561453908539632,
      "loss": 0.8476,
      "step": 92760
    },
    {
      "epoch": 1.6287153917730297,
      "grad_norm": 0.05902598661337365,
      "learning_rate": 0.00015613613575900676,
      "loss": 0.8525,
      "step": 92770
    },
    {
      "epoch": 1.6288909566530312,
      "grad_norm": 0.048001676880633906,
      "learning_rate": 0.00015612687996546486,
      "loss": 0.8489,
      "step": 92780
    },
    {
      "epoch": 1.6290665215330327,
      "grad_norm": 0.0478673309183264,
      "learning_rate": 0.00015611762347345483,
      "loss": 0.85,
      "step": 92790
    },
    {
      "epoch": 1.629242086413034,
      "grad_norm": 0.06589338471964809,
      "learning_rate": 0.00015610836628309396,
      "loss": 0.8441,
      "step": 92800
    },
    {
      "epoch": 1.6294176512930354,
      "grad_norm": 0.05964627486753238,
      "learning_rate": 0.0001560991083944995,
      "loss": 0.8576,
      "step": 92810
    },
    {
      "epoch": 1.6295932161730367,
      "grad_norm": 0.07594940725536976,
      "learning_rate": 0.0001560898498077888,
      "loss": 0.8443,
      "step": 92820
    },
    {
      "epoch": 1.6297687810530381,
      "grad_norm": 0.05065945871894749,
      "learning_rate": 0.00015608059052307915,
      "loss": 0.8432,
      "step": 92830
    },
    {
      "epoch": 1.6299443459330396,
      "grad_norm": 0.045733954687040446,
      "learning_rate": 0.00015607133054048783,
      "loss": 0.8525,
      "step": 92840
    },
    {
      "epoch": 1.630119910813041,
      "grad_norm": 0.042836778898670415,
      "learning_rate": 0.00015606206986013224,
      "loss": 0.855,
      "step": 92850
    },
    {
      "epoch": 1.6302954756930423,
      "grad_norm": 0.06059820391276282,
      "learning_rate": 0.00015605280848212966,
      "loss": 0.8508,
      "step": 92860
    },
    {
      "epoch": 1.6304710405730438,
      "grad_norm": 0.05527278156083065,
      "learning_rate": 0.00015604354640659743,
      "loss": 0.8521,
      "step": 92870
    },
    {
      "epoch": 1.630646605453045,
      "grad_norm": 0.05280593888780836,
      "learning_rate": 0.0001560342836336529,
      "loss": 0.844,
      "step": 92880
    },
    {
      "epoch": 1.6308221703330466,
      "grad_norm": 0.053537481813829516,
      "learning_rate": 0.00015602502016341348,
      "loss": 0.85,
      "step": 92890
    },
    {
      "epoch": 1.630997735213048,
      "grad_norm": 0.0711011242989417,
      "learning_rate": 0.0001560157559959965,
      "loss": 0.838,
      "step": 92900
    },
    {
      "epoch": 1.6311733000930495,
      "grad_norm": 0.05246343666878841,
      "learning_rate": 0.00015600649113151936,
      "loss": 0.8484,
      "step": 92910
    },
    {
      "epoch": 1.6313488649730508,
      "grad_norm": 0.09514382178138732,
      "learning_rate": 0.00015599722557009941,
      "loss": 0.8483,
      "step": 92920
    },
    {
      "epoch": 1.6315244298530522,
      "grad_norm": 0.060782909387991836,
      "learning_rate": 0.0001559879593118541,
      "loss": 0.8446,
      "step": 92930
    },
    {
      "epoch": 1.6316999947330535,
      "grad_norm": 0.08349018865739723,
      "learning_rate": 0.00015597869235690082,
      "loss": 0.8524,
      "step": 92940
    },
    {
      "epoch": 1.631875559613055,
      "grad_norm": 0.07676068845839809,
      "learning_rate": 0.00015596942470535693,
      "loss": 0.8507,
      "step": 92950
    },
    {
      "epoch": 1.6320511244930564,
      "grad_norm": 0.06490228821647631,
      "learning_rate": 0.00015596015635733988,
      "loss": 0.8478,
      "step": 92960
    },
    {
      "epoch": 1.632226689373058,
      "grad_norm": 0.061407681133224236,
      "learning_rate": 0.00015595088731296716,
      "loss": 0.8509,
      "step": 92970
    },
    {
      "epoch": 1.6324022542530592,
      "grad_norm": 0.05528066562560166,
      "learning_rate": 0.00015594161757235617,
      "loss": 0.8503,
      "step": 92980
    },
    {
      "epoch": 1.6325778191330607,
      "grad_norm": 0.06646586417675106,
      "learning_rate": 0.00015593234713562432,
      "loss": 0.8397,
      "step": 92990
    },
    {
      "epoch": 1.632753384013062,
      "grad_norm": 0.05357567509097248,
      "learning_rate": 0.00015592307600288913,
      "loss": 0.8454,
      "step": 93000
    },
    {
      "epoch": 1.6329289488930634,
      "grad_norm": 0.06800918316989468,
      "learning_rate": 0.00015591380417426804,
      "loss": 0.8505,
      "step": 93010
    },
    {
      "epoch": 1.6331045137730649,
      "grad_norm": 0.05796922906380319,
      "learning_rate": 0.00015590453164987853,
      "loss": 0.8535,
      "step": 93020
    },
    {
      "epoch": 1.6332800786530663,
      "grad_norm": 0.06953662039716438,
      "learning_rate": 0.00015589525842983805,
      "loss": 0.848,
      "step": 93030
    },
    {
      "epoch": 1.6334556435330676,
      "grad_norm": 0.05921364776863275,
      "learning_rate": 0.00015588598451426414,
      "loss": 0.8447,
      "step": 93040
    },
    {
      "epoch": 1.633631208413069,
      "grad_norm": 0.06279521326048645,
      "learning_rate": 0.0001558767099032743,
      "loss": 0.8455,
      "step": 93050
    },
    {
      "epoch": 1.6338067732930703,
      "grad_norm": 0.06335954920015711,
      "learning_rate": 0.00015586743459698602,
      "loss": 0.8456,
      "step": 93060
    },
    {
      "epoch": 1.6339823381730718,
      "grad_norm": 0.07522804216287529,
      "learning_rate": 0.00015585815859551686,
      "loss": 0.8507,
      "step": 93070
    },
    {
      "epoch": 1.6341579030530733,
      "grad_norm": 0.050495621690397285,
      "learning_rate": 0.00015584888189898428,
      "loss": 0.8475,
      "step": 93080
    },
    {
      "epoch": 1.6343334679330748,
      "grad_norm": 0.05904948401838302,
      "learning_rate": 0.00015583960450750586,
      "loss": 0.8545,
      "step": 93090
    },
    {
      "epoch": 1.634509032813076,
      "grad_norm": 0.05453140439933258,
      "learning_rate": 0.00015583032642119916,
      "loss": 0.8497,
      "step": 93100
    },
    {
      "epoch": 1.6346845976930775,
      "grad_norm": 0.06705514986193681,
      "learning_rate": 0.00015582104764018167,
      "loss": 0.8461,
      "step": 93110
    },
    {
      "epoch": 1.6348601625730788,
      "grad_norm": 0.06563125586449993,
      "learning_rate": 0.00015581176816457107,
      "loss": 0.853,
      "step": 93120
    },
    {
      "epoch": 1.6350357274530802,
      "grad_norm": 0.06402283246967556,
      "learning_rate": 0.00015580248799448477,
      "loss": 0.845,
      "step": 93130
    },
    {
      "epoch": 1.6352112923330817,
      "grad_norm": 0.04390886009995161,
      "learning_rate": 0.0001557932071300405,
      "loss": 0.8543,
      "step": 93140
    },
    {
      "epoch": 1.6353868572130832,
      "grad_norm": 0.10662917787309614,
      "learning_rate": 0.00015578392557135583,
      "loss": 0.8448,
      "step": 93150
    },
    {
      "epoch": 1.6355624220930844,
      "grad_norm": 0.04654190046370772,
      "learning_rate": 0.00015577464331854824,
      "loss": 0.8519,
      "step": 93160
    },
    {
      "epoch": 1.635737986973086,
      "grad_norm": 0.06630467115983776,
      "learning_rate": 0.0001557653603717355,
      "loss": 0.8542,
      "step": 93170
    },
    {
      "epoch": 1.6359135518530872,
      "grad_norm": 0.08799258814637599,
      "learning_rate": 0.0001557560767310351,
      "loss": 0.8481,
      "step": 93180
    },
    {
      "epoch": 1.6360891167330887,
      "grad_norm": 0.05039967452199262,
      "learning_rate": 0.00015574679239656471,
      "loss": 0.8536,
      "step": 93190
    },
    {
      "epoch": 1.6362646816130901,
      "grad_norm": 0.06279902911816634,
      "learning_rate": 0.00015573750736844202,
      "loss": 0.8513,
      "step": 93200
    },
    {
      "epoch": 1.6364402464930916,
      "grad_norm": 0.06017277823471987,
      "learning_rate": 0.00015572822164678455,
      "loss": 0.8484,
      "step": 93210
    },
    {
      "epoch": 1.636615811373093,
      "grad_norm": 0.05842352388928956,
      "learning_rate": 0.00015571893523171008,
      "loss": 0.8442,
      "step": 93220
    },
    {
      "epoch": 1.6367913762530943,
      "grad_norm": 0.058063863848250075,
      "learning_rate": 0.0001557096481233362,
      "loss": 0.8434,
      "step": 93230
    },
    {
      "epoch": 1.6369669411330956,
      "grad_norm": 0.056090845713304735,
      "learning_rate": 0.0001557003603217806,
      "loss": 0.8486,
      "step": 93240
    },
    {
      "epoch": 1.637142506013097,
      "grad_norm": 0.06565896056368768,
      "learning_rate": 0.00015569107182716093,
      "loss": 0.8532,
      "step": 93250
    },
    {
      "epoch": 1.6373180708930986,
      "grad_norm": 0.05310025301412718,
      "learning_rate": 0.00015568178263959488,
      "loss": 0.8398,
      "step": 93260
    },
    {
      "epoch": 1.6374936357731,
      "grad_norm": 0.053287086216140096,
      "learning_rate": 0.00015567249275920023,
      "loss": 0.8482,
      "step": 93270
    },
    {
      "epoch": 1.6376692006531015,
      "grad_norm": 0.07995294826187803,
      "learning_rate": 0.00015566320218609458,
      "loss": 0.8455,
      "step": 93280
    },
    {
      "epoch": 1.6378447655331028,
      "grad_norm": 0.06031950862694152,
      "learning_rate": 0.0001556539109203957,
      "loss": 0.8503,
      "step": 93290
    },
    {
      "epoch": 1.638020330413104,
      "grad_norm": 0.06577506622315833,
      "learning_rate": 0.00015564461896222127,
      "loss": 0.8432,
      "step": 93300
    },
    {
      "epoch": 1.6381958952931055,
      "grad_norm": 0.06687395733221028,
      "learning_rate": 0.00015563532631168904,
      "loss": 0.8517,
      "step": 93310
    },
    {
      "epoch": 1.638371460173107,
      "grad_norm": 0.06785014369017232,
      "learning_rate": 0.00015562603296891678,
      "loss": 0.8503,
      "step": 93320
    },
    {
      "epoch": 1.6385470250531085,
      "grad_norm": 0.0531955730099773,
      "learning_rate": 0.00015561673893402217,
      "loss": 0.8429,
      "step": 93330
    },
    {
      "epoch": 1.63872258993311,
      "grad_norm": 0.08283356934745809,
      "learning_rate": 0.00015560744420712306,
      "loss": 0.8449,
      "step": 93340
    },
    {
      "epoch": 1.6388981548131112,
      "grad_norm": 0.061719933800836534,
      "learning_rate": 0.00015559814878833716,
      "loss": 0.8496,
      "step": 93350
    },
    {
      "epoch": 1.6390737196931124,
      "grad_norm": 0.04819686682550963,
      "learning_rate": 0.0001555888526777822,
      "loss": 0.8394,
      "step": 93360
    },
    {
      "epoch": 1.639249284573114,
      "grad_norm": 0.049265222046182666,
      "learning_rate": 0.0001555795558755761,
      "loss": 0.8429,
      "step": 93370
    },
    {
      "epoch": 1.6394248494531154,
      "grad_norm": 0.05110838008035661,
      "learning_rate": 0.0001555702583818365,
      "loss": 0.8434,
      "step": 93380
    },
    {
      "epoch": 1.6396004143331169,
      "grad_norm": 0.05464654086886538,
      "learning_rate": 0.0001555609601966813,
      "loss": 0.8458,
      "step": 93390
    },
    {
      "epoch": 1.6397759792131184,
      "grad_norm": 0.04438792208593837,
      "learning_rate": 0.00015555166132022826,
      "loss": 0.8453,
      "step": 93400
    },
    {
      "epoch": 1.6399515440931196,
      "grad_norm": 0.08005731176164475,
      "learning_rate": 0.0001555423617525952,
      "loss": 0.856,
      "step": 93410
    },
    {
      "epoch": 1.6401271089731209,
      "grad_norm": 0.053448918489175615,
      "learning_rate": 0.00015553306149390002,
      "loss": 0.8435,
      "step": 93420
    },
    {
      "epoch": 1.6403026738531223,
      "grad_norm": 0.04861196482139708,
      "learning_rate": 0.00015552376054426043,
      "loss": 0.85,
      "step": 93430
    },
    {
      "epoch": 1.6404782387331238,
      "grad_norm": 0.05791412061152611,
      "learning_rate": 0.0001555144589037944,
      "loss": 0.8463,
      "step": 93440
    },
    {
      "epoch": 1.6406538036131253,
      "grad_norm": 0.09148201984048321,
      "learning_rate": 0.0001555051565726197,
      "loss": 0.8504,
      "step": 93450
    },
    {
      "epoch": 1.6408293684931268,
      "grad_norm": 0.05559436393345338,
      "learning_rate": 0.00015549585355085422,
      "loss": 0.8522,
      "step": 93460
    },
    {
      "epoch": 1.641004933373128,
      "grad_norm": 0.06479530829217506,
      "learning_rate": 0.00015548654983861584,
      "loss": 0.8452,
      "step": 93470
    },
    {
      "epoch": 1.6411804982531293,
      "grad_norm": 0.054691204195973876,
      "learning_rate": 0.00015547724543602245,
      "loss": 0.8432,
      "step": 93480
    },
    {
      "epoch": 1.6413560631331308,
      "grad_norm": 0.07430663417027324,
      "learning_rate": 0.0001554679403431919,
      "loss": 0.8454,
      "step": 93490
    },
    {
      "epoch": 1.6415316280131322,
      "grad_norm": 0.04644353638181911,
      "learning_rate": 0.0001554586345602421,
      "loss": 0.8484,
      "step": 93500
    },
    {
      "epoch": 1.6417071928931337,
      "grad_norm": 0.059197014048850756,
      "learning_rate": 0.000155449328087291,
      "loss": 0.8482,
      "step": 93510
    },
    {
      "epoch": 1.6418827577731352,
      "grad_norm": 0.06211160599815429,
      "learning_rate": 0.0001554400209244564,
      "loss": 0.8418,
      "step": 93520
    },
    {
      "epoch": 1.6420583226531364,
      "grad_norm": 0.07581239298340244,
      "learning_rate": 0.00015543071307185638,
      "loss": 0.8468,
      "step": 93530
    },
    {
      "epoch": 1.642233887533138,
      "grad_norm": 0.0664447821428876,
      "learning_rate": 0.00015542140452960877,
      "loss": 0.8433,
      "step": 93540
    },
    {
      "epoch": 1.6424094524131392,
      "grad_norm": 0.06474086576010334,
      "learning_rate": 0.0001554120952978315,
      "loss": 0.8511,
      "step": 93550
    },
    {
      "epoch": 1.6425850172931407,
      "grad_norm": 0.04850172578644998,
      "learning_rate": 0.0001554027853766426,
      "loss": 0.8497,
      "step": 93560
    },
    {
      "epoch": 1.6427605821731421,
      "grad_norm": 0.0617262399023797,
      "learning_rate": 0.00015539347476616,
      "loss": 0.8429,
      "step": 93570
    },
    {
      "epoch": 1.6429361470531436,
      "grad_norm": 0.05570664650435086,
      "learning_rate": 0.0001553841634665016,
      "loss": 0.8517,
      "step": 93580
    },
    {
      "epoch": 1.6431117119331449,
      "grad_norm": 0.06196171108592767,
      "learning_rate": 0.00015537485147778542,
      "loss": 0.851,
      "step": 93590
    },
    {
      "epoch": 1.6432872768131463,
      "grad_norm": 0.04555415117647718,
      "learning_rate": 0.00015536553880012949,
      "loss": 0.8511,
      "step": 93600
    },
    {
      "epoch": 1.6434628416931476,
      "grad_norm": 0.05795995480714154,
      "learning_rate": 0.00015535622543365175,
      "loss": 0.8452,
      "step": 93610
    },
    {
      "epoch": 1.643638406573149,
      "grad_norm": 0.04096322037153918,
      "learning_rate": 0.00015534691137847022,
      "loss": 0.8469,
      "step": 93620
    },
    {
      "epoch": 1.6438139714531506,
      "grad_norm": 0.052163042130873555,
      "learning_rate": 0.0001553375966347029,
      "loss": 0.8509,
      "step": 93630
    },
    {
      "epoch": 1.643989536333152,
      "grad_norm": 0.06383318608177115,
      "learning_rate": 0.00015532828120246782,
      "loss": 0.8381,
      "step": 93640
    },
    {
      "epoch": 1.6441651012131533,
      "grad_norm": 0.0647405964220709,
      "learning_rate": 0.000155318965081883,
      "loss": 0.8505,
      "step": 93650
    },
    {
      "epoch": 1.6443406660931548,
      "grad_norm": 0.05256998130176413,
      "learning_rate": 0.00015530964827306648,
      "loss": 0.8454,
      "step": 93660
    },
    {
      "epoch": 1.644516230973156,
      "grad_norm": 0.07846188712848794,
      "learning_rate": 0.00015530033077613633,
      "loss": 0.8494,
      "step": 93670
    },
    {
      "epoch": 1.6446917958531575,
      "grad_norm": 0.06589822488816825,
      "learning_rate": 0.00015529101259121057,
      "loss": 0.8421,
      "step": 93680
    },
    {
      "epoch": 1.644867360733159,
      "grad_norm": 0.05689007016667135,
      "learning_rate": 0.00015528169371840727,
      "loss": 0.8471,
      "step": 93690
    },
    {
      "epoch": 1.6450429256131605,
      "grad_norm": 0.052201534093170285,
      "learning_rate": 0.0001552723741578445,
      "loss": 0.8449,
      "step": 93700
    },
    {
      "epoch": 1.6452184904931617,
      "grad_norm": 0.05236402395271451,
      "learning_rate": 0.00015526305390964036,
      "loss": 0.8493,
      "step": 93710
    },
    {
      "epoch": 1.6453940553731632,
      "grad_norm": 0.05183680664718421,
      "learning_rate": 0.00015525373297391293,
      "loss": 0.8392,
      "step": 93720
    },
    {
      "epoch": 1.6455696202531644,
      "grad_norm": 0.07425804658727077,
      "learning_rate": 0.00015524441135078028,
      "loss": 0.8501,
      "step": 93730
    },
    {
      "epoch": 1.645745185133166,
      "grad_norm": 0.04579067205453503,
      "learning_rate": 0.00015523508904036057,
      "loss": 0.8584,
      "step": 93740
    },
    {
      "epoch": 1.6459207500131674,
      "grad_norm": 0.06107426372735321,
      "learning_rate": 0.0001552257660427719,
      "loss": 0.8437,
      "step": 93750
    },
    {
      "epoch": 1.6460963148931689,
      "grad_norm": 0.05544553027530088,
      "learning_rate": 0.00015521644235813234,
      "loss": 0.8524,
      "step": 93760
    },
    {
      "epoch": 1.6462718797731701,
      "grad_norm": 0.060843913468603426,
      "learning_rate": 0.0001552071179865601,
      "loss": 0.8451,
      "step": 93770
    },
    {
      "epoch": 1.6464474446531716,
      "grad_norm": 0.059856282129230325,
      "learning_rate": 0.0001551977929281733,
      "loss": 0.8479,
      "step": 93780
    },
    {
      "epoch": 1.6466230095331729,
      "grad_norm": 0.07649819939639745,
      "learning_rate": 0.00015518846718309004,
      "loss": 0.8533,
      "step": 93790
    },
    {
      "epoch": 1.6467985744131743,
      "grad_norm": 0.06440625372464633,
      "learning_rate": 0.00015517914075142852,
      "loss": 0.8538,
      "step": 93800
    },
    {
      "epoch": 1.6469741392931758,
      "grad_norm": 0.05001188564018746,
      "learning_rate": 0.00015516981363330692,
      "loss": 0.8604,
      "step": 93810
    },
    {
      "epoch": 1.6471497041731773,
      "grad_norm": 0.07793289713735553,
      "learning_rate": 0.00015516048582884338,
      "loss": 0.8502,
      "step": 93820
    },
    {
      "epoch": 1.6473252690531786,
      "grad_norm": 0.05307295104968642,
      "learning_rate": 0.00015515115733815614,
      "loss": 0.8502,
      "step": 93830
    },
    {
      "epoch": 1.64750083393318,
      "grad_norm": 0.07381462648311013,
      "learning_rate": 0.00015514182816136332,
      "loss": 0.8482,
      "step": 93840
    },
    {
      "epoch": 1.6476763988131813,
      "grad_norm": 0.08417456990063926,
      "learning_rate": 0.00015513249829858316,
      "loss": 0.848,
      "step": 93850
    },
    {
      "epoch": 1.6478519636931828,
      "grad_norm": 0.060698290657091605,
      "learning_rate": 0.00015512316774993388,
      "loss": 0.8494,
      "step": 93860
    },
    {
      "epoch": 1.6480275285731842,
      "grad_norm": 0.051960754627348706,
      "learning_rate": 0.0001551138365155337,
      "loss": 0.8485,
      "step": 93870
    },
    {
      "epoch": 1.6482030934531857,
      "grad_norm": 0.07414041524745645,
      "learning_rate": 0.00015510450459550087,
      "loss": 0.8539,
      "step": 93880
    },
    {
      "epoch": 1.648378658333187,
      "grad_norm": 0.06263852941481925,
      "learning_rate": 0.00015509517198995353,
      "loss": 0.8516,
      "step": 93890
    },
    {
      "epoch": 1.6485542232131885,
      "grad_norm": 0.06258250408379215,
      "learning_rate": 0.00015508583869901005,
      "loss": 0.8532,
      "step": 93900
    },
    {
      "epoch": 1.6487297880931897,
      "grad_norm": 0.062179561793404114,
      "learning_rate": 0.00015507650472278858,
      "loss": 0.8512,
      "step": 93910
    },
    {
      "epoch": 1.6489053529731912,
      "grad_norm": 0.07605777211527548,
      "learning_rate": 0.0001550671700614075,
      "loss": 0.8439,
      "step": 93920
    },
    {
      "epoch": 1.6490809178531927,
      "grad_norm": 0.05049319573563785,
      "learning_rate": 0.000155057834714985,
      "loss": 0.8496,
      "step": 93930
    },
    {
      "epoch": 1.6492564827331941,
      "grad_norm": 0.05319168856763329,
      "learning_rate": 0.00015504849868363927,
      "loss": 0.8407,
      "step": 93940
    },
    {
      "epoch": 1.6494320476131956,
      "grad_norm": 0.06390376083304793,
      "learning_rate": 0.0001550391619674888,
      "loss": 0.845,
      "step": 93950
    },
    {
      "epoch": 1.6496076124931969,
      "grad_norm": 0.07940219400139815,
      "learning_rate": 0.00015502982456665177,
      "loss": 0.8412,
      "step": 93960
    },
    {
      "epoch": 1.6497831773731981,
      "grad_norm": 0.07643779432546992,
      "learning_rate": 0.00015502048648124655,
      "loss": 0.8488,
      "step": 93970
    },
    {
      "epoch": 1.6499587422531996,
      "grad_norm": 0.04814489940064188,
      "learning_rate": 0.00015501114771139133,
      "loss": 0.8423,
      "step": 93980
    },
    {
      "epoch": 1.650134307133201,
      "grad_norm": 0.05757350434904797,
      "learning_rate": 0.0001550018082572046,
      "loss": 0.8477,
      "step": 93990
    },
    {
      "epoch": 1.6503098720132026,
      "grad_norm": 0.06305465977453316,
      "learning_rate": 0.00015499246811880458,
      "loss": 0.8394,
      "step": 94000
    },
    {
      "epoch": 1.650485436893204,
      "grad_norm": 0.07196151745045651,
      "learning_rate": 0.00015498312729630962,
      "loss": 0.8429,
      "step": 94010
    },
    {
      "epoch": 1.6506610017732053,
      "grad_norm": 0.09394280001404458,
      "learning_rate": 0.0001549737857898381,
      "loss": 0.8536,
      "step": 94020
    },
    {
      "epoch": 1.6508365666532065,
      "grad_norm": 0.05935638606684838,
      "learning_rate": 0.00015496444359950834,
      "loss": 0.8501,
      "step": 94030
    },
    {
      "epoch": 1.651012131533208,
      "grad_norm": 0.06308258834723679,
      "learning_rate": 0.00015495510072543878,
      "loss": 0.8477,
      "step": 94040
    },
    {
      "epoch": 1.6511876964132095,
      "grad_norm": 0.08647834662797356,
      "learning_rate": 0.00015494575716774775,
      "loss": 0.8441,
      "step": 94050
    },
    {
      "epoch": 1.651363261293211,
      "grad_norm": 0.048104074588721006,
      "learning_rate": 0.00015493641292655362,
      "loss": 0.8494,
      "step": 94060
    },
    {
      "epoch": 1.6515388261732125,
      "grad_norm": 0.05554638306732061,
      "learning_rate": 0.00015492706800197484,
      "loss": 0.8518,
      "step": 94070
    },
    {
      "epoch": 1.6517143910532137,
      "grad_norm": 0.059080115927504874,
      "learning_rate": 0.00015491772239412973,
      "loss": 0.8491,
      "step": 94080
    },
    {
      "epoch": 1.651889955933215,
      "grad_norm": 0.06895319034438686,
      "learning_rate": 0.00015490837610313677,
      "loss": 0.8504,
      "step": 94090
    },
    {
      "epoch": 1.6520655208132164,
      "grad_norm": 0.05458873426914941,
      "learning_rate": 0.00015489902912911433,
      "loss": 0.8486,
      "step": 94100
    },
    {
      "epoch": 1.652241085693218,
      "grad_norm": 0.05560483251210775,
      "learning_rate": 0.0001548896814721809,
      "loss": 0.8488,
      "step": 94110
    },
    {
      "epoch": 1.6524166505732194,
      "grad_norm": 0.048295533471829764,
      "learning_rate": 0.00015488033313245484,
      "loss": 0.8511,
      "step": 94120
    },
    {
      "epoch": 1.6525922154532209,
      "grad_norm": 0.06449020064957284,
      "learning_rate": 0.00015487098411005464,
      "loss": 0.8448,
      "step": 94130
    },
    {
      "epoch": 1.6527677803332221,
      "grad_norm": 0.0644914040603378,
      "learning_rate": 0.00015486163440509875,
      "loss": 0.8432,
      "step": 94140
    },
    {
      "epoch": 1.6529433452132234,
      "grad_norm": 0.06695852403175741,
      "learning_rate": 0.00015485228401770565,
      "loss": 0.841,
      "step": 94150
    },
    {
      "epoch": 1.6531189100932249,
      "grad_norm": 0.060649813792337824,
      "learning_rate": 0.00015484293294799377,
      "loss": 0.8433,
      "step": 94160
    },
    {
      "epoch": 1.6532944749732263,
      "grad_norm": 0.05154235776261945,
      "learning_rate": 0.0001548335811960816,
      "loss": 0.8485,
      "step": 94170
    },
    {
      "epoch": 1.6534700398532278,
      "grad_norm": 0.0662789750813923,
      "learning_rate": 0.00015482422876208764,
      "loss": 0.8526,
      "step": 94180
    },
    {
      "epoch": 1.6536456047332293,
      "grad_norm": 0.06637706936509075,
      "learning_rate": 0.00015481487564613045,
      "loss": 0.8474,
      "step": 94190
    },
    {
      "epoch": 1.6538211696132306,
      "grad_norm": 0.05486441454981343,
      "learning_rate": 0.0001548055218483284,
      "loss": 0.8444,
      "step": 94200
    },
    {
      "epoch": 1.6539967344932318,
      "grad_norm": 0.0585510457124072,
      "learning_rate": 0.00015479616736880007,
      "loss": 0.8522,
      "step": 94210
    },
    {
      "epoch": 1.6541722993732333,
      "grad_norm": 0.06860656321016921,
      "learning_rate": 0.00015478681220766404,
      "loss": 0.846,
      "step": 94220
    },
    {
      "epoch": 1.6543478642532348,
      "grad_norm": 0.06079191009960637,
      "learning_rate": 0.00015477745636503876,
      "loss": 0.8491,
      "step": 94230
    },
    {
      "epoch": 1.6545234291332362,
      "grad_norm": 0.09852596808255608,
      "learning_rate": 0.0001547680998410428,
      "loss": 0.8491,
      "step": 94240
    },
    {
      "epoch": 1.6546989940132377,
      "grad_norm": 0.07640035708138483,
      "learning_rate": 0.00015475874263579472,
      "loss": 0.8478,
      "step": 94250
    },
    {
      "epoch": 1.654874558893239,
      "grad_norm": 0.05733989744372962,
      "learning_rate": 0.00015474938474941308,
      "loss": 0.8518,
      "step": 94260
    },
    {
      "epoch": 1.6550501237732405,
      "grad_norm": 0.061768454165044215,
      "learning_rate": 0.0001547400261820164,
      "loss": 0.8468,
      "step": 94270
    },
    {
      "epoch": 1.6552256886532417,
      "grad_norm": 0.04913559889787357,
      "learning_rate": 0.00015473066693372334,
      "loss": 0.8482,
      "step": 94280
    },
    {
      "epoch": 1.6554012535332432,
      "grad_norm": 0.050182701872418564,
      "learning_rate": 0.00015472130700465238,
      "loss": 0.8512,
      "step": 94290
    },
    {
      "epoch": 1.6555768184132447,
      "grad_norm": 0.049712644347393184,
      "learning_rate": 0.0001547119463949222,
      "loss": 0.8515,
      "step": 94300
    },
    {
      "epoch": 1.6557523832932461,
      "grad_norm": 0.052355285993523404,
      "learning_rate": 0.00015470258510465136,
      "loss": 0.838,
      "step": 94310
    },
    {
      "epoch": 1.6559279481732474,
      "grad_norm": 0.06082019438078716,
      "learning_rate": 0.00015469322313395846,
      "loss": 0.8584,
      "step": 94320
    },
    {
      "epoch": 1.6561035130532489,
      "grad_norm": 0.06196627713563566,
      "learning_rate": 0.00015468386048296212,
      "loss": 0.8416,
      "step": 94330
    },
    {
      "epoch": 1.6562790779332501,
      "grad_norm": 0.051304798110873816,
      "learning_rate": 0.00015467449715178105,
      "loss": 0.8405,
      "step": 94340
    },
    {
      "epoch": 1.6564546428132516,
      "grad_norm": 0.07634714383886805,
      "learning_rate": 0.00015466513314053374,
      "loss": 0.8432,
      "step": 94350
    },
    {
      "epoch": 1.656630207693253,
      "grad_norm": 0.05902483665988435,
      "learning_rate": 0.00015465576844933893,
      "loss": 0.8464,
      "step": 94360
    },
    {
      "epoch": 1.6568057725732546,
      "grad_norm": 0.06203872666027397,
      "learning_rate": 0.00015464640307831526,
      "loss": 0.8488,
      "step": 94370
    },
    {
      "epoch": 1.6569813374532558,
      "grad_norm": 0.05083904660498888,
      "learning_rate": 0.00015463703702758136,
      "loss": 0.8501,
      "step": 94380
    },
    {
      "epoch": 1.6571569023332573,
      "grad_norm": 0.06845376288983532,
      "learning_rate": 0.00015462767029725597,
      "loss": 0.8392,
      "step": 94390
    },
    {
      "epoch": 1.6573324672132586,
      "grad_norm": 0.07025925724014305,
      "learning_rate": 0.00015461830288745766,
      "loss": 0.8533,
      "step": 94400
    },
    {
      "epoch": 1.65750803209326,
      "grad_norm": 0.051028909821571905,
      "learning_rate": 0.0001546089347983052,
      "loss": 0.846,
      "step": 94410
    },
    {
      "epoch": 1.6576835969732615,
      "grad_norm": 0.05551652465323953,
      "learning_rate": 0.00015459956602991727,
      "loss": 0.854,
      "step": 94420
    },
    {
      "epoch": 1.657859161853263,
      "grad_norm": 0.059167172109166125,
      "learning_rate": 0.0001545901965824126,
      "loss": 0.8437,
      "step": 94430
    },
    {
      "epoch": 1.6580347267332642,
      "grad_norm": 0.07247794241456089,
      "learning_rate": 0.00015458082645590979,
      "loss": 0.8441,
      "step": 94440
    },
    {
      "epoch": 1.6582102916132657,
      "grad_norm": 0.052087652191734564,
      "learning_rate": 0.00015457145565052767,
      "loss": 0.8463,
      "step": 94450
    },
    {
      "epoch": 1.658385856493267,
      "grad_norm": 0.05552839503503336,
      "learning_rate": 0.00015456208416638495,
      "loss": 0.8465,
      "step": 94460
    },
    {
      "epoch": 1.6585614213732685,
      "grad_norm": 0.07350336977256625,
      "learning_rate": 0.00015455271200360039,
      "loss": 0.8498,
      "step": 94470
    },
    {
      "epoch": 1.65873698625327,
      "grad_norm": 0.0778836169498377,
      "learning_rate": 0.00015454333916229264,
      "loss": 0.8531,
      "step": 94480
    },
    {
      "epoch": 1.6589125511332714,
      "grad_norm": 0.046910876804091124,
      "learning_rate": 0.00015453396564258057,
      "loss": 0.849,
      "step": 94490
    },
    {
      "epoch": 1.6590881160132727,
      "grad_norm": 0.049343776155634116,
      "learning_rate": 0.00015452459144458288,
      "loss": 0.8458,
      "step": 94500
    },
    {
      "epoch": 1.6592636808932741,
      "grad_norm": 0.0640951212401291,
      "learning_rate": 0.00015451521656841837,
      "loss": 0.8481,
      "step": 94510
    },
    {
      "epoch": 1.6594392457732754,
      "grad_norm": 0.0661579133645821,
      "learning_rate": 0.0001545058410142058,
      "loss": 0.8484,
      "step": 94520
    },
    {
      "epoch": 1.6596148106532769,
      "grad_norm": 0.058245136879074184,
      "learning_rate": 0.00015449646478206397,
      "loss": 0.8506,
      "step": 94530
    },
    {
      "epoch": 1.6597903755332784,
      "grad_norm": 0.0851085771278295,
      "learning_rate": 0.00015448708787211166,
      "loss": 0.8472,
      "step": 94540
    },
    {
      "epoch": 1.6599659404132798,
      "grad_norm": 0.08560657758088117,
      "learning_rate": 0.00015447771028446772,
      "loss": 0.8473,
      "step": 94550
    },
    {
      "epoch": 1.660141505293281,
      "grad_norm": 0.04913711471625984,
      "learning_rate": 0.00015446833201925093,
      "loss": 0.85,
      "step": 94560
    },
    {
      "epoch": 1.6603170701732826,
      "grad_norm": 0.058415037585008746,
      "learning_rate": 0.00015445895307658014,
      "loss": 0.8525,
      "step": 94570
    },
    {
      "epoch": 1.6604926350532838,
      "grad_norm": 0.06051588690920928,
      "learning_rate": 0.00015444957345657416,
      "loss": 0.8434,
      "step": 94580
    },
    {
      "epoch": 1.6606681999332853,
      "grad_norm": 0.06502999798943816,
      "learning_rate": 0.00015444019315935184,
      "loss": 0.8472,
      "step": 94590
    },
    {
      "epoch": 1.6608437648132868,
      "grad_norm": 0.06547394639168876,
      "learning_rate": 0.000154430812185032,
      "loss": 0.8483,
      "step": 94600
    },
    {
      "epoch": 1.6610193296932882,
      "grad_norm": 0.06040695763740297,
      "learning_rate": 0.0001544214305337336,
      "loss": 0.8437,
      "step": 94610
    },
    {
      "epoch": 1.6611948945732895,
      "grad_norm": 0.047260642966078285,
      "learning_rate": 0.0001544120482055754,
      "loss": 0.8474,
      "step": 94620
    },
    {
      "epoch": 1.661370459453291,
      "grad_norm": 0.06451493923212984,
      "learning_rate": 0.0001544026652006763,
      "loss": 0.8557,
      "step": 94630
    },
    {
      "epoch": 1.6615460243332922,
      "grad_norm": 0.06449058626848733,
      "learning_rate": 0.00015439328151915522,
      "loss": 0.8496,
      "step": 94640
    },
    {
      "epoch": 1.6617215892132937,
      "grad_norm": 0.0505237372961502,
      "learning_rate": 0.000154383897161131,
      "loss": 0.8447,
      "step": 94650
    },
    {
      "epoch": 1.6618971540932952,
      "grad_norm": 0.04730810971482457,
      "learning_rate": 0.00015437451212672258,
      "loss": 0.8496,
      "step": 94660
    },
    {
      "epoch": 1.6620727189732967,
      "grad_norm": 0.054795461713024074,
      "learning_rate": 0.00015436512641604888,
      "loss": 0.8513,
      "step": 94670
    },
    {
      "epoch": 1.6622482838532981,
      "grad_norm": 0.057114086931881365,
      "learning_rate": 0.00015435574002922882,
      "loss": 0.8467,
      "step": 94680
    },
    {
      "epoch": 1.6624238487332994,
      "grad_norm": 0.09154235157838572,
      "learning_rate": 0.00015434635296638128,
      "loss": 0.8476,
      "step": 94690
    },
    {
      "epoch": 1.6625994136133007,
      "grad_norm": 0.04729752760629243,
      "learning_rate": 0.0001543369652276252,
      "loss": 0.843,
      "step": 94700
    },
    {
      "epoch": 1.6627749784933021,
      "grad_norm": 0.06743616115739447,
      "learning_rate": 0.0001543275768130796,
      "loss": 0.8484,
      "step": 94710
    },
    {
      "epoch": 1.6629505433733036,
      "grad_norm": 0.06486637125207566,
      "learning_rate": 0.00015431818772286335,
      "loss": 0.841,
      "step": 94720
    },
    {
      "epoch": 1.663126108253305,
      "grad_norm": 0.06785441879303486,
      "learning_rate": 0.00015430879795709542,
      "loss": 0.8497,
      "step": 94730
    },
    {
      "epoch": 1.6633016731333066,
      "grad_norm": 0.060553837762360035,
      "learning_rate": 0.00015429940751589486,
      "loss": 0.8424,
      "step": 94740
    },
    {
      "epoch": 1.6634772380133078,
      "grad_norm": 0.05016028787067897,
      "learning_rate": 0.00015429001639938057,
      "loss": 0.8412,
      "step": 94750
    },
    {
      "epoch": 1.663652802893309,
      "grad_norm": 0.05855630634702015,
      "learning_rate": 0.00015428062460767154,
      "loss": 0.8428,
      "step": 94760
    },
    {
      "epoch": 1.6638283677733106,
      "grad_norm": 0.04856941013046328,
      "learning_rate": 0.00015427123214088682,
      "loss": 0.8504,
      "step": 94770
    },
    {
      "epoch": 1.664003932653312,
      "grad_norm": 0.056211501000091715,
      "learning_rate": 0.00015426183899914535,
      "loss": 0.8399,
      "step": 94780
    },
    {
      "epoch": 1.6641794975333135,
      "grad_norm": 0.048404433527417384,
      "learning_rate": 0.0001542524451825662,
      "loss": 0.8411,
      "step": 94790
    },
    {
      "epoch": 1.664355062413315,
      "grad_norm": 0.053438917253350426,
      "learning_rate": 0.00015424305069126836,
      "loss": 0.8409,
      "step": 94800
    },
    {
      "epoch": 1.6645306272933162,
      "grad_norm": 0.05591890080352559,
      "learning_rate": 0.00015423365552537084,
      "loss": 0.8562,
      "step": 94810
    },
    {
      "epoch": 1.6647061921733175,
      "grad_norm": 0.051064000941507884,
      "learning_rate": 0.0001542242596849927,
      "loss": 0.8409,
      "step": 94820
    },
    {
      "epoch": 1.664881757053319,
      "grad_norm": 0.06567919418937265,
      "learning_rate": 0.000154214863170253,
      "loss": 0.8486,
      "step": 94830
    },
    {
      "epoch": 1.6650573219333205,
      "grad_norm": 0.07367443657191486,
      "learning_rate": 0.00015420546598127077,
      "loss": 0.8505,
      "step": 94840
    },
    {
      "epoch": 1.665232886813322,
      "grad_norm": 0.05483704316121613,
      "learning_rate": 0.0001541960681181651,
      "loss": 0.85,
      "step": 94850
    },
    {
      "epoch": 1.6654084516933234,
      "grad_norm": 0.049390431039181774,
      "learning_rate": 0.00015418666958105508,
      "loss": 0.8481,
      "step": 94860
    },
    {
      "epoch": 1.6655840165733247,
      "grad_norm": 0.054306688278250144,
      "learning_rate": 0.00015417727037005973,
      "loss": 0.8514,
      "step": 94870
    },
    {
      "epoch": 1.665759581453326,
      "grad_norm": 0.05981539102518558,
      "learning_rate": 0.00015416787048529816,
      "loss": 0.8494,
      "step": 94880
    },
    {
      "epoch": 1.6659351463333274,
      "grad_norm": 0.04216374937669482,
      "learning_rate": 0.0001541584699268895,
      "loss": 0.8429,
      "step": 94890
    },
    {
      "epoch": 1.6661107112133289,
      "grad_norm": 0.046192287740402536,
      "learning_rate": 0.00015414906869495287,
      "loss": 0.8532,
      "step": 94900
    },
    {
      "epoch": 1.6662862760933304,
      "grad_norm": 0.07041764951375896,
      "learning_rate": 0.00015413966678960733,
      "loss": 0.8448,
      "step": 94910
    },
    {
      "epoch": 1.6664618409733318,
      "grad_norm": 0.07588508886355223,
      "learning_rate": 0.00015413026421097202,
      "loss": 0.8536,
      "step": 94920
    },
    {
      "epoch": 1.666637405853333,
      "grad_norm": 0.04860613827140448,
      "learning_rate": 0.00015412086095916607,
      "loss": 0.8531,
      "step": 94930
    },
    {
      "epoch": 1.6668129707333343,
      "grad_norm": 0.0412051524729264,
      "learning_rate": 0.00015411145703430866,
      "loss": 0.8438,
      "step": 94940
    },
    {
      "epoch": 1.6669885356133358,
      "grad_norm": 0.0633930385704807,
      "learning_rate": 0.0001541020524365189,
      "loss": 0.846,
      "step": 94950
    },
    {
      "epoch": 1.6671641004933373,
      "grad_norm": 0.07066803017219622,
      "learning_rate": 0.00015409264716591595,
      "loss": 0.8474,
      "step": 94960
    },
    {
      "epoch": 1.6673396653733388,
      "grad_norm": 0.05713335731903404,
      "learning_rate": 0.000154083241222619,
      "loss": 0.8505,
      "step": 94970
    },
    {
      "epoch": 1.6675152302533403,
      "grad_norm": 0.042073746343784524,
      "learning_rate": 0.00015407383460674717,
      "loss": 0.8576,
      "step": 94980
    },
    {
      "epoch": 1.6676907951333415,
      "grad_norm": 0.06370565942288797,
      "learning_rate": 0.00015406442731841972,
      "loss": 0.8445,
      "step": 94990
    },
    {
      "epoch": 1.667866360013343,
      "grad_norm": 0.05512551013490039,
      "learning_rate": 0.00015405501935775582,
      "loss": 0.8489,
      "step": 95000
    },
    {
      "epoch": 1.6680419248933442,
      "grad_norm": 0.07430079858041694,
      "learning_rate": 0.00015404561072487462,
      "loss": 0.8487,
      "step": 95010
    },
    {
      "epoch": 1.6682174897733457,
      "grad_norm": 0.06704008878450783,
      "learning_rate": 0.00015403620141989538,
      "loss": 0.8456,
      "step": 95020
    },
    {
      "epoch": 1.6683930546533472,
      "grad_norm": 0.05088199585829153,
      "learning_rate": 0.00015402679144293732,
      "loss": 0.841,
      "step": 95030
    },
    {
      "epoch": 1.6685686195333487,
      "grad_norm": 0.07355314004863245,
      "learning_rate": 0.00015401738079411965,
      "loss": 0.8449,
      "step": 95040
    },
    {
      "epoch": 1.66874418441335,
      "grad_norm": 0.06275682817240719,
      "learning_rate": 0.00015400796947356155,
      "loss": 0.8408,
      "step": 95050
    },
    {
      "epoch": 1.6689197492933514,
      "grad_norm": 0.059573583224777196,
      "learning_rate": 0.00015399855748138237,
      "loss": 0.8518,
      "step": 95060
    },
    {
      "epoch": 1.6690953141733527,
      "grad_norm": 0.0506518320831434,
      "learning_rate": 0.0001539891448177013,
      "loss": 0.8506,
      "step": 95070
    },
    {
      "epoch": 1.6692708790533541,
      "grad_norm": 0.06426923993348818,
      "learning_rate": 0.00015397973148263762,
      "loss": 0.853,
      "step": 95080
    },
    {
      "epoch": 1.6694464439333556,
      "grad_norm": 0.05071102500148647,
      "learning_rate": 0.00015397031747631059,
      "loss": 0.8596,
      "step": 95090
    },
    {
      "epoch": 1.669622008813357,
      "grad_norm": 0.04914567003463909,
      "learning_rate": 0.00015396090279883948,
      "loss": 0.8417,
      "step": 95100
    },
    {
      "epoch": 1.6697975736933584,
      "grad_norm": 0.058386213073507505,
      "learning_rate": 0.00015395148745034356,
      "loss": 0.8382,
      "step": 95110
    },
    {
      "epoch": 1.6699731385733598,
      "grad_norm": 0.06491351858113957,
      "learning_rate": 0.00015394207143094217,
      "loss": 0.8511,
      "step": 95120
    },
    {
      "epoch": 1.670148703453361,
      "grad_norm": 0.06224113794546137,
      "learning_rate": 0.00015393265474075455,
      "loss": 0.8473,
      "step": 95130
    },
    {
      "epoch": 1.6703242683333626,
      "grad_norm": 0.0585562309824553,
      "learning_rate": 0.00015392323737990012,
      "loss": 0.8479,
      "step": 95140
    },
    {
      "epoch": 1.670499833213364,
      "grad_norm": 0.05429424706494633,
      "learning_rate": 0.00015391381934849805,
      "loss": 0.8404,
      "step": 95150
    },
    {
      "epoch": 1.6706753980933655,
      "grad_norm": 0.06843827875008218,
      "learning_rate": 0.0001539044006466678,
      "loss": 0.8495,
      "step": 95160
    },
    {
      "epoch": 1.6708509629733668,
      "grad_norm": 0.06534873784759902,
      "learning_rate": 0.00015389498127452863,
      "loss": 0.8476,
      "step": 95170
    },
    {
      "epoch": 1.6710265278533682,
      "grad_norm": 0.06606457134853548,
      "learning_rate": 0.00015388556123219994,
      "loss": 0.8475,
      "step": 95180
    },
    {
      "epoch": 1.6712020927333695,
      "grad_norm": 0.06125756569073791,
      "learning_rate": 0.000153876140519801,
      "loss": 0.8595,
      "step": 95190
    },
    {
      "epoch": 1.671377657613371,
      "grad_norm": 0.07053746352002868,
      "learning_rate": 0.00015386671913745128,
      "loss": 0.855,
      "step": 95200
    },
    {
      "epoch": 1.6715532224933725,
      "grad_norm": 0.06120545884346809,
      "learning_rate": 0.0001538572970852701,
      "loss": 0.8519,
      "step": 95210
    },
    {
      "epoch": 1.671728787373374,
      "grad_norm": 0.07838430940555614,
      "learning_rate": 0.00015384787436337682,
      "loss": 0.8429,
      "step": 95220
    },
    {
      "epoch": 1.6719043522533752,
      "grad_norm": 0.061774839789035405,
      "learning_rate": 0.0001538384509718908,
      "loss": 0.8394,
      "step": 95230
    },
    {
      "epoch": 1.6720799171333767,
      "grad_norm": 0.053520478592809305,
      "learning_rate": 0.00015382902691093154,
      "loss": 0.8461,
      "step": 95240
    },
    {
      "epoch": 1.672255482013378,
      "grad_norm": 0.058158817898913535,
      "learning_rate": 0.00015381960218061836,
      "loss": 0.8486,
      "step": 95250
    },
    {
      "epoch": 1.6724310468933794,
      "grad_norm": 0.05923940267591351,
      "learning_rate": 0.0001538101767810707,
      "loss": 0.8519,
      "step": 95260
    },
    {
      "epoch": 1.6726066117733809,
      "grad_norm": 0.04877196405435707,
      "learning_rate": 0.00015380075071240798,
      "loss": 0.8466,
      "step": 95270
    },
    {
      "epoch": 1.6727821766533824,
      "grad_norm": 0.06851786003343383,
      "learning_rate": 0.00015379132397474963,
      "loss": 0.851,
      "step": 95280
    },
    {
      "epoch": 1.6729577415333836,
      "grad_norm": 0.048048576134409215,
      "learning_rate": 0.00015378189656821507,
      "loss": 0.8451,
      "step": 95290
    },
    {
      "epoch": 1.673133306413385,
      "grad_norm": 0.06441100145075235,
      "learning_rate": 0.00015377246849292378,
      "loss": 0.8503,
      "step": 95300
    },
    {
      "epoch": 1.6733088712933863,
      "grad_norm": 0.07299931793311638,
      "learning_rate": 0.00015376303974899523,
      "loss": 0.85,
      "step": 95310
    },
    {
      "epoch": 1.6734844361733878,
      "grad_norm": 0.06843065723408207,
      "learning_rate": 0.00015375361033654881,
      "loss": 0.8428,
      "step": 95320
    },
    {
      "epoch": 1.6736600010533893,
      "grad_norm": 0.061252414512759805,
      "learning_rate": 0.00015374418025570406,
      "loss": 0.8474,
      "step": 95330
    },
    {
      "epoch": 1.6738355659333908,
      "grad_norm": 0.0684872081491556,
      "learning_rate": 0.0001537347495065804,
      "loss": 0.8444,
      "step": 95340
    },
    {
      "epoch": 1.674011130813392,
      "grad_norm": 0.0752867684652765,
      "learning_rate": 0.00015372531808929737,
      "loss": 0.8424,
      "step": 95350
    },
    {
      "epoch": 1.6741866956933935,
      "grad_norm": 0.06798165438116349,
      "learning_rate": 0.0001537158860039745,
      "loss": 0.8465,
      "step": 95360
    },
    {
      "epoch": 1.6743622605733948,
      "grad_norm": 0.06310937524097925,
      "learning_rate": 0.00015370645325073123,
      "loss": 0.8461,
      "step": 95370
    },
    {
      "epoch": 1.6745378254533962,
      "grad_norm": 0.062022582181607604,
      "learning_rate": 0.0001536970198296871,
      "loss": 0.8461,
      "step": 95380
    },
    {
      "epoch": 1.6747133903333977,
      "grad_norm": 0.06219136708838761,
      "learning_rate": 0.0001536875857409616,
      "loss": 0.8484,
      "step": 95390
    },
    {
      "epoch": 1.6748889552133992,
      "grad_norm": 0.07287754470137178,
      "learning_rate": 0.0001536781509846743,
      "loss": 0.8498,
      "step": 95400
    },
    {
      "epoch": 1.6750645200934007,
      "grad_norm": 0.06648327186205415,
      "learning_rate": 0.00015366871556094473,
      "loss": 0.8483,
      "step": 95410
    },
    {
      "epoch": 1.675240084973402,
      "grad_norm": 0.06104804623844486,
      "learning_rate": 0.00015365927946989246,
      "loss": 0.855,
      "step": 95420
    },
    {
      "epoch": 1.6754156498534032,
      "grad_norm": 0.07156987550393536,
      "learning_rate": 0.00015364984271163705,
      "loss": 0.8421,
      "step": 95430
    },
    {
      "epoch": 1.6755912147334047,
      "grad_norm": 0.05330149035287825,
      "learning_rate": 0.00015364040528629798,
      "loss": 0.8472,
      "step": 95440
    },
    {
      "epoch": 1.6757667796134061,
      "grad_norm": 0.07240900563455192,
      "learning_rate": 0.00015363096719399496,
      "loss": 0.8462,
      "step": 95450
    },
    {
      "epoch": 1.6759423444934076,
      "grad_norm": 0.04792724640144473,
      "learning_rate": 0.00015362152843484742,
      "loss": 0.8483,
      "step": 95460
    },
    {
      "epoch": 1.676117909373409,
      "grad_norm": 0.05573833506318755,
      "learning_rate": 0.00015361208900897507,
      "loss": 0.8497,
      "step": 95470
    },
    {
      "epoch": 1.6762934742534104,
      "grad_norm": 0.06940693850166915,
      "learning_rate": 0.00015360264891649747,
      "loss": 0.8496,
      "step": 95480
    },
    {
      "epoch": 1.6764690391334116,
      "grad_norm": 0.05330598500458181,
      "learning_rate": 0.0001535932081575342,
      "loss": 0.8535,
      "step": 95490
    },
    {
      "epoch": 1.676644604013413,
      "grad_norm": 0.06355062326426408,
      "learning_rate": 0.00015358376673220493,
      "loss": 0.8496,
      "step": 95500
    },
    {
      "epoch": 1.6768201688934146,
      "grad_norm": 0.11479060705281197,
      "learning_rate": 0.00015357432464062926,
      "loss": 0.8479,
      "step": 95510
    },
    {
      "epoch": 1.676995733773416,
      "grad_norm": 0.04680207843694712,
      "learning_rate": 0.00015356488188292681,
      "loss": 0.8499,
      "step": 95520
    },
    {
      "epoch": 1.6771712986534175,
      "grad_norm": 0.05439626970373758,
      "learning_rate": 0.00015355543845921722,
      "loss": 0.8517,
      "step": 95530
    },
    {
      "epoch": 1.6773468635334188,
      "grad_norm": 0.05456687250982142,
      "learning_rate": 0.0001535459943696202,
      "loss": 0.8546,
      "step": 95540
    },
    {
      "epoch": 1.67752242841342,
      "grad_norm": 0.053634909926602345,
      "learning_rate": 0.00015353654961425533,
      "loss": 0.8517,
      "step": 95550
    },
    {
      "epoch": 1.6776979932934215,
      "grad_norm": 0.0773923600134467,
      "learning_rate": 0.00015352710419324231,
      "loss": 0.8527,
      "step": 95560
    },
    {
      "epoch": 1.677873558173423,
      "grad_norm": 0.06606235920627306,
      "learning_rate": 0.00015351765810670077,
      "loss": 0.8454,
      "step": 95570
    },
    {
      "epoch": 1.6780491230534245,
      "grad_norm": 0.05446328326585612,
      "learning_rate": 0.00015350821135475052,
      "loss": 0.8425,
      "step": 95580
    },
    {
      "epoch": 1.678224687933426,
      "grad_norm": 0.06313387935048015,
      "learning_rate": 0.00015349876393751114,
      "loss": 0.8529,
      "step": 95590
    },
    {
      "epoch": 1.6784002528134272,
      "grad_norm": 0.08051191038701772,
      "learning_rate": 0.00015348931585510233,
      "loss": 0.846,
      "step": 95600
    },
    {
      "epoch": 1.6785758176934285,
      "grad_norm": 0.05237718658569321,
      "learning_rate": 0.00015347986710764386,
      "loss": 0.8515,
      "step": 95610
    },
    {
      "epoch": 1.67875138257343,
      "grad_norm": 0.05203973058781705,
      "learning_rate": 0.0001534704176952554,
      "loss": 0.8462,
      "step": 95620
    },
    {
      "epoch": 1.6789269474534314,
      "grad_norm": 0.07672918258682997,
      "learning_rate": 0.00015346096761805668,
      "loss": 0.8499,
      "step": 95630
    },
    {
      "epoch": 1.6791025123334329,
      "grad_norm": 0.08622608620343247,
      "learning_rate": 0.00015345151687616747,
      "loss": 0.8549,
      "step": 95640
    },
    {
      "epoch": 1.6792780772134344,
      "grad_norm": 0.07599902932980918,
      "learning_rate": 0.00015344206546970748,
      "loss": 0.8494,
      "step": 95650
    },
    {
      "epoch": 1.6794536420934356,
      "grad_norm": 0.05353369189397644,
      "learning_rate": 0.00015343261339879644,
      "loss": 0.8456,
      "step": 95660
    },
    {
      "epoch": 1.6796292069734369,
      "grad_norm": 0.1123674735386603,
      "learning_rate": 0.0001534231606635542,
      "loss": 0.8517,
      "step": 95670
    },
    {
      "epoch": 1.6798047718534383,
      "grad_norm": 0.058523709188148636,
      "learning_rate": 0.0001534137072641004,
      "loss": 0.8535,
      "step": 95680
    },
    {
      "epoch": 1.6799803367334398,
      "grad_norm": 0.07731497630249619,
      "learning_rate": 0.0001534042532005549,
      "loss": 0.8472,
      "step": 95690
    },
    {
      "epoch": 1.6801559016134413,
      "grad_norm": 0.06224062651485618,
      "learning_rate": 0.0001533947984730375,
      "loss": 0.8471,
      "step": 95700
    },
    {
      "epoch": 1.6803314664934428,
      "grad_norm": 0.09533843131799141,
      "learning_rate": 0.00015338534308166797,
      "loss": 0.8445,
      "step": 95710
    },
    {
      "epoch": 1.680507031373444,
      "grad_norm": 0.06362895128683026,
      "learning_rate": 0.00015337588702656608,
      "loss": 0.8527,
      "step": 95720
    },
    {
      "epoch": 1.6806825962534455,
      "grad_norm": 0.05679390942250731,
      "learning_rate": 0.00015336643030785166,
      "loss": 0.8476,
      "step": 95730
    },
    {
      "epoch": 1.6808581611334468,
      "grad_norm": 0.05673447953000295,
      "learning_rate": 0.00015335697292564454,
      "loss": 0.8458,
      "step": 95740
    },
    {
      "epoch": 1.6810337260134482,
      "grad_norm": 0.05443225561591259,
      "learning_rate": 0.00015334751488006455,
      "loss": 0.8438,
      "step": 95750
    },
    {
      "epoch": 1.6812092908934497,
      "grad_norm": 0.05055958965335517,
      "learning_rate": 0.0001533380561712315,
      "loss": 0.8422,
      "step": 95760
    },
    {
      "epoch": 1.6813848557734512,
      "grad_norm": 0.0678798131184224,
      "learning_rate": 0.00015332859679926525,
      "loss": 0.855,
      "step": 95770
    },
    {
      "epoch": 1.6815604206534525,
      "grad_norm": 0.06453100921263666,
      "learning_rate": 0.00015331913676428565,
      "loss": 0.851,
      "step": 95780
    },
    {
      "epoch": 1.681735985533454,
      "grad_norm": 0.04748940838964168,
      "learning_rate": 0.00015330967606641256,
      "loss": 0.854,
      "step": 95790
    },
    {
      "epoch": 1.6819115504134552,
      "grad_norm": 0.0753851517040027,
      "learning_rate": 0.00015330021470576585,
      "loss": 0.8461,
      "step": 95800
    },
    {
      "epoch": 1.6820871152934567,
      "grad_norm": 0.0911950721169273,
      "learning_rate": 0.0001532907526824654,
      "loss": 0.8463,
      "step": 95810
    },
    {
      "epoch": 1.6822626801734581,
      "grad_norm": 0.07516061435204335,
      "learning_rate": 0.00015328128999663108,
      "loss": 0.8538,
      "step": 95820
    },
    {
      "epoch": 1.6824382450534596,
      "grad_norm": 0.059799068955432445,
      "learning_rate": 0.0001532718266483828,
      "loss": 0.851,
      "step": 95830
    },
    {
      "epoch": 1.6826138099334609,
      "grad_norm": 0.057756081694793404,
      "learning_rate": 0.00015326236263784045,
      "loss": 0.8485,
      "step": 95840
    },
    {
      "epoch": 1.6827893748134624,
      "grad_norm": 0.05902145850088131,
      "learning_rate": 0.000153252897965124,
      "loss": 0.8534,
      "step": 95850
    },
    {
      "epoch": 1.6829649396934636,
      "grad_norm": 0.07342560284232039,
      "learning_rate": 0.00015324343263035324,
      "loss": 0.8449,
      "step": 95860
    },
    {
      "epoch": 1.683140504573465,
      "grad_norm": 0.052068129114703966,
      "learning_rate": 0.00015323396663364822,
      "loss": 0.8446,
      "step": 95870
    },
    {
      "epoch": 1.6833160694534666,
      "grad_norm": 0.058193405558772726,
      "learning_rate": 0.00015322449997512883,
      "loss": 0.8388,
      "step": 95880
    },
    {
      "epoch": 1.683491634333468,
      "grad_norm": 0.049458443507647006,
      "learning_rate": 0.00015321503265491504,
      "loss": 0.8509,
      "step": 95890
    },
    {
      "epoch": 1.6836671992134693,
      "grad_norm": 0.04420036819366715,
      "learning_rate": 0.00015320556467312673,
      "loss": 0.8509,
      "step": 95900
    },
    {
      "epoch": 1.6838427640934708,
      "grad_norm": 0.04675505073036924,
      "learning_rate": 0.00015319609602988394,
      "loss": 0.8521,
      "step": 95910
    },
    {
      "epoch": 1.684018328973472,
      "grad_norm": 0.05789436408873242,
      "learning_rate": 0.0001531866267253066,
      "loss": 0.8435,
      "step": 95920
    },
    {
      "epoch": 1.6841938938534735,
      "grad_norm": 0.05395577530555509,
      "learning_rate": 0.0001531771567595147,
      "loss": 0.8523,
      "step": 95930
    },
    {
      "epoch": 1.684369458733475,
      "grad_norm": 0.07056038849606883,
      "learning_rate": 0.00015316768613262823,
      "loss": 0.8434,
      "step": 95940
    },
    {
      "epoch": 1.6845450236134765,
      "grad_norm": 0.04684354664580664,
      "learning_rate": 0.00015315821484476717,
      "loss": 0.8444,
      "step": 95950
    },
    {
      "epoch": 1.6847205884934777,
      "grad_norm": 0.049010913957714175,
      "learning_rate": 0.00015314874289605157,
      "loss": 0.8444,
      "step": 95960
    },
    {
      "epoch": 1.6848961533734792,
      "grad_norm": 0.04398224562191143,
      "learning_rate": 0.00015313927028660134,
      "loss": 0.8612,
      "step": 95970
    },
    {
      "epoch": 1.6850717182534805,
      "grad_norm": 0.08558575666911754,
      "learning_rate": 0.00015312979701653657,
      "loss": 0.8522,
      "step": 95980
    },
    {
      "epoch": 1.685247283133482,
      "grad_norm": 0.07735970110752302,
      "learning_rate": 0.00015312032308597735,
      "loss": 0.8415,
      "step": 95990
    },
    {
      "epoch": 1.6854228480134834,
      "grad_norm": 0.0646837853224522,
      "learning_rate": 0.00015311084849504357,
      "loss": 0.8596,
      "step": 96000
    },
    {
      "epoch": 1.6855984128934849,
      "grad_norm": 0.058455583545565896,
      "learning_rate": 0.00015310137324385542,
      "loss": 0.8446,
      "step": 96010
    },
    {
      "epoch": 1.6857739777734861,
      "grad_norm": 0.056098470236429394,
      "learning_rate": 0.00015309189733253282,
      "loss": 0.8524,
      "step": 96020
    },
    {
      "epoch": 1.6859495426534876,
      "grad_norm": 0.0476578238272854,
      "learning_rate": 0.00015308242076119595,
      "loss": 0.8512,
      "step": 96030
    },
    {
      "epoch": 1.6861251075334889,
      "grad_norm": 0.05147881136538946,
      "learning_rate": 0.0001530729435299648,
      "loss": 0.8487,
      "step": 96040
    },
    {
      "epoch": 1.6863006724134904,
      "grad_norm": 0.05665635325463302,
      "learning_rate": 0.0001530634656389595,
      "loss": 0.8457,
      "step": 96050
    },
    {
      "epoch": 1.6864762372934918,
      "grad_norm": 0.07736988525648364,
      "learning_rate": 0.00015305398708830007,
      "loss": 0.8345,
      "step": 96060
    },
    {
      "epoch": 1.6866518021734933,
      "grad_norm": 0.056845483251915935,
      "learning_rate": 0.00015304450787810667,
      "loss": 0.8478,
      "step": 96070
    },
    {
      "epoch": 1.6868273670534946,
      "grad_norm": 0.08621970172970536,
      "learning_rate": 0.00015303502800849937,
      "loss": 0.8561,
      "step": 96080
    },
    {
      "epoch": 1.687002931933496,
      "grad_norm": 0.06612637987095707,
      "learning_rate": 0.00015302554747959832,
      "loss": 0.8456,
      "step": 96090
    },
    {
      "epoch": 1.6871784968134973,
      "grad_norm": 0.05269822690879243,
      "learning_rate": 0.0001530160662915236,
      "loss": 0.8562,
      "step": 96100
    },
    {
      "epoch": 1.6873540616934988,
      "grad_norm": 0.05247681482313725,
      "learning_rate": 0.00015300658444439535,
      "loss": 0.8535,
      "step": 96110
    },
    {
      "epoch": 1.6875296265735003,
      "grad_norm": 0.05520266802168007,
      "learning_rate": 0.0001529971019383337,
      "loss": 0.8499,
      "step": 96120
    },
    {
      "epoch": 1.6877051914535017,
      "grad_norm": 0.05079496466075705,
      "learning_rate": 0.0001529876187734588,
      "loss": 0.8482,
      "step": 96130
    },
    {
      "epoch": 1.6878807563335032,
      "grad_norm": 0.1031477037999305,
      "learning_rate": 0.00015297813494989082,
      "loss": 0.8481,
      "step": 96140
    },
    {
      "epoch": 1.6880563212135045,
      "grad_norm": 0.04956608870164817,
      "learning_rate": 0.00015296865046774988,
      "loss": 0.8423,
      "step": 96150
    },
    {
      "epoch": 1.6882318860935057,
      "grad_norm": 0.06597760119517432,
      "learning_rate": 0.00015295916532715625,
      "loss": 0.8501,
      "step": 96160
    },
    {
      "epoch": 1.6884074509735072,
      "grad_norm": 0.08706964395327943,
      "learning_rate": 0.00015294967952822998,
      "loss": 0.8443,
      "step": 96170
    },
    {
      "epoch": 1.6885830158535087,
      "grad_norm": 0.055574154082522496,
      "learning_rate": 0.00015294019307109132,
      "loss": 0.8457,
      "step": 96180
    },
    {
      "epoch": 1.6887585807335102,
      "grad_norm": 0.05817518412213656,
      "learning_rate": 0.0001529307059558605,
      "loss": 0.8486,
      "step": 96190
    },
    {
      "epoch": 1.6889341456135116,
      "grad_norm": 0.05807971053085143,
      "learning_rate": 0.00015292121818265758,
      "loss": 0.85,
      "step": 96200
    },
    {
      "epoch": 1.6891097104935129,
      "grad_norm": 0.07352018798856931,
      "learning_rate": 0.00015291172975160294,
      "loss": 0.8486,
      "step": 96210
    },
    {
      "epoch": 1.6892852753735141,
      "grad_norm": 0.06676868817104249,
      "learning_rate": 0.00015290224066281675,
      "loss": 0.8414,
      "step": 96220
    },
    {
      "epoch": 1.6894608402535156,
      "grad_norm": 0.055505827217964876,
      "learning_rate": 0.00015289275091641925,
      "loss": 0.8575,
      "step": 96230
    },
    {
      "epoch": 1.689636405133517,
      "grad_norm": 0.09653288313157303,
      "learning_rate": 0.00015288326051253058,
      "loss": 0.855,
      "step": 96240
    },
    {
      "epoch": 1.6898119700135186,
      "grad_norm": 0.06005540229185079,
      "learning_rate": 0.0001528737694512711,
      "loss": 0.8473,
      "step": 96250
    },
    {
      "epoch": 1.68998753489352,
      "grad_norm": 0.05445235289326103,
      "learning_rate": 0.000152864277732761,
      "loss": 0.8519,
      "step": 96260
    },
    {
      "epoch": 1.6901630997735213,
      "grad_norm": 0.061782429595853464,
      "learning_rate": 0.00015285478535712057,
      "loss": 0.8457,
      "step": 96270
    },
    {
      "epoch": 1.6903386646535226,
      "grad_norm": 0.0655046105703986,
      "learning_rate": 0.0001528452923244701,
      "loss": 0.8509,
      "step": 96280
    },
    {
      "epoch": 1.690514229533524,
      "grad_norm": 0.06686896823782812,
      "learning_rate": 0.00015283579863492982,
      "loss": 0.8429,
      "step": 96290
    },
    {
      "epoch": 1.6906897944135255,
      "grad_norm": 0.08345840985561094,
      "learning_rate": 0.00015282630428862002,
      "loss": 0.846,
      "step": 96300
    },
    {
      "epoch": 1.690865359293527,
      "grad_norm": 0.05598847692007748,
      "learning_rate": 0.000152816809285661,
      "loss": 0.8491,
      "step": 96310
    },
    {
      "epoch": 1.6910409241735285,
      "grad_norm": 0.05575537931613184,
      "learning_rate": 0.0001528073136261731,
      "loss": 0.8512,
      "step": 96320
    },
    {
      "epoch": 1.6912164890535297,
      "grad_norm": 0.06171842138710155,
      "learning_rate": 0.00015279781731027662,
      "loss": 0.8456,
      "step": 96330
    },
    {
      "epoch": 1.691392053933531,
      "grad_norm": 0.04751745085695837,
      "learning_rate": 0.00015278832033809183,
      "loss": 0.8427,
      "step": 96340
    },
    {
      "epoch": 1.6915676188135325,
      "grad_norm": 0.05778163087762041,
      "learning_rate": 0.0001527788227097391,
      "loss": 0.8453,
      "step": 96350
    },
    {
      "epoch": 1.691743183693534,
      "grad_norm": 0.05416033228935911,
      "learning_rate": 0.0001527693244253388,
      "loss": 0.8448,
      "step": 96360
    },
    {
      "epoch": 1.6919187485735354,
      "grad_norm": 0.04948214536409534,
      "learning_rate": 0.00015275982548501122,
      "loss": 0.8516,
      "step": 96370
    },
    {
      "epoch": 1.692094313453537,
      "grad_norm": 0.05417735215233099,
      "learning_rate": 0.00015275032588887673,
      "loss": 0.8399,
      "step": 96380
    },
    {
      "epoch": 1.6922698783335381,
      "grad_norm": 0.05413940771508463,
      "learning_rate": 0.0001527408256370557,
      "loss": 0.8495,
      "step": 96390
    },
    {
      "epoch": 1.6924454432135394,
      "grad_norm": 0.06796335657124483,
      "learning_rate": 0.0001527313247296685,
      "loss": 0.8425,
      "step": 96400
    },
    {
      "epoch": 1.6926210080935409,
      "grad_norm": 0.04484990299952155,
      "learning_rate": 0.00015272182316683548,
      "loss": 0.8401,
      "step": 96410
    },
    {
      "epoch": 1.6927965729735424,
      "grad_norm": 0.052560842892518,
      "learning_rate": 0.00015271232094867706,
      "loss": 0.8517,
      "step": 96420
    },
    {
      "epoch": 1.6929721378535438,
      "grad_norm": 0.05599336106732716,
      "learning_rate": 0.0001527028180753136,
      "loss": 0.8434,
      "step": 96430
    },
    {
      "epoch": 1.6931477027335453,
      "grad_norm": 0.05842751937618904,
      "learning_rate": 0.00015269331454686555,
      "loss": 0.8493,
      "step": 96440
    },
    {
      "epoch": 1.6933232676135466,
      "grad_norm": 0.060900726305751905,
      "learning_rate": 0.0001526838103634533,
      "loss": 0.8446,
      "step": 96450
    },
    {
      "epoch": 1.693498832493548,
      "grad_norm": 0.0551389171565461,
      "learning_rate": 0.00015267430552519726,
      "loss": 0.8573,
      "step": 96460
    },
    {
      "epoch": 1.6936743973735493,
      "grad_norm": 0.06546728634941128,
      "learning_rate": 0.00015266480003221786,
      "loss": 0.8486,
      "step": 96470
    },
    {
      "epoch": 1.6938499622535508,
      "grad_norm": 0.08209163801877611,
      "learning_rate": 0.00015265529388463556,
      "loss": 0.8452,
      "step": 96480
    },
    {
      "epoch": 1.6940255271335523,
      "grad_norm": 0.05266890950945589,
      "learning_rate": 0.00015264578708257078,
      "loss": 0.8433,
      "step": 96490
    },
    {
      "epoch": 1.6942010920135537,
      "grad_norm": 0.10191373662920002,
      "learning_rate": 0.00015263627962614398,
      "loss": 0.8507,
      "step": 96500
    },
    {
      "epoch": 1.694376656893555,
      "grad_norm": 0.11211153293079006,
      "learning_rate": 0.00015262677151547558,
      "loss": 0.8494,
      "step": 96510
    },
    {
      "epoch": 1.6945522217735565,
      "grad_norm": 0.06401182431473064,
      "learning_rate": 0.00015261726275068616,
      "loss": 0.8451,
      "step": 96520
    },
    {
      "epoch": 1.6947277866535577,
      "grad_norm": 0.046550528789999414,
      "learning_rate": 0.00015260775333189607,
      "loss": 0.8421,
      "step": 96530
    },
    {
      "epoch": 1.6949033515335592,
      "grad_norm": 0.08728222971297829,
      "learning_rate": 0.00015259824325922585,
      "loss": 0.8447,
      "step": 96540
    },
    {
      "epoch": 1.6950789164135607,
      "grad_norm": 0.06498825139418961,
      "learning_rate": 0.00015258873253279606,
      "loss": 0.8374,
      "step": 96550
    },
    {
      "epoch": 1.6952544812935622,
      "grad_norm": 0.06957384071261191,
      "learning_rate": 0.00015257922115272712,
      "loss": 0.8452,
      "step": 96560
    },
    {
      "epoch": 1.6954300461735634,
      "grad_norm": 0.062241826806331606,
      "learning_rate": 0.00015256970911913957,
      "loss": 0.8363,
      "step": 96570
    },
    {
      "epoch": 1.6956056110535649,
      "grad_norm": 0.06415715931939064,
      "learning_rate": 0.0001525601964321539,
      "loss": 0.8493,
      "step": 96580
    },
    {
      "epoch": 1.6957811759335661,
      "grad_norm": 0.045724100671040026,
      "learning_rate": 0.00015255068309189064,
      "loss": 0.8571,
      "step": 96590
    },
    {
      "epoch": 1.6959567408135676,
      "grad_norm": 0.04681718095428107,
      "learning_rate": 0.00015254116909847038,
      "loss": 0.8393,
      "step": 96600
    },
    {
      "epoch": 1.696132305693569,
      "grad_norm": 0.06634265107436811,
      "learning_rate": 0.00015253165445201363,
      "loss": 0.8424,
      "step": 96610
    },
    {
      "epoch": 1.6963078705735706,
      "grad_norm": 0.05459139069820399,
      "learning_rate": 0.00015252213915264092,
      "loss": 0.8439,
      "step": 96620
    },
    {
      "epoch": 1.6964834354535718,
      "grad_norm": 0.0602764804554293,
      "learning_rate": 0.00015251262320047283,
      "loss": 0.8535,
      "step": 96630
    },
    {
      "epoch": 1.6966590003335733,
      "grad_norm": 0.04821388025388649,
      "learning_rate": 0.00015250310659562996,
      "loss": 0.8458,
      "step": 96640
    },
    {
      "epoch": 1.6968345652135746,
      "grad_norm": 0.05228930550179565,
      "learning_rate": 0.00015249358933823283,
      "loss": 0.846,
      "step": 96650
    },
    {
      "epoch": 1.697010130093576,
      "grad_norm": 0.04887150140854072,
      "learning_rate": 0.00015248407142840208,
      "loss": 0.8494,
      "step": 96660
    },
    {
      "epoch": 1.6971856949735775,
      "grad_norm": 0.06473050772818545,
      "learning_rate": 0.0001524745528662583,
      "loss": 0.8501,
      "step": 96670
    },
    {
      "epoch": 1.697361259853579,
      "grad_norm": 0.08292974962165586,
      "learning_rate": 0.00015246503365192203,
      "loss": 0.8492,
      "step": 96680
    },
    {
      "epoch": 1.6975368247335803,
      "grad_norm": 0.054037633892485455,
      "learning_rate": 0.00015245551378551393,
      "loss": 0.841,
      "step": 96690
    },
    {
      "epoch": 1.6977123896135817,
      "grad_norm": 0.06203524927204946,
      "learning_rate": 0.0001524459932671546,
      "loss": 0.8565,
      "step": 96700
    },
    {
      "epoch": 1.697887954493583,
      "grad_norm": 0.052700555943860895,
      "learning_rate": 0.00015243647209696467,
      "loss": 0.854,
      "step": 96710
    },
    {
      "epoch": 1.6980635193735845,
      "grad_norm": 0.05695772613794107,
      "learning_rate": 0.00015242695027506482,
      "loss": 0.8507,
      "step": 96720
    },
    {
      "epoch": 1.698239084253586,
      "grad_norm": 0.0747400954506372,
      "learning_rate": 0.00015241742780157562,
      "loss": 0.8442,
      "step": 96730
    },
    {
      "epoch": 1.6984146491335874,
      "grad_norm": 0.062193536117496305,
      "learning_rate": 0.00015240790467661774,
      "loss": 0.8416,
      "step": 96740
    },
    {
      "epoch": 1.6985902140135887,
      "grad_norm": 0.055259148204471024,
      "learning_rate": 0.00015239838090031187,
      "loss": 0.8437,
      "step": 96750
    },
    {
      "epoch": 1.6987657788935902,
      "grad_norm": 0.04530963705258118,
      "learning_rate": 0.0001523888564727787,
      "loss": 0.859,
      "step": 96760
    },
    {
      "epoch": 1.6989413437735914,
      "grad_norm": 0.06705087334145746,
      "learning_rate": 0.00015237933139413881,
      "loss": 0.8461,
      "step": 96770
    },
    {
      "epoch": 1.6991169086535929,
      "grad_norm": 0.05474999519941859,
      "learning_rate": 0.00015236980566451299,
      "loss": 0.8493,
      "step": 96780
    },
    {
      "epoch": 1.6992924735335944,
      "grad_norm": 0.057956140036380936,
      "learning_rate": 0.0001523602792840219,
      "loss": 0.8454,
      "step": 96790
    },
    {
      "epoch": 1.6994680384135958,
      "grad_norm": 0.0495773080281708,
      "learning_rate": 0.0001523507522527862,
      "loss": 0.8452,
      "step": 96800
    },
    {
      "epoch": 1.699643603293597,
      "grad_norm": 0.046600082266349645,
      "learning_rate": 0.00015234122457092662,
      "loss": 0.8498,
      "step": 96810
    },
    {
      "epoch": 1.6998191681735986,
      "grad_norm": 0.059998449426231185,
      "learning_rate": 0.00015233169623856392,
      "loss": 0.8496,
      "step": 96820
    },
    {
      "epoch": 1.6999947330535998,
      "grad_norm": 0.057900225516074384,
      "learning_rate": 0.00015232216725581877,
      "loss": 0.8429,
      "step": 96830
    },
    {
      "epoch": 1.7001702979336013,
      "grad_norm": 0.056153029710077094,
      "learning_rate": 0.0001523126376228119,
      "loss": 0.8466,
      "step": 96840
    },
    {
      "epoch": 1.7003458628136028,
      "grad_norm": 0.06070327524053241,
      "learning_rate": 0.00015230310733966408,
      "loss": 0.8485,
      "step": 96850
    },
    {
      "epoch": 1.7005214276936043,
      "grad_norm": 0.060566344893630684,
      "learning_rate": 0.00015229357640649611,
      "loss": 0.8455,
      "step": 96860
    },
    {
      "epoch": 1.7006969925736057,
      "grad_norm": 0.04643852814790019,
      "learning_rate": 0.00015228404482342867,
      "loss": 0.8557,
      "step": 96870
    },
    {
      "epoch": 1.700872557453607,
      "grad_norm": 0.04553816657605914,
      "learning_rate": 0.00015227451259058251,
      "loss": 0.8428,
      "step": 96880
    },
    {
      "epoch": 1.7010481223336082,
      "grad_norm": 0.07981304854822649,
      "learning_rate": 0.00015226497970807852,
      "loss": 0.8532,
      "step": 96890
    },
    {
      "epoch": 1.7012236872136097,
      "grad_norm": 0.05849282962822837,
      "learning_rate": 0.00015225544617603734,
      "loss": 0.8471,
      "step": 96900
    },
    {
      "epoch": 1.7013992520936112,
      "grad_norm": 0.05426745131305374,
      "learning_rate": 0.00015224591199457992,
      "loss": 0.8484,
      "step": 96910
    },
    {
      "epoch": 1.7015748169736127,
      "grad_norm": 0.06218703066181585,
      "learning_rate": 0.0001522363771638269,
      "loss": 0.8486,
      "step": 96920
    },
    {
      "epoch": 1.7017503818536142,
      "grad_norm": 0.08791812568717294,
      "learning_rate": 0.0001522268416838992,
      "loss": 0.847,
      "step": 96930
    },
    {
      "epoch": 1.7019259467336154,
      "grad_norm": 0.06566558096283807,
      "learning_rate": 0.00015221730555491756,
      "loss": 0.8449,
      "step": 96940
    },
    {
      "epoch": 1.7021015116136167,
      "grad_norm": 0.09674042422015741,
      "learning_rate": 0.00015220776877700287,
      "loss": 0.8433,
      "step": 96950
    },
    {
      "epoch": 1.7022770764936181,
      "grad_norm": 0.04248990719558457,
      "learning_rate": 0.00015219823135027597,
      "loss": 0.842,
      "step": 96960
    },
    {
      "epoch": 1.7024526413736196,
      "grad_norm": 0.057485654677226945,
      "learning_rate": 0.00015218869327485762,
      "loss": 0.847,
      "step": 96970
    },
    {
      "epoch": 1.702628206253621,
      "grad_norm": 0.0574137770080803,
      "learning_rate": 0.00015217915455086872,
      "loss": 0.8499,
      "step": 96980
    },
    {
      "epoch": 1.7028037711336226,
      "grad_norm": 0.05542091880855439,
      "learning_rate": 0.0001521696151784301,
      "loss": 0.8493,
      "step": 96990
    },
    {
      "epoch": 1.7029793360136238,
      "grad_norm": 0.06630582203921802,
      "learning_rate": 0.0001521600751576627,
      "loss": 0.8561,
      "step": 97000
    },
    {
      "epoch": 1.703154900893625,
      "grad_norm": 0.07199614064867678,
      "learning_rate": 0.00015215053448868734,
      "loss": 0.8532,
      "step": 97010
    },
    {
      "epoch": 1.7033304657736266,
      "grad_norm": 0.05367124904852347,
      "learning_rate": 0.00015214099317162487,
      "loss": 0.8418,
      "step": 97020
    },
    {
      "epoch": 1.703506030653628,
      "grad_norm": 0.04846972956662557,
      "learning_rate": 0.00015213145120659624,
      "loss": 0.8468,
      "step": 97030
    },
    {
      "epoch": 1.7036815955336295,
      "grad_norm": 0.049354321984957074,
      "learning_rate": 0.0001521219085937223,
      "loss": 0.8399,
      "step": 97040
    },
    {
      "epoch": 1.703857160413631,
      "grad_norm": 0.058078207010873184,
      "learning_rate": 0.000152112365333124,
      "loss": 0.856,
      "step": 97050
    },
    {
      "epoch": 1.7040327252936323,
      "grad_norm": 0.07702379027825147,
      "learning_rate": 0.00015210282142492224,
      "loss": 0.8434,
      "step": 97060
    },
    {
      "epoch": 1.7042082901736335,
      "grad_norm": 0.05348775652422329,
      "learning_rate": 0.00015209327686923794,
      "loss": 0.8458,
      "step": 97070
    },
    {
      "epoch": 1.704383855053635,
      "grad_norm": 0.05755630910175806,
      "learning_rate": 0.00015208373166619203,
      "loss": 0.8537,
      "step": 97080
    },
    {
      "epoch": 1.7045594199336365,
      "grad_norm": 0.04905357836689019,
      "learning_rate": 0.00015207418581590542,
      "loss": 0.8549,
      "step": 97090
    },
    {
      "epoch": 1.704734984813638,
      "grad_norm": 0.053901674504769174,
      "learning_rate": 0.00015206463931849911,
      "loss": 0.848,
      "step": 97100
    },
    {
      "epoch": 1.7049105496936394,
      "grad_norm": 0.04922649370274321,
      "learning_rate": 0.00015205509217409402,
      "loss": 0.8471,
      "step": 97110
    },
    {
      "epoch": 1.7050861145736407,
      "grad_norm": 0.07326526152282936,
      "learning_rate": 0.00015204554438281117,
      "loss": 0.842,
      "step": 97120
    },
    {
      "epoch": 1.705261679453642,
      "grad_norm": 0.06873451698597038,
      "learning_rate": 0.00015203599594477144,
      "loss": 0.8545,
      "step": 97130
    },
    {
      "epoch": 1.7054372443336434,
      "grad_norm": 0.05090565020169309,
      "learning_rate": 0.0001520264468600959,
      "loss": 0.8424,
      "step": 97140
    },
    {
      "epoch": 1.7056128092136449,
      "grad_norm": 0.05552458696105793,
      "learning_rate": 0.00015201689712890547,
      "loss": 0.8478,
      "step": 97150
    },
    {
      "epoch": 1.7057883740936464,
      "grad_norm": 0.0640305447375787,
      "learning_rate": 0.0001520073467513212,
      "loss": 0.8489,
      "step": 97160
    },
    {
      "epoch": 1.7059639389736478,
      "grad_norm": 0.05627944546993739,
      "learning_rate": 0.00015199779572746405,
      "loss": 0.8354,
      "step": 97170
    },
    {
      "epoch": 1.706139503853649,
      "grad_norm": 0.05628907161396378,
      "learning_rate": 0.00015198824405745505,
      "loss": 0.8436,
      "step": 97180
    },
    {
      "epoch": 1.7063150687336506,
      "grad_norm": 0.07530609252913746,
      "learning_rate": 0.00015197869174141525,
      "loss": 0.8422,
      "step": 97190
    },
    {
      "epoch": 1.7064906336136518,
      "grad_norm": 0.051074991984839745,
      "learning_rate": 0.00015196913877946563,
      "loss": 0.8418,
      "step": 97200
    },
    {
      "epoch": 1.7066661984936533,
      "grad_norm": 0.04579567871685326,
      "learning_rate": 0.00015195958517172725,
      "loss": 0.8507,
      "step": 97210
    },
    {
      "epoch": 1.7068417633736548,
      "grad_norm": 0.057851434183724154,
      "learning_rate": 0.00015195003091832116,
      "loss": 0.8466,
      "step": 97220
    },
    {
      "epoch": 1.7070173282536563,
      "grad_norm": 0.0718946159506703,
      "learning_rate": 0.00015194047601936847,
      "loss": 0.8409,
      "step": 97230
    },
    {
      "epoch": 1.7071928931336575,
      "grad_norm": 0.05158292239217141,
      "learning_rate": 0.00015193092047499014,
      "loss": 0.8503,
      "step": 97240
    },
    {
      "epoch": 1.707368458013659,
      "grad_norm": 0.06176464937168185,
      "learning_rate": 0.0001519213642853073,
      "loss": 0.8543,
      "step": 97250
    },
    {
      "epoch": 1.7075440228936603,
      "grad_norm": 0.0815471584692428,
      "learning_rate": 0.00015191180745044097,
      "loss": 0.8452,
      "step": 97260
    },
    {
      "epoch": 1.7077195877736617,
      "grad_norm": 0.04516434460157234,
      "learning_rate": 0.00015190224997051234,
      "loss": 0.848,
      "step": 97270
    },
    {
      "epoch": 1.7078951526536632,
      "grad_norm": 0.05435522560418124,
      "learning_rate": 0.00015189269184564242,
      "loss": 0.8463,
      "step": 97280
    },
    {
      "epoch": 1.7080707175336647,
      "grad_norm": 0.06781501584923312,
      "learning_rate": 0.00015188313307595237,
      "loss": 0.8463,
      "step": 97290
    },
    {
      "epoch": 1.708246282413666,
      "grad_norm": 0.06896512321811728,
      "learning_rate": 0.00015187357366156325,
      "loss": 0.8419,
      "step": 97300
    },
    {
      "epoch": 1.7084218472936674,
      "grad_norm": 0.06834619766394862,
      "learning_rate": 0.00015186401360259616,
      "loss": 0.8478,
      "step": 97310
    },
    {
      "epoch": 1.7085974121736687,
      "grad_norm": 0.06264427121814779,
      "learning_rate": 0.00015185445289917236,
      "loss": 0.8438,
      "step": 97320
    },
    {
      "epoch": 1.7087729770536702,
      "grad_norm": 0.06009883897394693,
      "learning_rate": 0.00015184489155141282,
      "loss": 0.8544,
      "step": 97330
    },
    {
      "epoch": 1.7089485419336716,
      "grad_norm": 0.06423539694661339,
      "learning_rate": 0.0001518353295594388,
      "loss": 0.8535,
      "step": 97340
    },
    {
      "epoch": 1.709124106813673,
      "grad_norm": 0.07209871891309425,
      "learning_rate": 0.0001518257669233714,
      "loss": 0.8467,
      "step": 97350
    },
    {
      "epoch": 1.7092996716936744,
      "grad_norm": 0.048110879210878206,
      "learning_rate": 0.00015181620364333178,
      "loss": 0.8505,
      "step": 97360
    },
    {
      "epoch": 1.7094752365736758,
      "grad_norm": 0.06938176658558055,
      "learning_rate": 0.00015180663971944115,
      "loss": 0.8446,
      "step": 97370
    },
    {
      "epoch": 1.709650801453677,
      "grad_norm": 0.047128069856970385,
      "learning_rate": 0.00015179707515182062,
      "loss": 0.8395,
      "step": 97380
    },
    {
      "epoch": 1.7098263663336786,
      "grad_norm": 0.07592155274503691,
      "learning_rate": 0.00015178750994059142,
      "loss": 0.8404,
      "step": 97390
    },
    {
      "epoch": 1.71000193121368,
      "grad_norm": 0.07280163078498068,
      "learning_rate": 0.0001517779440858748,
      "loss": 0.8525,
      "step": 97400
    },
    {
      "epoch": 1.7101774960936815,
      "grad_norm": 0.06084006311786117,
      "learning_rate": 0.00015176837758779183,
      "loss": 0.8514,
      "step": 97410
    },
    {
      "epoch": 1.7103530609736828,
      "grad_norm": 0.06665903883822748,
      "learning_rate": 0.0001517588104464638,
      "loss": 0.8448,
      "step": 97420
    },
    {
      "epoch": 1.7105286258536843,
      "grad_norm": 0.07179751429173435,
      "learning_rate": 0.0001517492426620119,
      "loss": 0.8516,
      "step": 97430
    },
    {
      "epoch": 1.7107041907336855,
      "grad_norm": 0.06683140238114314,
      "learning_rate": 0.0001517396742345574,
      "loss": 0.8432,
      "step": 97440
    },
    {
      "epoch": 1.710879755613687,
      "grad_norm": 0.06690344525670983,
      "learning_rate": 0.00015173010516422145,
      "loss": 0.8395,
      "step": 97450
    },
    {
      "epoch": 1.7110553204936885,
      "grad_norm": 0.06449953662530392,
      "learning_rate": 0.0001517205354511254,
      "loss": 0.8419,
      "step": 97460
    },
    {
      "epoch": 1.71123088537369,
      "grad_norm": 0.06959404164797721,
      "learning_rate": 0.00015171096509539045,
      "loss": 0.8362,
      "step": 97470
    },
    {
      "epoch": 1.7114064502536912,
      "grad_norm": 0.06943553561691121,
      "learning_rate": 0.0001517013940971378,
      "loss": 0.8525,
      "step": 97480
    },
    {
      "epoch": 1.7115820151336927,
      "grad_norm": 0.06328248546525903,
      "learning_rate": 0.00015169182245648884,
      "loss": 0.8518,
      "step": 97490
    },
    {
      "epoch": 1.711757580013694,
      "grad_norm": 0.08549568593175405,
      "learning_rate": 0.00015168225017356471,
      "loss": 0.8485,
      "step": 97500
    },
    {
      "epoch": 1.7119331448936954,
      "grad_norm": 0.06960141178321681,
      "learning_rate": 0.0001516726772484868,
      "loss": 0.8475,
      "step": 97510
    },
    {
      "epoch": 1.712108709773697,
      "grad_norm": 0.08496695567732052,
      "learning_rate": 0.00015166310368137632,
      "loss": 0.8402,
      "step": 97520
    },
    {
      "epoch": 1.7122842746536984,
      "grad_norm": 0.0640160459559346,
      "learning_rate": 0.00015165352947235465,
      "loss": 0.8501,
      "step": 97530
    },
    {
      "epoch": 1.7124598395336996,
      "grad_norm": 0.06925778335289651,
      "learning_rate": 0.00015164395462154303,
      "loss": 0.8503,
      "step": 97540
    },
    {
      "epoch": 1.712635404413701,
      "grad_norm": 0.0537319065163397,
      "learning_rate": 0.0001516343791290628,
      "loss": 0.8502,
      "step": 97550
    },
    {
      "epoch": 1.7128109692937024,
      "grad_norm": 0.07899014804940356,
      "learning_rate": 0.00015162480299503526,
      "loss": 0.845,
      "step": 97560
    },
    {
      "epoch": 1.7129865341737038,
      "grad_norm": 0.06860505434857679,
      "learning_rate": 0.00015161522621958182,
      "loss": 0.8434,
      "step": 97570
    },
    {
      "epoch": 1.7131620990537053,
      "grad_norm": 0.06582941069962257,
      "learning_rate": 0.00015160564880282372,
      "loss": 0.8457,
      "step": 97580
    },
    {
      "epoch": 1.7133376639337068,
      "grad_norm": 0.1444411138095392,
      "learning_rate": 0.00015159607074488235,
      "loss": 0.8466,
      "step": 97590
    },
    {
      "epoch": 1.7135132288137083,
      "grad_norm": 0.05393305766164613,
      "learning_rate": 0.00015158649204587908,
      "loss": 0.8398,
      "step": 97600
    },
    {
      "epoch": 1.7136887936937095,
      "grad_norm": 0.05914502897491321,
      "learning_rate": 0.00015157691270593527,
      "loss": 0.852,
      "step": 97610
    },
    {
      "epoch": 1.7138643585737108,
      "grad_norm": 0.0610326070649969,
      "learning_rate": 0.00015156733272517225,
      "loss": 0.8458,
      "step": 97620
    },
    {
      "epoch": 1.7140399234537123,
      "grad_norm": 0.07203019092968875,
      "learning_rate": 0.00015155775210371145,
      "loss": 0.8499,
      "step": 97630
    },
    {
      "epoch": 1.7142154883337137,
      "grad_norm": 0.07414454684997046,
      "learning_rate": 0.00015154817084167427,
      "loss": 0.8475,
      "step": 97640
    },
    {
      "epoch": 1.7143910532137152,
      "grad_norm": 0.06866403970633297,
      "learning_rate": 0.00015153858893918206,
      "loss": 0.8573,
      "step": 97650
    },
    {
      "epoch": 1.7145666180937167,
      "grad_norm": 0.08604900166023467,
      "learning_rate": 0.00015152900639635623,
      "loss": 0.8424,
      "step": 97660
    },
    {
      "epoch": 1.714742182973718,
      "grad_norm": 0.05617215135391306,
      "learning_rate": 0.0001515194232133182,
      "loss": 0.8436,
      "step": 97670
    },
    {
      "epoch": 1.7149177478537192,
      "grad_norm": 0.05800239081460891,
      "learning_rate": 0.0001515098393901894,
      "loss": 0.8468,
      "step": 97680
    },
    {
      "epoch": 1.7150933127337207,
      "grad_norm": 0.08011335604287692,
      "learning_rate": 0.0001515002549270913,
      "loss": 0.8483,
      "step": 97690
    },
    {
      "epoch": 1.7152688776137222,
      "grad_norm": 0.11123972771934064,
      "learning_rate": 0.00015149066982414526,
      "loss": 0.8473,
      "step": 97700
    },
    {
      "epoch": 1.7154444424937236,
      "grad_norm": 0.0567471883454452,
      "learning_rate": 0.00015148108408147274,
      "loss": 0.8416,
      "step": 97710
    },
    {
      "epoch": 1.715620007373725,
      "grad_norm": 0.06458314086989997,
      "learning_rate": 0.0001514714976991952,
      "loss": 0.8544,
      "step": 97720
    },
    {
      "epoch": 1.7157955722537264,
      "grad_norm": 0.05754184953355685,
      "learning_rate": 0.00015146191067743413,
      "loss": 0.8568,
      "step": 97730
    },
    {
      "epoch": 1.7159711371337276,
      "grad_norm": 0.05113665715830573,
      "learning_rate": 0.000151452323016311,
      "loss": 0.8511,
      "step": 97740
    },
    {
      "epoch": 1.716146702013729,
      "grad_norm": 0.06205559743317499,
      "learning_rate": 0.00015144273471594724,
      "loss": 0.849,
      "step": 97750
    },
    {
      "epoch": 1.7163222668937306,
      "grad_norm": 0.05460825503843154,
      "learning_rate": 0.00015143314577646438,
      "loss": 0.8448,
      "step": 97760
    },
    {
      "epoch": 1.716497831773732,
      "grad_norm": 0.06598672255920433,
      "learning_rate": 0.00015142355619798389,
      "loss": 0.8501,
      "step": 97770
    },
    {
      "epoch": 1.7166733966537335,
      "grad_norm": 0.05770046131034454,
      "learning_rate": 0.0001514139659806273,
      "loss": 0.853,
      "step": 97780
    },
    {
      "epoch": 1.7168489615337348,
      "grad_norm": 0.05957000131048351,
      "learning_rate": 0.0001514043751245161,
      "loss": 0.8451,
      "step": 97790
    },
    {
      "epoch": 1.717024526413736,
      "grad_norm": 0.051348686645230374,
      "learning_rate": 0.00015139478362977178,
      "loss": 0.8508,
      "step": 97800
    },
    {
      "epoch": 1.7172000912937375,
      "grad_norm": 0.06158751378237986,
      "learning_rate": 0.0001513851914965159,
      "loss": 0.8414,
      "step": 97810
    },
    {
      "epoch": 1.717375656173739,
      "grad_norm": 0.04212268388479152,
      "learning_rate": 0.00015137559872486998,
      "loss": 0.8481,
      "step": 97820
    },
    {
      "epoch": 1.7175512210537405,
      "grad_norm": 0.06371793908169576,
      "learning_rate": 0.00015136600531495556,
      "loss": 0.85,
      "step": 97830
    },
    {
      "epoch": 1.717726785933742,
      "grad_norm": 0.0577834639275934,
      "learning_rate": 0.00015135641126689424,
      "loss": 0.8527,
      "step": 97840
    },
    {
      "epoch": 1.7179023508137432,
      "grad_norm": 0.06913039443535143,
      "learning_rate": 0.0001513468165808075,
      "loss": 0.8488,
      "step": 97850
    },
    {
      "epoch": 1.7180779156937445,
      "grad_norm": 0.05206835825371254,
      "learning_rate": 0.000151337221256817,
      "loss": 0.8547,
      "step": 97860
    },
    {
      "epoch": 1.718253480573746,
      "grad_norm": 0.049430107612323464,
      "learning_rate": 0.00015132762529504421,
      "loss": 0.8499,
      "step": 97870
    },
    {
      "epoch": 1.7184290454537474,
      "grad_norm": 0.04945564506686379,
      "learning_rate": 0.00015131802869561078,
      "loss": 0.8491,
      "step": 97880
    },
    {
      "epoch": 1.718604610333749,
      "grad_norm": 0.043315430906840556,
      "learning_rate": 0.00015130843145863828,
      "loss": 0.845,
      "step": 97890
    },
    {
      "epoch": 1.7187801752137504,
      "grad_norm": 0.0480021777963715,
      "learning_rate": 0.00015129883358424827,
      "loss": 0.8478,
      "step": 97900
    },
    {
      "epoch": 1.7189557400937516,
      "grad_norm": 0.044041428477260926,
      "learning_rate": 0.0001512892350725625,
      "loss": 0.8372,
      "step": 97910
    },
    {
      "epoch": 1.719131304973753,
      "grad_norm": 0.05046376732321836,
      "learning_rate": 0.0001512796359237024,
      "loss": 0.8586,
      "step": 97920
    },
    {
      "epoch": 1.7193068698537544,
      "grad_norm": 0.06549957805479635,
      "learning_rate": 0.0001512700361377897,
      "loss": 0.8488,
      "step": 97930
    },
    {
      "epoch": 1.7194824347337558,
      "grad_norm": 0.05671870533216356,
      "learning_rate": 0.000151260435714946,
      "loss": 0.8441,
      "step": 97940
    },
    {
      "epoch": 1.7196579996137573,
      "grad_norm": 0.052509425481869605,
      "learning_rate": 0.00015125083465529298,
      "loss": 0.8506,
      "step": 97950
    },
    {
      "epoch": 1.7198335644937588,
      "grad_norm": 0.07854184336382379,
      "learning_rate": 0.00015124123295895222,
      "loss": 0.8403,
      "step": 97960
    },
    {
      "epoch": 1.72000912937376,
      "grad_norm": 0.061280758725232795,
      "learning_rate": 0.0001512316306260454,
      "loss": 0.8521,
      "step": 97970
    },
    {
      "epoch": 1.7201846942537615,
      "grad_norm": 0.06095638597521177,
      "learning_rate": 0.00015122202765669422,
      "loss": 0.8505,
      "step": 97980
    },
    {
      "epoch": 1.7203602591337628,
      "grad_norm": 0.059636760109282975,
      "learning_rate": 0.0001512124240510203,
      "loss": 0.8425,
      "step": 97990
    },
    {
      "epoch": 1.7205358240137643,
      "grad_norm": 0.05735675362163551,
      "learning_rate": 0.00015120281980914537,
      "loss": 0.8486,
      "step": 98000
    },
    {
      "epoch": 1.7207113888937657,
      "grad_norm": 0.053229757208893155,
      "learning_rate": 0.00015119321493119108,
      "loss": 0.8515,
      "step": 98010
    },
    {
      "epoch": 1.7208869537737672,
      "grad_norm": 0.09178269118652384,
      "learning_rate": 0.00015118360941727917,
      "loss": 0.8456,
      "step": 98020
    },
    {
      "epoch": 1.7210625186537685,
      "grad_norm": 0.07440545733216995,
      "learning_rate": 0.00015117400326753125,
      "loss": 0.8499,
      "step": 98030
    },
    {
      "epoch": 1.72123808353377,
      "grad_norm": 0.05858932919443485,
      "learning_rate": 0.00015116439648206915,
      "loss": 0.8546,
      "step": 98040
    },
    {
      "epoch": 1.7214136484137712,
      "grad_norm": 0.06337174441173236,
      "learning_rate": 0.00015115478906101447,
      "loss": 0.8503,
      "step": 98050
    },
    {
      "epoch": 1.7215892132937727,
      "grad_norm": 0.05964859620654927,
      "learning_rate": 0.00015114518100448906,
      "loss": 0.8504,
      "step": 98060
    },
    {
      "epoch": 1.7217647781737742,
      "grad_norm": 0.06078035909319704,
      "learning_rate": 0.00015113557231261456,
      "loss": 0.8437,
      "step": 98070
    },
    {
      "epoch": 1.7219403430537756,
      "grad_norm": 0.0657497772935001,
      "learning_rate": 0.00015112596298551278,
      "loss": 0.8547,
      "step": 98080
    },
    {
      "epoch": 1.722115907933777,
      "grad_norm": 0.050095497660575534,
      "learning_rate": 0.0001511163530233054,
      "loss": 0.8524,
      "step": 98090
    },
    {
      "epoch": 1.7222914728137784,
      "grad_norm": 0.05611491542911289,
      "learning_rate": 0.00015110674242611424,
      "loss": 0.8516,
      "step": 98100
    },
    {
      "epoch": 1.7224670376937796,
      "grad_norm": 0.08214345862642192,
      "learning_rate": 0.0001510971311940611,
      "loss": 0.8517,
      "step": 98110
    },
    {
      "epoch": 1.722642602573781,
      "grad_norm": 0.054228166412268027,
      "learning_rate": 0.00015108751932726763,
      "loss": 0.8475,
      "step": 98120
    },
    {
      "epoch": 1.7228181674537826,
      "grad_norm": 0.05903965700034488,
      "learning_rate": 0.00015107790682585573,
      "loss": 0.8532,
      "step": 98130
    },
    {
      "epoch": 1.722993732333784,
      "grad_norm": 0.055533487228533976,
      "learning_rate": 0.00015106829368994714,
      "loss": 0.835,
      "step": 98140
    },
    {
      "epoch": 1.7231692972137853,
      "grad_norm": 0.053564008361141086,
      "learning_rate": 0.0001510586799196637,
      "loss": 0.8505,
      "step": 98150
    },
    {
      "epoch": 1.7233448620937868,
      "grad_norm": 0.06708955020485145,
      "learning_rate": 0.00015104906551512717,
      "loss": 0.8518,
      "step": 98160
    },
    {
      "epoch": 1.723520426973788,
      "grad_norm": 0.050701243485894915,
      "learning_rate": 0.0001510394504764594,
      "loss": 0.8442,
      "step": 98170
    },
    {
      "epoch": 1.7236959918537895,
      "grad_norm": 0.05201715822176467,
      "learning_rate": 0.0001510298348037822,
      "loss": 0.8466,
      "step": 98180
    },
    {
      "epoch": 1.723871556733791,
      "grad_norm": 0.07532922201684264,
      "learning_rate": 0.0001510202184972174,
      "loss": 0.8457,
      "step": 98190
    },
    {
      "epoch": 1.7240471216137925,
      "grad_norm": 0.045883576467737164,
      "learning_rate": 0.00015101060155688686,
      "loss": 0.845,
      "step": 98200
    },
    {
      "epoch": 1.7242226864937937,
      "grad_norm": 0.058240260755423275,
      "learning_rate": 0.0001510009839829124,
      "loss": 0.8438,
      "step": 98210
    },
    {
      "epoch": 1.7243982513737952,
      "grad_norm": 0.058717622640629366,
      "learning_rate": 0.00015099136577541585,
      "loss": 0.8555,
      "step": 98220
    },
    {
      "epoch": 1.7245738162537965,
      "grad_norm": 0.061428623818586534,
      "learning_rate": 0.00015098174693451918,
      "loss": 0.8469,
      "step": 98230
    },
    {
      "epoch": 1.724749381133798,
      "grad_norm": 0.07440543065055699,
      "learning_rate": 0.00015097212746034415,
      "loss": 0.8519,
      "step": 98240
    },
    {
      "epoch": 1.7249249460137994,
      "grad_norm": 0.06627989800349525,
      "learning_rate": 0.00015096250735301275,
      "loss": 0.8525,
      "step": 98250
    },
    {
      "epoch": 1.725100510893801,
      "grad_norm": 0.05761612858917361,
      "learning_rate": 0.00015095288661264673,
      "loss": 0.854,
      "step": 98260
    },
    {
      "epoch": 1.7252760757738022,
      "grad_norm": 0.06372359396022848,
      "learning_rate": 0.00015094326523936813,
      "loss": 0.851,
      "step": 98270
    },
    {
      "epoch": 1.7254516406538036,
      "grad_norm": 0.0594459389638476,
      "learning_rate": 0.00015093364323329875,
      "loss": 0.853,
      "step": 98280
    },
    {
      "epoch": 1.7256272055338049,
      "grad_norm": 0.061158502792678725,
      "learning_rate": 0.00015092402059456052,
      "loss": 0.8451,
      "step": 98290
    },
    {
      "epoch": 1.7258027704138064,
      "grad_norm": 0.04968169337309075,
      "learning_rate": 0.00015091439732327542,
      "loss": 0.8545,
      "step": 98300
    },
    {
      "epoch": 1.7259783352938078,
      "grad_norm": 0.07158712895451569,
      "learning_rate": 0.00015090477341956533,
      "loss": 0.8473,
      "step": 98310
    },
    {
      "epoch": 1.7261539001738093,
      "grad_norm": 0.050813051433544025,
      "learning_rate": 0.00015089514888355217,
      "loss": 0.8498,
      "step": 98320
    },
    {
      "epoch": 1.7263294650538108,
      "grad_norm": 0.04713869336449074,
      "learning_rate": 0.00015088552371535796,
      "loss": 0.852,
      "step": 98330
    },
    {
      "epoch": 1.726505029933812,
      "grad_norm": 0.05793834039754169,
      "learning_rate": 0.0001508758979151046,
      "loss": 0.8419,
      "step": 98340
    },
    {
      "epoch": 1.7266805948138133,
      "grad_norm": 0.06659642410706508,
      "learning_rate": 0.00015086627148291402,
      "loss": 0.8505,
      "step": 98350
    },
    {
      "epoch": 1.7268561596938148,
      "grad_norm": 0.05989054917499796,
      "learning_rate": 0.00015085664441890823,
      "loss": 0.8506,
      "step": 98360
    },
    {
      "epoch": 1.7270317245738163,
      "grad_norm": 0.061260807817819375,
      "learning_rate": 0.00015084701672320926,
      "loss": 0.8447,
      "step": 98370
    },
    {
      "epoch": 1.7272072894538177,
      "grad_norm": 0.08723931419821015,
      "learning_rate": 0.00015083738839593894,
      "loss": 0.8495,
      "step": 98380
    },
    {
      "epoch": 1.7273828543338192,
      "grad_norm": 0.0706676973541325,
      "learning_rate": 0.0001508277594372194,
      "loss": 0.8417,
      "step": 98390
    },
    {
      "epoch": 1.7275584192138205,
      "grad_norm": 0.05362603381462821,
      "learning_rate": 0.00015081812984717258,
      "loss": 0.8426,
      "step": 98400
    },
    {
      "epoch": 1.7277339840938217,
      "grad_norm": 0.05851192581819552,
      "learning_rate": 0.00015080849962592055,
      "loss": 0.8499,
      "step": 98410
    },
    {
      "epoch": 1.7279095489738232,
      "grad_norm": 0.06177069901753289,
      "learning_rate": 0.00015079886877358527,
      "loss": 0.846,
      "step": 98420
    },
    {
      "epoch": 1.7280851138538247,
      "grad_norm": 0.05068705887466879,
      "learning_rate": 0.00015078923729028876,
      "loss": 0.8442,
      "step": 98430
    },
    {
      "epoch": 1.7282606787338262,
      "grad_norm": 0.05796676798420028,
      "learning_rate": 0.0001507796051761531,
      "loss": 0.8458,
      "step": 98440
    },
    {
      "epoch": 1.7284362436138276,
      "grad_norm": 0.052477489026851315,
      "learning_rate": 0.00015076997243130026,
      "loss": 0.8438,
      "step": 98450
    },
    {
      "epoch": 1.728611808493829,
      "grad_norm": 0.054469403100863144,
      "learning_rate": 0.00015076033905585238,
      "loss": 0.8507,
      "step": 98460
    },
    {
      "epoch": 1.7287873733738301,
      "grad_norm": 0.053563336313333966,
      "learning_rate": 0.00015075070504993142,
      "loss": 0.8488,
      "step": 98470
    },
    {
      "epoch": 1.7289629382538316,
      "grad_norm": 0.04981673172037838,
      "learning_rate": 0.0001507410704136595,
      "loss": 0.8472,
      "step": 98480
    },
    {
      "epoch": 1.729138503133833,
      "grad_norm": 0.050422251124851596,
      "learning_rate": 0.00015073143514715873,
      "loss": 0.8433,
      "step": 98490
    },
    {
      "epoch": 1.7293140680138346,
      "grad_norm": 0.07430827133895274,
      "learning_rate": 0.0001507217992505511,
      "loss": 0.8435,
      "step": 98500
    },
    {
      "epoch": 1.729489632893836,
      "grad_norm": 0.05763400623673126,
      "learning_rate": 0.00015071216272395874,
      "loss": 0.8493,
      "step": 98510
    },
    {
      "epoch": 1.7296651977738373,
      "grad_norm": 0.06162095558624064,
      "learning_rate": 0.00015070252556750378,
      "loss": 0.8535,
      "step": 98520
    },
    {
      "epoch": 1.7298407626538386,
      "grad_norm": 0.06281349190707212,
      "learning_rate": 0.0001506928877813083,
      "loss": 0.8513,
      "step": 98530
    },
    {
      "epoch": 1.73001632753384,
      "grad_norm": 0.07216847589303294,
      "learning_rate": 0.0001506832493654944,
      "loss": 0.8447,
      "step": 98540
    },
    {
      "epoch": 1.7301918924138415,
      "grad_norm": 0.04606963993901585,
      "learning_rate": 0.0001506736103201842,
      "loss": 0.8515,
      "step": 98550
    },
    {
      "epoch": 1.730367457293843,
      "grad_norm": 0.06399147001884023,
      "learning_rate": 0.00015066397064549984,
      "loss": 0.8524,
      "step": 98560
    },
    {
      "epoch": 1.7305430221738445,
      "grad_norm": 0.08222763861328515,
      "learning_rate": 0.00015065433034156344,
      "loss": 0.8443,
      "step": 98570
    },
    {
      "epoch": 1.7307185870538457,
      "grad_norm": 0.09014684113288732,
      "learning_rate": 0.00015064468940849716,
      "loss": 0.8531,
      "step": 98580
    },
    {
      "epoch": 1.730894151933847,
      "grad_norm": 0.048553338517865266,
      "learning_rate": 0.0001506350478464232,
      "loss": 0.8441,
      "step": 98590
    },
    {
      "epoch": 1.7310697168138485,
      "grad_norm": 0.05675486093690112,
      "learning_rate": 0.0001506254056554636,
      "loss": 0.8493,
      "step": 98600
    },
    {
      "epoch": 1.73124528169385,
      "grad_norm": 0.05750031826864967,
      "learning_rate": 0.00015061576283574067,
      "loss": 0.8456,
      "step": 98610
    },
    {
      "epoch": 1.7314208465738514,
      "grad_norm": 0.072611065959397,
      "learning_rate": 0.00015060611938737646,
      "loss": 0.8451,
      "step": 98620
    },
    {
      "epoch": 1.731596411453853,
      "grad_norm": 0.07019304851797256,
      "learning_rate": 0.00015059647531049324,
      "loss": 0.8564,
      "step": 98630
    },
    {
      "epoch": 1.7317719763338542,
      "grad_norm": 0.07042958565302529,
      "learning_rate": 0.00015058683060521316,
      "loss": 0.8443,
      "step": 98640
    },
    {
      "epoch": 1.7319475412138556,
      "grad_norm": 0.07228893696741615,
      "learning_rate": 0.00015057718527165844,
      "loss": 0.8449,
      "step": 98650
    },
    {
      "epoch": 1.732123106093857,
      "grad_norm": 0.05753958839569914,
      "learning_rate": 0.00015056753930995127,
      "loss": 0.8413,
      "step": 98660
    },
    {
      "epoch": 1.7322986709738584,
      "grad_norm": 0.07397959668206447,
      "learning_rate": 0.0001505578927202139,
      "loss": 0.8426,
      "step": 98670
    },
    {
      "epoch": 1.7324742358538598,
      "grad_norm": 0.05633411106005507,
      "learning_rate": 0.00015054824550256852,
      "loss": 0.8485,
      "step": 98680
    },
    {
      "epoch": 1.7326498007338613,
      "grad_norm": 0.047568517027621826,
      "learning_rate": 0.00015053859765713735,
      "loss": 0.8506,
      "step": 98690
    },
    {
      "epoch": 1.7328253656138626,
      "grad_norm": 0.05977623475778332,
      "learning_rate": 0.00015052894918404268,
      "loss": 0.8494,
      "step": 98700
    },
    {
      "epoch": 1.733000930493864,
      "grad_norm": 0.056797474151436905,
      "learning_rate": 0.00015051930008340675,
      "loss": 0.8448,
      "step": 98710
    },
    {
      "epoch": 1.7331764953738653,
      "grad_norm": 0.05452211820933728,
      "learning_rate": 0.00015050965035535177,
      "loss": 0.8509,
      "step": 98720
    },
    {
      "epoch": 1.7333520602538668,
      "grad_norm": 0.06281993723631585,
      "learning_rate": 0.0001505,
      "loss": 0.8554,
      "step": 98730
    },
    {
      "epoch": 1.7335276251338683,
      "grad_norm": 0.04507174734155711,
      "learning_rate": 0.00015049034901747378,
      "loss": 0.8476,
      "step": 98740
    },
    {
      "epoch": 1.7337031900138697,
      "grad_norm": 0.05155036661151256,
      "learning_rate": 0.00015048069740789533,
      "loss": 0.8434,
      "step": 98750
    },
    {
      "epoch": 1.733878754893871,
      "grad_norm": 0.06014055394113231,
      "learning_rate": 0.000150471045171387,
      "loss": 0.8398,
      "step": 98760
    },
    {
      "epoch": 1.7340543197738725,
      "grad_norm": 0.04978332956505653,
      "learning_rate": 0.00015046139230807105,
      "loss": 0.8574,
      "step": 98770
    },
    {
      "epoch": 1.7342298846538737,
      "grad_norm": 0.07288505217661696,
      "learning_rate": 0.00015045173881806975,
      "loss": 0.8513,
      "step": 98780
    },
    {
      "epoch": 1.7344054495338752,
      "grad_norm": 0.06502610395262082,
      "learning_rate": 0.00015044208470150545,
      "loss": 0.8498,
      "step": 98790
    },
    {
      "epoch": 1.7345810144138767,
      "grad_norm": 0.0625990790053323,
      "learning_rate": 0.00015043242995850046,
      "loss": 0.8529,
      "step": 98800
    },
    {
      "epoch": 1.7347565792938782,
      "grad_norm": 0.061329946149352776,
      "learning_rate": 0.0001504227745891771,
      "loss": 0.8371,
      "step": 98810
    },
    {
      "epoch": 1.7349321441738794,
      "grad_norm": 0.04463362014794127,
      "learning_rate": 0.00015041311859365775,
      "loss": 0.8477,
      "step": 98820
    },
    {
      "epoch": 1.735107709053881,
      "grad_norm": 0.05135968858192828,
      "learning_rate": 0.0001504034619720647,
      "loss": 0.8511,
      "step": 98830
    },
    {
      "epoch": 1.7352832739338822,
      "grad_norm": 0.05502900078444362,
      "learning_rate": 0.0001503938047245203,
      "loss": 0.8435,
      "step": 98840
    },
    {
      "epoch": 1.7354588388138836,
      "grad_norm": 0.05814079010173844,
      "learning_rate": 0.00015038414685114695,
      "loss": 0.8424,
      "step": 98850
    },
    {
      "epoch": 1.735634403693885,
      "grad_norm": 0.05917848986174673,
      "learning_rate": 0.00015037448835206695,
      "loss": 0.8396,
      "step": 98860
    },
    {
      "epoch": 1.7358099685738866,
      "grad_norm": 0.06328963192123692,
      "learning_rate": 0.00015036482922740275,
      "loss": 0.8429,
      "step": 98870
    },
    {
      "epoch": 1.7359855334538878,
      "grad_norm": 0.08323631268322775,
      "learning_rate": 0.00015035516947727674,
      "loss": 0.8471,
      "step": 98880
    },
    {
      "epoch": 1.7361610983338893,
      "grad_norm": 0.06468532609743846,
      "learning_rate": 0.00015034550910181122,
      "loss": 0.8515,
      "step": 98890
    },
    {
      "epoch": 1.7363366632138906,
      "grad_norm": 0.07374144043767022,
      "learning_rate": 0.00015033584810112866,
      "loss": 0.8447,
      "step": 98900
    },
    {
      "epoch": 1.736512228093892,
      "grad_norm": 0.0603905265021126,
      "learning_rate": 0.00015032618647535146,
      "loss": 0.8438,
      "step": 98910
    },
    {
      "epoch": 1.7366877929738935,
      "grad_norm": 0.07926063484322465,
      "learning_rate": 0.00015031652422460202,
      "loss": 0.8506,
      "step": 98920
    },
    {
      "epoch": 1.736863357853895,
      "grad_norm": 0.038898383927446274,
      "learning_rate": 0.00015030686134900276,
      "loss": 0.854,
      "step": 98930
    },
    {
      "epoch": 1.7370389227338963,
      "grad_norm": 0.06628666897035601,
      "learning_rate": 0.0001502971978486761,
      "loss": 0.8523,
      "step": 98940
    },
    {
      "epoch": 1.7372144876138977,
      "grad_norm": 0.05672838658921226,
      "learning_rate": 0.00015028753372374457,
      "loss": 0.8428,
      "step": 98950
    },
    {
      "epoch": 1.737390052493899,
      "grad_norm": 0.04795006147611145,
      "learning_rate": 0.00015027786897433044,
      "loss": 0.8468,
      "step": 98960
    },
    {
      "epoch": 1.7375656173739005,
      "grad_norm": 0.07213981366076495,
      "learning_rate": 0.00015026820360055634,
      "loss": 0.8563,
      "step": 98970
    },
    {
      "epoch": 1.737741182253902,
      "grad_norm": 0.07877007512515664,
      "learning_rate": 0.0001502585376025446,
      "loss": 0.8489,
      "step": 98980
    },
    {
      "epoch": 1.7379167471339034,
      "grad_norm": 0.06406492532983617,
      "learning_rate": 0.0001502488709804178,
      "loss": 0.8403,
      "step": 98990
    },
    {
      "epoch": 1.7380923120139047,
      "grad_norm": 0.06535492084124218,
      "learning_rate": 0.00015023920373429835,
      "loss": 0.8434,
      "step": 99000
    },
    {
      "epoch": 1.7382678768939062,
      "grad_norm": 0.09204183390962276,
      "learning_rate": 0.00015022953586430875,
      "loss": 0.8455,
      "step": 99010
    },
    {
      "epoch": 1.7384434417739074,
      "grad_norm": 0.05054325648062031,
      "learning_rate": 0.0001502198673705715,
      "loss": 0.8369,
      "step": 99020
    },
    {
      "epoch": 1.738619006653909,
      "grad_norm": 0.06486350576869376,
      "learning_rate": 0.00015021019825320908,
      "loss": 0.8526,
      "step": 99030
    },
    {
      "epoch": 1.7387945715339104,
      "grad_norm": 0.054398448771459336,
      "learning_rate": 0.00015020052851234404,
      "loss": 0.8488,
      "step": 99040
    },
    {
      "epoch": 1.7389701364139118,
      "grad_norm": 0.058786150484368624,
      "learning_rate": 0.00015019085814809888,
      "loss": 0.8472,
      "step": 99050
    },
    {
      "epoch": 1.7391457012939133,
      "grad_norm": 0.04908155724098363,
      "learning_rate": 0.0001501811871605961,
      "loss": 0.8406,
      "step": 99060
    },
    {
      "epoch": 1.7393212661739146,
      "grad_norm": 0.06678174491482253,
      "learning_rate": 0.00015017151554995824,
      "loss": 0.8554,
      "step": 99070
    },
    {
      "epoch": 1.7394968310539158,
      "grad_norm": 0.05986068730475089,
      "learning_rate": 0.00015016184331630787,
      "loss": 0.8431,
      "step": 99080
    },
    {
      "epoch": 1.7396723959339173,
      "grad_norm": 0.06281691313699242,
      "learning_rate": 0.0001501521704597675,
      "loss": 0.8417,
      "step": 99090
    },
    {
      "epoch": 1.7398479608139188,
      "grad_norm": 0.04925096516173562,
      "learning_rate": 0.00015014249698045977,
      "loss": 0.8496,
      "step": 99100
    },
    {
      "epoch": 1.7400235256939203,
      "grad_norm": 0.059614685169050746,
      "learning_rate": 0.00015013282287850715,
      "loss": 0.8483,
      "step": 99110
    },
    {
      "epoch": 1.7401990905739217,
      "grad_norm": 0.061657876772163674,
      "learning_rate": 0.00015012314815403225,
      "loss": 0.85,
      "step": 99120
    },
    {
      "epoch": 1.740374655453923,
      "grad_norm": 0.051076514945252026,
      "learning_rate": 0.0001501134728071576,
      "loss": 0.8454,
      "step": 99130
    },
    {
      "epoch": 1.7405502203339243,
      "grad_norm": 0.057935425478135585,
      "learning_rate": 0.00015010379683800593,
      "loss": 0.841,
      "step": 99140
    },
    {
      "epoch": 1.7407257852139257,
      "grad_norm": 0.04654527631280155,
      "learning_rate": 0.00015009412024669967,
      "loss": 0.8487,
      "step": 99150
    },
    {
      "epoch": 1.7409013500939272,
      "grad_norm": 0.06774620511629431,
      "learning_rate": 0.00015008444303336152,
      "loss": 0.8448,
      "step": 99160
    },
    {
      "epoch": 1.7410769149739287,
      "grad_norm": 0.05498508371825022,
      "learning_rate": 0.0001500747651981141,
      "loss": 0.8514,
      "step": 99170
    },
    {
      "epoch": 1.7412524798539302,
      "grad_norm": 0.06108917042964049,
      "learning_rate": 0.00015006508674107996,
      "loss": 0.8518,
      "step": 99180
    },
    {
      "epoch": 1.7414280447339314,
      "grad_norm": 0.0937239931002461,
      "learning_rate": 0.00015005540766238178,
      "loss": 0.8472,
      "step": 99190
    },
    {
      "epoch": 1.7416036096139327,
      "grad_norm": 0.06217339406245173,
      "learning_rate": 0.0001500457279621422,
      "loss": 0.85,
      "step": 99200
    },
    {
      "epoch": 1.7417791744939342,
      "grad_norm": 0.059329388729044676,
      "learning_rate": 0.0001500360476404838,
      "loss": 0.8444,
      "step": 99210
    },
    {
      "epoch": 1.7419547393739356,
      "grad_norm": 0.07505488799979677,
      "learning_rate": 0.00015002636669752932,
      "loss": 0.8524,
      "step": 99220
    },
    {
      "epoch": 1.7421303042539371,
      "grad_norm": 0.04597591716958494,
      "learning_rate": 0.00015001668513340137,
      "loss": 0.8435,
      "step": 99230
    },
    {
      "epoch": 1.7423058691339386,
      "grad_norm": 0.06361927376655267,
      "learning_rate": 0.00015000700294822263,
      "loss": 0.8521,
      "step": 99240
    },
    {
      "epoch": 1.7424814340139398,
      "grad_norm": 0.05311407550728778,
      "learning_rate": 0.00014999732014211575,
      "loss": 0.8438,
      "step": 99250
    },
    {
      "epoch": 1.742656998893941,
      "grad_norm": 0.060524813365248606,
      "learning_rate": 0.00014998763671520346,
      "loss": 0.8535,
      "step": 99260
    },
    {
      "epoch": 1.7428325637739426,
      "grad_norm": 0.05809119059650826,
      "learning_rate": 0.00014997795266760843,
      "loss": 0.855,
      "step": 99270
    },
    {
      "epoch": 1.743008128653944,
      "grad_norm": 0.07110139086799344,
      "learning_rate": 0.0001499682679994533,
      "loss": 0.8586,
      "step": 99280
    },
    {
      "epoch": 1.7431836935339455,
      "grad_norm": 0.04914840603573747,
      "learning_rate": 0.0001499585827108609,
      "loss": 0.8486,
      "step": 99290
    },
    {
      "epoch": 1.743359258413947,
      "grad_norm": 0.05286238862342092,
      "learning_rate": 0.00014994889680195383,
      "loss": 0.853,
      "step": 99300
    },
    {
      "epoch": 1.7435348232939483,
      "grad_norm": 0.05285327481353944,
      "learning_rate": 0.00014993921027285488,
      "loss": 0.8546,
      "step": 99310
    },
    {
      "epoch": 1.7437103881739495,
      "grad_norm": 0.07066219005255156,
      "learning_rate": 0.0001499295231236867,
      "loss": 0.8533,
      "step": 99320
    },
    {
      "epoch": 1.743885953053951,
      "grad_norm": 0.05910628855162143,
      "learning_rate": 0.00014991983535457218,
      "loss": 0.8593,
      "step": 99330
    },
    {
      "epoch": 1.7440615179339525,
      "grad_norm": 0.07826693959285057,
      "learning_rate": 0.00014991014696563387,
      "loss": 0.8466,
      "step": 99340
    },
    {
      "epoch": 1.744237082813954,
      "grad_norm": 0.05537173672246361,
      "learning_rate": 0.00014990045795699468,
      "loss": 0.8563,
      "step": 99350
    },
    {
      "epoch": 1.7444126476939554,
      "grad_norm": 0.06409578432790305,
      "learning_rate": 0.0001498907683287773,
      "loss": 0.8419,
      "step": 99360
    },
    {
      "epoch": 1.7445882125739567,
      "grad_norm": 0.04680093012280835,
      "learning_rate": 0.00014988107808110453,
      "loss": 0.8439,
      "step": 99370
    },
    {
      "epoch": 1.7447637774539582,
      "grad_norm": 0.08144459766388984,
      "learning_rate": 0.00014987138721409915,
      "loss": 0.8471,
      "step": 99380
    },
    {
      "epoch": 1.7449393423339594,
      "grad_norm": 0.06267057147946939,
      "learning_rate": 0.00014986169572788388,
      "loss": 0.8512,
      "step": 99390
    },
    {
      "epoch": 1.745114907213961,
      "grad_norm": 0.06272340917577833,
      "learning_rate": 0.00014985200362258158,
      "loss": 0.8456,
      "step": 99400
    },
    {
      "epoch": 1.7452904720939624,
      "grad_norm": 0.06566217899658852,
      "learning_rate": 0.00014984231089831506,
      "loss": 0.85,
      "step": 99410
    },
    {
      "epoch": 1.7454660369739639,
      "grad_norm": 0.08002125502648781,
      "learning_rate": 0.00014983261755520706,
      "loss": 0.8422,
      "step": 99420
    },
    {
      "epoch": 1.745641601853965,
      "grad_norm": 0.04149086205743743,
      "learning_rate": 0.00014982292359338046,
      "loss": 0.8424,
      "step": 99430
    },
    {
      "epoch": 1.7458171667339666,
      "grad_norm": 0.05916991697900539,
      "learning_rate": 0.00014981322901295806,
      "loss": 0.8511,
      "step": 99440
    },
    {
      "epoch": 1.7459927316139678,
      "grad_norm": 0.07000303749545639,
      "learning_rate": 0.0001498035338140627,
      "loss": 0.8537,
      "step": 99450
    },
    {
      "epoch": 1.7461682964939693,
      "grad_norm": 0.058808690171777624,
      "learning_rate": 0.00014979383799681718,
      "loss": 0.8434,
      "step": 99460
    },
    {
      "epoch": 1.7463438613739708,
      "grad_norm": 0.06498735864410474,
      "learning_rate": 0.00014978414156134438,
      "loss": 0.8442,
      "step": 99470
    },
    {
      "epoch": 1.7465194262539723,
      "grad_norm": 0.05126355991272636,
      "learning_rate": 0.00014977444450776724,
      "loss": 0.8505,
      "step": 99480
    },
    {
      "epoch": 1.7466949911339735,
      "grad_norm": 0.04863822052861131,
      "learning_rate": 0.00014976474683620845,
      "loss": 0.8494,
      "step": 99490
    },
    {
      "epoch": 1.746870556013975,
      "grad_norm": 0.054299575217619765,
      "learning_rate": 0.00014975504854679102,
      "loss": 0.8438,
      "step": 99500
    },
    {
      "epoch": 1.7470461208939763,
      "grad_norm": 0.04807307917258277,
      "learning_rate": 0.00014974534963963776,
      "loss": 0.8481,
      "step": 99510
    },
    {
      "epoch": 1.7472216857739777,
      "grad_norm": 0.052952041707364894,
      "learning_rate": 0.00014973565011487157,
      "loss": 0.8497,
      "step": 99520
    },
    {
      "epoch": 1.7473972506539792,
      "grad_norm": 0.07447129507475768,
      "learning_rate": 0.00014972594997261535,
      "loss": 0.8493,
      "step": 99530
    },
    {
      "epoch": 1.7475728155339807,
      "grad_norm": 0.05650589754176557,
      "learning_rate": 0.00014971624921299203,
      "loss": 0.8472,
      "step": 99540
    },
    {
      "epoch": 1.747748380413982,
      "grad_norm": 0.053034675398742626,
      "learning_rate": 0.0001497065478361245,
      "loss": 0.843,
      "step": 99550
    },
    {
      "epoch": 1.7479239452939834,
      "grad_norm": 0.05334475400590366,
      "learning_rate": 0.00014969684584213567,
      "loss": 0.8466,
      "step": 99560
    },
    {
      "epoch": 1.7480995101739847,
      "grad_norm": 0.0493346672869127,
      "learning_rate": 0.00014968714323114842,
      "loss": 0.8532,
      "step": 99570
    },
    {
      "epoch": 1.7482750750539862,
      "grad_norm": 0.04911919156632359,
      "learning_rate": 0.00014967744000328578,
      "loss": 0.8514,
      "step": 99580
    },
    {
      "epoch": 1.7484506399339876,
      "grad_norm": 0.04932001949682425,
      "learning_rate": 0.00014966773615867065,
      "loss": 0.8463,
      "step": 99590
    },
    {
      "epoch": 1.7486262048139891,
      "grad_norm": 0.05365447130394192,
      "learning_rate": 0.00014965803169742598,
      "loss": 0.8451,
      "step": 99600
    },
    {
      "epoch": 1.7488017696939904,
      "grad_norm": 0.0567565985736806,
      "learning_rate": 0.00014964832661967472,
      "loss": 0.859,
      "step": 99610
    },
    {
      "epoch": 1.7489773345739918,
      "grad_norm": 0.04531954882204411,
      "learning_rate": 0.0001496386209255398,
      "loss": 0.8452,
      "step": 99620
    },
    {
      "epoch": 1.749152899453993,
      "grad_norm": 0.06582661547721243,
      "learning_rate": 0.00014962891461514427,
      "loss": 0.8489,
      "step": 99630
    },
    {
      "epoch": 1.7493284643339946,
      "grad_norm": 0.059168221673006845,
      "learning_rate": 0.00014961920768861105,
      "loss": 0.8388,
      "step": 99640
    },
    {
      "epoch": 1.749504029213996,
      "grad_norm": 0.07057056516221366,
      "learning_rate": 0.0001496095001460632,
      "loss": 0.8406,
      "step": 99650
    },
    {
      "epoch": 1.7496795940939975,
      "grad_norm": 0.05433722455003561,
      "learning_rate": 0.00014959979198762358,
      "loss": 0.8487,
      "step": 99660
    },
    {
      "epoch": 1.7498551589739988,
      "grad_norm": 0.0710508127963252,
      "learning_rate": 0.00014959008321341533,
      "loss": 0.8464,
      "step": 99670
    },
    {
      "epoch": 1.7500307238540003,
      "grad_norm": 0.047301489699058334,
      "learning_rate": 0.00014958037382356145,
      "loss": 0.8452,
      "step": 99680
    },
    {
      "epoch": 1.7502062887340015,
      "grad_norm": 0.04475055213831268,
      "learning_rate": 0.00014957066381818488,
      "loss": 0.8474,
      "step": 99690
    },
    {
      "epoch": 1.750381853614003,
      "grad_norm": 0.05463109335700984,
      "learning_rate": 0.00014956095319740867,
      "loss": 0.8452,
      "step": 99700
    },
    {
      "epoch": 1.7505574184940045,
      "grad_norm": 0.06878409101547579,
      "learning_rate": 0.00014955124196135592,
      "loss": 0.8544,
      "step": 99710
    },
    {
      "epoch": 1.750732983374006,
      "grad_norm": 0.05012801216514237,
      "learning_rate": 0.00014954153011014962,
      "loss": 0.8441,
      "step": 99720
    },
    {
      "epoch": 1.7509085482540072,
      "grad_norm": 0.08219758977439091,
      "learning_rate": 0.00014953181764391286,
      "loss": 0.8466,
      "step": 99730
    },
    {
      "epoch": 1.7510841131340087,
      "grad_norm": 0.07092488762595216,
      "learning_rate": 0.0001495221045627686,
      "loss": 0.8511,
      "step": 99740
    },
    {
      "epoch": 1.75125967801401,
      "grad_norm": 0.05914972023226812,
      "learning_rate": 0.00014951239086684003,
      "loss": 0.8459,
      "step": 99750
    },
    {
      "epoch": 1.7514352428940114,
      "grad_norm": 0.07313104720578638,
      "learning_rate": 0.00014950267655625013,
      "loss": 0.8407,
      "step": 99760
    },
    {
      "epoch": 1.751610807774013,
      "grad_norm": 0.07709858248861258,
      "learning_rate": 0.00014949296163112205,
      "loss": 0.84,
      "step": 99770
    },
    {
      "epoch": 1.7517863726540144,
      "grad_norm": 0.08932769886023822,
      "learning_rate": 0.00014948324609157887,
      "loss": 0.8517,
      "step": 99780
    },
    {
      "epoch": 1.7519619375340159,
      "grad_norm": 0.05311146323252581,
      "learning_rate": 0.00014947352993774363,
      "loss": 0.8446,
      "step": 99790
    },
    {
      "epoch": 1.7521375024140171,
      "grad_norm": 0.05878548261690554,
      "learning_rate": 0.00014946381316973949,
      "loss": 0.8464,
      "step": 99800
    },
    {
      "epoch": 1.7523130672940184,
      "grad_norm": 0.06003053251791241,
      "learning_rate": 0.00014945409578768953,
      "loss": 0.841,
      "step": 99810
    },
    {
      "epoch": 1.7524886321740198,
      "grad_norm": 0.04652962401904225,
      "learning_rate": 0.00014944437779171695,
      "loss": 0.8461,
      "step": 99820
    },
    {
      "epoch": 1.7526641970540213,
      "grad_norm": 0.05141202943263027,
      "learning_rate": 0.00014943465918194478,
      "loss": 0.8495,
      "step": 99830
    },
    {
      "epoch": 1.7528397619340228,
      "grad_norm": 0.046499952022495815,
      "learning_rate": 0.00014942493995849618,
      "loss": 0.8463,
      "step": 99840
    },
    {
      "epoch": 1.7530153268140243,
      "grad_norm": 0.04853805008628242,
      "learning_rate": 0.00014941522012149435,
      "loss": 0.8434,
      "step": 99850
    },
    {
      "epoch": 1.7531908916940255,
      "grad_norm": 0.046672044019442666,
      "learning_rate": 0.00014940549967106234,
      "loss": 0.8485,
      "step": 99860
    },
    {
      "epoch": 1.7533664565740268,
      "grad_norm": 0.08292372204144013,
      "learning_rate": 0.00014939577860732345,
      "loss": 0.8451,
      "step": 99870
    },
    {
      "epoch": 1.7535420214540283,
      "grad_norm": 0.06969039489655623,
      "learning_rate": 0.00014938605693040068,
      "loss": 0.8398,
      "step": 99880
    },
    {
      "epoch": 1.7537175863340297,
      "grad_norm": 0.06677551880387926,
      "learning_rate": 0.00014937633464041735,
      "loss": 0.8592,
      "step": 99890
    },
    {
      "epoch": 1.7538931512140312,
      "grad_norm": 0.05475033707739741,
      "learning_rate": 0.00014936661173749663,
      "loss": 0.8466,
      "step": 99900
    },
    {
      "epoch": 1.7540687160940327,
      "grad_norm": 0.05036504940148524,
      "learning_rate": 0.0001493568882217616,
      "loss": 0.8501,
      "step": 99910
    },
    {
      "epoch": 1.754244280974034,
      "grad_norm": 0.05512513175458567,
      "learning_rate": 0.00014934716409333555,
      "loss": 0.8523,
      "step": 99920
    },
    {
      "epoch": 1.7544198458540352,
      "grad_norm": 0.04769026227140717,
      "learning_rate": 0.00014933743935234165,
      "loss": 0.8552,
      "step": 99930
    },
    {
      "epoch": 1.7545954107340367,
      "grad_norm": 0.07635643037061669,
      "learning_rate": 0.00014932771399890318,
      "loss": 0.8464,
      "step": 99940
    },
    {
      "epoch": 1.7547709756140382,
      "grad_norm": 0.04669067418617007,
      "learning_rate": 0.0001493179880331433,
      "loss": 0.849,
      "step": 99950
    },
    {
      "epoch": 1.7549465404940396,
      "grad_norm": 0.08934014879094156,
      "learning_rate": 0.0001493082614551852,
      "loss": 0.8477,
      "step": 99960
    },
    {
      "epoch": 1.7551221053740411,
      "grad_norm": 0.04799352907511265,
      "learning_rate": 0.0001492985342651522,
      "loss": 0.8491,
      "step": 99970
    },
    {
      "epoch": 1.7552976702540424,
      "grad_norm": 0.12135115781997591,
      "learning_rate": 0.0001492888064631675,
      "loss": 0.8429,
      "step": 99980
    },
    {
      "epoch": 1.7554732351340436,
      "grad_norm": 0.04495578884984961,
      "learning_rate": 0.0001492790780493544,
      "loss": 0.8436,
      "step": 99990
    },
    {
      "epoch": 1.755648800014045,
      "grad_norm": 0.048582984534774255,
      "learning_rate": 0.00014926934902383612,
      "loss": 0.8468,
      "step": 100000
    },
    {
      "epoch": 1.7558243648940466,
      "grad_norm": 0.05176372208027687,
      "learning_rate": 0.00014925961938673594,
      "loss": 0.8472,
      "step": 100010
    },
    {
      "epoch": 1.755999929774048,
      "grad_norm": 0.05039436918046425,
      "learning_rate": 0.00014924988913817713,
      "loss": 0.8453,
      "step": 100020
    },
    {
      "epoch": 1.7561754946540495,
      "grad_norm": 0.06165304427261012,
      "learning_rate": 0.00014924015827828295,
      "loss": 0.8426,
      "step": 100030
    },
    {
      "epoch": 1.7563510595340508,
      "grad_norm": 0.07770587040256656,
      "learning_rate": 0.00014923042680717674,
      "loss": 0.8433,
      "step": 100040
    },
    {
      "epoch": 1.756526624414052,
      "grad_norm": 0.05299270185481981,
      "learning_rate": 0.00014922069472498176,
      "loss": 0.8473,
      "step": 100050
    },
    {
      "epoch": 1.7567021892940535,
      "grad_norm": 0.06027716082753529,
      "learning_rate": 0.00014921096203182136,
      "loss": 0.8464,
      "step": 100060
    },
    {
      "epoch": 1.756877754174055,
      "grad_norm": 0.04703811901543525,
      "learning_rate": 0.00014920122872781884,
      "loss": 0.8403,
      "step": 100070
    },
    {
      "epoch": 1.7570533190540565,
      "grad_norm": 0.05599309915794096,
      "learning_rate": 0.0001491914948130975,
      "loss": 0.8495,
      "step": 100080
    },
    {
      "epoch": 1.757228883934058,
      "grad_norm": 0.06350336194644891,
      "learning_rate": 0.00014918176028778067,
      "loss": 0.8421,
      "step": 100090
    },
    {
      "epoch": 1.7574044488140592,
      "grad_norm": 0.05745831353746638,
      "learning_rate": 0.00014917202515199172,
      "loss": 0.8437,
      "step": 100100
    },
    {
      "epoch": 1.7575800136940607,
      "grad_norm": 0.04668624993273089,
      "learning_rate": 0.00014916228940585398,
      "loss": 0.8422,
      "step": 100110
    },
    {
      "epoch": 1.757755578574062,
      "grad_norm": 0.04659845552074928,
      "learning_rate": 0.00014915255304949083,
      "loss": 0.8467,
      "step": 100120
    },
    {
      "epoch": 1.7579311434540634,
      "grad_norm": 0.056292166959542984,
      "learning_rate": 0.00014914281608302554,
      "loss": 0.8507,
      "step": 100130
    },
    {
      "epoch": 1.758106708334065,
      "grad_norm": 0.04449088392484191,
      "learning_rate": 0.0001491330785065816,
      "loss": 0.8479,
      "step": 100140
    },
    {
      "epoch": 1.7582822732140664,
      "grad_norm": 0.058031254030030725,
      "learning_rate": 0.0001491233403202823,
      "loss": 0.8507,
      "step": 100150
    },
    {
      "epoch": 1.7584578380940676,
      "grad_norm": 0.045054691943088476,
      "learning_rate": 0.00014911360152425107,
      "loss": 0.8513,
      "step": 100160
    },
    {
      "epoch": 1.7586334029740691,
      "grad_norm": 0.0567007979379267,
      "learning_rate": 0.00014910386211861127,
      "loss": 0.8478,
      "step": 100170
    },
    {
      "epoch": 1.7588089678540704,
      "grad_norm": 0.07371419084725145,
      "learning_rate": 0.00014909412210348632,
      "loss": 0.8359,
      "step": 100180
    },
    {
      "epoch": 1.7589845327340718,
      "grad_norm": 0.08768198212548785,
      "learning_rate": 0.00014908438147899966,
      "loss": 0.85,
      "step": 100190
    },
    {
      "epoch": 1.7591600976140733,
      "grad_norm": 0.056764835307534775,
      "learning_rate": 0.00014907464024527462,
      "loss": 0.8469,
      "step": 100200
    },
    {
      "epoch": 1.7593356624940748,
      "grad_norm": 0.04886168454998745,
      "learning_rate": 0.00014906489840243474,
      "loss": 0.8465,
      "step": 100210
    },
    {
      "epoch": 1.759511227374076,
      "grad_norm": 0.06850907962987916,
      "learning_rate": 0.00014905515595060333,
      "loss": 0.8562,
      "step": 100220
    },
    {
      "epoch": 1.7596867922540775,
      "grad_norm": 0.07381106251839356,
      "learning_rate": 0.0001490454128899039,
      "loss": 0.8425,
      "step": 100230
    },
    {
      "epoch": 1.7598623571340788,
      "grad_norm": 0.05247618864156942,
      "learning_rate": 0.00014903566922045992,
      "loss": 0.8509,
      "step": 100240
    },
    {
      "epoch": 1.7600379220140803,
      "grad_norm": 0.07085769113607933,
      "learning_rate": 0.00014902592494239473,
      "loss": 0.8522,
      "step": 100250
    },
    {
      "epoch": 1.7602134868940817,
      "grad_norm": 0.04984498185638531,
      "learning_rate": 0.00014901618005583192,
      "loss": 0.8511,
      "step": 100260
    },
    {
      "epoch": 1.7603890517740832,
      "grad_norm": 0.05471020597216379,
      "learning_rate": 0.00014900643456089487,
      "loss": 0.8565,
      "step": 100270
    },
    {
      "epoch": 1.7605646166540845,
      "grad_norm": 0.054512240668882425,
      "learning_rate": 0.0001489966884577071,
      "loss": 0.8401,
      "step": 100280
    },
    {
      "epoch": 1.760740181534086,
      "grad_norm": 0.06112030800062166,
      "learning_rate": 0.00014898694174639215,
      "loss": 0.8517,
      "step": 100290
    },
    {
      "epoch": 1.7609157464140872,
      "grad_norm": 0.04902284573274158,
      "learning_rate": 0.00014897719442707335,
      "loss": 0.8518,
      "step": 100300
    },
    {
      "epoch": 1.7610913112940887,
      "grad_norm": 0.08821569782957299,
      "learning_rate": 0.00014896744649987439,
      "loss": 0.8436,
      "step": 100310
    },
    {
      "epoch": 1.7612668761740902,
      "grad_norm": 0.05248170484461329,
      "learning_rate": 0.00014895769796491865,
      "loss": 0.8451,
      "step": 100320
    },
    {
      "epoch": 1.7614424410540916,
      "grad_norm": 0.05413412495039793,
      "learning_rate": 0.0001489479488223297,
      "loss": 0.8538,
      "step": 100330
    },
    {
      "epoch": 1.761618005934093,
      "grad_norm": 0.06757569046279427,
      "learning_rate": 0.00014893819907223102,
      "loss": 0.8455,
      "step": 100340
    },
    {
      "epoch": 1.7617935708140944,
      "grad_norm": 0.06550038417899663,
      "learning_rate": 0.0001489284487147462,
      "loss": 0.8439,
      "step": 100350
    },
    {
      "epoch": 1.7619691356940956,
      "grad_norm": 0.06068355836460201,
      "learning_rate": 0.00014891869774999876,
      "loss": 0.8503,
      "step": 100360
    },
    {
      "epoch": 1.7621447005740971,
      "grad_norm": 0.05286240038847941,
      "learning_rate": 0.0001489089461781122,
      "loss": 0.8485,
      "step": 100370
    },
    {
      "epoch": 1.7623202654540986,
      "grad_norm": 0.064487530455774,
      "learning_rate": 0.00014889919399921014,
      "loss": 0.8466,
      "step": 100380
    },
    {
      "epoch": 1.7624958303341,
      "grad_norm": 0.07444212447438718,
      "learning_rate": 0.0001488894412134161,
      "loss": 0.8421,
      "step": 100390
    },
    {
      "epoch": 1.7626713952141013,
      "grad_norm": 0.06646532810093449,
      "learning_rate": 0.00014887968782085367,
      "loss": 0.8463,
      "step": 100400
    },
    {
      "epoch": 1.7628469600941028,
      "grad_norm": 0.056911191624414194,
      "learning_rate": 0.0001488699338216464,
      "loss": 0.8453,
      "step": 100410
    },
    {
      "epoch": 1.763022524974104,
      "grad_norm": 0.06293564655400136,
      "learning_rate": 0.0001488601792159179,
      "loss": 0.8529,
      "step": 100420
    },
    {
      "epoch": 1.7631980898541055,
      "grad_norm": 0.08176325476034543,
      "learning_rate": 0.00014885042400379178,
      "loss": 0.842,
      "step": 100430
    },
    {
      "epoch": 1.763373654734107,
      "grad_norm": 0.06608154789623585,
      "learning_rate": 0.00014884066818539158,
      "loss": 0.8448,
      "step": 100440
    },
    {
      "epoch": 1.7635492196141085,
      "grad_norm": 0.047199418899846926,
      "learning_rate": 0.00014883091176084098,
      "loss": 0.8475,
      "step": 100450
    },
    {
      "epoch": 1.7637247844941097,
      "grad_norm": 0.06284066574970712,
      "learning_rate": 0.00014882115473026356,
      "loss": 0.8567,
      "step": 100460
    },
    {
      "epoch": 1.7639003493741112,
      "grad_norm": 0.05371101731923757,
      "learning_rate": 0.0001488113970937829,
      "loss": 0.8543,
      "step": 100470
    },
    {
      "epoch": 1.7640759142541125,
      "grad_norm": 0.05203556876907882,
      "learning_rate": 0.0001488016388515227,
      "loss": 0.8449,
      "step": 100480
    },
    {
      "epoch": 1.764251479134114,
      "grad_norm": 0.06071773680845286,
      "learning_rate": 0.0001487918800036066,
      "loss": 0.8511,
      "step": 100490
    },
    {
      "epoch": 1.7644270440141154,
      "grad_norm": 0.07339285351702078,
      "learning_rate": 0.0001487821205501582,
      "loss": 0.845,
      "step": 100500
    },
    {
      "epoch": 1.764602608894117,
      "grad_norm": 0.054703087783145296,
      "learning_rate": 0.00014877236049130118,
      "loss": 0.8449,
      "step": 100510
    },
    {
      "epoch": 1.7647781737741184,
      "grad_norm": 0.057626888994876176,
      "learning_rate": 0.0001487625998271592,
      "loss": 0.8516,
      "step": 100520
    },
    {
      "epoch": 1.7649537386541196,
      "grad_norm": 0.057985516067943596,
      "learning_rate": 0.00014875283855785593,
      "loss": 0.8454,
      "step": 100530
    },
    {
      "epoch": 1.765129303534121,
      "grad_norm": 0.06469423097783478,
      "learning_rate": 0.00014874307668351502,
      "loss": 0.8461,
      "step": 100540
    },
    {
      "epoch": 1.7653048684141224,
      "grad_norm": 0.06404380760314742,
      "learning_rate": 0.0001487333142042602,
      "loss": 0.8487,
      "step": 100550
    },
    {
      "epoch": 1.7654804332941239,
      "grad_norm": 0.056386218532098194,
      "learning_rate": 0.00014872355112021513,
      "loss": 0.8462,
      "step": 100560
    },
    {
      "epoch": 1.7656559981741253,
      "grad_norm": 0.04321119408331251,
      "learning_rate": 0.00014871378743150352,
      "loss": 0.8495,
      "step": 100570
    },
    {
      "epoch": 1.7658315630541268,
      "grad_norm": 0.064112302646263,
      "learning_rate": 0.0001487040231382491,
      "loss": 0.8414,
      "step": 100580
    },
    {
      "epoch": 1.766007127934128,
      "grad_norm": 0.06425962487751692,
      "learning_rate": 0.0001486942582405755,
      "loss": 0.8453,
      "step": 100590
    },
    {
      "epoch": 1.7661826928141293,
      "grad_norm": 0.062313032238668506,
      "learning_rate": 0.0001486844927386066,
      "loss": 0.8392,
      "step": 100600
    },
    {
      "epoch": 1.7663582576941308,
      "grad_norm": 0.05313284477607172,
      "learning_rate": 0.00014867472663246598,
      "loss": 0.843,
      "step": 100610
    },
    {
      "epoch": 1.7665338225741323,
      "grad_norm": 0.094146501237338,
      "learning_rate": 0.00014866495992227745,
      "loss": 0.8558,
      "step": 100620
    },
    {
      "epoch": 1.7667093874541338,
      "grad_norm": 0.06353137042122002,
      "learning_rate": 0.00014865519260816477,
      "loss": 0.8546,
      "step": 100630
    },
    {
      "epoch": 1.7668849523341352,
      "grad_norm": 0.04700247750601395,
      "learning_rate": 0.00014864542469025165,
      "loss": 0.8488,
      "step": 100640
    },
    {
      "epoch": 1.7670605172141365,
      "grad_norm": 0.061492531136042694,
      "learning_rate": 0.00014863565616866185,
      "loss": 0.8388,
      "step": 100650
    },
    {
      "epoch": 1.7672360820941377,
      "grad_norm": 0.04645354614753374,
      "learning_rate": 0.00014862588704351914,
      "loss": 0.842,
      "step": 100660
    },
    {
      "epoch": 1.7674116469741392,
      "grad_norm": 0.0727807154995062,
      "learning_rate": 0.00014861611731494735,
      "loss": 0.846,
      "step": 100670
    },
    {
      "epoch": 1.7675872118541407,
      "grad_norm": 0.06709063910393628,
      "learning_rate": 0.0001486063469830702,
      "loss": 0.841,
      "step": 100680
    },
    {
      "epoch": 1.7677627767341422,
      "grad_norm": 0.050845295151977644,
      "learning_rate": 0.00014859657604801156,
      "loss": 0.8442,
      "step": 100690
    },
    {
      "epoch": 1.7679383416141436,
      "grad_norm": 0.05003359251736988,
      "learning_rate": 0.00014858680450989512,
      "loss": 0.8435,
      "step": 100700
    },
    {
      "epoch": 1.768113906494145,
      "grad_norm": 0.055903475971935034,
      "learning_rate": 0.0001485770323688448,
      "loss": 0.8397,
      "step": 100710
    },
    {
      "epoch": 1.7682894713741462,
      "grad_norm": 0.04983417034678124,
      "learning_rate": 0.00014856725962498433,
      "loss": 0.843,
      "step": 100720
    },
    {
      "epoch": 1.7684650362541476,
      "grad_norm": 0.058744119984620535,
      "learning_rate": 0.00014855748627843756,
      "loss": 0.8454,
      "step": 100730
    },
    {
      "epoch": 1.7686406011341491,
      "grad_norm": 0.0630039018145193,
      "learning_rate": 0.00014854771232932835,
      "loss": 0.8442,
      "step": 100740
    },
    {
      "epoch": 1.7688161660141506,
      "grad_norm": 0.07607105023826238,
      "learning_rate": 0.0001485379377777805,
      "loss": 0.851,
      "step": 100750
    },
    {
      "epoch": 1.768991730894152,
      "grad_norm": 0.05024397336875885,
      "learning_rate": 0.00014852816262391784,
      "loss": 0.8504,
      "step": 100760
    },
    {
      "epoch": 1.7691672957741533,
      "grad_norm": 0.05067327300843907,
      "learning_rate": 0.00014851838686786429,
      "loss": 0.844,
      "step": 100770
    },
    {
      "epoch": 1.7693428606541546,
      "grad_norm": 0.06241674059093423,
      "learning_rate": 0.00014850861050974364,
      "loss": 0.8394,
      "step": 100780
    },
    {
      "epoch": 1.769518425534156,
      "grad_norm": 0.04794849210018127,
      "learning_rate": 0.00014849883354967983,
      "loss": 0.8441,
      "step": 100790
    },
    {
      "epoch": 1.7696939904141575,
      "grad_norm": 0.06065690823980655,
      "learning_rate": 0.00014848905598779667,
      "loss": 0.8436,
      "step": 100800
    },
    {
      "epoch": 1.769869555294159,
      "grad_norm": 0.04936322614987246,
      "learning_rate": 0.00014847927782421806,
      "loss": 0.8474,
      "step": 100810
    },
    {
      "epoch": 1.7700451201741605,
      "grad_norm": 0.0618279587382625,
      "learning_rate": 0.00014846949905906792,
      "loss": 0.8545,
      "step": 100820
    },
    {
      "epoch": 1.7702206850541617,
      "grad_norm": 0.05910263160699115,
      "learning_rate": 0.00014845971969247013,
      "loss": 0.84,
      "step": 100830
    },
    {
      "epoch": 1.7703962499341632,
      "grad_norm": 0.05147808291324231,
      "learning_rate": 0.00014844993972454857,
      "loss": 0.8455,
      "step": 100840
    },
    {
      "epoch": 1.7705718148141645,
      "grad_norm": 0.06899171740338138,
      "learning_rate": 0.00014844015915542718,
      "loss": 0.851,
      "step": 100850
    },
    {
      "epoch": 1.770747379694166,
      "grad_norm": 0.05472539666052342,
      "learning_rate": 0.00014843037798522994,
      "loss": 0.8465,
      "step": 100860
    },
    {
      "epoch": 1.7709229445741674,
      "grad_norm": 0.05201139708088321,
      "learning_rate": 0.00014842059621408067,
      "loss": 0.8539,
      "step": 100870
    },
    {
      "epoch": 1.771098509454169,
      "grad_norm": 0.0590639585639497,
      "learning_rate": 0.00014841081384210338,
      "loss": 0.8392,
      "step": 100880
    },
    {
      "epoch": 1.7712740743341702,
      "grad_norm": 0.06235000952710594,
      "learning_rate": 0.00014840103086942197,
      "loss": 0.849,
      "step": 100890
    },
    {
      "epoch": 1.7714496392141716,
      "grad_norm": 0.05638156975131798,
      "learning_rate": 0.0001483912472961604,
      "loss": 0.8471,
      "step": 100900
    },
    {
      "epoch": 1.771625204094173,
      "grad_norm": 0.05578005632932971,
      "learning_rate": 0.00014838146312244269,
      "loss": 0.8508,
      "step": 100910
    },
    {
      "epoch": 1.7718007689741744,
      "grad_norm": 0.04523307547013058,
      "learning_rate": 0.00014837167834839276,
      "loss": 0.8493,
      "step": 100920
    },
    {
      "epoch": 1.7719763338541759,
      "grad_norm": 0.07163411894290149,
      "learning_rate": 0.00014836189297413454,
      "loss": 0.8447,
      "step": 100930
    },
    {
      "epoch": 1.7721518987341773,
      "grad_norm": 0.06775934715220515,
      "learning_rate": 0.0001483521069997921,
      "loss": 0.8413,
      "step": 100940
    },
    {
      "epoch": 1.7723274636141786,
      "grad_norm": 0.07516820385198386,
      "learning_rate": 0.00014834232042548938,
      "loss": 0.8499,
      "step": 100950
    },
    {
      "epoch": 1.77250302849418,
      "grad_norm": 0.04996942122537086,
      "learning_rate": 0.00014833253325135037,
      "loss": 0.8445,
      "step": 100960
    },
    {
      "epoch": 1.7726785933741813,
      "grad_norm": 0.056483648982587294,
      "learning_rate": 0.00014832274547749912,
      "loss": 0.8452,
      "step": 100970
    },
    {
      "epoch": 1.7728541582541828,
      "grad_norm": 0.05585237917456498,
      "learning_rate": 0.00014831295710405958,
      "loss": 0.8546,
      "step": 100980
    },
    {
      "epoch": 1.7730297231341843,
      "grad_norm": 0.08726791194681192,
      "learning_rate": 0.00014830316813115583,
      "loss": 0.8459,
      "step": 100990
    },
    {
      "epoch": 1.7732052880141858,
      "grad_norm": 0.05581989961921171,
      "learning_rate": 0.0001482933785589119,
      "loss": 0.8486,
      "step": 101000
    },
    {
      "epoch": 1.773380852894187,
      "grad_norm": 0.08220749099354357,
      "learning_rate": 0.00014828358838745175,
      "loss": 0.8514,
      "step": 101010
    },
    {
      "epoch": 1.7735564177741885,
      "grad_norm": 0.045175673862986626,
      "learning_rate": 0.0001482737976168995,
      "loss": 0.8485,
      "step": 101020
    },
    {
      "epoch": 1.7737319826541897,
      "grad_norm": 0.049875292374703506,
      "learning_rate": 0.00014826400624737916,
      "loss": 0.8509,
      "step": 101030
    },
    {
      "epoch": 1.7739075475341912,
      "grad_norm": 0.06044144183364798,
      "learning_rate": 0.00014825421427901484,
      "loss": 0.8501,
      "step": 101040
    },
    {
      "epoch": 1.7740831124141927,
      "grad_norm": 0.06702917999180036,
      "learning_rate": 0.00014824442171193053,
      "loss": 0.84,
      "step": 101050
    },
    {
      "epoch": 1.7742586772941942,
      "grad_norm": 0.07688567253273439,
      "learning_rate": 0.00014823462854625033,
      "loss": 0.8431,
      "step": 101060
    },
    {
      "epoch": 1.7744342421741954,
      "grad_norm": 0.048306066993318714,
      "learning_rate": 0.00014822483478209834,
      "loss": 0.8437,
      "step": 101070
    },
    {
      "epoch": 1.774609807054197,
      "grad_norm": 0.056960795923288246,
      "learning_rate": 0.0001482150404195987,
      "loss": 0.853,
      "step": 101080
    },
    {
      "epoch": 1.7747853719341982,
      "grad_norm": 0.0684016241423637,
      "learning_rate": 0.00014820524545887542,
      "loss": 0.8472,
      "step": 101090
    },
    {
      "epoch": 1.7749609368141996,
      "grad_norm": 0.0548586571888273,
      "learning_rate": 0.0001481954499000526,
      "loss": 0.8449,
      "step": 101100
    },
    {
      "epoch": 1.7751365016942011,
      "grad_norm": 0.054664972028250304,
      "learning_rate": 0.00014818565374325441,
      "loss": 0.836,
      "step": 101110
    },
    {
      "epoch": 1.7753120665742026,
      "grad_norm": 0.06549195241553156,
      "learning_rate": 0.00014817585698860495,
      "loss": 0.8451,
      "step": 101120
    },
    {
      "epoch": 1.7754876314542039,
      "grad_norm": 0.0615898116067138,
      "learning_rate": 0.0001481660596362283,
      "loss": 0.8455,
      "step": 101130
    },
    {
      "epoch": 1.7756631963342053,
      "grad_norm": 0.050525238111119766,
      "learning_rate": 0.0001481562616862487,
      "loss": 0.8359,
      "step": 101140
    },
    {
      "epoch": 1.7758387612142066,
      "grad_norm": 0.06543317997384115,
      "learning_rate": 0.00014814646313879016,
      "loss": 0.8517,
      "step": 101150
    },
    {
      "epoch": 1.776014326094208,
      "grad_norm": 0.07318967959777525,
      "learning_rate": 0.00014813666399397695,
      "loss": 0.846,
      "step": 101160
    },
    {
      "epoch": 1.7761898909742095,
      "grad_norm": 0.0501240776984155,
      "learning_rate": 0.00014812686425193315,
      "loss": 0.851,
      "step": 101170
    },
    {
      "epoch": 1.776365455854211,
      "grad_norm": 0.09731336800374236,
      "learning_rate": 0.00014811706391278296,
      "loss": 0.8408,
      "step": 101180
    },
    {
      "epoch": 1.7765410207342123,
      "grad_norm": 0.072232811738798,
      "learning_rate": 0.00014810726297665052,
      "loss": 0.839,
      "step": 101190
    },
    {
      "epoch": 1.7767165856142138,
      "grad_norm": 0.09995607463330111,
      "learning_rate": 0.00014809746144366003,
      "loss": 0.8495,
      "step": 101200
    },
    {
      "epoch": 1.776892150494215,
      "grad_norm": 0.0660836264462974,
      "learning_rate": 0.0001480876593139357,
      "loss": 0.8521,
      "step": 101210
    },
    {
      "epoch": 1.7770677153742165,
      "grad_norm": 0.06392327709092102,
      "learning_rate": 0.0001480778565876017,
      "loss": 0.851,
      "step": 101220
    },
    {
      "epoch": 1.777243280254218,
      "grad_norm": 0.06581778805393063,
      "learning_rate": 0.00014806805326478224,
      "loss": 0.8514,
      "step": 101230
    },
    {
      "epoch": 1.7774188451342194,
      "grad_norm": 0.04420255280691539,
      "learning_rate": 0.0001480582493456015,
      "loss": 0.8537,
      "step": 101240
    },
    {
      "epoch": 1.777594410014221,
      "grad_norm": 0.07008393464198444,
      "learning_rate": 0.00014804844483018372,
      "loss": 0.8548,
      "step": 101250
    },
    {
      "epoch": 1.7777699748942222,
      "grad_norm": 0.059887344745140514,
      "learning_rate": 0.00014803863971865315,
      "loss": 0.8454,
      "step": 101260
    },
    {
      "epoch": 1.7779455397742234,
      "grad_norm": 0.05046782718249097,
      "learning_rate": 0.000148028834011134,
      "loss": 0.853,
      "step": 101270
    },
    {
      "epoch": 1.778121104654225,
      "grad_norm": 0.05420184002693551,
      "learning_rate": 0.0001480190277077505,
      "loss": 0.85,
      "step": 101280
    },
    {
      "epoch": 1.7782966695342264,
      "grad_norm": 0.09376333104647253,
      "learning_rate": 0.0001480092208086269,
      "loss": 0.8494,
      "step": 101290
    },
    {
      "epoch": 1.7784722344142279,
      "grad_norm": 0.04946686505756815,
      "learning_rate": 0.00014799941331388746,
      "loss": 0.8411,
      "step": 101300
    },
    {
      "epoch": 1.7786477992942293,
      "grad_norm": 0.06190400029455918,
      "learning_rate": 0.00014798960522365647,
      "loss": 0.8494,
      "step": 101310
    },
    {
      "epoch": 1.7788233641742306,
      "grad_norm": 0.05823569898665038,
      "learning_rate": 0.00014797979653805816,
      "loss": 0.8462,
      "step": 101320
    },
    {
      "epoch": 1.7789989290542318,
      "grad_norm": 0.0718487488360185,
      "learning_rate": 0.00014796998725721683,
      "loss": 0.8535,
      "step": 101330
    },
    {
      "epoch": 1.7791744939342333,
      "grad_norm": 0.061922170145356524,
      "learning_rate": 0.00014796017738125675,
      "loss": 0.8549,
      "step": 101340
    },
    {
      "epoch": 1.7793500588142348,
      "grad_norm": 0.04583549245291089,
      "learning_rate": 0.0001479503669103022,
      "loss": 0.8438,
      "step": 101350
    },
    {
      "epoch": 1.7795256236942363,
      "grad_norm": 0.047970068895734845,
      "learning_rate": 0.00014794055584447754,
      "loss": 0.8459,
      "step": 101360
    },
    {
      "epoch": 1.7797011885742378,
      "grad_norm": 0.049584863851275904,
      "learning_rate": 0.00014793074418390704,
      "loss": 0.8432,
      "step": 101370
    },
    {
      "epoch": 1.779876753454239,
      "grad_norm": 0.05152334878264013,
      "learning_rate": 0.00014792093192871503,
      "loss": 0.852,
      "step": 101380
    },
    {
      "epoch": 1.7800523183342403,
      "grad_norm": 0.06557856343106847,
      "learning_rate": 0.0001479111190790258,
      "loss": 0.8493,
      "step": 101390
    },
    {
      "epoch": 1.7802278832142417,
      "grad_norm": 0.06324425931943413,
      "learning_rate": 0.00014790130563496371,
      "loss": 0.8396,
      "step": 101400
    },
    {
      "epoch": 1.7804034480942432,
      "grad_norm": 0.060978057819076475,
      "learning_rate": 0.00014789149159665304,
      "loss": 0.8448,
      "step": 101410
    },
    {
      "epoch": 1.7805790129742447,
      "grad_norm": 0.05403525395492698,
      "learning_rate": 0.00014788167696421826,
      "loss": 0.8471,
      "step": 101420
    },
    {
      "epoch": 1.7807545778542462,
      "grad_norm": 0.09556747109066605,
      "learning_rate": 0.0001478718617377836,
      "loss": 0.8463,
      "step": 101430
    },
    {
      "epoch": 1.7809301427342474,
      "grad_norm": 0.06018654051923663,
      "learning_rate": 0.0001478620459174735,
      "loss": 0.8451,
      "step": 101440
    },
    {
      "epoch": 1.7811057076142487,
      "grad_norm": 0.06430974489171216,
      "learning_rate": 0.00014785222950341228,
      "loss": 0.8408,
      "step": 101450
    },
    {
      "epoch": 1.7812812724942502,
      "grad_norm": 0.0559508951009287,
      "learning_rate": 0.00014784241249572432,
      "loss": 0.8451,
      "step": 101460
    },
    {
      "epoch": 1.7814568373742516,
      "grad_norm": 0.055620714110868504,
      "learning_rate": 0.00014783259489453403,
      "loss": 0.8477,
      "step": 101470
    },
    {
      "epoch": 1.7816324022542531,
      "grad_norm": 0.06577602787601951,
      "learning_rate": 0.0001478227766999658,
      "loss": 0.8489,
      "step": 101480
    },
    {
      "epoch": 1.7818079671342546,
      "grad_norm": 0.05292902977926135,
      "learning_rate": 0.000147812957912144,
      "loss": 0.8457,
      "step": 101490
    },
    {
      "epoch": 1.7819835320142559,
      "grad_norm": 0.04397960627941302,
      "learning_rate": 0.00014780313853119306,
      "loss": 0.8446,
      "step": 101500
    },
    {
      "epoch": 1.782159096894257,
      "grad_norm": 0.06679736825268157,
      "learning_rate": 0.00014779331855723735,
      "loss": 0.8475,
      "step": 101510
    },
    {
      "epoch": 1.7823346617742586,
      "grad_norm": 0.04961168977111472,
      "learning_rate": 0.00014778349799040138,
      "loss": 0.8393,
      "step": 101520
    },
    {
      "epoch": 1.78251022665426,
      "grad_norm": 0.06454017177268961,
      "learning_rate": 0.00014777367683080947,
      "loss": 0.8407,
      "step": 101530
    },
    {
      "epoch": 1.7826857915342615,
      "grad_norm": 0.07343562239440282,
      "learning_rate": 0.00014776385507858617,
      "loss": 0.8426,
      "step": 101540
    },
    {
      "epoch": 1.782861356414263,
      "grad_norm": 0.06359886007828307,
      "learning_rate": 0.00014775403273385585,
      "loss": 0.8467,
      "step": 101550
    },
    {
      "epoch": 1.7830369212942643,
      "grad_norm": 0.06024307554375262,
      "learning_rate": 0.00014774420979674294,
      "loss": 0.841,
      "step": 101560
    },
    {
      "epoch": 1.7832124861742658,
      "grad_norm": 0.0530713906080161,
      "learning_rate": 0.00014773438626737196,
      "loss": 0.8457,
      "step": 101570
    },
    {
      "epoch": 1.783388051054267,
      "grad_norm": 0.059397641105539786,
      "learning_rate": 0.00014772456214586733,
      "loss": 0.8422,
      "step": 101580
    },
    {
      "epoch": 1.7835636159342685,
      "grad_norm": 0.07803799157000206,
      "learning_rate": 0.00014771473743235353,
      "loss": 0.8459,
      "step": 101590
    },
    {
      "epoch": 1.78373918081427,
      "grad_norm": 0.06235103998629851,
      "learning_rate": 0.0001477049121269551,
      "loss": 0.8467,
      "step": 101600
    },
    {
      "epoch": 1.7839147456942714,
      "grad_norm": 0.06034672750187838,
      "learning_rate": 0.00014769508622979643,
      "loss": 0.848,
      "step": 101610
    },
    {
      "epoch": 1.7840903105742727,
      "grad_norm": 0.05770953330332583,
      "learning_rate": 0.0001476852597410021,
      "loss": 0.849,
      "step": 101620
    },
    {
      "epoch": 1.7842658754542742,
      "grad_norm": 0.055126047522359174,
      "learning_rate": 0.0001476754326606966,
      "loss": 0.8421,
      "step": 101630
    },
    {
      "epoch": 1.7844414403342754,
      "grad_norm": 0.05996327242886333,
      "learning_rate": 0.00014766560498900443,
      "loss": 0.8481,
      "step": 101640
    },
    {
      "epoch": 1.784617005214277,
      "grad_norm": 0.05246913226024682,
      "learning_rate": 0.00014765577672605008,
      "loss": 0.8399,
      "step": 101650
    },
    {
      "epoch": 1.7847925700942784,
      "grad_norm": 0.05494524213105213,
      "learning_rate": 0.00014764594787195807,
      "loss": 0.8473,
      "step": 101660
    },
    {
      "epoch": 1.7849681349742799,
      "grad_norm": 0.060479923560556044,
      "learning_rate": 0.000147636118426853,
      "loss": 0.8524,
      "step": 101670
    },
    {
      "epoch": 1.7851436998542811,
      "grad_norm": 0.05372316787884148,
      "learning_rate": 0.00014762628839085933,
      "loss": 0.8521,
      "step": 101680
    },
    {
      "epoch": 1.7853192647342826,
      "grad_norm": 0.06939730938543813,
      "learning_rate": 0.00014761645776410168,
      "loss": 0.8438,
      "step": 101690
    },
    {
      "epoch": 1.7854948296142839,
      "grad_norm": 0.07083962537369573,
      "learning_rate": 0.00014760662654670457,
      "loss": 0.8404,
      "step": 101700
    },
    {
      "epoch": 1.7856703944942853,
      "grad_norm": 0.05248351872528578,
      "learning_rate": 0.00014759679473879257,
      "loss": 0.8519,
      "step": 101710
    },
    {
      "epoch": 1.7858459593742868,
      "grad_norm": 0.07193517357938481,
      "learning_rate": 0.00014758696234049026,
      "loss": 0.8435,
      "step": 101720
    },
    {
      "epoch": 1.7860215242542883,
      "grad_norm": 0.05074871076308554,
      "learning_rate": 0.0001475771293519222,
      "loss": 0.8492,
      "step": 101730
    },
    {
      "epoch": 1.7861970891342895,
      "grad_norm": 0.06581942303531119,
      "learning_rate": 0.00014756729577321297,
      "loss": 0.8468,
      "step": 101740
    },
    {
      "epoch": 1.786372654014291,
      "grad_norm": 0.07734018984056511,
      "learning_rate": 0.00014755746160448717,
      "loss": 0.846,
      "step": 101750
    },
    {
      "epoch": 1.7865482188942923,
      "grad_norm": 0.06453578337164805,
      "learning_rate": 0.00014754762684586947,
      "loss": 0.8476,
      "step": 101760
    },
    {
      "epoch": 1.7867237837742938,
      "grad_norm": 0.05074343294943287,
      "learning_rate": 0.00014753779149748436,
      "loss": 0.8399,
      "step": 101770
    },
    {
      "epoch": 1.7868993486542952,
      "grad_norm": 0.05114710617510233,
      "learning_rate": 0.00014752795555945657,
      "loss": 0.8468,
      "step": 101780
    },
    {
      "epoch": 1.7870749135342967,
      "grad_norm": 0.056488428835735904,
      "learning_rate": 0.0001475181190319106,
      "loss": 0.8475,
      "step": 101790
    },
    {
      "epoch": 1.787250478414298,
      "grad_norm": 0.051217428808915104,
      "learning_rate": 0.00014750828191497116,
      "loss": 0.8555,
      "step": 101800
    },
    {
      "epoch": 1.7874260432942994,
      "grad_norm": 0.05375236945795411,
      "learning_rate": 0.0001474984442087629,
      "loss": 0.8453,
      "step": 101810
    },
    {
      "epoch": 1.7876016081743007,
      "grad_norm": 0.05269378744140611,
      "learning_rate": 0.00014748860591341044,
      "loss": 0.8431,
      "step": 101820
    },
    {
      "epoch": 1.7877771730543022,
      "grad_norm": 0.04814050045848598,
      "learning_rate": 0.00014747876702903842,
      "loss": 0.8438,
      "step": 101830
    },
    {
      "epoch": 1.7879527379343036,
      "grad_norm": 0.06697053664042446,
      "learning_rate": 0.00014746892755577152,
      "loss": 0.8416,
      "step": 101840
    },
    {
      "epoch": 1.7881283028143051,
      "grad_norm": 0.043050313054701646,
      "learning_rate": 0.00014745908749373438,
      "loss": 0.8401,
      "step": 101850
    },
    {
      "epoch": 1.7883038676943064,
      "grad_norm": 0.07047753940700073,
      "learning_rate": 0.0001474492468430517,
      "loss": 0.8533,
      "step": 101860
    },
    {
      "epoch": 1.7884794325743079,
      "grad_norm": 0.06298023151847378,
      "learning_rate": 0.0001474394056038482,
      "loss": 0.8464,
      "step": 101870
    },
    {
      "epoch": 1.7886549974543091,
      "grad_norm": 0.06643951785659453,
      "learning_rate": 0.0001474295637762485,
      "loss": 0.8461,
      "step": 101880
    },
    {
      "epoch": 1.7888305623343106,
      "grad_norm": 0.06444923881257214,
      "learning_rate": 0.00014741972136037733,
      "loss": 0.8406,
      "step": 101890
    },
    {
      "epoch": 1.789006127214312,
      "grad_norm": 0.05757107251178884,
      "learning_rate": 0.0001474098783563594,
      "loss": 0.8459,
      "step": 101900
    },
    {
      "epoch": 1.7891816920943135,
      "grad_norm": 0.056272644738241784,
      "learning_rate": 0.00014740003476431944,
      "loss": 0.8576,
      "step": 101910
    },
    {
      "epoch": 1.7893572569743148,
      "grad_norm": 0.0770077432479371,
      "learning_rate": 0.00014739019058438213,
      "loss": 0.8503,
      "step": 101920
    },
    {
      "epoch": 1.7895328218543163,
      "grad_norm": 0.05882316841005904,
      "learning_rate": 0.0001473803458166722,
      "loss": 0.8403,
      "step": 101930
    },
    {
      "epoch": 1.7897083867343175,
      "grad_norm": 0.055109512564710236,
      "learning_rate": 0.00014737050046131442,
      "loss": 0.841,
      "step": 101940
    },
    {
      "epoch": 1.789883951614319,
      "grad_norm": 0.04839498738869534,
      "learning_rate": 0.0001473606545184335,
      "loss": 0.8499,
      "step": 101950
    },
    {
      "epoch": 1.7900595164943205,
      "grad_norm": 0.0639570648007624,
      "learning_rate": 0.0001473508079881542,
      "loss": 0.8454,
      "step": 101960
    },
    {
      "epoch": 1.790235081374322,
      "grad_norm": 0.05862913615571413,
      "learning_rate": 0.00014734096087060127,
      "loss": 0.8565,
      "step": 101970
    },
    {
      "epoch": 1.7904106462543234,
      "grad_norm": 0.05466912141904713,
      "learning_rate": 0.0001473311131658995,
      "loss": 0.8447,
      "step": 101980
    },
    {
      "epoch": 1.7905862111343247,
      "grad_norm": 0.05277488870897125,
      "learning_rate": 0.00014732126487417363,
      "loss": 0.8503,
      "step": 101990
    },
    {
      "epoch": 1.790761776014326,
      "grad_norm": 0.0754790178391736,
      "learning_rate": 0.0001473114159955485,
      "loss": 0.8532,
      "step": 102000
    },
    {
      "epoch": 1.7909373408943274,
      "grad_norm": 0.060038014133459,
      "learning_rate": 0.00014730156653014882,
      "loss": 0.8453,
      "step": 102010
    },
    {
      "epoch": 1.791112905774329,
      "grad_norm": 0.05223794767526734,
      "learning_rate": 0.0001472917164780994,
      "loss": 0.8389,
      "step": 102020
    },
    {
      "epoch": 1.7912884706543304,
      "grad_norm": 0.07767486092138637,
      "learning_rate": 0.0001472818658395251,
      "loss": 0.8506,
      "step": 102030
    },
    {
      "epoch": 1.7914640355343319,
      "grad_norm": 0.06351692640950389,
      "learning_rate": 0.00014727201461455067,
      "loss": 0.8454,
      "step": 102040
    },
    {
      "epoch": 1.7916396004143331,
      "grad_norm": 0.05823319581637103,
      "learning_rate": 0.00014726216280330095,
      "loss": 0.8525,
      "step": 102050
    },
    {
      "epoch": 1.7918151652943344,
      "grad_norm": 0.04903576617211271,
      "learning_rate": 0.00014725231040590078,
      "loss": 0.8452,
      "step": 102060
    },
    {
      "epoch": 1.7919907301743359,
      "grad_norm": 0.04740030860691719,
      "learning_rate": 0.00014724245742247494,
      "loss": 0.8471,
      "step": 102070
    },
    {
      "epoch": 1.7921662950543373,
      "grad_norm": 0.04719420397710098,
      "learning_rate": 0.00014723260385314833,
      "loss": 0.8451,
      "step": 102080
    },
    {
      "epoch": 1.7923418599343388,
      "grad_norm": 0.06674261880379148,
      "learning_rate": 0.00014722274969804574,
      "loss": 0.8489,
      "step": 102090
    },
    {
      "epoch": 1.7925174248143403,
      "grad_norm": 0.05430970113027364,
      "learning_rate": 0.00014721289495729206,
      "loss": 0.8474,
      "step": 102100
    },
    {
      "epoch": 1.7926929896943415,
      "grad_norm": 0.06357294563170299,
      "learning_rate": 0.0001472030396310122,
      "loss": 0.8431,
      "step": 102110
    },
    {
      "epoch": 1.7928685545743428,
      "grad_norm": 0.055814020357151245,
      "learning_rate": 0.0001471931837193309,
      "loss": 0.8479,
      "step": 102120
    },
    {
      "epoch": 1.7930441194543443,
      "grad_norm": 0.05706718441674296,
      "learning_rate": 0.00014718332722237314,
      "loss": 0.854,
      "step": 102130
    },
    {
      "epoch": 1.7932196843343458,
      "grad_norm": 0.06901388599966833,
      "learning_rate": 0.00014717347014026378,
      "loss": 0.8461,
      "step": 102140
    },
    {
      "epoch": 1.7933952492143472,
      "grad_norm": 0.04366296156272304,
      "learning_rate": 0.00014716361247312768,
      "loss": 0.8443,
      "step": 102150
    },
    {
      "epoch": 1.7935708140943487,
      "grad_norm": 0.054147528180751525,
      "learning_rate": 0.00014715375422108976,
      "loss": 0.8498,
      "step": 102160
    },
    {
      "epoch": 1.79374637897435,
      "grad_norm": 0.05788233516747634,
      "learning_rate": 0.00014714389538427494,
      "loss": 0.8445,
      "step": 102170
    },
    {
      "epoch": 1.7939219438543512,
      "grad_norm": 0.0817055486579092,
      "learning_rate": 0.00014713403596280813,
      "loss": 0.8434,
      "step": 102180
    },
    {
      "epoch": 1.7940975087343527,
      "grad_norm": 0.047536495711713575,
      "learning_rate": 0.00014712417595681425,
      "loss": 0.8505,
      "step": 102190
    },
    {
      "epoch": 1.7942730736143542,
      "grad_norm": 0.048547500733285634,
      "learning_rate": 0.00014711431536641823,
      "loss": 0.8412,
      "step": 102200
    },
    {
      "epoch": 1.7944486384943557,
      "grad_norm": 0.047268529633787436,
      "learning_rate": 0.00014710445419174496,
      "loss": 0.8474,
      "step": 102210
    },
    {
      "epoch": 1.7946242033743571,
      "grad_norm": 0.06132348267493282,
      "learning_rate": 0.00014709459243291944,
      "loss": 0.8421,
      "step": 102220
    },
    {
      "epoch": 1.7947997682543584,
      "grad_norm": 0.05708710685207058,
      "learning_rate": 0.0001470847300900666,
      "loss": 0.8454,
      "step": 102230
    },
    {
      "epoch": 1.7949753331343596,
      "grad_norm": 0.05154143359238834,
      "learning_rate": 0.00014707486716331137,
      "loss": 0.8478,
      "step": 102240
    },
    {
      "epoch": 1.7951508980143611,
      "grad_norm": 0.05925004353643488,
      "learning_rate": 0.00014706500365277878,
      "loss": 0.85,
      "step": 102250
    },
    {
      "epoch": 1.7953264628943626,
      "grad_norm": 0.0726254967199455,
      "learning_rate": 0.00014705513955859373,
      "loss": 0.8412,
      "step": 102260
    },
    {
      "epoch": 1.795502027774364,
      "grad_norm": 0.06378930706408324,
      "learning_rate": 0.00014704527488088127,
      "loss": 0.8483,
      "step": 102270
    },
    {
      "epoch": 1.7956775926543656,
      "grad_norm": 0.06104439701786397,
      "learning_rate": 0.00014703540961976636,
      "loss": 0.846,
      "step": 102280
    },
    {
      "epoch": 1.7958531575343668,
      "grad_norm": 0.0632376086451732,
      "learning_rate": 0.00014702554377537394,
      "loss": 0.8576,
      "step": 102290
    },
    {
      "epoch": 1.7960287224143683,
      "grad_norm": 0.06101145643945044,
      "learning_rate": 0.0001470156773478291,
      "loss": 0.8496,
      "step": 102300
    },
    {
      "epoch": 1.7962042872943695,
      "grad_norm": 0.054766342471326585,
      "learning_rate": 0.0001470058103372568,
      "loss": 0.8492,
      "step": 102310
    },
    {
      "epoch": 1.796379852174371,
      "grad_norm": 0.057436648538460804,
      "learning_rate": 0.00014699594274378204,
      "loss": 0.8431,
      "step": 102320
    },
    {
      "epoch": 1.7965554170543725,
      "grad_norm": 0.0660988485255681,
      "learning_rate": 0.0001469860745675299,
      "loss": 0.8512,
      "step": 102330
    },
    {
      "epoch": 1.796730981934374,
      "grad_norm": 0.07351848577306126,
      "learning_rate": 0.0001469762058086254,
      "loss": 0.8508,
      "step": 102340
    },
    {
      "epoch": 1.7969065468143752,
      "grad_norm": 0.07747063522398563,
      "learning_rate": 0.00014696633646719355,
      "loss": 0.8471,
      "step": 102350
    },
    {
      "epoch": 1.7970821116943767,
      "grad_norm": 0.07062377756482936,
      "learning_rate": 0.00014695646654335937,
      "loss": 0.8531,
      "step": 102360
    },
    {
      "epoch": 1.797257676574378,
      "grad_norm": 0.04703631601040788,
      "learning_rate": 0.00014694659603724797,
      "loss": 0.8496,
      "step": 102370
    },
    {
      "epoch": 1.7974332414543794,
      "grad_norm": 0.06556615885234554,
      "learning_rate": 0.0001469367249489844,
      "loss": 0.8449,
      "step": 102380
    },
    {
      "epoch": 1.797608806334381,
      "grad_norm": 0.0655673188562377,
      "learning_rate": 0.0001469268532786937,
      "loss": 0.8469,
      "step": 102390
    },
    {
      "epoch": 1.7977843712143824,
      "grad_norm": 0.0559139939919967,
      "learning_rate": 0.000146916981026501,
      "loss": 0.8418,
      "step": 102400
    },
    {
      "epoch": 1.7979599360943836,
      "grad_norm": 0.050974723687880454,
      "learning_rate": 0.00014690710819253134,
      "loss": 0.8515,
      "step": 102410
    },
    {
      "epoch": 1.7981355009743851,
      "grad_norm": 0.07335299773173554,
      "learning_rate": 0.00014689723477690983,
      "loss": 0.8439,
      "step": 102420
    },
    {
      "epoch": 1.7983110658543864,
      "grad_norm": 0.04932177645946533,
      "learning_rate": 0.0001468873607797615,
      "loss": 0.8409,
      "step": 102430
    },
    {
      "epoch": 1.7984866307343879,
      "grad_norm": 0.05934820237665754,
      "learning_rate": 0.00014687748620121157,
      "loss": 0.8456,
      "step": 102440
    },
    {
      "epoch": 1.7986621956143893,
      "grad_norm": 0.05785393724458521,
      "learning_rate": 0.00014686761104138512,
      "loss": 0.8464,
      "step": 102450
    },
    {
      "epoch": 1.7988377604943908,
      "grad_norm": 0.0726213398144918,
      "learning_rate": 0.0001468577353004072,
      "loss": 0.8523,
      "step": 102460
    },
    {
      "epoch": 1.799013325374392,
      "grad_norm": 0.04414846024966255,
      "learning_rate": 0.000146847858978403,
      "loss": 0.8456,
      "step": 102470
    },
    {
      "epoch": 1.7991888902543935,
      "grad_norm": 0.06991075043830469,
      "learning_rate": 0.00014683798207549764,
      "loss": 0.8392,
      "step": 102480
    },
    {
      "epoch": 1.7993644551343948,
      "grad_norm": 0.0519246356328407,
      "learning_rate": 0.00014682810459181627,
      "loss": 0.8544,
      "step": 102490
    },
    {
      "epoch": 1.7995400200143963,
      "grad_norm": 0.06411567447937468,
      "learning_rate": 0.00014681822652748403,
      "loss": 0.8509,
      "step": 102500
    },
    {
      "epoch": 1.7997155848943978,
      "grad_norm": 0.06775988924371534,
      "learning_rate": 0.00014680834788262606,
      "loss": 0.8368,
      "step": 102510
    },
    {
      "epoch": 1.7998911497743992,
      "grad_norm": 0.06566946344125431,
      "learning_rate": 0.00014679846865736757,
      "loss": 0.8467,
      "step": 102520
    },
    {
      "epoch": 1.8000667146544005,
      "grad_norm": 0.05488017583106747,
      "learning_rate": 0.00014678858885183369,
      "loss": 0.8436,
      "step": 102530
    },
    {
      "epoch": 1.800242279534402,
      "grad_norm": 0.059157672065591305,
      "learning_rate": 0.00014677870846614962,
      "loss": 0.8458,
      "step": 102540
    },
    {
      "epoch": 1.8004178444144032,
      "grad_norm": 0.06247354606468738,
      "learning_rate": 0.0001467688275004405,
      "loss": 0.8569,
      "step": 102550
    },
    {
      "epoch": 1.8005934092944047,
      "grad_norm": 0.14427974213454114,
      "learning_rate": 0.00014675894595483164,
      "loss": 0.8502,
      "step": 102560
    },
    {
      "epoch": 1.8007689741744062,
      "grad_norm": 0.05669505632045852,
      "learning_rate": 0.00014674906382944813,
      "loss": 0.8511,
      "step": 102570
    },
    {
      "epoch": 1.8009445390544077,
      "grad_norm": 0.054482124949102026,
      "learning_rate": 0.0001467391811244152,
      "loss": 0.8443,
      "step": 102580
    },
    {
      "epoch": 1.801120103934409,
      "grad_norm": 0.047412894677076635,
      "learning_rate": 0.0001467292978398581,
      "loss": 0.8481,
      "step": 102590
    },
    {
      "epoch": 1.8012956688144104,
      "grad_norm": 0.06692624383561993,
      "learning_rate": 0.00014671941397590202,
      "loss": 0.8507,
      "step": 102600
    },
    {
      "epoch": 1.8014712336944116,
      "grad_norm": 0.05676479853088661,
      "learning_rate": 0.0001467095295326722,
      "loss": 0.8469,
      "step": 102610
    },
    {
      "epoch": 1.8016467985744131,
      "grad_norm": 0.06421243798284011,
      "learning_rate": 0.00014669964451029388,
      "loss": 0.8424,
      "step": 102620
    },
    {
      "epoch": 1.8018223634544146,
      "grad_norm": 0.050277080055847745,
      "learning_rate": 0.0001466897589088923,
      "loss": 0.8488,
      "step": 102630
    },
    {
      "epoch": 1.801997928334416,
      "grad_norm": 0.043581039997295005,
      "learning_rate": 0.00014667987272859275,
      "loss": 0.845,
      "step": 102640
    },
    {
      "epoch": 1.8021734932144173,
      "grad_norm": 0.06576789366523701,
      "learning_rate": 0.00014666998596952043,
      "loss": 0.8503,
      "step": 102650
    },
    {
      "epoch": 1.8023490580944188,
      "grad_norm": 0.05689923058792311,
      "learning_rate": 0.00014666009863180063,
      "loss": 0.8399,
      "step": 102660
    },
    {
      "epoch": 1.80252462297442,
      "grad_norm": 0.04834007031991282,
      "learning_rate": 0.00014665021071555865,
      "loss": 0.8497,
      "step": 102670
    },
    {
      "epoch": 1.8027001878544215,
      "grad_norm": 0.0601930436144198,
      "learning_rate": 0.00014664032222091972,
      "loss": 0.8502,
      "step": 102680
    },
    {
      "epoch": 1.802875752734423,
      "grad_norm": 0.05294405376999718,
      "learning_rate": 0.0001466304331480092,
      "loss": 0.8399,
      "step": 102690
    },
    {
      "epoch": 1.8030513176144245,
      "grad_norm": 0.07116214471597206,
      "learning_rate": 0.0001466205434969523,
      "loss": 0.8462,
      "step": 102700
    },
    {
      "epoch": 1.803226882494426,
      "grad_norm": 0.07211411370570404,
      "learning_rate": 0.00014661065326787445,
      "loss": 0.8469,
      "step": 102710
    },
    {
      "epoch": 1.8034024473744272,
      "grad_norm": 0.06452046660009543,
      "learning_rate": 0.0001466007624609008,
      "loss": 0.8512,
      "step": 102720
    },
    {
      "epoch": 1.8035780122544285,
      "grad_norm": 0.05218353548141175,
      "learning_rate": 0.00014659087107615678,
      "loss": 0.85,
      "step": 102730
    },
    {
      "epoch": 1.80375357713443,
      "grad_norm": 0.07336222857370284,
      "learning_rate": 0.0001465809791137677,
      "loss": 0.8504,
      "step": 102740
    },
    {
      "epoch": 1.8039291420144314,
      "grad_norm": 0.05059859873379582,
      "learning_rate": 0.00014657108657385888,
      "loss": 0.8504,
      "step": 102750
    },
    {
      "epoch": 1.804104706894433,
      "grad_norm": 0.05640139239242346,
      "learning_rate": 0.00014656119345655565,
      "loss": 0.852,
      "step": 102760
    },
    {
      "epoch": 1.8042802717744344,
      "grad_norm": 0.08111423002215667,
      "learning_rate": 0.00014655129976198336,
      "loss": 0.8428,
      "step": 102770
    },
    {
      "epoch": 1.8044558366544357,
      "grad_norm": 0.06986590586677832,
      "learning_rate": 0.00014654140549026737,
      "loss": 0.8529,
      "step": 102780
    },
    {
      "epoch": 1.804631401534437,
      "grad_norm": 0.051300561800302416,
      "learning_rate": 0.00014653151064153308,
      "loss": 0.8397,
      "step": 102790
    },
    {
      "epoch": 1.8048069664144384,
      "grad_norm": 0.059081122853188686,
      "learning_rate": 0.00014652161521590577,
      "loss": 0.8508,
      "step": 102800
    },
    {
      "epoch": 1.8049825312944399,
      "grad_norm": 0.05965470119356715,
      "learning_rate": 0.0001465117192135109,
      "loss": 0.842,
      "step": 102810
    },
    {
      "epoch": 1.8051580961744413,
      "grad_norm": 0.06652117929138983,
      "learning_rate": 0.0001465018226344738,
      "loss": 0.8488,
      "step": 102820
    },
    {
      "epoch": 1.8053336610544428,
      "grad_norm": 0.06043724070397839,
      "learning_rate": 0.00014649192547891993,
      "loss": 0.8454,
      "step": 102830
    },
    {
      "epoch": 1.805509225934444,
      "grad_norm": 0.06515304983465933,
      "learning_rate": 0.0001464820277469746,
      "loss": 0.842,
      "step": 102840
    },
    {
      "epoch": 1.8056847908144453,
      "grad_norm": 0.05515946603659303,
      "learning_rate": 0.00014647212943876329,
      "loss": 0.8512,
      "step": 102850
    },
    {
      "epoch": 1.8058603556944468,
      "grad_norm": 0.05125553243979122,
      "learning_rate": 0.00014646223055441143,
      "loss": 0.85,
      "step": 102860
    },
    {
      "epoch": 1.8060359205744483,
      "grad_norm": 0.07257174827967791,
      "learning_rate": 0.0001464523310940443,
      "loss": 0.851,
      "step": 102870
    },
    {
      "epoch": 1.8062114854544498,
      "grad_norm": 0.050301528848681804,
      "learning_rate": 0.00014644243105778746,
      "loss": 0.8526,
      "step": 102880
    },
    {
      "epoch": 1.8063870503344512,
      "grad_norm": 0.07874986404392209,
      "learning_rate": 0.00014643253044576633,
      "loss": 0.8467,
      "step": 102890
    },
    {
      "epoch": 1.8065626152144525,
      "grad_norm": 0.08173423673495543,
      "learning_rate": 0.00014642262925810632,
      "loss": 0.8407,
      "step": 102900
    },
    {
      "epoch": 1.8067381800944537,
      "grad_norm": 0.048612812189853055,
      "learning_rate": 0.0001464127274949329,
      "loss": 0.8343,
      "step": 102910
    },
    {
      "epoch": 1.8069137449744552,
      "grad_norm": 0.06180629965574668,
      "learning_rate": 0.00014640282515637152,
      "loss": 0.8472,
      "step": 102920
    },
    {
      "epoch": 1.8070893098544567,
      "grad_norm": 0.052224699014824134,
      "learning_rate": 0.00014639292224254765,
      "loss": 0.8427,
      "step": 102930
    },
    {
      "epoch": 1.8072648747344582,
      "grad_norm": 0.051427798024711154,
      "learning_rate": 0.00014638301875358672,
      "loss": 0.8473,
      "step": 102940
    },
    {
      "epoch": 1.8074404396144597,
      "grad_norm": 0.05434897718518453,
      "learning_rate": 0.00014637311468961427,
      "loss": 0.8423,
      "step": 102950
    },
    {
      "epoch": 1.807616004494461,
      "grad_norm": 0.10078962652361632,
      "learning_rate": 0.00014636321005075578,
      "loss": 0.8488,
      "step": 102960
    },
    {
      "epoch": 1.8077915693744622,
      "grad_norm": 0.056048542543374734,
      "learning_rate": 0.0001463533048371367,
      "loss": 0.8436,
      "step": 102970
    },
    {
      "epoch": 1.8079671342544636,
      "grad_norm": 0.07119010917231683,
      "learning_rate": 0.00014634339904888257,
      "loss": 0.845,
      "step": 102980
    },
    {
      "epoch": 1.8081426991344651,
      "grad_norm": 0.07040376552692931,
      "learning_rate": 0.00014633349268611885,
      "loss": 0.8405,
      "step": 102990
    },
    {
      "epoch": 1.8083182640144666,
      "grad_norm": 0.05576101393203609,
      "learning_rate": 0.00014632358574897112,
      "loss": 0.8482,
      "step": 103000
    },
    {
      "epoch": 1.808493828894468,
      "grad_norm": 0.04913832419290911,
      "learning_rate": 0.00014631367823756487,
      "loss": 0.845,
      "step": 103010
    },
    {
      "epoch": 1.8086693937744693,
      "grad_norm": 0.04826688961638336,
      "learning_rate": 0.00014630377015202562,
      "loss": 0.844,
      "step": 103020
    },
    {
      "epoch": 1.8088449586544708,
      "grad_norm": 0.08105926824640003,
      "learning_rate": 0.00014629386149247895,
      "loss": 0.8504,
      "step": 103030
    },
    {
      "epoch": 1.809020523534472,
      "grad_norm": 0.060588607808499784,
      "learning_rate": 0.00014628395225905035,
      "loss": 0.8412,
      "step": 103040
    },
    {
      "epoch": 1.8091960884144735,
      "grad_norm": 0.04641771206057309,
      "learning_rate": 0.0001462740424518654,
      "loss": 0.8477,
      "step": 103050
    },
    {
      "epoch": 1.809371653294475,
      "grad_norm": 0.050158882885466625,
      "learning_rate": 0.00014626413207104963,
      "loss": 0.8569,
      "step": 103060
    },
    {
      "epoch": 1.8095472181744765,
      "grad_norm": 0.05612182872990823,
      "learning_rate": 0.00014625422111672867,
      "loss": 0.8467,
      "step": 103070
    },
    {
      "epoch": 1.8097227830544778,
      "grad_norm": 0.06289540795904029,
      "learning_rate": 0.00014624430958902807,
      "loss": 0.8496,
      "step": 103080
    },
    {
      "epoch": 1.8098983479344792,
      "grad_norm": 0.06079955219843222,
      "learning_rate": 0.00014623439748807335,
      "loss": 0.8418,
      "step": 103090
    },
    {
      "epoch": 1.8100739128144805,
      "grad_norm": 0.06268025854893741,
      "learning_rate": 0.00014622448481399017,
      "loss": 0.8358,
      "step": 103100
    },
    {
      "epoch": 1.810249477694482,
      "grad_norm": 0.05474224260917414,
      "learning_rate": 0.00014621457156690408,
      "loss": 0.8399,
      "step": 103110
    },
    {
      "epoch": 1.8104250425744834,
      "grad_norm": 0.05288035806388404,
      "learning_rate": 0.00014620465774694072,
      "loss": 0.8456,
      "step": 103120
    },
    {
      "epoch": 1.810600607454485,
      "grad_norm": 0.054428753205340207,
      "learning_rate": 0.00014619474335422568,
      "loss": 0.842,
      "step": 103130
    },
    {
      "epoch": 1.8107761723344862,
      "grad_norm": 0.05635648795165411,
      "learning_rate": 0.00014618482838888458,
      "loss": 0.8525,
      "step": 103140
    },
    {
      "epoch": 1.8109517372144877,
      "grad_norm": 0.05623448223007993,
      "learning_rate": 0.00014617491285104305,
      "loss": 0.8585,
      "step": 103150
    },
    {
      "epoch": 1.811127302094489,
      "grad_norm": 0.055697183491576714,
      "learning_rate": 0.00014616499674082668,
      "loss": 0.8498,
      "step": 103160
    },
    {
      "epoch": 1.8113028669744904,
      "grad_norm": 0.06817063349173819,
      "learning_rate": 0.0001461550800583612,
      "loss": 0.8416,
      "step": 103170
    },
    {
      "epoch": 1.8114784318544919,
      "grad_norm": 0.07173965895085183,
      "learning_rate": 0.00014614516280377218,
      "loss": 0.8542,
      "step": 103180
    },
    {
      "epoch": 1.8116539967344933,
      "grad_norm": 0.060798855373964955,
      "learning_rate": 0.0001461352449771853,
      "loss": 0.8447,
      "step": 103190
    },
    {
      "epoch": 1.8118295616144946,
      "grad_norm": 0.06041484436336685,
      "learning_rate": 0.0001461253265787262,
      "loss": 0.8448,
      "step": 103200
    },
    {
      "epoch": 1.812005126494496,
      "grad_norm": 0.07348035062438414,
      "learning_rate": 0.00014611540760852055,
      "loss": 0.8485,
      "step": 103210
    },
    {
      "epoch": 1.8121806913744973,
      "grad_norm": 0.045164826863984514,
      "learning_rate": 0.00014610548806669407,
      "loss": 0.8547,
      "step": 103220
    },
    {
      "epoch": 1.8123562562544988,
      "grad_norm": 0.04596292703991063,
      "learning_rate": 0.00014609556795337243,
      "loss": 0.8402,
      "step": 103230
    },
    {
      "epoch": 1.8125318211345003,
      "grad_norm": 0.0638509902013268,
      "learning_rate": 0.00014608564726868126,
      "loss": 0.8413,
      "step": 103240
    },
    {
      "epoch": 1.8127073860145018,
      "grad_norm": 0.05772036069088936,
      "learning_rate": 0.00014607572601274633,
      "loss": 0.8502,
      "step": 103250
    },
    {
      "epoch": 1.812882950894503,
      "grad_norm": 0.06361603793744569,
      "learning_rate": 0.00014606580418569326,
      "loss": 0.8395,
      "step": 103260
    },
    {
      "epoch": 1.8130585157745045,
      "grad_norm": 0.07136299763125004,
      "learning_rate": 0.0001460558817876479,
      "loss": 0.8457,
      "step": 103270
    },
    {
      "epoch": 1.8132340806545058,
      "grad_norm": 0.052942732908552184,
      "learning_rate": 0.0001460459588187358,
      "loss": 0.8516,
      "step": 103280
    },
    {
      "epoch": 1.8134096455345072,
      "grad_norm": 0.07404416061992061,
      "learning_rate": 0.00014603603527908282,
      "loss": 0.8463,
      "step": 103290
    },
    {
      "epoch": 1.8135852104145087,
      "grad_norm": 0.0482229821194097,
      "learning_rate": 0.0001460261111688146,
      "loss": 0.8491,
      "step": 103300
    },
    {
      "epoch": 1.8137607752945102,
      "grad_norm": 0.048241560696166697,
      "learning_rate": 0.00014601618648805696,
      "loss": 0.857,
      "step": 103310
    },
    {
      "epoch": 1.8139363401745114,
      "grad_norm": 0.04609429241029781,
      "learning_rate": 0.0001460062612369356,
      "loss": 0.8467,
      "step": 103320
    },
    {
      "epoch": 1.814111905054513,
      "grad_norm": 0.08982020736651947,
      "learning_rate": 0.00014599633541557627,
      "loss": 0.8424,
      "step": 103330
    },
    {
      "epoch": 1.8142874699345142,
      "grad_norm": 0.07795169605573436,
      "learning_rate": 0.00014598640902410476,
      "loss": 0.8477,
      "step": 103340
    },
    {
      "epoch": 1.8144630348145157,
      "grad_norm": 0.0543265724755221,
      "learning_rate": 0.00014597648206264688,
      "loss": 0.8487,
      "step": 103350
    },
    {
      "epoch": 1.8146385996945171,
      "grad_norm": 0.06255385112448957,
      "learning_rate": 0.00014596655453132825,
      "loss": 0.8452,
      "step": 103360
    },
    {
      "epoch": 1.8148141645745186,
      "grad_norm": 0.06035038773929551,
      "learning_rate": 0.00014595662643027482,
      "loss": 0.8443,
      "step": 103370
    },
    {
      "epoch": 1.81498972945452,
      "grad_norm": 0.09771322139211523,
      "learning_rate": 0.00014594669775961228,
      "loss": 0.8437,
      "step": 103380
    },
    {
      "epoch": 1.8151652943345213,
      "grad_norm": 0.06570542889063487,
      "learning_rate": 0.00014593676851946648,
      "loss": 0.8438,
      "step": 103390
    },
    {
      "epoch": 1.8153408592145226,
      "grad_norm": 0.06373262775142734,
      "learning_rate": 0.00014592683870996321,
      "loss": 0.8515,
      "step": 103400
    },
    {
      "epoch": 1.815516424094524,
      "grad_norm": 0.09405414117872206,
      "learning_rate": 0.0001459169083312283,
      "loss": 0.8559,
      "step": 103410
    },
    {
      "epoch": 1.8156919889745256,
      "grad_norm": 0.0535352041598483,
      "learning_rate": 0.00014590697738338756,
      "loss": 0.8451,
      "step": 103420
    },
    {
      "epoch": 1.815867553854527,
      "grad_norm": 0.06907554168511587,
      "learning_rate": 0.00014589704586656679,
      "loss": 0.8488,
      "step": 103430
    },
    {
      "epoch": 1.8160431187345285,
      "grad_norm": 0.04574795761773199,
      "learning_rate": 0.00014588711378089183,
      "loss": 0.8448,
      "step": 103440
    },
    {
      "epoch": 1.8162186836145298,
      "grad_norm": 0.0783453909586768,
      "learning_rate": 0.00014587718112648855,
      "loss": 0.8467,
      "step": 103450
    },
    {
      "epoch": 1.816394248494531,
      "grad_norm": 0.050707831457041955,
      "learning_rate": 0.0001458672479034828,
      "loss": 0.8544,
      "step": 103460
    },
    {
      "epoch": 1.8165698133745325,
      "grad_norm": 0.07770511811979791,
      "learning_rate": 0.0001458573141120004,
      "loss": 0.8431,
      "step": 103470
    },
    {
      "epoch": 1.816745378254534,
      "grad_norm": 0.07697485645267162,
      "learning_rate": 0.00014584737975216724,
      "loss": 0.8449,
      "step": 103480
    },
    {
      "epoch": 1.8169209431345354,
      "grad_norm": 0.06805441738519118,
      "learning_rate": 0.00014583744482410922,
      "loss": 0.8475,
      "step": 103490
    },
    {
      "epoch": 1.817096508014537,
      "grad_norm": 0.061114649497454675,
      "learning_rate": 0.00014582750932795214,
      "loss": 0.839,
      "step": 103500
    },
    {
      "epoch": 1.8172720728945382,
      "grad_norm": 0.046417769343516796,
      "learning_rate": 0.0001458175732638219,
      "loss": 0.8454,
      "step": 103510
    },
    {
      "epoch": 1.8174476377745394,
      "grad_norm": 0.059762433345334445,
      "learning_rate": 0.00014580763663184446,
      "loss": 0.8445,
      "step": 103520
    },
    {
      "epoch": 1.817623202654541,
      "grad_norm": 0.0996982836794326,
      "learning_rate": 0.00014579769943214566,
      "loss": 0.8454,
      "step": 103530
    },
    {
      "epoch": 1.8177987675345424,
      "grad_norm": 0.06967210442291118,
      "learning_rate": 0.00014578776166485145,
      "loss": 0.8426,
      "step": 103540
    },
    {
      "epoch": 1.8179743324145439,
      "grad_norm": 0.062291572659844786,
      "learning_rate": 0.00014577782333008768,
      "loss": 0.8496,
      "step": 103550
    },
    {
      "epoch": 1.8181498972945453,
      "grad_norm": 0.05732654100637381,
      "learning_rate": 0.00014576788442798033,
      "loss": 0.8474,
      "step": 103560
    },
    {
      "epoch": 1.8183254621745466,
      "grad_norm": 0.046065874302085856,
      "learning_rate": 0.0001457579449586553,
      "loss": 0.8456,
      "step": 103570
    },
    {
      "epoch": 1.8185010270545479,
      "grad_norm": 0.056433662657315746,
      "learning_rate": 0.0001457480049222385,
      "loss": 0.8494,
      "step": 103580
    },
    {
      "epoch": 1.8186765919345493,
      "grad_norm": 0.05667465641432079,
      "learning_rate": 0.00014573806431885595,
      "loss": 0.8475,
      "step": 103590
    },
    {
      "epoch": 1.8188521568145508,
      "grad_norm": 0.048114032338695344,
      "learning_rate": 0.00014572812314863351,
      "loss": 0.85,
      "step": 103600
    },
    {
      "epoch": 1.8190277216945523,
      "grad_norm": 0.05609438972161714,
      "learning_rate": 0.00014571818141169722,
      "loss": 0.8465,
      "step": 103610
    },
    {
      "epoch": 1.8192032865745538,
      "grad_norm": 0.05325899060752258,
      "learning_rate": 0.00014570823910817297,
      "loss": 0.8536,
      "step": 103620
    },
    {
      "epoch": 1.819378851454555,
      "grad_norm": 0.1062252917676987,
      "learning_rate": 0.00014569829623818677,
      "loss": 0.8458,
      "step": 103630
    },
    {
      "epoch": 1.8195544163345563,
      "grad_norm": 0.044830391611341364,
      "learning_rate": 0.00014568835280186456,
      "loss": 0.8446,
      "step": 103640
    },
    {
      "epoch": 1.8197299812145578,
      "grad_norm": 0.08624428908472713,
      "learning_rate": 0.0001456784087993324,
      "loss": 0.8394,
      "step": 103650
    },
    {
      "epoch": 1.8199055460945592,
      "grad_norm": 0.06985088289492071,
      "learning_rate": 0.00014566846423071623,
      "loss": 0.8506,
      "step": 103660
    },
    {
      "epoch": 1.8200811109745607,
      "grad_norm": 0.04816300984936778,
      "learning_rate": 0.00014565851909614203,
      "loss": 0.8498,
      "step": 103670
    },
    {
      "epoch": 1.8202566758545622,
      "grad_norm": 0.05384698487434245,
      "learning_rate": 0.00014564857339573588,
      "loss": 0.8449,
      "step": 103680
    },
    {
      "epoch": 1.8204322407345634,
      "grad_norm": 0.0574895200156774,
      "learning_rate": 0.00014563862712962372,
      "loss": 0.8445,
      "step": 103690
    },
    {
      "epoch": 1.8206078056145647,
      "grad_norm": 0.06803156302934985,
      "learning_rate": 0.0001456286802979316,
      "loss": 0.8436,
      "step": 103700
    },
    {
      "epoch": 1.8207833704945662,
      "grad_norm": 0.06248569976459023,
      "learning_rate": 0.00014561873290078555,
      "loss": 0.8463,
      "step": 103710
    },
    {
      "epoch": 1.8209589353745677,
      "grad_norm": 0.07293376816098275,
      "learning_rate": 0.0001456087849383116,
      "loss": 0.8508,
      "step": 103720
    },
    {
      "epoch": 1.8211345002545691,
      "grad_norm": 0.04967432585243372,
      "learning_rate": 0.0001455988364106358,
      "loss": 0.8474,
      "step": 103730
    },
    {
      "epoch": 1.8213100651345706,
      "grad_norm": 0.05308784942116989,
      "learning_rate": 0.00014558888731788417,
      "loss": 0.8403,
      "step": 103740
    },
    {
      "epoch": 1.8214856300145719,
      "grad_norm": 0.04926429358967063,
      "learning_rate": 0.00014557893766018284,
      "loss": 0.8498,
      "step": 103750
    },
    {
      "epoch": 1.8216611948945733,
      "grad_norm": 0.05706302677269536,
      "learning_rate": 0.00014556898743765782,
      "loss": 0.8514,
      "step": 103760
    },
    {
      "epoch": 1.8218367597745746,
      "grad_norm": 0.07560068378554091,
      "learning_rate": 0.00014555903665043515,
      "loss": 0.8518,
      "step": 103770
    },
    {
      "epoch": 1.822012324654576,
      "grad_norm": 0.051747842584976604,
      "learning_rate": 0.00014554908529864098,
      "loss": 0.845,
      "step": 103780
    },
    {
      "epoch": 1.8221878895345776,
      "grad_norm": 0.06610801474062032,
      "learning_rate": 0.00014553913338240132,
      "loss": 0.8482,
      "step": 103790
    },
    {
      "epoch": 1.822363454414579,
      "grad_norm": 0.1007452586187125,
      "learning_rate": 0.00014552918090184234,
      "loss": 0.8486,
      "step": 103800
    },
    {
      "epoch": 1.8225390192945803,
      "grad_norm": 0.048416038547672006,
      "learning_rate": 0.00014551922785709012,
      "loss": 0.8465,
      "step": 103810
    },
    {
      "epoch": 1.8227145841745818,
      "grad_norm": 0.060798949682430846,
      "learning_rate": 0.00014550927424827073,
      "loss": 0.8498,
      "step": 103820
    },
    {
      "epoch": 1.822890149054583,
      "grad_norm": 0.06032790417445101,
      "learning_rate": 0.00014549932007551028,
      "loss": 0.8525,
      "step": 103830
    },
    {
      "epoch": 1.8230657139345845,
      "grad_norm": 0.060918439787335535,
      "learning_rate": 0.00014548936533893494,
      "loss": 0.8521,
      "step": 103840
    },
    {
      "epoch": 1.823241278814586,
      "grad_norm": 0.07031275462114858,
      "learning_rate": 0.0001454794100386708,
      "loss": 0.8557,
      "step": 103850
    },
    {
      "epoch": 1.8234168436945875,
      "grad_norm": 0.05412675602387381,
      "learning_rate": 0.00014546945417484404,
      "loss": 0.8485,
      "step": 103860
    },
    {
      "epoch": 1.8235924085745887,
      "grad_norm": 0.0629298751025017,
      "learning_rate": 0.00014545949774758077,
      "loss": 0.8544,
      "step": 103870
    },
    {
      "epoch": 1.8237679734545902,
      "grad_norm": 0.05737261387869819,
      "learning_rate": 0.00014544954075700712,
      "loss": 0.8511,
      "step": 103880
    },
    {
      "epoch": 1.8239435383345914,
      "grad_norm": 0.08761124953938763,
      "learning_rate": 0.00014543958320324929,
      "loss": 0.842,
      "step": 103890
    },
    {
      "epoch": 1.824119103214593,
      "grad_norm": 0.05443669725867993,
      "learning_rate": 0.0001454296250864334,
      "loss": 0.8528,
      "step": 103900
    },
    {
      "epoch": 1.8242946680945944,
      "grad_norm": 0.049672424946825354,
      "learning_rate": 0.00014541966640668565,
      "loss": 0.8457,
      "step": 103910
    },
    {
      "epoch": 1.8244702329745959,
      "grad_norm": 0.06543816745771697,
      "learning_rate": 0.00014540970716413223,
      "loss": 0.8404,
      "step": 103920
    },
    {
      "epoch": 1.8246457978545971,
      "grad_norm": 0.08685339768181459,
      "learning_rate": 0.0001453997473588993,
      "loss": 0.8482,
      "step": 103930
    },
    {
      "epoch": 1.8248213627345986,
      "grad_norm": 0.07795145854049687,
      "learning_rate": 0.00014538978699111307,
      "loss": 0.8493,
      "step": 103940
    },
    {
      "epoch": 1.8249969276145999,
      "grad_norm": 0.052798915744215524,
      "learning_rate": 0.00014537982606089973,
      "loss": 0.8591,
      "step": 103950
    },
    {
      "epoch": 1.8251724924946013,
      "grad_norm": 0.06390329998294687,
      "learning_rate": 0.00014536986456838545,
      "loss": 0.8518,
      "step": 103960
    },
    {
      "epoch": 1.8253480573746028,
      "grad_norm": 0.06305254683220209,
      "learning_rate": 0.00014535990251369652,
      "loss": 0.8447,
      "step": 103970
    },
    {
      "epoch": 1.8255236222546043,
      "grad_norm": 0.07192618989329394,
      "learning_rate": 0.00014534993989695914,
      "loss": 0.8442,
      "step": 103980
    },
    {
      "epoch": 1.8256991871346056,
      "grad_norm": 0.06620298961029075,
      "learning_rate": 0.0001453399767182995,
      "loss": 0.8454,
      "step": 103990
    },
    {
      "epoch": 1.825874752014607,
      "grad_norm": 0.0677387431315837,
      "learning_rate": 0.00014533001297784384,
      "loss": 0.8506,
      "step": 104000
    },
    {
      "epoch": 1.8260503168946083,
      "grad_norm": 0.052340678303456185,
      "learning_rate": 0.00014532004867571846,
      "loss": 0.8471,
      "step": 104010
    },
    {
      "epoch": 1.8262258817746098,
      "grad_norm": 0.05311824483018857,
      "learning_rate": 0.00014531008381204953,
      "loss": 0.846,
      "step": 104020
    },
    {
      "epoch": 1.8264014466546112,
      "grad_norm": 0.07318437678902039,
      "learning_rate": 0.00014530011838696338,
      "loss": 0.861,
      "step": 104030
    },
    {
      "epoch": 1.8265770115346127,
      "grad_norm": 0.06047287150884961,
      "learning_rate": 0.00014529015240058622,
      "loss": 0.8486,
      "step": 104040
    },
    {
      "epoch": 1.826752576414614,
      "grad_norm": 0.04954628516358834,
      "learning_rate": 0.00014528018585304434,
      "loss": 0.8466,
      "step": 104050
    },
    {
      "epoch": 1.8269281412946154,
      "grad_norm": 0.053673465785127515,
      "learning_rate": 0.000145270218744464,
      "loss": 0.8495,
      "step": 104060
    },
    {
      "epoch": 1.8271037061746167,
      "grad_norm": 0.06757328523726161,
      "learning_rate": 0.00014526025107497155,
      "loss": 0.8437,
      "step": 104070
    },
    {
      "epoch": 1.8272792710546182,
      "grad_norm": 0.06034596865910602,
      "learning_rate": 0.0001452502828446932,
      "loss": 0.8464,
      "step": 104080
    },
    {
      "epoch": 1.8274548359346197,
      "grad_norm": 0.0716904074468678,
      "learning_rate": 0.00014524031405375532,
      "loss": 0.8565,
      "step": 104090
    },
    {
      "epoch": 1.8276304008146211,
      "grad_norm": 0.07205661444400599,
      "learning_rate": 0.00014523034470228415,
      "loss": 0.8476,
      "step": 104100
    },
    {
      "epoch": 1.8278059656946226,
      "grad_norm": 0.07102809453948053,
      "learning_rate": 0.00014522037479040604,
      "loss": 0.8468,
      "step": 104110
    },
    {
      "epoch": 1.8279815305746239,
      "grad_norm": 0.059026014660531645,
      "learning_rate": 0.00014521040431824734,
      "loss": 0.8538,
      "step": 104120
    },
    {
      "epoch": 1.8281570954546251,
      "grad_norm": 0.056868598603061614,
      "learning_rate": 0.0001452004332859343,
      "loss": 0.8495,
      "step": 104130
    },
    {
      "epoch": 1.8283326603346266,
      "grad_norm": 0.06350616035524086,
      "learning_rate": 0.0001451904616935933,
      "loss": 0.8401,
      "step": 104140
    },
    {
      "epoch": 1.828508225214628,
      "grad_norm": 0.050935435255591306,
      "learning_rate": 0.00014518048954135072,
      "loss": 0.8477,
      "step": 104150
    },
    {
      "epoch": 1.8286837900946296,
      "grad_norm": 0.0846137130155374,
      "learning_rate": 0.00014517051682933288,
      "loss": 0.8416,
      "step": 104160
    },
    {
      "epoch": 1.828859354974631,
      "grad_norm": 0.05492541274331719,
      "learning_rate": 0.00014516054355766608,
      "loss": 0.8485,
      "step": 104170
    },
    {
      "epoch": 1.8290349198546323,
      "grad_norm": 0.057451721713236314,
      "learning_rate": 0.00014515056972647675,
      "loss": 0.849,
      "step": 104180
    },
    {
      "epoch": 1.8292104847346335,
      "grad_norm": 0.054695019296509134,
      "learning_rate": 0.00014514059533589124,
      "loss": 0.8458,
      "step": 104190
    },
    {
      "epoch": 1.829386049614635,
      "grad_norm": 0.06226453329089998,
      "learning_rate": 0.00014513062038603593,
      "loss": 0.8388,
      "step": 104200
    },
    {
      "epoch": 1.8295616144946365,
      "grad_norm": 0.05851732027763638,
      "learning_rate": 0.00014512064487703715,
      "loss": 0.853,
      "step": 104210
    },
    {
      "epoch": 1.829737179374638,
      "grad_norm": 0.0557548377784978,
      "learning_rate": 0.00014511066880902143,
      "loss": 0.8446,
      "step": 104220
    },
    {
      "epoch": 1.8299127442546395,
      "grad_norm": 0.0625415624747608,
      "learning_rate": 0.00014510069218211504,
      "loss": 0.8486,
      "step": 104230
    },
    {
      "epoch": 1.8300883091346407,
      "grad_norm": 0.0608376327166691,
      "learning_rate": 0.0001450907149964444,
      "loss": 0.8458,
      "step": 104240
    },
    {
      "epoch": 1.830263874014642,
      "grad_norm": 0.053990028882840616,
      "learning_rate": 0.00014508073725213598,
      "loss": 0.8491,
      "step": 104250
    },
    {
      "epoch": 1.8304394388946434,
      "grad_norm": 0.07280071513517536,
      "learning_rate": 0.0001450707589493162,
      "loss": 0.8605,
      "step": 104260
    },
    {
      "epoch": 1.830615003774645,
      "grad_norm": 0.0527541627035653,
      "learning_rate": 0.00014506078008811143,
      "loss": 0.8462,
      "step": 104270
    },
    {
      "epoch": 1.8307905686546464,
      "grad_norm": 0.046081626570998485,
      "learning_rate": 0.0001450508006686481,
      "loss": 0.8481,
      "step": 104280
    },
    {
      "epoch": 1.8309661335346479,
      "grad_norm": 0.0521331325165148,
      "learning_rate": 0.00014504082069105275,
      "loss": 0.8423,
      "step": 104290
    },
    {
      "epoch": 1.8311416984146491,
      "grad_norm": 0.059447253967566956,
      "learning_rate": 0.00014503084015545173,
      "loss": 0.8521,
      "step": 104300
    },
    {
      "epoch": 1.8313172632946504,
      "grad_norm": 0.04759922706850495,
      "learning_rate": 0.0001450208590619715,
      "loss": 0.8509,
      "step": 104310
    },
    {
      "epoch": 1.8314928281746519,
      "grad_norm": 0.06496509959826567,
      "learning_rate": 0.0001450108774107386,
      "loss": 0.8393,
      "step": 104320
    },
    {
      "epoch": 1.8316683930546533,
      "grad_norm": 0.06927111942250183,
      "learning_rate": 0.0001450008952018794,
      "loss": 0.8596,
      "step": 104330
    },
    {
      "epoch": 1.8318439579346548,
      "grad_norm": 0.040314147060411615,
      "learning_rate": 0.00014499091243552046,
      "loss": 0.8522,
      "step": 104340
    },
    {
      "epoch": 1.8320195228146563,
      "grad_norm": 0.06897843693927244,
      "learning_rate": 0.0001449809291117882,
      "loss": 0.8428,
      "step": 104350
    },
    {
      "epoch": 1.8321950876946576,
      "grad_norm": 0.052562630634805416,
      "learning_rate": 0.00014497094523080917,
      "loss": 0.8488,
      "step": 104360
    },
    {
      "epoch": 1.8323706525746588,
      "grad_norm": 0.05628542132656844,
      "learning_rate": 0.00014496096079270982,
      "loss": 0.8429,
      "step": 104370
    },
    {
      "epoch": 1.8325462174546603,
      "grad_norm": 0.04074547051931621,
      "learning_rate": 0.00014495097579761667,
      "loss": 0.8454,
      "step": 104380
    },
    {
      "epoch": 1.8327217823346618,
      "grad_norm": 0.06279640709000898,
      "learning_rate": 0.00014494099024565623,
      "loss": 0.843,
      "step": 104390
    },
    {
      "epoch": 1.8328973472146632,
      "grad_norm": 0.05854862808552525,
      "learning_rate": 0.00014493100413695502,
      "loss": 0.8393,
      "step": 104400
    },
    {
      "epoch": 1.8330729120946647,
      "grad_norm": 0.06004371835786594,
      "learning_rate": 0.00014492101747163956,
      "loss": 0.8517,
      "step": 104410
    },
    {
      "epoch": 1.833248476974666,
      "grad_norm": 0.05485700263495913,
      "learning_rate": 0.00014491103024983644,
      "loss": 0.8505,
      "step": 104420
    },
    {
      "epoch": 1.8334240418546672,
      "grad_norm": 0.07989898774971158,
      "learning_rate": 0.00014490104247167214,
      "loss": 0.8422,
      "step": 104430
    },
    {
      "epoch": 1.8335996067346687,
      "grad_norm": 0.0698563321365467,
      "learning_rate": 0.0001448910541372732,
      "loss": 0.8428,
      "step": 104440
    },
    {
      "epoch": 1.8337751716146702,
      "grad_norm": 0.043560890355799166,
      "learning_rate": 0.00014488106524676616,
      "loss": 0.846,
      "step": 104450
    },
    {
      "epoch": 1.8339507364946717,
      "grad_norm": 0.06311623539340162,
      "learning_rate": 0.00014487107580027765,
      "loss": 0.8441,
      "step": 104460
    },
    {
      "epoch": 1.8341263013746731,
      "grad_norm": 0.08966435225073092,
      "learning_rate": 0.0001448610857979342,
      "loss": 0.8419,
      "step": 104470
    },
    {
      "epoch": 1.8343018662546744,
      "grad_norm": 0.05564065564271927,
      "learning_rate": 0.00014485109523986237,
      "loss": 0.8449,
      "step": 104480
    },
    {
      "epoch": 1.8344774311346759,
      "grad_norm": 0.04541515744824186,
      "learning_rate": 0.00014484110412618878,
      "loss": 0.8398,
      "step": 104490
    },
    {
      "epoch": 1.8346529960146771,
      "grad_norm": 0.08655378863361059,
      "learning_rate": 0.00014483111245704,
      "loss": 0.8408,
      "step": 104500
    },
    {
      "epoch": 1.8348285608946786,
      "grad_norm": 0.06549743382682997,
      "learning_rate": 0.00014482112023254262,
      "loss": 0.8453,
      "step": 104510
    },
    {
      "epoch": 1.83500412577468,
      "grad_norm": 0.0455265368866188,
      "learning_rate": 0.00014481112745282323,
      "loss": 0.8534,
      "step": 104520
    },
    {
      "epoch": 1.8351796906546816,
      "grad_norm": 0.06164994946019855,
      "learning_rate": 0.00014480113411800849,
      "loss": 0.8476,
      "step": 104530
    },
    {
      "epoch": 1.8353552555346828,
      "grad_norm": 0.05145123138243053,
      "learning_rate": 0.00014479114022822496,
      "loss": 0.8447,
      "step": 104540
    },
    {
      "epoch": 1.8355308204146843,
      "grad_norm": 0.07988407001286241,
      "learning_rate": 0.0001447811457835993,
      "loss": 0.8497,
      "step": 104550
    },
    {
      "epoch": 1.8357063852946855,
      "grad_norm": 0.050550514712438946,
      "learning_rate": 0.00014477115078425813,
      "loss": 0.8466,
      "step": 104560
    },
    {
      "epoch": 1.835881950174687,
      "grad_norm": 0.04905350095380034,
      "learning_rate": 0.00014476115523032808,
      "loss": 0.8464,
      "step": 104570
    },
    {
      "epoch": 1.8360575150546885,
      "grad_norm": 0.06477172367640323,
      "learning_rate": 0.00014475115912193583,
      "loss": 0.8448,
      "step": 104580
    },
    {
      "epoch": 1.83623307993469,
      "grad_norm": 0.05561516936538364,
      "learning_rate": 0.00014474116245920797,
      "loss": 0.8548,
      "step": 104590
    },
    {
      "epoch": 1.8364086448146912,
      "grad_norm": 0.05257069576767229,
      "learning_rate": 0.00014473116524227125,
      "loss": 0.8488,
      "step": 104600
    },
    {
      "epoch": 1.8365842096946927,
      "grad_norm": 0.0743718011767674,
      "learning_rate": 0.00014472116747125228,
      "loss": 0.8477,
      "step": 104610
    },
    {
      "epoch": 1.836759774574694,
      "grad_norm": 0.06879615834479477,
      "learning_rate": 0.0001447111691462777,
      "loss": 0.848,
      "step": 104620
    },
    {
      "epoch": 1.8369353394546954,
      "grad_norm": 0.08145576560373398,
      "learning_rate": 0.00014470117026747426,
      "loss": 0.8393,
      "step": 104630
    },
    {
      "epoch": 1.837110904334697,
      "grad_norm": 0.06896568366437907,
      "learning_rate": 0.0001446911708349686,
      "loss": 0.8502,
      "step": 104640
    },
    {
      "epoch": 1.8372864692146984,
      "grad_norm": 0.06373185742560591,
      "learning_rate": 0.00014468117084888747,
      "loss": 0.8415,
      "step": 104650
    },
    {
      "epoch": 1.8374620340946997,
      "grad_norm": 0.04782265542222571,
      "learning_rate": 0.0001446711703093575,
      "loss": 0.8463,
      "step": 104660
    },
    {
      "epoch": 1.8376375989747011,
      "grad_norm": 0.04434110808190677,
      "learning_rate": 0.00014466116921650542,
      "loss": 0.8529,
      "step": 104670
    },
    {
      "epoch": 1.8378131638547024,
      "grad_norm": 0.06840731925541023,
      "learning_rate": 0.00014465116757045798,
      "loss": 0.8424,
      "step": 104680
    },
    {
      "epoch": 1.8379887287347039,
      "grad_norm": 0.05480411771743443,
      "learning_rate": 0.00014464116537134188,
      "loss": 0.8513,
      "step": 104690
    },
    {
      "epoch": 1.8381642936147053,
      "grad_norm": 0.05461782728309248,
      "learning_rate": 0.00014463116261928383,
      "loss": 0.8448,
      "step": 104700
    },
    {
      "epoch": 1.8383398584947068,
      "grad_norm": 0.0472545170022371,
      "learning_rate": 0.00014462115931441062,
      "loss": 0.8484,
      "step": 104710
    },
    {
      "epoch": 1.838515423374708,
      "grad_norm": 0.07138042002633492,
      "learning_rate": 0.00014461115545684895,
      "loss": 0.8467,
      "step": 104720
    },
    {
      "epoch": 1.8386909882547096,
      "grad_norm": 0.05438005598449055,
      "learning_rate": 0.0001446011510467256,
      "loss": 0.8453,
      "step": 104730
    },
    {
      "epoch": 1.8388665531347108,
      "grad_norm": 0.05781638094644989,
      "learning_rate": 0.00014459114608416724,
      "loss": 0.8418,
      "step": 104740
    },
    {
      "epoch": 1.8390421180147123,
      "grad_norm": 0.0630991080054205,
      "learning_rate": 0.0001445811405693008,
      "loss": 0.853,
      "step": 104750
    },
    {
      "epoch": 1.8392176828947138,
      "grad_norm": 0.06936775904450057,
      "learning_rate": 0.0001445711345022529,
      "loss": 0.8471,
      "step": 104760
    },
    {
      "epoch": 1.8393932477747152,
      "grad_norm": 0.05002906764594328,
      "learning_rate": 0.00014456112788315039,
      "loss": 0.8478,
      "step": 104770
    },
    {
      "epoch": 1.8395688126547165,
      "grad_norm": 0.060422605626508044,
      "learning_rate": 0.00014455112071212002,
      "loss": 0.8519,
      "step": 104780
    },
    {
      "epoch": 1.839744377534718,
      "grad_norm": 0.0544593676342436,
      "learning_rate": 0.00014454111298928862,
      "loss": 0.8442,
      "step": 104790
    },
    {
      "epoch": 1.8399199424147192,
      "grad_norm": 0.06879640382305899,
      "learning_rate": 0.000144531104714783,
      "loss": 0.849,
      "step": 104800
    },
    {
      "epoch": 1.8400955072947207,
      "grad_norm": 0.05698487199157572,
      "learning_rate": 0.0001445210958887299,
      "loss": 0.8478,
      "step": 104810
    },
    {
      "epoch": 1.8402710721747222,
      "grad_norm": 0.04301142574596063,
      "learning_rate": 0.00014451108651125617,
      "loss": 0.851,
      "step": 104820
    },
    {
      "epoch": 1.8404466370547237,
      "grad_norm": 0.06264058971448595,
      "learning_rate": 0.00014450107658248867,
      "loss": 0.8462,
      "step": 104830
    },
    {
      "epoch": 1.8406222019347251,
      "grad_norm": 0.059121319490239424,
      "learning_rate": 0.00014449106610255419,
      "loss": 0.8417,
      "step": 104840
    },
    {
      "epoch": 1.8407977668147264,
      "grad_norm": 0.05161203675654514,
      "learning_rate": 0.00014448105507157955,
      "loss": 0.8437,
      "step": 104850
    },
    {
      "epoch": 1.8409733316947277,
      "grad_norm": 0.07734212081986895,
      "learning_rate": 0.0001444710434896916,
      "loss": 0.8473,
      "step": 104860
    },
    {
      "epoch": 1.8411488965747291,
      "grad_norm": 0.052306153187998874,
      "learning_rate": 0.00014446103135701724,
      "loss": 0.8511,
      "step": 104870
    },
    {
      "epoch": 1.8413244614547306,
      "grad_norm": 0.05556103771278905,
      "learning_rate": 0.00014445101867368324,
      "loss": 0.8452,
      "step": 104880
    },
    {
      "epoch": 1.841500026334732,
      "grad_norm": 0.06644153510186514,
      "learning_rate": 0.00014444100543981652,
      "loss": 0.8486,
      "step": 104890
    },
    {
      "epoch": 1.8416755912147336,
      "grad_norm": 0.06229308302956273,
      "learning_rate": 0.00014443099165554395,
      "loss": 0.8441,
      "step": 104900
    },
    {
      "epoch": 1.8418511560947348,
      "grad_norm": 0.0538453567973435,
      "learning_rate": 0.00014442097732099237,
      "loss": 0.8469,
      "step": 104910
    },
    {
      "epoch": 1.842026720974736,
      "grad_norm": 0.04979485007305208,
      "learning_rate": 0.0001444109624362887,
      "loss": 0.8483,
      "step": 104920
    },
    {
      "epoch": 1.8422022858547376,
      "grad_norm": 0.05609681774493647,
      "learning_rate": 0.0001444009470015598,
      "loss": 0.8416,
      "step": 104930
    },
    {
      "epoch": 1.842377850734739,
      "grad_norm": 0.0677096049023443,
      "learning_rate": 0.0001443909310169326,
      "loss": 0.8473,
      "step": 104940
    },
    {
      "epoch": 1.8425534156147405,
      "grad_norm": 0.050771945494740324,
      "learning_rate": 0.00014438091448253402,
      "loss": 0.8487,
      "step": 104950
    },
    {
      "epoch": 1.842728980494742,
      "grad_norm": 0.08077315440810526,
      "learning_rate": 0.0001443708973984909,
      "loss": 0.8449,
      "step": 104960
    },
    {
      "epoch": 1.8429045453747432,
      "grad_norm": 0.04788276840885666,
      "learning_rate": 0.00014436087976493018,
      "loss": 0.8551,
      "step": 104970
    },
    {
      "epoch": 1.8430801102547445,
      "grad_norm": 0.04611985249668424,
      "learning_rate": 0.00014435086158197882,
      "loss": 0.8478,
      "step": 104980
    },
    {
      "epoch": 1.843255675134746,
      "grad_norm": 0.05695919540248361,
      "learning_rate": 0.00014434084284976374,
      "loss": 0.8487,
      "step": 104990
    },
    {
      "epoch": 1.8434312400147475,
      "grad_norm": 0.06440783717999957,
      "learning_rate": 0.00014433082356841187,
      "loss": 0.8429,
      "step": 105000
    },
    {
      "epoch": 1.843606804894749,
      "grad_norm": 0.05995257118819511,
      "learning_rate": 0.00014432080373805018,
      "loss": 0.8522,
      "step": 105010
    },
    {
      "epoch": 1.8437823697747504,
      "grad_norm": 0.052830448904488826,
      "learning_rate": 0.0001443107833588056,
      "loss": 0.8489,
      "step": 105020
    },
    {
      "epoch": 1.8439579346547517,
      "grad_norm": 0.05382948905359375,
      "learning_rate": 0.00014430076243080505,
      "loss": 0.8446,
      "step": 105030
    },
    {
      "epoch": 1.844133499534753,
      "grad_norm": 0.06459877621160837,
      "learning_rate": 0.00014429074095417558,
      "loss": 0.8498,
      "step": 105040
    },
    {
      "epoch": 1.8443090644147544,
      "grad_norm": 0.06390736210079573,
      "learning_rate": 0.00014428071892904414,
      "loss": 0.8488,
      "step": 105050
    },
    {
      "epoch": 1.8444846292947559,
      "grad_norm": 0.050275395075164236,
      "learning_rate": 0.00014427069635553769,
      "loss": 0.8523,
      "step": 105060
    },
    {
      "epoch": 1.8446601941747574,
      "grad_norm": 0.06282258932396985,
      "learning_rate": 0.0001442606732337832,
      "loss": 0.8477,
      "step": 105070
    },
    {
      "epoch": 1.8448357590547588,
      "grad_norm": 0.06978338871855795,
      "learning_rate": 0.0001442506495639077,
      "loss": 0.8521,
      "step": 105080
    },
    {
      "epoch": 1.84501132393476,
      "grad_norm": 0.06574508133677207,
      "learning_rate": 0.0001442406253460382,
      "loss": 0.8409,
      "step": 105090
    },
    {
      "epoch": 1.8451868888147613,
      "grad_norm": 0.042234097029942186,
      "learning_rate": 0.00014423060058030166,
      "loss": 0.846,
      "step": 105100
    },
    {
      "epoch": 1.8453624536947628,
      "grad_norm": 0.05805944144678811,
      "learning_rate": 0.00014422057526682514,
      "loss": 0.8523,
      "step": 105110
    },
    {
      "epoch": 1.8455380185747643,
      "grad_norm": 0.09689149220574289,
      "learning_rate": 0.00014421054940573567,
      "loss": 0.8479,
      "step": 105120
    },
    {
      "epoch": 1.8457135834547658,
      "grad_norm": 0.04878881701861149,
      "learning_rate": 0.0001442005229971602,
      "loss": 0.8481,
      "step": 105130
    },
    {
      "epoch": 1.8458891483347672,
      "grad_norm": 0.050600700989645045,
      "learning_rate": 0.00014419049604122588,
      "loss": 0.8496,
      "step": 105140
    },
    {
      "epoch": 1.8460647132147685,
      "grad_norm": 0.05528096176878486,
      "learning_rate": 0.00014418046853805966,
      "loss": 0.8453,
      "step": 105150
    },
    {
      "epoch": 1.8462402780947698,
      "grad_norm": 0.05634514658153842,
      "learning_rate": 0.00014417044048778866,
      "loss": 0.8452,
      "step": 105160
    },
    {
      "epoch": 1.8464158429747712,
      "grad_norm": 0.05574479742509345,
      "learning_rate": 0.00014416041189053988,
      "loss": 0.8454,
      "step": 105170
    },
    {
      "epoch": 1.8465914078547727,
      "grad_norm": 0.06019476224043916,
      "learning_rate": 0.00014415038274644042,
      "loss": 0.8436,
      "step": 105180
    },
    {
      "epoch": 1.8467669727347742,
      "grad_norm": 0.048108837661600144,
      "learning_rate": 0.00014414035305561733,
      "loss": 0.8505,
      "step": 105190
    },
    {
      "epoch": 1.8469425376147757,
      "grad_norm": 0.06348523513364851,
      "learning_rate": 0.00014413032281819773,
      "loss": 0.8435,
      "step": 105200
    },
    {
      "epoch": 1.847118102494777,
      "grad_norm": 0.07684988397168108,
      "learning_rate": 0.00014412029203430865,
      "loss": 0.8505,
      "step": 105210
    },
    {
      "epoch": 1.8472936673747784,
      "grad_norm": 0.06609883496574573,
      "learning_rate": 0.0001441102607040772,
      "loss": 0.8499,
      "step": 105220
    },
    {
      "epoch": 1.8474692322547797,
      "grad_norm": 0.06879414803741012,
      "learning_rate": 0.0001441002288276305,
      "loss": 0.8529,
      "step": 105230
    },
    {
      "epoch": 1.8476447971347811,
      "grad_norm": 0.04570523598852523,
      "learning_rate": 0.00014409019640509562,
      "loss": 0.8521,
      "step": 105240
    },
    {
      "epoch": 1.8478203620147826,
      "grad_norm": 0.04965360089226083,
      "learning_rate": 0.00014408016343659972,
      "loss": 0.8494,
      "step": 105250
    },
    {
      "epoch": 1.847995926894784,
      "grad_norm": 0.06003013477885608,
      "learning_rate": 0.0001440701299222699,
      "loss": 0.8414,
      "step": 105260
    },
    {
      "epoch": 1.8481714917747853,
      "grad_norm": 0.056641403449427856,
      "learning_rate": 0.00014406009586223324,
      "loss": 0.8439,
      "step": 105270
    },
    {
      "epoch": 1.8483470566547868,
      "grad_norm": 0.07828698446618373,
      "learning_rate": 0.00014405006125661692,
      "loss": 0.8372,
      "step": 105280
    },
    {
      "epoch": 1.848522621534788,
      "grad_norm": 0.05972810326846312,
      "learning_rate": 0.00014404002610554813,
      "loss": 0.8508,
      "step": 105290
    },
    {
      "epoch": 1.8486981864147896,
      "grad_norm": 0.069053087963169,
      "learning_rate": 0.00014402999040915389,
      "loss": 0.8461,
      "step": 105300
    },
    {
      "epoch": 1.848873751294791,
      "grad_norm": 0.07462298935890625,
      "learning_rate": 0.00014401995416756145,
      "loss": 0.8558,
      "step": 105310
    },
    {
      "epoch": 1.8490493161747925,
      "grad_norm": 0.056263524130788445,
      "learning_rate": 0.00014400991738089795,
      "loss": 0.8407,
      "step": 105320
    },
    {
      "epoch": 1.8492248810547938,
      "grad_norm": 0.05431571271889149,
      "learning_rate": 0.00014399988004929056,
      "loss": 0.8395,
      "step": 105330
    },
    {
      "epoch": 1.8494004459347952,
      "grad_norm": 0.051791898240005387,
      "learning_rate": 0.00014398984217286643,
      "loss": 0.8431,
      "step": 105340
    },
    {
      "epoch": 1.8495760108147965,
      "grad_norm": 0.05308405441913459,
      "learning_rate": 0.0001439798037517528,
      "loss": 0.8467,
      "step": 105350
    },
    {
      "epoch": 1.849751575694798,
      "grad_norm": 0.07238559933530972,
      "learning_rate": 0.00014396976478607678,
      "loss": 0.8503,
      "step": 105360
    },
    {
      "epoch": 1.8499271405747995,
      "grad_norm": 0.08060978660065447,
      "learning_rate": 0.00014395972527596562,
      "loss": 0.8444,
      "step": 105370
    },
    {
      "epoch": 1.850102705454801,
      "grad_norm": 0.06380114733084927,
      "learning_rate": 0.0001439496852215465,
      "loss": 0.846,
      "step": 105380
    },
    {
      "epoch": 1.8502782703348022,
      "grad_norm": 0.05562760747515547,
      "learning_rate": 0.00014393964462294665,
      "loss": 0.8534,
      "step": 105390
    },
    {
      "epoch": 1.8504538352148037,
      "grad_norm": 0.0485590948949104,
      "learning_rate": 0.00014392960348029324,
      "loss": 0.8454,
      "step": 105400
    },
    {
      "epoch": 1.850629400094805,
      "grad_norm": 0.04578927672877621,
      "learning_rate": 0.00014391956179371355,
      "loss": 0.8509,
      "step": 105410
    },
    {
      "epoch": 1.8508049649748064,
      "grad_norm": 0.047992567544259276,
      "learning_rate": 0.0001439095195633348,
      "loss": 0.8512,
      "step": 105420
    },
    {
      "epoch": 1.8509805298548079,
      "grad_norm": 0.0612133038027644,
      "learning_rate": 0.0001438994767892842,
      "loss": 0.8421,
      "step": 105430
    },
    {
      "epoch": 1.8511560947348094,
      "grad_norm": 0.04918445908685571,
      "learning_rate": 0.000143889433471689,
      "loss": 0.8475,
      "step": 105440
    },
    {
      "epoch": 1.8513316596148106,
      "grad_norm": 0.08897320873537727,
      "learning_rate": 0.00014387938961067648,
      "loss": 0.8421,
      "step": 105450
    },
    {
      "epoch": 1.851507224494812,
      "grad_norm": 0.057743006353822394,
      "learning_rate": 0.00014386934520637385,
      "loss": 0.8511,
      "step": 105460
    },
    {
      "epoch": 1.8516827893748133,
      "grad_norm": 0.059395532796334666,
      "learning_rate": 0.0001438593002589084,
      "loss": 0.8515,
      "step": 105470
    },
    {
      "epoch": 1.8518583542548148,
      "grad_norm": 0.06841509792693741,
      "learning_rate": 0.00014384925476840745,
      "loss": 0.8475,
      "step": 105480
    },
    {
      "epoch": 1.8520339191348163,
      "grad_norm": 0.05009044907997425,
      "learning_rate": 0.0001438392087349982,
      "loss": 0.8406,
      "step": 105490
    },
    {
      "epoch": 1.8522094840148178,
      "grad_norm": 0.049482034222281625,
      "learning_rate": 0.00014382916215880794,
      "loss": 0.844,
      "step": 105500
    },
    {
      "epoch": 1.852385048894819,
      "grad_norm": 0.06792225582727836,
      "learning_rate": 0.00014381911503996405,
      "loss": 0.8451,
      "step": 105510
    },
    {
      "epoch": 1.8525606137748205,
      "grad_norm": 0.06994152798722306,
      "learning_rate": 0.00014380906737859368,
      "loss": 0.8463,
      "step": 105520
    },
    {
      "epoch": 1.8527361786548218,
      "grad_norm": 0.06055368410006964,
      "learning_rate": 0.00014379901917482428,
      "loss": 0.8419,
      "step": 105530
    },
    {
      "epoch": 1.8529117435348232,
      "grad_norm": 0.05733804173170315,
      "learning_rate": 0.00014378897042878307,
      "loss": 0.8458,
      "step": 105540
    },
    {
      "epoch": 1.8530873084148247,
      "grad_norm": 0.058000389828197545,
      "learning_rate": 0.00014377892114059742,
      "loss": 0.8478,
      "step": 105550
    },
    {
      "epoch": 1.8532628732948262,
      "grad_norm": 0.05025373137179316,
      "learning_rate": 0.00014376887131039467,
      "loss": 0.8509,
      "step": 105560
    },
    {
      "epoch": 1.8534384381748277,
      "grad_norm": 0.06326491809135787,
      "learning_rate": 0.00014375882093830208,
      "loss": 0.8455,
      "step": 105570
    },
    {
      "epoch": 1.853614003054829,
      "grad_norm": 0.04326927491664561,
      "learning_rate": 0.00014374877002444708,
      "loss": 0.8499,
      "step": 105580
    },
    {
      "epoch": 1.8537895679348302,
      "grad_norm": 0.06149130752774975,
      "learning_rate": 0.00014373871856895694,
      "loss": 0.8561,
      "step": 105590
    },
    {
      "epoch": 1.8539651328148317,
      "grad_norm": 0.04931187010374151,
      "learning_rate": 0.00014372866657195904,
      "loss": 0.842,
      "step": 105600
    },
    {
      "epoch": 1.8541406976948331,
      "grad_norm": 0.0529182904041342,
      "learning_rate": 0.00014371861403358074,
      "loss": 0.853,
      "step": 105610
    },
    {
      "epoch": 1.8543162625748346,
      "grad_norm": 0.06553724259871956,
      "learning_rate": 0.00014370856095394946,
      "loss": 0.8408,
      "step": 105620
    },
    {
      "epoch": 1.854491827454836,
      "grad_norm": 0.07176489522510181,
      "learning_rate": 0.00014369850733319248,
      "loss": 0.8476,
      "step": 105630
    },
    {
      "epoch": 1.8546673923348374,
      "grad_norm": 0.06385213511496166,
      "learning_rate": 0.00014368845317143725,
      "loss": 0.8442,
      "step": 105640
    },
    {
      "epoch": 1.8548429572148386,
      "grad_norm": 0.07032120088265846,
      "learning_rate": 0.00014367839846881112,
      "loss": 0.8544,
      "step": 105650
    },
    {
      "epoch": 1.85501852209484,
      "grad_norm": 0.06458903577006424,
      "learning_rate": 0.0001436683432254415,
      "loss": 0.8514,
      "step": 105660
    },
    {
      "epoch": 1.8551940869748416,
      "grad_norm": 0.047885956073720044,
      "learning_rate": 0.00014365828744145582,
      "loss": 0.8364,
      "step": 105670
    },
    {
      "epoch": 1.855369651854843,
      "grad_norm": 0.056563015813032226,
      "learning_rate": 0.00014364823111698147,
      "loss": 0.8459,
      "step": 105680
    },
    {
      "epoch": 1.8555452167348445,
      "grad_norm": 0.07881521268803729,
      "learning_rate": 0.00014363817425214583,
      "loss": 0.8472,
      "step": 105690
    },
    {
      "epoch": 1.8557207816148458,
      "grad_norm": 0.0504751637851391,
      "learning_rate": 0.00014362811684707637,
      "loss": 0.8493,
      "step": 105700
    },
    {
      "epoch": 1.855896346494847,
      "grad_norm": 0.0628142103849103,
      "learning_rate": 0.00014361805890190045,
      "loss": 0.8416,
      "step": 105710
    },
    {
      "epoch": 1.8560719113748485,
      "grad_norm": 0.04520123292801607,
      "learning_rate": 0.00014360800041674562,
      "loss": 0.8418,
      "step": 105720
    },
    {
      "epoch": 1.85624747625485,
      "grad_norm": 0.06115456241514542,
      "learning_rate": 0.00014359794139173922,
      "loss": 0.8436,
      "step": 105730
    },
    {
      "epoch": 1.8564230411348515,
      "grad_norm": 0.05502703561893415,
      "learning_rate": 0.00014358788182700877,
      "loss": 0.8478,
      "step": 105740
    },
    {
      "epoch": 1.856598606014853,
      "grad_norm": 0.0668340131696173,
      "learning_rate": 0.0001435778217226817,
      "loss": 0.8493,
      "step": 105750
    },
    {
      "epoch": 1.8567741708948542,
      "grad_norm": 0.05952188609868521,
      "learning_rate": 0.00014356776107888542,
      "loss": 0.8508,
      "step": 105760
    },
    {
      "epoch": 1.8569497357748554,
      "grad_norm": 0.07717501901512717,
      "learning_rate": 0.00014355769989574747,
      "loss": 0.8404,
      "step": 105770
    },
    {
      "epoch": 1.857125300654857,
      "grad_norm": 0.08427664902579136,
      "learning_rate": 0.00014354763817339529,
      "loss": 0.8469,
      "step": 105780
    },
    {
      "epoch": 1.8573008655348584,
      "grad_norm": 0.05908911822536438,
      "learning_rate": 0.00014353757591195642,
      "loss": 0.8503,
      "step": 105790
    },
    {
      "epoch": 1.8574764304148599,
      "grad_norm": 0.06498241932218926,
      "learning_rate": 0.0001435275131115583,
      "loss": 0.8498,
      "step": 105800
    },
    {
      "epoch": 1.8576519952948614,
      "grad_norm": 0.05198884428792769,
      "learning_rate": 0.00014351744977232842,
      "loss": 0.8418,
      "step": 105810
    },
    {
      "epoch": 1.8578275601748626,
      "grad_norm": 0.05188555309641177,
      "learning_rate": 0.0001435073858943943,
      "loss": 0.8492,
      "step": 105820
    },
    {
      "epoch": 1.8580031250548639,
      "grad_norm": 0.0709295247386001,
      "learning_rate": 0.00014349732147788348,
      "loss": 0.8488,
      "step": 105830
    },
    {
      "epoch": 1.8581786899348653,
      "grad_norm": 0.06260994986469656,
      "learning_rate": 0.00014348725652292346,
      "loss": 0.8428,
      "step": 105840
    },
    {
      "epoch": 1.8583542548148668,
      "grad_norm": 0.05559356657050467,
      "learning_rate": 0.00014347719102964174,
      "loss": 0.8482,
      "step": 105850
    },
    {
      "epoch": 1.8585298196948683,
      "grad_norm": 0.05518184823135107,
      "learning_rate": 0.00014346712499816586,
      "loss": 0.8489,
      "step": 105860
    },
    {
      "epoch": 1.8587053845748698,
      "grad_norm": 0.05041999233228258,
      "learning_rate": 0.0001434570584286234,
      "loss": 0.8461,
      "step": 105870
    },
    {
      "epoch": 1.858880949454871,
      "grad_norm": 0.04860185562823752,
      "learning_rate": 0.0001434469913211418,
      "loss": 0.8573,
      "step": 105880
    },
    {
      "epoch": 1.8590565143348723,
      "grad_norm": 0.045906976520861995,
      "learning_rate": 0.00014343692367584877,
      "loss": 0.838,
      "step": 105890
    },
    {
      "epoch": 1.8592320792148738,
      "grad_norm": 0.04867132825814745,
      "learning_rate": 0.00014342685549287174,
      "loss": 0.8462,
      "step": 105900
    },
    {
      "epoch": 1.8594076440948752,
      "grad_norm": 0.05346777791073834,
      "learning_rate": 0.00014341678677233833,
      "loss": 0.845,
      "step": 105910
    },
    {
      "epoch": 1.8595832089748767,
      "grad_norm": 0.10865122499614244,
      "learning_rate": 0.00014340671751437612,
      "loss": 0.8534,
      "step": 105920
    },
    {
      "epoch": 1.8597587738548782,
      "grad_norm": 0.04812895058936261,
      "learning_rate": 0.00014339664771911262,
      "loss": 0.8435,
      "step": 105930
    },
    {
      "epoch": 1.8599343387348795,
      "grad_norm": 0.06073722792840988,
      "learning_rate": 0.0001433865773866755,
      "loss": 0.8442,
      "step": 105940
    },
    {
      "epoch": 1.860109903614881,
      "grad_norm": 0.048152760263412656,
      "learning_rate": 0.00014337650651719233,
      "loss": 0.8455,
      "step": 105950
    },
    {
      "epoch": 1.8602854684948822,
      "grad_norm": 0.08461173132057581,
      "learning_rate": 0.00014336643511079068,
      "loss": 0.8483,
      "step": 105960
    },
    {
      "epoch": 1.8604610333748837,
      "grad_norm": 0.055151699199706194,
      "learning_rate": 0.0001433563631675982,
      "loss": 0.8494,
      "step": 105970
    },
    {
      "epoch": 1.8606365982548851,
      "grad_norm": 0.06852641796368732,
      "learning_rate": 0.00014334629068774246,
      "loss": 0.8464,
      "step": 105980
    },
    {
      "epoch": 1.8608121631348866,
      "grad_norm": 0.05039410347012162,
      "learning_rate": 0.00014333621767135112,
      "loss": 0.8429,
      "step": 105990
    },
    {
      "epoch": 1.8609877280148879,
      "grad_norm": 0.05270753421936715,
      "learning_rate": 0.00014332614411855177,
      "loss": 0.8505,
      "step": 106000
    },
    {
      "epoch": 1.8611632928948894,
      "grad_norm": 0.05007712127983891,
      "learning_rate": 0.0001433160700294721,
      "loss": 0.8451,
      "step": 106010
    },
    {
      "epoch": 1.8613388577748906,
      "grad_norm": 0.05474355403358141,
      "learning_rate": 0.00014330599540423965,
      "loss": 0.8432,
      "step": 106020
    },
    {
      "epoch": 1.861514422654892,
      "grad_norm": 0.06424179647035486,
      "learning_rate": 0.00014329592024298215,
      "loss": 0.841,
      "step": 106030
    },
    {
      "epoch": 1.8616899875348936,
      "grad_norm": 0.054897714499368845,
      "learning_rate": 0.00014328584454582728,
      "loss": 0.8443,
      "step": 106040
    },
    {
      "epoch": 1.861865552414895,
      "grad_norm": 0.08094435461837184,
      "learning_rate": 0.0001432757683129026,
      "loss": 0.85,
      "step": 106050
    },
    {
      "epoch": 1.8620411172948963,
      "grad_norm": 0.0489076394863472,
      "learning_rate": 0.00014326569154433588,
      "loss": 0.852,
      "step": 106060
    },
    {
      "epoch": 1.8622166821748978,
      "grad_norm": 0.055851275170473204,
      "learning_rate": 0.00014325561424025473,
      "loss": 0.8461,
      "step": 106070
    },
    {
      "epoch": 1.862392247054899,
      "grad_norm": 0.055846653203605476,
      "learning_rate": 0.00014324553640078684,
      "loss": 0.8413,
      "step": 106080
    },
    {
      "epoch": 1.8625678119349005,
      "grad_norm": 0.05795555425517053,
      "learning_rate": 0.00014323545802605994,
      "loss": 0.8422,
      "step": 106090
    },
    {
      "epoch": 1.862743376814902,
      "grad_norm": 0.053848750324656774,
      "learning_rate": 0.00014322537911620165,
      "loss": 0.8419,
      "step": 106100
    },
    {
      "epoch": 1.8629189416949035,
      "grad_norm": 0.06082115786942185,
      "learning_rate": 0.00014321529967133972,
      "loss": 0.8414,
      "step": 106110
    },
    {
      "epoch": 1.8630945065749047,
      "grad_norm": 0.04536819587977698,
      "learning_rate": 0.00014320521969160186,
      "loss": 0.8511,
      "step": 106120
    },
    {
      "epoch": 1.8632700714549062,
      "grad_norm": 0.05804081068897011,
      "learning_rate": 0.00014319513917711578,
      "loss": 0.8475,
      "step": 106130
    },
    {
      "epoch": 1.8634456363349075,
      "grad_norm": 0.059863783339413786,
      "learning_rate": 0.0001431850581280092,
      "loss": 0.8438,
      "step": 106140
    },
    {
      "epoch": 1.863621201214909,
      "grad_norm": 0.05850482433688861,
      "learning_rate": 0.00014317497654440985,
      "loss": 0.8435,
      "step": 106150
    },
    {
      "epoch": 1.8637967660949104,
      "grad_norm": 0.08743121459884803,
      "learning_rate": 0.00014316489442644547,
      "loss": 0.8487,
      "step": 106160
    },
    {
      "epoch": 1.8639723309749119,
      "grad_norm": 0.05456994854308207,
      "learning_rate": 0.00014315481177424376,
      "loss": 0.8521,
      "step": 106170
    },
    {
      "epoch": 1.8641478958549131,
      "grad_norm": 0.0662911114432518,
      "learning_rate": 0.00014314472858793255,
      "loss": 0.8467,
      "step": 106180
    },
    {
      "epoch": 1.8643234607349146,
      "grad_norm": 0.06001968944784056,
      "learning_rate": 0.00014313464486763957,
      "loss": 0.8432,
      "step": 106190
    },
    {
      "epoch": 1.8644990256149159,
      "grad_norm": 0.05792799944250671,
      "learning_rate": 0.0001431245606134925,
      "loss": 0.8509,
      "step": 106200
    },
    {
      "epoch": 1.8646745904949174,
      "grad_norm": 0.05158513975245155,
      "learning_rate": 0.00014311447582561923,
      "loss": 0.8398,
      "step": 106210
    },
    {
      "epoch": 1.8648501553749188,
      "grad_norm": 0.04893682102015972,
      "learning_rate": 0.00014310439050414746,
      "loss": 0.8494,
      "step": 106220
    },
    {
      "epoch": 1.8650257202549203,
      "grad_norm": 0.052089546400028115,
      "learning_rate": 0.00014309430464920497,
      "loss": 0.8502,
      "step": 106230
    },
    {
      "epoch": 1.8652012851349216,
      "grad_norm": 0.06148195231914974,
      "learning_rate": 0.00014308421826091962,
      "loss": 0.8563,
      "step": 106240
    },
    {
      "epoch": 1.865376850014923,
      "grad_norm": 0.07035013334097538,
      "learning_rate": 0.0001430741313394191,
      "loss": 0.8446,
      "step": 106250
    },
    {
      "epoch": 1.8655524148949243,
      "grad_norm": 0.06553126762720474,
      "learning_rate": 0.0001430640438848313,
      "loss": 0.8455,
      "step": 106260
    },
    {
      "epoch": 1.8657279797749258,
      "grad_norm": 0.056809769280939004,
      "learning_rate": 0.00014305395589728402,
      "loss": 0.8485,
      "step": 106270
    },
    {
      "epoch": 1.8659035446549272,
      "grad_norm": 0.08196862181374277,
      "learning_rate": 0.00014304386737690506,
      "loss": 0.8507,
      "step": 106280
    },
    {
      "epoch": 1.8660791095349287,
      "grad_norm": 0.05724522921159814,
      "learning_rate": 0.0001430337783238222,
      "loss": 0.8445,
      "step": 106290
    },
    {
      "epoch": 1.8662546744149302,
      "grad_norm": 0.0639744094990267,
      "learning_rate": 0.00014302368873816336,
      "loss": 0.8386,
      "step": 106300
    },
    {
      "epoch": 1.8664302392949315,
      "grad_norm": 0.05048336654108011,
      "learning_rate": 0.0001430135986200563,
      "loss": 0.8478,
      "step": 106310
    },
    {
      "epoch": 1.8666058041749327,
      "grad_norm": 0.07035333767761513,
      "learning_rate": 0.00014300350796962892,
      "loss": 0.8491,
      "step": 106320
    },
    {
      "epoch": 1.8667813690549342,
      "grad_norm": 0.06906543606468618,
      "learning_rate": 0.00014299341678700904,
      "loss": 0.8493,
      "step": 106330
    },
    {
      "epoch": 1.8669569339349357,
      "grad_norm": 0.07519086704381173,
      "learning_rate": 0.00014298332507232448,
      "loss": 0.8485,
      "step": 106340
    },
    {
      "epoch": 1.8671324988149371,
      "grad_norm": 0.05427978819979762,
      "learning_rate": 0.00014297323282570318,
      "loss": 0.8526,
      "step": 106350
    },
    {
      "epoch": 1.8673080636949386,
      "grad_norm": 0.05731549885886311,
      "learning_rate": 0.00014296314004727298,
      "loss": 0.8396,
      "step": 106360
    },
    {
      "epoch": 1.8674836285749399,
      "grad_norm": 0.06107116922718381,
      "learning_rate": 0.00014295304673716172,
      "loss": 0.8412,
      "step": 106370
    },
    {
      "epoch": 1.8676591934549411,
      "grad_norm": 0.04861769694047454,
      "learning_rate": 0.00014294295289549739,
      "loss": 0.8459,
      "step": 106380
    },
    {
      "epoch": 1.8678347583349426,
      "grad_norm": 0.06375853880432544,
      "learning_rate": 0.00014293285852240773,
      "loss": 0.8483,
      "step": 106390
    },
    {
      "epoch": 1.868010323214944,
      "grad_norm": 0.05217875770225602,
      "learning_rate": 0.00014292276361802076,
      "loss": 0.841,
      "step": 106400
    },
    {
      "epoch": 1.8681858880949456,
      "grad_norm": 0.05078627976329698,
      "learning_rate": 0.0001429126681824643,
      "loss": 0.8451,
      "step": 106410
    },
    {
      "epoch": 1.868361452974947,
      "grad_norm": 0.06910009442831427,
      "learning_rate": 0.00014290257221586633,
      "loss": 0.844,
      "step": 106420
    },
    {
      "epoch": 1.8685370178549483,
      "grad_norm": 0.06905871657021134,
      "learning_rate": 0.00014289247571835475,
      "loss": 0.8464,
      "step": 106430
    },
    {
      "epoch": 1.8687125827349496,
      "grad_norm": 0.07317570828928004,
      "learning_rate": 0.00014288237869005744,
      "loss": 0.8488,
      "step": 106440
    },
    {
      "epoch": 1.868888147614951,
      "grad_norm": 0.11072606347813466,
      "learning_rate": 0.00014287228113110242,
      "loss": 0.8491,
      "step": 106450
    },
    {
      "epoch": 1.8690637124949525,
      "grad_norm": 0.06054434908992929,
      "learning_rate": 0.0001428621830416175,
      "loss": 0.8421,
      "step": 106460
    },
    {
      "epoch": 1.869239277374954,
      "grad_norm": 0.06629311997281169,
      "learning_rate": 0.00014285208442173076,
      "loss": 0.8422,
      "step": 106470
    },
    {
      "epoch": 1.8694148422549555,
      "grad_norm": 0.058903305878965063,
      "learning_rate": 0.00014284198527157008,
      "loss": 0.8549,
      "step": 106480
    },
    {
      "epoch": 1.8695904071349567,
      "grad_norm": 0.04849563544425149,
      "learning_rate": 0.0001428318855912634,
      "loss": 0.8465,
      "step": 106490
    },
    {
      "epoch": 1.869765972014958,
      "grad_norm": 0.057268247192382646,
      "learning_rate": 0.00014282178538093876,
      "loss": 0.8444,
      "step": 106500
    },
    {
      "epoch": 1.8699415368949595,
      "grad_norm": 0.05744425080670455,
      "learning_rate": 0.00014281168464072404,
      "loss": 0.8531,
      "step": 106510
    },
    {
      "epoch": 1.870117101774961,
      "grad_norm": 0.059870508191846604,
      "learning_rate": 0.00014280158337074726,
      "loss": 0.8362,
      "step": 106520
    },
    {
      "epoch": 1.8702926666549624,
      "grad_norm": 0.05860086733234563,
      "learning_rate": 0.00014279148157113645,
      "loss": 0.8469,
      "step": 106530
    },
    {
      "epoch": 1.8704682315349639,
      "grad_norm": 0.05742856494832758,
      "learning_rate": 0.00014278137924201954,
      "loss": 0.8434,
      "step": 106540
    },
    {
      "epoch": 1.8706437964149651,
      "grad_norm": 0.08241862264768168,
      "learning_rate": 0.00014277127638352453,
      "loss": 0.858,
      "step": 106550
    },
    {
      "epoch": 1.8708193612949664,
      "grad_norm": 0.05187892899933325,
      "learning_rate": 0.00014276117299577942,
      "loss": 0.8488,
      "step": 106560
    },
    {
      "epoch": 1.8709949261749679,
      "grad_norm": 0.05221914509180794,
      "learning_rate": 0.00014275106907891228,
      "loss": 0.8551,
      "step": 106570
    },
    {
      "epoch": 1.8711704910549694,
      "grad_norm": 0.06941507300408725,
      "learning_rate": 0.00014274096463305112,
      "loss": 0.8482,
      "step": 106580
    },
    {
      "epoch": 1.8713460559349708,
      "grad_norm": 0.06786575612933661,
      "learning_rate": 0.0001427308596583239,
      "loss": 0.8448,
      "step": 106590
    },
    {
      "epoch": 1.8715216208149723,
      "grad_norm": 0.03994729824373997,
      "learning_rate": 0.0001427207541548587,
      "loss": 0.852,
      "step": 106600
    },
    {
      "epoch": 1.8716971856949736,
      "grad_norm": 0.04275986311730804,
      "learning_rate": 0.00014271064812278354,
      "loss": 0.8437,
      "step": 106610
    },
    {
      "epoch": 1.8718727505749748,
      "grad_norm": 0.06467919332615514,
      "learning_rate": 0.0001427005415622265,
      "loss": 0.8492,
      "step": 106620
    },
    {
      "epoch": 1.8720483154549763,
      "grad_norm": 0.08777611258016828,
      "learning_rate": 0.0001426904344733156,
      "loss": 0.8414,
      "step": 106630
    },
    {
      "epoch": 1.8722238803349778,
      "grad_norm": 0.06770187452825405,
      "learning_rate": 0.00014268032685617888,
      "loss": 0.842,
      "step": 106640
    },
    {
      "epoch": 1.8723994452149793,
      "grad_norm": 0.12117309276788532,
      "learning_rate": 0.00014267021871094444,
      "loss": 0.8438,
      "step": 106650
    },
    {
      "epoch": 1.8725750100949807,
      "grad_norm": 0.05409034649371126,
      "learning_rate": 0.00014266011003774033,
      "loss": 0.8386,
      "step": 106660
    },
    {
      "epoch": 1.872750574974982,
      "grad_norm": 0.06100023461840248,
      "learning_rate": 0.00014265000083669467,
      "loss": 0.8481,
      "step": 106670
    },
    {
      "epoch": 1.8729261398549835,
      "grad_norm": 0.07236629086388514,
      "learning_rate": 0.00014263989110793553,
      "loss": 0.8512,
      "step": 106680
    },
    {
      "epoch": 1.8731017047349847,
      "grad_norm": 0.05009996645943287,
      "learning_rate": 0.00014262978085159095,
      "loss": 0.8489,
      "step": 106690
    },
    {
      "epoch": 1.8732772696149862,
      "grad_norm": 0.052623287048906474,
      "learning_rate": 0.0001426196700677891,
      "loss": 0.8503,
      "step": 106700
    },
    {
      "epoch": 1.8734528344949877,
      "grad_norm": 0.06621136824025649,
      "learning_rate": 0.00014260955875665802,
      "loss": 0.8499,
      "step": 106710
    },
    {
      "epoch": 1.8736283993749892,
      "grad_norm": 0.05201835732314325,
      "learning_rate": 0.00014259944691832587,
      "loss": 0.8472,
      "step": 106720
    },
    {
      "epoch": 1.8738039642549904,
      "grad_norm": 0.058640741103336884,
      "learning_rate": 0.00014258933455292075,
      "loss": 0.8449,
      "step": 106730
    },
    {
      "epoch": 1.8739795291349919,
      "grad_norm": 0.05207213865743045,
      "learning_rate": 0.0001425792216605708,
      "loss": 0.8492,
      "step": 106740
    },
    {
      "epoch": 1.8741550940149931,
      "grad_norm": 0.07606162917623273,
      "learning_rate": 0.00014256910824140414,
      "loss": 0.8476,
      "step": 106750
    },
    {
      "epoch": 1.8743306588949946,
      "grad_norm": 0.0501523854528693,
      "learning_rate": 0.00014255899429554893,
      "loss": 0.8477,
      "step": 106760
    },
    {
      "epoch": 1.874506223774996,
      "grad_norm": 0.06521310385406806,
      "learning_rate": 0.00014254887982313328,
      "loss": 0.844,
      "step": 106770
    },
    {
      "epoch": 1.8746817886549976,
      "grad_norm": 0.05710473299849508,
      "learning_rate": 0.00014253876482428536,
      "loss": 0.8368,
      "step": 106780
    },
    {
      "epoch": 1.8748573535349988,
      "grad_norm": 0.05649736123456589,
      "learning_rate": 0.00014252864929913334,
      "loss": 0.8472,
      "step": 106790
    },
    {
      "epoch": 1.8750329184150003,
      "grad_norm": 0.051546050644041955,
      "learning_rate": 0.00014251853324780535,
      "loss": 0.8429,
      "step": 106800
    },
    {
      "epoch": 1.8752084832950016,
      "grad_norm": 0.055776196820393996,
      "learning_rate": 0.00014250841667042962,
      "loss": 0.8404,
      "step": 106810
    },
    {
      "epoch": 1.875384048175003,
      "grad_norm": 0.05447227103761869,
      "learning_rate": 0.0001424982995671343,
      "loss": 0.8468,
      "step": 106820
    },
    {
      "epoch": 1.8755596130550045,
      "grad_norm": 0.06321192990996816,
      "learning_rate": 0.00014248818193804752,
      "loss": 0.843,
      "step": 106830
    },
    {
      "epoch": 1.875735177935006,
      "grad_norm": 0.051129771131792214,
      "learning_rate": 0.00014247806378329755,
      "loss": 0.8431,
      "step": 106840
    },
    {
      "epoch": 1.8759107428150072,
      "grad_norm": 0.05128504330676382,
      "learning_rate": 0.00014246794510301258,
      "loss": 0.8467,
      "step": 106850
    },
    {
      "epoch": 1.8760863076950087,
      "grad_norm": 0.04762875475581306,
      "learning_rate": 0.00014245782589732077,
      "loss": 0.8403,
      "step": 106860
    },
    {
      "epoch": 1.87626187257501,
      "grad_norm": 0.061494135503067975,
      "learning_rate": 0.00014244770616635038,
      "loss": 0.8428,
      "step": 106870
    },
    {
      "epoch": 1.8764374374550115,
      "grad_norm": 0.08505308630518457,
      "learning_rate": 0.00014243758591022963,
      "loss": 0.8449,
      "step": 106880
    },
    {
      "epoch": 1.876613002335013,
      "grad_norm": 0.055887250062141805,
      "learning_rate": 0.00014242746512908668,
      "loss": 0.8493,
      "step": 106890
    },
    {
      "epoch": 1.8767885672150144,
      "grad_norm": 0.06941012518619032,
      "learning_rate": 0.00014241734382304983,
      "loss": 0.8419,
      "step": 106900
    },
    {
      "epoch": 1.8769641320950157,
      "grad_norm": 0.05062300614049672,
      "learning_rate": 0.0001424072219922473,
      "loss": 0.8534,
      "step": 106910
    },
    {
      "epoch": 1.8771396969750171,
      "grad_norm": 0.04685103142956085,
      "learning_rate": 0.00014239709963680732,
      "loss": 0.8428,
      "step": 106920
    },
    {
      "epoch": 1.8773152618550184,
      "grad_norm": 0.056837822228893164,
      "learning_rate": 0.00014238697675685814,
      "loss": 0.8555,
      "step": 106930
    },
    {
      "epoch": 1.8774908267350199,
      "grad_norm": 0.07059317980987856,
      "learning_rate": 0.00014237685335252807,
      "loss": 0.8632,
      "step": 106940
    },
    {
      "epoch": 1.8776663916150214,
      "grad_norm": 0.05713618580226438,
      "learning_rate": 0.0001423667294239453,
      "loss": 0.8431,
      "step": 106950
    },
    {
      "epoch": 1.8778419564950228,
      "grad_norm": 0.061105083315512385,
      "learning_rate": 0.00014235660497123818,
      "loss": 0.8536,
      "step": 106960
    },
    {
      "epoch": 1.878017521375024,
      "grad_norm": 0.05931098888587571,
      "learning_rate": 0.00014234647999453493,
      "loss": 0.8523,
      "step": 106970
    },
    {
      "epoch": 1.8781930862550256,
      "grad_norm": 0.06097482247807706,
      "learning_rate": 0.00014233635449396382,
      "loss": 0.842,
      "step": 106980
    },
    {
      "epoch": 1.8783686511350268,
      "grad_norm": 0.047817585078135334,
      "learning_rate": 0.00014232622846965324,
      "loss": 0.8465,
      "step": 106990
    },
    {
      "epoch": 1.8785442160150283,
      "grad_norm": 0.04086416968896691,
      "learning_rate": 0.0001423161019217314,
      "loss": 0.8465,
      "step": 107000
    },
    {
      "epoch": 1.8787197808950298,
      "grad_norm": 0.06999391716564102,
      "learning_rate": 0.00014230597485032662,
      "loss": 0.8457,
      "step": 107010
    },
    {
      "epoch": 1.8788953457750313,
      "grad_norm": 0.06261146237952987,
      "learning_rate": 0.0001422958472555672,
      "loss": 0.8485,
      "step": 107020
    },
    {
      "epoch": 1.8790709106550327,
      "grad_norm": 0.06653264978714801,
      "learning_rate": 0.0001422857191375815,
      "loss": 0.8448,
      "step": 107030
    },
    {
      "epoch": 1.879246475535034,
      "grad_norm": 0.06419026542779141,
      "learning_rate": 0.00014227559049649787,
      "loss": 0.8491,
      "step": 107040
    },
    {
      "epoch": 1.8794220404150352,
      "grad_norm": 0.07086030047204199,
      "learning_rate": 0.00014226546133244453,
      "loss": 0.8527,
      "step": 107050
    },
    {
      "epoch": 1.8795976052950367,
      "grad_norm": 0.04884721304963423,
      "learning_rate": 0.00014225533164554993,
      "loss": 0.8412,
      "step": 107060
    },
    {
      "epoch": 1.8797731701750382,
      "grad_norm": 0.05047404509011464,
      "learning_rate": 0.00014224520143594235,
      "loss": 0.8569,
      "step": 107070
    },
    {
      "epoch": 1.8799487350550397,
      "grad_norm": 0.050915546279555954,
      "learning_rate": 0.0001422350707037502,
      "loss": 0.8449,
      "step": 107080
    },
    {
      "epoch": 1.8801242999350412,
      "grad_norm": 0.08279230265072478,
      "learning_rate": 0.00014222493944910178,
      "loss": 0.8499,
      "step": 107090
    },
    {
      "epoch": 1.8802998648150424,
      "grad_norm": 0.05979296312296291,
      "learning_rate": 0.00014221480767212546,
      "loss": 0.8421,
      "step": 107100
    },
    {
      "epoch": 1.8804754296950437,
      "grad_norm": 0.05725569416508,
      "learning_rate": 0.00014220467537294965,
      "loss": 0.851,
      "step": 107110
    },
    {
      "epoch": 1.8806509945750451,
      "grad_norm": 0.056874237072009635,
      "learning_rate": 0.00014219454255170268,
      "loss": 0.8469,
      "step": 107120
    },
    {
      "epoch": 1.8808265594550466,
      "grad_norm": 0.07656319910571784,
      "learning_rate": 0.00014218440920851297,
      "loss": 0.8541,
      "step": 107130
    },
    {
      "epoch": 1.881002124335048,
      "grad_norm": 0.06474535908425746,
      "learning_rate": 0.00014217427534350888,
      "loss": 0.8464,
      "step": 107140
    },
    {
      "epoch": 1.8811776892150496,
      "grad_norm": 0.06186571706960451,
      "learning_rate": 0.00014216414095681888,
      "loss": 0.8485,
      "step": 107150
    },
    {
      "epoch": 1.8813532540950508,
      "grad_norm": 0.07320283249497198,
      "learning_rate": 0.0001421540060485713,
      "loss": 0.848,
      "step": 107160
    },
    {
      "epoch": 1.881528818975052,
      "grad_norm": 0.0730864233251961,
      "learning_rate": 0.00014214387061889462,
      "loss": 0.8406,
      "step": 107170
    },
    {
      "epoch": 1.8817043838550536,
      "grad_norm": 0.05476893690260848,
      "learning_rate": 0.00014213373466791718,
      "loss": 0.8542,
      "step": 107180
    },
    {
      "epoch": 1.881879948735055,
      "grad_norm": 0.07025244372421353,
      "learning_rate": 0.0001421235981957674,
      "loss": 0.8567,
      "step": 107190
    },
    {
      "epoch": 1.8820555136150565,
      "grad_norm": 0.05379338310814346,
      "learning_rate": 0.0001421134612025738,
      "loss": 0.8475,
      "step": 107200
    },
    {
      "epoch": 1.882231078495058,
      "grad_norm": 0.04722433418647219,
      "learning_rate": 0.00014210332368846476,
      "loss": 0.8412,
      "step": 107210
    },
    {
      "epoch": 1.8824066433750593,
      "grad_norm": 0.05767838207549319,
      "learning_rate": 0.0001420931856535687,
      "loss": 0.8473,
      "step": 107220
    },
    {
      "epoch": 1.8825822082550605,
      "grad_norm": 0.07112488756503287,
      "learning_rate": 0.00014208304709801414,
      "loss": 0.8499,
      "step": 107230
    },
    {
      "epoch": 1.882757773135062,
      "grad_norm": 0.06744766689324587,
      "learning_rate": 0.00014207290802192946,
      "loss": 0.8468,
      "step": 107240
    },
    {
      "epoch": 1.8829333380150635,
      "grad_norm": 0.05413876776208277,
      "learning_rate": 0.0001420627684254432,
      "loss": 0.8576,
      "step": 107250
    },
    {
      "epoch": 1.883108902895065,
      "grad_norm": 0.07420114260325573,
      "learning_rate": 0.0001420526283086838,
      "loss": 0.8486,
      "step": 107260
    },
    {
      "epoch": 1.8832844677750664,
      "grad_norm": 0.04801823388160254,
      "learning_rate": 0.0001420424876717797,
      "loss": 0.8505,
      "step": 107270
    },
    {
      "epoch": 1.8834600326550677,
      "grad_norm": 0.07221812508160749,
      "learning_rate": 0.00014203234651485945,
      "loss": 0.8433,
      "step": 107280
    },
    {
      "epoch": 1.883635597535069,
      "grad_norm": 0.04684675881877558,
      "learning_rate": 0.00014202220483805144,
      "loss": 0.8436,
      "step": 107290
    },
    {
      "epoch": 1.8838111624150704,
      "grad_norm": 0.062126216312158436,
      "learning_rate": 0.0001420120626414843,
      "loss": 0.8407,
      "step": 107300
    },
    {
      "epoch": 1.8839867272950719,
      "grad_norm": 0.059683336107710615,
      "learning_rate": 0.00014200191992528644,
      "loss": 0.8437,
      "step": 107310
    },
    {
      "epoch": 1.8841622921750734,
      "grad_norm": 0.049621776165936916,
      "learning_rate": 0.0001419917766895864,
      "loss": 0.847,
      "step": 107320
    },
    {
      "epoch": 1.8843378570550748,
      "grad_norm": 0.054866762711107776,
      "learning_rate": 0.00014198163293451272,
      "loss": 0.8453,
      "step": 107330
    },
    {
      "epoch": 1.884513421935076,
      "grad_norm": 0.06883726828216678,
      "learning_rate": 0.00014197148866019387,
      "loss": 0.8547,
      "step": 107340
    },
    {
      "epoch": 1.8846889868150773,
      "grad_norm": 0.05777454013858833,
      "learning_rate": 0.00014196134386675838,
      "loss": 0.8466,
      "step": 107350
    },
    {
      "epoch": 1.8848645516950788,
      "grad_norm": 0.04714107499095379,
      "learning_rate": 0.00014195119855433484,
      "loss": 0.8469,
      "step": 107360
    },
    {
      "epoch": 1.8850401165750803,
      "grad_norm": 0.04822490221063911,
      "learning_rate": 0.00014194105272305178,
      "loss": 0.8461,
      "step": 107370
    },
    {
      "epoch": 1.8852156814550818,
      "grad_norm": 0.05755602217200438,
      "learning_rate": 0.00014193090637303773,
      "loss": 0.8526,
      "step": 107380
    },
    {
      "epoch": 1.8853912463350833,
      "grad_norm": 0.07767271905554261,
      "learning_rate": 0.00014192075950442124,
      "loss": 0.8425,
      "step": 107390
    },
    {
      "epoch": 1.8855668112150845,
      "grad_norm": 0.06326985499505673,
      "learning_rate": 0.00014191061211733086,
      "loss": 0.8396,
      "step": 107400
    },
    {
      "epoch": 1.885742376095086,
      "grad_norm": 0.06557530274851213,
      "learning_rate": 0.00014190046421189522,
      "loss": 0.8453,
      "step": 107410
    },
    {
      "epoch": 1.8859179409750872,
      "grad_norm": 0.05094212131436509,
      "learning_rate": 0.00014189031578824284,
      "loss": 0.8506,
      "step": 107420
    },
    {
      "epoch": 1.8860935058550887,
      "grad_norm": 0.06670241020488299,
      "learning_rate": 0.00014188016684650237,
      "loss": 0.842,
      "step": 107430
    },
    {
      "epoch": 1.8862690707350902,
      "grad_norm": 0.07395802676178292,
      "learning_rate": 0.00014187001738680233,
      "loss": 0.8379,
      "step": 107440
    },
    {
      "epoch": 1.8864446356150917,
      "grad_norm": 0.0835616541873766,
      "learning_rate": 0.00014185986740927132,
      "loss": 0.8363,
      "step": 107450
    },
    {
      "epoch": 1.886620200495093,
      "grad_norm": 0.06240295038700513,
      "learning_rate": 0.00014184971691403792,
      "loss": 0.8469,
      "step": 107460
    },
    {
      "epoch": 1.8867957653750944,
      "grad_norm": 0.12309507692733078,
      "learning_rate": 0.00014183956590123085,
      "loss": 0.8418,
      "step": 107470
    },
    {
      "epoch": 1.8869713302550957,
      "grad_norm": 0.0643614759255706,
      "learning_rate": 0.0001418294143709786,
      "loss": 0.8444,
      "step": 107480
    },
    {
      "epoch": 1.8871468951350971,
      "grad_norm": 0.055088249613504475,
      "learning_rate": 0.00014181926232340988,
      "loss": 0.8469,
      "step": 107490
    },
    {
      "epoch": 1.8873224600150986,
      "grad_norm": 0.060458193505361635,
      "learning_rate": 0.0001418091097586533,
      "loss": 0.8358,
      "step": 107500
    },
    {
      "epoch": 1.8874980248951,
      "grad_norm": 0.06231488983709582,
      "learning_rate": 0.0001417989566768374,
      "loss": 0.8451,
      "step": 107510
    },
    {
      "epoch": 1.8876735897751014,
      "grad_norm": 0.07789856960182835,
      "learning_rate": 0.00014178880307809096,
      "loss": 0.8454,
      "step": 107520
    },
    {
      "epoch": 1.8878491546551028,
      "grad_norm": 0.07616474544878739,
      "learning_rate": 0.00014177864896254254,
      "loss": 0.8547,
      "step": 107530
    },
    {
      "epoch": 1.888024719535104,
      "grad_norm": 0.05300663868160519,
      "learning_rate": 0.00014176849433032086,
      "loss": 0.8516,
      "step": 107540
    },
    {
      "epoch": 1.8882002844151056,
      "grad_norm": 0.053665085414996046,
      "learning_rate": 0.00014175833918155455,
      "loss": 0.8469,
      "step": 107550
    },
    {
      "epoch": 1.888375849295107,
      "grad_norm": 0.04924250682414715,
      "learning_rate": 0.00014174818351637224,
      "loss": 0.8574,
      "step": 107560
    },
    {
      "epoch": 1.8885514141751085,
      "grad_norm": 0.06290635314423378,
      "learning_rate": 0.00014173802733490263,
      "loss": 0.8451,
      "step": 107570
    },
    {
      "epoch": 1.8887269790551098,
      "grad_norm": 0.06732807430868974,
      "learning_rate": 0.00014172787063727443,
      "loss": 0.847,
      "step": 107580
    },
    {
      "epoch": 1.8889025439351113,
      "grad_norm": 0.05172818029006134,
      "learning_rate": 0.00014171771342361627,
      "loss": 0.8516,
      "step": 107590
    },
    {
      "epoch": 1.8890781088151125,
      "grad_norm": 0.06351186401231626,
      "learning_rate": 0.0001417075556940569,
      "loss": 0.8422,
      "step": 107600
    },
    {
      "epoch": 1.889253673695114,
      "grad_norm": 0.057293301440097075,
      "learning_rate": 0.000141697397448725,
      "loss": 0.8478,
      "step": 107610
    },
    {
      "epoch": 1.8894292385751155,
      "grad_norm": 0.0673215092221544,
      "learning_rate": 0.00014168723868774927,
      "loss": 0.8395,
      "step": 107620
    },
    {
      "epoch": 1.889604803455117,
      "grad_norm": 0.04823121354755642,
      "learning_rate": 0.0001416770794112584,
      "loss": 0.8482,
      "step": 107630
    },
    {
      "epoch": 1.8897803683351182,
      "grad_norm": 0.059311494600987495,
      "learning_rate": 0.0001416669196193812,
      "loss": 0.8441,
      "step": 107640
    },
    {
      "epoch": 1.8899559332151197,
      "grad_norm": 0.0840162613643754,
      "learning_rate": 0.00014165675931224626,
      "loss": 0.8499,
      "step": 107650
    },
    {
      "epoch": 1.890131498095121,
      "grad_norm": 0.048877394575244884,
      "learning_rate": 0.00014164659848998247,
      "loss": 0.8539,
      "step": 107660
    },
    {
      "epoch": 1.8903070629751224,
      "grad_norm": 0.061874923136603706,
      "learning_rate": 0.00014163643715271843,
      "loss": 0.8429,
      "step": 107670
    },
    {
      "epoch": 1.8904826278551239,
      "grad_norm": 0.05633089843058496,
      "learning_rate": 0.00014162627530058298,
      "loss": 0.8324,
      "step": 107680
    },
    {
      "epoch": 1.8906581927351254,
      "grad_norm": 0.05115625617394214,
      "learning_rate": 0.0001416161129337048,
      "loss": 0.8472,
      "step": 107690
    },
    {
      "epoch": 1.8908337576151266,
      "grad_norm": 0.06915576366332989,
      "learning_rate": 0.00014160595005221268,
      "loss": 0.8471,
      "step": 107700
    },
    {
      "epoch": 1.891009322495128,
      "grad_norm": 0.05039608585626326,
      "learning_rate": 0.00014159578665623543,
      "loss": 0.8472,
      "step": 107710
    },
    {
      "epoch": 1.8911848873751294,
      "grad_norm": 0.05418986849129086,
      "learning_rate": 0.00014158562274590178,
      "loss": 0.8493,
      "step": 107720
    },
    {
      "epoch": 1.8913604522551308,
      "grad_norm": 0.0637601971100921,
      "learning_rate": 0.0001415754583213405,
      "loss": 0.8464,
      "step": 107730
    },
    {
      "epoch": 1.8915360171351323,
      "grad_norm": 0.0520977026215685,
      "learning_rate": 0.00014156529338268036,
      "loss": 0.8418,
      "step": 107740
    },
    {
      "epoch": 1.8917115820151338,
      "grad_norm": 0.048837147787480155,
      "learning_rate": 0.0001415551279300502,
      "loss": 0.8474,
      "step": 107750
    },
    {
      "epoch": 1.8918871468951353,
      "grad_norm": 0.06491819592977356,
      "learning_rate": 0.0001415449619635788,
      "loss": 0.8531,
      "step": 107760
    },
    {
      "epoch": 1.8920627117751365,
      "grad_norm": 0.04781228622129684,
      "learning_rate": 0.00014153479548339497,
      "loss": 0.844,
      "step": 107770
    },
    {
      "epoch": 1.8922382766551378,
      "grad_norm": 0.0670408520734645,
      "learning_rate": 0.0001415246284896275,
      "loss": 0.8562,
      "step": 107780
    },
    {
      "epoch": 1.8924138415351393,
      "grad_norm": 0.05855727320240304,
      "learning_rate": 0.00014151446098240522,
      "loss": 0.8405,
      "step": 107790
    },
    {
      "epoch": 1.8925894064151407,
      "grad_norm": 0.05332822163093553,
      "learning_rate": 0.00014150429296185696,
      "loss": 0.8566,
      "step": 107800
    },
    {
      "epoch": 1.8927649712951422,
      "grad_norm": 0.04945567835346476,
      "learning_rate": 0.00014149412442811154,
      "loss": 0.8525,
      "step": 107810
    },
    {
      "epoch": 1.8929405361751437,
      "grad_norm": 0.053944223908817014,
      "learning_rate": 0.00014148395538129778,
      "loss": 0.8512,
      "step": 107820
    },
    {
      "epoch": 1.893116101055145,
      "grad_norm": 0.048007295765710396,
      "learning_rate": 0.00014147378582154461,
      "loss": 0.8452,
      "step": 107830
    },
    {
      "epoch": 1.8932916659351462,
      "grad_norm": 0.07807557004147252,
      "learning_rate": 0.00014146361574898076,
      "loss": 0.8347,
      "step": 107840
    },
    {
      "epoch": 1.8934672308151477,
      "grad_norm": 0.05084074038597083,
      "learning_rate": 0.00014145344516373518,
      "loss": 0.8514,
      "step": 107850
    },
    {
      "epoch": 1.8936427956951492,
      "grad_norm": 0.06734713327522247,
      "learning_rate": 0.00014144327406593666,
      "loss": 0.8482,
      "step": 107860
    },
    {
      "epoch": 1.8938183605751506,
      "grad_norm": 0.07760737518094328,
      "learning_rate": 0.00014143310245571413,
      "loss": 0.8449,
      "step": 107870
    },
    {
      "epoch": 1.893993925455152,
      "grad_norm": 0.06337041605995325,
      "learning_rate": 0.00014142293033319646,
      "loss": 0.8447,
      "step": 107880
    },
    {
      "epoch": 1.8941694903351534,
      "grad_norm": 0.06403281990195644,
      "learning_rate": 0.0001414127576985125,
      "loss": 0.8536,
      "step": 107890
    },
    {
      "epoch": 1.8943450552151546,
      "grad_norm": 0.08533104214737414,
      "learning_rate": 0.0001414025845517911,
      "loss": 0.8496,
      "step": 107900
    },
    {
      "epoch": 1.894520620095156,
      "grad_norm": 0.05500028585551226,
      "learning_rate": 0.0001413924108931613,
      "loss": 0.8516,
      "step": 107910
    },
    {
      "epoch": 1.8946961849751576,
      "grad_norm": 0.048955904620565614,
      "learning_rate": 0.00014138223672275187,
      "loss": 0.8458,
      "step": 107920
    },
    {
      "epoch": 1.894871749855159,
      "grad_norm": 0.05491596207331266,
      "learning_rate": 0.00014137206204069174,
      "loss": 0.8467,
      "step": 107930
    },
    {
      "epoch": 1.8950473147351605,
      "grad_norm": 0.05260274246286499,
      "learning_rate": 0.0001413618868471099,
      "loss": 0.8522,
      "step": 107940
    },
    {
      "epoch": 1.8952228796151618,
      "grad_norm": 0.04268190440183228,
      "learning_rate": 0.00014135171114213518,
      "loss": 0.8492,
      "step": 107950
    },
    {
      "epoch": 1.895398444495163,
      "grad_norm": 0.052742469417618634,
      "learning_rate": 0.00014134153492589656,
      "loss": 0.8455,
      "step": 107960
    },
    {
      "epoch": 1.8955740093751645,
      "grad_norm": 0.051877583604409236,
      "learning_rate": 0.00014133135819852295,
      "loss": 0.8534,
      "step": 107970
    },
    {
      "epoch": 1.895749574255166,
      "grad_norm": 0.06891095149725714,
      "learning_rate": 0.0001413211809601433,
      "loss": 0.8481,
      "step": 107980
    },
    {
      "epoch": 1.8959251391351675,
      "grad_norm": 0.06327544022969567,
      "learning_rate": 0.00014131100321088658,
      "loss": 0.8443,
      "step": 107990
    },
    {
      "epoch": 1.896100704015169,
      "grad_norm": 0.06886207825649504,
      "learning_rate": 0.00014130082495088171,
      "loss": 0.8393,
      "step": 108000
    },
    {
      "epoch": 1.8962762688951702,
      "grad_norm": 0.04671398416622909,
      "learning_rate": 0.00014129064618025768,
      "loss": 0.8508,
      "step": 108010
    },
    {
      "epoch": 1.8964518337751715,
      "grad_norm": 0.05240672431194136,
      "learning_rate": 0.0001412804668991434,
      "loss": 0.8425,
      "step": 108020
    },
    {
      "epoch": 1.896627398655173,
      "grad_norm": 0.07130189768631692,
      "learning_rate": 0.00014127028710766794,
      "loss": 0.8539,
      "step": 108030
    },
    {
      "epoch": 1.8968029635351744,
      "grad_norm": 0.05013359050011324,
      "learning_rate": 0.00014126010680596017,
      "loss": 0.8518,
      "step": 108040
    },
    {
      "epoch": 1.896978528415176,
      "grad_norm": 0.06674445799024864,
      "learning_rate": 0.00014124992599414917,
      "loss": 0.8572,
      "step": 108050
    },
    {
      "epoch": 1.8971540932951774,
      "grad_norm": 0.08033225024067958,
      "learning_rate": 0.0001412397446723639,
      "loss": 0.8462,
      "step": 108060
    },
    {
      "epoch": 1.8973296581751786,
      "grad_norm": 0.047149167130092304,
      "learning_rate": 0.00014122956284073332,
      "loss": 0.8437,
      "step": 108070
    },
    {
      "epoch": 1.8975052230551799,
      "grad_norm": 0.06644263957529475,
      "learning_rate": 0.0001412193804993865,
      "loss": 0.8456,
      "step": 108080
    },
    {
      "epoch": 1.8976807879351814,
      "grad_norm": 0.10469896484280211,
      "learning_rate": 0.00014120919764845237,
      "loss": 0.8428,
      "step": 108090
    },
    {
      "epoch": 1.8978563528151828,
      "grad_norm": 0.05441651620206813,
      "learning_rate": 0.00014119901428806008,
      "loss": 0.841,
      "step": 108100
    },
    {
      "epoch": 1.8980319176951843,
      "grad_norm": 0.0629302947625853,
      "learning_rate": 0.0001411888304183385,
      "loss": 0.8524,
      "step": 108110
    },
    {
      "epoch": 1.8982074825751858,
      "grad_norm": 0.09052496886701328,
      "learning_rate": 0.00014117864603941678,
      "loss": 0.8483,
      "step": 108120
    },
    {
      "epoch": 1.898383047455187,
      "grad_norm": 0.046545689701206365,
      "learning_rate": 0.0001411684611514239,
      "loss": 0.8473,
      "step": 108130
    },
    {
      "epoch": 1.8985586123351885,
      "grad_norm": 0.05506670920975157,
      "learning_rate": 0.0001411582757544889,
      "loss": 0.8466,
      "step": 108140
    },
    {
      "epoch": 1.8987341772151898,
      "grad_norm": 0.07989729092275612,
      "learning_rate": 0.0001411480898487409,
      "loss": 0.8434,
      "step": 108150
    },
    {
      "epoch": 1.8989097420951913,
      "grad_norm": 0.04895512959265241,
      "learning_rate": 0.00014113790343430886,
      "loss": 0.8444,
      "step": 108160
    },
    {
      "epoch": 1.8990853069751927,
      "grad_norm": 0.05815092475242129,
      "learning_rate": 0.0001411277165113219,
      "loss": 0.8472,
      "step": 108170
    },
    {
      "epoch": 1.8992608718551942,
      "grad_norm": 0.05642132449511782,
      "learning_rate": 0.00014111752907990913,
      "loss": 0.8437,
      "step": 108180
    },
    {
      "epoch": 1.8994364367351955,
      "grad_norm": 0.06242399528410662,
      "learning_rate": 0.0001411073411401995,
      "loss": 0.8462,
      "step": 108190
    },
    {
      "epoch": 1.899612001615197,
      "grad_norm": 0.06411775789754191,
      "learning_rate": 0.00014109715269232221,
      "loss": 0.8507,
      "step": 108200
    },
    {
      "epoch": 1.8997875664951982,
      "grad_norm": 0.05989055344931941,
      "learning_rate": 0.00014108696373640634,
      "loss": 0.849,
      "step": 108210
    },
    {
      "epoch": 1.8999631313751997,
      "grad_norm": 0.04999944208962908,
      "learning_rate": 0.0001410767742725809,
      "loss": 0.8454,
      "step": 108220
    },
    {
      "epoch": 1.9001386962552012,
      "grad_norm": 0.061125782778037596,
      "learning_rate": 0.00014106658430097508,
      "loss": 0.8453,
      "step": 108230
    },
    {
      "epoch": 1.9003142611352026,
      "grad_norm": 0.058493903619202384,
      "learning_rate": 0.00014105639382171797,
      "loss": 0.8409,
      "step": 108240
    },
    {
      "epoch": 1.9004898260152039,
      "grad_norm": 0.05465425728049224,
      "learning_rate": 0.00014104620283493866,
      "loss": 0.8435,
      "step": 108250
    },
    {
      "epoch": 1.9006653908952054,
      "grad_norm": 0.0752546653543732,
      "learning_rate": 0.00014103601134076625,
      "loss": 0.838,
      "step": 108260
    },
    {
      "epoch": 1.9008409557752066,
      "grad_norm": 0.05950518474494803,
      "learning_rate": 0.00014102581933932994,
      "loss": 0.8495,
      "step": 108270
    },
    {
      "epoch": 1.901016520655208,
      "grad_norm": 0.08677785983640798,
      "learning_rate": 0.00014101562683075885,
      "loss": 0.8475,
      "step": 108280
    },
    {
      "epoch": 1.9011920855352096,
      "grad_norm": 0.056737299995773274,
      "learning_rate": 0.00014100543381518208,
      "loss": 0.8484,
      "step": 108290
    },
    {
      "epoch": 1.901367650415211,
      "grad_norm": 0.06150333002641156,
      "learning_rate": 0.00014099524029272876,
      "loss": 0.8406,
      "step": 108300
    },
    {
      "epoch": 1.9015432152952123,
      "grad_norm": 0.055500822469179395,
      "learning_rate": 0.0001409850462635281,
      "loss": 0.8538,
      "step": 108310
    },
    {
      "epoch": 1.9017187801752138,
      "grad_norm": 0.06849014010795729,
      "learning_rate": 0.00014097485172770929,
      "loss": 0.8523,
      "step": 108320
    },
    {
      "epoch": 1.901894345055215,
      "grad_norm": 0.06510865282463428,
      "learning_rate": 0.00014096465668540138,
      "loss": 0.8409,
      "step": 108330
    },
    {
      "epoch": 1.9020699099352165,
      "grad_norm": 0.07215788809050788,
      "learning_rate": 0.00014095446113673366,
      "loss": 0.8532,
      "step": 108340
    },
    {
      "epoch": 1.902245474815218,
      "grad_norm": 0.04894762146029988,
      "learning_rate": 0.00014094426508183525,
      "loss": 0.8411,
      "step": 108350
    },
    {
      "epoch": 1.9024210396952195,
      "grad_norm": 0.06381695592213568,
      "learning_rate": 0.00014093406852083533,
      "loss": 0.8471,
      "step": 108360
    },
    {
      "epoch": 1.9025966045752207,
      "grad_norm": 0.0433647993516044,
      "learning_rate": 0.0001409238714538631,
      "loss": 0.8514,
      "step": 108370
    },
    {
      "epoch": 1.9027721694552222,
      "grad_norm": 0.04800673712762564,
      "learning_rate": 0.00014091367388104778,
      "loss": 0.843,
      "step": 108380
    },
    {
      "epoch": 1.9029477343352235,
      "grad_norm": 0.04452944757244378,
      "learning_rate": 0.00014090347580251856,
      "loss": 0.8403,
      "step": 108390
    },
    {
      "epoch": 1.903123299215225,
      "grad_norm": 0.07097054075556722,
      "learning_rate": 0.00014089327721840465,
      "loss": 0.8435,
      "step": 108400
    },
    {
      "epoch": 1.9032988640952264,
      "grad_norm": 0.05514961772378471,
      "learning_rate": 0.00014088307812883528,
      "loss": 0.8435,
      "step": 108410
    },
    {
      "epoch": 1.903474428975228,
      "grad_norm": 0.10057120002010742,
      "learning_rate": 0.0001408728785339397,
      "loss": 0.8486,
      "step": 108420
    },
    {
      "epoch": 1.9036499938552291,
      "grad_norm": 0.05689106817574387,
      "learning_rate": 0.00014086267843384703,
      "loss": 0.8476,
      "step": 108430
    },
    {
      "epoch": 1.9038255587352306,
      "grad_norm": 0.0622920908222074,
      "learning_rate": 0.00014085247782868665,
      "loss": 0.8379,
      "step": 108440
    },
    {
      "epoch": 1.9040011236152319,
      "grad_norm": 0.051952681743626196,
      "learning_rate": 0.0001408422767185877,
      "loss": 0.8422,
      "step": 108450
    },
    {
      "epoch": 1.9041766884952334,
      "grad_norm": 0.05840574954758984,
      "learning_rate": 0.00014083207510367954,
      "loss": 0.8416,
      "step": 108460
    },
    {
      "epoch": 1.9043522533752348,
      "grad_norm": 0.04790376409757451,
      "learning_rate": 0.00014082187298409125,
      "loss": 0.8452,
      "step": 108470
    },
    {
      "epoch": 1.9045278182552363,
      "grad_norm": 0.06876775747933524,
      "learning_rate": 0.00014081167035995224,
      "loss": 0.8447,
      "step": 108480
    },
    {
      "epoch": 1.9047033831352378,
      "grad_norm": 0.05461691620414574,
      "learning_rate": 0.00014080146723139176,
      "loss": 0.8426,
      "step": 108490
    },
    {
      "epoch": 1.904878948015239,
      "grad_norm": 0.05606552498528673,
      "learning_rate": 0.00014079126359853905,
      "loss": 0.8502,
      "step": 108500
    },
    {
      "epoch": 1.9050545128952403,
      "grad_norm": 0.04862585245920811,
      "learning_rate": 0.0001407810594615234,
      "loss": 0.8457,
      "step": 108510
    },
    {
      "epoch": 1.9052300777752418,
      "grad_norm": 0.04389575161882601,
      "learning_rate": 0.0001407708548204741,
      "loss": 0.8541,
      "step": 108520
    },
    {
      "epoch": 1.9054056426552433,
      "grad_norm": 0.059297975787292734,
      "learning_rate": 0.00014076064967552044,
      "loss": 0.8461,
      "step": 108530
    },
    {
      "epoch": 1.9055812075352447,
      "grad_norm": 0.05466549868710481,
      "learning_rate": 0.00014075044402679174,
      "loss": 0.8457,
      "step": 108540
    },
    {
      "epoch": 1.9057567724152462,
      "grad_norm": 0.051362282021734505,
      "learning_rate": 0.00014074023787441728,
      "loss": 0.8451,
      "step": 108550
    },
    {
      "epoch": 1.9059323372952475,
      "grad_norm": 0.06356141288763777,
      "learning_rate": 0.00014073003121852642,
      "loss": 0.8367,
      "step": 108560
    },
    {
      "epoch": 1.9061079021752487,
      "grad_norm": 0.05168308454784958,
      "learning_rate": 0.00014071982405924844,
      "loss": 0.8436,
      "step": 108570
    },
    {
      "epoch": 1.9062834670552502,
      "grad_norm": 0.049354065173214466,
      "learning_rate": 0.00014070961639671268,
      "loss": 0.8435,
      "step": 108580
    },
    {
      "epoch": 1.9064590319352517,
      "grad_norm": 0.059741485079001844,
      "learning_rate": 0.00014069940823104848,
      "loss": 0.842,
      "step": 108590
    },
    {
      "epoch": 1.9066345968152532,
      "grad_norm": 0.055434290797836454,
      "learning_rate": 0.0001406891995623851,
      "loss": 0.847,
      "step": 108600
    },
    {
      "epoch": 1.9068101616952546,
      "grad_norm": 0.050470466246593605,
      "learning_rate": 0.00014067899039085205,
      "loss": 0.8525,
      "step": 108610
    },
    {
      "epoch": 1.906985726575256,
      "grad_norm": 0.0550273245575837,
      "learning_rate": 0.00014066878071657853,
      "loss": 0.8437,
      "step": 108620
    },
    {
      "epoch": 1.9071612914552571,
      "grad_norm": 0.07000598880629592,
      "learning_rate": 0.000140658570539694,
      "loss": 0.8449,
      "step": 108630
    },
    {
      "epoch": 1.9073368563352586,
      "grad_norm": 0.06665386800025123,
      "learning_rate": 0.0001406483598603277,
      "loss": 0.8448,
      "step": 108640
    },
    {
      "epoch": 1.90751242121526,
      "grad_norm": 0.05739585879047912,
      "learning_rate": 0.00014063814867860914,
      "loss": 0.8378,
      "step": 108650
    },
    {
      "epoch": 1.9076879860952616,
      "grad_norm": 0.07393926304853243,
      "learning_rate": 0.00014062793699466764,
      "loss": 0.8458,
      "step": 108660
    },
    {
      "epoch": 1.907863550975263,
      "grad_norm": 0.04636745848572456,
      "learning_rate": 0.00014061772480863257,
      "loss": 0.8471,
      "step": 108670
    },
    {
      "epoch": 1.9080391158552643,
      "grad_norm": 0.056886732161853075,
      "learning_rate": 0.00014060751212063334,
      "loss": 0.8418,
      "step": 108680
    },
    {
      "epoch": 1.9082146807352656,
      "grad_norm": 0.06010053147826944,
      "learning_rate": 0.00014059729893079937,
      "loss": 0.8459,
      "step": 108690
    },
    {
      "epoch": 1.908390245615267,
      "grad_norm": 0.05570348717273528,
      "learning_rate": 0.00014058708523925997,
      "loss": 0.8452,
      "step": 108700
    },
    {
      "epoch": 1.9085658104952685,
      "grad_norm": 0.043798744690329544,
      "learning_rate": 0.00014057687104614464,
      "loss": 0.8476,
      "step": 108710
    },
    {
      "epoch": 1.90874137537527,
      "grad_norm": 0.05674659432419544,
      "learning_rate": 0.00014056665635158277,
      "loss": 0.8356,
      "step": 108720
    },
    {
      "epoch": 1.9089169402552715,
      "grad_norm": 0.040027467941607435,
      "learning_rate": 0.0001405564411557038,
      "loss": 0.8473,
      "step": 108730
    },
    {
      "epoch": 1.9090925051352727,
      "grad_norm": 0.07320322339357423,
      "learning_rate": 0.00014054622545863713,
      "loss": 0.8512,
      "step": 108740
    },
    {
      "epoch": 1.909268070015274,
      "grad_norm": 0.05807409866479108,
      "learning_rate": 0.0001405360092605122,
      "loss": 0.8474,
      "step": 108750
    },
    {
      "epoch": 1.9094436348952755,
      "grad_norm": 0.06973971760343531,
      "learning_rate": 0.00014052579256145848,
      "loss": 0.853,
      "step": 108760
    },
    {
      "epoch": 1.909619199775277,
      "grad_norm": 0.06048710126583099,
      "learning_rate": 0.00014051557536160536,
      "loss": 0.8477,
      "step": 108770
    },
    {
      "epoch": 1.9097947646552784,
      "grad_norm": 0.06662035588948004,
      "learning_rate": 0.00014050535766108233,
      "loss": 0.8407,
      "step": 108780
    },
    {
      "epoch": 1.90997032953528,
      "grad_norm": 0.0566361794557259,
      "learning_rate": 0.00014049513946001887,
      "loss": 0.848,
      "step": 108790
    },
    {
      "epoch": 1.9101458944152812,
      "grad_norm": 0.048099938614052797,
      "learning_rate": 0.00014048492075854443,
      "loss": 0.851,
      "step": 108800
    },
    {
      "epoch": 1.9103214592952824,
      "grad_norm": 0.05825740718999886,
      "learning_rate": 0.00014047470155678844,
      "loss": 0.8431,
      "step": 108810
    },
    {
      "epoch": 1.9104970241752839,
      "grad_norm": 0.05128485400278828,
      "learning_rate": 0.00014046448185488038,
      "loss": 0.845,
      "step": 108820
    },
    {
      "epoch": 1.9106725890552854,
      "grad_norm": 0.062375680735361615,
      "learning_rate": 0.00014045426165294986,
      "loss": 0.8491,
      "step": 108830
    },
    {
      "epoch": 1.9108481539352868,
      "grad_norm": 0.0695809499700619,
      "learning_rate": 0.0001404440409511262,
      "loss": 0.8455,
      "step": 108840
    },
    {
      "epoch": 1.9110237188152883,
      "grad_norm": 0.04966323015519648,
      "learning_rate": 0.00014043381974953906,
      "loss": 0.8455,
      "step": 108850
    },
    {
      "epoch": 1.9111992836952896,
      "grad_norm": 0.0543653883322364,
      "learning_rate": 0.00014042359804831785,
      "loss": 0.8505,
      "step": 108860
    },
    {
      "epoch": 1.911374848575291,
      "grad_norm": 0.07059276273912259,
      "learning_rate": 0.00014041337584759208,
      "loss": 0.8448,
      "step": 108870
    },
    {
      "epoch": 1.9115504134552923,
      "grad_norm": 0.06050247972385024,
      "learning_rate": 0.0001404031531474913,
      "loss": 0.8402,
      "step": 108880
    },
    {
      "epoch": 1.9117259783352938,
      "grad_norm": 0.05192358779536679,
      "learning_rate": 0.00014039292994814497,
      "loss": 0.8478,
      "step": 108890
    },
    {
      "epoch": 1.9119015432152953,
      "grad_norm": 0.06754939331277159,
      "learning_rate": 0.00014038270624968272,
      "loss": 0.8405,
      "step": 108900
    },
    {
      "epoch": 1.9120771080952967,
      "grad_norm": 0.06729861697934697,
      "learning_rate": 0.000140372482052234,
      "loss": 0.849,
      "step": 108910
    },
    {
      "epoch": 1.912252672975298,
      "grad_norm": 0.057269650318284425,
      "learning_rate": 0.0001403622573559284,
      "loss": 0.8542,
      "step": 108920
    },
    {
      "epoch": 1.9124282378552995,
      "grad_norm": 0.06792431400727635,
      "learning_rate": 0.00014035203216089548,
      "loss": 0.8474,
      "step": 108930
    },
    {
      "epoch": 1.9126038027353007,
      "grad_norm": 0.05210475335255505,
      "learning_rate": 0.00014034180646726473,
      "loss": 0.8478,
      "step": 108940
    },
    {
      "epoch": 1.9127793676153022,
      "grad_norm": 0.07137871548764076,
      "learning_rate": 0.00014033158027516578,
      "loss": 0.8436,
      "step": 108950
    },
    {
      "epoch": 1.9129549324953037,
      "grad_norm": 0.0685170917587989,
      "learning_rate": 0.00014032135358472815,
      "loss": 0.846,
      "step": 108960
    },
    {
      "epoch": 1.9131304973753052,
      "grad_norm": 0.06395082215294239,
      "learning_rate": 0.00014031112639608146,
      "loss": 0.8449,
      "step": 108970
    },
    {
      "epoch": 1.9133060622553064,
      "grad_norm": 0.06472451947959022,
      "learning_rate": 0.00014030089870935525,
      "loss": 0.8457,
      "step": 108980
    },
    {
      "epoch": 1.913481627135308,
      "grad_norm": 0.06508392025049282,
      "learning_rate": 0.0001402906705246791,
      "loss": 0.8429,
      "step": 108990
    },
    {
      "epoch": 1.9136571920153091,
      "grad_norm": 0.0648303100569029,
      "learning_rate": 0.00014028044184218267,
      "loss": 0.8534,
      "step": 109000
    },
    {
      "epoch": 1.9138327568953106,
      "grad_norm": 0.049307810041276386,
      "learning_rate": 0.00014027021266199547,
      "loss": 0.8445,
      "step": 109010
    },
    {
      "epoch": 1.914008321775312,
      "grad_norm": 0.07176977930937767,
      "learning_rate": 0.0001402599829842472,
      "loss": 0.8471,
      "step": 109020
    },
    {
      "epoch": 1.9141838866553136,
      "grad_norm": 0.06177694039333116,
      "learning_rate": 0.00014024975280906737,
      "loss": 0.8456,
      "step": 109030
    },
    {
      "epoch": 1.9143594515353148,
      "grad_norm": 0.06999889748006373,
      "learning_rate": 0.00014023952213658566,
      "loss": 0.8465,
      "step": 109040
    },
    {
      "epoch": 1.9145350164153163,
      "grad_norm": 0.07429515754065107,
      "learning_rate": 0.00014022929096693172,
      "loss": 0.8484,
      "step": 109050
    },
    {
      "epoch": 1.9147105812953176,
      "grad_norm": 0.0644837563595987,
      "learning_rate": 0.0001402190593002351,
      "loss": 0.853,
      "step": 109060
    },
    {
      "epoch": 1.914886146175319,
      "grad_norm": 0.05921236745748017,
      "learning_rate": 0.0001402088271366255,
      "loss": 0.8456,
      "step": 109070
    },
    {
      "epoch": 1.9150617110553205,
      "grad_norm": 0.04656384317855633,
      "learning_rate": 0.00014019859447623255,
      "loss": 0.845,
      "step": 109080
    },
    {
      "epoch": 1.915237275935322,
      "grad_norm": 0.05206015243711889,
      "learning_rate": 0.00014018836131918588,
      "loss": 0.8481,
      "step": 109090
    },
    {
      "epoch": 1.9154128408153233,
      "grad_norm": 0.06351582137420657,
      "learning_rate": 0.0001401781276656152,
      "loss": 0.846,
      "step": 109100
    },
    {
      "epoch": 1.9155884056953247,
      "grad_norm": 0.07567837322953253,
      "learning_rate": 0.00014016789351565007,
      "loss": 0.8481,
      "step": 109110
    },
    {
      "epoch": 1.915763970575326,
      "grad_norm": 0.07370846910471293,
      "learning_rate": 0.0001401576588694203,
      "loss": 0.8477,
      "step": 109120
    },
    {
      "epoch": 1.9159395354553275,
      "grad_norm": 0.059045983361408004,
      "learning_rate": 0.00014014742372705545,
      "loss": 0.8473,
      "step": 109130
    },
    {
      "epoch": 1.916115100335329,
      "grad_norm": 0.08185170310352748,
      "learning_rate": 0.00014013718808868523,
      "loss": 0.8442,
      "step": 109140
    },
    {
      "epoch": 1.9162906652153304,
      "grad_norm": 0.0611465855854901,
      "learning_rate": 0.00014012695195443934,
      "loss": 0.8424,
      "step": 109150
    },
    {
      "epoch": 1.9164662300953317,
      "grad_norm": 0.06606384355049631,
      "learning_rate": 0.00014011671532444746,
      "loss": 0.8507,
      "step": 109160
    },
    {
      "epoch": 1.9166417949753332,
      "grad_norm": 0.04626462745905083,
      "learning_rate": 0.00014010647819883934,
      "loss": 0.8516,
      "step": 109170
    },
    {
      "epoch": 1.9168173598553344,
      "grad_norm": 0.07024736439805652,
      "learning_rate": 0.0001400962405777446,
      "loss": 0.8412,
      "step": 109180
    },
    {
      "epoch": 1.916992924735336,
      "grad_norm": 0.051064654704056994,
      "learning_rate": 0.00014008600246129304,
      "loss": 0.8413,
      "step": 109190
    },
    {
      "epoch": 1.9171684896153374,
      "grad_norm": 0.06400710338605019,
      "learning_rate": 0.00014007576384961434,
      "loss": 0.8434,
      "step": 109200
    },
    {
      "epoch": 1.9173440544953388,
      "grad_norm": 0.06078629904104021,
      "learning_rate": 0.0001400655247428382,
      "loss": 0.8412,
      "step": 109210
    },
    {
      "epoch": 1.9175196193753403,
      "grad_norm": 0.040037408112853556,
      "learning_rate": 0.00014005528514109437,
      "loss": 0.8474,
      "step": 109220
    },
    {
      "epoch": 1.9176951842553416,
      "grad_norm": 0.05394502441971714,
      "learning_rate": 0.00014004504504451257,
      "loss": 0.841,
      "step": 109230
    },
    {
      "epoch": 1.9178707491353428,
      "grad_norm": 0.056065289916888264,
      "learning_rate": 0.0001400348044532226,
      "loss": 0.8462,
      "step": 109240
    },
    {
      "epoch": 1.9180463140153443,
      "grad_norm": 0.05271756565695092,
      "learning_rate": 0.0001400245633673542,
      "loss": 0.8491,
      "step": 109250
    },
    {
      "epoch": 1.9182218788953458,
      "grad_norm": 0.05527894040234634,
      "learning_rate": 0.00014001432178703703,
      "loss": 0.8505,
      "step": 109260
    },
    {
      "epoch": 1.9183974437753473,
      "grad_norm": 0.05637605731152893,
      "learning_rate": 0.00014000407971240098,
      "loss": 0.85,
      "step": 109270
    },
    {
      "epoch": 1.9185730086553487,
      "grad_norm": 0.07000617243914982,
      "learning_rate": 0.00013999383714357575,
      "loss": 0.8491,
      "step": 109280
    },
    {
      "epoch": 1.91874857353535,
      "grad_norm": 0.05542083997295546,
      "learning_rate": 0.00013998359408069115,
      "loss": 0.8509,
      "step": 109290
    },
    {
      "epoch": 1.9189241384153513,
      "grad_norm": 0.045338515380453144,
      "learning_rate": 0.00013997335052387696,
      "loss": 0.8553,
      "step": 109300
    },
    {
      "epoch": 1.9190997032953527,
      "grad_norm": 0.06074213584416928,
      "learning_rate": 0.0001399631064732629,
      "loss": 0.8605,
      "step": 109310
    },
    {
      "epoch": 1.9192752681753542,
      "grad_norm": 0.06432284858839514,
      "learning_rate": 0.00013995286192897884,
      "loss": 0.8422,
      "step": 109320
    },
    {
      "epoch": 1.9194508330553557,
      "grad_norm": 0.05767675634166488,
      "learning_rate": 0.0001399426168911545,
      "loss": 0.8446,
      "step": 109330
    },
    {
      "epoch": 1.9196263979353572,
      "grad_norm": 0.055823504261203916,
      "learning_rate": 0.0001399323713599198,
      "loss": 0.8416,
      "step": 109340
    },
    {
      "epoch": 1.9198019628153584,
      "grad_norm": 0.05636997761846393,
      "learning_rate": 0.00013992212533540447,
      "loss": 0.8448,
      "step": 109350
    },
    {
      "epoch": 1.9199775276953597,
      "grad_norm": 0.061167589081324934,
      "learning_rate": 0.00013991187881773838,
      "loss": 0.84,
      "step": 109360
    },
    {
      "epoch": 1.9201530925753612,
      "grad_norm": 0.07063325495232811,
      "learning_rate": 0.0001399016318070513,
      "loss": 0.8561,
      "step": 109370
    },
    {
      "epoch": 1.9203286574553626,
      "grad_norm": 0.07356901460173183,
      "learning_rate": 0.0001398913843034731,
      "loss": 0.8441,
      "step": 109380
    },
    {
      "epoch": 1.920504222335364,
      "grad_norm": 0.08150419853141022,
      "learning_rate": 0.00013988113630713361,
      "loss": 0.8429,
      "step": 109390
    },
    {
      "epoch": 1.9206797872153656,
      "grad_norm": 0.043364890412105575,
      "learning_rate": 0.00013987088781816266,
      "loss": 0.8426,
      "step": 109400
    },
    {
      "epoch": 1.9208553520953668,
      "grad_norm": 0.0545539331577491,
      "learning_rate": 0.00013986063883669015,
      "loss": 0.8459,
      "step": 109410
    },
    {
      "epoch": 1.921030916975368,
      "grad_norm": 0.05165759900007163,
      "learning_rate": 0.00013985038936284588,
      "loss": 0.8481,
      "step": 109420
    },
    {
      "epoch": 1.9212064818553696,
      "grad_norm": 0.054786925532448974,
      "learning_rate": 0.0001398401393967597,
      "loss": 0.8539,
      "step": 109430
    },
    {
      "epoch": 1.921382046735371,
      "grad_norm": 0.057168807941532236,
      "learning_rate": 0.00013982988893856152,
      "loss": 0.8514,
      "step": 109440
    },
    {
      "epoch": 1.9215576116153725,
      "grad_norm": 0.05167601235074014,
      "learning_rate": 0.0001398196379883812,
      "loss": 0.8392,
      "step": 109450
    },
    {
      "epoch": 1.921733176495374,
      "grad_norm": 0.05557451034383708,
      "learning_rate": 0.00013980938654634866,
      "loss": 0.846,
      "step": 109460
    },
    {
      "epoch": 1.9219087413753753,
      "grad_norm": 0.06526894633094976,
      "learning_rate": 0.00013979913461259372,
      "loss": 0.8468,
      "step": 109470
    },
    {
      "epoch": 1.9220843062553765,
      "grad_norm": 0.06215766399190211,
      "learning_rate": 0.00013978888218724636,
      "loss": 0.8404,
      "step": 109480
    },
    {
      "epoch": 1.922259871135378,
      "grad_norm": 0.061054069116031874,
      "learning_rate": 0.00013977862927043637,
      "loss": 0.847,
      "step": 109490
    },
    {
      "epoch": 1.9224354360153795,
      "grad_norm": 0.05264543287121562,
      "learning_rate": 0.00013976837586229373,
      "loss": 0.8493,
      "step": 109500
    },
    {
      "epoch": 1.922611000895381,
      "grad_norm": 0.07293935715336002,
      "learning_rate": 0.00013975812196294834,
      "loss": 0.8521,
      "step": 109510
    },
    {
      "epoch": 1.9227865657753824,
      "grad_norm": 0.048246841580725945,
      "learning_rate": 0.00013974786757253014,
      "loss": 0.8477,
      "step": 109520
    },
    {
      "epoch": 1.9229621306553837,
      "grad_norm": 0.05283540645870476,
      "learning_rate": 0.000139737612691169,
      "loss": 0.8495,
      "step": 109530
    },
    {
      "epoch": 1.923137695535385,
      "grad_norm": 0.05730390611405973,
      "learning_rate": 0.00013972735731899492,
      "loss": 0.8456,
      "step": 109540
    },
    {
      "epoch": 1.9233132604153864,
      "grad_norm": 0.0538844349981491,
      "learning_rate": 0.0001397171014561378,
      "loss": 0.8485,
      "step": 109550
    },
    {
      "epoch": 1.923488825295388,
      "grad_norm": 0.05032730956745802,
      "learning_rate": 0.00013970684510272754,
      "loss": 0.8426,
      "step": 109560
    },
    {
      "epoch": 1.9236643901753894,
      "grad_norm": 0.06343390128446444,
      "learning_rate": 0.00013969658825889412,
      "loss": 0.8489,
      "step": 109570
    },
    {
      "epoch": 1.9238399550553908,
      "grad_norm": 0.06483015831361037,
      "learning_rate": 0.0001396863309247676,
      "loss": 0.854,
      "step": 109580
    },
    {
      "epoch": 1.924015519935392,
      "grad_norm": 0.0555310418351682,
      "learning_rate": 0.0001396760731004778,
      "loss": 0.8628,
      "step": 109590
    },
    {
      "epoch": 1.9241910848153936,
      "grad_norm": 0.046000328435455914,
      "learning_rate": 0.00013966581478615472,
      "loss": 0.8427,
      "step": 109600
    },
    {
      "epoch": 1.9243666496953948,
      "grad_norm": 0.046976298806344734,
      "learning_rate": 0.0001396555559819284,
      "loss": 0.8433,
      "step": 109610
    },
    {
      "epoch": 1.9245422145753963,
      "grad_norm": 0.05438040493530965,
      "learning_rate": 0.00013964529668792874,
      "loss": 0.8472,
      "step": 109620
    },
    {
      "epoch": 1.9247177794553978,
      "grad_norm": 0.05905506738325855,
      "learning_rate": 0.00013963503690428582,
      "loss": 0.8521,
      "step": 109630
    },
    {
      "epoch": 1.9248933443353993,
      "grad_norm": 0.05055191510228411,
      "learning_rate": 0.00013962477663112953,
      "loss": 0.8469,
      "step": 109640
    },
    {
      "epoch": 1.9250689092154005,
      "grad_norm": 0.07253488823013744,
      "learning_rate": 0.0001396145158685899,
      "loss": 0.8458,
      "step": 109650
    },
    {
      "epoch": 1.925244474095402,
      "grad_norm": 0.04859578234710109,
      "learning_rate": 0.000139604254616797,
      "loss": 0.8429,
      "step": 109660
    },
    {
      "epoch": 1.9254200389754033,
      "grad_norm": 0.08736948800487222,
      "learning_rate": 0.00013959399287588077,
      "loss": 0.8439,
      "step": 109670
    },
    {
      "epoch": 1.9255956038554047,
      "grad_norm": 0.05348680421067987,
      "learning_rate": 0.00013958373064597125,
      "loss": 0.8466,
      "step": 109680
    },
    {
      "epoch": 1.9257711687354062,
      "grad_norm": 0.06695755715051982,
      "learning_rate": 0.00013957346792719847,
      "loss": 0.8514,
      "step": 109690
    },
    {
      "epoch": 1.9259467336154077,
      "grad_norm": 0.051682733815016424,
      "learning_rate": 0.0001395632047196925,
      "loss": 0.8461,
      "step": 109700
    },
    {
      "epoch": 1.926122298495409,
      "grad_norm": 0.06197047889673589,
      "learning_rate": 0.00013955294102358328,
      "loss": 0.8487,
      "step": 109710
    },
    {
      "epoch": 1.9262978633754104,
      "grad_norm": 0.06252049921222398,
      "learning_rate": 0.00013954267683900093,
      "loss": 0.8466,
      "step": 109720
    },
    {
      "epoch": 1.9264734282554117,
      "grad_norm": 0.053533256831242326,
      "learning_rate": 0.00013953241216607549,
      "loss": 0.8482,
      "step": 109730
    },
    {
      "epoch": 1.9266489931354132,
      "grad_norm": 0.048783418057774336,
      "learning_rate": 0.00013952214700493697,
      "loss": 0.8423,
      "step": 109740
    },
    {
      "epoch": 1.9268245580154146,
      "grad_norm": 0.06079491353531496,
      "learning_rate": 0.0001395118813557155,
      "loss": 0.852,
      "step": 109750
    },
    {
      "epoch": 1.9270001228954161,
      "grad_norm": 0.07017594789607368,
      "learning_rate": 0.00013950161521854111,
      "loss": 0.8552,
      "step": 109760
    },
    {
      "epoch": 1.9271756877754174,
      "grad_norm": 0.06121378688487306,
      "learning_rate": 0.00013949134859354384,
      "loss": 0.8451,
      "step": 109770
    },
    {
      "epoch": 1.9273512526554188,
      "grad_norm": 0.07295320066324712,
      "learning_rate": 0.00013948108148085386,
      "loss": 0.8535,
      "step": 109780
    },
    {
      "epoch": 1.92752681753542,
      "grad_norm": 0.04691421691606526,
      "learning_rate": 0.00013947081388060117,
      "loss": 0.8485,
      "step": 109790
    },
    {
      "epoch": 1.9277023824154216,
      "grad_norm": 0.0490133396146313,
      "learning_rate": 0.0001394605457929159,
      "loss": 0.8381,
      "step": 109800
    },
    {
      "epoch": 1.927877947295423,
      "grad_norm": 0.05565900301449928,
      "learning_rate": 0.00013945027721792817,
      "loss": 0.8419,
      "step": 109810
    },
    {
      "epoch": 1.9280535121754245,
      "grad_norm": 0.04745193580482123,
      "learning_rate": 0.00013944000815576803,
      "loss": 0.8431,
      "step": 109820
    },
    {
      "epoch": 1.9282290770554258,
      "grad_norm": 0.0504740322233134,
      "learning_rate": 0.0001394297386065656,
      "loss": 0.8507,
      "step": 109830
    },
    {
      "epoch": 1.9284046419354273,
      "grad_norm": 0.05834547578309637,
      "learning_rate": 0.00013941946857045102,
      "loss": 0.8493,
      "step": 109840
    },
    {
      "epoch": 1.9285802068154285,
      "grad_norm": 0.05564761016223016,
      "learning_rate": 0.00013940919804755446,
      "loss": 0.8463,
      "step": 109850
    },
    {
      "epoch": 1.92875577169543,
      "grad_norm": 0.053876190800339716,
      "learning_rate": 0.00013939892703800595,
      "loss": 0.8488,
      "step": 109860
    },
    {
      "epoch": 1.9289313365754315,
      "grad_norm": 0.05492983571158568,
      "learning_rate": 0.00013938865554193573,
      "loss": 0.838,
      "step": 109870
    },
    {
      "epoch": 1.929106901455433,
      "grad_norm": 0.06891115450340746,
      "learning_rate": 0.00013937838355947385,
      "loss": 0.8442,
      "step": 109880
    },
    {
      "epoch": 1.9292824663354342,
      "grad_norm": 0.05210652973139689,
      "learning_rate": 0.0001393681110907505,
      "loss": 0.8494,
      "step": 109890
    },
    {
      "epoch": 1.9294580312154357,
      "grad_norm": 0.05603195976531089,
      "learning_rate": 0.0001393578381358958,
      "loss": 0.8291,
      "step": 109900
    },
    {
      "epoch": 1.929633596095437,
      "grad_norm": 0.041173388289455964,
      "learning_rate": 0.00013934756469503995,
      "loss": 0.8492,
      "step": 109910
    },
    {
      "epoch": 1.9298091609754384,
      "grad_norm": 0.07105663005916103,
      "learning_rate": 0.00013933729076831314,
      "loss": 0.8424,
      "step": 109920
    },
    {
      "epoch": 1.92998472585544,
      "grad_norm": 0.07890869179027006,
      "learning_rate": 0.0001393270163558455,
      "loss": 0.8464,
      "step": 109930
    },
    {
      "epoch": 1.9301602907354414,
      "grad_norm": 0.05041698930204177,
      "learning_rate": 0.00013931674145776718,
      "loss": 0.8468,
      "step": 109940
    },
    {
      "epoch": 1.9303358556154429,
      "grad_norm": 0.056166543079045235,
      "learning_rate": 0.00013930646607420845,
      "loss": 0.843,
      "step": 109950
    },
    {
      "epoch": 1.930511420495444,
      "grad_norm": 0.04607659360209689,
      "learning_rate": 0.00013929619020529942,
      "loss": 0.8436,
      "step": 109960
    },
    {
      "epoch": 1.9306869853754454,
      "grad_norm": 0.059381698668602656,
      "learning_rate": 0.00013928591385117036,
      "loss": 0.8441,
      "step": 109970
    },
    {
      "epoch": 1.9308625502554468,
      "grad_norm": 0.0778028469933006,
      "learning_rate": 0.00013927563701195142,
      "loss": 0.8404,
      "step": 109980
    },
    {
      "epoch": 1.9310381151354483,
      "grad_norm": 0.05764913897474398,
      "learning_rate": 0.0001392653596877728,
      "loss": 0.8387,
      "step": 109990
    },
    {
      "epoch": 1.9312136800154498,
      "grad_norm": 0.08215842570877105,
      "learning_rate": 0.00013925508187876475,
      "loss": 0.8504,
      "step": 110000
    },
    {
      "epoch": 1.9313892448954513,
      "grad_norm": 0.0644207058567212,
      "learning_rate": 0.00013924480358505747,
      "loss": 0.8584,
      "step": 110010
    },
    {
      "epoch": 1.9315648097754525,
      "grad_norm": 0.055364481097454454,
      "learning_rate": 0.00013923452480678122,
      "loss": 0.8479,
      "step": 110020
    },
    {
      "epoch": 1.9317403746554538,
      "grad_norm": 0.05851137704356893,
      "learning_rate": 0.0001392242455440662,
      "loss": 0.8507,
      "step": 110030
    },
    {
      "epoch": 1.9319159395354553,
      "grad_norm": 0.059985345578728415,
      "learning_rate": 0.00013921396579704267,
      "loss": 0.8499,
      "step": 110040
    },
    {
      "epoch": 1.9320915044154567,
      "grad_norm": 0.07607149495968114,
      "learning_rate": 0.00013920368556584087,
      "loss": 0.8561,
      "step": 110050
    },
    {
      "epoch": 1.9322670692954582,
      "grad_norm": 0.05051007656277145,
      "learning_rate": 0.00013919340485059106,
      "loss": 0.8475,
      "step": 110060
    },
    {
      "epoch": 1.9324426341754597,
      "grad_norm": 0.056546594192532225,
      "learning_rate": 0.00013918312365142346,
      "loss": 0.8448,
      "step": 110070
    },
    {
      "epoch": 1.932618199055461,
      "grad_norm": 0.050334877426547006,
      "learning_rate": 0.00013917284196846838,
      "loss": 0.8554,
      "step": 110080
    },
    {
      "epoch": 1.9327937639354622,
      "grad_norm": 0.0634736074918317,
      "learning_rate": 0.0001391625598018561,
      "loss": 0.838,
      "step": 110090
    },
    {
      "epoch": 1.9329693288154637,
      "grad_norm": 0.05483337694115247,
      "learning_rate": 0.00013915227715171686,
      "loss": 0.842,
      "step": 110100
    },
    {
      "epoch": 1.9331448936954652,
      "grad_norm": 0.05097409870360858,
      "learning_rate": 0.00013914199401818093,
      "loss": 0.8434,
      "step": 110110
    },
    {
      "epoch": 1.9333204585754666,
      "grad_norm": 0.05815932542587577,
      "learning_rate": 0.0001391317104013787,
      "loss": 0.8434,
      "step": 110120
    },
    {
      "epoch": 1.9334960234554681,
      "grad_norm": 0.052847771976825636,
      "learning_rate": 0.0001391214263014403,
      "loss": 0.8421,
      "step": 110130
    },
    {
      "epoch": 1.9336715883354694,
      "grad_norm": 0.08553797004293283,
      "learning_rate": 0.00013911114171849618,
      "loss": 0.8447,
      "step": 110140
    },
    {
      "epoch": 1.9338471532154706,
      "grad_norm": 0.04233605013806068,
      "learning_rate": 0.00013910085665267658,
      "loss": 0.8487,
      "step": 110150
    },
    {
      "epoch": 1.934022718095472,
      "grad_norm": 0.07155425528622072,
      "learning_rate": 0.0001390905711041118,
      "loss": 0.8485,
      "step": 110160
    },
    {
      "epoch": 1.9341982829754736,
      "grad_norm": 0.0732868573175537,
      "learning_rate": 0.0001390802850729322,
      "loss": 0.843,
      "step": 110170
    },
    {
      "epoch": 1.934373847855475,
      "grad_norm": 0.050632813856611315,
      "learning_rate": 0.0001390699985592681,
      "loss": 0.8479,
      "step": 110180
    },
    {
      "epoch": 1.9345494127354765,
      "grad_norm": 0.05967548786588748,
      "learning_rate": 0.0001390597115632498,
      "loss": 0.8506,
      "step": 110190
    },
    {
      "epoch": 1.9347249776154778,
      "grad_norm": 0.05182076271245335,
      "learning_rate": 0.00013904942408500767,
      "loss": 0.852,
      "step": 110200
    },
    {
      "epoch": 1.934900542495479,
      "grad_norm": 0.056142480256224606,
      "learning_rate": 0.00013903913612467206,
      "loss": 0.8376,
      "step": 110210
    },
    {
      "epoch": 1.9350761073754805,
      "grad_norm": 0.04969720547337604,
      "learning_rate": 0.0001390288476823733,
      "loss": 0.8516,
      "step": 110220
    },
    {
      "epoch": 1.935251672255482,
      "grad_norm": 0.06205925030680653,
      "learning_rate": 0.00013901855875824175,
      "loss": 0.8398,
      "step": 110230
    },
    {
      "epoch": 1.9354272371354835,
      "grad_norm": 0.054358827800129836,
      "learning_rate": 0.0001390082693524078,
      "loss": 0.848,
      "step": 110240
    },
    {
      "epoch": 1.935602802015485,
      "grad_norm": 0.06906513094921642,
      "learning_rate": 0.00013899797946500172,
      "loss": 0.8418,
      "step": 110250
    },
    {
      "epoch": 1.9357783668954862,
      "grad_norm": 0.08708874220530889,
      "learning_rate": 0.000138987689096154,
      "loss": 0.8382,
      "step": 110260
    },
    {
      "epoch": 1.9359539317754875,
      "grad_norm": 0.07483690366785595,
      "learning_rate": 0.00013897739824599497,
      "loss": 0.838,
      "step": 110270
    },
    {
      "epoch": 1.936129496655489,
      "grad_norm": 0.055521690406955794,
      "learning_rate": 0.00013896710691465507,
      "loss": 0.8472,
      "step": 110280
    },
    {
      "epoch": 1.9363050615354904,
      "grad_norm": 0.09102722329746905,
      "learning_rate": 0.0001389568151022646,
      "loss": 0.8434,
      "step": 110290
    },
    {
      "epoch": 1.936480626415492,
      "grad_norm": 0.04957268099073826,
      "learning_rate": 0.00013894652280895401,
      "loss": 0.8473,
      "step": 110300
    },
    {
      "epoch": 1.9366561912954934,
      "grad_norm": 0.059473850385543196,
      "learning_rate": 0.00013893623003485373,
      "loss": 0.8425,
      "step": 110310
    },
    {
      "epoch": 1.9368317561754946,
      "grad_norm": 0.06053463426270177,
      "learning_rate": 0.00013892593678009415,
      "loss": 0.8482,
      "step": 110320
    },
    {
      "epoch": 1.9370073210554961,
      "grad_norm": 0.04568089749588158,
      "learning_rate": 0.00013891564304480567,
      "loss": 0.8465,
      "step": 110330
    },
    {
      "epoch": 1.9371828859354974,
      "grad_norm": 0.06143853888972699,
      "learning_rate": 0.0001389053488291187,
      "loss": 0.839,
      "step": 110340
    },
    {
      "epoch": 1.9373584508154988,
      "grad_norm": 0.06778017752627759,
      "learning_rate": 0.0001388950541331637,
      "loss": 0.8546,
      "step": 110350
    },
    {
      "epoch": 1.9375340156955003,
      "grad_norm": 0.06541387657628221,
      "learning_rate": 0.0001388847589570711,
      "loss": 0.8541,
      "step": 110360
    },
    {
      "epoch": 1.9377095805755018,
      "grad_norm": 0.0492002750964673,
      "learning_rate": 0.00013887446330097136,
      "loss": 0.8482,
      "step": 110370
    },
    {
      "epoch": 1.937885145455503,
      "grad_norm": 0.06129444877002491,
      "learning_rate": 0.00013886416716499492,
      "loss": 0.8435,
      "step": 110380
    },
    {
      "epoch": 1.9380607103355045,
      "grad_norm": 0.05518731498555092,
      "learning_rate": 0.00013885387054927224,
      "loss": 0.8479,
      "step": 110390
    },
    {
      "epoch": 1.9382362752155058,
      "grad_norm": 0.07131631111995335,
      "learning_rate": 0.00013884357345393374,
      "loss": 0.8432,
      "step": 110400
    },
    {
      "epoch": 1.9384118400955073,
      "grad_norm": 0.0801187176384715,
      "learning_rate": 0.0001388332758791099,
      "loss": 0.843,
      "step": 110410
    },
    {
      "epoch": 1.9385874049755087,
      "grad_norm": 0.06456239092853251,
      "learning_rate": 0.00013882297782493122,
      "loss": 0.8451,
      "step": 110420
    },
    {
      "epoch": 1.9387629698555102,
      "grad_norm": 0.047566465989190884,
      "learning_rate": 0.00013881267929152818,
      "loss": 0.8407,
      "step": 110430
    },
    {
      "epoch": 1.9389385347355115,
      "grad_norm": 0.06723719921566795,
      "learning_rate": 0.00013880238027903124,
      "loss": 0.8448,
      "step": 110440
    },
    {
      "epoch": 1.939114099615513,
      "grad_norm": 0.048114590133407346,
      "learning_rate": 0.00013879208078757086,
      "loss": 0.8517,
      "step": 110450
    },
    {
      "epoch": 1.9392896644955142,
      "grad_norm": 0.06503803358546217,
      "learning_rate": 0.0001387817808172776,
      "loss": 0.8456,
      "step": 110460
    },
    {
      "epoch": 1.9394652293755157,
      "grad_norm": 0.05882121869710032,
      "learning_rate": 0.00013877148036828193,
      "loss": 0.8461,
      "step": 110470
    },
    {
      "epoch": 1.9396407942555172,
      "grad_norm": 0.06594265824971789,
      "learning_rate": 0.00013876117944071437,
      "loss": 0.8474,
      "step": 110480
    },
    {
      "epoch": 1.9398163591355186,
      "grad_norm": 0.04677869249307817,
      "learning_rate": 0.00013875087803470545,
      "loss": 0.8424,
      "step": 110490
    },
    {
      "epoch": 1.93999192401552,
      "grad_norm": 0.051964072768149336,
      "learning_rate": 0.00013874057615038564,
      "loss": 0.8484,
      "step": 110500
    },
    {
      "epoch": 1.9401674888955214,
      "grad_norm": 0.046364603562904644,
      "learning_rate": 0.0001387302737878855,
      "loss": 0.845,
      "step": 110510
    },
    {
      "epoch": 1.9403430537755226,
      "grad_norm": 0.05873337809483829,
      "learning_rate": 0.00013871997094733556,
      "loss": 0.8433,
      "step": 110520
    },
    {
      "epoch": 1.940518618655524,
      "grad_norm": 0.0489641474617125,
      "learning_rate": 0.00013870966762886638,
      "loss": 0.8542,
      "step": 110530
    },
    {
      "epoch": 1.9406941835355256,
      "grad_norm": 0.07046771358684527,
      "learning_rate": 0.00013869936383260843,
      "loss": 0.8434,
      "step": 110540
    },
    {
      "epoch": 1.940869748415527,
      "grad_norm": 0.04696090753137456,
      "learning_rate": 0.0001386890595586924,
      "loss": 0.8446,
      "step": 110550
    },
    {
      "epoch": 1.9410453132955283,
      "grad_norm": 0.05853005643756745,
      "learning_rate": 0.00013867875480724873,
      "loss": 0.855,
      "step": 110560
    },
    {
      "epoch": 1.9412208781755298,
      "grad_norm": 0.043471358290616144,
      "learning_rate": 0.00013866844957840797,
      "loss": 0.8496,
      "step": 110570
    },
    {
      "epoch": 1.941396443055531,
      "grad_norm": 0.04515198528544044,
      "learning_rate": 0.0001386581438723008,
      "loss": 0.8478,
      "step": 110580
    },
    {
      "epoch": 1.9415720079355325,
      "grad_norm": 0.05324858016657406,
      "learning_rate": 0.00013864783768905768,
      "loss": 0.8491,
      "step": 110590
    },
    {
      "epoch": 1.941747572815534,
      "grad_norm": 0.08524420360595526,
      "learning_rate": 0.00013863753102880924,
      "loss": 0.8472,
      "step": 110600
    },
    {
      "epoch": 1.9419231376955355,
      "grad_norm": 0.05452026837492923,
      "learning_rate": 0.0001386272238916861,
      "loss": 0.8501,
      "step": 110610
    },
    {
      "epoch": 1.9420987025755367,
      "grad_norm": 0.09513483607386232,
      "learning_rate": 0.00013861691627781883,
      "loss": 0.8473,
      "step": 110620
    },
    {
      "epoch": 1.9422742674555382,
      "grad_norm": 0.054360144956843715,
      "learning_rate": 0.000138606608187338,
      "loss": 0.8425,
      "step": 110630
    },
    {
      "epoch": 1.9424498323355395,
      "grad_norm": 0.05073077666801863,
      "learning_rate": 0.00013859629962037422,
      "loss": 0.8501,
      "step": 110640
    },
    {
      "epoch": 1.942625397215541,
      "grad_norm": 0.06391493835532694,
      "learning_rate": 0.00013858599057705814,
      "loss": 0.8473,
      "step": 110650
    },
    {
      "epoch": 1.9428009620955424,
      "grad_norm": 0.049415378522348924,
      "learning_rate": 0.00013857568105752035,
      "loss": 0.8552,
      "step": 110660
    },
    {
      "epoch": 1.942976526975544,
      "grad_norm": 0.05512735365108504,
      "learning_rate": 0.0001385653710618915,
      "loss": 0.847,
      "step": 110670
    },
    {
      "epoch": 1.9431520918555454,
      "grad_norm": 0.05396216139192124,
      "learning_rate": 0.00013855506059030218,
      "loss": 0.846,
      "step": 110680
    },
    {
      "epoch": 1.9433276567355466,
      "grad_norm": 0.11540291208205888,
      "learning_rate": 0.00013854474964288303,
      "loss": 0.8464,
      "step": 110690
    },
    {
      "epoch": 1.943503221615548,
      "grad_norm": 0.05282965074713177,
      "learning_rate": 0.00013853443821976473,
      "loss": 0.8412,
      "step": 110700
    },
    {
      "epoch": 1.9436787864955494,
      "grad_norm": 0.04938275950187081,
      "learning_rate": 0.00013852412632107787,
      "loss": 0.8401,
      "step": 110710
    },
    {
      "epoch": 1.9438543513755508,
      "grad_norm": 0.1053093634857138,
      "learning_rate": 0.0001385138139469532,
      "loss": 0.8426,
      "step": 110720
    },
    {
      "epoch": 1.9440299162555523,
      "grad_norm": 0.06211397431521468,
      "learning_rate": 0.00013850350109752128,
      "loss": 0.8465,
      "step": 110730
    },
    {
      "epoch": 1.9442054811355538,
      "grad_norm": 0.06471545038863896,
      "learning_rate": 0.0001384931877729128,
      "loss": 0.85,
      "step": 110740
    },
    {
      "epoch": 1.944381046015555,
      "grad_norm": 0.057918616330289635,
      "learning_rate": 0.00013848287397325848,
      "loss": 0.846,
      "step": 110750
    },
    {
      "epoch": 1.9445566108955563,
      "grad_norm": 0.05946825186977022,
      "learning_rate": 0.00013847255969868892,
      "loss": 0.842,
      "step": 110760
    },
    {
      "epoch": 1.9447321757755578,
      "grad_norm": 0.046267994122825555,
      "learning_rate": 0.00013846224494933488,
      "loss": 0.8498,
      "step": 110770
    },
    {
      "epoch": 1.9449077406555593,
      "grad_norm": 0.04900158334707771,
      "learning_rate": 0.00013845192972532702,
      "loss": 0.8518,
      "step": 110780
    },
    {
      "epoch": 1.9450833055355607,
      "grad_norm": 0.054208597130305104,
      "learning_rate": 0.00013844161402679604,
      "loss": 0.84,
      "step": 110790
    },
    {
      "epoch": 1.9452588704155622,
      "grad_norm": 0.05314854600360138,
      "learning_rate": 0.00013843129785387262,
      "loss": 0.8471,
      "step": 110800
    },
    {
      "epoch": 1.9454344352955635,
      "grad_norm": 0.06662735167191434,
      "learning_rate": 0.00013842098120668745,
      "loss": 0.8432,
      "step": 110810
    },
    {
      "epoch": 1.9456100001755647,
      "grad_norm": 0.06542019234872798,
      "learning_rate": 0.0001384106640853713,
      "loss": 0.836,
      "step": 110820
    },
    {
      "epoch": 1.9457855650555662,
      "grad_norm": 0.07337907760309358,
      "learning_rate": 0.0001384003464900549,
      "loss": 0.8466,
      "step": 110830
    },
    {
      "epoch": 1.9459611299355677,
      "grad_norm": 0.05767330555197751,
      "learning_rate": 0.0001383900284208689,
      "loss": 0.8411,
      "step": 110840
    },
    {
      "epoch": 1.9461366948155692,
      "grad_norm": 0.09528948033454833,
      "learning_rate": 0.0001383797098779441,
      "loss": 0.8459,
      "step": 110850
    },
    {
      "epoch": 1.9463122596955706,
      "grad_norm": 0.05531033053246291,
      "learning_rate": 0.0001383693908614112,
      "loss": 0.8398,
      "step": 110860
    },
    {
      "epoch": 1.946487824575572,
      "grad_norm": 0.04316658181336026,
      "learning_rate": 0.00013835907137140093,
      "loss": 0.8459,
      "step": 110870
    },
    {
      "epoch": 1.9466633894555732,
      "grad_norm": 0.06599577482938304,
      "learning_rate": 0.0001383487514080441,
      "loss": 0.8426,
      "step": 110880
    },
    {
      "epoch": 1.9468389543355746,
      "grad_norm": 0.043773838352917614,
      "learning_rate": 0.00013833843097147146,
      "loss": 0.8492,
      "step": 110890
    },
    {
      "epoch": 1.9470145192155761,
      "grad_norm": 0.0613839366703142,
      "learning_rate": 0.00013832811006181367,
      "loss": 0.8498,
      "step": 110900
    },
    {
      "epoch": 1.9471900840955776,
      "grad_norm": 0.08432225639555833,
      "learning_rate": 0.0001383177886792016,
      "loss": 0.8521,
      "step": 110910
    },
    {
      "epoch": 1.947365648975579,
      "grad_norm": 0.047474200986894866,
      "learning_rate": 0.00013830746682376603,
      "loss": 0.8549,
      "step": 110920
    },
    {
      "epoch": 1.9475412138555803,
      "grad_norm": 0.054698344460265266,
      "learning_rate": 0.00013829714449563766,
      "loss": 0.8457,
      "step": 110930
    },
    {
      "epoch": 1.9477167787355816,
      "grad_norm": 0.06654912234448594,
      "learning_rate": 0.00013828682169494735,
      "loss": 0.8449,
      "step": 110940
    },
    {
      "epoch": 1.947892343615583,
      "grad_norm": 0.049443629112518046,
      "learning_rate": 0.00013827649842182587,
      "loss": 0.847,
      "step": 110950
    },
    {
      "epoch": 1.9480679084955845,
      "grad_norm": 0.05911688569446259,
      "learning_rate": 0.000138266174676404,
      "loss": 0.8435,
      "step": 110960
    },
    {
      "epoch": 1.948243473375586,
      "grad_norm": 0.06707515792637,
      "learning_rate": 0.00013825585045881253,
      "loss": 0.8503,
      "step": 110970
    },
    {
      "epoch": 1.9484190382555875,
      "grad_norm": 0.06218555251206529,
      "learning_rate": 0.00013824552576918232,
      "loss": 0.8511,
      "step": 110980
    },
    {
      "epoch": 1.9485946031355887,
      "grad_norm": 0.07272450904394485,
      "learning_rate": 0.00013823520060764414,
      "loss": 0.8472,
      "step": 110990
    },
    {
      "epoch": 1.94877016801559,
      "grad_norm": 0.04967469027415058,
      "learning_rate": 0.00013822487497432885,
      "loss": 0.8487,
      "step": 111000
    },
    {
      "epoch": 1.9489457328955915,
      "grad_norm": 0.06821883623636296,
      "learning_rate": 0.00013821454886936728,
      "loss": 0.8553,
      "step": 111010
    },
    {
      "epoch": 1.949121297775593,
      "grad_norm": 0.05175521027511555,
      "learning_rate": 0.0001382042222928902,
      "loss": 0.8403,
      "step": 111020
    },
    {
      "epoch": 1.9492968626555944,
      "grad_norm": 0.05170647933957972,
      "learning_rate": 0.00013819389524502852,
      "loss": 0.8447,
      "step": 111030
    },
    {
      "epoch": 1.949472427535596,
      "grad_norm": 0.054873140532061465,
      "learning_rate": 0.00013818356772591305,
      "loss": 0.8383,
      "step": 111040
    },
    {
      "epoch": 1.9496479924155972,
      "grad_norm": 0.06431816070658887,
      "learning_rate": 0.00013817323973567465,
      "loss": 0.8493,
      "step": 111050
    },
    {
      "epoch": 1.9498235572955986,
      "grad_norm": 0.04444172519907014,
      "learning_rate": 0.00013816291127444422,
      "loss": 0.84,
      "step": 111060
    },
    {
      "epoch": 1.9499991221756,
      "grad_norm": 0.051309588789921194,
      "learning_rate": 0.00013815258234235254,
      "loss": 0.8505,
      "step": 111070
    },
    {
      "epoch": 1.9501746870556014,
      "grad_norm": 0.06573534223375747,
      "learning_rate": 0.00013814225293953052,
      "loss": 0.842,
      "step": 111080
    },
    {
      "epoch": 1.9503502519356029,
      "grad_norm": 0.0675857645123154,
      "learning_rate": 0.00013813192306610903,
      "loss": 0.8481,
      "step": 111090
    },
    {
      "epoch": 1.9505258168156043,
      "grad_norm": 0.05682141282059442,
      "learning_rate": 0.00013812159272221896,
      "loss": 0.8489,
      "step": 111100
    },
    {
      "epoch": 1.9507013816956056,
      "grad_norm": 0.06217729963056017,
      "learning_rate": 0.0001381112619079912,
      "loss": 0.8444,
      "step": 111110
    },
    {
      "epoch": 1.950876946575607,
      "grad_norm": 0.04751208094930493,
      "learning_rate": 0.00013810093062355667,
      "loss": 0.837,
      "step": 111120
    },
    {
      "epoch": 1.9510525114556083,
      "grad_norm": 0.05091998410470348,
      "learning_rate": 0.00013809059886904623,
      "loss": 0.8456,
      "step": 111130
    },
    {
      "epoch": 1.9512280763356098,
      "grad_norm": 0.05798100042864082,
      "learning_rate": 0.00013808026664459075,
      "loss": 0.8569,
      "step": 111140
    },
    {
      "epoch": 1.9514036412156113,
      "grad_norm": 0.04211340649263376,
      "learning_rate": 0.0001380699339503212,
      "loss": 0.8479,
      "step": 111150
    },
    {
      "epoch": 1.9515792060956128,
      "grad_norm": 0.0543682623445383,
      "learning_rate": 0.00013805960078636852,
      "loss": 0.843,
      "step": 111160
    },
    {
      "epoch": 1.951754770975614,
      "grad_norm": 0.10325494897322293,
      "learning_rate": 0.00013804926715286358,
      "loss": 0.8469,
      "step": 111170
    },
    {
      "epoch": 1.9519303358556155,
      "grad_norm": 0.06659514440094427,
      "learning_rate": 0.00013803893304993733,
      "loss": 0.8458,
      "step": 111180
    },
    {
      "epoch": 1.9521059007356167,
      "grad_norm": 0.04266441328691606,
      "learning_rate": 0.00013802859847772067,
      "loss": 0.8388,
      "step": 111190
    },
    {
      "epoch": 1.9522814656156182,
      "grad_norm": 0.04988117664058472,
      "learning_rate": 0.00013801826343634458,
      "loss": 0.8477,
      "step": 111200
    },
    {
      "epoch": 1.9524570304956197,
      "grad_norm": 0.06384853441538592,
      "learning_rate": 0.00013800792792594005,
      "loss": 0.84,
      "step": 111210
    },
    {
      "epoch": 1.9526325953756212,
      "grad_norm": 0.047131634168125006,
      "learning_rate": 0.0001379975919466379,
      "loss": 0.8482,
      "step": 111220
    },
    {
      "epoch": 1.9528081602556224,
      "grad_norm": 0.03993722350107053,
      "learning_rate": 0.00013798725549856925,
      "loss": 0.8446,
      "step": 111230
    },
    {
      "epoch": 1.952983725135624,
      "grad_norm": 0.056306085263318775,
      "learning_rate": 0.0001379769185818649,
      "loss": 0.8449,
      "step": 111240
    },
    {
      "epoch": 1.9531592900156252,
      "grad_norm": 0.06439511924802489,
      "learning_rate": 0.00013796658119665593,
      "loss": 0.8479,
      "step": 111250
    },
    {
      "epoch": 1.9533348548956266,
      "grad_norm": 0.07884682817670598,
      "learning_rate": 0.00013795624334307334,
      "loss": 0.8504,
      "step": 111260
    },
    {
      "epoch": 1.9535104197756281,
      "grad_norm": 0.05548761980499351,
      "learning_rate": 0.000137945905021248,
      "loss": 0.8423,
      "step": 111270
    },
    {
      "epoch": 1.9536859846556296,
      "grad_norm": 0.06971279514820031,
      "learning_rate": 0.000137935566231311,
      "loss": 0.8469,
      "step": 111280
    },
    {
      "epoch": 1.9538615495356308,
      "grad_norm": 0.05832756348475977,
      "learning_rate": 0.00013792522697339332,
      "loss": 0.8483,
      "step": 111290
    },
    {
      "epoch": 1.9540371144156323,
      "grad_norm": 0.054820127882666984,
      "learning_rate": 0.00013791488724762592,
      "loss": 0.8379,
      "step": 111300
    },
    {
      "epoch": 1.9542126792956336,
      "grad_norm": 0.05456210953624253,
      "learning_rate": 0.00013790454705413978,
      "loss": 0.8491,
      "step": 111310
    },
    {
      "epoch": 1.954388244175635,
      "grad_norm": 0.059374741276292546,
      "learning_rate": 0.00013789420639306597,
      "loss": 0.8498,
      "step": 111320
    },
    {
      "epoch": 1.9545638090556365,
      "grad_norm": 0.05086896019944371,
      "learning_rate": 0.00013788386526453553,
      "loss": 0.8389,
      "step": 111330
    },
    {
      "epoch": 1.954739373935638,
      "grad_norm": 0.05081394123106769,
      "learning_rate": 0.00013787352366867944,
      "loss": 0.8522,
      "step": 111340
    },
    {
      "epoch": 1.9549149388156393,
      "grad_norm": 0.05744919775767495,
      "learning_rate": 0.00013786318160562873,
      "loss": 0.8493,
      "step": 111350
    },
    {
      "epoch": 1.9550905036956407,
      "grad_norm": 0.06616854035134763,
      "learning_rate": 0.00013785283907551444,
      "loss": 0.8456,
      "step": 111360
    },
    {
      "epoch": 1.955266068575642,
      "grad_norm": 0.047916096529324805,
      "learning_rate": 0.0001378424960784676,
      "loss": 0.8524,
      "step": 111370
    },
    {
      "epoch": 1.9554416334556435,
      "grad_norm": 0.06033553490430679,
      "learning_rate": 0.00013783215261461932,
      "loss": 0.8483,
      "step": 111380
    },
    {
      "epoch": 1.955617198335645,
      "grad_norm": 0.05790408478531029,
      "learning_rate": 0.00013782180868410058,
      "loss": 0.8436,
      "step": 111390
    },
    {
      "epoch": 1.9557927632156464,
      "grad_norm": 0.0558236560889582,
      "learning_rate": 0.00013781146428704249,
      "loss": 0.844,
      "step": 111400
    },
    {
      "epoch": 1.955968328095648,
      "grad_norm": 0.0574266909693405,
      "learning_rate": 0.00013780111942357604,
      "loss": 0.846,
      "step": 111410
    },
    {
      "epoch": 1.9561438929756492,
      "grad_norm": 0.06630645954575247,
      "learning_rate": 0.00013779077409383238,
      "loss": 0.8377,
      "step": 111420
    },
    {
      "epoch": 1.9563194578556504,
      "grad_norm": 0.05294057787728783,
      "learning_rate": 0.00013778042829794258,
      "loss": 0.855,
      "step": 111430
    },
    {
      "epoch": 1.956495022735652,
      "grad_norm": 0.05209037192016049,
      "learning_rate": 0.00013777008203603766,
      "loss": 0.8487,
      "step": 111440
    },
    {
      "epoch": 1.9566705876156534,
      "grad_norm": 0.04902073569170586,
      "learning_rate": 0.00013775973530824883,
      "loss": 0.8392,
      "step": 111450
    },
    {
      "epoch": 1.9568461524956549,
      "grad_norm": 0.04148426298274345,
      "learning_rate": 0.00013774938811470707,
      "loss": 0.8483,
      "step": 111460
    },
    {
      "epoch": 1.9570217173756563,
      "grad_norm": 0.06196535876573423,
      "learning_rate": 0.00013773904045554351,
      "loss": 0.843,
      "step": 111470
    },
    {
      "epoch": 1.9571972822556576,
      "grad_norm": 0.04889457941551729,
      "learning_rate": 0.00013772869233088923,
      "loss": 0.8425,
      "step": 111480
    },
    {
      "epoch": 1.9573728471356588,
      "grad_norm": 0.07889997696083662,
      "learning_rate": 0.00013771834374087542,
      "loss": 0.8411,
      "step": 111490
    },
    {
      "epoch": 1.9575484120156603,
      "grad_norm": 0.04564574881801453,
      "learning_rate": 0.0001377079946856332,
      "loss": 0.851,
      "step": 111500
    },
    {
      "epoch": 1.9577239768956618,
      "grad_norm": 0.049490382507343694,
      "learning_rate": 0.00013769764516529358,
      "loss": 0.849,
      "step": 111510
    },
    {
      "epoch": 1.9578995417756633,
      "grad_norm": 0.054334320191869845,
      "learning_rate": 0.0001376872951799878,
      "loss": 0.8512,
      "step": 111520
    },
    {
      "epoch": 1.9580751066556648,
      "grad_norm": 0.05710555754741688,
      "learning_rate": 0.00013767694472984694,
      "loss": 0.855,
      "step": 111530
    },
    {
      "epoch": 1.958250671535666,
      "grad_norm": 0.06884234692258692,
      "learning_rate": 0.00013766659381500215,
      "loss": 0.845,
      "step": 111540
    },
    {
      "epoch": 1.9584262364156673,
      "grad_norm": 0.05764003128176208,
      "learning_rate": 0.00013765624243558463,
      "loss": 0.8439,
      "step": 111550
    },
    {
      "epoch": 1.9586018012956687,
      "grad_norm": 0.06915201123688364,
      "learning_rate": 0.00013764589059172546,
      "loss": 0.8507,
      "step": 111560
    },
    {
      "epoch": 1.9587773661756702,
      "grad_norm": 0.055482567730472075,
      "learning_rate": 0.00013763553828355588,
      "loss": 0.8486,
      "step": 111570
    },
    {
      "epoch": 1.9589529310556717,
      "grad_norm": 0.05130169807393473,
      "learning_rate": 0.00013762518551120695,
      "loss": 0.8554,
      "step": 111580
    },
    {
      "epoch": 1.9591284959356732,
      "grad_norm": 0.05859079667471612,
      "learning_rate": 0.00013761483227480993,
      "loss": 0.8429,
      "step": 111590
    },
    {
      "epoch": 1.9593040608156744,
      "grad_norm": 0.04607285720226849,
      "learning_rate": 0.00013760447857449597,
      "loss": 0.8585,
      "step": 111600
    },
    {
      "epoch": 1.9594796256956757,
      "grad_norm": 0.08667571554579434,
      "learning_rate": 0.0001375941244103962,
      "loss": 0.8401,
      "step": 111610
    },
    {
      "epoch": 1.9596551905756772,
      "grad_norm": 0.0837991878738869,
      "learning_rate": 0.00013758376978264194,
      "loss": 0.8596,
      "step": 111620
    },
    {
      "epoch": 1.9598307554556786,
      "grad_norm": 0.05735243482677296,
      "learning_rate": 0.0001375734146913643,
      "loss": 0.8373,
      "step": 111630
    },
    {
      "epoch": 1.9600063203356801,
      "grad_norm": 0.04761647211153599,
      "learning_rate": 0.00013756305913669445,
      "loss": 0.852,
      "step": 111640
    },
    {
      "epoch": 1.9601818852156816,
      "grad_norm": 0.06219658495593204,
      "learning_rate": 0.00013755270311876364,
      "loss": 0.8334,
      "step": 111650
    },
    {
      "epoch": 1.9603574500956829,
      "grad_norm": 0.07376402181450545,
      "learning_rate": 0.00013754234663770307,
      "loss": 0.841,
      "step": 111660
    },
    {
      "epoch": 1.960533014975684,
      "grad_norm": 0.05190014774494503,
      "learning_rate": 0.00013753198969364397,
      "loss": 0.8539,
      "step": 111670
    },
    {
      "epoch": 1.9607085798556856,
      "grad_norm": 0.05315925522893024,
      "learning_rate": 0.00013752163228671754,
      "loss": 0.851,
      "step": 111680
    },
    {
      "epoch": 1.960884144735687,
      "grad_norm": 0.06666761720435306,
      "learning_rate": 0.0001375112744170551,
      "loss": 0.8547,
      "step": 111690
    },
    {
      "epoch": 1.9610597096156885,
      "grad_norm": 0.06553831956176707,
      "learning_rate": 0.00013750091608478773,
      "loss": 0.8474,
      "step": 111700
    },
    {
      "epoch": 1.96123527449569,
      "grad_norm": 0.05389466632573982,
      "learning_rate": 0.00013749055729004678,
      "loss": 0.8537,
      "step": 111710
    },
    {
      "epoch": 1.9614108393756913,
      "grad_norm": 0.04755621928665611,
      "learning_rate": 0.00013748019803296353,
      "loss": 0.8415,
      "step": 111720
    },
    {
      "epoch": 1.9615864042556925,
      "grad_norm": 0.06609845973689565,
      "learning_rate": 0.00013746983831366911,
      "loss": 0.8559,
      "step": 111730
    },
    {
      "epoch": 1.961761969135694,
      "grad_norm": 0.0639902368494352,
      "learning_rate": 0.00013745947813229492,
      "loss": 0.8481,
      "step": 111740
    },
    {
      "epoch": 1.9619375340156955,
      "grad_norm": 0.06090942230032175,
      "learning_rate": 0.0001374491174889721,
      "loss": 0.8515,
      "step": 111750
    },
    {
      "epoch": 1.962113098895697,
      "grad_norm": 0.05010329637736998,
      "learning_rate": 0.00013743875638383196,
      "loss": 0.8441,
      "step": 111760
    },
    {
      "epoch": 1.9622886637756984,
      "grad_norm": 0.06701389864253225,
      "learning_rate": 0.00013742839481700582,
      "loss": 0.8468,
      "step": 111770
    },
    {
      "epoch": 1.9624642286556997,
      "grad_norm": 0.05781902753120717,
      "learning_rate": 0.00013741803278862494,
      "loss": 0.843,
      "step": 111780
    },
    {
      "epoch": 1.9626397935357012,
      "grad_norm": 0.05085217351125225,
      "learning_rate": 0.0001374076702988206,
      "loss": 0.8446,
      "step": 111790
    },
    {
      "epoch": 1.9628153584157024,
      "grad_norm": 0.05772546283327184,
      "learning_rate": 0.00013739730734772413,
      "loss": 0.8403,
      "step": 111800
    },
    {
      "epoch": 1.962990923295704,
      "grad_norm": 0.06945430522377904,
      "learning_rate": 0.00013738694393546677,
      "loss": 0.8462,
      "step": 111810
    },
    {
      "epoch": 1.9631664881757054,
      "grad_norm": 0.05208059240649248,
      "learning_rate": 0.00013737658006217986,
      "loss": 0.8437,
      "step": 111820
    },
    {
      "epoch": 1.9633420530557069,
      "grad_norm": 0.09142016896988486,
      "learning_rate": 0.0001373662157279947,
      "loss": 0.8449,
      "step": 111830
    },
    {
      "epoch": 1.9635176179357081,
      "grad_norm": 0.061436105075631776,
      "learning_rate": 0.00013735585093304265,
      "loss": 0.8479,
      "step": 111840
    },
    {
      "epoch": 1.9636931828157096,
      "grad_norm": 0.05581460051234504,
      "learning_rate": 0.00013734548567745497,
      "loss": 0.838,
      "step": 111850
    },
    {
      "epoch": 1.9638687476957108,
      "grad_norm": 0.05685588091062141,
      "learning_rate": 0.00013733511996136304,
      "loss": 0.843,
      "step": 111860
    },
    {
      "epoch": 1.9640443125757123,
      "grad_norm": 0.07599634910143273,
      "learning_rate": 0.00013732475378489816,
      "loss": 0.8427,
      "step": 111870
    },
    {
      "epoch": 1.9642198774557138,
      "grad_norm": 0.04979434157805368,
      "learning_rate": 0.00013731438714819166,
      "loss": 0.8379,
      "step": 111880
    },
    {
      "epoch": 1.9643954423357153,
      "grad_norm": 0.05938079840790157,
      "learning_rate": 0.00013730402005137496,
      "loss": 0.8462,
      "step": 111890
    },
    {
      "epoch": 1.9645710072157165,
      "grad_norm": 0.07252541563748643,
      "learning_rate": 0.00013729365249457934,
      "loss": 0.8473,
      "step": 111900
    },
    {
      "epoch": 1.964746572095718,
      "grad_norm": 0.05589546804008972,
      "learning_rate": 0.00013728328447793622,
      "loss": 0.8553,
      "step": 111910
    },
    {
      "epoch": 1.9649221369757193,
      "grad_norm": 0.05801023838666999,
      "learning_rate": 0.00013727291600157694,
      "loss": 0.8467,
      "step": 111920
    },
    {
      "epoch": 1.9650977018557207,
      "grad_norm": 0.04524414239400619,
      "learning_rate": 0.00013726254706563279,
      "loss": 0.8533,
      "step": 111930
    },
    {
      "epoch": 1.9652732667357222,
      "grad_norm": 0.048569899294953256,
      "learning_rate": 0.0001372521776702353,
      "loss": 0.8536,
      "step": 111940
    },
    {
      "epoch": 1.9654488316157237,
      "grad_norm": 0.05675421069427916,
      "learning_rate": 0.00013724180781551576,
      "loss": 0.8451,
      "step": 111950
    },
    {
      "epoch": 1.965624396495725,
      "grad_norm": 0.04428275246229521,
      "learning_rate": 0.00013723143750160553,
      "loss": 0.8471,
      "step": 111960
    },
    {
      "epoch": 1.9657999613757264,
      "grad_norm": 0.06515163323687068,
      "learning_rate": 0.00013722106672863605,
      "loss": 0.8459,
      "step": 111970
    },
    {
      "epoch": 1.9659755262557277,
      "grad_norm": 0.04549247944855436,
      "learning_rate": 0.00013721069549673876,
      "loss": 0.8456,
      "step": 111980
    },
    {
      "epoch": 1.9661510911357292,
      "grad_norm": 0.050607779214252246,
      "learning_rate": 0.00013720032380604497,
      "loss": 0.8569,
      "step": 111990
    },
    {
      "epoch": 1.9663266560157306,
      "grad_norm": 0.06105337184140935,
      "learning_rate": 0.00013718995165668614,
      "loss": 0.8418,
      "step": 112000
    },
    {
      "epoch": 1.9665022208957321,
      "grad_norm": 0.04909532717851437,
      "learning_rate": 0.0001371795790487937,
      "loss": 0.8532,
      "step": 112010
    },
    {
      "epoch": 1.9666777857757334,
      "grad_norm": 0.07152726501538492,
      "learning_rate": 0.00013716920598249907,
      "loss": 0.8465,
      "step": 112020
    },
    {
      "epoch": 1.9668533506557349,
      "grad_norm": 0.05531608623245482,
      "learning_rate": 0.00013715883245793366,
      "loss": 0.8446,
      "step": 112030
    },
    {
      "epoch": 1.9670289155357361,
      "grad_norm": 0.08141642137353286,
      "learning_rate": 0.00013714845847522891,
      "loss": 0.8485,
      "step": 112040
    },
    {
      "epoch": 1.9672044804157376,
      "grad_norm": 0.05569285191434445,
      "learning_rate": 0.00013713808403451624,
      "loss": 0.8462,
      "step": 112050
    },
    {
      "epoch": 1.967380045295739,
      "grad_norm": 0.10253804804456446,
      "learning_rate": 0.00013712770913592716,
      "loss": 0.8525,
      "step": 112060
    },
    {
      "epoch": 1.9675556101757405,
      "grad_norm": 0.05665519156320135,
      "learning_rate": 0.00013711733377959304,
      "loss": 0.8515,
      "step": 112070
    },
    {
      "epoch": 1.9677311750557418,
      "grad_norm": 0.05156344241792496,
      "learning_rate": 0.00013710695796564546,
      "loss": 0.8495,
      "step": 112080
    },
    {
      "epoch": 1.9679067399357433,
      "grad_norm": 0.057602998794603656,
      "learning_rate": 0.0001370965816942157,
      "loss": 0.8498,
      "step": 112090
    },
    {
      "epoch": 1.9680823048157445,
      "grad_norm": 0.044158164936276216,
      "learning_rate": 0.00013708620496543538,
      "loss": 0.8474,
      "step": 112100
    },
    {
      "epoch": 1.968257869695746,
      "grad_norm": 0.05022559477354006,
      "learning_rate": 0.00013707582777943594,
      "loss": 0.8466,
      "step": 112110
    },
    {
      "epoch": 1.9684334345757475,
      "grad_norm": 0.056847037859170654,
      "learning_rate": 0.00013706545013634878,
      "loss": 0.8506,
      "step": 112120
    },
    {
      "epoch": 1.968608999455749,
      "grad_norm": 0.05358143267516846,
      "learning_rate": 0.00013705507203630554,
      "loss": 0.8439,
      "step": 112130
    },
    {
      "epoch": 1.9687845643357504,
      "grad_norm": 0.06274214332666163,
      "learning_rate": 0.00013704469347943762,
      "loss": 0.8435,
      "step": 112140
    },
    {
      "epoch": 1.9689601292157517,
      "grad_norm": 0.049815264143497066,
      "learning_rate": 0.0001370343144658765,
      "loss": 0.8455,
      "step": 112150
    },
    {
      "epoch": 1.969135694095753,
      "grad_norm": 0.06035834975044513,
      "learning_rate": 0.0001370239349957537,
      "loss": 0.8353,
      "step": 112160
    },
    {
      "epoch": 1.9693112589757544,
      "grad_norm": 0.05868245528217591,
      "learning_rate": 0.00013701355506920074,
      "loss": 0.8413,
      "step": 112170
    },
    {
      "epoch": 1.969486823855756,
      "grad_norm": 0.06134697673325138,
      "learning_rate": 0.00013700317468634917,
      "loss": 0.8383,
      "step": 112180
    },
    {
      "epoch": 1.9696623887357574,
      "grad_norm": 0.055422040190051855,
      "learning_rate": 0.00013699279384733046,
      "loss": 0.8452,
      "step": 112190
    },
    {
      "epoch": 1.9698379536157589,
      "grad_norm": 0.053009749135581284,
      "learning_rate": 0.00013698241255227615,
      "loss": 0.8411,
      "step": 112200
    },
    {
      "epoch": 1.9700135184957601,
      "grad_norm": 0.05532207092116294,
      "learning_rate": 0.00013697203080131776,
      "loss": 0.8457,
      "step": 112210
    },
    {
      "epoch": 1.9701890833757614,
      "grad_norm": 0.06225810359290014,
      "learning_rate": 0.00013696164859458684,
      "loss": 0.8482,
      "step": 112220
    },
    {
      "epoch": 1.9703646482557629,
      "grad_norm": 0.06508954611006998,
      "learning_rate": 0.00013695126593221498,
      "loss": 0.8427,
      "step": 112230
    },
    {
      "epoch": 1.9705402131357643,
      "grad_norm": 0.05064508561542393,
      "learning_rate": 0.00013694088281433365,
      "loss": 0.8567,
      "step": 112240
    },
    {
      "epoch": 1.9707157780157658,
      "grad_norm": 0.04745319522216446,
      "learning_rate": 0.00013693049924107448,
      "loss": 0.8465,
      "step": 112250
    },
    {
      "epoch": 1.9708913428957673,
      "grad_norm": 0.06296571117194988,
      "learning_rate": 0.000136920115212569,
      "loss": 0.8504,
      "step": 112260
    },
    {
      "epoch": 1.9710669077757685,
      "grad_norm": 0.08091111467866258,
      "learning_rate": 0.00013690973072894873,
      "loss": 0.8529,
      "step": 112270
    },
    {
      "epoch": 1.9712424726557698,
      "grad_norm": 0.06184837537969197,
      "learning_rate": 0.00013689934579034536,
      "loss": 0.8459,
      "step": 112280
    },
    {
      "epoch": 1.9714180375357713,
      "grad_norm": 0.051942807004240484,
      "learning_rate": 0.00013688896039689032,
      "loss": 0.8541,
      "step": 112290
    },
    {
      "epoch": 1.9715936024157728,
      "grad_norm": 0.06665885391822368,
      "learning_rate": 0.00013687857454871533,
      "loss": 0.8488,
      "step": 112300
    },
    {
      "epoch": 1.9717691672957742,
      "grad_norm": 0.05234023747761811,
      "learning_rate": 0.0001368681882459519,
      "loss": 0.8488,
      "step": 112310
    },
    {
      "epoch": 1.9719447321757757,
      "grad_norm": 0.05344553054385083,
      "learning_rate": 0.00013685780148873171,
      "loss": 0.8493,
      "step": 112320
    },
    {
      "epoch": 1.972120297055777,
      "grad_norm": 0.06448900193207546,
      "learning_rate": 0.0001368474142771862,
      "loss": 0.8452,
      "step": 112330
    },
    {
      "epoch": 1.9722958619357782,
      "grad_norm": 0.04640013560323404,
      "learning_rate": 0.00013683702661144713,
      "loss": 0.8439,
      "step": 112340
    },
    {
      "epoch": 1.9724714268157797,
      "grad_norm": 0.0586605637525434,
      "learning_rate": 0.0001368266384916461,
      "loss": 0.8387,
      "step": 112350
    },
    {
      "epoch": 1.9726469916957812,
      "grad_norm": 0.08152566482918713,
      "learning_rate": 0.00013681624991791465,
      "loss": 0.843,
      "step": 112360
    },
    {
      "epoch": 1.9728225565757826,
      "grad_norm": 0.05641902023485922,
      "learning_rate": 0.00013680586089038447,
      "loss": 0.8488,
      "step": 112370
    },
    {
      "epoch": 1.9729981214557841,
      "grad_norm": 0.056450206311494264,
      "learning_rate": 0.00013679547140918712,
      "loss": 0.853,
      "step": 112380
    },
    {
      "epoch": 1.9731736863357854,
      "grad_norm": 0.06712940119443858,
      "learning_rate": 0.00013678508147445435,
      "loss": 0.8448,
      "step": 112390
    },
    {
      "epoch": 1.9733492512157866,
      "grad_norm": 0.059442025640018,
      "learning_rate": 0.00013677469108631772,
      "loss": 0.8478,
      "step": 112400
    },
    {
      "epoch": 1.9735248160957881,
      "grad_norm": 0.05494829901521695,
      "learning_rate": 0.0001367643002449089,
      "loss": 0.8527,
      "step": 112410
    },
    {
      "epoch": 1.9737003809757896,
      "grad_norm": 0.07532894017308209,
      "learning_rate": 0.00013675390895035954,
      "loss": 0.8497,
      "step": 112420
    },
    {
      "epoch": 1.973875945855791,
      "grad_norm": 0.04827355376336661,
      "learning_rate": 0.00013674351720280133,
      "loss": 0.8517,
      "step": 112430
    },
    {
      "epoch": 1.9740515107357925,
      "grad_norm": 0.05325019925914543,
      "learning_rate": 0.00013673312500236585,
      "loss": 0.8482,
      "step": 112440
    },
    {
      "epoch": 1.9742270756157938,
      "grad_norm": 0.04516517864573724,
      "learning_rate": 0.00013672273234918486,
      "loss": 0.8468,
      "step": 112450
    },
    {
      "epoch": 1.974402640495795,
      "grad_norm": 0.062616923633693,
      "learning_rate": 0.00013671233924339,
      "loss": 0.8518,
      "step": 112460
    },
    {
      "epoch": 1.9745782053757965,
      "grad_norm": 0.06138062110513502,
      "learning_rate": 0.00013670194568511293,
      "loss": 0.8509,
      "step": 112470
    },
    {
      "epoch": 1.974753770255798,
      "grad_norm": 0.06641328348026769,
      "learning_rate": 0.00013669155167448543,
      "loss": 0.8442,
      "step": 112480
    },
    {
      "epoch": 1.9749293351357995,
      "grad_norm": 0.05254735037647124,
      "learning_rate": 0.00013668115721163907,
      "loss": 0.8545,
      "step": 112490
    },
    {
      "epoch": 1.975104900015801,
      "grad_norm": 0.05014516364830169,
      "learning_rate": 0.0001366707622967056,
      "loss": 0.842,
      "step": 112500
    },
    {
      "epoch": 1.9752804648958022,
      "grad_norm": 0.045213294320378614,
      "learning_rate": 0.00013666036692981678,
      "loss": 0.8513,
      "step": 112510
    },
    {
      "epoch": 1.9754560297758037,
      "grad_norm": 0.06142220035936826,
      "learning_rate": 0.0001366499711111043,
      "loss": 0.8515,
      "step": 112520
    },
    {
      "epoch": 1.975631594655805,
      "grad_norm": 0.05311307592934878,
      "learning_rate": 0.0001366395748406998,
      "loss": 0.8408,
      "step": 112530
    },
    {
      "epoch": 1.9758071595358064,
      "grad_norm": 0.06067453259053971,
      "learning_rate": 0.0001366291781187351,
      "loss": 0.8483,
      "step": 112540
    },
    {
      "epoch": 1.975982724415808,
      "grad_norm": 0.04705721514124499,
      "learning_rate": 0.00013661878094534184,
      "loss": 0.8467,
      "step": 112550
    },
    {
      "epoch": 1.9761582892958094,
      "grad_norm": 0.05608046049114521,
      "learning_rate": 0.0001366083833206518,
      "loss": 0.8457,
      "step": 112560
    },
    {
      "epoch": 1.9763338541758106,
      "grad_norm": 0.07109562141525895,
      "learning_rate": 0.00013659798524479676,
      "loss": 0.853,
      "step": 112570
    },
    {
      "epoch": 1.9765094190558121,
      "grad_norm": 0.05425655221369997,
      "learning_rate": 0.00013658758671790836,
      "loss": 0.8363,
      "step": 112580
    },
    {
      "epoch": 1.9766849839358134,
      "grad_norm": 0.06076979571738597,
      "learning_rate": 0.0001365771877401185,
      "loss": 0.8516,
      "step": 112590
    },
    {
      "epoch": 1.9768605488158149,
      "grad_norm": 0.05899682487385788,
      "learning_rate": 0.0001365667883115588,
      "loss": 0.8482,
      "step": 112600
    },
    {
      "epoch": 1.9770361136958163,
      "grad_norm": 0.056984966705559005,
      "learning_rate": 0.00013655638843236106,
      "loss": 0.8498,
      "step": 112610
    },
    {
      "epoch": 1.9772116785758178,
      "grad_norm": 0.06154549184322054,
      "learning_rate": 0.00013654598810265714,
      "loss": 0.8407,
      "step": 112620
    },
    {
      "epoch": 1.977387243455819,
      "grad_norm": 0.05305198344610314,
      "learning_rate": 0.00013653558732257864,
      "loss": 0.8488,
      "step": 112630
    },
    {
      "epoch": 1.9775628083358205,
      "grad_norm": 0.050025536757220344,
      "learning_rate": 0.00013652518609225754,
      "loss": 0.8499,
      "step": 112640
    },
    {
      "epoch": 1.9777383732158218,
      "grad_norm": 0.06258042626368888,
      "learning_rate": 0.00013651478441182547,
      "loss": 0.8421,
      "step": 112650
    },
    {
      "epoch": 1.9779139380958233,
      "grad_norm": 0.04461053313000469,
      "learning_rate": 0.00013650438228141428,
      "loss": 0.8536,
      "step": 112660
    },
    {
      "epoch": 1.9780895029758248,
      "grad_norm": 0.06529034180677364,
      "learning_rate": 0.00013649397970115577,
      "loss": 0.8391,
      "step": 112670
    },
    {
      "epoch": 1.9782650678558262,
      "grad_norm": 0.07589689807920065,
      "learning_rate": 0.0001364835766711817,
      "loss": 0.8432,
      "step": 112680
    },
    {
      "epoch": 1.9784406327358275,
      "grad_norm": 0.04867637467160566,
      "learning_rate": 0.00013647317319162393,
      "loss": 0.8521,
      "step": 112690
    },
    {
      "epoch": 1.978616197615829,
      "grad_norm": 0.05630312766723142,
      "learning_rate": 0.00013646276926261428,
      "loss": 0.8445,
      "step": 112700
    },
    {
      "epoch": 1.9787917624958302,
      "grad_norm": 0.048956343834586837,
      "learning_rate": 0.00013645236488428453,
      "loss": 0.8452,
      "step": 112710
    },
    {
      "epoch": 1.9789673273758317,
      "grad_norm": 0.06688692944673606,
      "learning_rate": 0.0001364419600567665,
      "loss": 0.8435,
      "step": 112720
    },
    {
      "epoch": 1.9791428922558332,
      "grad_norm": 0.05997518761702158,
      "learning_rate": 0.00013643155478019206,
      "loss": 0.8514,
      "step": 112730
    },
    {
      "epoch": 1.9793184571358347,
      "grad_norm": 0.06524179363941135,
      "learning_rate": 0.00013642114905469305,
      "loss": 0.8499,
      "step": 112740
    },
    {
      "epoch": 1.979494022015836,
      "grad_norm": 0.047889682893501345,
      "learning_rate": 0.00013641074288040127,
      "loss": 0.8504,
      "step": 112750
    },
    {
      "epoch": 1.9796695868958374,
      "grad_norm": 0.06476962900790542,
      "learning_rate": 0.00013640033625744862,
      "loss": 0.8453,
      "step": 112760
    },
    {
      "epoch": 1.9798451517758386,
      "grad_norm": 0.08940209374492061,
      "learning_rate": 0.0001363899291859669,
      "loss": 0.844,
      "step": 112770
    },
    {
      "epoch": 1.9800207166558401,
      "grad_norm": 0.05380295403122007,
      "learning_rate": 0.00013637952166608802,
      "loss": 0.8466,
      "step": 112780
    },
    {
      "epoch": 1.9801962815358416,
      "grad_norm": 0.04696816727002857,
      "learning_rate": 0.00013636911369794378,
      "loss": 0.8457,
      "step": 112790
    },
    {
      "epoch": 1.980371846415843,
      "grad_norm": 0.048097653270389916,
      "learning_rate": 0.00013635870528166613,
      "loss": 0.8485,
      "step": 112800
    },
    {
      "epoch": 1.9805474112958443,
      "grad_norm": 0.05345709559411261,
      "learning_rate": 0.0001363482964173869,
      "loss": 0.8466,
      "step": 112810
    },
    {
      "epoch": 1.9807229761758458,
      "grad_norm": 0.06359764668736587,
      "learning_rate": 0.00013633788710523795,
      "loss": 0.8523,
      "step": 112820
    },
    {
      "epoch": 1.980898541055847,
      "grad_norm": 0.05095872464464077,
      "learning_rate": 0.00013632747734535126,
      "loss": 0.8409,
      "step": 112830
    },
    {
      "epoch": 1.9810741059358485,
      "grad_norm": 0.0605210526562662,
      "learning_rate": 0.0001363170671378586,
      "loss": 0.8483,
      "step": 112840
    },
    {
      "epoch": 1.98124967081585,
      "grad_norm": 0.07275166049036463,
      "learning_rate": 0.0001363066564828919,
      "loss": 0.8469,
      "step": 112850
    },
    {
      "epoch": 1.9814252356958515,
      "grad_norm": 0.05694802572171876,
      "learning_rate": 0.0001362962453805832,
      "loss": 0.8456,
      "step": 112860
    },
    {
      "epoch": 1.981600800575853,
      "grad_norm": 0.06331686768126674,
      "learning_rate": 0.00013628583383106424,
      "loss": 0.8514,
      "step": 112870
    },
    {
      "epoch": 1.9817763654558542,
      "grad_norm": 0.06567382078638014,
      "learning_rate": 0.00013627542183446703,
      "loss": 0.8496,
      "step": 112880
    },
    {
      "epoch": 1.9819519303358555,
      "grad_norm": 0.0589973633253366,
      "learning_rate": 0.00013626500939092345,
      "loss": 0.8529,
      "step": 112890
    },
    {
      "epoch": 1.982127495215857,
      "grad_norm": 0.052477192636918545,
      "learning_rate": 0.00013625459650056543,
      "loss": 0.8493,
      "step": 112900
    },
    {
      "epoch": 1.9823030600958584,
      "grad_norm": 0.046796588699126763,
      "learning_rate": 0.00013624418316352492,
      "loss": 0.8477,
      "step": 112910
    },
    {
      "epoch": 1.98247862497586,
      "grad_norm": 0.0520443631579435,
      "learning_rate": 0.00013623376937993385,
      "loss": 0.8483,
      "step": 112920
    },
    {
      "epoch": 1.9826541898558614,
      "grad_norm": 0.06478219818683789,
      "learning_rate": 0.00013622335514992424,
      "loss": 0.8477,
      "step": 112930
    },
    {
      "epoch": 1.9828297547358626,
      "grad_norm": 0.05176410026108621,
      "learning_rate": 0.0001362129404736279,
      "loss": 0.8457,
      "step": 112940
    },
    {
      "epoch": 1.983005319615864,
      "grad_norm": 0.05096199104229896,
      "learning_rate": 0.00013620252535117688,
      "loss": 0.847,
      "step": 112950
    },
    {
      "epoch": 1.9831808844958654,
      "grad_norm": 0.04649112624620297,
      "learning_rate": 0.0001361921097827031,
      "loss": 0.8442,
      "step": 112960
    },
    {
      "epoch": 1.9833564493758669,
      "grad_norm": 0.04745320822544568,
      "learning_rate": 0.00013618169376833852,
      "loss": 0.845,
      "step": 112970
    },
    {
      "epoch": 1.9835320142558683,
      "grad_norm": 0.0874565540704111,
      "learning_rate": 0.00013617127730821516,
      "loss": 0.8455,
      "step": 112980
    },
    {
      "epoch": 1.9837075791358698,
      "grad_norm": 0.06487104145718568,
      "learning_rate": 0.000136160860402465,
      "loss": 0.8528,
      "step": 112990
    },
    {
      "epoch": 1.983883144015871,
      "grad_norm": 0.06132186082515845,
      "learning_rate": 0.00013615044305122,
      "loss": 0.8425,
      "step": 113000
    },
    {
      "epoch": 1.9840587088958723,
      "grad_norm": 0.052583871069146235,
      "learning_rate": 0.0001361400252546121,
      "loss": 0.8501,
      "step": 113010
    },
    {
      "epoch": 1.9842342737758738,
      "grad_norm": 0.04862285707600157,
      "learning_rate": 0.0001361296070127734,
      "loss": 0.8333,
      "step": 113020
    },
    {
      "epoch": 1.9844098386558753,
      "grad_norm": 0.0472821307519736,
      "learning_rate": 0.0001361191883258358,
      "loss": 0.8422,
      "step": 113030
    },
    {
      "epoch": 1.9845854035358768,
      "grad_norm": 0.05567995079914193,
      "learning_rate": 0.00013610876919393144,
      "loss": 0.8397,
      "step": 113040
    },
    {
      "epoch": 1.9847609684158782,
      "grad_norm": 0.05893082532672756,
      "learning_rate": 0.00013609834961719218,
      "loss": 0.8515,
      "step": 113050
    },
    {
      "epoch": 1.9849365332958795,
      "grad_norm": 0.047785088558486614,
      "learning_rate": 0.0001360879295957501,
      "loss": 0.8411,
      "step": 113060
    },
    {
      "epoch": 1.9851120981758807,
      "grad_norm": 0.05698853413723527,
      "learning_rate": 0.0001360775091297372,
      "loss": 0.8492,
      "step": 113070
    },
    {
      "epoch": 1.9852876630558822,
      "grad_norm": 0.07032625120067681,
      "learning_rate": 0.00013606708821928558,
      "loss": 0.8408,
      "step": 113080
    },
    {
      "epoch": 1.9854632279358837,
      "grad_norm": 0.06140224469173712,
      "learning_rate": 0.00013605666686452724,
      "loss": 0.8426,
      "step": 113090
    },
    {
      "epoch": 1.9856387928158852,
      "grad_norm": 0.06461662001836235,
      "learning_rate": 0.0001360462450655942,
      "loss": 0.8408,
      "step": 113100
    },
    {
      "epoch": 1.9858143576958867,
      "grad_norm": 0.05357033268772333,
      "learning_rate": 0.0001360358228226185,
      "loss": 0.8392,
      "step": 113110
    },
    {
      "epoch": 1.985989922575888,
      "grad_norm": 0.05280731136624155,
      "learning_rate": 0.00013602540013573223,
      "loss": 0.8452,
      "step": 113120
    },
    {
      "epoch": 1.9861654874558892,
      "grad_norm": 0.05083691871196495,
      "learning_rate": 0.00013601497700506745,
      "loss": 0.8429,
      "step": 113130
    },
    {
      "epoch": 1.9863410523358906,
      "grad_norm": 0.05982377455675431,
      "learning_rate": 0.00013600455343075612,
      "loss": 0.8459,
      "step": 113140
    },
    {
      "epoch": 1.9865166172158921,
      "grad_norm": 0.06679549430252404,
      "learning_rate": 0.00013599412941293047,
      "loss": 0.8384,
      "step": 113150
    },
    {
      "epoch": 1.9866921820958936,
      "grad_norm": 0.062478230366772886,
      "learning_rate": 0.00013598370495172251,
      "loss": 0.8514,
      "step": 113160
    },
    {
      "epoch": 1.986867746975895,
      "grad_norm": 0.05927878761538843,
      "learning_rate": 0.00013597328004726425,
      "loss": 0.8436,
      "step": 113170
    },
    {
      "epoch": 1.9870433118558963,
      "grad_norm": 0.056186843238671555,
      "learning_rate": 0.0001359628546996878,
      "loss": 0.8453,
      "step": 113180
    },
    {
      "epoch": 1.9872188767358976,
      "grad_norm": 0.0489662957662357,
      "learning_rate": 0.00013595242890912534,
      "loss": 0.851,
      "step": 113190
    },
    {
      "epoch": 1.987394441615899,
      "grad_norm": 0.05545039656461873,
      "learning_rate": 0.0001359420026757089,
      "loss": 0.8468,
      "step": 113200
    },
    {
      "epoch": 1.9875700064959005,
      "grad_norm": 0.05497134062028974,
      "learning_rate": 0.00013593157599957055,
      "loss": 0.846,
      "step": 113210
    },
    {
      "epoch": 1.987745571375902,
      "grad_norm": 0.05588599263776602,
      "learning_rate": 0.0001359211488808425,
      "loss": 0.8403,
      "step": 113220
    },
    {
      "epoch": 1.9879211362559035,
      "grad_norm": 0.06951052116175246,
      "learning_rate": 0.00013591072131965675,
      "loss": 0.8475,
      "step": 113230
    },
    {
      "epoch": 1.9880967011359048,
      "grad_norm": 0.05856090499716973,
      "learning_rate": 0.00013590029331614548,
      "loss": 0.8512,
      "step": 113240
    },
    {
      "epoch": 1.9882722660159062,
      "grad_norm": 0.053546904069701216,
      "learning_rate": 0.00013588986487044082,
      "loss": 0.8438,
      "step": 113250
    },
    {
      "epoch": 1.9884478308959075,
      "grad_norm": 0.04801920089532098,
      "learning_rate": 0.00013587943598267487,
      "loss": 0.8476,
      "step": 113260
    },
    {
      "epoch": 1.988623395775909,
      "grad_norm": 0.07622268684859984,
      "learning_rate": 0.0001358690066529798,
      "loss": 0.8468,
      "step": 113270
    },
    {
      "epoch": 1.9887989606559104,
      "grad_norm": 0.06363166585750886,
      "learning_rate": 0.00013585857688148774,
      "loss": 0.8498,
      "step": 113280
    },
    {
      "epoch": 1.988974525535912,
      "grad_norm": 0.07435371805308183,
      "learning_rate": 0.0001358481466683308,
      "loss": 0.8501,
      "step": 113290
    },
    {
      "epoch": 1.9891500904159132,
      "grad_norm": 0.06290964644000084,
      "learning_rate": 0.00013583771601364115,
      "loss": 0.8452,
      "step": 113300
    },
    {
      "epoch": 1.9893256552959147,
      "grad_norm": 0.04790562376505452,
      "learning_rate": 0.000135827284917551,
      "loss": 0.8444,
      "step": 113310
    },
    {
      "epoch": 1.989501220175916,
      "grad_norm": 0.052830277630511804,
      "learning_rate": 0.00013581685338019244,
      "loss": 0.8457,
      "step": 113320
    },
    {
      "epoch": 1.9896767850559174,
      "grad_norm": 0.05864565921898938,
      "learning_rate": 0.00013580642140169772,
      "loss": 0.8494,
      "step": 113330
    },
    {
      "epoch": 1.9898523499359189,
      "grad_norm": 0.05687451092108016,
      "learning_rate": 0.00013579598898219897,
      "loss": 0.8461,
      "step": 113340
    },
    {
      "epoch": 1.9900279148159203,
      "grad_norm": 0.05553310759628339,
      "learning_rate": 0.00013578555612182833,
      "loss": 0.8482,
      "step": 113350
    },
    {
      "epoch": 1.9902034796959216,
      "grad_norm": 0.06605189641358583,
      "learning_rate": 0.000135775122820718,
      "loss": 0.8415,
      "step": 113360
    },
    {
      "epoch": 1.990379044575923,
      "grad_norm": 0.06038150575719236,
      "learning_rate": 0.0001357646890790003,
      "loss": 0.8458,
      "step": 113370
    },
    {
      "epoch": 1.9905546094559243,
      "grad_norm": 0.07408070250145019,
      "learning_rate": 0.00013575425489680728,
      "loss": 0.8487,
      "step": 113380
    },
    {
      "epoch": 1.9907301743359258,
      "grad_norm": 0.07869457394329508,
      "learning_rate": 0.00013574382027427116,
      "loss": 0.8493,
      "step": 113390
    },
    {
      "epoch": 1.9909057392159273,
      "grad_norm": 0.07000849879809619,
      "learning_rate": 0.0001357333852115242,
      "loss": 0.8426,
      "step": 113400
    },
    {
      "epoch": 1.9910813040959288,
      "grad_norm": 0.050163243763052125,
      "learning_rate": 0.00013572294970869858,
      "loss": 0.8426,
      "step": 113410
    },
    {
      "epoch": 1.99125686897593,
      "grad_norm": 0.05111133452457369,
      "learning_rate": 0.00013571251376592655,
      "loss": 0.8444,
      "step": 113420
    },
    {
      "epoch": 1.9914324338559315,
      "grad_norm": 0.04787731569194388,
      "learning_rate": 0.00013570207738334032,
      "loss": 0.8456,
      "step": 113430
    },
    {
      "epoch": 1.9916079987359327,
      "grad_norm": 0.08786222168563011,
      "learning_rate": 0.00013569164056107213,
      "loss": 0.8507,
      "step": 113440
    },
    {
      "epoch": 1.9917835636159342,
      "grad_norm": 0.050146745491604,
      "learning_rate": 0.00013568120329925419,
      "loss": 0.8477,
      "step": 113450
    },
    {
      "epoch": 1.9919591284959357,
      "grad_norm": 0.07170616617284797,
      "learning_rate": 0.00013567076559801875,
      "loss": 0.8498,
      "step": 113460
    },
    {
      "epoch": 1.9921346933759372,
      "grad_norm": 0.05019008208724075,
      "learning_rate": 0.00013566032745749806,
      "loss": 0.8452,
      "step": 113470
    },
    {
      "epoch": 1.9923102582559384,
      "grad_norm": 0.04752802994267239,
      "learning_rate": 0.00013564988887782437,
      "loss": 0.848,
      "step": 113480
    },
    {
      "epoch": 1.99248582313594,
      "grad_norm": 0.04473179781088009,
      "learning_rate": 0.00013563944985912996,
      "loss": 0.8417,
      "step": 113490
    },
    {
      "epoch": 1.9926613880159412,
      "grad_norm": 0.06503432033884965,
      "learning_rate": 0.00013562901040154712,
      "loss": 0.8404,
      "step": 113500
    },
    {
      "epoch": 1.9928369528959426,
      "grad_norm": 0.06509192068622259,
      "learning_rate": 0.00013561857050520805,
      "loss": 0.8534,
      "step": 113510
    },
    {
      "epoch": 1.9930125177759441,
      "grad_norm": 0.05307628776245263,
      "learning_rate": 0.00013560813017024506,
      "loss": 0.8408,
      "step": 113520
    },
    {
      "epoch": 1.9931880826559456,
      "grad_norm": 0.05240323420658199,
      "learning_rate": 0.00013559768939679038,
      "loss": 0.8588,
      "step": 113530
    },
    {
      "epoch": 1.9933636475359469,
      "grad_norm": 0.05475029428534364,
      "learning_rate": 0.00013558724818497642,
      "loss": 0.8438,
      "step": 113540
    },
    {
      "epoch": 1.9935392124159483,
      "grad_norm": 0.05833286532931981,
      "learning_rate": 0.00013557680653493537,
      "loss": 0.8309,
      "step": 113550
    },
    {
      "epoch": 1.9937147772959496,
      "grad_norm": 0.14077713735813713,
      "learning_rate": 0.00013556636444679956,
      "loss": 0.8498,
      "step": 113560
    },
    {
      "epoch": 1.993890342175951,
      "grad_norm": 0.049583402173552,
      "learning_rate": 0.00013555592192070125,
      "loss": 0.8409,
      "step": 113570
    },
    {
      "epoch": 1.9940659070559525,
      "grad_norm": 0.05273531365049017,
      "learning_rate": 0.0001355454789567728,
      "loss": 0.8519,
      "step": 113580
    },
    {
      "epoch": 1.994241471935954,
      "grad_norm": 0.051443593076406675,
      "learning_rate": 0.00013553503555514655,
      "loss": 0.8495,
      "step": 113590
    },
    {
      "epoch": 1.9944170368159555,
      "grad_norm": 0.055307706721711924,
      "learning_rate": 0.00013552459171595473,
      "loss": 0.8466,
      "step": 113600
    },
    {
      "epoch": 1.9945926016959568,
      "grad_norm": 0.06382103446475906,
      "learning_rate": 0.0001355141474393298,
      "loss": 0.8492,
      "step": 113610
    },
    {
      "epoch": 1.994768166575958,
      "grad_norm": 0.06311067394725677,
      "learning_rate": 0.00013550370272540394,
      "loss": 0.8474,
      "step": 113620
    },
    {
      "epoch": 1.9949437314559595,
      "grad_norm": 0.049940949473648466,
      "learning_rate": 0.00013549325757430958,
      "loss": 0.8386,
      "step": 113630
    },
    {
      "epoch": 1.995119296335961,
      "grad_norm": 0.04888330499386549,
      "learning_rate": 0.000135482811986179,
      "loss": 0.8442,
      "step": 113640
    },
    {
      "epoch": 1.9952948612159624,
      "grad_norm": 0.05084791964167364,
      "learning_rate": 0.0001354723659611446,
      "loss": 0.8469,
      "step": 113650
    },
    {
      "epoch": 1.995470426095964,
      "grad_norm": 0.06494654435429727,
      "learning_rate": 0.00013546191949933873,
      "loss": 0.839,
      "step": 113660
    },
    {
      "epoch": 1.9956459909759652,
      "grad_norm": 0.060790159166661695,
      "learning_rate": 0.00013545147260089372,
      "loss": 0.8372,
      "step": 113670
    },
    {
      "epoch": 1.9958215558559664,
      "grad_norm": 0.0750173180028978,
      "learning_rate": 0.000135441025265942,
      "loss": 0.846,
      "step": 113680
    },
    {
      "epoch": 1.995997120735968,
      "grad_norm": 0.05890827825219573,
      "learning_rate": 0.00013543057749461582,
      "loss": 0.8551,
      "step": 113690
    },
    {
      "epoch": 1.9961726856159694,
      "grad_norm": 0.05154412548586042,
      "learning_rate": 0.00013542012928704762,
      "loss": 0.8429,
      "step": 113700
    },
    {
      "epoch": 1.9963482504959709,
      "grad_norm": 0.0795098287776006,
      "learning_rate": 0.0001354096806433698,
      "loss": 0.8481,
      "step": 113710
    },
    {
      "epoch": 1.9965238153759723,
      "grad_norm": 0.04770524572030175,
      "learning_rate": 0.00013539923156371479,
      "loss": 0.8485,
      "step": 113720
    },
    {
      "epoch": 1.9966993802559736,
      "grad_norm": 0.061915312129725085,
      "learning_rate": 0.00013538878204821488,
      "loss": 0.8463,
      "step": 113730
    },
    {
      "epoch": 1.9968749451359749,
      "grad_norm": 0.04545566232544252,
      "learning_rate": 0.0001353783320970025,
      "loss": 0.8428,
      "step": 113740
    },
    {
      "epoch": 1.9970505100159763,
      "grad_norm": 0.04865147066821,
      "learning_rate": 0.00013536788171021006,
      "loss": 0.8492,
      "step": 113750
    },
    {
      "epoch": 1.9972260748959778,
      "grad_norm": 0.05772005922730127,
      "learning_rate": 0.00013535743088796998,
      "loss": 0.8437,
      "step": 113760
    },
    {
      "epoch": 1.9974016397759793,
      "grad_norm": 0.052887716663862286,
      "learning_rate": 0.00013534697963041466,
      "loss": 0.8481,
      "step": 113770
    },
    {
      "epoch": 1.9975772046559808,
      "grad_norm": 0.05409510889263886,
      "learning_rate": 0.00013533652793767657,
      "loss": 0.8474,
      "step": 113780
    },
    {
      "epoch": 1.997752769535982,
      "grad_norm": 0.049500497001010106,
      "learning_rate": 0.00013532607580988806,
      "loss": 0.8513,
      "step": 113790
    },
    {
      "epoch": 1.9979283344159833,
      "grad_norm": 0.07757090466959704,
      "learning_rate": 0.0001353156232471816,
      "loss": 0.8474,
      "step": 113800
    },
    {
      "epoch": 1.9981038992959848,
      "grad_norm": 0.061776725536211566,
      "learning_rate": 0.0001353051702496896,
      "loss": 0.8372,
      "step": 113810
    },
    {
      "epoch": 1.9982794641759862,
      "grad_norm": 0.06178439615651428,
      "learning_rate": 0.0001352947168175445,
      "loss": 0.8464,
      "step": 113820
    },
    {
      "epoch": 1.9984550290559877,
      "grad_norm": 0.07912462738574236,
      "learning_rate": 0.00013528426295087882,
      "loss": 0.8436,
      "step": 113830
    },
    {
      "epoch": 1.9986305939359892,
      "grad_norm": 0.05231637641634119,
      "learning_rate": 0.00013527380864982493,
      "loss": 0.8433,
      "step": 113840
    },
    {
      "epoch": 1.9988061588159904,
      "grad_norm": 0.0662815104211963,
      "learning_rate": 0.0001352633539145153,
      "loss": 0.846,
      "step": 113850
    },
    {
      "epoch": 1.9989817236959917,
      "grad_norm": 0.0458997055838155,
      "learning_rate": 0.0001352528987450824,
      "loss": 0.8398,
      "step": 113860
    },
    {
      "epoch": 1.9991572885759932,
      "grad_norm": 0.07614035736272887,
      "learning_rate": 0.00013524244314165875,
      "loss": 0.844,
      "step": 113870
    },
    {
      "epoch": 1.9993328534559947,
      "grad_norm": 0.06490601276969686,
      "learning_rate": 0.00013523198710437676,
      "loss": 0.8393,
      "step": 113880
    },
    {
      "epoch": 1.9995084183359961,
      "grad_norm": 0.05782898867264072,
      "learning_rate": 0.00013522153063336896,
      "loss": 0.8512,
      "step": 113890
    },
    {
      "epoch": 1.9996839832159976,
      "grad_norm": 0.04503327316327342,
      "learning_rate": 0.0001352110737287678,
      "loss": 0.8485,
      "step": 113900
    },
    {
      "epoch": 1.9998595480959989,
      "grad_norm": 0.07053254884046814,
      "learning_rate": 0.00013520061639070574,
      "loss": 0.8464,
      "step": 113910
    },
    {
      "epoch": 2.000035112976,
      "grad_norm": 0.0606326042215767,
      "learning_rate": 0.00013519015861931534,
      "loss": 0.843,
      "step": 113920
    },
    {
      "epoch": 2.0002106778560016,
      "grad_norm": 0.058836147346170244,
      "learning_rate": 0.00013517970041472907,
      "loss": 0.8522,
      "step": 113930
    },
    {
      "epoch": 2.000386242736003,
      "grad_norm": 0.044514210430623855,
      "learning_rate": 0.00013516924177707946,
      "loss": 0.8456,
      "step": 113940
    },
    {
      "epoch": 2.0005618076160046,
      "grad_norm": 0.05777230690172198,
      "learning_rate": 0.00013515878270649906,
      "loss": 0.8451,
      "step": 113950
    },
    {
      "epoch": 2.000737372496006,
      "grad_norm": 0.0502243830580969,
      "learning_rate": 0.00013514832320312022,
      "loss": 0.854,
      "step": 113960
    },
    {
      "epoch": 2.0009129373760075,
      "grad_norm": 0.06955115754152005,
      "learning_rate": 0.00013513786326707568,
      "loss": 0.8539,
      "step": 113970
    },
    {
      "epoch": 2.0010885022560085,
      "grad_norm": 0.04790961120870193,
      "learning_rate": 0.0001351274028984978,
      "loss": 0.8446,
      "step": 113980
    },
    {
      "epoch": 2.00126406713601,
      "grad_norm": 0.06665954622937104,
      "learning_rate": 0.0001351169420975192,
      "loss": 0.8463,
      "step": 113990
    },
    {
      "epoch": 2.0014396320160115,
      "grad_norm": 0.06310645904546874,
      "learning_rate": 0.0001351064808642724,
      "loss": 0.8441,
      "step": 114000
    },
    {
      "epoch": 2.001615196896013,
      "grad_norm": 0.044112943451601806,
      "learning_rate": 0.00013509601919889,
      "loss": 0.8525,
      "step": 114010
    },
    {
      "epoch": 2.0017907617760144,
      "grad_norm": 0.06093453613962986,
      "learning_rate": 0.0001350855571015045,
      "loss": 0.8429,
      "step": 114020
    },
    {
      "epoch": 2.001966326656016,
      "grad_norm": 0.06150864102800133,
      "learning_rate": 0.0001350750945722484,
      "loss": 0.8429,
      "step": 114030
    },
    {
      "epoch": 2.002141891536017,
      "grad_norm": 0.05229750838209784,
      "learning_rate": 0.00013506463161125435,
      "loss": 0.8443,
      "step": 114040
    },
    {
      "epoch": 2.0023174564160184,
      "grad_norm": 0.07207908972500637,
      "learning_rate": 0.0001350541682186549,
      "loss": 0.8479,
      "step": 114050
    },
    {
      "epoch": 2.00249302129602,
      "grad_norm": 0.061654797840156296,
      "learning_rate": 0.0001350437043945826,
      "loss": 0.8476,
      "step": 114060
    },
    {
      "epoch": 2.0026685861760214,
      "grad_norm": 0.05993847115616903,
      "learning_rate": 0.00013503324013917006,
      "loss": 0.8499,
      "step": 114070
    },
    {
      "epoch": 2.002844151056023,
      "grad_norm": 0.052604870392052284,
      "learning_rate": 0.00013502277545254982,
      "loss": 0.8415,
      "step": 114080
    },
    {
      "epoch": 2.0030197159360243,
      "grad_norm": 0.06461825526203878,
      "learning_rate": 0.0001350123103348545,
      "loss": 0.8473,
      "step": 114090
    },
    {
      "epoch": 2.0031952808160254,
      "grad_norm": 0.060306963996115626,
      "learning_rate": 0.0001350018447862167,
      "loss": 0.8529,
      "step": 114100
    },
    {
      "epoch": 2.003370845696027,
      "grad_norm": 0.08836145389322259,
      "learning_rate": 0.000134991378806769,
      "loss": 0.8458,
      "step": 114110
    },
    {
      "epoch": 2.0035464105760283,
      "grad_norm": 0.09286037855120095,
      "learning_rate": 0.00013498091239664403,
      "loss": 0.8471,
      "step": 114120
    },
    {
      "epoch": 2.00372197545603,
      "grad_norm": 0.06145041644904422,
      "learning_rate": 0.00013497044555597438,
      "loss": 0.85,
      "step": 114130
    },
    {
      "epoch": 2.0038975403360313,
      "grad_norm": 0.049034754672637605,
      "learning_rate": 0.0001349599782848927,
      "loss": 0.8379,
      "step": 114140
    },
    {
      "epoch": 2.0040731052160328,
      "grad_norm": 0.061322893966892555,
      "learning_rate": 0.00013494951058353152,
      "loss": 0.8477,
      "step": 114150
    },
    {
      "epoch": 2.004248670096034,
      "grad_norm": 0.05212551232791151,
      "learning_rate": 0.00013493904245202355,
      "loss": 0.8382,
      "step": 114160
    },
    {
      "epoch": 2.0044242349760353,
      "grad_norm": 0.04923527604975416,
      "learning_rate": 0.00013492857389050145,
      "loss": 0.8424,
      "step": 114170
    },
    {
      "epoch": 2.0045997998560368,
      "grad_norm": 0.05638140199273135,
      "learning_rate": 0.0001349181048990978,
      "loss": 0.8476,
      "step": 114180
    },
    {
      "epoch": 2.0047753647360382,
      "grad_norm": 0.06512216120070365,
      "learning_rate": 0.00013490763547794528,
      "loss": 0.8412,
      "step": 114190
    },
    {
      "epoch": 2.0049509296160397,
      "grad_norm": 0.053732995663984706,
      "learning_rate": 0.00013489716562717648,
      "loss": 0.8494,
      "step": 114200
    },
    {
      "epoch": 2.005126494496041,
      "grad_norm": 0.05985599047769587,
      "learning_rate": 0.0001348866953469241,
      "loss": 0.8518,
      "step": 114210
    },
    {
      "epoch": 2.0053020593760422,
      "grad_norm": 0.06915741579989299,
      "learning_rate": 0.00013487622463732084,
      "loss": 0.8454,
      "step": 114220
    },
    {
      "epoch": 2.0054776242560437,
      "grad_norm": 0.049260058539230524,
      "learning_rate": 0.0001348657534984993,
      "loss": 0.8447,
      "step": 114230
    },
    {
      "epoch": 2.005653189136045,
      "grad_norm": 0.07321162544386986,
      "learning_rate": 0.00013485528193059215,
      "loss": 0.8537,
      "step": 114240
    },
    {
      "epoch": 2.0058287540160467,
      "grad_norm": 0.04416521541643712,
      "learning_rate": 0.0001348448099337321,
      "loss": 0.8562,
      "step": 114250
    },
    {
      "epoch": 2.006004318896048,
      "grad_norm": 0.051233888064597444,
      "learning_rate": 0.00013483433750805182,
      "loss": 0.8489,
      "step": 114260
    },
    {
      "epoch": 2.0061798837760496,
      "grad_norm": 0.07776902567710309,
      "learning_rate": 0.00013482386465368403,
      "loss": 0.8443,
      "step": 114270
    },
    {
      "epoch": 2.0063554486560506,
      "grad_norm": 0.05624012643898958,
      "learning_rate": 0.00013481339137076133,
      "loss": 0.8513,
      "step": 114280
    },
    {
      "epoch": 2.006531013536052,
      "grad_norm": 0.06466818814883778,
      "learning_rate": 0.00013480291765941658,
      "loss": 0.8452,
      "step": 114290
    },
    {
      "epoch": 2.0067065784160536,
      "grad_norm": 0.060343804075031295,
      "learning_rate": 0.00013479244351978227,
      "loss": 0.8458,
      "step": 114300
    },
    {
      "epoch": 2.006882143296055,
      "grad_norm": 0.07616588654708759,
      "learning_rate": 0.0001347819689519913,
      "loss": 0.8464,
      "step": 114310
    },
    {
      "epoch": 2.0070577081760566,
      "grad_norm": 0.06208738496621342,
      "learning_rate": 0.00013477149395617628,
      "loss": 0.8458,
      "step": 114320
    },
    {
      "epoch": 2.007233273056058,
      "grad_norm": 0.04949669835570067,
      "learning_rate": 0.00013476101853246994,
      "loss": 0.8421,
      "step": 114330
    },
    {
      "epoch": 2.007408837936059,
      "grad_norm": 0.08528483250846157,
      "learning_rate": 0.00013475054268100506,
      "loss": 0.8535,
      "step": 114340
    },
    {
      "epoch": 2.0075844028160605,
      "grad_norm": 0.29097020814177377,
      "learning_rate": 0.00013474006640191435,
      "loss": 0.8452,
      "step": 114350
    },
    {
      "epoch": 2.007759967696062,
      "grad_norm": 0.05365254475189281,
      "learning_rate": 0.0001347295896953305,
      "loss": 0.8406,
      "step": 114360
    },
    {
      "epoch": 2.0079355325760635,
      "grad_norm": 0.05382612429457709,
      "learning_rate": 0.0001347191125613863,
      "loss": 0.8535,
      "step": 114370
    },
    {
      "epoch": 2.008111097456065,
      "grad_norm": 0.05424787582843832,
      "learning_rate": 0.00013470863500021445,
      "loss": 0.8496,
      "step": 114380
    },
    {
      "epoch": 2.0082866623360665,
      "grad_norm": 0.05137209334914331,
      "learning_rate": 0.00013469815701194776,
      "loss": 0.8456,
      "step": 114390
    },
    {
      "epoch": 2.0084622272160675,
      "grad_norm": 0.05126529508172749,
      "learning_rate": 0.00013468767859671897,
      "loss": 0.8479,
      "step": 114400
    },
    {
      "epoch": 2.008637792096069,
      "grad_norm": 0.062175359237212374,
      "learning_rate": 0.00013467719975466085,
      "loss": 0.8458,
      "step": 114410
    },
    {
      "epoch": 2.0088133569760704,
      "grad_norm": 0.06270036048786773,
      "learning_rate": 0.00013466672048590613,
      "loss": 0.8372,
      "step": 114420
    },
    {
      "epoch": 2.008988921856072,
      "grad_norm": 0.07065660048532044,
      "learning_rate": 0.0001346562407905876,
      "loss": 0.8448,
      "step": 114430
    },
    {
      "epoch": 2.0091644867360734,
      "grad_norm": 0.08576976190254285,
      "learning_rate": 0.00013464576066883805,
      "loss": 0.8484,
      "step": 114440
    },
    {
      "epoch": 2.009340051616075,
      "grad_norm": 0.0721886926160602,
      "learning_rate": 0.00013463528012079028,
      "loss": 0.8478,
      "step": 114450
    },
    {
      "epoch": 2.009515616496076,
      "grad_norm": 0.050288629384035985,
      "learning_rate": 0.00013462479914657708,
      "loss": 0.8412,
      "step": 114460
    },
    {
      "epoch": 2.0096911813760774,
      "grad_norm": 0.05126094509481073,
      "learning_rate": 0.0001346143177463312,
      "loss": 0.8506,
      "step": 114470
    },
    {
      "epoch": 2.009866746256079,
      "grad_norm": 0.04262736938493137,
      "learning_rate": 0.0001346038359201855,
      "loss": 0.8481,
      "step": 114480
    },
    {
      "epoch": 2.0100423111360803,
      "grad_norm": 0.0431808176996709,
      "learning_rate": 0.0001345933536682727,
      "loss": 0.8484,
      "step": 114490
    },
    {
      "epoch": 2.010217876016082,
      "grad_norm": 0.073128948536251,
      "learning_rate": 0.0001345828709907257,
      "loss": 0.8468,
      "step": 114500
    },
    {
      "epoch": 2.0103934408960833,
      "grad_norm": 0.06269174452867857,
      "learning_rate": 0.0001345723878876773,
      "loss": 0.8482,
      "step": 114510
    },
    {
      "epoch": 2.0105690057760848,
      "grad_norm": 0.04559977122566213,
      "learning_rate": 0.00013456190435926036,
      "loss": 0.8514,
      "step": 114520
    },
    {
      "epoch": 2.010744570656086,
      "grad_norm": 0.045154179152875146,
      "learning_rate": 0.0001345514204056076,
      "loss": 0.8419,
      "step": 114530
    },
    {
      "epoch": 2.0109201355360873,
      "grad_norm": 0.052095697519621016,
      "learning_rate": 0.00013454093602685194,
      "loss": 0.8432,
      "step": 114540
    },
    {
      "epoch": 2.0110957004160888,
      "grad_norm": 0.04975314424321105,
      "learning_rate": 0.00013453045122312614,
      "loss": 0.8444,
      "step": 114550
    },
    {
      "epoch": 2.0112712652960902,
      "grad_norm": 0.05745755712307473,
      "learning_rate": 0.00013451996599456314,
      "loss": 0.849,
      "step": 114560
    },
    {
      "epoch": 2.0114468301760917,
      "grad_norm": 0.06201977290482736,
      "learning_rate": 0.00013450948034129576,
      "loss": 0.846,
      "step": 114570
    },
    {
      "epoch": 2.011622395056093,
      "grad_norm": 0.05308793158567267,
      "learning_rate": 0.00013449899426345682,
      "loss": 0.8512,
      "step": 114580
    },
    {
      "epoch": 2.0117979599360942,
      "grad_norm": 0.04944267162345111,
      "learning_rate": 0.00013448850776117918,
      "loss": 0.8506,
      "step": 114590
    },
    {
      "epoch": 2.0119735248160957,
      "grad_norm": 0.04476102678855464,
      "learning_rate": 0.00013447802083459576,
      "loss": 0.8334,
      "step": 114600
    },
    {
      "epoch": 2.012149089696097,
      "grad_norm": 0.05606315520393755,
      "learning_rate": 0.0001344675334838394,
      "loss": 0.8412,
      "step": 114610
    },
    {
      "epoch": 2.0123246545760987,
      "grad_norm": 0.05685905632906982,
      "learning_rate": 0.00013445704570904296,
      "loss": 0.8496,
      "step": 114620
    },
    {
      "epoch": 2.0125002194561,
      "grad_norm": 0.0734451718413507,
      "learning_rate": 0.0001344465575103394,
      "loss": 0.8422,
      "step": 114630
    },
    {
      "epoch": 2.0126757843361016,
      "grad_norm": 0.05473215171140964,
      "learning_rate": 0.00013443606888786146,
      "loss": 0.8556,
      "step": 114640
    },
    {
      "epoch": 2.0128513492161026,
      "grad_norm": 0.05294759787718478,
      "learning_rate": 0.00013442557984174217,
      "loss": 0.8396,
      "step": 114650
    },
    {
      "epoch": 2.013026914096104,
      "grad_norm": 0.05347678201464379,
      "learning_rate": 0.00013441509037211434,
      "loss": 0.8454,
      "step": 114660
    },
    {
      "epoch": 2.0132024789761056,
      "grad_norm": 0.059919025087596445,
      "learning_rate": 0.00013440460047911093,
      "loss": 0.8503,
      "step": 114670
    },
    {
      "epoch": 2.013378043856107,
      "grad_norm": 0.05095874032247942,
      "learning_rate": 0.00013439411016286486,
      "loss": 0.8494,
      "step": 114680
    },
    {
      "epoch": 2.0135536087361086,
      "grad_norm": 0.06607182344166838,
      "learning_rate": 0.00013438361942350898,
      "loss": 0.8465,
      "step": 114690
    },
    {
      "epoch": 2.01372917361611,
      "grad_norm": 0.057785898319845784,
      "learning_rate": 0.00013437312826117625,
      "loss": 0.8526,
      "step": 114700
    },
    {
      "epoch": 2.013904738496111,
      "grad_norm": 0.0781211512870403,
      "learning_rate": 0.00013436263667599959,
      "loss": 0.8561,
      "step": 114710
    },
    {
      "epoch": 2.0140803033761125,
      "grad_norm": 0.06452074092178503,
      "learning_rate": 0.00013435214466811193,
      "loss": 0.8534,
      "step": 114720
    },
    {
      "epoch": 2.014255868256114,
      "grad_norm": 0.04572098735602519,
      "learning_rate": 0.0001343416522376462,
      "loss": 0.8453,
      "step": 114730
    },
    {
      "epoch": 2.0144314331361155,
      "grad_norm": 0.05654720246574256,
      "learning_rate": 0.00013433115938473538,
      "loss": 0.851,
      "step": 114740
    },
    {
      "epoch": 2.014606998016117,
      "grad_norm": 0.05815307110010233,
      "learning_rate": 0.00013432066610951233,
      "loss": 0.855,
      "step": 114750
    },
    {
      "epoch": 2.0147825628961185,
      "grad_norm": 0.07833316220779164,
      "learning_rate": 0.00013431017241211007,
      "loss": 0.8412,
      "step": 114760
    },
    {
      "epoch": 2.0149581277761195,
      "grad_norm": 0.06269227788139098,
      "learning_rate": 0.00013429967829266154,
      "loss": 0.8515,
      "step": 114770
    },
    {
      "epoch": 2.015133692656121,
      "grad_norm": 0.06583459935517852,
      "learning_rate": 0.00013428918375129974,
      "loss": 0.8531,
      "step": 114780
    },
    {
      "epoch": 2.0153092575361224,
      "grad_norm": 0.05590005707062792,
      "learning_rate": 0.00013427868878815758,
      "loss": 0.8491,
      "step": 114790
    },
    {
      "epoch": 2.015484822416124,
      "grad_norm": 0.057171093498978,
      "learning_rate": 0.00013426819340336808,
      "loss": 0.8441,
      "step": 114800
    },
    {
      "epoch": 2.0156603872961254,
      "grad_norm": 0.0673890045774469,
      "learning_rate": 0.00013425769759706414,
      "loss": 0.8466,
      "step": 114810
    },
    {
      "epoch": 2.015835952176127,
      "grad_norm": 0.09249265438384394,
      "learning_rate": 0.00013424720136937882,
      "loss": 0.8437,
      "step": 114820
    },
    {
      "epoch": 2.016011517056128,
      "grad_norm": 0.06066608688458283,
      "learning_rate": 0.00013423670472044507,
      "loss": 0.8519,
      "step": 114830
    },
    {
      "epoch": 2.0161870819361294,
      "grad_norm": 0.06628922226591351,
      "learning_rate": 0.00013422620765039594,
      "loss": 0.8442,
      "step": 114840
    },
    {
      "epoch": 2.016362646816131,
      "grad_norm": 0.06389109337263413,
      "learning_rate": 0.00013421571015936437,
      "loss": 0.8504,
      "step": 114850
    },
    {
      "epoch": 2.0165382116961323,
      "grad_norm": 0.04792761571629373,
      "learning_rate": 0.0001342052122474834,
      "loss": 0.8451,
      "step": 114860
    },
    {
      "epoch": 2.016713776576134,
      "grad_norm": 0.058215515000586075,
      "learning_rate": 0.00013419471391488603,
      "loss": 0.8469,
      "step": 114870
    },
    {
      "epoch": 2.0168893414561353,
      "grad_norm": 0.053044099058293624,
      "learning_rate": 0.00013418421516170524,
      "loss": 0.8491,
      "step": 114880
    },
    {
      "epoch": 2.0170649063361363,
      "grad_norm": 0.04929495799488826,
      "learning_rate": 0.0001341737159880741,
      "loss": 0.8495,
      "step": 114890
    },
    {
      "epoch": 2.017240471216138,
      "grad_norm": 0.06880087881004598,
      "learning_rate": 0.00013416321639412562,
      "loss": 0.8439,
      "step": 114900
    },
    {
      "epoch": 2.0174160360961393,
      "grad_norm": 0.052465731460408206,
      "learning_rate": 0.00013415271637999286,
      "loss": 0.8452,
      "step": 114910
    },
    {
      "epoch": 2.0175916009761408,
      "grad_norm": 0.0665734467835106,
      "learning_rate": 0.00013414221594580883,
      "loss": 0.8422,
      "step": 114920
    },
    {
      "epoch": 2.0177671658561422,
      "grad_norm": 0.06740423812929837,
      "learning_rate": 0.00013413171509170653,
      "loss": 0.8406,
      "step": 114930
    },
    {
      "epoch": 2.0179427307361437,
      "grad_norm": 0.062172901040825145,
      "learning_rate": 0.00013412121381781904,
      "loss": 0.8426,
      "step": 114940
    },
    {
      "epoch": 2.0181182956161448,
      "grad_norm": 0.08665204389214046,
      "learning_rate": 0.00013411071212427947,
      "loss": 0.8462,
      "step": 114950
    },
    {
      "epoch": 2.0182938604961462,
      "grad_norm": 0.055718351648822356,
      "learning_rate": 0.0001341002100112208,
      "loss": 0.8503,
      "step": 114960
    },
    {
      "epoch": 2.0184694253761477,
      "grad_norm": 0.06020372985501083,
      "learning_rate": 0.00013408970747877616,
      "loss": 0.846,
      "step": 114970
    },
    {
      "epoch": 2.018644990256149,
      "grad_norm": 0.053606728240258604,
      "learning_rate": 0.00013407920452707857,
      "loss": 0.8512,
      "step": 114980
    },
    {
      "epoch": 2.0188205551361507,
      "grad_norm": 0.06222561441000506,
      "learning_rate": 0.0001340687011562611,
      "loss": 0.8428,
      "step": 114990
    },
    {
      "epoch": 2.018996120016152,
      "grad_norm": 0.05435892529729845,
      "learning_rate": 0.00013405819736645686,
      "loss": 0.8489,
      "step": 115000
    },
    {
      "epoch": 2.019171684896153,
      "grad_norm": 0.04496920930288166,
      "learning_rate": 0.0001340476931577989,
      "loss": 0.8514,
      "step": 115010
    },
    {
      "epoch": 2.0193472497761547,
      "grad_norm": 0.06001170005726854,
      "learning_rate": 0.00013403718853042038,
      "loss": 0.8467,
      "step": 115020
    },
    {
      "epoch": 2.019522814656156,
      "grad_norm": 0.05831129050299066,
      "learning_rate": 0.00013402668348445432,
      "loss": 0.8475,
      "step": 115030
    },
    {
      "epoch": 2.0196983795361576,
      "grad_norm": 0.046602917701931985,
      "learning_rate": 0.00013401617802003387,
      "loss": 0.8427,
      "step": 115040
    },
    {
      "epoch": 2.019873944416159,
      "grad_norm": 0.054032970087813804,
      "learning_rate": 0.00013400567213729211,
      "loss": 0.8448,
      "step": 115050
    },
    {
      "epoch": 2.0200495092961606,
      "grad_norm": 0.058670470769464224,
      "learning_rate": 0.00013399516583636215,
      "loss": 0.8419,
      "step": 115060
    },
    {
      "epoch": 2.0202250741761616,
      "grad_norm": 0.07714800254442357,
      "learning_rate": 0.00013398465911737713,
      "loss": 0.8507,
      "step": 115070
    },
    {
      "epoch": 2.020400639056163,
      "grad_norm": 0.05243889645617048,
      "learning_rate": 0.00013397415198047018,
      "loss": 0.846,
      "step": 115080
    },
    {
      "epoch": 2.0205762039361645,
      "grad_norm": 0.05485107780486826,
      "learning_rate": 0.00013396364442577434,
      "loss": 0.8471,
      "step": 115090
    },
    {
      "epoch": 2.020751768816166,
      "grad_norm": 0.06190259203889378,
      "learning_rate": 0.00013395313645342286,
      "loss": 0.8487,
      "step": 115100
    },
    {
      "epoch": 2.0209273336961675,
      "grad_norm": 0.04696627911334669,
      "learning_rate": 0.0001339426280635488,
      "loss": 0.8485,
      "step": 115110
    },
    {
      "epoch": 2.021102898576169,
      "grad_norm": 0.08685885702083879,
      "learning_rate": 0.00013393211925628534,
      "loss": 0.8491,
      "step": 115120
    },
    {
      "epoch": 2.02127846345617,
      "grad_norm": 0.05454289342799106,
      "learning_rate": 0.0001339216100317656,
      "loss": 0.8387,
      "step": 115130
    },
    {
      "epoch": 2.0214540283361715,
      "grad_norm": 0.05337349083695242,
      "learning_rate": 0.0001339111003901228,
      "loss": 0.8476,
      "step": 115140
    },
    {
      "epoch": 2.021629593216173,
      "grad_norm": 0.058476467604367095,
      "learning_rate": 0.00013390059033149,
      "loss": 0.8471,
      "step": 115150
    },
    {
      "epoch": 2.0218051580961744,
      "grad_norm": 0.05058513493130815,
      "learning_rate": 0.00013389007985600043,
      "loss": 0.8451,
      "step": 115160
    },
    {
      "epoch": 2.021980722976176,
      "grad_norm": 0.06623775978297361,
      "learning_rate": 0.00013387956896378721,
      "loss": 0.8456,
      "step": 115170
    },
    {
      "epoch": 2.0221562878561774,
      "grad_norm": 0.05972348019226066,
      "learning_rate": 0.0001338690576549836,
      "loss": 0.8483,
      "step": 115180
    },
    {
      "epoch": 2.0223318527361784,
      "grad_norm": 0.05722414462023936,
      "learning_rate": 0.00013385854592972273,
      "loss": 0.8415,
      "step": 115190
    },
    {
      "epoch": 2.02250741761618,
      "grad_norm": 0.074708931394996,
      "learning_rate": 0.00013384803378813778,
      "loss": 0.8509,
      "step": 115200
    },
    {
      "epoch": 2.0226829824961814,
      "grad_norm": 0.05985430729877331,
      "learning_rate": 0.00013383752123036194,
      "loss": 0.8482,
      "step": 115210
    },
    {
      "epoch": 2.022858547376183,
      "grad_norm": 0.06988622583481423,
      "learning_rate": 0.00013382700825652837,
      "loss": 0.8533,
      "step": 115220
    },
    {
      "epoch": 2.0230341122561843,
      "grad_norm": 0.06480394334015178,
      "learning_rate": 0.00013381649486677037,
      "loss": 0.8475,
      "step": 115230
    },
    {
      "epoch": 2.023209677136186,
      "grad_norm": 0.0560321081684587,
      "learning_rate": 0.0001338059810612211,
      "loss": 0.8483,
      "step": 115240
    },
    {
      "epoch": 2.0233852420161873,
      "grad_norm": 0.06913458446008973,
      "learning_rate": 0.00013379546684001375,
      "loss": 0.8447,
      "step": 115250
    },
    {
      "epoch": 2.0235608068961883,
      "grad_norm": 0.06700175372101036,
      "learning_rate": 0.00013378495220328152,
      "loss": 0.8415,
      "step": 115260
    },
    {
      "epoch": 2.02373637177619,
      "grad_norm": 0.05873468343397788,
      "learning_rate": 0.00013377443715115766,
      "loss": 0.84,
      "step": 115270
    },
    {
      "epoch": 2.0239119366561913,
      "grad_norm": 0.054762037451959705,
      "learning_rate": 0.00013376392168377546,
      "loss": 0.8397,
      "step": 115280
    },
    {
      "epoch": 2.0240875015361928,
      "grad_norm": 0.04676265574368405,
      "learning_rate": 0.00013375340580126804,
      "loss": 0.842,
      "step": 115290
    },
    {
      "epoch": 2.0242630664161942,
      "grad_norm": 0.055341930259218144,
      "learning_rate": 0.0001337428895037687,
      "loss": 0.8511,
      "step": 115300
    },
    {
      "epoch": 2.0244386312961957,
      "grad_norm": 0.047587102510171564,
      "learning_rate": 0.00013373237279141068,
      "loss": 0.8459,
      "step": 115310
    },
    {
      "epoch": 2.0246141961761968,
      "grad_norm": 0.05659087013807943,
      "learning_rate": 0.00013372185566432724,
      "loss": 0.8543,
      "step": 115320
    },
    {
      "epoch": 2.0247897610561982,
      "grad_norm": 0.048208045573658355,
      "learning_rate": 0.0001337113381226516,
      "loss": 0.8456,
      "step": 115330
    },
    {
      "epoch": 2.0249653259361997,
      "grad_norm": 0.04919676323508185,
      "learning_rate": 0.00013370082016651703,
      "loss": 0.8408,
      "step": 115340
    },
    {
      "epoch": 2.025140890816201,
      "grad_norm": 0.05787053045918171,
      "learning_rate": 0.00013369030179605676,
      "loss": 0.8383,
      "step": 115350
    },
    {
      "epoch": 2.0253164556962027,
      "grad_norm": 0.04660183490528461,
      "learning_rate": 0.0001336797830114042,
      "loss": 0.8416,
      "step": 115360
    },
    {
      "epoch": 2.025492020576204,
      "grad_norm": 0.05928035975055669,
      "learning_rate": 0.00013366926381269247,
      "loss": 0.8445,
      "step": 115370
    },
    {
      "epoch": 2.025667585456205,
      "grad_norm": 0.05282926019146055,
      "learning_rate": 0.00013365874420005492,
      "loss": 0.8497,
      "step": 115380
    },
    {
      "epoch": 2.0258431503362067,
      "grad_norm": 0.04811007436059055,
      "learning_rate": 0.0001336482241736248,
      "loss": 0.8467,
      "step": 115390
    },
    {
      "epoch": 2.026018715216208,
      "grad_norm": 0.04749536452083036,
      "learning_rate": 0.00013363770373353546,
      "loss": 0.8437,
      "step": 115400
    },
    {
      "epoch": 2.0261942800962096,
      "grad_norm": 0.06518996281320225,
      "learning_rate": 0.00013362718287992012,
      "loss": 0.847,
      "step": 115410
    },
    {
      "epoch": 2.026369844976211,
      "grad_norm": 0.059305433976548284,
      "learning_rate": 0.00013361666161291216,
      "loss": 0.8502,
      "step": 115420
    },
    {
      "epoch": 2.0265454098562126,
      "grad_norm": 0.04821230808203083,
      "learning_rate": 0.00013360613993264483,
      "loss": 0.8379,
      "step": 115430
    },
    {
      "epoch": 2.0267209747362136,
      "grad_norm": 0.06640324361371842,
      "learning_rate": 0.00013359561783925146,
      "loss": 0.8429,
      "step": 115440
    },
    {
      "epoch": 2.026896539616215,
      "grad_norm": 0.05189061545948313,
      "learning_rate": 0.00013358509533286533,
      "loss": 0.8438,
      "step": 115450
    },
    {
      "epoch": 2.0270721044962166,
      "grad_norm": 0.06265166799819974,
      "learning_rate": 0.0001335745724136198,
      "loss": 0.8512,
      "step": 115460
    },
    {
      "epoch": 2.027247669376218,
      "grad_norm": 0.05987172363399539,
      "learning_rate": 0.00013356404908164825,
      "loss": 0.8386,
      "step": 115470
    },
    {
      "epoch": 2.0274232342562195,
      "grad_norm": 0.06343860259308753,
      "learning_rate": 0.00013355352533708393,
      "loss": 0.8378,
      "step": 115480
    },
    {
      "epoch": 2.027598799136221,
      "grad_norm": 0.05248691689378751,
      "learning_rate": 0.0001335430011800602,
      "loss": 0.8473,
      "step": 115490
    },
    {
      "epoch": 2.027774364016222,
      "grad_norm": 0.06930778911479617,
      "learning_rate": 0.00013353247661071043,
      "loss": 0.8448,
      "step": 115500
    },
    {
      "epoch": 2.0279499288962235,
      "grad_norm": 0.07012236197470272,
      "learning_rate": 0.0001335219516291679,
      "loss": 0.8513,
      "step": 115510
    },
    {
      "epoch": 2.028125493776225,
      "grad_norm": 0.0721906435922019,
      "learning_rate": 0.000133511426235566,
      "loss": 0.8435,
      "step": 115520
    },
    {
      "epoch": 2.0283010586562265,
      "grad_norm": 0.07284340100786416,
      "learning_rate": 0.00013350090043003817,
      "loss": 0.8455,
      "step": 115530
    },
    {
      "epoch": 2.028476623536228,
      "grad_norm": 0.06104836535884297,
      "learning_rate": 0.0001334903742127177,
      "loss": 0.8493,
      "step": 115540
    },
    {
      "epoch": 2.0286521884162294,
      "grad_norm": 0.04364158793472262,
      "learning_rate": 0.0001334798475837379,
      "loss": 0.8482,
      "step": 115550
    },
    {
      "epoch": 2.0288277532962304,
      "grad_norm": 0.050887870211579075,
      "learning_rate": 0.00013346932054323222,
      "loss": 0.8412,
      "step": 115560
    },
    {
      "epoch": 2.029003318176232,
      "grad_norm": 0.04773907218193663,
      "learning_rate": 0.00013345879309133404,
      "loss": 0.8454,
      "step": 115570
    },
    {
      "epoch": 2.0291788830562334,
      "grad_norm": 0.05345660103102053,
      "learning_rate": 0.00013344826522817673,
      "loss": 0.841,
      "step": 115580
    },
    {
      "epoch": 2.029354447936235,
      "grad_norm": 0.06247476754439574,
      "learning_rate": 0.0001334377369538937,
      "loss": 0.847,
      "step": 115590
    },
    {
      "epoch": 2.0295300128162364,
      "grad_norm": 0.062369824021922225,
      "learning_rate": 0.00013342720826861832,
      "loss": 0.8448,
      "step": 115600
    },
    {
      "epoch": 2.029705577696238,
      "grad_norm": 0.04840204781153707,
      "learning_rate": 0.00013341667917248398,
      "loss": 0.8525,
      "step": 115610
    },
    {
      "epoch": 2.029881142576239,
      "grad_norm": 0.04772455444270228,
      "learning_rate": 0.00013340614966562408,
      "loss": 0.8465,
      "step": 115620
    },
    {
      "epoch": 2.0300567074562403,
      "grad_norm": 0.056017273502248825,
      "learning_rate": 0.0001333956197481721,
      "loss": 0.8492,
      "step": 115630
    },
    {
      "epoch": 2.030232272336242,
      "grad_norm": 0.046453222512950774,
      "learning_rate": 0.00013338508942026137,
      "loss": 0.8562,
      "step": 115640
    },
    {
      "epoch": 2.0304078372162433,
      "grad_norm": 0.0483932539119913,
      "learning_rate": 0.0001333745586820254,
      "loss": 0.8459,
      "step": 115650
    },
    {
      "epoch": 2.0305834020962448,
      "grad_norm": 0.05727429214033881,
      "learning_rate": 0.00013336402753359755,
      "loss": 0.8497,
      "step": 115660
    },
    {
      "epoch": 2.0307589669762462,
      "grad_norm": 0.058478551270248265,
      "learning_rate": 0.00013335349597511128,
      "loss": 0.8454,
      "step": 115670
    },
    {
      "epoch": 2.0309345318562473,
      "grad_norm": 0.08264964847794652,
      "learning_rate": 0.00013334296400669996,
      "loss": 0.8488,
      "step": 115680
    },
    {
      "epoch": 2.0311100967362488,
      "grad_norm": 0.06769458548360013,
      "learning_rate": 0.0001333324316284971,
      "loss": 0.8442,
      "step": 115690
    },
    {
      "epoch": 2.0312856616162502,
      "grad_norm": 0.06123485408426817,
      "learning_rate": 0.00013332189884063616,
      "loss": 0.8426,
      "step": 115700
    },
    {
      "epoch": 2.0314612264962517,
      "grad_norm": 0.04830995500866124,
      "learning_rate": 0.00013331136564325057,
      "loss": 0.8587,
      "step": 115710
    },
    {
      "epoch": 2.031636791376253,
      "grad_norm": 0.05807789595576253,
      "learning_rate": 0.00013330083203647378,
      "loss": 0.8363,
      "step": 115720
    },
    {
      "epoch": 2.0318123562562547,
      "grad_norm": 0.06291654448699847,
      "learning_rate": 0.00013329029802043924,
      "loss": 0.8431,
      "step": 115730
    },
    {
      "epoch": 2.0319879211362557,
      "grad_norm": 0.0554771292764982,
      "learning_rate": 0.00013327976359528047,
      "loss": 0.849,
      "step": 115740
    },
    {
      "epoch": 2.032163486016257,
      "grad_norm": 0.06242007315761199,
      "learning_rate": 0.00013326922876113088,
      "loss": 0.844,
      "step": 115750
    },
    {
      "epoch": 2.0323390508962587,
      "grad_norm": 0.07609717724976298,
      "learning_rate": 0.00013325869351812398,
      "loss": 0.8552,
      "step": 115760
    },
    {
      "epoch": 2.03251461577626,
      "grad_norm": 0.05656908145305979,
      "learning_rate": 0.00013324815786639328,
      "loss": 0.8519,
      "step": 115770
    },
    {
      "epoch": 2.0326901806562616,
      "grad_norm": 0.059050945524467484,
      "learning_rate": 0.00013323762180607223,
      "loss": 0.8488,
      "step": 115780
    },
    {
      "epoch": 2.032865745536263,
      "grad_norm": 0.05437253578124569,
      "learning_rate": 0.0001332270853372943,
      "loss": 0.8491,
      "step": 115790
    },
    {
      "epoch": 2.033041310416264,
      "grad_norm": 0.05508091614683385,
      "learning_rate": 0.00013321654846019305,
      "loss": 0.8403,
      "step": 115800
    },
    {
      "epoch": 2.0332168752962656,
      "grad_norm": 0.04808424088507107,
      "learning_rate": 0.0001332060111749019,
      "loss": 0.8454,
      "step": 115810
    },
    {
      "epoch": 2.033392440176267,
      "grad_norm": 0.06110371474594997,
      "learning_rate": 0.00013319547348155452,
      "loss": 0.8516,
      "step": 115820
    },
    {
      "epoch": 2.0335680050562686,
      "grad_norm": 0.08323995711292181,
      "learning_rate": 0.00013318493538028425,
      "loss": 0.8467,
      "step": 115830
    },
    {
      "epoch": 2.03374356993627,
      "grad_norm": 0.049480264693308275,
      "learning_rate": 0.0001331743968712247,
      "loss": 0.8509,
      "step": 115840
    },
    {
      "epoch": 2.0339191348162715,
      "grad_norm": 0.05607720482521311,
      "learning_rate": 0.00013316385795450938,
      "loss": 0.8524,
      "step": 115850
    },
    {
      "epoch": 2.0340946996962725,
      "grad_norm": 0.04698563640020885,
      "learning_rate": 0.00013315331863027178,
      "loss": 0.8456,
      "step": 115860
    },
    {
      "epoch": 2.034270264576274,
      "grad_norm": 0.05841459968649828,
      "learning_rate": 0.0001331427788986455,
      "loss": 0.8395,
      "step": 115870
    },
    {
      "epoch": 2.0344458294562755,
      "grad_norm": 0.05181330525803007,
      "learning_rate": 0.00013313223875976407,
      "loss": 0.8414,
      "step": 115880
    },
    {
      "epoch": 2.034621394336277,
      "grad_norm": 0.05583067659385761,
      "learning_rate": 0.000133121698213761,
      "loss": 0.8486,
      "step": 115890
    },
    {
      "epoch": 2.0347969592162785,
      "grad_norm": 0.051815545555068024,
      "learning_rate": 0.00013311115726076983,
      "loss": 0.8424,
      "step": 115900
    },
    {
      "epoch": 2.03497252409628,
      "grad_norm": 0.07933082776830837,
      "learning_rate": 0.00013310061590092418,
      "loss": 0.851,
      "step": 115910
    },
    {
      "epoch": 2.035148088976281,
      "grad_norm": 0.048959342121175954,
      "learning_rate": 0.00013309007413435755,
      "loss": 0.8453,
      "step": 115920
    },
    {
      "epoch": 2.0353236538562824,
      "grad_norm": 0.06601770264334351,
      "learning_rate": 0.00013307953196120355,
      "loss": 0.8411,
      "step": 115930
    },
    {
      "epoch": 2.035499218736284,
      "grad_norm": 0.057703558929822395,
      "learning_rate": 0.00013306898938159574,
      "loss": 0.8463,
      "step": 115940
    },
    {
      "epoch": 2.0356747836162854,
      "grad_norm": 0.05055575743545241,
      "learning_rate": 0.00013305844639566766,
      "loss": 0.8475,
      "step": 115950
    },
    {
      "epoch": 2.035850348496287,
      "grad_norm": 0.06440967136082601,
      "learning_rate": 0.00013304790300355294,
      "loss": 0.8466,
      "step": 115960
    },
    {
      "epoch": 2.0360259133762884,
      "grad_norm": 0.0534708983795741,
      "learning_rate": 0.0001330373592053851,
      "loss": 0.8487,
      "step": 115970
    },
    {
      "epoch": 2.03620147825629,
      "grad_norm": 0.07710463290540008,
      "learning_rate": 0.00013302681500129782,
      "loss": 0.8386,
      "step": 115980
    },
    {
      "epoch": 2.036377043136291,
      "grad_norm": 0.06552634507337075,
      "learning_rate": 0.0001330162703914247,
      "loss": 0.8446,
      "step": 115990
    },
    {
      "epoch": 2.0365526080162923,
      "grad_norm": 0.05114791020433911,
      "learning_rate": 0.00013300572537589924,
      "loss": 0.8449,
      "step": 116000
    },
    {
      "epoch": 2.036728172896294,
      "grad_norm": 0.06631196863416539,
      "learning_rate": 0.00013299517995485512,
      "loss": 0.8479,
      "step": 116010
    },
    {
      "epoch": 2.0369037377762953,
      "grad_norm": 0.07016295146607304,
      "learning_rate": 0.00013298463412842596,
      "loss": 0.8493,
      "step": 116020
    },
    {
      "epoch": 2.0370793026562968,
      "grad_norm": 0.08126247624912136,
      "learning_rate": 0.0001329740878967453,
      "loss": 0.8469,
      "step": 116030
    },
    {
      "epoch": 2.0372548675362983,
      "grad_norm": 0.047751456064162326,
      "learning_rate": 0.00013296354125994685,
      "loss": 0.8482,
      "step": 116040
    },
    {
      "epoch": 2.0374304324162993,
      "grad_norm": 0.06139718837707965,
      "learning_rate": 0.00013295299421816421,
      "loss": 0.8519,
      "step": 116050
    },
    {
      "epoch": 2.0376059972963008,
      "grad_norm": 0.057790784702353666,
      "learning_rate": 0.00013294244677153105,
      "loss": 0.8426,
      "step": 116060
    },
    {
      "epoch": 2.0377815621763022,
      "grad_norm": 0.06385447157295256,
      "learning_rate": 0.00013293189892018086,
      "loss": 0.8419,
      "step": 116070
    },
    {
      "epoch": 2.0379571270563037,
      "grad_norm": 0.052351381233236446,
      "learning_rate": 0.0001329213506642475,
      "loss": 0.846,
      "step": 116080
    },
    {
      "epoch": 2.038132691936305,
      "grad_norm": 0.05053340485579395,
      "learning_rate": 0.00013291080200386444,
      "loss": 0.8421,
      "step": 116090
    },
    {
      "epoch": 2.0383082568163067,
      "grad_norm": 0.04618696882272358,
      "learning_rate": 0.00013290025293916547,
      "loss": 0.8449,
      "step": 116100
    },
    {
      "epoch": 2.0384838216963077,
      "grad_norm": 0.046102424827990804,
      "learning_rate": 0.00013288970347028413,
      "loss": 0.8395,
      "step": 116110
    },
    {
      "epoch": 2.038659386576309,
      "grad_norm": 0.05917354650635874,
      "learning_rate": 0.00013287915359735415,
      "loss": 0.8426,
      "step": 116120
    },
    {
      "epoch": 2.0388349514563107,
      "grad_norm": 0.05319090158827418,
      "learning_rate": 0.00013286860332050915,
      "loss": 0.8448,
      "step": 116130
    },
    {
      "epoch": 2.039010516336312,
      "grad_norm": 0.043565520765816095,
      "learning_rate": 0.00013285805263988287,
      "loss": 0.8531,
      "step": 116140
    },
    {
      "epoch": 2.0391860812163136,
      "grad_norm": 0.06266152075779538,
      "learning_rate": 0.00013284750155560893,
      "loss": 0.8527,
      "step": 116150
    },
    {
      "epoch": 2.039361646096315,
      "grad_norm": 0.05447159353044122,
      "learning_rate": 0.0001328369500678211,
      "loss": 0.8455,
      "step": 116160
    },
    {
      "epoch": 2.039537210976316,
      "grad_norm": 0.05460489749810612,
      "learning_rate": 0.00013282639817665298,
      "loss": 0.8416,
      "step": 116170
    },
    {
      "epoch": 2.0397127758563176,
      "grad_norm": 0.0660507308029412,
      "learning_rate": 0.0001328158458822383,
      "loss": 0.8476,
      "step": 116180
    },
    {
      "epoch": 2.039888340736319,
      "grad_norm": 0.07031905587848508,
      "learning_rate": 0.00013280529318471072,
      "loss": 0.8414,
      "step": 116190
    },
    {
      "epoch": 2.0400639056163206,
      "grad_norm": 0.07107260916300454,
      "learning_rate": 0.000132794740084204,
      "loss": 0.8501,
      "step": 116200
    },
    {
      "epoch": 2.040239470496322,
      "grad_norm": 0.055892353157972516,
      "learning_rate": 0.00013278418658085186,
      "loss": 0.8474,
      "step": 116210
    },
    {
      "epoch": 2.0404150353763235,
      "grad_norm": 0.05555582160874119,
      "learning_rate": 0.00013277363267478798,
      "loss": 0.835,
      "step": 116220
    },
    {
      "epoch": 2.0405906002563245,
      "grad_norm": 0.04455648663860785,
      "learning_rate": 0.00013276307836614608,
      "loss": 0.8621,
      "step": 116230
    },
    {
      "epoch": 2.040766165136326,
      "grad_norm": 0.10316197217055278,
      "learning_rate": 0.00013275252365505983,
      "loss": 0.8434,
      "step": 116240
    },
    {
      "epoch": 2.0409417300163275,
      "grad_norm": 0.057645345104377375,
      "learning_rate": 0.00013274196854166307,
      "loss": 0.8499,
      "step": 116250
    },
    {
      "epoch": 2.041117294896329,
      "grad_norm": 0.06191808016936995,
      "learning_rate": 0.0001327314130260895,
      "loss": 0.8421,
      "step": 116260
    },
    {
      "epoch": 2.0412928597763305,
      "grad_norm": 0.05573585853557654,
      "learning_rate": 0.00013272085710847284,
      "loss": 0.85,
      "step": 116270
    },
    {
      "epoch": 2.041468424656332,
      "grad_norm": 0.06133067029503345,
      "learning_rate": 0.00013271030078894683,
      "loss": 0.8478,
      "step": 116280
    },
    {
      "epoch": 2.041643989536333,
      "grad_norm": 0.05296574885081548,
      "learning_rate": 0.00013269974406764524,
      "loss": 0.8484,
      "step": 116290
    },
    {
      "epoch": 2.0418195544163344,
      "grad_norm": 0.05401344316572221,
      "learning_rate": 0.00013268918694470181,
      "loss": 0.8498,
      "step": 116300
    },
    {
      "epoch": 2.041995119296336,
      "grad_norm": 0.06569200524393778,
      "learning_rate": 0.0001326786294202503,
      "loss": 0.8473,
      "step": 116310
    },
    {
      "epoch": 2.0421706841763374,
      "grad_norm": 0.05916541825529002,
      "learning_rate": 0.0001326680714944245,
      "loss": 0.8494,
      "step": 116320
    },
    {
      "epoch": 2.042346249056339,
      "grad_norm": 0.057139597819307886,
      "learning_rate": 0.00013265751316735816,
      "loss": 0.8437,
      "step": 116330
    },
    {
      "epoch": 2.0425218139363404,
      "grad_norm": 0.04955425771372743,
      "learning_rate": 0.00013264695443918504,
      "loss": 0.8453,
      "step": 116340
    },
    {
      "epoch": 2.0426973788163414,
      "grad_norm": 0.04899030553920241,
      "learning_rate": 0.00013263639531003898,
      "loss": 0.8493,
      "step": 116350
    },
    {
      "epoch": 2.042872943696343,
      "grad_norm": 0.048653393954722284,
      "learning_rate": 0.00013262583578005368,
      "loss": 0.841,
      "step": 116360
    },
    {
      "epoch": 2.0430485085763443,
      "grad_norm": 0.05621177560871276,
      "learning_rate": 0.000132615275849363,
      "loss": 0.8521,
      "step": 116370
    },
    {
      "epoch": 2.043224073456346,
      "grad_norm": 0.07844356908001517,
      "learning_rate": 0.00013260471551810072,
      "loss": 0.8534,
      "step": 116380
    },
    {
      "epoch": 2.0433996383363473,
      "grad_norm": 0.07609001094324307,
      "learning_rate": 0.00013259415478640067,
      "loss": 0.8458,
      "step": 116390
    },
    {
      "epoch": 2.043575203216349,
      "grad_norm": 0.09192768634504529,
      "learning_rate": 0.00013258359365439658,
      "loss": 0.8503,
      "step": 116400
    },
    {
      "epoch": 2.04375076809635,
      "grad_norm": 0.07695274918346011,
      "learning_rate": 0.00013257303212222234,
      "loss": 0.8479,
      "step": 116410
    },
    {
      "epoch": 2.0439263329763513,
      "grad_norm": 0.06288379811974534,
      "learning_rate": 0.00013256247019001168,
      "loss": 0.8464,
      "step": 116420
    },
    {
      "epoch": 2.0441018978563528,
      "grad_norm": 0.04121207161087905,
      "learning_rate": 0.00013255190785789852,
      "loss": 0.8507,
      "step": 116430
    },
    {
      "epoch": 2.0442774627363542,
      "grad_norm": 0.04861349478423367,
      "learning_rate": 0.00013254134512601662,
      "loss": 0.8414,
      "step": 116440
    },
    {
      "epoch": 2.0444530276163557,
      "grad_norm": 0.06151609859613146,
      "learning_rate": 0.00013253078199449987,
      "loss": 0.8436,
      "step": 116450
    },
    {
      "epoch": 2.044628592496357,
      "grad_norm": 0.06083081639432223,
      "learning_rate": 0.00013252021846348202,
      "loss": 0.8499,
      "step": 116460
    },
    {
      "epoch": 2.0448041573763582,
      "grad_norm": 0.06300657262902802,
      "learning_rate": 0.000132509654533097,
      "loss": 0.8431,
      "step": 116470
    },
    {
      "epoch": 2.0449797222563597,
      "grad_norm": 0.054216995482307365,
      "learning_rate": 0.00013249909020347856,
      "loss": 0.8506,
      "step": 116480
    },
    {
      "epoch": 2.045155287136361,
      "grad_norm": 0.0643038824707139,
      "learning_rate": 0.00013248852547476065,
      "loss": 0.8369,
      "step": 116490
    },
    {
      "epoch": 2.0453308520163627,
      "grad_norm": 0.05644508359568222,
      "learning_rate": 0.0001324779603470771,
      "loss": 0.8473,
      "step": 116500
    },
    {
      "epoch": 2.045506416896364,
      "grad_norm": 0.05398078839871528,
      "learning_rate": 0.00013246739482056173,
      "loss": 0.8376,
      "step": 116510
    },
    {
      "epoch": 2.0456819817763656,
      "grad_norm": 0.051990521889777305,
      "learning_rate": 0.0001324568288953485,
      "loss": 0.8411,
      "step": 116520
    },
    {
      "epoch": 2.0458575466563667,
      "grad_norm": 0.08074660320320019,
      "learning_rate": 0.00013244626257157113,
      "loss": 0.85,
      "step": 116530
    },
    {
      "epoch": 2.046033111536368,
      "grad_norm": 0.06144587869852296,
      "learning_rate": 0.00013243569584936362,
      "loss": 0.8437,
      "step": 116540
    },
    {
      "epoch": 2.0462086764163696,
      "grad_norm": 0.06124286058685424,
      "learning_rate": 0.00013242512872885983,
      "loss": 0.8426,
      "step": 116550
    },
    {
      "epoch": 2.046384241296371,
      "grad_norm": 0.04826526630618794,
      "learning_rate": 0.00013241456121019366,
      "loss": 0.8481,
      "step": 116560
    },
    {
      "epoch": 2.0465598061763726,
      "grad_norm": 0.05625447415279807,
      "learning_rate": 0.00013240399329349895,
      "loss": 0.8478,
      "step": 116570
    },
    {
      "epoch": 2.046735371056374,
      "grad_norm": 0.04265975327780129,
      "learning_rate": 0.0001323934249789096,
      "loss": 0.8553,
      "step": 116580
    },
    {
      "epoch": 2.046910935936375,
      "grad_norm": 0.05077316213276536,
      "learning_rate": 0.00013238285626655956,
      "loss": 0.851,
      "step": 116590
    },
    {
      "epoch": 2.0470865008163766,
      "grad_norm": 0.06188178339830002,
      "learning_rate": 0.00013237228715658273,
      "loss": 0.8497,
      "step": 116600
    },
    {
      "epoch": 2.047262065696378,
      "grad_norm": 0.057509419825723405,
      "learning_rate": 0.00013236171764911302,
      "loss": 0.8448,
      "step": 116610
    },
    {
      "epoch": 2.0474376305763795,
      "grad_norm": 0.052022263716679305,
      "learning_rate": 0.00013235114774428434,
      "loss": 0.8516,
      "step": 116620
    },
    {
      "epoch": 2.047613195456381,
      "grad_norm": 0.09215195672826766,
      "learning_rate": 0.00013234057744223058,
      "loss": 0.8398,
      "step": 116630
    },
    {
      "epoch": 2.0477887603363825,
      "grad_norm": 0.05132276002732203,
      "learning_rate": 0.0001323300067430857,
      "loss": 0.8446,
      "step": 116640
    },
    {
      "epoch": 2.0479643252163835,
      "grad_norm": 0.06338349255615106,
      "learning_rate": 0.00013231943564698365,
      "loss": 0.8484,
      "step": 116650
    },
    {
      "epoch": 2.048139890096385,
      "grad_norm": 0.08014154596627829,
      "learning_rate": 0.0001323088641540583,
      "loss": 0.8425,
      "step": 116660
    },
    {
      "epoch": 2.0483154549763865,
      "grad_norm": 0.04420946453317965,
      "learning_rate": 0.0001322982922644437,
      "loss": 0.8482,
      "step": 116670
    },
    {
      "epoch": 2.048491019856388,
      "grad_norm": 0.05606991624814584,
      "learning_rate": 0.00013228771997827373,
      "loss": 0.8512,
      "step": 116680
    },
    {
      "epoch": 2.0486665847363894,
      "grad_norm": 0.05411675367312648,
      "learning_rate": 0.00013227714729568236,
      "loss": 0.8465,
      "step": 116690
    },
    {
      "epoch": 2.048842149616391,
      "grad_norm": 0.048158042863548425,
      "learning_rate": 0.0001322665742168035,
      "loss": 0.849,
      "step": 116700
    },
    {
      "epoch": 2.049017714496392,
      "grad_norm": 0.06056048117599479,
      "learning_rate": 0.00013225600074177112,
      "loss": 0.8585,
      "step": 116710
    },
    {
      "epoch": 2.0491932793763934,
      "grad_norm": 0.06830059597440585,
      "learning_rate": 0.0001322454268707193,
      "loss": 0.8454,
      "step": 116720
    },
    {
      "epoch": 2.049368844256395,
      "grad_norm": 0.05409421175982465,
      "learning_rate": 0.00013223485260378188,
      "loss": 0.8462,
      "step": 116730
    },
    {
      "epoch": 2.0495444091363964,
      "grad_norm": 0.04800039165187589,
      "learning_rate": 0.00013222427794109292,
      "loss": 0.8513,
      "step": 116740
    },
    {
      "epoch": 2.049719974016398,
      "grad_norm": 0.053660159594438504,
      "learning_rate": 0.00013221370288278637,
      "loss": 0.8557,
      "step": 116750
    },
    {
      "epoch": 2.0498955388963993,
      "grad_norm": 0.05242607072412516,
      "learning_rate": 0.0001322031274289962,
      "loss": 0.843,
      "step": 116760
    },
    {
      "epoch": 2.050071103776401,
      "grad_norm": 0.05094070408924866,
      "learning_rate": 0.0001321925515798565,
      "loss": 0.8457,
      "step": 116770
    },
    {
      "epoch": 2.050246668656402,
      "grad_norm": 0.06020719097212672,
      "learning_rate": 0.00013218197533550113,
      "loss": 0.8454,
      "step": 116780
    },
    {
      "epoch": 2.0504222335364033,
      "grad_norm": 0.0845442361807617,
      "learning_rate": 0.00013217139869606417,
      "loss": 0.8477,
      "step": 116790
    },
    {
      "epoch": 2.0505977984164048,
      "grad_norm": 0.0617076540077912,
      "learning_rate": 0.00013216082166167957,
      "loss": 0.8504,
      "step": 116800
    },
    {
      "epoch": 2.0507733632964062,
      "grad_norm": 0.07895446693153199,
      "learning_rate": 0.00013215024423248145,
      "loss": 0.8431,
      "step": 116810
    },
    {
      "epoch": 2.0509489281764077,
      "grad_norm": 0.05019213213758843,
      "learning_rate": 0.00013213966640860373,
      "loss": 0.8464,
      "step": 116820
    },
    {
      "epoch": 2.051124493056409,
      "grad_norm": 0.08107089982981344,
      "learning_rate": 0.00013212908819018048,
      "loss": 0.8429,
      "step": 116830
    },
    {
      "epoch": 2.0513000579364102,
      "grad_norm": 0.05832598153035355,
      "learning_rate": 0.00013211850957734576,
      "loss": 0.8434,
      "step": 116840
    },
    {
      "epoch": 2.0514756228164117,
      "grad_norm": 0.06625837677210851,
      "learning_rate": 0.0001321079305702335,
      "loss": 0.8464,
      "step": 116850
    },
    {
      "epoch": 2.051651187696413,
      "grad_norm": 0.04776299835760795,
      "learning_rate": 0.00013209735116897781,
      "loss": 0.8418,
      "step": 116860
    },
    {
      "epoch": 2.0518267525764147,
      "grad_norm": 0.06978582788693996,
      "learning_rate": 0.00013208677137371272,
      "loss": 0.8447,
      "step": 116870
    },
    {
      "epoch": 2.052002317456416,
      "grad_norm": 0.08732448344240398,
      "learning_rate": 0.00013207619118457228,
      "loss": 0.8429,
      "step": 116880
    },
    {
      "epoch": 2.0521778823364176,
      "grad_norm": 0.06254202261491476,
      "learning_rate": 0.00013206561060169052,
      "loss": 0.8442,
      "step": 116890
    },
    {
      "epoch": 2.0523534472164187,
      "grad_norm": 0.05285086836895094,
      "learning_rate": 0.0001320550296252016,
      "loss": 0.8441,
      "step": 116900
    },
    {
      "epoch": 2.05252901209642,
      "grad_norm": 0.0503620800079201,
      "learning_rate": 0.00013204444825523946,
      "loss": 0.85,
      "step": 116910
    },
    {
      "epoch": 2.0527045769764216,
      "grad_norm": 0.06008033063326506,
      "learning_rate": 0.0001320338664919382,
      "loss": 0.8453,
      "step": 116920
    },
    {
      "epoch": 2.052880141856423,
      "grad_norm": 0.050234672109880175,
      "learning_rate": 0.00013202328433543187,
      "loss": 0.8567,
      "step": 116930
    },
    {
      "epoch": 2.0530557067364246,
      "grad_norm": 0.05778919015963096,
      "learning_rate": 0.00013201270178585464,
      "loss": 0.8512,
      "step": 116940
    },
    {
      "epoch": 2.053231271616426,
      "grad_norm": 0.07014813109728066,
      "learning_rate": 0.00013200211884334054,
      "loss": 0.847,
      "step": 116950
    },
    {
      "epoch": 2.053406836496427,
      "grad_norm": 0.07730391193767756,
      "learning_rate": 0.00013199153550802362,
      "loss": 0.8497,
      "step": 116960
    },
    {
      "epoch": 2.0535824013764286,
      "grad_norm": 0.05135339323569785,
      "learning_rate": 0.00013198095178003797,
      "loss": 0.8448,
      "step": 116970
    },
    {
      "epoch": 2.05375796625643,
      "grad_norm": 0.08866676068461232,
      "learning_rate": 0.0001319703676595178,
      "loss": 0.8505,
      "step": 116980
    },
    {
      "epoch": 2.0539335311364315,
      "grad_norm": 0.051504735699953355,
      "learning_rate": 0.0001319597831465971,
      "loss": 0.8478,
      "step": 116990
    },
    {
      "epoch": 2.054109096016433,
      "grad_norm": 0.055775260413131494,
      "learning_rate": 0.00013194919824141,
      "loss": 0.8399,
      "step": 117000
    },
    {
      "epoch": 2.0542846608964345,
      "grad_norm": 0.06268497421104081,
      "learning_rate": 0.00013193861294409068,
      "loss": 0.8491,
      "step": 117010
    },
    {
      "epoch": 2.0544602257764355,
      "grad_norm": 0.0527786127006905,
      "learning_rate": 0.00013192802725477313,
      "loss": 0.8494,
      "step": 117020
    },
    {
      "epoch": 2.054635790656437,
      "grad_norm": 0.08092653836116273,
      "learning_rate": 0.0001319174411735916,
      "loss": 0.8397,
      "step": 117030
    },
    {
      "epoch": 2.0548113555364385,
      "grad_norm": 0.056513994499245664,
      "learning_rate": 0.00013190685470068012,
      "loss": 0.8456,
      "step": 117040
    },
    {
      "epoch": 2.05498692041644,
      "grad_norm": 0.04403965014468588,
      "learning_rate": 0.00013189626783617292,
      "loss": 0.8442,
      "step": 117050
    },
    {
      "epoch": 2.0551624852964414,
      "grad_norm": 0.048100450151004795,
      "learning_rate": 0.00013188568058020408,
      "loss": 0.8468,
      "step": 117060
    },
    {
      "epoch": 2.055338050176443,
      "grad_norm": 0.0687114888222507,
      "learning_rate": 0.0001318750929329077,
      "loss": 0.8513,
      "step": 117070
    },
    {
      "epoch": 2.055513615056444,
      "grad_norm": 0.05583709159471642,
      "learning_rate": 0.00013186450489441803,
      "loss": 0.8504,
      "step": 117080
    },
    {
      "epoch": 2.0556891799364454,
      "grad_norm": 0.06150920063138464,
      "learning_rate": 0.00013185391646486914,
      "loss": 0.8399,
      "step": 117090
    },
    {
      "epoch": 2.055864744816447,
      "grad_norm": 0.050187381075212915,
      "learning_rate": 0.0001318433276443952,
      "loss": 0.8482,
      "step": 117100
    },
    {
      "epoch": 2.0560403096964484,
      "grad_norm": 0.06081301717443002,
      "learning_rate": 0.00013183273843313042,
      "loss": 0.8437,
      "step": 117110
    },
    {
      "epoch": 2.05621587457645,
      "grad_norm": 0.05440428022774388,
      "learning_rate": 0.0001318221488312089,
      "loss": 0.852,
      "step": 117120
    },
    {
      "epoch": 2.0563914394564513,
      "grad_norm": 0.055174173248240366,
      "learning_rate": 0.00013181155883876488,
      "loss": 0.8451,
      "step": 117130
    },
    {
      "epoch": 2.0565670043364523,
      "grad_norm": 0.06182150727201478,
      "learning_rate": 0.00013180096845593248,
      "loss": 0.8374,
      "step": 117140
    },
    {
      "epoch": 2.056742569216454,
      "grad_norm": 0.059485033311463445,
      "learning_rate": 0.00013179037768284597,
      "loss": 0.8519,
      "step": 117150
    },
    {
      "epoch": 2.0569181340964553,
      "grad_norm": 0.05276210178432408,
      "learning_rate": 0.0001317797865196394,
      "loss": 0.8482,
      "step": 117160
    },
    {
      "epoch": 2.0570936989764568,
      "grad_norm": 0.04685972075628747,
      "learning_rate": 0.00013176919496644702,
      "loss": 0.8488,
      "step": 117170
    },
    {
      "epoch": 2.0572692638564583,
      "grad_norm": 0.06646078241010726,
      "learning_rate": 0.00013175860302340312,
      "loss": 0.8399,
      "step": 117180
    },
    {
      "epoch": 2.0574448287364597,
      "grad_norm": 0.09665695608295388,
      "learning_rate": 0.00013174801069064179,
      "loss": 0.8461,
      "step": 117190
    },
    {
      "epoch": 2.0576203936164608,
      "grad_norm": 0.06095798254836928,
      "learning_rate": 0.00013173741796829725,
      "loss": 0.8465,
      "step": 117200
    },
    {
      "epoch": 2.0577959584964622,
      "grad_norm": 0.08214773134898144,
      "learning_rate": 0.00013172682485650377,
      "loss": 0.8465,
      "step": 117210
    },
    {
      "epoch": 2.0579715233764637,
      "grad_norm": 0.04749023393663511,
      "learning_rate": 0.0001317162313553955,
      "loss": 0.841,
      "step": 117220
    },
    {
      "epoch": 2.058147088256465,
      "grad_norm": 0.06684910127792876,
      "learning_rate": 0.0001317056374651067,
      "loss": 0.8514,
      "step": 117230
    },
    {
      "epoch": 2.0583226531364667,
      "grad_norm": 0.06306280836075377,
      "learning_rate": 0.00013169504318577161,
      "loss": 0.8432,
      "step": 117240
    },
    {
      "epoch": 2.058498218016468,
      "grad_norm": 0.060018147719329715,
      "learning_rate": 0.00013168444851752444,
      "loss": 0.8498,
      "step": 117250
    },
    {
      "epoch": 2.058673782896469,
      "grad_norm": 0.04645489561283139,
      "learning_rate": 0.0001316738534604994,
      "loss": 0.8471,
      "step": 117260
    },
    {
      "epoch": 2.0588493477764707,
      "grad_norm": 0.05983014448167691,
      "learning_rate": 0.00013166325801483074,
      "loss": 0.8492,
      "step": 117270
    },
    {
      "epoch": 2.059024912656472,
      "grad_norm": 0.05836171615952909,
      "learning_rate": 0.00013165266218065276,
      "loss": 0.8495,
      "step": 117280
    },
    {
      "epoch": 2.0592004775364736,
      "grad_norm": 0.05628495813643304,
      "learning_rate": 0.0001316420659580997,
      "loss": 0.8551,
      "step": 117290
    },
    {
      "epoch": 2.059376042416475,
      "grad_norm": 0.04807645485968771,
      "learning_rate": 0.00013163146934730576,
      "loss": 0.8492,
      "step": 117300
    },
    {
      "epoch": 2.0595516072964766,
      "grad_norm": 0.08626536291749594,
      "learning_rate": 0.0001316208723484052,
      "loss": 0.8432,
      "step": 117310
    },
    {
      "epoch": 2.0597271721764776,
      "grad_norm": 0.053627977994988085,
      "learning_rate": 0.00013161027496153238,
      "loss": 0.8482,
      "step": 117320
    },
    {
      "epoch": 2.059902737056479,
      "grad_norm": 0.06905659020750213,
      "learning_rate": 0.00013159967718682144,
      "loss": 0.8438,
      "step": 117330
    },
    {
      "epoch": 2.0600783019364806,
      "grad_norm": 0.05698726476192725,
      "learning_rate": 0.00013158907902440677,
      "loss": 0.848,
      "step": 117340
    },
    {
      "epoch": 2.060253866816482,
      "grad_norm": 0.0644578509057911,
      "learning_rate": 0.00013157848047442262,
      "loss": 0.8442,
      "step": 117350
    },
    {
      "epoch": 2.0604294316964835,
      "grad_norm": 0.05535545499088577,
      "learning_rate": 0.00013156788153700325,
      "loss": 0.8449,
      "step": 117360
    },
    {
      "epoch": 2.060604996576485,
      "grad_norm": 0.07270484399188647,
      "learning_rate": 0.00013155728221228295,
      "loss": 0.8416,
      "step": 117370
    },
    {
      "epoch": 2.060780561456486,
      "grad_norm": 0.07362176247010538,
      "learning_rate": 0.00013154668250039603,
      "loss": 0.852,
      "step": 117380
    },
    {
      "epoch": 2.0609561263364875,
      "grad_norm": 0.06282169931919038,
      "learning_rate": 0.00013153608240147678,
      "loss": 0.8383,
      "step": 117390
    },
    {
      "epoch": 2.061131691216489,
      "grad_norm": 0.06238277698393166,
      "learning_rate": 0.00013152548191565957,
      "loss": 0.8477,
      "step": 117400
    },
    {
      "epoch": 2.0613072560964905,
      "grad_norm": 0.08803594550315069,
      "learning_rate": 0.0001315148810430786,
      "loss": 0.8528,
      "step": 117410
    },
    {
      "epoch": 2.061482820976492,
      "grad_norm": 0.06832259739252615,
      "learning_rate": 0.00013150427978386825,
      "loss": 0.8403,
      "step": 117420
    },
    {
      "epoch": 2.0616583858564934,
      "grad_norm": 0.05810143877432337,
      "learning_rate": 0.00013149367813816283,
      "loss": 0.8534,
      "step": 117430
    },
    {
      "epoch": 2.061833950736495,
      "grad_norm": 0.06232509240224349,
      "learning_rate": 0.00013148307610609666,
      "loss": 0.8458,
      "step": 117440
    },
    {
      "epoch": 2.062009515616496,
      "grad_norm": 0.047940697218336506,
      "learning_rate": 0.0001314724736878041,
      "loss": 0.8427,
      "step": 117450
    },
    {
      "epoch": 2.0621850804964974,
      "grad_norm": 0.04421746022950881,
      "learning_rate": 0.00013146187088341942,
      "loss": 0.8464,
      "step": 117460
    },
    {
      "epoch": 2.062360645376499,
      "grad_norm": 0.05316127648732322,
      "learning_rate": 0.00013145126769307703,
      "loss": 0.8548,
      "step": 117470
    },
    {
      "epoch": 2.0625362102565004,
      "grad_norm": 0.055929061941069474,
      "learning_rate": 0.00013144066411691123,
      "loss": 0.8498,
      "step": 117480
    },
    {
      "epoch": 2.062711775136502,
      "grad_norm": 0.05649436224928336,
      "learning_rate": 0.00013143006015505637,
      "loss": 0.8523,
      "step": 117490
    },
    {
      "epoch": 2.0628873400165033,
      "grad_norm": 0.05703313569068454,
      "learning_rate": 0.0001314194558076468,
      "loss": 0.8406,
      "step": 117500
    },
    {
      "epoch": 2.0630629048965043,
      "grad_norm": 0.06513264830100284,
      "learning_rate": 0.00013140885107481692,
      "loss": 0.8473,
      "step": 117510
    },
    {
      "epoch": 2.063238469776506,
      "grad_norm": 0.0659363644351434,
      "learning_rate": 0.0001313982459567011,
      "loss": 0.8442,
      "step": 117520
    },
    {
      "epoch": 2.0634140346565073,
      "grad_norm": 0.04387751325633889,
      "learning_rate": 0.00013138764045343365,
      "loss": 0.8425,
      "step": 117530
    },
    {
      "epoch": 2.063589599536509,
      "grad_norm": 0.07555991854982856,
      "learning_rate": 0.00013137703456514896,
      "loss": 0.8493,
      "step": 117540
    },
    {
      "epoch": 2.0637651644165103,
      "grad_norm": 0.09374892710684568,
      "learning_rate": 0.0001313664282919814,
      "loss": 0.8488,
      "step": 117550
    },
    {
      "epoch": 2.0639407292965117,
      "grad_norm": 0.06642892492447659,
      "learning_rate": 0.0001313558216340654,
      "loss": 0.853,
      "step": 117560
    },
    {
      "epoch": 2.0641162941765128,
      "grad_norm": 0.07090316859647429,
      "learning_rate": 0.00013134521459153532,
      "loss": 0.8443,
      "step": 117570
    },
    {
      "epoch": 2.0642918590565142,
      "grad_norm": 0.05603762961315771,
      "learning_rate": 0.0001313346071645256,
      "loss": 0.8371,
      "step": 117580
    },
    {
      "epoch": 2.0644674239365157,
      "grad_norm": 0.057858323781356205,
      "learning_rate": 0.00013132399935317056,
      "loss": 0.8472,
      "step": 117590
    },
    {
      "epoch": 2.064642988816517,
      "grad_norm": 0.07448642261942436,
      "learning_rate": 0.0001313133911576046,
      "loss": 0.8459,
      "step": 117600
    },
    {
      "epoch": 2.0648185536965187,
      "grad_norm": 0.0702654014668517,
      "learning_rate": 0.00013130278257796217,
      "loss": 0.8392,
      "step": 117610
    },
    {
      "epoch": 2.06499411857652,
      "grad_norm": 0.04775846753050002,
      "learning_rate": 0.00013129217361437773,
      "loss": 0.8448,
      "step": 117620
    },
    {
      "epoch": 2.065169683456521,
      "grad_norm": 0.05566351784856461,
      "learning_rate": 0.00013128156426698564,
      "loss": 0.8484,
      "step": 117630
    },
    {
      "epoch": 2.0653452483365227,
      "grad_norm": 0.04773423477891228,
      "learning_rate": 0.0001312709545359203,
      "loss": 0.848,
      "step": 117640
    },
    {
      "epoch": 2.065520813216524,
      "grad_norm": 0.055725062102700194,
      "learning_rate": 0.00013126034442131616,
      "loss": 0.8497,
      "step": 117650
    },
    {
      "epoch": 2.0656963780965256,
      "grad_norm": 0.05709561322861614,
      "learning_rate": 0.00013124973392330766,
      "loss": 0.8415,
      "step": 117660
    },
    {
      "epoch": 2.065871942976527,
      "grad_norm": 0.06911996911883099,
      "learning_rate": 0.00013123912304202924,
      "loss": 0.848,
      "step": 117670
    },
    {
      "epoch": 2.0660475078565286,
      "grad_norm": 0.06643285651695771,
      "learning_rate": 0.00013122851177761532,
      "loss": 0.849,
      "step": 117680
    },
    {
      "epoch": 2.0662230727365296,
      "grad_norm": 0.076268772533281,
      "learning_rate": 0.0001312179001302004,
      "loss": 0.8563,
      "step": 117690
    },
    {
      "epoch": 2.066398637616531,
      "grad_norm": 0.0522771140148735,
      "learning_rate": 0.00013120728809991886,
      "loss": 0.8485,
      "step": 117700
    },
    {
      "epoch": 2.0665742024965326,
      "grad_norm": 0.06906749270320302,
      "learning_rate": 0.0001311966756869052,
      "loss": 0.8474,
      "step": 117710
    },
    {
      "epoch": 2.066749767376534,
      "grad_norm": 0.06263984022231062,
      "learning_rate": 0.00013118606289129386,
      "loss": 0.8394,
      "step": 117720
    },
    {
      "epoch": 2.0669253322565355,
      "grad_norm": 0.05879904387961159,
      "learning_rate": 0.00013117544971321936,
      "loss": 0.8466,
      "step": 117730
    },
    {
      "epoch": 2.067100897136537,
      "grad_norm": 0.051766424652675454,
      "learning_rate": 0.00013116483615281606,
      "loss": 0.8504,
      "step": 117740
    },
    {
      "epoch": 2.067276462016538,
      "grad_norm": 0.06359597956084846,
      "learning_rate": 0.0001311542222102186,
      "loss": 0.8436,
      "step": 117750
    },
    {
      "epoch": 2.0674520268965395,
      "grad_norm": 0.05610101919326903,
      "learning_rate": 0.00013114360788556132,
      "loss": 0.8469,
      "step": 117760
    },
    {
      "epoch": 2.067627591776541,
      "grad_norm": 0.0534630064185257,
      "learning_rate": 0.0001311329931789787,
      "loss": 0.8446,
      "step": 117770
    },
    {
      "epoch": 2.0678031566565425,
      "grad_norm": 0.056290531078137356,
      "learning_rate": 0.00013112237809060533,
      "loss": 0.8424,
      "step": 117780
    },
    {
      "epoch": 2.067978721536544,
      "grad_norm": 0.049862569672064046,
      "learning_rate": 0.00013111176262057568,
      "loss": 0.8434,
      "step": 117790
    },
    {
      "epoch": 2.0681542864165454,
      "grad_norm": 0.04968111852191073,
      "learning_rate": 0.00013110114676902418,
      "loss": 0.8393,
      "step": 117800
    },
    {
      "epoch": 2.0683298512965465,
      "grad_norm": 0.0468386811004812,
      "learning_rate": 0.0001310905305360854,
      "loss": 0.8572,
      "step": 117810
    },
    {
      "epoch": 2.068505416176548,
      "grad_norm": 0.06506616279207952,
      "learning_rate": 0.00013107991392189382,
      "loss": 0.843,
      "step": 117820
    },
    {
      "epoch": 2.0686809810565494,
      "grad_norm": 0.04993473710354421,
      "learning_rate": 0.000131069296926584,
      "loss": 0.8431,
      "step": 117830
    },
    {
      "epoch": 2.068856545936551,
      "grad_norm": 0.06279240433223755,
      "learning_rate": 0.0001310586795502904,
      "loss": 0.8454,
      "step": 117840
    },
    {
      "epoch": 2.0690321108165524,
      "grad_norm": 0.053092119505256004,
      "learning_rate": 0.00013104806179314754,
      "loss": 0.8462,
      "step": 117850
    },
    {
      "epoch": 2.069207675696554,
      "grad_norm": 0.0609474927688844,
      "learning_rate": 0.00013103744365529005,
      "loss": 0.8447,
      "step": 117860
    },
    {
      "epoch": 2.069383240576555,
      "grad_norm": 0.04828540515928716,
      "learning_rate": 0.00013102682513685233,
      "loss": 0.8478,
      "step": 117870
    },
    {
      "epoch": 2.0695588054565563,
      "grad_norm": 0.06785773591237065,
      "learning_rate": 0.00013101620623796902,
      "loss": 0.8376,
      "step": 117880
    },
    {
      "epoch": 2.069734370336558,
      "grad_norm": 0.06572799022444999,
      "learning_rate": 0.00013100558695877462,
      "loss": 0.8534,
      "step": 117890
    },
    {
      "epoch": 2.0699099352165593,
      "grad_norm": 0.06381464243707702,
      "learning_rate": 0.0001309949672994036,
      "loss": 0.8366,
      "step": 117900
    },
    {
      "epoch": 2.070085500096561,
      "grad_norm": 0.05114644355809604,
      "learning_rate": 0.00013098434725999068,
      "loss": 0.8454,
      "step": 117910
    },
    {
      "epoch": 2.0702610649765623,
      "grad_norm": 0.051372027634937624,
      "learning_rate": 0.00013097372684067038,
      "loss": 0.8477,
      "step": 117920
    },
    {
      "epoch": 2.0704366298565633,
      "grad_norm": 0.06252157647542729,
      "learning_rate": 0.00013096310604157713,
      "loss": 0.841,
      "step": 117930
    },
    {
      "epoch": 2.0706121947365648,
      "grad_norm": 0.04968907396902904,
      "learning_rate": 0.00013095248486284558,
      "loss": 0.85,
      "step": 117940
    },
    {
      "epoch": 2.0707877596165662,
      "grad_norm": 0.06193185325779506,
      "learning_rate": 0.00013094186330461037,
      "loss": 0.8436,
      "step": 117950
    },
    {
      "epoch": 2.0709633244965677,
      "grad_norm": 0.04979606195755423,
      "learning_rate": 0.00013093124136700596,
      "loss": 0.8447,
      "step": 117960
    },
    {
      "epoch": 2.071138889376569,
      "grad_norm": 0.07014103113125919,
      "learning_rate": 0.00013092061905016703,
      "loss": 0.8517,
      "step": 117970
    },
    {
      "epoch": 2.0713144542565707,
      "grad_norm": 0.08426600380381069,
      "learning_rate": 0.0001309099963542281,
      "loss": 0.8497,
      "step": 117980
    },
    {
      "epoch": 2.0714900191365717,
      "grad_norm": 0.07930857408223287,
      "learning_rate": 0.00013089937327932376,
      "loss": 0.8454,
      "step": 117990
    },
    {
      "epoch": 2.071665584016573,
      "grad_norm": 0.05583155242913865,
      "learning_rate": 0.00013088874982558868,
      "loss": 0.8408,
      "step": 118000
    },
    {
      "epoch": 2.0718411488965747,
      "grad_norm": 0.05376811591804228,
      "learning_rate": 0.00013087812599315736,
      "loss": 0.838,
      "step": 118010
    },
    {
      "epoch": 2.072016713776576,
      "grad_norm": 0.06899213811454095,
      "learning_rate": 0.00013086750178216446,
      "loss": 0.8468,
      "step": 118020
    },
    {
      "epoch": 2.0721922786565776,
      "grad_norm": 0.07197068911579907,
      "learning_rate": 0.00013085687719274465,
      "loss": 0.8373,
      "step": 118030
    },
    {
      "epoch": 2.072367843536579,
      "grad_norm": 0.04939052749696648,
      "learning_rate": 0.00013084625222503242,
      "loss": 0.8434,
      "step": 118040
    },
    {
      "epoch": 2.07254340841658,
      "grad_norm": 0.05966030372195348,
      "learning_rate": 0.00013083562687916247,
      "loss": 0.8467,
      "step": 118050
    },
    {
      "epoch": 2.0727189732965816,
      "grad_norm": 0.04853340362247222,
      "learning_rate": 0.00013082500115526942,
      "loss": 0.8461,
      "step": 118060
    },
    {
      "epoch": 2.072894538176583,
      "grad_norm": 0.056299221822315025,
      "learning_rate": 0.0001308143750534879,
      "loss": 0.846,
      "step": 118070
    },
    {
      "epoch": 2.0730701030565846,
      "grad_norm": 0.05473082342763264,
      "learning_rate": 0.00013080374857395253,
      "loss": 0.8433,
      "step": 118080
    },
    {
      "epoch": 2.073245667936586,
      "grad_norm": 0.05647424957894315,
      "learning_rate": 0.00013079312171679794,
      "loss": 0.8512,
      "step": 118090
    },
    {
      "epoch": 2.0734212328165875,
      "grad_norm": 0.07209340114511861,
      "learning_rate": 0.0001307824944821588,
      "loss": 0.845,
      "step": 118100
    },
    {
      "epoch": 2.073596797696589,
      "grad_norm": 0.06360679555317258,
      "learning_rate": 0.00013077186687016973,
      "loss": 0.8485,
      "step": 118110
    },
    {
      "epoch": 2.07377236257659,
      "grad_norm": 0.06348647797796021,
      "learning_rate": 0.00013076123888096537,
      "loss": 0.849,
      "step": 118120
    },
    {
      "epoch": 2.0739479274565915,
      "grad_norm": 0.045860652147728935,
      "learning_rate": 0.00013075061051468048,
      "loss": 0.8409,
      "step": 118130
    },
    {
      "epoch": 2.074123492336593,
      "grad_norm": 0.05852068411308971,
      "learning_rate": 0.00013073998177144962,
      "loss": 0.8461,
      "step": 118140
    },
    {
      "epoch": 2.0742990572165945,
      "grad_norm": 0.06830147591800605,
      "learning_rate": 0.00013072935265140753,
      "loss": 0.8437,
      "step": 118150
    },
    {
      "epoch": 2.074474622096596,
      "grad_norm": 0.06672198765029379,
      "learning_rate": 0.0001307187231546888,
      "loss": 0.8494,
      "step": 118160
    },
    {
      "epoch": 2.074650186976597,
      "grad_norm": 0.06288729578773182,
      "learning_rate": 0.00013070809328142815,
      "loss": 0.8452,
      "step": 118170
    },
    {
      "epoch": 2.0748257518565985,
      "grad_norm": 0.07343908060771329,
      "learning_rate": 0.00013069746303176027,
      "loss": 0.8503,
      "step": 118180
    },
    {
      "epoch": 2.0750013167366,
      "grad_norm": 0.04944791184807578,
      "learning_rate": 0.00013068683240581984,
      "loss": 0.8413,
      "step": 118190
    },
    {
      "epoch": 2.0751768816166014,
      "grad_norm": 0.05589806142155868,
      "learning_rate": 0.0001306762014037416,
      "loss": 0.8473,
      "step": 118200
    },
    {
      "epoch": 2.075352446496603,
      "grad_norm": 0.05085676597573957,
      "learning_rate": 0.00013066557002566016,
      "loss": 0.844,
      "step": 118210
    },
    {
      "epoch": 2.0755280113766044,
      "grad_norm": 0.06695396334319004,
      "learning_rate": 0.00013065493827171026,
      "loss": 0.8525,
      "step": 118220
    },
    {
      "epoch": 2.075703576256606,
      "grad_norm": 0.06160484016234548,
      "learning_rate": 0.0001306443061420266,
      "loss": 0.8482,
      "step": 118230
    },
    {
      "epoch": 2.075879141136607,
      "grad_norm": 0.05013514532133713,
      "learning_rate": 0.0001306336736367439,
      "loss": 0.8549,
      "step": 118240
    },
    {
      "epoch": 2.0760547060166084,
      "grad_norm": 0.09175940893952178,
      "learning_rate": 0.0001306230407559969,
      "loss": 0.8505,
      "step": 118250
    },
    {
      "epoch": 2.07623027089661,
      "grad_norm": 0.05997439304405328,
      "learning_rate": 0.00013061240749992028,
      "loss": 0.8428,
      "step": 118260
    },
    {
      "epoch": 2.0764058357766113,
      "grad_norm": 0.04738890772238235,
      "learning_rate": 0.00013060177386864881,
      "loss": 0.848,
      "step": 118270
    },
    {
      "epoch": 2.076581400656613,
      "grad_norm": 0.04170677429624133,
      "learning_rate": 0.00013059113986231715,
      "loss": 0.8442,
      "step": 118280
    },
    {
      "epoch": 2.0767569655366143,
      "grad_norm": 0.05423571512654899,
      "learning_rate": 0.00013058050548106007,
      "loss": 0.8384,
      "step": 118290
    },
    {
      "epoch": 2.0769325304166153,
      "grad_norm": 0.05725474898382126,
      "learning_rate": 0.00013056987072501236,
      "loss": 0.8445,
      "step": 118300
    },
    {
      "epoch": 2.0771080952966168,
      "grad_norm": 0.05611621702683366,
      "learning_rate": 0.00013055923559430872,
      "loss": 0.8507,
      "step": 118310
    },
    {
      "epoch": 2.0772836601766183,
      "grad_norm": 0.07002228811568009,
      "learning_rate": 0.00013054860008908386,
      "loss": 0.8545,
      "step": 118320
    },
    {
      "epoch": 2.0774592250566197,
      "grad_norm": 0.04908871339609314,
      "learning_rate": 0.00013053796420947261,
      "loss": 0.8425,
      "step": 118330
    },
    {
      "epoch": 2.077634789936621,
      "grad_norm": 0.04695805276124948,
      "learning_rate": 0.00013052732795560968,
      "loss": 0.8457,
      "step": 118340
    },
    {
      "epoch": 2.0778103548166227,
      "grad_norm": 0.048136985618847015,
      "learning_rate": 0.00013051669132762984,
      "loss": 0.839,
      "step": 118350
    },
    {
      "epoch": 2.0779859196966237,
      "grad_norm": 0.04672687878391958,
      "learning_rate": 0.00013050605432566787,
      "loss": 0.844,
      "step": 118360
    },
    {
      "epoch": 2.078161484576625,
      "grad_norm": 0.07708039680252794,
      "learning_rate": 0.00013049541694985854,
      "loss": 0.8465,
      "step": 118370
    },
    {
      "epoch": 2.0783370494566267,
      "grad_norm": 0.05080157488052207,
      "learning_rate": 0.00013048477920033662,
      "loss": 0.8457,
      "step": 118380
    },
    {
      "epoch": 2.078512614336628,
      "grad_norm": 0.055787916595614266,
      "learning_rate": 0.0001304741410772369,
      "loss": 0.8425,
      "step": 118390
    },
    {
      "epoch": 2.0786881792166296,
      "grad_norm": 0.05644821785070292,
      "learning_rate": 0.00013046350258069417,
      "loss": 0.8541,
      "step": 118400
    },
    {
      "epoch": 2.078863744096631,
      "grad_norm": 0.07410387094627895,
      "learning_rate": 0.00013045286371084317,
      "loss": 0.8443,
      "step": 118410
    },
    {
      "epoch": 2.079039308976632,
      "grad_norm": 0.054601319248659276,
      "learning_rate": 0.0001304422244678188,
      "loss": 0.8509,
      "step": 118420
    },
    {
      "epoch": 2.0792148738566336,
      "grad_norm": 0.07200270745868438,
      "learning_rate": 0.0001304315848517558,
      "loss": 0.8518,
      "step": 118430
    },
    {
      "epoch": 2.079390438736635,
      "grad_norm": 0.06647522303922543,
      "learning_rate": 0.00013042094486278896,
      "loss": 0.8512,
      "step": 118440
    },
    {
      "epoch": 2.0795660036166366,
      "grad_norm": 0.05419612894348638,
      "learning_rate": 0.0001304103045010531,
      "loss": 0.852,
      "step": 118450
    },
    {
      "epoch": 2.079741568496638,
      "grad_norm": 0.05746220263040244,
      "learning_rate": 0.00013039966376668307,
      "loss": 0.8484,
      "step": 118460
    },
    {
      "epoch": 2.0799171333766395,
      "grad_norm": 0.06417366374717477,
      "learning_rate": 0.00013038902265981366,
      "loss": 0.8498,
      "step": 118470
    },
    {
      "epoch": 2.0800926982566406,
      "grad_norm": 0.06874477077958871,
      "learning_rate": 0.0001303783811805797,
      "loss": 0.8392,
      "step": 118480
    },
    {
      "epoch": 2.080268263136642,
      "grad_norm": 0.04878451626997051,
      "learning_rate": 0.00013036773932911602,
      "loss": 0.8419,
      "step": 118490
    },
    {
      "epoch": 2.0804438280166435,
      "grad_norm": 0.05295461265971048,
      "learning_rate": 0.00013035709710555743,
      "loss": 0.8478,
      "step": 118500
    },
    {
      "epoch": 2.080619392896645,
      "grad_norm": 0.05566465215360677,
      "learning_rate": 0.00013034645451003883,
      "loss": 0.8508,
      "step": 118510
    },
    {
      "epoch": 2.0807949577766465,
      "grad_norm": 0.055895124126773825,
      "learning_rate": 0.00013033581154269497,
      "loss": 0.8415,
      "step": 118520
    },
    {
      "epoch": 2.080970522656648,
      "grad_norm": 0.07352915255953853,
      "learning_rate": 0.0001303251682036608,
      "loss": 0.8416,
      "step": 118530
    },
    {
      "epoch": 2.081146087536649,
      "grad_norm": 0.05644313845276074,
      "learning_rate": 0.00013031452449307115,
      "loss": 0.8486,
      "step": 118540
    },
    {
      "epoch": 2.0813216524166505,
      "grad_norm": 0.05789895821678486,
      "learning_rate": 0.0001303038804110608,
      "loss": 0.854,
      "step": 118550
    },
    {
      "epoch": 2.081497217296652,
      "grad_norm": 0.04451223502273039,
      "learning_rate": 0.00013029323595776471,
      "loss": 0.8455,
      "step": 118560
    },
    {
      "epoch": 2.0816727821766534,
      "grad_norm": 0.0472715382705794,
      "learning_rate": 0.00013028259113331768,
      "loss": 0.8471,
      "step": 118570
    },
    {
      "epoch": 2.081848347056655,
      "grad_norm": 0.06057419246399166,
      "learning_rate": 0.00013027194593785463,
      "loss": 0.8472,
      "step": 118580
    },
    {
      "epoch": 2.0820239119366564,
      "grad_norm": 0.038454005591746665,
      "learning_rate": 0.0001302613003715104,
      "loss": 0.8512,
      "step": 118590
    },
    {
      "epoch": 2.0821994768166574,
      "grad_norm": 0.0523775493078008,
      "learning_rate": 0.00013025065443441993,
      "loss": 0.8499,
      "step": 118600
    },
    {
      "epoch": 2.082375041696659,
      "grad_norm": 0.06044270087472284,
      "learning_rate": 0.00013024000812671805,
      "loss": 0.8483,
      "step": 118610
    },
    {
      "epoch": 2.0825506065766604,
      "grad_norm": 0.0444396072175837,
      "learning_rate": 0.00013022936144853963,
      "loss": 0.8388,
      "step": 118620
    },
    {
      "epoch": 2.082726171456662,
      "grad_norm": 0.0523506241054172,
      "learning_rate": 0.00013021871440001962,
      "loss": 0.8507,
      "step": 118630
    },
    {
      "epoch": 2.0829017363366633,
      "grad_norm": 0.047146156806259346,
      "learning_rate": 0.00013020806698129286,
      "loss": 0.8468,
      "step": 118640
    },
    {
      "epoch": 2.083077301216665,
      "grad_norm": 0.05945772526434581,
      "learning_rate": 0.00013019741919249433,
      "loss": 0.8419,
      "step": 118650
    },
    {
      "epoch": 2.083252866096666,
      "grad_norm": 0.06684334870738119,
      "learning_rate": 0.0001301867710337589,
      "loss": 0.8392,
      "step": 118660
    },
    {
      "epoch": 2.0834284309766673,
      "grad_norm": 0.07477604644118806,
      "learning_rate": 0.00013017612250522148,
      "loss": 0.8474,
      "step": 118670
    },
    {
      "epoch": 2.083603995856669,
      "grad_norm": 0.06423417350365533,
      "learning_rate": 0.000130165473607017,
      "loss": 0.8414,
      "step": 118680
    },
    {
      "epoch": 2.0837795607366703,
      "grad_norm": 0.0536944698029669,
      "learning_rate": 0.00013015482433928038,
      "loss": 0.8448,
      "step": 118690
    },
    {
      "epoch": 2.0839551256166717,
      "grad_norm": 0.04551269512970482,
      "learning_rate": 0.0001301441747021465,
      "loss": 0.8483,
      "step": 118700
    },
    {
      "epoch": 2.084130690496673,
      "grad_norm": 0.060944129837585306,
      "learning_rate": 0.0001301335246957504,
      "loss": 0.8496,
      "step": 118710
    },
    {
      "epoch": 2.0843062553766742,
      "grad_norm": 0.051701522725089225,
      "learning_rate": 0.000130122874320227,
      "loss": 0.8486,
      "step": 118720
    },
    {
      "epoch": 2.0844818202566757,
      "grad_norm": 0.06206857779416364,
      "learning_rate": 0.00013011222357571113,
      "loss": 0.8481,
      "step": 118730
    },
    {
      "epoch": 2.084657385136677,
      "grad_norm": 0.06482821548729743,
      "learning_rate": 0.0001301015724623378,
      "loss": 0.8447,
      "step": 118740
    },
    {
      "epoch": 2.0848329500166787,
      "grad_norm": 0.05120482776346574,
      "learning_rate": 0.000130090920980242,
      "loss": 0.8442,
      "step": 118750
    },
    {
      "epoch": 2.08500851489668,
      "grad_norm": 0.05424320640413045,
      "learning_rate": 0.00013008026912955867,
      "loss": 0.8449,
      "step": 118760
    },
    {
      "epoch": 2.0851840797766816,
      "grad_norm": 0.06726648813085498,
      "learning_rate": 0.00013006961691042272,
      "loss": 0.8499,
      "step": 118770
    },
    {
      "epoch": 2.0853596446566827,
      "grad_norm": 0.06255846654018957,
      "learning_rate": 0.0001300589643229692,
      "loss": 0.8449,
      "step": 118780
    },
    {
      "epoch": 2.085535209536684,
      "grad_norm": 0.04957427466986319,
      "learning_rate": 0.000130048311367333,
      "loss": 0.8395,
      "step": 118790
    },
    {
      "epoch": 2.0857107744166856,
      "grad_norm": 0.04741331489984336,
      "learning_rate": 0.00013003765804364914,
      "loss": 0.8403,
      "step": 118800
    },
    {
      "epoch": 2.085886339296687,
      "grad_norm": 0.059287024615710934,
      "learning_rate": 0.00013002700435205255,
      "loss": 0.8475,
      "step": 118810
    },
    {
      "epoch": 2.0860619041766886,
      "grad_norm": 0.05298919594229593,
      "learning_rate": 0.0001300163502926783,
      "loss": 0.8437,
      "step": 118820
    },
    {
      "epoch": 2.08623746905669,
      "grad_norm": 0.058677960794733704,
      "learning_rate": 0.0001300056958656613,
      "loss": 0.8421,
      "step": 118830
    },
    {
      "epoch": 2.086413033936691,
      "grad_norm": 0.0899647270889374,
      "learning_rate": 0.0001299950410711366,
      "loss": 0.8478,
      "step": 118840
    },
    {
      "epoch": 2.0865885988166926,
      "grad_norm": 0.05651325758564085,
      "learning_rate": 0.00012998438590923917,
      "loss": 0.8426,
      "step": 118850
    },
    {
      "epoch": 2.086764163696694,
      "grad_norm": 0.047521045145038174,
      "learning_rate": 0.00012997373038010397,
      "loss": 0.8504,
      "step": 118860
    },
    {
      "epoch": 2.0869397285766955,
      "grad_norm": 0.055074486689884156,
      "learning_rate": 0.00012996307448386607,
      "loss": 0.8387,
      "step": 118870
    },
    {
      "epoch": 2.087115293456697,
      "grad_norm": 0.05790227964044634,
      "learning_rate": 0.00012995241822066054,
      "loss": 0.8542,
      "step": 118880
    },
    {
      "epoch": 2.0872908583366985,
      "grad_norm": 0.04792861610532143,
      "learning_rate": 0.00012994176159062223,
      "loss": 0.8466,
      "step": 118890
    },
    {
      "epoch": 2.0874664232167,
      "grad_norm": 0.05088836514725931,
      "learning_rate": 0.0001299311045938863,
      "loss": 0.8433,
      "step": 118900
    },
    {
      "epoch": 2.087641988096701,
      "grad_norm": 0.058928982018121856,
      "learning_rate": 0.00012992044723058772,
      "loss": 0.8456,
      "step": 118910
    },
    {
      "epoch": 2.0878175529767025,
      "grad_norm": 0.14996525471979177,
      "learning_rate": 0.0001299097895008615,
      "loss": 0.8356,
      "step": 118920
    },
    {
      "epoch": 2.087993117856704,
      "grad_norm": 0.057527029766592915,
      "learning_rate": 0.00012989913140484276,
      "loss": 0.8467,
      "step": 118930
    },
    {
      "epoch": 2.0881686827367054,
      "grad_norm": 0.062479208861357834,
      "learning_rate": 0.0001298884729426665,
      "loss": 0.8517,
      "step": 118940
    },
    {
      "epoch": 2.088344247616707,
      "grad_norm": 0.0598885970714651,
      "learning_rate": 0.0001298778141144677,
      "loss": 0.8479,
      "step": 118950
    },
    {
      "epoch": 2.0885198124967084,
      "grad_norm": 0.047131678062513546,
      "learning_rate": 0.00012986715492038146,
      "loss": 0.8389,
      "step": 118960
    },
    {
      "epoch": 2.0886953773767094,
      "grad_norm": 0.05034395564671304,
      "learning_rate": 0.00012985649536054287,
      "loss": 0.8494,
      "step": 118970
    },
    {
      "epoch": 2.088870942256711,
      "grad_norm": 0.08800170177354992,
      "learning_rate": 0.00012984583543508693,
      "loss": 0.8469,
      "step": 118980
    },
    {
      "epoch": 2.0890465071367124,
      "grad_norm": 0.0813289232223802,
      "learning_rate": 0.00012983517514414874,
      "loss": 0.84,
      "step": 118990
    },
    {
      "epoch": 2.089222072016714,
      "grad_norm": 0.07425434050543239,
      "learning_rate": 0.00012982451448786335,
      "loss": 0.8505,
      "step": 119000
    },
    {
      "epoch": 2.0893976368967153,
      "grad_norm": 0.08887626486826802,
      "learning_rate": 0.0001298138534663658,
      "loss": 0.8415,
      "step": 119010
    },
    {
      "epoch": 2.089573201776717,
      "grad_norm": 0.05114025308548986,
      "learning_rate": 0.00012980319207979126,
      "loss": 0.8433,
      "step": 119020
    },
    {
      "epoch": 2.089748766656718,
      "grad_norm": 0.05989045765072727,
      "learning_rate": 0.00012979253032827472,
      "loss": 0.8432,
      "step": 119030
    },
    {
      "epoch": 2.0899243315367193,
      "grad_norm": 0.06750992564981907,
      "learning_rate": 0.00012978186821195132,
      "loss": 0.8462,
      "step": 119040
    },
    {
      "epoch": 2.090099896416721,
      "grad_norm": 0.043925868510633265,
      "learning_rate": 0.00012977120573095616,
      "loss": 0.8437,
      "step": 119050
    },
    {
      "epoch": 2.0902754612967223,
      "grad_norm": 0.058341948424190555,
      "learning_rate": 0.00012976054288542426,
      "loss": 0.8489,
      "step": 119060
    },
    {
      "epoch": 2.0904510261767237,
      "grad_norm": 0.043171138671992675,
      "learning_rate": 0.00012974987967549078,
      "loss": 0.8453,
      "step": 119070
    },
    {
      "epoch": 2.090626591056725,
      "grad_norm": 0.05530716073458449,
      "learning_rate": 0.00012973921610129083,
      "loss": 0.8439,
      "step": 119080
    },
    {
      "epoch": 2.0908021559367262,
      "grad_norm": 0.0598933000504145,
      "learning_rate": 0.0001297285521629595,
      "loss": 0.8443,
      "step": 119090
    },
    {
      "epoch": 2.0909777208167277,
      "grad_norm": 0.05296229222252606,
      "learning_rate": 0.0001297178878606319,
      "loss": 0.8435,
      "step": 119100
    },
    {
      "epoch": 2.091153285696729,
      "grad_norm": 0.06317692273949202,
      "learning_rate": 0.00012970722319444316,
      "loss": 0.849,
      "step": 119110
    },
    {
      "epoch": 2.0913288505767307,
      "grad_norm": 0.05176946521246532,
      "learning_rate": 0.0001296965581645284,
      "loss": 0.8403,
      "step": 119120
    },
    {
      "epoch": 2.091504415456732,
      "grad_norm": 0.05479756542991588,
      "learning_rate": 0.00012968589277102276,
      "loss": 0.8527,
      "step": 119130
    },
    {
      "epoch": 2.0916799803367336,
      "grad_norm": 0.048594726292483126,
      "learning_rate": 0.00012967522701406135,
      "loss": 0.8507,
      "step": 119140
    },
    {
      "epoch": 2.0918555452167347,
      "grad_norm": 0.04464203382567726,
      "learning_rate": 0.00012966456089377932,
      "loss": 0.8554,
      "step": 119150
    },
    {
      "epoch": 2.092031110096736,
      "grad_norm": 0.0501921797257226,
      "learning_rate": 0.00012965389441031182,
      "loss": 0.8475,
      "step": 119160
    },
    {
      "epoch": 2.0922066749767376,
      "grad_norm": 0.08725694317022412,
      "learning_rate": 0.00012964322756379399,
      "loss": 0.8396,
      "step": 119170
    },
    {
      "epoch": 2.092382239856739,
      "grad_norm": 0.07343568388306458,
      "learning_rate": 0.00012963256035436094,
      "loss": 0.8492,
      "step": 119180
    },
    {
      "epoch": 2.0925578047367406,
      "grad_norm": 0.05239381402519818,
      "learning_rate": 0.00012962189278214792,
      "loss": 0.8422,
      "step": 119190
    },
    {
      "epoch": 2.092733369616742,
      "grad_norm": 0.05894281284536644,
      "learning_rate": 0.00012961122484728995,
      "loss": 0.844,
      "step": 119200
    },
    {
      "epoch": 2.092908934496743,
      "grad_norm": 0.060772315019222214,
      "learning_rate": 0.00012960055654992234,
      "loss": 0.8534,
      "step": 119210
    },
    {
      "epoch": 2.0930844993767446,
      "grad_norm": 0.07340050083328287,
      "learning_rate": 0.00012958988789018024,
      "loss": 0.8413,
      "step": 119220
    },
    {
      "epoch": 2.093260064256746,
      "grad_norm": 0.054145156728760746,
      "learning_rate": 0.0001295792188681987,
      "loss": 0.8448,
      "step": 119230
    },
    {
      "epoch": 2.0934356291367475,
      "grad_norm": 0.06329412344500263,
      "learning_rate": 0.000129568549484113,
      "loss": 0.8475,
      "step": 119240
    },
    {
      "epoch": 2.093611194016749,
      "grad_norm": 0.06025345153675746,
      "learning_rate": 0.00012955787973805834,
      "loss": 0.8436,
      "step": 119250
    },
    {
      "epoch": 2.0937867588967505,
      "grad_norm": 0.06009372035563773,
      "learning_rate": 0.00012954720963016976,
      "loss": 0.8387,
      "step": 119260
    },
    {
      "epoch": 2.0939623237767515,
      "grad_norm": 0.057737614941348686,
      "learning_rate": 0.0001295365391605827,
      "loss": 0.8502,
      "step": 119270
    },
    {
      "epoch": 2.094137888656753,
      "grad_norm": 0.052606189309477906,
      "learning_rate": 0.00012952586832943217,
      "loss": 0.8405,
      "step": 119280
    },
    {
      "epoch": 2.0943134535367545,
      "grad_norm": 0.06512257055977859,
      "learning_rate": 0.00012951519713685341,
      "loss": 0.8416,
      "step": 119290
    },
    {
      "epoch": 2.094489018416756,
      "grad_norm": 0.06261862558368836,
      "learning_rate": 0.00012950452558298166,
      "loss": 0.8384,
      "step": 119300
    },
    {
      "epoch": 2.0946645832967574,
      "grad_norm": 0.07478950630296173,
      "learning_rate": 0.00012949385366795206,
      "loss": 0.8428,
      "step": 119310
    },
    {
      "epoch": 2.094840148176759,
      "grad_norm": 0.04501536766710995,
      "learning_rate": 0.0001294831813918999,
      "loss": 0.8411,
      "step": 119320
    },
    {
      "epoch": 2.09501571305676,
      "grad_norm": 0.05209167569663294,
      "learning_rate": 0.00012947250875496038,
      "loss": 0.8384,
      "step": 119330
    },
    {
      "epoch": 2.0951912779367614,
      "grad_norm": 0.05954363979486522,
      "learning_rate": 0.00012946183575726874,
      "loss": 0.8419,
      "step": 119340
    },
    {
      "epoch": 2.095366842816763,
      "grad_norm": 0.05540306952595872,
      "learning_rate": 0.00012945116239896013,
      "loss": 0.8461,
      "step": 119350
    },
    {
      "epoch": 2.0955424076967644,
      "grad_norm": 0.055658991680472454,
      "learning_rate": 0.00012944048868016988,
      "loss": 0.8411,
      "step": 119360
    },
    {
      "epoch": 2.095717972576766,
      "grad_norm": 0.07415580308536401,
      "learning_rate": 0.0001294298146010332,
      "loss": 0.8457,
      "step": 119370
    },
    {
      "epoch": 2.0958935374567673,
      "grad_norm": 0.05875727495888005,
      "learning_rate": 0.00012941914016168527,
      "loss": 0.8493,
      "step": 119380
    },
    {
      "epoch": 2.0960691023367684,
      "grad_norm": 0.05849005890618339,
      "learning_rate": 0.00012940846536226144,
      "loss": 0.8494,
      "step": 119390
    },
    {
      "epoch": 2.09624466721677,
      "grad_norm": 0.06283955390024872,
      "learning_rate": 0.00012939779020289686,
      "loss": 0.8439,
      "step": 119400
    },
    {
      "epoch": 2.0964202320967713,
      "grad_norm": 0.051115569405675625,
      "learning_rate": 0.00012938711468372688,
      "loss": 0.8412,
      "step": 119410
    },
    {
      "epoch": 2.096595796976773,
      "grad_norm": 0.06397583207457816,
      "learning_rate": 0.0001293764388048867,
      "loss": 0.8498,
      "step": 119420
    },
    {
      "epoch": 2.0967713618567743,
      "grad_norm": 0.06577406594790697,
      "learning_rate": 0.00012936576256651156,
      "loss": 0.8449,
      "step": 119430
    },
    {
      "epoch": 2.0969469267367757,
      "grad_norm": 0.060924461291992436,
      "learning_rate": 0.00012935508596873685,
      "loss": 0.8528,
      "step": 119440
    },
    {
      "epoch": 2.0971224916167768,
      "grad_norm": 0.054585407820245097,
      "learning_rate": 0.0001293444090116977,
      "loss": 0.8496,
      "step": 119450
    },
    {
      "epoch": 2.0972980564967783,
      "grad_norm": 0.049489830812382354,
      "learning_rate": 0.00012933373169552953,
      "loss": 0.8443,
      "step": 119460
    },
    {
      "epoch": 2.0974736213767797,
      "grad_norm": 0.047651022594085574,
      "learning_rate": 0.00012932305402036745,
      "loss": 0.8477,
      "step": 119470
    },
    {
      "epoch": 2.097649186256781,
      "grad_norm": 0.044677718240568154,
      "learning_rate": 0.00012931237598634693,
      "loss": 0.8484,
      "step": 119480
    },
    {
      "epoch": 2.0978247511367827,
      "grad_norm": 0.04985374384716653,
      "learning_rate": 0.00012930169759360312,
      "loss": 0.8501,
      "step": 119490
    },
    {
      "epoch": 2.098000316016784,
      "grad_norm": 0.06536676565661524,
      "learning_rate": 0.00012929101884227143,
      "loss": 0.8477,
      "step": 119500
    },
    {
      "epoch": 2.098175880896785,
      "grad_norm": 0.05227888575025538,
      "learning_rate": 0.0001292803397324871,
      "loss": 0.8443,
      "step": 119510
    },
    {
      "epoch": 2.0983514457767867,
      "grad_norm": 0.05741158065631572,
      "learning_rate": 0.00012926966026438543,
      "loss": 0.8446,
      "step": 119520
    },
    {
      "epoch": 2.098527010656788,
      "grad_norm": 0.0555165870991368,
      "learning_rate": 0.00012925898043810173,
      "loss": 0.8456,
      "step": 119530
    },
    {
      "epoch": 2.0987025755367896,
      "grad_norm": 0.06065128829062834,
      "learning_rate": 0.00012924830025377136,
      "loss": 0.8422,
      "step": 119540
    },
    {
      "epoch": 2.098878140416791,
      "grad_norm": 0.056264711310855905,
      "learning_rate": 0.00012923761971152958,
      "loss": 0.8474,
      "step": 119550
    },
    {
      "epoch": 2.0990537052967926,
      "grad_norm": 0.07159646306618508,
      "learning_rate": 0.00012922693881151182,
      "loss": 0.8407,
      "step": 119560
    },
    {
      "epoch": 2.099229270176794,
      "grad_norm": 0.05803496818593126,
      "learning_rate": 0.00012921625755385328,
      "loss": 0.8448,
      "step": 119570
    },
    {
      "epoch": 2.099404835056795,
      "grad_norm": 0.05559012646563602,
      "learning_rate": 0.00012920557593868934,
      "loss": 0.8489,
      "step": 119580
    },
    {
      "epoch": 2.0995803999367966,
      "grad_norm": 0.06787745363183025,
      "learning_rate": 0.00012919489396615537,
      "loss": 0.8429,
      "step": 119590
    },
    {
      "epoch": 2.099755964816798,
      "grad_norm": 0.06769318875925688,
      "learning_rate": 0.00012918421163638665,
      "loss": 0.8452,
      "step": 119600
    },
    {
      "epoch": 2.0999315296967995,
      "grad_norm": 0.04795884508436189,
      "learning_rate": 0.00012917352894951863,
      "loss": 0.8416,
      "step": 119610
    },
    {
      "epoch": 2.100107094576801,
      "grad_norm": 0.055808325945215106,
      "learning_rate": 0.00012916284590568656,
      "loss": 0.8491,
      "step": 119620
    },
    {
      "epoch": 2.100282659456802,
      "grad_norm": 0.06505995974052305,
      "learning_rate": 0.00012915216250502587,
      "loss": 0.8484,
      "step": 119630
    },
    {
      "epoch": 2.1004582243368035,
      "grad_norm": 0.05415543982621401,
      "learning_rate": 0.00012914147874767184,
      "loss": 0.8426,
      "step": 119640
    },
    {
      "epoch": 2.100633789216805,
      "grad_norm": 0.056809176685881965,
      "learning_rate": 0.00012913079463375992,
      "loss": 0.842,
      "step": 119650
    },
    {
      "epoch": 2.1008093540968065,
      "grad_norm": 0.06017411354021387,
      "learning_rate": 0.00012912011016342538,
      "loss": 0.8459,
      "step": 119660
    },
    {
      "epoch": 2.100984918976808,
      "grad_norm": 0.05263617757758233,
      "learning_rate": 0.00012910942533680373,
      "loss": 0.8435,
      "step": 119670
    },
    {
      "epoch": 2.1011604838568094,
      "grad_norm": 0.055137533743825996,
      "learning_rate": 0.00012909874015403023,
      "loss": 0.8556,
      "step": 119680
    },
    {
      "epoch": 2.101336048736811,
      "grad_norm": 0.0640425422246378,
      "learning_rate": 0.00012908805461524033,
      "loss": 0.8582,
      "step": 119690
    },
    {
      "epoch": 2.101511613616812,
      "grad_norm": 0.06329322398041959,
      "learning_rate": 0.00012907736872056936,
      "loss": 0.842,
      "step": 119700
    },
    {
      "epoch": 2.1016871784968134,
      "grad_norm": 0.0776990387663314,
      "learning_rate": 0.00012906668247015277,
      "loss": 0.8421,
      "step": 119710
    },
    {
      "epoch": 2.101862743376815,
      "grad_norm": 0.061480815767956054,
      "learning_rate": 0.00012905599586412594,
      "loss": 0.8422,
      "step": 119720
    },
    {
      "epoch": 2.1020383082568164,
      "grad_norm": 0.052681379775825786,
      "learning_rate": 0.0001290453089026243,
      "loss": 0.8499,
      "step": 119730
    },
    {
      "epoch": 2.102213873136818,
      "grad_norm": 0.07958251706693292,
      "learning_rate": 0.00012903462158578316,
      "loss": 0.8479,
      "step": 119740
    },
    {
      "epoch": 2.1023894380168193,
      "grad_norm": 0.07647845522021782,
      "learning_rate": 0.00012902393391373798,
      "loss": 0.8441,
      "step": 119750
    },
    {
      "epoch": 2.1025650028968204,
      "grad_norm": 0.060657759464783294,
      "learning_rate": 0.00012901324588662422,
      "loss": 0.8465,
      "step": 119760
    },
    {
      "epoch": 2.102740567776822,
      "grad_norm": 0.04467368215676108,
      "learning_rate": 0.00012900255750457725,
      "loss": 0.8415,
      "step": 119770
    },
    {
      "epoch": 2.1029161326568233,
      "grad_norm": 0.09038001991499593,
      "learning_rate": 0.00012899186876773252,
      "loss": 0.8505,
      "step": 119780
    },
    {
      "epoch": 2.103091697536825,
      "grad_norm": 0.05781097887935857,
      "learning_rate": 0.00012898117967622547,
      "loss": 0.8414,
      "step": 119790
    },
    {
      "epoch": 2.1032672624168263,
      "grad_norm": 0.07698853382539673,
      "learning_rate": 0.0001289704902301915,
      "loss": 0.8497,
      "step": 119800
    },
    {
      "epoch": 2.1034428272968277,
      "grad_norm": 0.044097110603553175,
      "learning_rate": 0.00012895980042976603,
      "loss": 0.845,
      "step": 119810
    },
    {
      "epoch": 2.1036183921768288,
      "grad_norm": 0.05465333765803815,
      "learning_rate": 0.00012894911027508458,
      "loss": 0.8408,
      "step": 119820
    },
    {
      "epoch": 2.1037939570568303,
      "grad_norm": 0.04826868810991105,
      "learning_rate": 0.00012893841976628252,
      "loss": 0.8507,
      "step": 119830
    },
    {
      "epoch": 2.1039695219368317,
      "grad_norm": 0.051415061001360486,
      "learning_rate": 0.00012892772890349532,
      "loss": 0.8449,
      "step": 119840
    },
    {
      "epoch": 2.104145086816833,
      "grad_norm": 0.06598451844707298,
      "learning_rate": 0.00012891703768685842,
      "loss": 0.8549,
      "step": 119850
    },
    {
      "epoch": 2.1043206516968347,
      "grad_norm": 0.05332056004641161,
      "learning_rate": 0.00012890634611650736,
      "loss": 0.8459,
      "step": 119860
    },
    {
      "epoch": 2.104496216576836,
      "grad_norm": 0.06793736112583504,
      "learning_rate": 0.0001288956541925775,
      "loss": 0.8498,
      "step": 119870
    },
    {
      "epoch": 2.104671781456837,
      "grad_norm": 0.06387928480327408,
      "learning_rate": 0.00012888496191520437,
      "loss": 0.8435,
      "step": 119880
    },
    {
      "epoch": 2.1048473463368387,
      "grad_norm": 0.056989975833406994,
      "learning_rate": 0.0001288742692845234,
      "loss": 0.8503,
      "step": 119890
    },
    {
      "epoch": 2.10502291121684,
      "grad_norm": 0.059923914731861906,
      "learning_rate": 0.00012886357630067017,
      "loss": 0.8463,
      "step": 119900
    },
    {
      "epoch": 2.1051984760968416,
      "grad_norm": 0.08033871531544783,
      "learning_rate": 0.00012885288296378005,
      "loss": 0.8515,
      "step": 119910
    },
    {
      "epoch": 2.105374040976843,
      "grad_norm": 0.04665483280237207,
      "learning_rate": 0.00012884218927398854,
      "loss": 0.837,
      "step": 119920
    },
    {
      "epoch": 2.1055496058568446,
      "grad_norm": 0.05424304557822585,
      "learning_rate": 0.00012883149523143115,
      "loss": 0.8501,
      "step": 119930
    },
    {
      "epoch": 2.1057251707368456,
      "grad_norm": 0.06732183700281143,
      "learning_rate": 0.0001288208008362434,
      "loss": 0.8426,
      "step": 119940
    },
    {
      "epoch": 2.105900735616847,
      "grad_norm": 0.0719062333884648,
      "learning_rate": 0.00012881010608856082,
      "loss": 0.846,
      "step": 119950
    },
    {
      "epoch": 2.1060763004968486,
      "grad_norm": 0.0493532925888085,
      "learning_rate": 0.0001287994109885188,
      "loss": 0.8404,
      "step": 119960
    },
    {
      "epoch": 2.10625186537685,
      "grad_norm": 0.047877687344783953,
      "learning_rate": 0.00012878871553625296,
      "loss": 0.8537,
      "step": 119970
    },
    {
      "epoch": 2.1064274302568515,
      "grad_norm": 0.05949690335462631,
      "learning_rate": 0.00012877801973189872,
      "loss": 0.8543,
      "step": 119980
    },
    {
      "epoch": 2.106602995136853,
      "grad_norm": 0.04639837670129455,
      "learning_rate": 0.0001287673235755917,
      "loss": 0.8395,
      "step": 119990
    },
    {
      "epoch": 2.106778560016854,
      "grad_norm": 0.07720085082431946,
      "learning_rate": 0.00012875662706746734,
      "loss": 0.851,
      "step": 120000
    },
    {
      "epoch": 2.1069541248968555,
      "grad_norm": 0.11848386472478743,
      "learning_rate": 0.00012874593020766118,
      "loss": 0.8472,
      "step": 120010
    },
    {
      "epoch": 2.107129689776857,
      "grad_norm": 0.0775428983436027,
      "learning_rate": 0.00012873523299630882,
      "loss": 0.8474,
      "step": 120020
    },
    {
      "epoch": 2.1073052546568585,
      "grad_norm": 0.04961760284948472,
      "learning_rate": 0.00012872453543354568,
      "loss": 0.8453,
      "step": 120030
    },
    {
      "epoch": 2.10748081953686,
      "grad_norm": 0.0697490045795182,
      "learning_rate": 0.00012871383751950739,
      "loss": 0.8477,
      "step": 120040
    },
    {
      "epoch": 2.1076563844168614,
      "grad_norm": 0.05116042415217239,
      "learning_rate": 0.00012870313925432945,
      "loss": 0.8452,
      "step": 120050
    },
    {
      "epoch": 2.1078319492968625,
      "grad_norm": 0.07631964339491805,
      "learning_rate": 0.00012869244063814744,
      "loss": 0.846,
      "step": 120060
    },
    {
      "epoch": 2.108007514176864,
      "grad_norm": 0.09231692875797248,
      "learning_rate": 0.00012868174167109694,
      "loss": 0.8512,
      "step": 120070
    },
    {
      "epoch": 2.1081830790568654,
      "grad_norm": 0.06201152392546709,
      "learning_rate": 0.00012867104235331342,
      "loss": 0.8348,
      "step": 120080
    },
    {
      "epoch": 2.108358643936867,
      "grad_norm": 0.05149266023143844,
      "learning_rate": 0.00012866034268493252,
      "loss": 0.8436,
      "step": 120090
    },
    {
      "epoch": 2.1085342088168684,
      "grad_norm": 0.05492878981827273,
      "learning_rate": 0.00012864964266608972,
      "loss": 0.854,
      "step": 120100
    },
    {
      "epoch": 2.10870977369687,
      "grad_norm": 0.0652201064531521,
      "learning_rate": 0.0001286389422969207,
      "loss": 0.8479,
      "step": 120110
    },
    {
      "epoch": 2.108885338576871,
      "grad_norm": 0.05323424349719322,
      "learning_rate": 0.00012862824157756098,
      "loss": 0.8421,
      "step": 120120
    },
    {
      "epoch": 2.1090609034568724,
      "grad_norm": 0.055684206972616855,
      "learning_rate": 0.0001286175405081461,
      "loss": 0.8401,
      "step": 120130
    },
    {
      "epoch": 2.109236468336874,
      "grad_norm": 0.06254448674136937,
      "learning_rate": 0.00012860683908881175,
      "loss": 0.8505,
      "step": 120140
    },
    {
      "epoch": 2.1094120332168753,
      "grad_norm": 0.05958595786606714,
      "learning_rate": 0.00012859613731969343,
      "loss": 0.8438,
      "step": 120150
    },
    {
      "epoch": 2.109587598096877,
      "grad_norm": 0.05400044598537587,
      "learning_rate": 0.00012858543520092677,
      "loss": 0.8482,
      "step": 120160
    },
    {
      "epoch": 2.1097631629768783,
      "grad_norm": 0.06617817637436536,
      "learning_rate": 0.00012857473273264735,
      "loss": 0.842,
      "step": 120170
    },
    {
      "epoch": 2.1099387278568793,
      "grad_norm": 0.06127411261332291,
      "learning_rate": 0.00012856402991499082,
      "loss": 0.8437,
      "step": 120180
    },
    {
      "epoch": 2.110114292736881,
      "grad_norm": 0.05767425611677834,
      "learning_rate": 0.0001285533267480927,
      "loss": 0.8487,
      "step": 120190
    },
    {
      "epoch": 2.1102898576168823,
      "grad_norm": 0.0804391859803979,
      "learning_rate": 0.00012854262323208862,
      "loss": 0.847,
      "step": 120200
    },
    {
      "epoch": 2.1104654224968837,
      "grad_norm": 0.055995745240848084,
      "learning_rate": 0.00012853191936711432,
      "loss": 0.8496,
      "step": 120210
    },
    {
      "epoch": 2.110640987376885,
      "grad_norm": 0.06374635922692969,
      "learning_rate": 0.00012852121515330523,
      "loss": 0.8481,
      "step": 120220
    },
    {
      "epoch": 2.1108165522568867,
      "grad_norm": 0.05526077787021195,
      "learning_rate": 0.0001285105105907971,
      "loss": 0.845,
      "step": 120230
    },
    {
      "epoch": 2.1109921171368877,
      "grad_norm": 0.05283924750285396,
      "learning_rate": 0.0001284998056797256,
      "loss": 0.8514,
      "step": 120240
    },
    {
      "epoch": 2.111167682016889,
      "grad_norm": 0.06821721565522322,
      "learning_rate": 0.0001284891004202262,
      "loss": 0.8413,
      "step": 120250
    },
    {
      "epoch": 2.1113432468968907,
      "grad_norm": 0.05323742140707625,
      "learning_rate": 0.00012847839481243466,
      "loss": 0.8459,
      "step": 120260
    },
    {
      "epoch": 2.111518811776892,
      "grad_norm": 0.047503945701279894,
      "learning_rate": 0.0001284676888564866,
      "loss": 0.8392,
      "step": 120270
    },
    {
      "epoch": 2.1116943766568936,
      "grad_norm": 0.05091347852080487,
      "learning_rate": 0.00012845698255251764,
      "loss": 0.8511,
      "step": 120280
    },
    {
      "epoch": 2.111869941536895,
      "grad_norm": 0.0514573703576317,
      "learning_rate": 0.00012844627590066348,
      "loss": 0.8499,
      "step": 120290
    },
    {
      "epoch": 2.112045506416896,
      "grad_norm": 0.06635009141854468,
      "learning_rate": 0.0001284355689010597,
      "loss": 0.8435,
      "step": 120300
    },
    {
      "epoch": 2.1122210712968976,
      "grad_norm": 0.04800060241562691,
      "learning_rate": 0.00012842486155384205,
      "loss": 0.8448,
      "step": 120310
    },
    {
      "epoch": 2.112396636176899,
      "grad_norm": 0.0499585480168948,
      "learning_rate": 0.00012841415385914613,
      "loss": 0.8498,
      "step": 120320
    },
    {
      "epoch": 2.1125722010569006,
      "grad_norm": 0.05559164721219993,
      "learning_rate": 0.00012840344581710765,
      "loss": 0.8532,
      "step": 120330
    },
    {
      "epoch": 2.112747765936902,
      "grad_norm": 0.044874555252453374,
      "learning_rate": 0.0001283927374278622,
      "loss": 0.8531,
      "step": 120340
    },
    {
      "epoch": 2.1129233308169035,
      "grad_norm": 0.06833185160794958,
      "learning_rate": 0.00012838202869154558,
      "loss": 0.8507,
      "step": 120350
    },
    {
      "epoch": 2.113098895696905,
      "grad_norm": 0.05370908769795035,
      "learning_rate": 0.00012837131960829337,
      "loss": 0.8461,
      "step": 120360
    },
    {
      "epoch": 2.113274460576906,
      "grad_norm": 0.05614456972074223,
      "learning_rate": 0.0001283606101782413,
      "loss": 0.851,
      "step": 120370
    },
    {
      "epoch": 2.1134500254569075,
      "grad_norm": 0.053994180132530795,
      "learning_rate": 0.00012834990040152503,
      "loss": 0.8455,
      "step": 120380
    },
    {
      "epoch": 2.113625590336909,
      "grad_norm": 0.05872084053861498,
      "learning_rate": 0.00012833919027828032,
      "loss": 0.8454,
      "step": 120390
    },
    {
      "epoch": 2.1138011552169105,
      "grad_norm": 0.05150998789942413,
      "learning_rate": 0.0001283284798086428,
      "loss": 0.8418,
      "step": 120400
    },
    {
      "epoch": 2.113976720096912,
      "grad_norm": 0.05712309312530378,
      "learning_rate": 0.00012831776899274825,
      "loss": 0.84,
      "step": 120410
    },
    {
      "epoch": 2.1141522849769134,
      "grad_norm": 0.059926695723715825,
      "learning_rate": 0.00012830705783073229,
      "loss": 0.849,
      "step": 120420
    },
    {
      "epoch": 2.1143278498569145,
      "grad_norm": 0.04469794571336374,
      "learning_rate": 0.00012829634632273069,
      "loss": 0.8486,
      "step": 120430
    },
    {
      "epoch": 2.114503414736916,
      "grad_norm": 0.06053941460389534,
      "learning_rate": 0.00012828563446887916,
      "loss": 0.8421,
      "step": 120440
    },
    {
      "epoch": 2.1146789796169174,
      "grad_norm": 0.0680898910302209,
      "learning_rate": 0.00012827492226931338,
      "loss": 0.8396,
      "step": 120450
    },
    {
      "epoch": 2.114854544496919,
      "grad_norm": 0.0484127206714241,
      "learning_rate": 0.00012826420972416913,
      "loss": 0.8385,
      "step": 120460
    },
    {
      "epoch": 2.1150301093769204,
      "grad_norm": 0.04654008738403729,
      "learning_rate": 0.00012825349683358214,
      "loss": 0.8426,
      "step": 120470
    },
    {
      "epoch": 2.115205674256922,
      "grad_norm": 0.05733258067605185,
      "learning_rate": 0.00012824278359768808,
      "loss": 0.8467,
      "step": 120480
    },
    {
      "epoch": 2.115381239136923,
      "grad_norm": 0.0533479426204141,
      "learning_rate": 0.00012823207001662275,
      "loss": 0.8445,
      "step": 120490
    },
    {
      "epoch": 2.1155568040169244,
      "grad_norm": 0.092325818912977,
      "learning_rate": 0.00012822135609052187,
      "loss": 0.8389,
      "step": 120500
    },
    {
      "epoch": 2.115732368896926,
      "grad_norm": 0.07534653691018381,
      "learning_rate": 0.0001282106418195212,
      "loss": 0.8357,
      "step": 120510
    },
    {
      "epoch": 2.1159079337769273,
      "grad_norm": 0.04871616304876778,
      "learning_rate": 0.0001281999272037565,
      "loss": 0.8448,
      "step": 120520
    },
    {
      "epoch": 2.116083498656929,
      "grad_norm": 0.04837426104389736,
      "learning_rate": 0.00012818921224336347,
      "loss": 0.8455,
      "step": 120530
    },
    {
      "epoch": 2.1162590635369303,
      "grad_norm": 0.05471868937951037,
      "learning_rate": 0.00012817849693847792,
      "loss": 0.8469,
      "step": 120540
    },
    {
      "epoch": 2.1164346284169313,
      "grad_norm": 0.05326301528890997,
      "learning_rate": 0.0001281677812892356,
      "loss": 0.8485,
      "step": 120550
    },
    {
      "epoch": 2.116610193296933,
      "grad_norm": 0.04789783484648204,
      "learning_rate": 0.00012815706529577232,
      "loss": 0.8487,
      "step": 120560
    },
    {
      "epoch": 2.1167857581769343,
      "grad_norm": 0.04925082366714438,
      "learning_rate": 0.00012814634895822377,
      "loss": 0.8422,
      "step": 120570
    },
    {
      "epoch": 2.1169613230569357,
      "grad_norm": 0.04421277663818635,
      "learning_rate": 0.00012813563227672585,
      "loss": 0.8515,
      "step": 120580
    },
    {
      "epoch": 2.117136887936937,
      "grad_norm": 0.05002867921670386,
      "learning_rate": 0.0001281249152514142,
      "loss": 0.844,
      "step": 120590
    },
    {
      "epoch": 2.1173124528169387,
      "grad_norm": 0.0498596924182204,
      "learning_rate": 0.0001281141978824247,
      "loss": 0.8442,
      "step": 120600
    },
    {
      "epoch": 2.1174880176969397,
      "grad_norm": 0.0613022116747638,
      "learning_rate": 0.0001281034801698931,
      "loss": 0.8418,
      "step": 120610
    },
    {
      "epoch": 2.117663582576941,
      "grad_norm": 0.0557323827904395,
      "learning_rate": 0.0001280927621139552,
      "loss": 0.835,
      "step": 120620
    },
    {
      "epoch": 2.1178391474569427,
      "grad_norm": 0.052802398628669076,
      "learning_rate": 0.00012808204371474687,
      "loss": 0.8392,
      "step": 120630
    },
    {
      "epoch": 2.118014712336944,
      "grad_norm": 0.05345063534616969,
      "learning_rate": 0.0001280713249724038,
      "loss": 0.8482,
      "step": 120640
    },
    {
      "epoch": 2.1181902772169456,
      "grad_norm": 0.06146079217569216,
      "learning_rate": 0.00012806060588706187,
      "loss": 0.8462,
      "step": 120650
    },
    {
      "epoch": 2.118365842096947,
      "grad_norm": 0.08142762403119296,
      "learning_rate": 0.00012804988645885687,
      "loss": 0.8375,
      "step": 120660
    },
    {
      "epoch": 2.118541406976948,
      "grad_norm": 0.04707940276618875,
      "learning_rate": 0.00012803916668792465,
      "loss": 0.8433,
      "step": 120670
    },
    {
      "epoch": 2.1187169718569496,
      "grad_norm": 0.06834024825386517,
      "learning_rate": 0.000128028446574401,
      "loss": 0.8418,
      "step": 120680
    },
    {
      "epoch": 2.118892536736951,
      "grad_norm": 0.06456022619399358,
      "learning_rate": 0.00012801772611842173,
      "loss": 0.8524,
      "step": 120690
    },
    {
      "epoch": 2.1190681016169526,
      "grad_norm": 0.07876484514183339,
      "learning_rate": 0.00012800700532012273,
      "loss": 0.8498,
      "step": 120700
    },
    {
      "epoch": 2.119243666496954,
      "grad_norm": 0.05002167554150724,
      "learning_rate": 0.00012799628417963977,
      "loss": 0.8397,
      "step": 120710
    },
    {
      "epoch": 2.1194192313769555,
      "grad_norm": 0.05972542470910255,
      "learning_rate": 0.00012798556269710872,
      "loss": 0.8491,
      "step": 120720
    },
    {
      "epoch": 2.1195947962569566,
      "grad_norm": 0.059743684518182724,
      "learning_rate": 0.0001279748408726654,
      "loss": 0.8442,
      "step": 120730
    },
    {
      "epoch": 2.119770361136958,
      "grad_norm": 0.05346207544712484,
      "learning_rate": 0.00012796411870644568,
      "loss": 0.8458,
      "step": 120740
    },
    {
      "epoch": 2.1199459260169595,
      "grad_norm": 0.051467842497595354,
      "learning_rate": 0.00012795339619858543,
      "loss": 0.8449,
      "step": 120750
    },
    {
      "epoch": 2.120121490896961,
      "grad_norm": 0.05502775104849383,
      "learning_rate": 0.0001279426733492205,
      "loss": 0.8518,
      "step": 120760
    },
    {
      "epoch": 2.1202970557769625,
      "grad_norm": 0.05448512029548603,
      "learning_rate": 0.0001279319501584867,
      "loss": 0.8446,
      "step": 120770
    },
    {
      "epoch": 2.120472620656964,
      "grad_norm": 0.04496531958479392,
      "learning_rate": 0.00012792122662651993,
      "loss": 0.8441,
      "step": 120780
    },
    {
      "epoch": 2.120648185536965,
      "grad_norm": 0.0520769890161646,
      "learning_rate": 0.00012791050275345607,
      "loss": 0.8468,
      "step": 120790
    },
    {
      "epoch": 2.1208237504169665,
      "grad_norm": 0.04185506043679376,
      "learning_rate": 0.000127899778539431,
      "loss": 0.8493,
      "step": 120800
    },
    {
      "epoch": 2.120999315296968,
      "grad_norm": 0.06324232729460372,
      "learning_rate": 0.00012788905398458054,
      "loss": 0.8475,
      "step": 120810
    },
    {
      "epoch": 2.1211748801769694,
      "grad_norm": 0.04787170721276395,
      "learning_rate": 0.00012787832908904068,
      "loss": 0.8431,
      "step": 120820
    },
    {
      "epoch": 2.121350445056971,
      "grad_norm": 0.05809126716204946,
      "learning_rate": 0.00012786760385294715,
      "loss": 0.841,
      "step": 120830
    },
    {
      "epoch": 2.1215260099369724,
      "grad_norm": 0.05847671316649162,
      "learning_rate": 0.000127856878276436,
      "loss": 0.845,
      "step": 120840
    },
    {
      "epoch": 2.1217015748169734,
      "grad_norm": 0.045302080482299535,
      "learning_rate": 0.00012784615235964303,
      "loss": 0.8483,
      "step": 120850
    },
    {
      "epoch": 2.121877139696975,
      "grad_norm": 0.04410287494310764,
      "learning_rate": 0.00012783542610270417,
      "loss": 0.8441,
      "step": 120860
    },
    {
      "epoch": 2.1220527045769764,
      "grad_norm": 0.06286640068732423,
      "learning_rate": 0.00012782469950575534,
      "loss": 0.8425,
      "step": 120870
    },
    {
      "epoch": 2.122228269456978,
      "grad_norm": 0.06106314189582124,
      "learning_rate": 0.00012781397256893238,
      "loss": 0.8485,
      "step": 120880
    },
    {
      "epoch": 2.1224038343369793,
      "grad_norm": 0.05386118987079417,
      "learning_rate": 0.00012780324529237128,
      "loss": 0.8543,
      "step": 120890
    },
    {
      "epoch": 2.122579399216981,
      "grad_norm": 0.06865444486780099,
      "learning_rate": 0.0001277925176762079,
      "loss": 0.8439,
      "step": 120900
    },
    {
      "epoch": 2.122754964096982,
      "grad_norm": 0.04916098393037035,
      "learning_rate": 0.00012778178972057817,
      "loss": 0.8411,
      "step": 120910
    },
    {
      "epoch": 2.1229305289769833,
      "grad_norm": 0.08748638737033435,
      "learning_rate": 0.0001277710614256181,
      "loss": 0.8431,
      "step": 120920
    },
    {
      "epoch": 2.123106093856985,
      "grad_norm": 0.07121657338992499,
      "learning_rate": 0.0001277603327914635,
      "loss": 0.843,
      "step": 120930
    },
    {
      "epoch": 2.1232816587369863,
      "grad_norm": 0.056237356547752505,
      "learning_rate": 0.0001277496038182504,
      "loss": 0.8438,
      "step": 120940
    },
    {
      "epoch": 2.1234572236169877,
      "grad_norm": 0.05143683552673948,
      "learning_rate": 0.00012773887450611465,
      "loss": 0.8521,
      "step": 120950
    },
    {
      "epoch": 2.123632788496989,
      "grad_norm": 0.058638204337942035,
      "learning_rate": 0.00012772814485519223,
      "loss": 0.8463,
      "step": 120960
    },
    {
      "epoch": 2.1238083533769903,
      "grad_norm": 0.044994579788062386,
      "learning_rate": 0.0001277174148656191,
      "loss": 0.8415,
      "step": 120970
    },
    {
      "epoch": 2.1239839182569917,
      "grad_norm": 0.04931286165918005,
      "learning_rate": 0.00012770668453753123,
      "loss": 0.8516,
      "step": 120980
    },
    {
      "epoch": 2.124159483136993,
      "grad_norm": 0.06059540905291427,
      "learning_rate": 0.00012769595387106454,
      "loss": 0.8443,
      "step": 120990
    },
    {
      "epoch": 2.1243350480169947,
      "grad_norm": 0.04989914196665394,
      "learning_rate": 0.00012768522286635498,
      "loss": 0.8407,
      "step": 121000
    },
    {
      "epoch": 2.124510612896996,
      "grad_norm": 0.05699358562273494,
      "learning_rate": 0.00012767449152353857,
      "loss": 0.85,
      "step": 121010
    },
    {
      "epoch": 2.1246861777769976,
      "grad_norm": 0.07154571271018864,
      "learning_rate": 0.0001276637598427512,
      "loss": 0.8563,
      "step": 121020
    },
    {
      "epoch": 2.124861742656999,
      "grad_norm": 0.05102595240415249,
      "learning_rate": 0.00012765302782412888,
      "loss": 0.8481,
      "step": 121030
    },
    {
      "epoch": 2.125037307537,
      "grad_norm": 0.05911004215512329,
      "learning_rate": 0.00012764229546780763,
      "loss": 0.8433,
      "step": 121040
    },
    {
      "epoch": 2.1252128724170016,
      "grad_norm": 0.043139953510372905,
      "learning_rate": 0.00012763156277392334,
      "loss": 0.8411,
      "step": 121050
    },
    {
      "epoch": 2.125388437297003,
      "grad_norm": 0.04874995424381429,
      "learning_rate": 0.00012762082974261208,
      "loss": 0.8406,
      "step": 121060
    },
    {
      "epoch": 2.1255640021770046,
      "grad_norm": 0.05894145788657715,
      "learning_rate": 0.00012761009637400977,
      "loss": 0.8479,
      "step": 121070
    },
    {
      "epoch": 2.125739567057006,
      "grad_norm": 0.07746761075596896,
      "learning_rate": 0.00012759936266825247,
      "loss": 0.8506,
      "step": 121080
    },
    {
      "epoch": 2.125915131937007,
      "grad_norm": 0.08277006502975455,
      "learning_rate": 0.00012758862862547618,
      "loss": 0.8449,
      "step": 121090
    },
    {
      "epoch": 2.1260906968170086,
      "grad_norm": 0.05992735167901212,
      "learning_rate": 0.0001275778942458168,
      "loss": 0.8494,
      "step": 121100
    },
    {
      "epoch": 2.12626626169701,
      "grad_norm": 0.0571665284827362,
      "learning_rate": 0.00012756715952941047,
      "loss": 0.8452,
      "step": 121110
    },
    {
      "epoch": 2.1264418265770115,
      "grad_norm": 0.056317448253266726,
      "learning_rate": 0.00012755642447639308,
      "loss": 0.8439,
      "step": 121120
    },
    {
      "epoch": 2.126617391457013,
      "grad_norm": 0.06003236902337512,
      "learning_rate": 0.00012754568908690075,
      "loss": 0.85,
      "step": 121130
    },
    {
      "epoch": 2.1267929563370145,
      "grad_norm": 0.047557460797166724,
      "learning_rate": 0.0001275349533610694,
      "loss": 0.8446,
      "step": 121140
    },
    {
      "epoch": 2.126968521217016,
      "grad_norm": 0.054727725923606696,
      "learning_rate": 0.00012752421729903514,
      "loss": 0.8546,
      "step": 121150
    },
    {
      "epoch": 2.127144086097017,
      "grad_norm": 0.05374675776467751,
      "learning_rate": 0.00012751348090093398,
      "loss": 0.847,
      "step": 121160
    },
    {
      "epoch": 2.1273196509770185,
      "grad_norm": 0.06087439283356885,
      "learning_rate": 0.0001275027441669019,
      "loss": 0.8403,
      "step": 121170
    },
    {
      "epoch": 2.12749521585702,
      "grad_norm": 0.057487291774440216,
      "learning_rate": 0.00012749200709707498,
      "loss": 0.8474,
      "step": 121180
    },
    {
      "epoch": 2.1276707807370214,
      "grad_norm": 0.04907112720130079,
      "learning_rate": 0.00012748126969158925,
      "loss": 0.8411,
      "step": 121190
    },
    {
      "epoch": 2.127846345617023,
      "grad_norm": 0.06567519968936898,
      "learning_rate": 0.00012747053195058078,
      "loss": 0.8447,
      "step": 121200
    },
    {
      "epoch": 2.1280219104970244,
      "grad_norm": 0.0720864372133087,
      "learning_rate": 0.00012745979387418561,
      "loss": 0.8409,
      "step": 121210
    },
    {
      "epoch": 2.1281974753770254,
      "grad_norm": 0.053132547949080414,
      "learning_rate": 0.00012744905546253975,
      "loss": 0.8417,
      "step": 121220
    },
    {
      "epoch": 2.128373040257027,
      "grad_norm": 0.06458251270445044,
      "learning_rate": 0.0001274383167157793,
      "loss": 0.8468,
      "step": 121230
    },
    {
      "epoch": 2.1285486051370284,
      "grad_norm": 0.0636278241897984,
      "learning_rate": 0.00012742757763404026,
      "loss": 0.8487,
      "step": 121240
    },
    {
      "epoch": 2.12872417001703,
      "grad_norm": 0.052576106112149294,
      "learning_rate": 0.0001274168382174588,
      "loss": 0.8556,
      "step": 121250
    },
    {
      "epoch": 2.1288997348970313,
      "grad_norm": 0.06937656808786613,
      "learning_rate": 0.00012740609846617092,
      "loss": 0.8466,
      "step": 121260
    },
    {
      "epoch": 2.129075299777033,
      "grad_norm": 0.060072200441324566,
      "learning_rate": 0.00012739535838031274,
      "loss": 0.84,
      "step": 121270
    },
    {
      "epoch": 2.129250864657034,
      "grad_norm": 0.08293624581075938,
      "learning_rate": 0.0001273846179600203,
      "loss": 0.8429,
      "step": 121280
    },
    {
      "epoch": 2.1294264295370353,
      "grad_norm": 0.09017318650155191,
      "learning_rate": 0.00012737387720542965,
      "loss": 0.8427,
      "step": 121290
    },
    {
      "epoch": 2.129601994417037,
      "grad_norm": 0.06565781018814018,
      "learning_rate": 0.00012736313611667698,
      "loss": 0.8462,
      "step": 121300
    },
    {
      "epoch": 2.1297775592970383,
      "grad_norm": 0.053488956328955124,
      "learning_rate": 0.0001273523946938983,
      "loss": 0.8461,
      "step": 121310
    },
    {
      "epoch": 2.1299531241770397,
      "grad_norm": 0.058931913459728265,
      "learning_rate": 0.00012734165293722973,
      "loss": 0.8427,
      "step": 121320
    },
    {
      "epoch": 2.1301286890570412,
      "grad_norm": 0.06224932675176705,
      "learning_rate": 0.00012733091084680736,
      "loss": 0.8418,
      "step": 121330
    },
    {
      "epoch": 2.1303042539370423,
      "grad_norm": 0.04536557503075575,
      "learning_rate": 0.00012732016842276732,
      "loss": 0.8414,
      "step": 121340
    },
    {
      "epoch": 2.1304798188170437,
      "grad_norm": 0.04895075545377758,
      "learning_rate": 0.00012730942566524568,
      "loss": 0.8397,
      "step": 121350
    },
    {
      "epoch": 2.130655383697045,
      "grad_norm": 0.09879464323539412,
      "learning_rate": 0.00012729868257437858,
      "loss": 0.8472,
      "step": 121360
    },
    {
      "epoch": 2.1308309485770467,
      "grad_norm": 0.05833389434955276,
      "learning_rate": 0.00012728793915030214,
      "loss": 0.8458,
      "step": 121370
    },
    {
      "epoch": 2.131006513457048,
      "grad_norm": 0.048872779012008545,
      "learning_rate": 0.00012727719539315247,
      "loss": 0.8426,
      "step": 121380
    },
    {
      "epoch": 2.1311820783370496,
      "grad_norm": 0.05700759308312006,
      "learning_rate": 0.00012726645130306568,
      "loss": 0.8482,
      "step": 121390
    },
    {
      "epoch": 2.1313576432170507,
      "grad_norm": 0.06892270936933463,
      "learning_rate": 0.00012725570688017795,
      "loss": 0.8469,
      "step": 121400
    },
    {
      "epoch": 2.131533208097052,
      "grad_norm": 0.07467425315040163,
      "learning_rate": 0.00012724496212462535,
      "loss": 0.8444,
      "step": 121410
    },
    {
      "epoch": 2.1317087729770536,
      "grad_norm": 0.05680642434103941,
      "learning_rate": 0.0001272342170365441,
      "loss": 0.8444,
      "step": 121420
    },
    {
      "epoch": 2.131884337857055,
      "grad_norm": 0.08797298692466085,
      "learning_rate": 0.00012722347161607024,
      "loss": 0.8448,
      "step": 121430
    },
    {
      "epoch": 2.1320599027370566,
      "grad_norm": 0.06718051286637969,
      "learning_rate": 0.00012721272586334,
      "loss": 0.8496,
      "step": 121440
    },
    {
      "epoch": 2.132235467617058,
      "grad_norm": 0.06524732592239288,
      "learning_rate": 0.00012720197977848948,
      "loss": 0.8445,
      "step": 121450
    },
    {
      "epoch": 2.132411032497059,
      "grad_norm": 0.056767385445960605,
      "learning_rate": 0.00012719123336165487,
      "loss": 0.8448,
      "step": 121460
    },
    {
      "epoch": 2.1325865973770606,
      "grad_norm": 0.059069090892374436,
      "learning_rate": 0.00012718048661297231,
      "loss": 0.8472,
      "step": 121470
    },
    {
      "epoch": 2.132762162257062,
      "grad_norm": 0.07982275053857242,
      "learning_rate": 0.00012716973953257795,
      "loss": 0.8475,
      "step": 121480
    },
    {
      "epoch": 2.1329377271370635,
      "grad_norm": 0.053828731364708046,
      "learning_rate": 0.000127158992120608,
      "loss": 0.8538,
      "step": 121490
    },
    {
      "epoch": 2.133113292017065,
      "grad_norm": 0.053085481375353245,
      "learning_rate": 0.00012714824437719862,
      "loss": 0.8452,
      "step": 121500
    },
    {
      "epoch": 2.1332888568970665,
      "grad_norm": 0.05769245641347034,
      "learning_rate": 0.00012713749630248594,
      "loss": 0.849,
      "step": 121510
    },
    {
      "epoch": 2.1334644217770675,
      "grad_norm": 0.0628167361560404,
      "learning_rate": 0.00012712674789660617,
      "loss": 0.8477,
      "step": 121520
    },
    {
      "epoch": 2.133639986657069,
      "grad_norm": 0.051877009840877876,
      "learning_rate": 0.00012711599915969548,
      "loss": 0.8504,
      "step": 121530
    },
    {
      "epoch": 2.1338155515370705,
      "grad_norm": 0.0497730165284881,
      "learning_rate": 0.00012710525009189012,
      "loss": 0.8532,
      "step": 121540
    },
    {
      "epoch": 2.133991116417072,
      "grad_norm": 0.0750875159992833,
      "learning_rate": 0.00012709450069332625,
      "loss": 0.8394,
      "step": 121550
    },
    {
      "epoch": 2.1341666812970734,
      "grad_norm": 0.05030537431276347,
      "learning_rate": 0.00012708375096414,
      "loss": 0.8477,
      "step": 121560
    },
    {
      "epoch": 2.134342246177075,
      "grad_norm": 0.07515800141164694,
      "learning_rate": 0.00012707300090446764,
      "loss": 0.8404,
      "step": 121570
    },
    {
      "epoch": 2.134517811057076,
      "grad_norm": 0.0624666954611381,
      "learning_rate": 0.0001270622505144454,
      "loss": 0.8472,
      "step": 121580
    },
    {
      "epoch": 2.1346933759370774,
      "grad_norm": 0.06288740846719176,
      "learning_rate": 0.0001270514997942094,
      "loss": 0.8417,
      "step": 121590
    },
    {
      "epoch": 2.134868940817079,
      "grad_norm": 0.05040645760598401,
      "learning_rate": 0.00012704074874389596,
      "loss": 0.8455,
      "step": 121600
    },
    {
      "epoch": 2.1350445056970804,
      "grad_norm": 0.0532483718112495,
      "learning_rate": 0.0001270299973636412,
      "loss": 0.8379,
      "step": 121610
    },
    {
      "epoch": 2.135220070577082,
      "grad_norm": 0.052868114313295855,
      "learning_rate": 0.0001270192456535814,
      "loss": 0.85,
      "step": 121620
    },
    {
      "epoch": 2.1353956354570833,
      "grad_norm": 0.0510002351359912,
      "learning_rate": 0.00012700849361385277,
      "loss": 0.8456,
      "step": 121630
    },
    {
      "epoch": 2.1355712003370844,
      "grad_norm": 0.06452661195290794,
      "learning_rate": 0.00012699774124459156,
      "loss": 0.8512,
      "step": 121640
    },
    {
      "epoch": 2.135746765217086,
      "grad_norm": 0.04744673713645255,
      "learning_rate": 0.00012698698854593393,
      "loss": 0.844,
      "step": 121650
    },
    {
      "epoch": 2.1359223300970873,
      "grad_norm": 0.04913355677577753,
      "learning_rate": 0.00012697623551801622,
      "loss": 0.8366,
      "step": 121660
    },
    {
      "epoch": 2.136097894977089,
      "grad_norm": 0.05339341361307408,
      "learning_rate": 0.00012696548216097465,
      "loss": 0.8457,
      "step": 121670
    },
    {
      "epoch": 2.1362734598570903,
      "grad_norm": 0.05983972237817765,
      "learning_rate": 0.00012695472847494538,
      "loss": 0.8425,
      "step": 121680
    },
    {
      "epoch": 2.1364490247370918,
      "grad_norm": 0.060022299891216144,
      "learning_rate": 0.00012694397446006477,
      "loss": 0.8415,
      "step": 121690
    },
    {
      "epoch": 2.1366245896170932,
      "grad_norm": 0.045921953824434064,
      "learning_rate": 0.00012693322011646902,
      "loss": 0.8547,
      "step": 121700
    },
    {
      "epoch": 2.1368001544970943,
      "grad_norm": 0.05623920617071648,
      "learning_rate": 0.0001269224654442944,
      "loss": 0.8479,
      "step": 121710
    },
    {
      "epoch": 2.1369757193770957,
      "grad_norm": 0.05420956876919341,
      "learning_rate": 0.0001269117104436772,
      "loss": 0.8475,
      "step": 121720
    },
    {
      "epoch": 2.137151284257097,
      "grad_norm": 0.041691592224262004,
      "learning_rate": 0.0001269009551147536,
      "loss": 0.8493,
      "step": 121730
    },
    {
      "epoch": 2.1373268491370987,
      "grad_norm": 0.062381535569252834,
      "learning_rate": 0.00012689019945766,
      "loss": 0.845,
      "step": 121740
    },
    {
      "epoch": 2.1375024140171,
      "grad_norm": 0.049057228841299545,
      "learning_rate": 0.00012687944347253258,
      "loss": 0.8501,
      "step": 121750
    },
    {
      "epoch": 2.137677978897101,
      "grad_norm": 0.045106639520723475,
      "learning_rate": 0.00012686868715950763,
      "loss": 0.8385,
      "step": 121760
    },
    {
      "epoch": 2.1378535437771027,
      "grad_norm": 0.05267462105795237,
      "learning_rate": 0.00012685793051872153,
      "loss": 0.8495,
      "step": 121770
    },
    {
      "epoch": 2.138029108657104,
      "grad_norm": 0.06173883210335855,
      "learning_rate": 0.0001268471735503104,
      "loss": 0.8436,
      "step": 121780
    },
    {
      "epoch": 2.1382046735371056,
      "grad_norm": 0.053920561240786535,
      "learning_rate": 0.0001268364162544107,
      "loss": 0.8448,
      "step": 121790
    },
    {
      "epoch": 2.138380238417107,
      "grad_norm": 0.05012129913394474,
      "learning_rate": 0.00012682565863115862,
      "loss": 0.8445,
      "step": 121800
    },
    {
      "epoch": 2.1385558032971086,
      "grad_norm": 0.07831914220490918,
      "learning_rate": 0.00012681490068069056,
      "loss": 0.843,
      "step": 121810
    },
    {
      "epoch": 2.13873136817711,
      "grad_norm": 0.05983074888379308,
      "learning_rate": 0.00012680414240314266,
      "loss": 0.8415,
      "step": 121820
    },
    {
      "epoch": 2.138906933057111,
      "grad_norm": 0.05662485089780951,
      "learning_rate": 0.0001267933837986514,
      "loss": 0.8506,
      "step": 121830
    },
    {
      "epoch": 2.1390824979371126,
      "grad_norm": 0.05127954270965801,
      "learning_rate": 0.000126782624867353,
      "loss": 0.8418,
      "step": 121840
    },
    {
      "epoch": 2.139258062817114,
      "grad_norm": 0.056898847858386586,
      "learning_rate": 0.00012677186560938378,
      "loss": 0.8467,
      "step": 121850
    },
    {
      "epoch": 2.1394336276971155,
      "grad_norm": 0.041614681709839216,
      "learning_rate": 0.00012676110602488015,
      "loss": 0.8565,
      "step": 121860
    },
    {
      "epoch": 2.139609192577117,
      "grad_norm": 0.05065493048824311,
      "learning_rate": 0.0001267503461139783,
      "loss": 0.8385,
      "step": 121870
    },
    {
      "epoch": 2.139784757457118,
      "grad_norm": 0.07479262508633547,
      "learning_rate": 0.00012673958587681465,
      "loss": 0.8446,
      "step": 121880
    },
    {
      "epoch": 2.1399603223371195,
      "grad_norm": 0.05895899819262954,
      "learning_rate": 0.00012672882531352553,
      "loss": 0.8545,
      "step": 121890
    },
    {
      "epoch": 2.140135887217121,
      "grad_norm": 0.06897900497619065,
      "learning_rate": 0.00012671806442424724,
      "loss": 0.8461,
      "step": 121900
    },
    {
      "epoch": 2.1403114520971225,
      "grad_norm": 0.05971589701391146,
      "learning_rate": 0.00012670730320911617,
      "loss": 0.8451,
      "step": 121910
    },
    {
      "epoch": 2.140487016977124,
      "grad_norm": 0.057832430433844825,
      "learning_rate": 0.0001266965416682686,
      "loss": 0.8505,
      "step": 121920
    },
    {
      "epoch": 2.1406625818571254,
      "grad_norm": 0.043374746657688965,
      "learning_rate": 0.00012668577980184096,
      "loss": 0.8423,
      "step": 121930
    },
    {
      "epoch": 2.140838146737127,
      "grad_norm": 0.044150327259708265,
      "learning_rate": 0.00012667501760996958,
      "loss": 0.8471,
      "step": 121940
    },
    {
      "epoch": 2.141013711617128,
      "grad_norm": 0.05668230100539992,
      "learning_rate": 0.0001266642550927908,
      "loss": 0.8422,
      "step": 121950
    },
    {
      "epoch": 2.1411892764971294,
      "grad_norm": 0.04781377722242677,
      "learning_rate": 0.00012665349225044097,
      "loss": 0.8516,
      "step": 121960
    },
    {
      "epoch": 2.141364841377131,
      "grad_norm": 0.08308324231755497,
      "learning_rate": 0.00012664272908305646,
      "loss": 0.8476,
      "step": 121970
    },
    {
      "epoch": 2.1415404062571324,
      "grad_norm": 0.05400825809611451,
      "learning_rate": 0.0001266319655907737,
      "loss": 0.8424,
      "step": 121980
    },
    {
      "epoch": 2.141715971137134,
      "grad_norm": 0.06515672195379303,
      "learning_rate": 0.00012662120177372898,
      "loss": 0.8432,
      "step": 121990
    },
    {
      "epoch": 2.1418915360171353,
      "grad_norm": 0.06425784240917502,
      "learning_rate": 0.00012661043763205872,
      "loss": 0.8447,
      "step": 122000
    },
    {
      "epoch": 2.1420671008971364,
      "grad_norm": 0.05637481893243189,
      "learning_rate": 0.00012659967316589938,
      "loss": 0.8426,
      "step": 122010
    },
    {
      "epoch": 2.142242665777138,
      "grad_norm": 0.06109153155541413,
      "learning_rate": 0.00012658890837538719,
      "loss": 0.848,
      "step": 122020
    },
    {
      "epoch": 2.1424182306571393,
      "grad_norm": 0.07007960119523468,
      "learning_rate": 0.0001265781432606587,
      "loss": 0.8463,
      "step": 122030
    },
    {
      "epoch": 2.142593795537141,
      "grad_norm": 0.050094164044648115,
      "learning_rate": 0.00012656737782185018,
      "loss": 0.8373,
      "step": 122040
    },
    {
      "epoch": 2.1427693604171423,
      "grad_norm": 0.04658511953362383,
      "learning_rate": 0.0001265566120590981,
      "loss": 0.8453,
      "step": 122050
    },
    {
      "epoch": 2.1429449252971438,
      "grad_norm": 0.0555397156679044,
      "learning_rate": 0.00012654584597253886,
      "loss": 0.8376,
      "step": 122060
    },
    {
      "epoch": 2.143120490177145,
      "grad_norm": 0.05183612206850376,
      "learning_rate": 0.0001265350795623088,
      "loss": 0.8454,
      "step": 122070
    },
    {
      "epoch": 2.1432960550571463,
      "grad_norm": 0.05991558779222775,
      "learning_rate": 0.00012652431282854446,
      "loss": 0.8484,
      "step": 122080
    },
    {
      "epoch": 2.1434716199371477,
      "grad_norm": 0.04330223921377113,
      "learning_rate": 0.00012651354577138213,
      "loss": 0.8542,
      "step": 122090
    },
    {
      "epoch": 2.143647184817149,
      "grad_norm": 0.08281452034897126,
      "learning_rate": 0.00012650277839095831,
      "loss": 0.8467,
      "step": 122100
    },
    {
      "epoch": 2.1438227496971507,
      "grad_norm": 0.05261789885832064,
      "learning_rate": 0.00012649201068740945,
      "loss": 0.8438,
      "step": 122110
    },
    {
      "epoch": 2.143998314577152,
      "grad_norm": 0.05286165582929605,
      "learning_rate": 0.00012648124266087185,
      "loss": 0.8436,
      "step": 122120
    },
    {
      "epoch": 2.144173879457153,
      "grad_norm": 0.045449611343035025,
      "learning_rate": 0.00012647047431148207,
      "loss": 0.8472,
      "step": 122130
    },
    {
      "epoch": 2.1443494443371547,
      "grad_norm": 0.05957479969236508,
      "learning_rate": 0.00012645970563937648,
      "loss": 0.8418,
      "step": 122140
    },
    {
      "epoch": 2.144525009217156,
      "grad_norm": 0.04566570898303258,
      "learning_rate": 0.00012644893664469156,
      "loss": 0.8434,
      "step": 122150
    },
    {
      "epoch": 2.1447005740971576,
      "grad_norm": 0.06234597847940553,
      "learning_rate": 0.00012643816732756368,
      "loss": 0.8434,
      "step": 122160
    },
    {
      "epoch": 2.144876138977159,
      "grad_norm": 0.05077508017621042,
      "learning_rate": 0.00012642739768812942,
      "loss": 0.8505,
      "step": 122170
    },
    {
      "epoch": 2.1450517038571606,
      "grad_norm": 0.06115892797971896,
      "learning_rate": 0.00012641662772652513,
      "loss": 0.8427,
      "step": 122180
    },
    {
      "epoch": 2.1452272687371616,
      "grad_norm": 0.06418232687655422,
      "learning_rate": 0.00012640585744288723,
      "loss": 0.8441,
      "step": 122190
    },
    {
      "epoch": 2.145402833617163,
      "grad_norm": 0.056253494741729915,
      "learning_rate": 0.00012639508683735232,
      "loss": 0.8421,
      "step": 122200
    },
    {
      "epoch": 2.1455783984971646,
      "grad_norm": 0.052564378731507495,
      "learning_rate": 0.00012638431591005677,
      "loss": 0.8461,
      "step": 122210
    },
    {
      "epoch": 2.145753963377166,
      "grad_norm": 0.057955258975678575,
      "learning_rate": 0.0001263735446611371,
      "loss": 0.847,
      "step": 122220
    },
    {
      "epoch": 2.1459295282571675,
      "grad_norm": 0.057378961528014534,
      "learning_rate": 0.00012636277309072976,
      "loss": 0.8489,
      "step": 122230
    },
    {
      "epoch": 2.146105093137169,
      "grad_norm": 0.05269813119111132,
      "learning_rate": 0.00012635200119897117,
      "loss": 0.8465,
      "step": 122240
    },
    {
      "epoch": 2.14628065801717,
      "grad_norm": 0.054666038594552775,
      "learning_rate": 0.0001263412289859979,
      "loss": 0.8406,
      "step": 122250
    },
    {
      "epoch": 2.1464562228971715,
      "grad_norm": 0.05511157330298314,
      "learning_rate": 0.00012633045645194638,
      "loss": 0.851,
      "step": 122260
    },
    {
      "epoch": 2.146631787777173,
      "grad_norm": 0.055120575672913875,
      "learning_rate": 0.0001263196835969531,
      "loss": 0.8509,
      "step": 122270
    },
    {
      "epoch": 2.1468073526571745,
      "grad_norm": 0.048713549072839105,
      "learning_rate": 0.0001263089104211547,
      "loss": 0.8489,
      "step": 122280
    },
    {
      "epoch": 2.146982917537176,
      "grad_norm": 0.06139539222663221,
      "learning_rate": 0.00012629813692468744,
      "loss": 0.8469,
      "step": 122290
    },
    {
      "epoch": 2.1471584824171774,
      "grad_norm": 0.04720071451965171,
      "learning_rate": 0.00012628736310768795,
      "loss": 0.84,
      "step": 122300
    },
    {
      "epoch": 2.1473340472971785,
      "grad_norm": 0.059677953320660726,
      "learning_rate": 0.00012627658897029277,
      "loss": 0.8494,
      "step": 122310
    },
    {
      "epoch": 2.14750961217718,
      "grad_norm": 0.05094274065752045,
      "learning_rate": 0.00012626581451263832,
      "loss": 0.8459,
      "step": 122320
    },
    {
      "epoch": 2.1476851770571814,
      "grad_norm": 0.056672922249999376,
      "learning_rate": 0.00012625503973486116,
      "loss": 0.8467,
      "step": 122330
    },
    {
      "epoch": 2.147860741937183,
      "grad_norm": 0.052459868051554524,
      "learning_rate": 0.00012624426463709782,
      "loss": 0.8394,
      "step": 122340
    },
    {
      "epoch": 2.1480363068171844,
      "grad_norm": 0.08510818871732356,
      "learning_rate": 0.00012623348921948484,
      "loss": 0.844,
      "step": 122350
    },
    {
      "epoch": 2.148211871697186,
      "grad_norm": 0.05583045469835623,
      "learning_rate": 0.00012622271348215866,
      "loss": 0.844,
      "step": 122360
    },
    {
      "epoch": 2.148387436577187,
      "grad_norm": 0.04652216579151652,
      "learning_rate": 0.0001262119374252559,
      "loss": 0.8489,
      "step": 122370
    },
    {
      "epoch": 2.1485630014571884,
      "grad_norm": 0.059556016936233934,
      "learning_rate": 0.00012620116104891305,
      "loss": 0.8479,
      "step": 122380
    },
    {
      "epoch": 2.14873856633719,
      "grad_norm": 0.04518820831933173,
      "learning_rate": 0.00012619038435326664,
      "loss": 0.8478,
      "step": 122390
    },
    {
      "epoch": 2.1489141312171913,
      "grad_norm": 0.08706766072593099,
      "learning_rate": 0.00012617960733845326,
      "loss": 0.8463,
      "step": 122400
    },
    {
      "epoch": 2.149089696097193,
      "grad_norm": 0.04382815815947329,
      "learning_rate": 0.0001261688300046094,
      "loss": 0.8426,
      "step": 122410
    },
    {
      "epoch": 2.1492652609771943,
      "grad_norm": 0.052601616389815,
      "learning_rate": 0.0001261580523518717,
      "loss": 0.8447,
      "step": 122420
    },
    {
      "epoch": 2.1494408258571953,
      "grad_norm": 0.09976124812563328,
      "learning_rate": 0.00012614727438037658,
      "loss": 0.8462,
      "step": 122430
    },
    {
      "epoch": 2.149616390737197,
      "grad_norm": 0.09997674844236455,
      "learning_rate": 0.00012613649609026072,
      "loss": 0.8447,
      "step": 122440
    },
    {
      "epoch": 2.1497919556171983,
      "grad_norm": 0.05008535556970504,
      "learning_rate": 0.00012612571748166066,
      "loss": 0.8509,
      "step": 122450
    },
    {
      "epoch": 2.1499675204971997,
      "grad_norm": 0.05406101587645634,
      "learning_rate": 0.0001261149385547129,
      "loss": 0.8471,
      "step": 122460
    },
    {
      "epoch": 2.1501430853772012,
      "grad_norm": 0.06892209918433047,
      "learning_rate": 0.00012610415930955406,
      "loss": 0.8468,
      "step": 122470
    },
    {
      "epoch": 2.1503186502572027,
      "grad_norm": 0.05942190181735698,
      "learning_rate": 0.0001260933797463207,
      "loss": 0.8434,
      "step": 122480
    },
    {
      "epoch": 2.150494215137204,
      "grad_norm": 0.06358348531901072,
      "learning_rate": 0.00012608259986514946,
      "loss": 0.8442,
      "step": 122490
    },
    {
      "epoch": 2.150669780017205,
      "grad_norm": 0.05656266479823076,
      "learning_rate": 0.00012607181966617682,
      "loss": 0.8444,
      "step": 122500
    },
    {
      "epoch": 2.1508453448972067,
      "grad_norm": 0.05708493733728223,
      "learning_rate": 0.00012606103914953947,
      "loss": 0.846,
      "step": 122510
    },
    {
      "epoch": 2.151020909777208,
      "grad_norm": 0.05812347736233087,
      "learning_rate": 0.00012605025831537392,
      "loss": 0.8453,
      "step": 122520
    },
    {
      "epoch": 2.1511964746572096,
      "grad_norm": 0.05197321013549247,
      "learning_rate": 0.00012603947716381678,
      "loss": 0.8506,
      "step": 122530
    },
    {
      "epoch": 2.151372039537211,
      "grad_norm": 0.04817383420698873,
      "learning_rate": 0.00012602869569500468,
      "loss": 0.8465,
      "step": 122540
    },
    {
      "epoch": 2.151547604417212,
      "grad_norm": 0.05916210859272396,
      "learning_rate": 0.0001260179139090742,
      "loss": 0.837,
      "step": 122550
    },
    {
      "epoch": 2.1517231692972136,
      "grad_norm": 0.06343025198297428,
      "learning_rate": 0.00012600713180616196,
      "loss": 0.8441,
      "step": 122560
    },
    {
      "epoch": 2.151898734177215,
      "grad_norm": 0.04953946538727044,
      "learning_rate": 0.00012599634938640462,
      "loss": 0.845,
      "step": 122570
    },
    {
      "epoch": 2.1520742990572166,
      "grad_norm": 0.04621058091292335,
      "learning_rate": 0.00012598556664993865,
      "loss": 0.8392,
      "step": 122580
    },
    {
      "epoch": 2.152249863937218,
      "grad_norm": 0.07060981525282273,
      "learning_rate": 0.00012597478359690082,
      "loss": 0.8462,
      "step": 122590
    },
    {
      "epoch": 2.1524254288172195,
      "grad_norm": 0.053263150881923176,
      "learning_rate": 0.0001259640002274277,
      "loss": 0.8407,
      "step": 122600
    },
    {
      "epoch": 2.152600993697221,
      "grad_norm": 0.0705206887480015,
      "learning_rate": 0.00012595321654165585,
      "loss": 0.8426,
      "step": 122610
    },
    {
      "epoch": 2.152776558577222,
      "grad_norm": 0.05640936225135474,
      "learning_rate": 0.00012594243253972207,
      "loss": 0.8369,
      "step": 122620
    },
    {
      "epoch": 2.1529521234572235,
      "grad_norm": 0.09945043045481991,
      "learning_rate": 0.00012593164822176283,
      "loss": 0.8494,
      "step": 122630
    },
    {
      "epoch": 2.153127688337225,
      "grad_norm": 0.05878893658114796,
      "learning_rate": 0.00012592086358791482,
      "loss": 0.8406,
      "step": 122640
    },
    {
      "epoch": 2.1533032532172265,
      "grad_norm": 0.04678317035051498,
      "learning_rate": 0.0001259100786383147,
      "loss": 0.8418,
      "step": 122650
    },
    {
      "epoch": 2.153478818097228,
      "grad_norm": 0.0459123683897107,
      "learning_rate": 0.00012589929337309912,
      "loss": 0.8519,
      "step": 122660
    },
    {
      "epoch": 2.1536543829772294,
      "grad_norm": 0.049907685630294595,
      "learning_rate": 0.0001258885077924047,
      "loss": 0.8484,
      "step": 122670
    },
    {
      "epoch": 2.1538299478572305,
      "grad_norm": 0.050935928531704984,
      "learning_rate": 0.00012587772189636814,
      "loss": 0.8487,
      "step": 122680
    },
    {
      "epoch": 2.154005512737232,
      "grad_norm": 0.07696801893654553,
      "learning_rate": 0.0001258669356851261,
      "loss": 0.8541,
      "step": 122690
    },
    {
      "epoch": 2.1541810776172334,
      "grad_norm": 0.0688353112860496,
      "learning_rate": 0.00012585614915881516,
      "loss": 0.8527,
      "step": 122700
    },
    {
      "epoch": 2.154356642497235,
      "grad_norm": 0.07577347980263846,
      "learning_rate": 0.00012584536231757208,
      "loss": 0.8402,
      "step": 122710
    },
    {
      "epoch": 2.1545322073772364,
      "grad_norm": 0.042024828477999994,
      "learning_rate": 0.00012583457516153346,
      "loss": 0.8422,
      "step": 122720
    },
    {
      "epoch": 2.154707772257238,
      "grad_norm": 0.04406056863707316,
      "learning_rate": 0.00012582378769083606,
      "loss": 0.8528,
      "step": 122730
    },
    {
      "epoch": 2.154883337137239,
      "grad_norm": 0.09021554353527346,
      "learning_rate": 0.0001258129999056165,
      "loss": 0.844,
      "step": 122740
    },
    {
      "epoch": 2.1550589020172404,
      "grad_norm": 0.05287361083521872,
      "learning_rate": 0.00012580221180601148,
      "loss": 0.8474,
      "step": 122750
    },
    {
      "epoch": 2.155234466897242,
      "grad_norm": 0.05473318537889592,
      "learning_rate": 0.00012579142339215767,
      "loss": 0.8539,
      "step": 122760
    },
    {
      "epoch": 2.1554100317772433,
      "grad_norm": 0.053829202439265866,
      "learning_rate": 0.00012578063466419176,
      "loss": 0.8438,
      "step": 122770
    },
    {
      "epoch": 2.155585596657245,
      "grad_norm": 0.0970537609150588,
      "learning_rate": 0.00012576984562225047,
      "loss": 0.8526,
      "step": 122780
    },
    {
      "epoch": 2.1557611615372463,
      "grad_norm": 0.062152059649286634,
      "learning_rate": 0.00012575905626647051,
      "loss": 0.8449,
      "step": 122790
    },
    {
      "epoch": 2.1559367264172473,
      "grad_norm": 0.08194447026435826,
      "learning_rate": 0.00012574826659698856,
      "loss": 0.8528,
      "step": 122800
    },
    {
      "epoch": 2.156112291297249,
      "grad_norm": 0.06555298533572194,
      "learning_rate": 0.0001257374766139413,
      "loss": 0.8498,
      "step": 122810
    },
    {
      "epoch": 2.1562878561772503,
      "grad_norm": 0.05450481350545309,
      "learning_rate": 0.0001257266863174655,
      "loss": 0.8446,
      "step": 122820
    },
    {
      "epoch": 2.1564634210572518,
      "grad_norm": 0.06537404862425224,
      "learning_rate": 0.00012571589570769783,
      "loss": 0.8438,
      "step": 122830
    },
    {
      "epoch": 2.1566389859372532,
      "grad_norm": 0.057752259951987864,
      "learning_rate": 0.00012570510478477503,
      "loss": 0.8419,
      "step": 122840
    },
    {
      "epoch": 2.1568145508172547,
      "grad_norm": 0.04934172814569647,
      "learning_rate": 0.00012569431354883382,
      "loss": 0.8444,
      "step": 122850
    },
    {
      "epoch": 2.1569901156972557,
      "grad_norm": 0.059858435091536995,
      "learning_rate": 0.00012568352200001092,
      "loss": 0.8547,
      "step": 122860
    },
    {
      "epoch": 2.157165680577257,
      "grad_norm": 0.0452909810479934,
      "learning_rate": 0.00012567273013844307,
      "loss": 0.8492,
      "step": 122870
    },
    {
      "epoch": 2.1573412454572587,
      "grad_norm": 0.053925688718312534,
      "learning_rate": 0.000125661937964267,
      "loss": 0.8478,
      "step": 122880
    },
    {
      "epoch": 2.15751681033726,
      "grad_norm": 0.045564125432769,
      "learning_rate": 0.00012565114547761943,
      "loss": 0.8492,
      "step": 122890
    },
    {
      "epoch": 2.1576923752172616,
      "grad_norm": 0.04653895956202837,
      "learning_rate": 0.00012564035267863713,
      "loss": 0.8452,
      "step": 122900
    },
    {
      "epoch": 2.157867940097263,
      "grad_norm": 0.06161567557037066,
      "learning_rate": 0.0001256295595674569,
      "loss": 0.8445,
      "step": 122910
    },
    {
      "epoch": 2.158043504977264,
      "grad_norm": 0.06476290698907873,
      "learning_rate": 0.00012561876614421535,
      "loss": 0.8474,
      "step": 122920
    },
    {
      "epoch": 2.1582190698572656,
      "grad_norm": 0.048937156795896865,
      "learning_rate": 0.00012560797240904935,
      "loss": 0.8464,
      "step": 122930
    },
    {
      "epoch": 2.158394634737267,
      "grad_norm": 0.05086648272963853,
      "learning_rate": 0.00012559717836209563,
      "loss": 0.8471,
      "step": 122940
    },
    {
      "epoch": 2.1585701996172686,
      "grad_norm": 0.04829066574372977,
      "learning_rate": 0.0001255863840034909,
      "loss": 0.8536,
      "step": 122950
    },
    {
      "epoch": 2.15874576449727,
      "grad_norm": 0.05765167798897579,
      "learning_rate": 0.00012557558933337204,
      "loss": 0.8404,
      "step": 122960
    },
    {
      "epoch": 2.1589213293772715,
      "grad_norm": 0.05030253783243984,
      "learning_rate": 0.0001255647943518757,
      "loss": 0.8442,
      "step": 122970
    },
    {
      "epoch": 2.1590968942572726,
      "grad_norm": 0.05767912040180794,
      "learning_rate": 0.00012555399905913873,
      "loss": 0.8448,
      "step": 122980
    },
    {
      "epoch": 2.159272459137274,
      "grad_norm": 0.0637417362077183,
      "learning_rate": 0.0001255432034552979,
      "loss": 0.8506,
      "step": 122990
    },
    {
      "epoch": 2.1594480240172755,
      "grad_norm": 0.08153551978695114,
      "learning_rate": 0.00012553240754048994,
      "loss": 0.8497,
      "step": 123000
    },
    {
      "epoch": 2.159623588897277,
      "grad_norm": 0.06675163498915841,
      "learning_rate": 0.0001255216113148517,
      "loss": 0.8464,
      "step": 123010
    },
    {
      "epoch": 2.1597991537772785,
      "grad_norm": 0.04587275272346365,
      "learning_rate": 0.00012551081477851994,
      "loss": 0.8432,
      "step": 123020
    },
    {
      "epoch": 2.15997471865728,
      "grad_norm": 0.06031215515388103,
      "learning_rate": 0.00012550001793163148,
      "loss": 0.8424,
      "step": 123030
    },
    {
      "epoch": 2.160150283537281,
      "grad_norm": 0.04516047979616498,
      "learning_rate": 0.00012548922077432306,
      "loss": 0.8477,
      "step": 123040
    },
    {
      "epoch": 2.1603258484172825,
      "grad_norm": 0.045989521734667624,
      "learning_rate": 0.00012547842330673157,
      "loss": 0.8457,
      "step": 123050
    },
    {
      "epoch": 2.160501413297284,
      "grad_norm": 0.04883277531094337,
      "learning_rate": 0.00012546762552899368,
      "loss": 0.8401,
      "step": 123060
    },
    {
      "epoch": 2.1606769781772854,
      "grad_norm": 0.049191545954761065,
      "learning_rate": 0.00012545682744124635,
      "loss": 0.8497,
      "step": 123070
    },
    {
      "epoch": 2.160852543057287,
      "grad_norm": 0.045593609210218926,
      "learning_rate": 0.00012544602904362634,
      "loss": 0.8453,
      "step": 123080
    },
    {
      "epoch": 2.1610281079372884,
      "grad_norm": 0.059883756748748095,
      "learning_rate": 0.00012543523033627042,
      "loss": 0.8485,
      "step": 123090
    },
    {
      "epoch": 2.1612036728172894,
      "grad_norm": 0.044976177361046046,
      "learning_rate": 0.00012542443131931548,
      "loss": 0.8425,
      "step": 123100
    },
    {
      "epoch": 2.161379237697291,
      "grad_norm": 0.06184425734460933,
      "learning_rate": 0.00012541363199289828,
      "loss": 0.8462,
      "step": 123110
    },
    {
      "epoch": 2.1615548025772924,
      "grad_norm": 0.05593073168708114,
      "learning_rate": 0.0001254028323571557,
      "loss": 0.8429,
      "step": 123120
    },
    {
      "epoch": 2.161730367457294,
      "grad_norm": 0.06971427864116034,
      "learning_rate": 0.00012539203241222456,
      "loss": 0.8482,
      "step": 123130
    },
    {
      "epoch": 2.1619059323372953,
      "grad_norm": 0.06242565953601442,
      "learning_rate": 0.0001253812321582417,
      "loss": 0.8477,
      "step": 123140
    },
    {
      "epoch": 2.162081497217297,
      "grad_norm": 0.04491850539715488,
      "learning_rate": 0.00012537043159534396,
      "loss": 0.8524,
      "step": 123150
    },
    {
      "epoch": 2.1622570620972983,
      "grad_norm": 0.05563688745234271,
      "learning_rate": 0.00012535963072366814,
      "loss": 0.85,
      "step": 123160
    },
    {
      "epoch": 2.1624326269772993,
      "grad_norm": 0.0611989365859918,
      "learning_rate": 0.0001253488295433512,
      "loss": 0.8424,
      "step": 123170
    },
    {
      "epoch": 2.162608191857301,
      "grad_norm": 0.056068738686743745,
      "learning_rate": 0.00012533802805452985,
      "loss": 0.8466,
      "step": 123180
    },
    {
      "epoch": 2.1627837567373023,
      "grad_norm": 0.057269354489344754,
      "learning_rate": 0.00012532722625734107,
      "loss": 0.8478,
      "step": 123190
    },
    {
      "epoch": 2.1629593216173038,
      "grad_norm": 0.06369443027272213,
      "learning_rate": 0.00012531642415192167,
      "loss": 0.8435,
      "step": 123200
    },
    {
      "epoch": 2.1631348864973052,
      "grad_norm": 0.06505753874853729,
      "learning_rate": 0.00012530562173840849,
      "loss": 0.8398,
      "step": 123210
    },
    {
      "epoch": 2.1633104513773063,
      "grad_norm": 0.05082455639676722,
      "learning_rate": 0.00012529481901693844,
      "loss": 0.8372,
      "step": 123220
    },
    {
      "epoch": 2.1634860162573077,
      "grad_norm": 0.04854103658516852,
      "learning_rate": 0.00012528401598764833,
      "loss": 0.8508,
      "step": 123230
    },
    {
      "epoch": 2.163661581137309,
      "grad_norm": 0.057505223975851436,
      "learning_rate": 0.00012527321265067516,
      "loss": 0.8406,
      "step": 123240
    },
    {
      "epoch": 2.1638371460173107,
      "grad_norm": 0.06693783669132056,
      "learning_rate": 0.0001252624090061557,
      "loss": 0.8504,
      "step": 123250
    },
    {
      "epoch": 2.164012710897312,
      "grad_norm": 0.07940519119044576,
      "learning_rate": 0.00012525160505422685,
      "loss": 0.8357,
      "step": 123260
    },
    {
      "epoch": 2.1641882757773137,
      "grad_norm": 0.0614080039982501,
      "learning_rate": 0.00012524080079502555,
      "loss": 0.8497,
      "step": 123270
    },
    {
      "epoch": 2.164363840657315,
      "grad_norm": 0.047181637722750726,
      "learning_rate": 0.00012522999622868863,
      "loss": 0.8446,
      "step": 123280
    },
    {
      "epoch": 2.164539405537316,
      "grad_norm": 0.07182754052455327,
      "learning_rate": 0.00012521919135535304,
      "loss": 0.849,
      "step": 123290
    },
    {
      "epoch": 2.1647149704173176,
      "grad_norm": 0.04454859955951449,
      "learning_rate": 0.00012520838617515565,
      "loss": 0.848,
      "step": 123300
    },
    {
      "epoch": 2.164890535297319,
      "grad_norm": 0.06574497102174334,
      "learning_rate": 0.00012519758068823333,
      "loss": 0.855,
      "step": 123310
    },
    {
      "epoch": 2.1650661001773206,
      "grad_norm": 0.06005489652659552,
      "learning_rate": 0.00012518677489472307,
      "loss": 0.8474,
      "step": 123320
    },
    {
      "epoch": 2.165241665057322,
      "grad_norm": 0.05388424570768922,
      "learning_rate": 0.0001251759687947617,
      "loss": 0.8455,
      "step": 123330
    },
    {
      "epoch": 2.165417229937323,
      "grad_norm": 0.059018544713113096,
      "learning_rate": 0.0001251651623884862,
      "loss": 0.8424,
      "step": 123340
    },
    {
      "epoch": 2.1655927948173246,
      "grad_norm": 0.05256314635782841,
      "learning_rate": 0.00012515435567603345,
      "loss": 0.8328,
      "step": 123350
    },
    {
      "epoch": 2.165768359697326,
      "grad_norm": 0.06056426446387385,
      "learning_rate": 0.00012514354865754038,
      "loss": 0.8379,
      "step": 123360
    },
    {
      "epoch": 2.1659439245773275,
      "grad_norm": 0.05477629759328827,
      "learning_rate": 0.00012513274133314395,
      "loss": 0.8423,
      "step": 123370
    },
    {
      "epoch": 2.166119489457329,
      "grad_norm": 0.059053477505771015,
      "learning_rate": 0.00012512193370298102,
      "loss": 0.8551,
      "step": 123380
    },
    {
      "epoch": 2.1662950543373305,
      "grad_norm": 0.06311385385411306,
      "learning_rate": 0.0001251111257671886,
      "loss": 0.8394,
      "step": 123390
    },
    {
      "epoch": 2.166470619217332,
      "grad_norm": 0.05822866871566747,
      "learning_rate": 0.00012510031752590358,
      "loss": 0.8476,
      "step": 123400
    },
    {
      "epoch": 2.166646184097333,
      "grad_norm": 0.06829294346106853,
      "learning_rate": 0.00012508950897926292,
      "loss": 0.8487,
      "step": 123410
    },
    {
      "epoch": 2.1668217489773345,
      "grad_norm": 0.06356539122858655,
      "learning_rate": 0.0001250787001274036,
      "loss": 0.8424,
      "step": 123420
    },
    {
      "epoch": 2.166997313857336,
      "grad_norm": 0.06478674937609141,
      "learning_rate": 0.00012506789097046247,
      "loss": 0.8448,
      "step": 123430
    },
    {
      "epoch": 2.1671728787373374,
      "grad_norm": 0.07808897287527729,
      "learning_rate": 0.0001250570815085766,
      "loss": 0.8519,
      "step": 123440
    },
    {
      "epoch": 2.167348443617339,
      "grad_norm": 0.05433814547957261,
      "learning_rate": 0.00012504627174188287,
      "loss": 0.8438,
      "step": 123450
    },
    {
      "epoch": 2.1675240084973404,
      "grad_norm": 0.0631556568939575,
      "learning_rate": 0.00012503546167051826,
      "loss": 0.8491,
      "step": 123460
    },
    {
      "epoch": 2.1676995733773414,
      "grad_norm": 0.05943117199218949,
      "learning_rate": 0.00012502465129461977,
      "loss": 0.8428,
      "step": 123470
    },
    {
      "epoch": 2.167875138257343,
      "grad_norm": 0.06507241623617935,
      "learning_rate": 0.00012501384061432434,
      "loss": 0.8451,
      "step": 123480
    },
    {
      "epoch": 2.1680507031373444,
      "grad_norm": 0.06361072455034496,
      "learning_rate": 0.00012500302962976893,
      "loss": 0.8517,
      "step": 123490
    },
    {
      "epoch": 2.168226268017346,
      "grad_norm": 0.06402255512225936,
      "learning_rate": 0.00012499221834109053,
      "loss": 0.8408,
      "step": 123500
    },
    {
      "epoch": 2.1684018328973473,
      "grad_norm": 0.05548907026850045,
      "learning_rate": 0.00012498140674842615,
      "loss": 0.8518,
      "step": 123510
    },
    {
      "epoch": 2.168577397777349,
      "grad_norm": 0.07947448772909059,
      "learning_rate": 0.00012497059485191273,
      "loss": 0.8436,
      "step": 123520
    },
    {
      "epoch": 2.16875296265735,
      "grad_norm": 0.045572009436122995,
      "learning_rate": 0.0001249597826516873,
      "loss": 0.84,
      "step": 123530
    },
    {
      "epoch": 2.1689285275373513,
      "grad_norm": 0.07358715312892573,
      "learning_rate": 0.00012494897014788683,
      "loss": 0.8442,
      "step": 123540
    },
    {
      "epoch": 2.169104092417353,
      "grad_norm": 0.06645140275702627,
      "learning_rate": 0.0001249381573406483,
      "loss": 0.8386,
      "step": 123550
    },
    {
      "epoch": 2.1692796572973543,
      "grad_norm": 0.05634174523086861,
      "learning_rate": 0.00012492734423010876,
      "loss": 0.8479,
      "step": 123560
    },
    {
      "epoch": 2.1694552221773558,
      "grad_norm": 0.05573696773519324,
      "learning_rate": 0.00012491653081640513,
      "loss": 0.846,
      "step": 123570
    },
    {
      "epoch": 2.1696307870573572,
      "grad_norm": 0.07663758710610388,
      "learning_rate": 0.00012490571709967453,
      "loss": 0.8493,
      "step": 123580
    },
    {
      "epoch": 2.1698063519373583,
      "grad_norm": 0.04700266760723505,
      "learning_rate": 0.0001248949030800539,
      "loss": 0.8463,
      "step": 123590
    },
    {
      "epoch": 2.1699819168173597,
      "grad_norm": 0.05225052774266155,
      "learning_rate": 0.00012488408875768023,
      "loss": 0.8463,
      "step": 123600
    },
    {
      "epoch": 2.1701574816973612,
      "grad_norm": 0.06193158481254446,
      "learning_rate": 0.00012487327413269062,
      "loss": 0.842,
      "step": 123610
    },
    {
      "epoch": 2.1703330465773627,
      "grad_norm": 0.04697554980432962,
      "learning_rate": 0.000124862459205222,
      "loss": 0.8545,
      "step": 123620
    },
    {
      "epoch": 2.170508611457364,
      "grad_norm": 0.056739524341885776,
      "learning_rate": 0.0001248516439754115,
      "loss": 0.8422,
      "step": 123630
    },
    {
      "epoch": 2.1706841763373657,
      "grad_norm": 0.08742530181387005,
      "learning_rate": 0.00012484082844339608,
      "loss": 0.8412,
      "step": 123640
    },
    {
      "epoch": 2.1708597412173667,
      "grad_norm": 0.07013446733397781,
      "learning_rate": 0.0001248300126093128,
      "loss": 0.8526,
      "step": 123650
    },
    {
      "epoch": 2.171035306097368,
      "grad_norm": 0.05553531618005677,
      "learning_rate": 0.00012481919647329873,
      "loss": 0.8492,
      "step": 123660
    },
    {
      "epoch": 2.1712108709773696,
      "grad_norm": 0.051912876363076794,
      "learning_rate": 0.0001248083800354908,
      "loss": 0.8449,
      "step": 123670
    },
    {
      "epoch": 2.171386435857371,
      "grad_norm": 0.06386722963633953,
      "learning_rate": 0.00012479756329602617,
      "loss": 0.8441,
      "step": 123680
    },
    {
      "epoch": 2.1715620007373726,
      "grad_norm": 0.045878547726121245,
      "learning_rate": 0.00012478674625504183,
      "loss": 0.8451,
      "step": 123690
    },
    {
      "epoch": 2.171737565617374,
      "grad_norm": 0.05947409673084285,
      "learning_rate": 0.0001247759289126749,
      "loss": 0.8488,
      "step": 123700
    },
    {
      "epoch": 2.171913130497375,
      "grad_norm": 0.05373187644567895,
      "learning_rate": 0.0001247651112690624,
      "loss": 0.8344,
      "step": 123710
    },
    {
      "epoch": 2.1720886953773766,
      "grad_norm": 0.06487902252994918,
      "learning_rate": 0.00012475429332434133,
      "loss": 0.841,
      "step": 123720
    },
    {
      "epoch": 2.172264260257378,
      "grad_norm": 0.05693690294324708,
      "learning_rate": 0.00012474347507864883,
      "loss": 0.8466,
      "step": 123730
    },
    {
      "epoch": 2.1724398251373795,
      "grad_norm": 0.06380406904541203,
      "learning_rate": 0.00012473265653212195,
      "loss": 0.8399,
      "step": 123740
    },
    {
      "epoch": 2.172615390017381,
      "grad_norm": 0.04708442362705157,
      "learning_rate": 0.0001247218376848978,
      "loss": 0.846,
      "step": 123750
    },
    {
      "epoch": 2.1727909548973825,
      "grad_norm": 0.05921448394526895,
      "learning_rate": 0.00012471101853711336,
      "loss": 0.8486,
      "step": 123760
    },
    {
      "epoch": 2.1729665197773835,
      "grad_norm": 0.06391038415863211,
      "learning_rate": 0.0001247001990889058,
      "loss": 0.845,
      "step": 123770
    },
    {
      "epoch": 2.173142084657385,
      "grad_norm": 0.09526912248201307,
      "learning_rate": 0.0001246893793404122,
      "loss": 0.8481,
      "step": 123780
    },
    {
      "epoch": 2.1733176495373865,
      "grad_norm": 0.0519164137466698,
      "learning_rate": 0.00012467855929176958,
      "loss": 0.8416,
      "step": 123790
    },
    {
      "epoch": 2.173493214417388,
      "grad_norm": 0.058377102564079186,
      "learning_rate": 0.00012466773894311513,
      "loss": 0.8476,
      "step": 123800
    },
    {
      "epoch": 2.1736687792973894,
      "grad_norm": 0.05677061349491721,
      "learning_rate": 0.00012465691829458583,
      "loss": 0.846,
      "step": 123810
    },
    {
      "epoch": 2.173844344177391,
      "grad_norm": 0.05369148146952836,
      "learning_rate": 0.0001246460973463189,
      "loss": 0.8518,
      "step": 123820
    },
    {
      "epoch": 2.174019909057392,
      "grad_norm": 0.05165790745881904,
      "learning_rate": 0.00012463527609845134,
      "loss": 0.8444,
      "step": 123830
    },
    {
      "epoch": 2.1741954739373934,
      "grad_norm": 0.060151186607304834,
      "learning_rate": 0.0001246244545511203,
      "loss": 0.843,
      "step": 123840
    },
    {
      "epoch": 2.174371038817395,
      "grad_norm": 0.06910895875769794,
      "learning_rate": 0.0001246136327044629,
      "loss": 0.8429,
      "step": 123850
    },
    {
      "epoch": 2.1745466036973964,
      "grad_norm": 0.056964762466887454,
      "learning_rate": 0.0001246028105586163,
      "loss": 0.8406,
      "step": 123860
    },
    {
      "epoch": 2.174722168577398,
      "grad_norm": 0.0542786903149581,
      "learning_rate": 0.00012459198811371752,
      "loss": 0.8513,
      "step": 123870
    },
    {
      "epoch": 2.1748977334573993,
      "grad_norm": 0.08090676072541725,
      "learning_rate": 0.00012458116536990374,
      "loss": 0.8434,
      "step": 123880
    },
    {
      "epoch": 2.1750732983374004,
      "grad_norm": 0.14160801463940786,
      "learning_rate": 0.00012457034232731207,
      "loss": 0.8471,
      "step": 123890
    },
    {
      "epoch": 2.175248863217402,
      "grad_norm": 0.07653226554955579,
      "learning_rate": 0.00012455951898607961,
      "loss": 0.8521,
      "step": 123900
    },
    {
      "epoch": 2.1754244280974033,
      "grad_norm": 0.05255207404237297,
      "learning_rate": 0.00012454869534634356,
      "loss": 0.8435,
      "step": 123910
    },
    {
      "epoch": 2.175599992977405,
      "grad_norm": 0.06543145640924555,
      "learning_rate": 0.00012453787140824103,
      "loss": 0.8497,
      "step": 123920
    },
    {
      "epoch": 2.1757755578574063,
      "grad_norm": 0.07256036919523151,
      "learning_rate": 0.00012452704717190918,
      "loss": 0.8552,
      "step": 123930
    },
    {
      "epoch": 2.1759511227374078,
      "grad_norm": 0.04723772983124345,
      "learning_rate": 0.0001245162226374851,
      "loss": 0.8472,
      "step": 123940
    },
    {
      "epoch": 2.1761266876174092,
      "grad_norm": 0.057120876966958285,
      "learning_rate": 0.00012450539780510598,
      "loss": 0.8441,
      "step": 123950
    },
    {
      "epoch": 2.1763022524974103,
      "grad_norm": 0.05362422174651468,
      "learning_rate": 0.000124494572674909,
      "loss": 0.8489,
      "step": 123960
    },
    {
      "epoch": 2.1764778173774117,
      "grad_norm": 0.07465561882241571,
      "learning_rate": 0.00012448374724703125,
      "loss": 0.8478,
      "step": 123970
    },
    {
      "epoch": 2.1766533822574132,
      "grad_norm": 0.045602415122536204,
      "learning_rate": 0.00012447292152160993,
      "loss": 0.8451,
      "step": 123980
    },
    {
      "epoch": 2.1768289471374147,
      "grad_norm": 0.04607392409797833,
      "learning_rate": 0.00012446209549878218,
      "loss": 0.8443,
      "step": 123990
    },
    {
      "epoch": 2.177004512017416,
      "grad_norm": 0.04890620727085508,
      "learning_rate": 0.0001244512691786852,
      "loss": 0.8475,
      "step": 124000
    },
    {
      "epoch": 2.177180076897417,
      "grad_norm": 0.054246675579755675,
      "learning_rate": 0.00012444044256145612,
      "loss": 0.8509,
      "step": 124010
    },
    {
      "epoch": 2.1773556417774187,
      "grad_norm": 0.04379280732011523,
      "learning_rate": 0.00012442961564723218,
      "loss": 0.8504,
      "step": 124020
    },
    {
      "epoch": 2.17753120665742,
      "grad_norm": 0.05093156004712589,
      "learning_rate": 0.00012441878843615046,
      "loss": 0.8582,
      "step": 124030
    },
    {
      "epoch": 2.1777067715374216,
      "grad_norm": 0.053671132305850695,
      "learning_rate": 0.00012440796092834828,
      "loss": 0.85,
      "step": 124040
    },
    {
      "epoch": 2.177882336417423,
      "grad_norm": 0.06431349002105705,
      "learning_rate": 0.0001243971331239627,
      "loss": 0.8478,
      "step": 124050
    },
    {
      "epoch": 2.1780579012974246,
      "grad_norm": 0.04986083115892917,
      "learning_rate": 0.00012438630502313095,
      "loss": 0.8486,
      "step": 124060
    },
    {
      "epoch": 2.178233466177426,
      "grad_norm": 0.05362001593394558,
      "learning_rate": 0.00012437547662599028,
      "loss": 0.8502,
      "step": 124070
    },
    {
      "epoch": 2.178409031057427,
      "grad_norm": 0.06353331901311271,
      "learning_rate": 0.0001243646479326778,
      "loss": 0.8417,
      "step": 124080
    },
    {
      "epoch": 2.1785845959374286,
      "grad_norm": 0.056419079152094306,
      "learning_rate": 0.00012435381894333075,
      "loss": 0.8449,
      "step": 124090
    },
    {
      "epoch": 2.17876016081743,
      "grad_norm": 0.07978454010103818,
      "learning_rate": 0.00012434298965808638,
      "loss": 0.8487,
      "step": 124100
    },
    {
      "epoch": 2.1789357256974315,
      "grad_norm": 0.048076991343196845,
      "learning_rate": 0.00012433216007708183,
      "loss": 0.8496,
      "step": 124110
    },
    {
      "epoch": 2.179111290577433,
      "grad_norm": 0.04703085929474038,
      "learning_rate": 0.00012432133020045435,
      "loss": 0.8498,
      "step": 124120
    },
    {
      "epoch": 2.1792868554574345,
      "grad_norm": 0.057300506462325144,
      "learning_rate": 0.00012431050002834111,
      "loss": 0.8524,
      "step": 124130
    },
    {
      "epoch": 2.1794624203374355,
      "grad_norm": 0.05639639576668472,
      "learning_rate": 0.00012429966956087942,
      "loss": 0.8466,
      "step": 124140
    },
    {
      "epoch": 2.179637985217437,
      "grad_norm": 0.04696763244393638,
      "learning_rate": 0.00012428883879820644,
      "loss": 0.8525,
      "step": 124150
    },
    {
      "epoch": 2.1798135500974385,
      "grad_norm": 0.06309630483796173,
      "learning_rate": 0.0001242780077404594,
      "loss": 0.851,
      "step": 124160
    },
    {
      "epoch": 2.17998911497744,
      "grad_norm": 0.06435072369298316,
      "learning_rate": 0.00012426717638777551,
      "loss": 0.8395,
      "step": 124170
    },
    {
      "epoch": 2.1801646798574414,
      "grad_norm": 0.09548616792871292,
      "learning_rate": 0.00012425634474029206,
      "loss": 0.8451,
      "step": 124180
    },
    {
      "epoch": 2.180340244737443,
      "grad_norm": 0.052627675024609744,
      "learning_rate": 0.00012424551279814627,
      "loss": 0.8521,
      "step": 124190
    },
    {
      "epoch": 2.180515809617444,
      "grad_norm": 0.04863765858051558,
      "learning_rate": 0.00012423468056147537,
      "loss": 0.8552,
      "step": 124200
    },
    {
      "epoch": 2.1806913744974454,
      "grad_norm": 0.06982796803111618,
      "learning_rate": 0.00012422384803041663,
      "loss": 0.8478,
      "step": 124210
    },
    {
      "epoch": 2.180866939377447,
      "grad_norm": 0.04835613608075363,
      "learning_rate": 0.0001242130152051073,
      "loss": 0.8472,
      "step": 124220
    },
    {
      "epoch": 2.1810425042574484,
      "grad_norm": 0.04815058462228203,
      "learning_rate": 0.00012420218208568457,
      "loss": 0.8471,
      "step": 124230
    },
    {
      "epoch": 2.18121806913745,
      "grad_norm": 0.0544713681628411,
      "learning_rate": 0.00012419134867228576,
      "loss": 0.8489,
      "step": 124240
    },
    {
      "epoch": 2.1813936340174513,
      "grad_norm": 0.0484822876831519,
      "learning_rate": 0.00012418051496504814,
      "loss": 0.8434,
      "step": 124250
    },
    {
      "epoch": 2.1815691988974524,
      "grad_norm": 0.06111312864960844,
      "learning_rate": 0.00012416968096410895,
      "loss": 0.844,
      "step": 124260
    },
    {
      "epoch": 2.181744763777454,
      "grad_norm": 0.06878806977171939,
      "learning_rate": 0.00012415884666960546,
      "loss": 0.8455,
      "step": 124270
    },
    {
      "epoch": 2.1819203286574553,
      "grad_norm": 0.0438092388533409,
      "learning_rate": 0.0001241480120816749,
      "loss": 0.8561,
      "step": 124280
    },
    {
      "epoch": 2.182095893537457,
      "grad_norm": 0.05533468143341329,
      "learning_rate": 0.00012413717720045466,
      "loss": 0.8478,
      "step": 124290
    },
    {
      "epoch": 2.1822714584174583,
      "grad_norm": 0.042214065010058775,
      "learning_rate": 0.00012412634202608192,
      "loss": 0.8478,
      "step": 124300
    },
    {
      "epoch": 2.1824470232974598,
      "grad_norm": 0.08879199941465796,
      "learning_rate": 0.00012411550655869397,
      "loss": 0.8496,
      "step": 124310
    },
    {
      "epoch": 2.182622588177461,
      "grad_norm": 0.06221116538184695,
      "learning_rate": 0.00012410467079842816,
      "loss": 0.8423,
      "step": 124320
    },
    {
      "epoch": 2.1827981530574623,
      "grad_norm": 0.07666458266636257,
      "learning_rate": 0.00012409383474542172,
      "loss": 0.844,
      "step": 124330
    },
    {
      "epoch": 2.1829737179374638,
      "grad_norm": 0.07460389627093264,
      "learning_rate": 0.00012408299839981196,
      "loss": 0.8494,
      "step": 124340
    },
    {
      "epoch": 2.1831492828174652,
      "grad_norm": 0.05543471257627494,
      "learning_rate": 0.00012407216176173622,
      "loss": 0.8436,
      "step": 124350
    },
    {
      "epoch": 2.1833248476974667,
      "grad_norm": 0.06106819196889984,
      "learning_rate": 0.00012406132483133174,
      "loss": 0.8388,
      "step": 124360
    },
    {
      "epoch": 2.183500412577468,
      "grad_norm": 0.06775570594240572,
      "learning_rate": 0.00012405048760873587,
      "loss": 0.8461,
      "step": 124370
    },
    {
      "epoch": 2.183675977457469,
      "grad_norm": 0.05598915062536177,
      "learning_rate": 0.0001240396500940859,
      "loss": 0.8519,
      "step": 124380
    },
    {
      "epoch": 2.1838515423374707,
      "grad_norm": 0.05491776060283023,
      "learning_rate": 0.00012402881228751916,
      "loss": 0.8518,
      "step": 124390
    },
    {
      "epoch": 2.184027107217472,
      "grad_norm": 0.05716590290636651,
      "learning_rate": 0.00012401797418917294,
      "loss": 0.8405,
      "step": 124400
    },
    {
      "epoch": 2.1842026720974737,
      "grad_norm": 0.11126827939467303,
      "learning_rate": 0.00012400713579918457,
      "loss": 0.8385,
      "step": 124410
    },
    {
      "epoch": 2.184378236977475,
      "grad_norm": 0.048945369557086546,
      "learning_rate": 0.0001239962971176914,
      "loss": 0.8508,
      "step": 124420
    },
    {
      "epoch": 2.1845538018574766,
      "grad_norm": 0.0733136288547381,
      "learning_rate": 0.0001239854581448307,
      "loss": 0.8518,
      "step": 124430
    },
    {
      "epoch": 2.1847293667374776,
      "grad_norm": 0.04949587556250827,
      "learning_rate": 0.0001239746188807399,
      "loss": 0.857,
      "step": 124440
    },
    {
      "epoch": 2.184904931617479,
      "grad_norm": 0.08384599385636037,
      "learning_rate": 0.00012396377932555623,
      "loss": 0.8504,
      "step": 124450
    },
    {
      "epoch": 2.1850804964974806,
      "grad_norm": 0.05707377086070871,
      "learning_rate": 0.0001239529394794171,
      "loss": 0.8367,
      "step": 124460
    },
    {
      "epoch": 2.185256061377482,
      "grad_norm": 0.053669842841630226,
      "learning_rate": 0.00012394209934245978,
      "loss": 0.8483,
      "step": 124470
    },
    {
      "epoch": 2.1854316262574836,
      "grad_norm": 0.05230729913594521,
      "learning_rate": 0.00012393125891482171,
      "loss": 0.8443,
      "step": 124480
    },
    {
      "epoch": 2.185607191137485,
      "grad_norm": 0.0533293088886247,
      "learning_rate": 0.0001239204181966402,
      "loss": 0.8406,
      "step": 124490
    },
    {
      "epoch": 2.185782756017486,
      "grad_norm": 0.054164319471502784,
      "learning_rate": 0.00012390957718805257,
      "loss": 0.8403,
      "step": 124500
    },
    {
      "epoch": 2.1859583208974875,
      "grad_norm": 0.06204044833086338,
      "learning_rate": 0.00012389873588919623,
      "loss": 0.8388,
      "step": 124510
    },
    {
      "epoch": 2.186133885777489,
      "grad_norm": 0.06683614347790409,
      "learning_rate": 0.00012388789430020847,
      "loss": 0.8476,
      "step": 124520
    },
    {
      "epoch": 2.1863094506574905,
      "grad_norm": 0.06120919276672839,
      "learning_rate": 0.00012387705242122675,
      "loss": 0.8439,
      "step": 124530
    },
    {
      "epoch": 2.186485015537492,
      "grad_norm": 0.06109698335312943,
      "learning_rate": 0.00012386621025238837,
      "loss": 0.8409,
      "step": 124540
    },
    {
      "epoch": 2.1866605804174934,
      "grad_norm": 0.06959552684227063,
      "learning_rate": 0.0001238553677938308,
      "loss": 0.841,
      "step": 124550
    },
    {
      "epoch": 2.1868361452974945,
      "grad_norm": 0.0524760032244521,
      "learning_rate": 0.00012384452504569127,
      "loss": 0.8455,
      "step": 124560
    },
    {
      "epoch": 2.187011710177496,
      "grad_norm": 0.05884108742890781,
      "learning_rate": 0.0001238336820081072,
      "loss": 0.8399,
      "step": 124570
    },
    {
      "epoch": 2.1871872750574974,
      "grad_norm": 0.06346271336278474,
      "learning_rate": 0.00012382283868121605,
      "loss": 0.8464,
      "step": 124580
    },
    {
      "epoch": 2.187362839937499,
      "grad_norm": 0.057504717273171455,
      "learning_rate": 0.00012381199506515512,
      "loss": 0.8426,
      "step": 124590
    },
    {
      "epoch": 2.1875384048175004,
      "grad_norm": 0.06529070901163926,
      "learning_rate": 0.0001238011511600619,
      "loss": 0.8454,
      "step": 124600
    },
    {
      "epoch": 2.187713969697502,
      "grad_norm": 0.06331793044008517,
      "learning_rate": 0.0001237903069660737,
      "loss": 0.8422,
      "step": 124610
    },
    {
      "epoch": 2.1878895345775033,
      "grad_norm": 0.06832443647920729,
      "learning_rate": 0.00012377946248332792,
      "loss": 0.8494,
      "step": 124620
    },
    {
      "epoch": 2.1880650994575044,
      "grad_norm": 0.048776861731735156,
      "learning_rate": 0.00012376861771196202,
      "loss": 0.8386,
      "step": 124630
    },
    {
      "epoch": 2.188240664337506,
      "grad_norm": 0.07169812032552944,
      "learning_rate": 0.00012375777265211333,
      "loss": 0.8498,
      "step": 124640
    },
    {
      "epoch": 2.1884162292175073,
      "grad_norm": 0.07225922269930124,
      "learning_rate": 0.00012374692730391932,
      "loss": 0.8465,
      "step": 124650
    },
    {
      "epoch": 2.188591794097509,
      "grad_norm": 0.04104672223341215,
      "learning_rate": 0.0001237360816675174,
      "loss": 0.8451,
      "step": 124660
    },
    {
      "epoch": 2.1887673589775103,
      "grad_norm": 0.05695683053575399,
      "learning_rate": 0.00012372523574304494,
      "loss": 0.8494,
      "step": 124670
    },
    {
      "epoch": 2.1889429238575113,
      "grad_norm": 0.04816120329970796,
      "learning_rate": 0.0001237143895306394,
      "loss": 0.8489,
      "step": 124680
    },
    {
      "epoch": 2.189118488737513,
      "grad_norm": 0.04423390303753773,
      "learning_rate": 0.00012370354303043815,
      "loss": 0.8408,
      "step": 124690
    },
    {
      "epoch": 2.1892940536175143,
      "grad_norm": 0.05406996108359208,
      "learning_rate": 0.00012369269624257873,
      "loss": 0.8513,
      "step": 124700
    },
    {
      "epoch": 2.1894696184975158,
      "grad_norm": 0.05862277206954831,
      "learning_rate": 0.00012368184916719844,
      "loss": 0.8485,
      "step": 124710
    },
    {
      "epoch": 2.1896451833775172,
      "grad_norm": 0.05772796682316606,
      "learning_rate": 0.00012367100180443483,
      "loss": 0.853,
      "step": 124720
    },
    {
      "epoch": 2.1898207482575187,
      "grad_norm": 0.050587670555889926,
      "learning_rate": 0.00012366015415442525,
      "loss": 0.8416,
      "step": 124730
    },
    {
      "epoch": 2.18999631313752,
      "grad_norm": 0.07073559777962278,
      "learning_rate": 0.00012364930621730716,
      "loss": 0.8494,
      "step": 124740
    },
    {
      "epoch": 2.1901718780175212,
      "grad_norm": 0.05871586037914789,
      "learning_rate": 0.00012363845799321806,
      "loss": 0.8389,
      "step": 124750
    },
    {
      "epoch": 2.1903474428975227,
      "grad_norm": 0.05730734696712359,
      "learning_rate": 0.0001236276094822953,
      "loss": 0.8347,
      "step": 124760
    },
    {
      "epoch": 2.190523007777524,
      "grad_norm": 0.05996872474304952,
      "learning_rate": 0.00012361676068467644,
      "loss": 0.8444,
      "step": 124770
    },
    {
      "epoch": 2.1906985726575257,
      "grad_norm": 0.0518479738809413,
      "learning_rate": 0.00012360591160049888,
      "loss": 0.8385,
      "step": 124780
    },
    {
      "epoch": 2.190874137537527,
      "grad_norm": 0.05952806614851918,
      "learning_rate": 0.00012359506222990008,
      "loss": 0.8276,
      "step": 124790
    },
    {
      "epoch": 2.191049702417528,
      "grad_norm": 0.05852579679051964,
      "learning_rate": 0.00012358421257301753,
      "loss": 0.8469,
      "step": 124800
    },
    {
      "epoch": 2.1912252672975296,
      "grad_norm": 0.04848725955274247,
      "learning_rate": 0.00012357336262998865,
      "loss": 0.8453,
      "step": 124810
    },
    {
      "epoch": 2.191400832177531,
      "grad_norm": 0.045949866241660114,
      "learning_rate": 0.00012356251240095093,
      "loss": 0.8495,
      "step": 124820
    },
    {
      "epoch": 2.1915763970575326,
      "grad_norm": 0.06354117190313383,
      "learning_rate": 0.00012355166188604192,
      "loss": 0.8475,
      "step": 124830
    },
    {
      "epoch": 2.191751961937534,
      "grad_norm": 0.05381139733017475,
      "learning_rate": 0.00012354081108539896,
      "loss": 0.8432,
      "step": 124840
    },
    {
      "epoch": 2.1919275268175356,
      "grad_norm": 0.06361854333189433,
      "learning_rate": 0.00012352995999915964,
      "loss": 0.8437,
      "step": 124850
    },
    {
      "epoch": 2.192103091697537,
      "grad_norm": 0.058228716539792356,
      "learning_rate": 0.00012351910862746138,
      "loss": 0.8388,
      "step": 124860
    },
    {
      "epoch": 2.192278656577538,
      "grad_norm": 0.07456602380192723,
      "learning_rate": 0.00012350825697044174,
      "loss": 0.849,
      "step": 124870
    },
    {
      "epoch": 2.1924542214575395,
      "grad_norm": 0.061739578946245655,
      "learning_rate": 0.00012349740502823814,
      "loss": 0.8459,
      "step": 124880
    },
    {
      "epoch": 2.192629786337541,
      "grad_norm": 0.07450989205189124,
      "learning_rate": 0.0001234865528009881,
      "loss": 0.8419,
      "step": 124890
    },
    {
      "epoch": 2.1928053512175425,
      "grad_norm": 0.06421089159694598,
      "learning_rate": 0.00012347570028882916,
      "loss": 0.847,
      "step": 124900
    },
    {
      "epoch": 2.192980916097544,
      "grad_norm": 0.05643411795373679,
      "learning_rate": 0.00012346484749189876,
      "loss": 0.8434,
      "step": 124910
    },
    {
      "epoch": 2.1931564809775455,
      "grad_norm": 0.07492104307522619,
      "learning_rate": 0.00012345399441033447,
      "loss": 0.844,
      "step": 124920
    },
    {
      "epoch": 2.1933320458575465,
      "grad_norm": 0.07521773989896895,
      "learning_rate": 0.00012344314104427372,
      "loss": 0.8525,
      "step": 124930
    },
    {
      "epoch": 2.193507610737548,
      "grad_norm": 0.044729428391283604,
      "learning_rate": 0.0001234322873938541,
      "loss": 0.8461,
      "step": 124940
    },
    {
      "epoch": 2.1936831756175494,
      "grad_norm": 0.04876994067571262,
      "learning_rate": 0.0001234214334592131,
      "loss": 0.8522,
      "step": 124950
    },
    {
      "epoch": 2.193858740497551,
      "grad_norm": 0.04506581021993961,
      "learning_rate": 0.00012341057924048822,
      "loss": 0.8506,
      "step": 124960
    },
    {
      "epoch": 2.1940343053775524,
      "grad_norm": 0.04979531119707928,
      "learning_rate": 0.00012339972473781704,
      "loss": 0.8443,
      "step": 124970
    },
    {
      "epoch": 2.194209870257554,
      "grad_norm": 0.15702013656830915,
      "learning_rate": 0.00012338886995133703,
      "loss": 0.8499,
      "step": 124980
    },
    {
      "epoch": 2.194385435137555,
      "grad_norm": 0.053146515720667364,
      "learning_rate": 0.00012337801488118574,
      "loss": 0.8371,
      "step": 124990
    },
    {
      "epoch": 2.1945610000175564,
      "grad_norm": 0.061224050383658496,
      "learning_rate": 0.00012336715952750076,
      "loss": 0.8515,
      "step": 125000
    },
    {
      "epoch": 2.194736564897558,
      "grad_norm": 0.06629138656234636,
      "learning_rate": 0.00012335630389041953,
      "loss": 0.8479,
      "step": 125010
    },
    {
      "epoch": 2.1949121297775593,
      "grad_norm": 0.05300960359027961,
      "learning_rate": 0.0001233454479700797,
      "loss": 0.8487,
      "step": 125020
    },
    {
      "epoch": 2.195087694657561,
      "grad_norm": 0.055743151249398354,
      "learning_rate": 0.0001233345917666187,
      "loss": 0.8415,
      "step": 125030
    },
    {
      "epoch": 2.1952632595375623,
      "grad_norm": 0.059800211743205205,
      "learning_rate": 0.00012332373528017417,
      "loss": 0.8484,
      "step": 125040
    },
    {
      "epoch": 2.1954388244175633,
      "grad_norm": 0.06961328300368419,
      "learning_rate": 0.00012331287851088363,
      "loss": 0.8541,
      "step": 125050
    },
    {
      "epoch": 2.195614389297565,
      "grad_norm": 0.050941711498586346,
      "learning_rate": 0.00012330202145888466,
      "loss": 0.8456,
      "step": 125060
    },
    {
      "epoch": 2.1957899541775663,
      "grad_norm": 0.0497845061241817,
      "learning_rate": 0.0001232911641243148,
      "loss": 0.8431,
      "step": 125070
    },
    {
      "epoch": 2.1959655190575678,
      "grad_norm": 0.06516888247510592,
      "learning_rate": 0.00012328030650731158,
      "loss": 0.8441,
      "step": 125080
    },
    {
      "epoch": 2.1961410839375692,
      "grad_norm": 0.056651977927015595,
      "learning_rate": 0.00012326944860801265,
      "loss": 0.8525,
      "step": 125090
    },
    {
      "epoch": 2.1963166488175707,
      "grad_norm": 0.05260301829185189,
      "learning_rate": 0.00012325859042655552,
      "loss": 0.8458,
      "step": 125100
    },
    {
      "epoch": 2.1964922136975717,
      "grad_norm": 0.051478051535946284,
      "learning_rate": 0.0001232477319630778,
      "loss": 0.8503,
      "step": 125110
    },
    {
      "epoch": 2.1966677785775732,
      "grad_norm": 0.06019713037447807,
      "learning_rate": 0.00012323687321771703,
      "loss": 0.8439,
      "step": 125120
    },
    {
      "epoch": 2.1968433434575747,
      "grad_norm": 0.04970495064692608,
      "learning_rate": 0.0001232260141906108,
      "loss": 0.8431,
      "step": 125130
    },
    {
      "epoch": 2.197018908337576,
      "grad_norm": 0.05441628837102619,
      "learning_rate": 0.00012321515488189678,
      "loss": 0.8441,
      "step": 125140
    },
    {
      "epoch": 2.1971944732175777,
      "grad_norm": 0.049183790723191206,
      "learning_rate": 0.00012320429529171244,
      "loss": 0.8456,
      "step": 125150
    },
    {
      "epoch": 2.197370038097579,
      "grad_norm": 0.05617163181709665,
      "learning_rate": 0.00012319343542019543,
      "loss": 0.8429,
      "step": 125160
    },
    {
      "epoch": 2.19754560297758,
      "grad_norm": 0.06751041642333382,
      "learning_rate": 0.00012318257526748334,
      "loss": 0.8455,
      "step": 125170
    },
    {
      "epoch": 2.1977211678575816,
      "grad_norm": 0.07411392524987319,
      "learning_rate": 0.00012317171483371375,
      "loss": 0.8404,
      "step": 125180
    },
    {
      "epoch": 2.197896732737583,
      "grad_norm": 0.0527861840285676,
      "learning_rate": 0.00012316085411902432,
      "loss": 0.8453,
      "step": 125190
    },
    {
      "epoch": 2.1980722976175846,
      "grad_norm": 0.06590777969665389,
      "learning_rate": 0.00012314999312355258,
      "loss": 0.8389,
      "step": 125200
    },
    {
      "epoch": 2.198247862497586,
      "grad_norm": 0.0599513115938885,
      "learning_rate": 0.00012313913184743624,
      "loss": 0.8405,
      "step": 125210
    },
    {
      "epoch": 2.1984234273775876,
      "grad_norm": 0.06106611470021757,
      "learning_rate": 0.0001231282702908128,
      "loss": 0.8453,
      "step": 125220
    },
    {
      "epoch": 2.1985989922575886,
      "grad_norm": 0.10117208943172988,
      "learning_rate": 0.00012311740845381995,
      "loss": 0.8503,
      "step": 125230
    },
    {
      "epoch": 2.19877455713759,
      "grad_norm": 0.05165086137557909,
      "learning_rate": 0.00012310654633659532,
      "loss": 0.8459,
      "step": 125240
    },
    {
      "epoch": 2.1989501220175915,
      "grad_norm": 0.09034258197437707,
      "learning_rate": 0.00012309568393927648,
      "loss": 0.8464,
      "step": 125250
    },
    {
      "epoch": 2.199125686897593,
      "grad_norm": 0.06317236885176475,
      "learning_rate": 0.0001230848212620011,
      "loss": 0.84,
      "step": 125260
    },
    {
      "epoch": 2.1993012517775945,
      "grad_norm": 0.08407934057377935,
      "learning_rate": 0.0001230739583049068,
      "loss": 0.8467,
      "step": 125270
    },
    {
      "epoch": 2.199476816657596,
      "grad_norm": 0.061975442945358526,
      "learning_rate": 0.00012306309506813126,
      "loss": 0.8496,
      "step": 125280
    },
    {
      "epoch": 2.199652381537597,
      "grad_norm": 0.06781003575958125,
      "learning_rate": 0.00012305223155181203,
      "loss": 0.8419,
      "step": 125290
    },
    {
      "epoch": 2.1998279464175985,
      "grad_norm": 0.061315674002411766,
      "learning_rate": 0.0001230413677560868,
      "loss": 0.848,
      "step": 125300
    },
    {
      "epoch": 2.2000035112976,
      "grad_norm": 0.06567647760789999,
      "learning_rate": 0.00012303050368109325,
      "loss": 0.8443,
      "step": 125310
    },
    {
      "epoch": 2.2001790761776014,
      "grad_norm": 0.05733471971736188,
      "learning_rate": 0.00012301963932696895,
      "loss": 0.8478,
      "step": 125320
    },
    {
      "epoch": 2.200354641057603,
      "grad_norm": 0.0573010098286156,
      "learning_rate": 0.00012300877469385163,
      "loss": 0.8449,
      "step": 125330
    },
    {
      "epoch": 2.2005302059376044,
      "grad_norm": 0.05747132760890244,
      "learning_rate": 0.00012299790978187892,
      "loss": 0.8463,
      "step": 125340
    },
    {
      "epoch": 2.2007057708176054,
      "grad_norm": 0.0648700490782898,
      "learning_rate": 0.00012298704459118846,
      "loss": 0.8466,
      "step": 125350
    },
    {
      "epoch": 2.200881335697607,
      "grad_norm": 0.057046422741351084,
      "learning_rate": 0.00012297617912191794,
      "loss": 0.8318,
      "step": 125360
    },
    {
      "epoch": 2.2010569005776084,
      "grad_norm": 0.06687825196909523,
      "learning_rate": 0.000122965313374205,
      "loss": 0.8435,
      "step": 125370
    },
    {
      "epoch": 2.20123246545761,
      "grad_norm": 0.05219632397206237,
      "learning_rate": 0.00012295444734818737,
      "loss": 0.8529,
      "step": 125380
    },
    {
      "epoch": 2.2014080303376113,
      "grad_norm": 0.07331635524076144,
      "learning_rate": 0.00012294358104400266,
      "loss": 0.8401,
      "step": 125390
    },
    {
      "epoch": 2.201583595217613,
      "grad_norm": 0.05662271633042301,
      "learning_rate": 0.00012293271446178855,
      "loss": 0.8446,
      "step": 125400
    },
    {
      "epoch": 2.2017591600976143,
      "grad_norm": 0.05539631324940245,
      "learning_rate": 0.00012292184760168278,
      "loss": 0.8474,
      "step": 125410
    },
    {
      "epoch": 2.2019347249776153,
      "grad_norm": 0.05393086095220049,
      "learning_rate": 0.00012291098046382298,
      "loss": 0.8447,
      "step": 125420
    },
    {
      "epoch": 2.202110289857617,
      "grad_norm": 0.06558529740814115,
      "learning_rate": 0.00012290011304834687,
      "loss": 0.8432,
      "step": 125430
    },
    {
      "epoch": 2.2022858547376183,
      "grad_norm": 0.05036473399127462,
      "learning_rate": 0.0001228892453553921,
      "loss": 0.843,
      "step": 125440
    },
    {
      "epoch": 2.2024614196176198,
      "grad_norm": 0.06128331735382623,
      "learning_rate": 0.00012287837738509646,
      "loss": 0.8403,
      "step": 125450
    },
    {
      "epoch": 2.2026369844976212,
      "grad_norm": 0.0821845996812044,
      "learning_rate": 0.00012286750913759756,
      "loss": 0.8372,
      "step": 125460
    },
    {
      "epoch": 2.2028125493776223,
      "grad_norm": 0.05990086243924384,
      "learning_rate": 0.0001228566406130331,
      "loss": 0.8488,
      "step": 125470
    },
    {
      "epoch": 2.2029881142576238,
      "grad_norm": 0.07698032749352471,
      "learning_rate": 0.00012284577181154085,
      "loss": 0.8456,
      "step": 125480
    },
    {
      "epoch": 2.2031636791376252,
      "grad_norm": 0.048995946593386575,
      "learning_rate": 0.00012283490273325845,
      "loss": 0.8415,
      "step": 125490
    },
    {
      "epoch": 2.2033392440176267,
      "grad_norm": 0.050733362123132876,
      "learning_rate": 0.00012282403337832365,
      "loss": 0.8474,
      "step": 125500
    },
    {
      "epoch": 2.203514808897628,
      "grad_norm": 0.07306194442978454,
      "learning_rate": 0.0001228131637468742,
      "loss": 0.8412,
      "step": 125510
    },
    {
      "epoch": 2.2036903737776297,
      "grad_norm": 0.06245423144868065,
      "learning_rate": 0.00012280229383904777,
      "loss": 0.8412,
      "step": 125520
    },
    {
      "epoch": 2.203865938657631,
      "grad_norm": 0.05511804509860098,
      "learning_rate": 0.00012279142365498208,
      "loss": 0.8431,
      "step": 125530
    },
    {
      "epoch": 2.204041503537632,
      "grad_norm": 0.04622685627168382,
      "learning_rate": 0.00012278055319481487,
      "loss": 0.846,
      "step": 125540
    },
    {
      "epoch": 2.2042170684176337,
      "grad_norm": 0.07082454472505893,
      "learning_rate": 0.00012276968245868391,
      "loss": 0.834,
      "step": 125550
    },
    {
      "epoch": 2.204392633297635,
      "grad_norm": 0.046995905838768615,
      "learning_rate": 0.0001227588114467269,
      "loss": 0.8505,
      "step": 125560
    },
    {
      "epoch": 2.2045681981776366,
      "grad_norm": 0.07278887090022329,
      "learning_rate": 0.00012274794015908156,
      "loss": 0.8504,
      "step": 125570
    },
    {
      "epoch": 2.204743763057638,
      "grad_norm": 0.06551369508671201,
      "learning_rate": 0.00012273706859588569,
      "loss": 0.8445,
      "step": 125580
    },
    {
      "epoch": 2.2049193279376396,
      "grad_norm": 0.07609966660076732,
      "learning_rate": 0.00012272619675727692,
      "loss": 0.8418,
      "step": 125590
    },
    {
      "epoch": 2.2050948928176406,
      "grad_norm": 0.06689123011749996,
      "learning_rate": 0.00012271532464339313,
      "loss": 0.8467,
      "step": 125600
    },
    {
      "epoch": 2.205270457697642,
      "grad_norm": 0.04210652157157535,
      "learning_rate": 0.00012270445225437198,
      "loss": 0.8484,
      "step": 125610
    },
    {
      "epoch": 2.2054460225776436,
      "grad_norm": 0.04956489980164874,
      "learning_rate": 0.00012269357959035127,
      "loss": 0.8466,
      "step": 125620
    },
    {
      "epoch": 2.205621587457645,
      "grad_norm": 0.06518604230397582,
      "learning_rate": 0.0001226827066514688,
      "loss": 0.8371,
      "step": 125630
    },
    {
      "epoch": 2.2057971523376465,
      "grad_norm": 0.04965756110611197,
      "learning_rate": 0.00012267183343786221,
      "loss": 0.8463,
      "step": 125640
    },
    {
      "epoch": 2.205972717217648,
      "grad_norm": 0.05862716575130776,
      "learning_rate": 0.00012266095994966936,
      "loss": 0.8424,
      "step": 125650
    },
    {
      "epoch": 2.206148282097649,
      "grad_norm": 0.04056365706950594,
      "learning_rate": 0.00012265008618702803,
      "loss": 0.8484,
      "step": 125660
    },
    {
      "epoch": 2.2063238469776505,
      "grad_norm": 0.04750040146834537,
      "learning_rate": 0.00012263921215007592,
      "loss": 0.8433,
      "step": 125670
    },
    {
      "epoch": 2.206499411857652,
      "grad_norm": 0.05971387613135431,
      "learning_rate": 0.00012262833783895086,
      "loss": 0.8413,
      "step": 125680
    },
    {
      "epoch": 2.2066749767376534,
      "grad_norm": 0.05490319039989867,
      "learning_rate": 0.00012261746325379063,
      "loss": 0.8453,
      "step": 125690
    },
    {
      "epoch": 2.206850541617655,
      "grad_norm": 0.07802674220132272,
      "learning_rate": 0.00012260658839473296,
      "loss": 0.8491,
      "step": 125700
    },
    {
      "epoch": 2.2070261064976564,
      "grad_norm": 0.05672089388712474,
      "learning_rate": 0.00012259571326191568,
      "loss": 0.8463,
      "step": 125710
    },
    {
      "epoch": 2.2072016713776574,
      "grad_norm": 0.06635830308129866,
      "learning_rate": 0.0001225848378554766,
      "loss": 0.85,
      "step": 125720
    },
    {
      "epoch": 2.207377236257659,
      "grad_norm": 0.05346426411813844,
      "learning_rate": 0.00012257396217555349,
      "loss": 0.8458,
      "step": 125730
    },
    {
      "epoch": 2.2075528011376604,
      "grad_norm": 0.05256035196097201,
      "learning_rate": 0.00012256308622228413,
      "loss": 0.8526,
      "step": 125740
    },
    {
      "epoch": 2.207728366017662,
      "grad_norm": 0.061090920080678165,
      "learning_rate": 0.0001225522099958063,
      "loss": 0.8453,
      "step": 125750
    },
    {
      "epoch": 2.2079039308976633,
      "grad_norm": 0.06083036814929553,
      "learning_rate": 0.00012254133349625786,
      "loss": 0.8545,
      "step": 125760
    },
    {
      "epoch": 2.208079495777665,
      "grad_norm": 0.054618645146653017,
      "learning_rate": 0.00012253045672377664,
      "loss": 0.8369,
      "step": 125770
    },
    {
      "epoch": 2.208255060657666,
      "grad_norm": 0.055949130011165284,
      "learning_rate": 0.00012251957967850037,
      "loss": 0.8446,
      "step": 125780
    },
    {
      "epoch": 2.2084306255376673,
      "grad_norm": 0.0460511466644566,
      "learning_rate": 0.0001225087023605669,
      "loss": 0.8465,
      "step": 125790
    },
    {
      "epoch": 2.208606190417669,
      "grad_norm": 0.05387253807842644,
      "learning_rate": 0.00012249782477011408,
      "loss": 0.8438,
      "step": 125800
    },
    {
      "epoch": 2.2087817552976703,
      "grad_norm": 0.058319548848021166,
      "learning_rate": 0.00012248694690727966,
      "loss": 0.8475,
      "step": 125810
    },
    {
      "epoch": 2.2089573201776718,
      "grad_norm": 0.04787920684208883,
      "learning_rate": 0.00012247606877220153,
      "loss": 0.8497,
      "step": 125820
    },
    {
      "epoch": 2.2091328850576732,
      "grad_norm": 0.06443302738764249,
      "learning_rate": 0.00012246519036501746,
      "loss": 0.8438,
      "step": 125830
    },
    {
      "epoch": 2.2093084499376743,
      "grad_norm": 0.06723664384036332,
      "learning_rate": 0.00012245431168586535,
      "loss": 0.8401,
      "step": 125840
    },
    {
      "epoch": 2.2094840148176758,
      "grad_norm": 0.06132274617518927,
      "learning_rate": 0.00012244343273488302,
      "loss": 0.8335,
      "step": 125850
    },
    {
      "epoch": 2.2096595796976772,
      "grad_norm": 0.05644585427636818,
      "learning_rate": 0.00012243255351220822,
      "loss": 0.8486,
      "step": 125860
    },
    {
      "epoch": 2.2098351445776787,
      "grad_norm": 0.056816390465292375,
      "learning_rate": 0.0001224216740179789,
      "loss": 0.8482,
      "step": 125870
    },
    {
      "epoch": 2.21001070945768,
      "grad_norm": 0.05463519191612593,
      "learning_rate": 0.00012241079425233286,
      "loss": 0.846,
      "step": 125880
    },
    {
      "epoch": 2.2101862743376817,
      "grad_norm": 0.058992668618387,
      "learning_rate": 0.00012239991421540794,
      "loss": 0.8485,
      "step": 125890
    },
    {
      "epoch": 2.2103618392176827,
      "grad_norm": 0.053769683662493,
      "learning_rate": 0.00012238903390734203,
      "loss": 0.842,
      "step": 125900
    },
    {
      "epoch": 2.210537404097684,
      "grad_norm": 0.05335001081905554,
      "learning_rate": 0.00012237815332827296,
      "loss": 0.8488,
      "step": 125910
    },
    {
      "epoch": 2.2107129689776857,
      "grad_norm": 0.08882324544724791,
      "learning_rate": 0.00012236727247833857,
      "loss": 0.8506,
      "step": 125920
    },
    {
      "epoch": 2.210888533857687,
      "grad_norm": 0.058324125372274487,
      "learning_rate": 0.00012235639135767677,
      "loss": 0.8358,
      "step": 125930
    },
    {
      "epoch": 2.2110640987376886,
      "grad_norm": 0.05728458036400814,
      "learning_rate": 0.00012234550996642537,
      "loss": 0.8461,
      "step": 125940
    },
    {
      "epoch": 2.21123966361769,
      "grad_norm": 0.06869062245046836,
      "learning_rate": 0.00012233462830472225,
      "loss": 0.8455,
      "step": 125950
    },
    {
      "epoch": 2.211415228497691,
      "grad_norm": 0.049680874213661914,
      "learning_rate": 0.00012232374637270534,
      "loss": 0.8405,
      "step": 125960
    },
    {
      "epoch": 2.2115907933776926,
      "grad_norm": 0.053337751162601386,
      "learning_rate": 0.00012231286417051248,
      "loss": 0.8369,
      "step": 125970
    },
    {
      "epoch": 2.211766358257694,
      "grad_norm": 0.05537271130590268,
      "learning_rate": 0.00012230198169828154,
      "loss": 0.8458,
      "step": 125980
    },
    {
      "epoch": 2.2119419231376956,
      "grad_norm": 0.07133347002202642,
      "learning_rate": 0.00012229109895615041,
      "loss": 0.8564,
      "step": 125990
    },
    {
      "epoch": 2.212117488017697,
      "grad_norm": 0.061176276188035714,
      "learning_rate": 0.00012228021594425694,
      "loss": 0.8549,
      "step": 126000
    },
    {
      "epoch": 2.2122930528976985,
      "grad_norm": 0.07602315571020539,
      "learning_rate": 0.0001222693326627391,
      "loss": 0.8391,
      "step": 126010
    },
    {
      "epoch": 2.2124686177776995,
      "grad_norm": 0.06249039776680722,
      "learning_rate": 0.00012225844911173473,
      "loss": 0.8478,
      "step": 126020
    },
    {
      "epoch": 2.212644182657701,
      "grad_norm": 0.04655733895851769,
      "learning_rate": 0.0001222475652913817,
      "loss": 0.8422,
      "step": 126030
    },
    {
      "epoch": 2.2128197475377025,
      "grad_norm": 0.06552088603653033,
      "learning_rate": 0.000122236681201818,
      "loss": 0.8463,
      "step": 126040
    },
    {
      "epoch": 2.212995312417704,
      "grad_norm": 0.059192355323315765,
      "learning_rate": 0.00012222579684318142,
      "loss": 0.8401,
      "step": 126050
    },
    {
      "epoch": 2.2131708772977055,
      "grad_norm": 0.05694046352028415,
      "learning_rate": 0.00012221491221560998,
      "loss": 0.8411,
      "step": 126060
    },
    {
      "epoch": 2.213346442177707,
      "grad_norm": 0.05691325805312341,
      "learning_rate": 0.0001222040273192415,
      "loss": 0.8443,
      "step": 126070
    },
    {
      "epoch": 2.2135220070577084,
      "grad_norm": 0.058102169878842626,
      "learning_rate": 0.00012219314215421395,
      "loss": 0.8421,
      "step": 126080
    },
    {
      "epoch": 2.2136975719377094,
      "grad_norm": 0.04827077825591236,
      "learning_rate": 0.00012218225672066524,
      "loss": 0.8395,
      "step": 126090
    },
    {
      "epoch": 2.213873136817711,
      "grad_norm": 0.047618621170581056,
      "learning_rate": 0.00012217137101873323,
      "loss": 0.8444,
      "step": 126100
    },
    {
      "epoch": 2.2140487016977124,
      "grad_norm": 0.05334565107204412,
      "learning_rate": 0.00012216048504855596,
      "loss": 0.8435,
      "step": 126110
    },
    {
      "epoch": 2.214224266577714,
      "grad_norm": 0.04536185241880609,
      "learning_rate": 0.00012214959881027122,
      "loss": 0.8441,
      "step": 126120
    },
    {
      "epoch": 2.2143998314577154,
      "grad_norm": 0.05777115785077156,
      "learning_rate": 0.00012213871230401705,
      "loss": 0.8432,
      "step": 126130
    },
    {
      "epoch": 2.2145753963377164,
      "grad_norm": 0.06228377848543966,
      "learning_rate": 0.00012212782552993137,
      "loss": 0.8367,
      "step": 126140
    },
    {
      "epoch": 2.214750961217718,
      "grad_norm": 0.05075513739031121,
      "learning_rate": 0.000122116938488152,
      "loss": 0.8409,
      "step": 126150
    },
    {
      "epoch": 2.2149265260977193,
      "grad_norm": 0.05817278099598366,
      "learning_rate": 0.00012210605117881707,
      "loss": 0.8432,
      "step": 126160
    },
    {
      "epoch": 2.215102090977721,
      "grad_norm": 0.05295496694444222,
      "learning_rate": 0.00012209516360206437,
      "loss": 0.8451,
      "step": 126170
    },
    {
      "epoch": 2.2152776558577223,
      "grad_norm": 0.09323557715991254,
      "learning_rate": 0.00012208427575803192,
      "loss": 0.8418,
      "step": 126180
    },
    {
      "epoch": 2.2154532207377238,
      "grad_norm": 0.0537539429914354,
      "learning_rate": 0.00012207338764685767,
      "loss": 0.846,
      "step": 126190
    },
    {
      "epoch": 2.2156287856177252,
      "grad_norm": 0.06573737807092007,
      "learning_rate": 0.00012206249926867954,
      "loss": 0.8414,
      "step": 126200
    },
    {
      "epoch": 2.2158043504977263,
      "grad_norm": 0.07288256548536974,
      "learning_rate": 0.00012205161062363553,
      "loss": 0.8431,
      "step": 126210
    },
    {
      "epoch": 2.2159799153777278,
      "grad_norm": 0.07123957953719558,
      "learning_rate": 0.00012204072171186353,
      "loss": 0.8454,
      "step": 126220
    },
    {
      "epoch": 2.2161554802577292,
      "grad_norm": 0.07004823700841521,
      "learning_rate": 0.0001220298325335016,
      "loss": 0.8576,
      "step": 126230
    },
    {
      "epoch": 2.2163310451377307,
      "grad_norm": 0.05533944641770923,
      "learning_rate": 0.00012201894308868763,
      "loss": 0.8523,
      "step": 126240
    },
    {
      "epoch": 2.216506610017732,
      "grad_norm": 0.05325957913678593,
      "learning_rate": 0.00012200805337755964,
      "loss": 0.84,
      "step": 126250
    },
    {
      "epoch": 2.2166821748977332,
      "grad_norm": 0.062291746086394595,
      "learning_rate": 0.00012199716340025562,
      "loss": 0.8446,
      "step": 126260
    },
    {
      "epoch": 2.2168577397777347,
      "grad_norm": 0.052076041633373545,
      "learning_rate": 0.00012198627315691345,
      "loss": 0.8418,
      "step": 126270
    },
    {
      "epoch": 2.217033304657736,
      "grad_norm": 0.051228659134223635,
      "learning_rate": 0.00012197538264767124,
      "loss": 0.8447,
      "step": 126280
    },
    {
      "epoch": 2.2172088695377377,
      "grad_norm": 0.05090819212002249,
      "learning_rate": 0.00012196449187266685,
      "loss": 0.8437,
      "step": 126290
    },
    {
      "epoch": 2.217384434417739,
      "grad_norm": 0.0454109009190558,
      "learning_rate": 0.00012195360083203837,
      "loss": 0.848,
      "step": 126300
    },
    {
      "epoch": 2.2175599992977406,
      "grad_norm": 0.04859826158450606,
      "learning_rate": 0.00012194270952592377,
      "loss": 0.8468,
      "step": 126310
    },
    {
      "epoch": 2.217735564177742,
      "grad_norm": 0.04827383549640412,
      "learning_rate": 0.00012193181795446095,
      "loss": 0.8466,
      "step": 126320
    },
    {
      "epoch": 2.217911129057743,
      "grad_norm": 0.04461621830822064,
      "learning_rate": 0.00012192092611778807,
      "loss": 0.845,
      "step": 126330
    },
    {
      "epoch": 2.2180866939377446,
      "grad_norm": 0.05859748355994447,
      "learning_rate": 0.000121910034016043,
      "loss": 0.8467,
      "step": 126340
    },
    {
      "epoch": 2.218262258817746,
      "grad_norm": 0.059394069683367835,
      "learning_rate": 0.00012189914164936382,
      "loss": 0.8457,
      "step": 126350
    },
    {
      "epoch": 2.2184378236977476,
      "grad_norm": 0.07942198955600516,
      "learning_rate": 0.00012188824901788851,
      "loss": 0.8405,
      "step": 126360
    },
    {
      "epoch": 2.218613388577749,
      "grad_norm": 0.06395199360690917,
      "learning_rate": 0.00012187735612175504,
      "loss": 0.8478,
      "step": 126370
    },
    {
      "epoch": 2.2187889534577505,
      "grad_norm": 0.046814827531287906,
      "learning_rate": 0.00012186646296110149,
      "loss": 0.842,
      "step": 126380
    },
    {
      "epoch": 2.2189645183377515,
      "grad_norm": 0.0684136061567483,
      "learning_rate": 0.00012185556953606586,
      "loss": 0.8469,
      "step": 126390
    },
    {
      "epoch": 2.219140083217753,
      "grad_norm": 0.06758837407902178,
      "learning_rate": 0.00012184467584678616,
      "loss": 0.8473,
      "step": 126400
    },
    {
      "epoch": 2.2193156480977545,
      "grad_norm": 0.0828426595774835,
      "learning_rate": 0.00012183378189340042,
      "loss": 0.8412,
      "step": 126410
    },
    {
      "epoch": 2.219491212977756,
      "grad_norm": 0.06526645148277187,
      "learning_rate": 0.00012182288767604668,
      "loss": 0.8357,
      "step": 126420
    },
    {
      "epoch": 2.2196667778577575,
      "grad_norm": 0.049842220838381045,
      "learning_rate": 0.00012181199319486296,
      "loss": 0.8403,
      "step": 126430
    },
    {
      "epoch": 2.219842342737759,
      "grad_norm": 0.0539725460119618,
      "learning_rate": 0.00012180109844998728,
      "loss": 0.8424,
      "step": 126440
    },
    {
      "epoch": 2.22001790761776,
      "grad_norm": 0.061508797196667,
      "learning_rate": 0.00012179020344155773,
      "loss": 0.8403,
      "step": 126450
    },
    {
      "epoch": 2.2201934724977614,
      "grad_norm": 0.061168748295878456,
      "learning_rate": 0.0001217793081697123,
      "loss": 0.8435,
      "step": 126460
    },
    {
      "epoch": 2.220369037377763,
      "grad_norm": 0.06485893815468528,
      "learning_rate": 0.00012176841263458907,
      "loss": 0.8444,
      "step": 126470
    },
    {
      "epoch": 2.2205446022577644,
      "grad_norm": 0.05034276875293499,
      "learning_rate": 0.00012175751683632606,
      "loss": 0.8499,
      "step": 126480
    },
    {
      "epoch": 2.220720167137766,
      "grad_norm": 0.06548199859907763,
      "learning_rate": 0.00012174662077506129,
      "loss": 0.8369,
      "step": 126490
    },
    {
      "epoch": 2.2208957320177674,
      "grad_norm": 0.04593495378947425,
      "learning_rate": 0.00012173572445093293,
      "loss": 0.8456,
      "step": 126500
    },
    {
      "epoch": 2.2210712968977684,
      "grad_norm": 0.045744289489457905,
      "learning_rate": 0.0001217248278640789,
      "loss": 0.8413,
      "step": 126510
    },
    {
      "epoch": 2.22124686177777,
      "grad_norm": 0.05951027161702524,
      "learning_rate": 0.00012171393101463738,
      "loss": 0.8413,
      "step": 126520
    },
    {
      "epoch": 2.2214224266577713,
      "grad_norm": 0.06424175949058222,
      "learning_rate": 0.00012170303390274638,
      "loss": 0.8501,
      "step": 126530
    },
    {
      "epoch": 2.221597991537773,
      "grad_norm": 0.06002808848016759,
      "learning_rate": 0.00012169213652854396,
      "loss": 0.8471,
      "step": 126540
    },
    {
      "epoch": 2.2217735564177743,
      "grad_norm": 0.0505488578300717,
      "learning_rate": 0.0001216812388921682,
      "loss": 0.8424,
      "step": 126550
    },
    {
      "epoch": 2.2219491212977758,
      "grad_norm": 0.1429328371098002,
      "learning_rate": 0.00012167034099375718,
      "loss": 0.8528,
      "step": 126560
    },
    {
      "epoch": 2.222124686177777,
      "grad_norm": 0.047233463558937414,
      "learning_rate": 0.00012165944283344898,
      "loss": 0.8435,
      "step": 126570
    },
    {
      "epoch": 2.2223002510577783,
      "grad_norm": 0.06954293731363158,
      "learning_rate": 0.0001216485444113817,
      "loss": 0.8516,
      "step": 126580
    },
    {
      "epoch": 2.2224758159377798,
      "grad_norm": 0.054206520764593605,
      "learning_rate": 0.0001216376457276934,
      "loss": 0.8473,
      "step": 126590
    },
    {
      "epoch": 2.2226513808177812,
      "grad_norm": 0.04874093361036773,
      "learning_rate": 0.00012162674678252215,
      "loss": 0.8504,
      "step": 126600
    },
    {
      "epoch": 2.2228269456977827,
      "grad_norm": 0.05207421707499434,
      "learning_rate": 0.00012161584757600608,
      "loss": 0.8466,
      "step": 126610
    },
    {
      "epoch": 2.223002510577784,
      "grad_norm": 0.06338965954588952,
      "learning_rate": 0.00012160494810828327,
      "loss": 0.8442,
      "step": 126620
    },
    {
      "epoch": 2.2231780754577852,
      "grad_norm": 0.05591294857931998,
      "learning_rate": 0.0001215940483794918,
      "loss": 0.8462,
      "step": 126630
    },
    {
      "epoch": 2.2233536403377867,
      "grad_norm": 0.0621792012067224,
      "learning_rate": 0.00012158314838976981,
      "loss": 0.8447,
      "step": 126640
    },
    {
      "epoch": 2.223529205217788,
      "grad_norm": 0.05331296408453664,
      "learning_rate": 0.00012157224813925538,
      "loss": 0.8436,
      "step": 126650
    },
    {
      "epoch": 2.2237047700977897,
      "grad_norm": 0.05685037123672975,
      "learning_rate": 0.00012156134762808662,
      "loss": 0.8488,
      "step": 126660
    },
    {
      "epoch": 2.223880334977791,
      "grad_norm": 0.05537803300502391,
      "learning_rate": 0.00012155044685640167,
      "loss": 0.8457,
      "step": 126670
    },
    {
      "epoch": 2.2240558998577926,
      "grad_norm": 0.05079050571256507,
      "learning_rate": 0.0001215395458243386,
      "loss": 0.8416,
      "step": 126680
    },
    {
      "epoch": 2.2242314647377937,
      "grad_norm": 0.05715163435197782,
      "learning_rate": 0.00012152864453203554,
      "loss": 0.8521,
      "step": 126690
    },
    {
      "epoch": 2.224407029617795,
      "grad_norm": 0.08843125716107106,
      "learning_rate": 0.00012151774297963063,
      "loss": 0.8427,
      "step": 126700
    },
    {
      "epoch": 2.2245825944977966,
      "grad_norm": 0.07611433116226245,
      "learning_rate": 0.00012150684116726198,
      "loss": 0.8432,
      "step": 126710
    },
    {
      "epoch": 2.224758159377798,
      "grad_norm": 0.06354914418885407,
      "learning_rate": 0.00012149593909506772,
      "loss": 0.8398,
      "step": 126720
    },
    {
      "epoch": 2.2249337242577996,
      "grad_norm": 0.06861252538379717,
      "learning_rate": 0.000121485036763186,
      "loss": 0.8394,
      "step": 126730
    },
    {
      "epoch": 2.225109289137801,
      "grad_norm": 0.05338294286142157,
      "learning_rate": 0.0001214741341717549,
      "loss": 0.8455,
      "step": 126740
    },
    {
      "epoch": 2.225284854017802,
      "grad_norm": 0.05350894275836725,
      "learning_rate": 0.00012146323132091262,
      "loss": 0.8454,
      "step": 126750
    },
    {
      "epoch": 2.2254604188978035,
      "grad_norm": 0.0695870259562744,
      "learning_rate": 0.00012145232821079729,
      "loss": 0.8444,
      "step": 126760
    },
    {
      "epoch": 2.225635983777805,
      "grad_norm": 0.05358854761925729,
      "learning_rate": 0.00012144142484154704,
      "loss": 0.8414,
      "step": 126770
    },
    {
      "epoch": 2.2258115486578065,
      "grad_norm": 0.05469690436933419,
      "learning_rate": 0.00012143052121329999,
      "loss": 0.8463,
      "step": 126780
    },
    {
      "epoch": 2.225987113537808,
      "grad_norm": 0.058186260558330274,
      "learning_rate": 0.00012141961732619435,
      "loss": 0.8412,
      "step": 126790
    },
    {
      "epoch": 2.2261626784178095,
      "grad_norm": 0.053162088032489015,
      "learning_rate": 0.00012140871318036824,
      "loss": 0.8462,
      "step": 126800
    },
    {
      "epoch": 2.2263382432978105,
      "grad_norm": 0.06756330812642146,
      "learning_rate": 0.00012139780877595984,
      "loss": 0.8405,
      "step": 126810
    },
    {
      "epoch": 2.226513808177812,
      "grad_norm": 0.049634778961152815,
      "learning_rate": 0.00012138690411310725,
      "loss": 0.8487,
      "step": 126820
    },
    {
      "epoch": 2.2266893730578134,
      "grad_norm": 0.05190590211886879,
      "learning_rate": 0.00012137599919194869,
      "loss": 0.8507,
      "step": 126830
    },
    {
      "epoch": 2.226864937937815,
      "grad_norm": 0.06303888580385243,
      "learning_rate": 0.00012136509401262235,
      "loss": 0.8484,
      "step": 126840
    },
    {
      "epoch": 2.2270405028178164,
      "grad_norm": 0.08153305411877568,
      "learning_rate": 0.00012135418857526633,
      "loss": 0.8458,
      "step": 126850
    },
    {
      "epoch": 2.227216067697818,
      "grad_norm": 0.08049074541294166,
      "learning_rate": 0.00012134328288001885,
      "loss": 0.8372,
      "step": 126860
    },
    {
      "epoch": 2.2273916325778194,
      "grad_norm": 0.052023189048536346,
      "learning_rate": 0.0001213323769270181,
      "loss": 0.8413,
      "step": 126870
    },
    {
      "epoch": 2.2275671974578204,
      "grad_norm": 0.05475484043582049,
      "learning_rate": 0.0001213214707164022,
      "loss": 0.8384,
      "step": 126880
    },
    {
      "epoch": 2.227742762337822,
      "grad_norm": 0.0818886718792576,
      "learning_rate": 0.00012131056424830937,
      "loss": 0.8588,
      "step": 126890
    },
    {
      "epoch": 2.2279183272178233,
      "grad_norm": 0.06755558330779185,
      "learning_rate": 0.00012129965752287784,
      "loss": 0.8418,
      "step": 126900
    },
    {
      "epoch": 2.228093892097825,
      "grad_norm": 0.05026721468648753,
      "learning_rate": 0.00012128875054024574,
      "loss": 0.8518,
      "step": 126910
    },
    {
      "epoch": 2.2282694569778263,
      "grad_norm": 0.06254677141938834,
      "learning_rate": 0.00012127784330055125,
      "loss": 0.8382,
      "step": 126920
    },
    {
      "epoch": 2.2284450218578273,
      "grad_norm": 0.053669949354754685,
      "learning_rate": 0.00012126693580393262,
      "loss": 0.8476,
      "step": 126930
    },
    {
      "epoch": 2.228620586737829,
      "grad_norm": 0.05960322953139219,
      "learning_rate": 0.00012125602805052807,
      "loss": 0.8434,
      "step": 126940
    },
    {
      "epoch": 2.2287961516178303,
      "grad_norm": 0.05254167017081585,
      "learning_rate": 0.0001212451200404757,
      "loss": 0.8485,
      "step": 126950
    },
    {
      "epoch": 2.2289717164978318,
      "grad_norm": 0.05068568847728551,
      "learning_rate": 0.0001212342117739138,
      "loss": 0.8526,
      "step": 126960
    },
    {
      "epoch": 2.2291472813778332,
      "grad_norm": 0.06901731372316801,
      "learning_rate": 0.00012122330325098058,
      "loss": 0.8392,
      "step": 126970
    },
    {
      "epoch": 2.2293228462578347,
      "grad_norm": 0.04651029935805866,
      "learning_rate": 0.00012121239447181423,
      "loss": 0.8497,
      "step": 126980
    },
    {
      "epoch": 2.229498411137836,
      "grad_norm": 0.05203396572661705,
      "learning_rate": 0.00012120148543655297,
      "loss": 0.8451,
      "step": 126990
    },
    {
      "epoch": 2.2296739760178372,
      "grad_norm": 0.05957030913687146,
      "learning_rate": 0.00012119057614533498,
      "loss": 0.8486,
      "step": 127000
    },
    {
      "epoch": 2.2298495408978387,
      "grad_norm": 0.06426275082970383,
      "learning_rate": 0.00012117966659829856,
      "loss": 0.847,
      "step": 127010
    },
    {
      "epoch": 2.23002510577784,
      "grad_norm": 0.09740694752366445,
      "learning_rate": 0.00012116875679558187,
      "loss": 0.852,
      "step": 127020
    },
    {
      "epoch": 2.2302006706578417,
      "grad_norm": 0.06183828889707933,
      "learning_rate": 0.0001211578467373232,
      "loss": 0.8488,
      "step": 127030
    },
    {
      "epoch": 2.230376235537843,
      "grad_norm": 0.05043877802563578,
      "learning_rate": 0.00012114693642366072,
      "loss": 0.8366,
      "step": 127040
    },
    {
      "epoch": 2.2305518004178446,
      "grad_norm": 0.051978325806673685,
      "learning_rate": 0.00012113602585473268,
      "loss": 0.849,
      "step": 127050
    },
    {
      "epoch": 2.2307273652978457,
      "grad_norm": 0.04063493361903021,
      "learning_rate": 0.00012112511503067738,
      "loss": 0.8492,
      "step": 127060
    },
    {
      "epoch": 2.230902930177847,
      "grad_norm": 0.050096779605621866,
      "learning_rate": 0.00012111420395163297,
      "loss": 0.8567,
      "step": 127070
    },
    {
      "epoch": 2.2310784950578486,
      "grad_norm": 0.05017020258850516,
      "learning_rate": 0.00012110329261773776,
      "loss": 0.8528,
      "step": 127080
    },
    {
      "epoch": 2.23125405993785,
      "grad_norm": 0.050876975008473185,
      "learning_rate": 0.00012109238102912996,
      "loss": 0.8427,
      "step": 127090
    },
    {
      "epoch": 2.2314296248178516,
      "grad_norm": 0.057987899619577815,
      "learning_rate": 0.0001210814691859479,
      "loss": 0.8388,
      "step": 127100
    },
    {
      "epoch": 2.231605189697853,
      "grad_norm": 0.05043450875782372,
      "learning_rate": 0.00012107055708832975,
      "loss": 0.8417,
      "step": 127110
    },
    {
      "epoch": 2.231780754577854,
      "grad_norm": 0.052397492491604074,
      "learning_rate": 0.00012105964473641376,
      "loss": 0.845,
      "step": 127120
    },
    {
      "epoch": 2.2319563194578556,
      "grad_norm": 0.0642220400619057,
      "learning_rate": 0.00012104873213033824,
      "loss": 0.8527,
      "step": 127130
    },
    {
      "epoch": 2.232131884337857,
      "grad_norm": 0.053397180817605576,
      "learning_rate": 0.00012103781927024145,
      "loss": 0.8466,
      "step": 127140
    },
    {
      "epoch": 2.2323074492178585,
      "grad_norm": 0.09446753028256093,
      "learning_rate": 0.00012102690615626167,
      "loss": 0.8363,
      "step": 127150
    },
    {
      "epoch": 2.23248301409786,
      "grad_norm": 0.14015779149747495,
      "learning_rate": 0.00012101599278853712,
      "loss": 0.8509,
      "step": 127160
    },
    {
      "epoch": 2.2326585789778615,
      "grad_norm": 0.05699227329853976,
      "learning_rate": 0.00012100507916720614,
      "loss": 0.8429,
      "step": 127170
    },
    {
      "epoch": 2.2328341438578625,
      "grad_norm": 0.05057897529303181,
      "learning_rate": 0.00012099416529240695,
      "loss": 0.8457,
      "step": 127180
    },
    {
      "epoch": 2.233009708737864,
      "grad_norm": 0.051850279455429846,
      "learning_rate": 0.00012098325116427784,
      "loss": 0.8504,
      "step": 127190
    },
    {
      "epoch": 2.2331852736178655,
      "grad_norm": 0.048130844961995774,
      "learning_rate": 0.00012097233678295712,
      "loss": 0.8382,
      "step": 127200
    },
    {
      "epoch": 2.233360838497867,
      "grad_norm": 0.05786962030938025,
      "learning_rate": 0.00012096142214858308,
      "loss": 0.8461,
      "step": 127210
    },
    {
      "epoch": 2.2335364033778684,
      "grad_norm": 0.05219772283483234,
      "learning_rate": 0.000120950507261294,
      "loss": 0.8517,
      "step": 127220
    },
    {
      "epoch": 2.23371196825787,
      "grad_norm": 0.05125757420415853,
      "learning_rate": 0.00012093959212122814,
      "loss": 0.8454,
      "step": 127230
    },
    {
      "epoch": 2.233887533137871,
      "grad_norm": 0.06337737455899584,
      "learning_rate": 0.00012092867672852384,
      "loss": 0.8431,
      "step": 127240
    },
    {
      "epoch": 2.2340630980178724,
      "grad_norm": 0.06240386819017549,
      "learning_rate": 0.00012091776108331938,
      "loss": 0.848,
      "step": 127250
    },
    {
      "epoch": 2.234238662897874,
      "grad_norm": 0.04820925002540306,
      "learning_rate": 0.00012090684518575309,
      "loss": 0.8388,
      "step": 127260
    },
    {
      "epoch": 2.2344142277778754,
      "grad_norm": 0.07092715891410631,
      "learning_rate": 0.00012089592903596323,
      "loss": 0.8429,
      "step": 127270
    },
    {
      "epoch": 2.234589792657877,
      "grad_norm": 0.04851370776953453,
      "learning_rate": 0.00012088501263408816,
      "loss": 0.8411,
      "step": 127280
    },
    {
      "epoch": 2.2347653575378783,
      "grad_norm": 0.056370651516150026,
      "learning_rate": 0.00012087409598026616,
      "loss": 0.8445,
      "step": 127290
    },
    {
      "epoch": 2.2349409224178793,
      "grad_norm": 0.055406029301865944,
      "learning_rate": 0.00012086317907463556,
      "loss": 0.8466,
      "step": 127300
    },
    {
      "epoch": 2.235116487297881,
      "grad_norm": 0.06741774078954495,
      "learning_rate": 0.00012085226191733465,
      "loss": 0.8423,
      "step": 127310
    },
    {
      "epoch": 2.2352920521778823,
      "grad_norm": 0.049896059664963244,
      "learning_rate": 0.00012084134450850182,
      "loss": 0.8478,
      "step": 127320
    },
    {
      "epoch": 2.2354676170578838,
      "grad_norm": 0.057817205338778016,
      "learning_rate": 0.00012083042684827532,
      "loss": 0.84,
      "step": 127330
    },
    {
      "epoch": 2.2356431819378852,
      "grad_norm": 0.07149054434998792,
      "learning_rate": 0.0001208195089367935,
      "loss": 0.8464,
      "step": 127340
    },
    {
      "epoch": 2.2358187468178867,
      "grad_norm": 0.06225726158944956,
      "learning_rate": 0.00012080859077419473,
      "loss": 0.849,
      "step": 127350
    },
    {
      "epoch": 2.2359943116978878,
      "grad_norm": 0.05003494012216922,
      "learning_rate": 0.0001207976723606173,
      "loss": 0.8448,
      "step": 127360
    },
    {
      "epoch": 2.2361698765778892,
      "grad_norm": 0.05616719051380266,
      "learning_rate": 0.00012078675369619959,
      "loss": 0.8468,
      "step": 127370
    },
    {
      "epoch": 2.2363454414578907,
      "grad_norm": 0.05386309344251496,
      "learning_rate": 0.00012077583478107992,
      "loss": 0.8394,
      "step": 127380
    },
    {
      "epoch": 2.236521006337892,
      "grad_norm": 0.05572780073858405,
      "learning_rate": 0.0001207649156153966,
      "loss": 0.8513,
      "step": 127390
    },
    {
      "epoch": 2.2366965712178937,
      "grad_norm": 0.07718166836074371,
      "learning_rate": 0.00012075399619928805,
      "loss": 0.8373,
      "step": 127400
    },
    {
      "epoch": 2.236872136097895,
      "grad_norm": 0.06672499207061582,
      "learning_rate": 0.00012074307653289254,
      "loss": 0.8376,
      "step": 127410
    },
    {
      "epoch": 2.237047700977896,
      "grad_norm": 0.07108159699886057,
      "learning_rate": 0.00012073215661634851,
      "loss": 0.852,
      "step": 127420
    },
    {
      "epoch": 2.2372232658578977,
      "grad_norm": 0.04008863013393428,
      "learning_rate": 0.00012072123644979423,
      "loss": 0.8383,
      "step": 127430
    },
    {
      "epoch": 2.237398830737899,
      "grad_norm": 0.048878392621651764,
      "learning_rate": 0.00012071031603336814,
      "loss": 0.853,
      "step": 127440
    },
    {
      "epoch": 2.2375743956179006,
      "grad_norm": 0.040562453410576675,
      "learning_rate": 0.00012069939536720858,
      "loss": 0.8438,
      "step": 127450
    },
    {
      "epoch": 2.237749960497902,
      "grad_norm": 0.06841936846319334,
      "learning_rate": 0.00012068847445145385,
      "loss": 0.8547,
      "step": 127460
    },
    {
      "epoch": 2.2379255253779036,
      "grad_norm": 0.062043539179198776,
      "learning_rate": 0.00012067755328624242,
      "loss": 0.8483,
      "step": 127470
    },
    {
      "epoch": 2.2381010902579046,
      "grad_norm": 0.0515222887281822,
      "learning_rate": 0.00012066663187171261,
      "loss": 0.8438,
      "step": 127480
    },
    {
      "epoch": 2.238276655137906,
      "grad_norm": 0.048078104386411954,
      "learning_rate": 0.00012065571020800279,
      "loss": 0.8398,
      "step": 127490
    },
    {
      "epoch": 2.2384522200179076,
      "grad_norm": 0.06331724834574518,
      "learning_rate": 0.00012064478829525139,
      "loss": 0.8406,
      "step": 127500
    },
    {
      "epoch": 2.238627784897909,
      "grad_norm": 0.05408857775549483,
      "learning_rate": 0.00012063386613359671,
      "loss": 0.8513,
      "step": 127510
    },
    {
      "epoch": 2.2388033497779105,
      "grad_norm": 0.05420122880186484,
      "learning_rate": 0.00012062294372317721,
      "loss": 0.8446,
      "step": 127520
    },
    {
      "epoch": 2.238978914657912,
      "grad_norm": 0.04161104660398838,
      "learning_rate": 0.00012061202106413125,
      "loss": 0.8432,
      "step": 127530
    },
    {
      "epoch": 2.2391544795379135,
      "grad_norm": 0.04592771082959264,
      "learning_rate": 0.00012060109815659721,
      "loss": 0.8487,
      "step": 127540
    },
    {
      "epoch": 2.2393300444179145,
      "grad_norm": 0.04161974344010487,
      "learning_rate": 0.00012059017500071354,
      "loss": 0.8456,
      "step": 127550
    },
    {
      "epoch": 2.239505609297916,
      "grad_norm": 0.0549208471819303,
      "learning_rate": 0.00012057925159661856,
      "loss": 0.847,
      "step": 127560
    },
    {
      "epoch": 2.2396811741779175,
      "grad_norm": 0.061230256495136845,
      "learning_rate": 0.00012056832794445072,
      "loss": 0.8396,
      "step": 127570
    },
    {
      "epoch": 2.239856739057919,
      "grad_norm": 0.050607266488797556,
      "learning_rate": 0.00012055740404434841,
      "loss": 0.8414,
      "step": 127580
    },
    {
      "epoch": 2.2400323039379204,
      "grad_norm": 0.07557476795461483,
      "learning_rate": 0.00012054647989645007,
      "loss": 0.8511,
      "step": 127590
    },
    {
      "epoch": 2.2402078688179214,
      "grad_norm": 0.04487835281780756,
      "learning_rate": 0.00012053555550089403,
      "loss": 0.8359,
      "step": 127600
    },
    {
      "epoch": 2.240383433697923,
      "grad_norm": 0.06741257746376572,
      "learning_rate": 0.0001205246308578188,
      "loss": 0.8493,
      "step": 127610
    },
    {
      "epoch": 2.2405589985779244,
      "grad_norm": 0.050863118017829244,
      "learning_rate": 0.00012051370596736275,
      "loss": 0.8433,
      "step": 127620
    },
    {
      "epoch": 2.240734563457926,
      "grad_norm": 0.06005233088979729,
      "learning_rate": 0.00012050278082966429,
      "loss": 0.8418,
      "step": 127630
    },
    {
      "epoch": 2.2409101283379274,
      "grad_norm": 0.06577786908152802,
      "learning_rate": 0.00012049185544486187,
      "loss": 0.8494,
      "step": 127640
    },
    {
      "epoch": 2.241085693217929,
      "grad_norm": 0.058503192094923794,
      "learning_rate": 0.00012048092981309391,
      "loss": 0.8441,
      "step": 127650
    },
    {
      "epoch": 2.2412612580979303,
      "grad_norm": 0.0590534186115401,
      "learning_rate": 0.00012047000393449883,
      "loss": 0.8491,
      "step": 127660
    },
    {
      "epoch": 2.2414368229779313,
      "grad_norm": 0.07115154207453026,
      "learning_rate": 0.0001204590778092151,
      "loss": 0.8438,
      "step": 127670
    },
    {
      "epoch": 2.241612387857933,
      "grad_norm": 0.08093564099255758,
      "learning_rate": 0.00012044815143738106,
      "loss": 0.8474,
      "step": 127680
    },
    {
      "epoch": 2.2417879527379343,
      "grad_norm": 0.051001426231876555,
      "learning_rate": 0.00012043722481913526,
      "loss": 0.8449,
      "step": 127690
    },
    {
      "epoch": 2.2419635176179358,
      "grad_norm": 0.04811234529075256,
      "learning_rate": 0.00012042629795461609,
      "loss": 0.8419,
      "step": 127700
    },
    {
      "epoch": 2.2421390824979373,
      "grad_norm": 0.05945939353237796,
      "learning_rate": 0.00012041537084396199,
      "loss": 0.8469,
      "step": 127710
    },
    {
      "epoch": 2.2423146473779383,
      "grad_norm": 0.050002411317756686,
      "learning_rate": 0.00012040444348731146,
      "loss": 0.8451,
      "step": 127720
    },
    {
      "epoch": 2.2424902122579398,
      "grad_norm": 0.08294293376518933,
      "learning_rate": 0.00012039351588480285,
      "loss": 0.8521,
      "step": 127730
    },
    {
      "epoch": 2.2426657771379412,
      "grad_norm": 0.05417032205793188,
      "learning_rate": 0.00012038258803657472,
      "loss": 0.8523,
      "step": 127740
    },
    {
      "epoch": 2.2428413420179427,
      "grad_norm": 0.062023417435839764,
      "learning_rate": 0.00012037165994276545,
      "loss": 0.8514,
      "step": 127750
    },
    {
      "epoch": 2.243016906897944,
      "grad_norm": 0.05556749597666896,
      "learning_rate": 0.00012036073160351358,
      "loss": 0.8491,
      "step": 127760
    },
    {
      "epoch": 2.2431924717779457,
      "grad_norm": 0.06576451850135229,
      "learning_rate": 0.00012034980301895749,
      "loss": 0.8407,
      "step": 127770
    },
    {
      "epoch": 2.243368036657947,
      "grad_norm": 0.06110549067518934,
      "learning_rate": 0.0001203388741892357,
      "loss": 0.8457,
      "step": 127780
    },
    {
      "epoch": 2.243543601537948,
      "grad_norm": 0.05854086673256772,
      "learning_rate": 0.00012032794511448669,
      "loss": 0.8463,
      "step": 127790
    },
    {
      "epoch": 2.2437191664179497,
      "grad_norm": 0.0452600777635768,
      "learning_rate": 0.0001203170157948489,
      "loss": 0.8487,
      "step": 127800
    },
    {
      "epoch": 2.243894731297951,
      "grad_norm": 0.06230528180066106,
      "learning_rate": 0.0001203060862304608,
      "loss": 0.8497,
      "step": 127810
    },
    {
      "epoch": 2.2440702961779526,
      "grad_norm": 0.050887777695409635,
      "learning_rate": 0.0001202951564214609,
      "loss": 0.8458,
      "step": 127820
    },
    {
      "epoch": 2.244245861057954,
      "grad_norm": 0.05131143702692972,
      "learning_rate": 0.00012028422636798766,
      "loss": 0.8441,
      "step": 127830
    },
    {
      "epoch": 2.2444214259379556,
      "grad_norm": 0.04679862232566087,
      "learning_rate": 0.00012027329607017961,
      "loss": 0.8511,
      "step": 127840
    },
    {
      "epoch": 2.2445969908179566,
      "grad_norm": 0.07384193750348465,
      "learning_rate": 0.00012026236552817516,
      "loss": 0.846,
      "step": 127850
    },
    {
      "epoch": 2.244772555697958,
      "grad_norm": 0.04834016687838362,
      "learning_rate": 0.00012025143474211289,
      "loss": 0.8433,
      "step": 127860
    },
    {
      "epoch": 2.2449481205779596,
      "grad_norm": 0.04618925652587663,
      "learning_rate": 0.0001202405037121312,
      "loss": 0.8449,
      "step": 127870
    },
    {
      "epoch": 2.245123685457961,
      "grad_norm": 0.05392338022026461,
      "learning_rate": 0.0001202295724383687,
      "loss": 0.8494,
      "step": 127880
    },
    {
      "epoch": 2.2452992503379625,
      "grad_norm": 0.05367631157373896,
      "learning_rate": 0.0001202186409209638,
      "loss": 0.8412,
      "step": 127890
    },
    {
      "epoch": 2.245474815217964,
      "grad_norm": 0.054883681539805704,
      "learning_rate": 0.00012020770916005504,
      "loss": 0.8356,
      "step": 127900
    },
    {
      "epoch": 2.245650380097965,
      "grad_norm": 0.06484077660874309,
      "learning_rate": 0.0001201967771557809,
      "loss": 0.8381,
      "step": 127910
    },
    {
      "epoch": 2.2458259449779665,
      "grad_norm": 0.07652584890937958,
      "learning_rate": 0.00012018584490827995,
      "loss": 0.844,
      "step": 127920
    },
    {
      "epoch": 2.246001509857968,
      "grad_norm": 0.050771552571713674,
      "learning_rate": 0.00012017491241769066,
      "loss": 0.8476,
      "step": 127930
    },
    {
      "epoch": 2.2461770747379695,
      "grad_norm": 0.06577278789734202,
      "learning_rate": 0.00012016397968415153,
      "loss": 0.8425,
      "step": 127940
    },
    {
      "epoch": 2.246352639617971,
      "grad_norm": 0.05219347956392632,
      "learning_rate": 0.00012015304670780115,
      "loss": 0.848,
      "step": 127950
    },
    {
      "epoch": 2.2465282044979724,
      "grad_norm": 0.047634636200193474,
      "learning_rate": 0.00012014211348877798,
      "loss": 0.8428,
      "step": 127960
    },
    {
      "epoch": 2.2467037693779734,
      "grad_norm": 0.04196046724706411,
      "learning_rate": 0.00012013118002722054,
      "loss": 0.8498,
      "step": 127970
    },
    {
      "epoch": 2.246879334257975,
      "grad_norm": 0.05014883572543192,
      "learning_rate": 0.00012012024632326742,
      "loss": 0.8536,
      "step": 127980
    },
    {
      "epoch": 2.2470548991379764,
      "grad_norm": 0.057760193843324086,
      "learning_rate": 0.00012010931237705707,
      "loss": 0.8393,
      "step": 127990
    },
    {
      "epoch": 2.247230464017978,
      "grad_norm": 0.06211715323321185,
      "learning_rate": 0.00012009837818872808,
      "loss": 0.8462,
      "step": 128000
    },
    {
      "epoch": 2.2474060288979794,
      "grad_norm": 0.048407856334641335,
      "learning_rate": 0.00012008744375841903,
      "loss": 0.8526,
      "step": 128010
    },
    {
      "epoch": 2.247581593777981,
      "grad_norm": 0.05652547433648526,
      "learning_rate": 0.00012007650908626835,
      "loss": 0.8483,
      "step": 128020
    },
    {
      "epoch": 2.247757158657982,
      "grad_norm": 0.04483441715001763,
      "learning_rate": 0.00012006557417241469,
      "loss": 0.844,
      "step": 128030
    },
    {
      "epoch": 2.2479327235379833,
      "grad_norm": 0.05064505364581861,
      "learning_rate": 0.00012005463901699649,
      "loss": 0.8403,
      "step": 128040
    },
    {
      "epoch": 2.248108288417985,
      "grad_norm": 0.05509192719538145,
      "learning_rate": 0.00012004370362015242,
      "loss": 0.8407,
      "step": 128050
    },
    {
      "epoch": 2.2482838532979863,
      "grad_norm": 0.055436586314752505,
      "learning_rate": 0.00012003276798202095,
      "loss": 0.842,
      "step": 128060
    },
    {
      "epoch": 2.248459418177988,
      "grad_norm": 0.06358946168356959,
      "learning_rate": 0.00012002183210274066,
      "loss": 0.8554,
      "step": 128070
    },
    {
      "epoch": 2.2486349830579893,
      "grad_norm": 0.07174397724627936,
      "learning_rate": 0.00012001089598245012,
      "loss": 0.8387,
      "step": 128080
    },
    {
      "epoch": 2.2488105479379903,
      "grad_norm": 0.09524476515321627,
      "learning_rate": 0.0001199999596212879,
      "loss": 0.858,
      "step": 128090
    },
    {
      "epoch": 2.2489861128179918,
      "grad_norm": 0.04632378523523165,
      "learning_rate": 0.00011998902301939252,
      "loss": 0.8414,
      "step": 128100
    },
    {
      "epoch": 2.2491616776979932,
      "grad_norm": 0.04873433943446527,
      "learning_rate": 0.00011997808617690258,
      "loss": 0.8416,
      "step": 128110
    },
    {
      "epoch": 2.2493372425779947,
      "grad_norm": 0.054624505429070616,
      "learning_rate": 0.00011996714909395665,
      "loss": 0.8423,
      "step": 128120
    },
    {
      "epoch": 2.249512807457996,
      "grad_norm": 0.0566616030348163,
      "learning_rate": 0.00011995621177069334,
      "loss": 0.8391,
      "step": 128130
    },
    {
      "epoch": 2.2496883723379977,
      "grad_norm": 0.04297834493865435,
      "learning_rate": 0.00011994527420725116,
      "loss": 0.8421,
      "step": 128140
    },
    {
      "epoch": 2.2498639372179987,
      "grad_norm": 0.06599351450731214,
      "learning_rate": 0.00011993433640376877,
      "loss": 0.8452,
      "step": 128150
    },
    {
      "epoch": 2.250039502098,
      "grad_norm": 0.0711317104756666,
      "learning_rate": 0.00011992339836038465,
      "loss": 0.8458,
      "step": 128160
    },
    {
      "epoch": 2.2502150669780017,
      "grad_norm": 0.06529629568761311,
      "learning_rate": 0.00011991246007723747,
      "loss": 0.8406,
      "step": 128170
    },
    {
      "epoch": 2.250390631858003,
      "grad_norm": 0.05008751918166789,
      "learning_rate": 0.00011990152155446581,
      "loss": 0.8446,
      "step": 128180
    },
    {
      "epoch": 2.2505661967380046,
      "grad_norm": 0.047516861812579206,
      "learning_rate": 0.00011989058279220824,
      "loss": 0.8427,
      "step": 128190
    },
    {
      "epoch": 2.250741761618006,
      "grad_norm": 0.06785645912376191,
      "learning_rate": 0.00011987964379060337,
      "loss": 0.8379,
      "step": 128200
    },
    {
      "epoch": 2.2509173264980076,
      "grad_norm": 0.051872200204142925,
      "learning_rate": 0.00011986870454978978,
      "loss": 0.849,
      "step": 128210
    },
    {
      "epoch": 2.2510928913780086,
      "grad_norm": 0.06372857168697321,
      "learning_rate": 0.0001198577650699061,
      "loss": 0.8487,
      "step": 128220
    },
    {
      "epoch": 2.25126845625801,
      "grad_norm": 0.06796493771244702,
      "learning_rate": 0.0001198468253510909,
      "loss": 0.8497,
      "step": 128230
    },
    {
      "epoch": 2.2514440211380116,
      "grad_norm": 0.05471695906352941,
      "learning_rate": 0.00011983588539348282,
      "loss": 0.8486,
      "step": 128240
    },
    {
      "epoch": 2.251619586018013,
      "grad_norm": 0.050153442746337144,
      "learning_rate": 0.0001198249451972205,
      "loss": 0.8434,
      "step": 128250
    },
    {
      "epoch": 2.2517951508980145,
      "grad_norm": 0.06623710628659894,
      "learning_rate": 0.00011981400476244246,
      "loss": 0.8444,
      "step": 128260
    },
    {
      "epoch": 2.2519707157780156,
      "grad_norm": 0.061799800584016844,
      "learning_rate": 0.00011980306408928743,
      "loss": 0.8466,
      "step": 128270
    },
    {
      "epoch": 2.252146280658017,
      "grad_norm": 0.06187129068418657,
      "learning_rate": 0.00011979212317789392,
      "loss": 0.8448,
      "step": 128280
    },
    {
      "epoch": 2.2523218455380185,
      "grad_norm": 0.05838236552609495,
      "learning_rate": 0.00011978118202840062,
      "loss": 0.8444,
      "step": 128290
    },
    {
      "epoch": 2.25249741041802,
      "grad_norm": 0.10139481726466139,
      "learning_rate": 0.0001197702406409462,
      "loss": 0.8489,
      "step": 128300
    },
    {
      "epoch": 2.2526729752980215,
      "grad_norm": 0.05767646991297251,
      "learning_rate": 0.00011975929901566918,
      "loss": 0.8381,
      "step": 128310
    },
    {
      "epoch": 2.252848540178023,
      "grad_norm": 0.04663109369127097,
      "learning_rate": 0.00011974835715270825,
      "loss": 0.8496,
      "step": 128320
    },
    {
      "epoch": 2.2530241050580244,
      "grad_norm": 0.06738488341629936,
      "learning_rate": 0.00011973741505220203,
      "loss": 0.8453,
      "step": 128330
    },
    {
      "epoch": 2.2531996699380255,
      "grad_norm": 0.05587725114651585,
      "learning_rate": 0.0001197264727142892,
      "loss": 0.8469,
      "step": 128340
    },
    {
      "epoch": 2.253375234818027,
      "grad_norm": 0.061583466364615536,
      "learning_rate": 0.0001197155301391084,
      "loss": 0.8555,
      "step": 128350
    },
    {
      "epoch": 2.2535507996980284,
      "grad_norm": 0.05737171752425201,
      "learning_rate": 0.00011970458732679819,
      "loss": 0.8515,
      "step": 128360
    },
    {
      "epoch": 2.25372636457803,
      "grad_norm": 0.07621597394769182,
      "learning_rate": 0.00011969364427749727,
      "loss": 0.8389,
      "step": 128370
    },
    {
      "epoch": 2.2539019294580314,
      "grad_norm": 0.05819792472221031,
      "learning_rate": 0.00011968270099134433,
      "loss": 0.8361,
      "step": 128380
    },
    {
      "epoch": 2.2540774943380324,
      "grad_norm": 0.07056257788937743,
      "learning_rate": 0.00011967175746847797,
      "loss": 0.8478,
      "step": 128390
    },
    {
      "epoch": 2.254253059218034,
      "grad_norm": 0.04515653624560417,
      "learning_rate": 0.00011966081370903688,
      "loss": 0.8517,
      "step": 128400
    },
    {
      "epoch": 2.2544286240980353,
      "grad_norm": 0.0649929536184501,
      "learning_rate": 0.00011964986971315968,
      "loss": 0.8448,
      "step": 128410
    },
    {
      "epoch": 2.254604188978037,
      "grad_norm": 0.04489463433818073,
      "learning_rate": 0.0001196389254809851,
      "loss": 0.8472,
      "step": 128420
    },
    {
      "epoch": 2.2547797538580383,
      "grad_norm": 0.059994200173014475,
      "learning_rate": 0.00011962798101265171,
      "loss": 0.8543,
      "step": 128430
    },
    {
      "epoch": 2.25495531873804,
      "grad_norm": 0.0475703998049247,
      "learning_rate": 0.00011961703630829826,
      "loss": 0.8427,
      "step": 128440
    },
    {
      "epoch": 2.2551308836180413,
      "grad_norm": 0.05790113153554155,
      "learning_rate": 0.00011960609136806338,
      "loss": 0.8448,
      "step": 128450
    },
    {
      "epoch": 2.2553064484980423,
      "grad_norm": 0.05757026380798913,
      "learning_rate": 0.00011959514619208576,
      "loss": 0.854,
      "step": 128460
    },
    {
      "epoch": 2.2554820133780438,
      "grad_norm": 0.06405550238337962,
      "learning_rate": 0.00011958420078050408,
      "loss": 0.8467,
      "step": 128470
    },
    {
      "epoch": 2.2556575782580452,
      "grad_norm": 0.051013608384510106,
      "learning_rate": 0.000119573255133457,
      "loss": 0.8366,
      "step": 128480
    },
    {
      "epoch": 2.2558331431380467,
      "grad_norm": 0.07210407954662754,
      "learning_rate": 0.00011956230925108323,
      "loss": 0.8509,
      "step": 128490
    },
    {
      "epoch": 2.256008708018048,
      "grad_norm": 0.09117962769974118,
      "learning_rate": 0.00011955136313352142,
      "loss": 0.8378,
      "step": 128500
    },
    {
      "epoch": 2.2561842728980492,
      "grad_norm": 0.05171681448831035,
      "learning_rate": 0.00011954041678091032,
      "loss": 0.8535,
      "step": 128510
    },
    {
      "epoch": 2.2563598377780507,
      "grad_norm": 0.12385349977343448,
      "learning_rate": 0.0001195294701933886,
      "loss": 0.8561,
      "step": 128520
    },
    {
      "epoch": 2.256535402658052,
      "grad_norm": 0.051481911535265996,
      "learning_rate": 0.0001195185233710949,
      "loss": 0.8509,
      "step": 128530
    },
    {
      "epoch": 2.2567109675380537,
      "grad_norm": 0.05677915412963972,
      "learning_rate": 0.00011950757631416796,
      "loss": 0.843,
      "step": 128540
    },
    {
      "epoch": 2.256886532418055,
      "grad_norm": 0.047572117502051,
      "learning_rate": 0.0001194966290227465,
      "loss": 0.8483,
      "step": 128550
    },
    {
      "epoch": 2.2570620972980566,
      "grad_norm": 0.047085237980671324,
      "learning_rate": 0.0001194856814969692,
      "loss": 0.8483,
      "step": 128560
    },
    {
      "epoch": 2.257237662178058,
      "grad_norm": 0.04338139919355072,
      "learning_rate": 0.00011947473373697477,
      "loss": 0.837,
      "step": 128570
    },
    {
      "epoch": 2.257413227058059,
      "grad_norm": 0.041388391937648684,
      "learning_rate": 0.00011946378574290191,
      "loss": 0.845,
      "step": 128580
    },
    {
      "epoch": 2.2575887919380606,
      "grad_norm": 0.04563434591388101,
      "learning_rate": 0.00011945283751488938,
      "loss": 0.848,
      "step": 128590
    },
    {
      "epoch": 2.257764356818062,
      "grad_norm": 0.062396487350686425,
      "learning_rate": 0.00011944188905307583,
      "loss": 0.8452,
      "step": 128600
    },
    {
      "epoch": 2.2579399216980636,
      "grad_norm": 0.0588803562004438,
      "learning_rate": 0.00011943094035760003,
      "loss": 0.8402,
      "step": 128610
    },
    {
      "epoch": 2.258115486578065,
      "grad_norm": 0.066373537622859,
      "learning_rate": 0.00011941999142860065,
      "loss": 0.8443,
      "step": 128620
    },
    {
      "epoch": 2.258291051458066,
      "grad_norm": 0.06280810540122657,
      "learning_rate": 0.0001194090422662165,
      "loss": 0.8453,
      "step": 128630
    },
    {
      "epoch": 2.2584666163380676,
      "grad_norm": 0.06676546185459425,
      "learning_rate": 0.00011939809287058626,
      "loss": 0.8439,
      "step": 128640
    },
    {
      "epoch": 2.258642181218069,
      "grad_norm": 0.060672489982477006,
      "learning_rate": 0.00011938714324184859,
      "loss": 0.8401,
      "step": 128650
    },
    {
      "epoch": 2.2588177460980705,
      "grad_norm": 0.05056339385951393,
      "learning_rate": 0.00011937619338014236,
      "loss": 0.8421,
      "step": 128660
    },
    {
      "epoch": 2.258993310978072,
      "grad_norm": 0.055790992082679666,
      "learning_rate": 0.00011936524328560617,
      "loss": 0.8415,
      "step": 128670
    },
    {
      "epoch": 2.2591688758580735,
      "grad_norm": 0.09056493606918961,
      "learning_rate": 0.00011935429295837888,
      "loss": 0.845,
      "step": 128680
    },
    {
      "epoch": 2.259344440738075,
      "grad_norm": 0.069733168422254,
      "learning_rate": 0.00011934334239859916,
      "loss": 0.846,
      "step": 128690
    },
    {
      "epoch": 2.259520005618076,
      "grad_norm": 0.040228691703988566,
      "learning_rate": 0.00011933239160640582,
      "loss": 0.8452,
      "step": 128700
    },
    {
      "epoch": 2.2596955704980775,
      "grad_norm": 0.06961700494381752,
      "learning_rate": 0.00011932144058193753,
      "loss": 0.8445,
      "step": 128710
    },
    {
      "epoch": 2.259871135378079,
      "grad_norm": 0.05596952710503413,
      "learning_rate": 0.00011931048932533304,
      "loss": 0.8565,
      "step": 128720
    },
    {
      "epoch": 2.2600467002580804,
      "grad_norm": 0.04689436572418216,
      "learning_rate": 0.0001192995378367312,
      "loss": 0.8483,
      "step": 128730
    },
    {
      "epoch": 2.260222265138082,
      "grad_norm": 0.054401472174705555,
      "learning_rate": 0.0001192885861162707,
      "loss": 0.8414,
      "step": 128740
    },
    {
      "epoch": 2.2603978300180834,
      "grad_norm": 0.07803545844379477,
      "learning_rate": 0.00011927763416409028,
      "loss": 0.8444,
      "step": 128750
    },
    {
      "epoch": 2.2605733948980844,
      "grad_norm": 0.06192113681842398,
      "learning_rate": 0.00011926668198032876,
      "loss": 0.8443,
      "step": 128760
    },
    {
      "epoch": 2.260748959778086,
      "grad_norm": 0.057075963234159255,
      "learning_rate": 0.00011925572956512485,
      "loss": 0.8424,
      "step": 128770
    },
    {
      "epoch": 2.2609245246580874,
      "grad_norm": 0.06687259583707748,
      "learning_rate": 0.00011924477691861736,
      "loss": 0.8438,
      "step": 128780
    },
    {
      "epoch": 2.261100089538089,
      "grad_norm": 0.08156946047359688,
      "learning_rate": 0.00011923382404094505,
      "loss": 0.8515,
      "step": 128790
    },
    {
      "epoch": 2.2612756544180903,
      "grad_norm": 0.06636762341431478,
      "learning_rate": 0.00011922287093224668,
      "loss": 0.858,
      "step": 128800
    },
    {
      "epoch": 2.261451219298092,
      "grad_norm": 0.0675332470849134,
      "learning_rate": 0.00011921191759266108,
      "loss": 0.8423,
      "step": 128810
    },
    {
      "epoch": 2.261626784178093,
      "grad_norm": 0.06689572991419672,
      "learning_rate": 0.00011920096402232697,
      "loss": 0.8405,
      "step": 128820
    },
    {
      "epoch": 2.2618023490580943,
      "grad_norm": 0.05223008113191984,
      "learning_rate": 0.00011919001022138319,
      "loss": 0.8517,
      "step": 128830
    },
    {
      "epoch": 2.2619779139380958,
      "grad_norm": 0.05488393026727925,
      "learning_rate": 0.00011917905618996843,
      "loss": 0.84,
      "step": 128840
    },
    {
      "epoch": 2.2621534788180973,
      "grad_norm": 0.051427643869491994,
      "learning_rate": 0.00011916810192822157,
      "loss": 0.8522,
      "step": 128850
    },
    {
      "epoch": 2.2623290436980987,
      "grad_norm": 0.051293824899979586,
      "learning_rate": 0.00011915714743628142,
      "loss": 0.854,
      "step": 128860
    },
    {
      "epoch": 2.2625046085781,
      "grad_norm": 0.052748337123286644,
      "learning_rate": 0.00011914619271428667,
      "loss": 0.8505,
      "step": 128870
    },
    {
      "epoch": 2.2626801734581017,
      "grad_norm": 0.056719268216248814,
      "learning_rate": 0.00011913523776237622,
      "loss": 0.8468,
      "step": 128880
    },
    {
      "epoch": 2.2628557383381027,
      "grad_norm": 0.04965753379087251,
      "learning_rate": 0.0001191242825806888,
      "loss": 0.855,
      "step": 128890
    },
    {
      "epoch": 2.263031303218104,
      "grad_norm": 0.053292097875048745,
      "learning_rate": 0.00011911332716936329,
      "loss": 0.8392,
      "step": 128900
    },
    {
      "epoch": 2.2632068680981057,
      "grad_norm": 0.050790342624577836,
      "learning_rate": 0.00011910237152853843,
      "loss": 0.8465,
      "step": 128910
    },
    {
      "epoch": 2.263382432978107,
      "grad_norm": 0.0521877062923961,
      "learning_rate": 0.00011909141565835305,
      "loss": 0.845,
      "step": 128920
    },
    {
      "epoch": 2.2635579978581086,
      "grad_norm": 0.06382062226312961,
      "learning_rate": 0.00011908045955894598,
      "loss": 0.8488,
      "step": 128930
    },
    {
      "epoch": 2.2637335627381097,
      "grad_norm": 0.05340611316476193,
      "learning_rate": 0.000119069503230456,
      "loss": 0.8398,
      "step": 128940
    },
    {
      "epoch": 2.263909127618111,
      "grad_norm": 0.06559357088304377,
      "learning_rate": 0.00011905854667302197,
      "loss": 0.8448,
      "step": 128950
    },
    {
      "epoch": 2.2640846924981126,
      "grad_norm": 0.06479807681978746,
      "learning_rate": 0.00011904758988678269,
      "loss": 0.8499,
      "step": 128960
    },
    {
      "epoch": 2.264260257378114,
      "grad_norm": 0.04393978046740033,
      "learning_rate": 0.00011903663287187699,
      "loss": 0.8402,
      "step": 128970
    },
    {
      "epoch": 2.2644358222581156,
      "grad_norm": 0.06086719383965293,
      "learning_rate": 0.00011902567562844369,
      "loss": 0.8396,
      "step": 128980
    },
    {
      "epoch": 2.264611387138117,
      "grad_norm": 0.04981113300432262,
      "learning_rate": 0.00011901471815662162,
      "loss": 0.8393,
      "step": 128990
    },
    {
      "epoch": 2.2647869520181185,
      "grad_norm": 0.05122712513331021,
      "learning_rate": 0.00011900376045654965,
      "loss": 0.845,
      "step": 129000
    },
    {
      "epoch": 2.2649625168981196,
      "grad_norm": 0.04311122617790364,
      "learning_rate": 0.0001189928025283665,
      "loss": 0.8413,
      "step": 129010
    },
    {
      "epoch": 2.265138081778121,
      "grad_norm": 0.04999881742061369,
      "learning_rate": 0.00011898184437221117,
      "loss": 0.8438,
      "step": 129020
    },
    {
      "epoch": 2.2653136466581225,
      "grad_norm": 0.06288071637446098,
      "learning_rate": 0.00011897088598822241,
      "loss": 0.8464,
      "step": 129030
    },
    {
      "epoch": 2.265489211538124,
      "grad_norm": 0.04995700196822464,
      "learning_rate": 0.00011895992737653907,
      "loss": 0.8433,
      "step": 129040
    },
    {
      "epoch": 2.2656647764181255,
      "grad_norm": 0.045328697418117556,
      "learning_rate": 0.00011894896853730002,
      "loss": 0.8547,
      "step": 129050
    },
    {
      "epoch": 2.2658403412981265,
      "grad_norm": 0.052531732684747534,
      "learning_rate": 0.00011893800947064411,
      "loss": 0.8498,
      "step": 129060
    },
    {
      "epoch": 2.266015906178128,
      "grad_norm": 0.06645497773617823,
      "learning_rate": 0.00011892705017671018,
      "loss": 0.8411,
      "step": 129070
    },
    {
      "epoch": 2.2661914710581295,
      "grad_norm": 0.05732260616410579,
      "learning_rate": 0.00011891609065563707,
      "loss": 0.845,
      "step": 129080
    },
    {
      "epoch": 2.266367035938131,
      "grad_norm": 0.046365735790160595,
      "learning_rate": 0.00011890513090756364,
      "loss": 0.8475,
      "step": 129090
    },
    {
      "epoch": 2.2665426008181324,
      "grad_norm": 0.04604328642341464,
      "learning_rate": 0.0001188941709326288,
      "loss": 0.8456,
      "step": 129100
    },
    {
      "epoch": 2.266718165698134,
      "grad_norm": 0.06850647325222535,
      "learning_rate": 0.00011888321073097135,
      "loss": 0.8413,
      "step": 129110
    },
    {
      "epoch": 2.2668937305781354,
      "grad_norm": 0.07088630697786258,
      "learning_rate": 0.00011887225030273024,
      "loss": 0.8433,
      "step": 129120
    },
    {
      "epoch": 2.2670692954581364,
      "grad_norm": 0.059783499091725985,
      "learning_rate": 0.00011886128964804426,
      "loss": 0.8495,
      "step": 129130
    },
    {
      "epoch": 2.267244860338138,
      "grad_norm": 0.0524798732890788,
      "learning_rate": 0.00011885032876705233,
      "loss": 0.8423,
      "step": 129140
    },
    {
      "epoch": 2.2674204252181394,
      "grad_norm": 0.05453577132980929,
      "learning_rate": 0.00011883936765989334,
      "loss": 0.8495,
      "step": 129150
    },
    {
      "epoch": 2.267595990098141,
      "grad_norm": 0.06481405471189107,
      "learning_rate": 0.00011882840632670611,
      "loss": 0.847,
      "step": 129160
    },
    {
      "epoch": 2.2677715549781423,
      "grad_norm": 0.0618537446003354,
      "learning_rate": 0.00011881744476762957,
      "loss": 0.8457,
      "step": 129170
    },
    {
      "epoch": 2.2679471198581433,
      "grad_norm": 0.050764107989486544,
      "learning_rate": 0.00011880648298280254,
      "loss": 0.8446,
      "step": 129180
    },
    {
      "epoch": 2.268122684738145,
      "grad_norm": 0.044845734562794115,
      "learning_rate": 0.00011879552097236403,
      "loss": 0.8482,
      "step": 129190
    },
    {
      "epoch": 2.2682982496181463,
      "grad_norm": 0.06359242211063387,
      "learning_rate": 0.00011878455873645285,
      "loss": 0.8447,
      "step": 129200
    },
    {
      "epoch": 2.268473814498148,
      "grad_norm": 0.058670792500878755,
      "learning_rate": 0.00011877359627520785,
      "loss": 0.8522,
      "step": 129210
    },
    {
      "epoch": 2.2686493793781493,
      "grad_norm": 0.054645783743175115,
      "learning_rate": 0.00011876263358876802,
      "loss": 0.8476,
      "step": 129220
    },
    {
      "epoch": 2.2688249442581507,
      "grad_norm": 0.051404881765793574,
      "learning_rate": 0.0001187516706772722,
      "loss": 0.8508,
      "step": 129230
    },
    {
      "epoch": 2.269000509138152,
      "grad_norm": 0.05039239493434629,
      "learning_rate": 0.00011874070754085933,
      "loss": 0.8301,
      "step": 129240
    },
    {
      "epoch": 2.2691760740181532,
      "grad_norm": 0.06626321855193014,
      "learning_rate": 0.0001187297441796683,
      "loss": 0.8419,
      "step": 129250
    },
    {
      "epoch": 2.2693516388981547,
      "grad_norm": 0.05344582657729159,
      "learning_rate": 0.00011871878059383799,
      "loss": 0.8385,
      "step": 129260
    },
    {
      "epoch": 2.269527203778156,
      "grad_norm": 0.05335136078793268,
      "learning_rate": 0.00011870781678350735,
      "loss": 0.8414,
      "step": 129270
    },
    {
      "epoch": 2.2697027686581577,
      "grad_norm": 0.06172758884923604,
      "learning_rate": 0.00011869685274881524,
      "loss": 0.8406,
      "step": 129280
    },
    {
      "epoch": 2.269878333538159,
      "grad_norm": 0.06618398432430601,
      "learning_rate": 0.00011868588848990065,
      "loss": 0.8472,
      "step": 129290
    },
    {
      "epoch": 2.27005389841816,
      "grad_norm": 0.06207034681398319,
      "learning_rate": 0.00011867492400690245,
      "loss": 0.8383,
      "step": 129300
    },
    {
      "epoch": 2.2702294632981617,
      "grad_norm": 0.06419387391632278,
      "learning_rate": 0.00011866395929995959,
      "loss": 0.8403,
      "step": 129310
    },
    {
      "epoch": 2.270405028178163,
      "grad_norm": 0.060039639100398626,
      "learning_rate": 0.00011865299436921098,
      "loss": 0.8485,
      "step": 129320
    },
    {
      "epoch": 2.2705805930581646,
      "grad_norm": 0.0679823066805295,
      "learning_rate": 0.00011864202921479552,
      "loss": 0.8384,
      "step": 129330
    },
    {
      "epoch": 2.270756157938166,
      "grad_norm": 0.06832794344263073,
      "learning_rate": 0.00011863106383685221,
      "loss": 0.8401,
      "step": 129340
    },
    {
      "epoch": 2.2709317228181676,
      "grad_norm": 0.06358627326632904,
      "learning_rate": 0.00011862009823551991,
      "loss": 0.8465,
      "step": 129350
    },
    {
      "epoch": 2.271107287698169,
      "grad_norm": 0.050006104767864455,
      "learning_rate": 0.0001186091324109376,
      "loss": 0.8451,
      "step": 129360
    },
    {
      "epoch": 2.27128285257817,
      "grad_norm": 0.06332275528794809,
      "learning_rate": 0.00011859816636324423,
      "loss": 0.8526,
      "step": 129370
    },
    {
      "epoch": 2.2714584174581716,
      "grad_norm": 0.045909283492552265,
      "learning_rate": 0.00011858720009257866,
      "loss": 0.8477,
      "step": 129380
    },
    {
      "epoch": 2.271633982338173,
      "grad_norm": 0.056710918397080826,
      "learning_rate": 0.00011857623359907996,
      "loss": 0.8404,
      "step": 129390
    },
    {
      "epoch": 2.2718095472181745,
      "grad_norm": 0.07986701405353673,
      "learning_rate": 0.00011856526688288697,
      "loss": 0.8467,
      "step": 129400
    },
    {
      "epoch": 2.271985112098176,
      "grad_norm": 0.058751586458308025,
      "learning_rate": 0.00011855429994413872,
      "loss": 0.8494,
      "step": 129410
    },
    {
      "epoch": 2.2721606769781775,
      "grad_norm": 0.056924664107089636,
      "learning_rate": 0.00011854333278297409,
      "loss": 0.8416,
      "step": 129420
    },
    {
      "epoch": 2.2723362418581785,
      "grad_norm": 0.05195922820050257,
      "learning_rate": 0.00011853236539953208,
      "loss": 0.8399,
      "step": 129430
    },
    {
      "epoch": 2.27251180673818,
      "grad_norm": 0.05887344258579035,
      "learning_rate": 0.00011852139779395165,
      "loss": 0.8548,
      "step": 129440
    },
    {
      "epoch": 2.2726873716181815,
      "grad_norm": 0.05113652805241019,
      "learning_rate": 0.00011851042996637174,
      "loss": 0.847,
      "step": 129450
    },
    {
      "epoch": 2.272862936498183,
      "grad_norm": 0.06754436247590408,
      "learning_rate": 0.00011849946191693133,
      "loss": 0.8387,
      "step": 129460
    },
    {
      "epoch": 2.2730385013781844,
      "grad_norm": 0.05423731463884294,
      "learning_rate": 0.00011848849364576937,
      "loss": 0.8373,
      "step": 129470
    },
    {
      "epoch": 2.273214066258186,
      "grad_norm": 0.045430220203175796,
      "learning_rate": 0.00011847752515302488,
      "loss": 0.8445,
      "step": 129480
    },
    {
      "epoch": 2.273389631138187,
      "grad_norm": 0.04471582660871143,
      "learning_rate": 0.00011846655643883677,
      "loss": 0.8398,
      "step": 129490
    },
    {
      "epoch": 2.2735651960181884,
      "grad_norm": 0.07043807658896724,
      "learning_rate": 0.00011845558750334405,
      "loss": 0.8523,
      "step": 129500
    },
    {
      "epoch": 2.27374076089819,
      "grad_norm": 0.07890608960812155,
      "learning_rate": 0.00011844461834668568,
      "loss": 0.8468,
      "step": 129510
    },
    {
      "epoch": 2.2739163257781914,
      "grad_norm": 0.0511565270027119,
      "learning_rate": 0.00011843364896900067,
      "loss": 0.8433,
      "step": 129520
    },
    {
      "epoch": 2.274091890658193,
      "grad_norm": 0.055853148133836426,
      "learning_rate": 0.00011842267937042797,
      "loss": 0.8503,
      "step": 129530
    },
    {
      "epoch": 2.2742674555381943,
      "grad_norm": 0.046149389880351464,
      "learning_rate": 0.0001184117095511066,
      "loss": 0.8381,
      "step": 129540
    },
    {
      "epoch": 2.274443020418196,
      "grad_norm": 0.07004655837883492,
      "learning_rate": 0.00011840073951117551,
      "loss": 0.8403,
      "step": 129550
    },
    {
      "epoch": 2.274618585298197,
      "grad_norm": 0.057593870100165846,
      "learning_rate": 0.00011838976925077374,
      "loss": 0.84,
      "step": 129560
    },
    {
      "epoch": 2.2747941501781983,
      "grad_norm": 0.051371258155804414,
      "learning_rate": 0.00011837879877004026,
      "loss": 0.8394,
      "step": 129570
    },
    {
      "epoch": 2.2749697150582,
      "grad_norm": 0.046230762401157026,
      "learning_rate": 0.00011836782806911407,
      "loss": 0.8401,
      "step": 129580
    },
    {
      "epoch": 2.2751452799382013,
      "grad_norm": 0.051863447218863344,
      "learning_rate": 0.00011835685714813414,
      "loss": 0.846,
      "step": 129590
    },
    {
      "epoch": 2.2753208448182027,
      "grad_norm": 0.05719602078398104,
      "learning_rate": 0.00011834588600723952,
      "loss": 0.8418,
      "step": 129600
    },
    {
      "epoch": 2.2754964096982038,
      "grad_norm": 0.06544422152431038,
      "learning_rate": 0.0001183349146465692,
      "loss": 0.8434,
      "step": 129610
    },
    {
      "epoch": 2.2756719745782052,
      "grad_norm": 0.06285023466739272,
      "learning_rate": 0.00011832394306626219,
      "loss": 0.8448,
      "step": 129620
    },
    {
      "epoch": 2.2758475394582067,
      "grad_norm": 0.044911014470704655,
      "learning_rate": 0.00011831297126645751,
      "loss": 0.8412,
      "step": 129630
    },
    {
      "epoch": 2.276023104338208,
      "grad_norm": 0.0479992437868216,
      "learning_rate": 0.00011830199924729417,
      "loss": 0.8489,
      "step": 129640
    },
    {
      "epoch": 2.2761986692182097,
      "grad_norm": 0.05117141817460023,
      "learning_rate": 0.00011829102700891119,
      "loss": 0.8429,
      "step": 129650
    },
    {
      "epoch": 2.276374234098211,
      "grad_norm": 0.06416624383069706,
      "learning_rate": 0.00011828005455144753,
      "loss": 0.8352,
      "step": 129660
    },
    {
      "epoch": 2.2765497989782126,
      "grad_norm": 0.065749076321207,
      "learning_rate": 0.00011826908187504231,
      "loss": 0.8318,
      "step": 129670
    },
    {
      "epoch": 2.2767253638582137,
      "grad_norm": 0.06279565600557396,
      "learning_rate": 0.00011825810897983452,
      "loss": 0.8514,
      "step": 129680
    },
    {
      "epoch": 2.276900928738215,
      "grad_norm": 0.05739575792114748,
      "learning_rate": 0.00011824713586596314,
      "loss": 0.8428,
      "step": 129690
    },
    {
      "epoch": 2.2770764936182166,
      "grad_norm": 0.0717804458339436,
      "learning_rate": 0.00011823616253356729,
      "loss": 0.8435,
      "step": 129700
    },
    {
      "epoch": 2.277252058498218,
      "grad_norm": 0.053729285338981386,
      "learning_rate": 0.00011822518898278594,
      "loss": 0.8504,
      "step": 129710
    },
    {
      "epoch": 2.2774276233782196,
      "grad_norm": 0.05559735601709704,
      "learning_rate": 0.00011821421521375815,
      "loss": 0.8396,
      "step": 129720
    },
    {
      "epoch": 2.2776031882582206,
      "grad_norm": 0.05150264886612537,
      "learning_rate": 0.00011820324122662294,
      "loss": 0.8505,
      "step": 129730
    },
    {
      "epoch": 2.277778753138222,
      "grad_norm": 0.047813267147626765,
      "learning_rate": 0.00011819226702151938,
      "loss": 0.846,
      "step": 129740
    },
    {
      "epoch": 2.2779543180182236,
      "grad_norm": 0.0466886593579533,
      "learning_rate": 0.0001181812925985865,
      "loss": 0.8446,
      "step": 129750
    },
    {
      "epoch": 2.278129882898225,
      "grad_norm": 0.07339992714141004,
      "learning_rate": 0.00011817031795796335,
      "loss": 0.8548,
      "step": 129760
    },
    {
      "epoch": 2.2783054477782265,
      "grad_norm": 0.06314719042527846,
      "learning_rate": 0.00011815934309978895,
      "loss": 0.8472,
      "step": 129770
    },
    {
      "epoch": 2.278481012658228,
      "grad_norm": 0.05314529959875097,
      "learning_rate": 0.00011814836802420244,
      "loss": 0.8472,
      "step": 129780
    },
    {
      "epoch": 2.2786565775382295,
      "grad_norm": 0.0514973275901852,
      "learning_rate": 0.00011813739273134276,
      "loss": 0.8448,
      "step": 129790
    },
    {
      "epoch": 2.2788321424182305,
      "grad_norm": 0.060307078092409705,
      "learning_rate": 0.00011812641722134908,
      "loss": 0.8425,
      "step": 129800
    },
    {
      "epoch": 2.279007707298232,
      "grad_norm": 0.05577619142405697,
      "learning_rate": 0.0001181154414943604,
      "loss": 0.847,
      "step": 129810
    },
    {
      "epoch": 2.2791832721782335,
      "grad_norm": 0.05057814469937263,
      "learning_rate": 0.00011810446555051581,
      "loss": 0.8495,
      "step": 129820
    },
    {
      "epoch": 2.279358837058235,
      "grad_norm": 0.06582348791286638,
      "learning_rate": 0.00011809348938995432,
      "loss": 0.8524,
      "step": 129830
    },
    {
      "epoch": 2.2795344019382364,
      "grad_norm": 0.05118899656874646,
      "learning_rate": 0.00011808251301281505,
      "loss": 0.8473,
      "step": 129840
    },
    {
      "epoch": 2.2797099668182375,
      "grad_norm": 0.04250705938769466,
      "learning_rate": 0.00011807153641923712,
      "loss": 0.8493,
      "step": 129850
    },
    {
      "epoch": 2.279885531698239,
      "grad_norm": 0.05805244291247016,
      "learning_rate": 0.0001180605596093595,
      "loss": 0.8431,
      "step": 129860
    },
    {
      "epoch": 2.2800610965782404,
      "grad_norm": 0.06231957840555816,
      "learning_rate": 0.00011804958258332133,
      "loss": 0.8383,
      "step": 129870
    },
    {
      "epoch": 2.280236661458242,
      "grad_norm": 0.05472550870322227,
      "learning_rate": 0.00011803860534126169,
      "loss": 0.8385,
      "step": 129880
    },
    {
      "epoch": 2.2804122263382434,
      "grad_norm": 0.05419459268270777,
      "learning_rate": 0.00011802762788331965,
      "loss": 0.8424,
      "step": 129890
    },
    {
      "epoch": 2.280587791218245,
      "grad_norm": 0.0632593393844415,
      "learning_rate": 0.0001180166502096343,
      "loss": 0.8478,
      "step": 129900
    },
    {
      "epoch": 2.2807633560982463,
      "grad_norm": 0.0600316395811481,
      "learning_rate": 0.00011800567232034472,
      "loss": 0.8497,
      "step": 129910
    },
    {
      "epoch": 2.2809389209782474,
      "grad_norm": 0.04796001099166595,
      "learning_rate": 0.00011799469421559004,
      "loss": 0.8543,
      "step": 129920
    },
    {
      "epoch": 2.281114485858249,
      "grad_norm": 0.049883508043356846,
      "learning_rate": 0.00011798371589550932,
      "loss": 0.8417,
      "step": 129930
    },
    {
      "epoch": 2.2812900507382503,
      "grad_norm": 0.050490564478985485,
      "learning_rate": 0.00011797273736024163,
      "loss": 0.8448,
      "step": 129940
    },
    {
      "epoch": 2.281465615618252,
      "grad_norm": 0.060264121720865725,
      "learning_rate": 0.00011796175860992617,
      "loss": 0.8427,
      "step": 129950
    },
    {
      "epoch": 2.2816411804982533,
      "grad_norm": 0.07362048454170758,
      "learning_rate": 0.00011795077964470191,
      "loss": 0.8413,
      "step": 129960
    },
    {
      "epoch": 2.2818167453782543,
      "grad_norm": 0.07903753733159112,
      "learning_rate": 0.00011793980046470805,
      "loss": 0.8493,
      "step": 129970
    },
    {
      "epoch": 2.2819923102582558,
      "grad_norm": 0.059410934266897295,
      "learning_rate": 0.00011792882107008367,
      "loss": 0.8456,
      "step": 129980
    },
    {
      "epoch": 2.2821678751382573,
      "grad_norm": 0.049542601398926175,
      "learning_rate": 0.00011791784146096791,
      "loss": 0.8511,
      "step": 129990
    },
    {
      "epoch": 2.2823434400182587,
      "grad_norm": 0.049202953978799466,
      "learning_rate": 0.00011790686163749984,
      "loss": 0.84,
      "step": 130000
    },
    {
      "epoch": 2.28251900489826,
      "grad_norm": 0.05690058778022031,
      "learning_rate": 0.00011789588159981857,
      "loss": 0.8479,
      "step": 130010
    },
    {
      "epoch": 2.2826945697782617,
      "grad_norm": 0.06180183839489033,
      "learning_rate": 0.00011788490134806327,
      "loss": 0.8439,
      "step": 130020
    },
    {
      "epoch": 2.282870134658263,
      "grad_norm": 0.06771560229280559,
      "learning_rate": 0.00011787392088237302,
      "loss": 0.8385,
      "step": 130030
    },
    {
      "epoch": 2.283045699538264,
      "grad_norm": 0.052878148800075235,
      "learning_rate": 0.00011786294020288694,
      "loss": 0.845,
      "step": 130040
    },
    {
      "epoch": 2.2832212644182657,
      "grad_norm": 0.062006793770580694,
      "learning_rate": 0.00011785195930974422,
      "loss": 0.8457,
      "step": 130050
    },
    {
      "epoch": 2.283396829298267,
      "grad_norm": 0.04911647141595334,
      "learning_rate": 0.00011784097820308393,
      "loss": 0.846,
      "step": 130060
    },
    {
      "epoch": 2.2835723941782686,
      "grad_norm": 0.044417913639716144,
      "learning_rate": 0.00011782999688304521,
      "loss": 0.8509,
      "step": 130070
    },
    {
      "epoch": 2.28374795905827,
      "grad_norm": 0.0623140963752634,
      "learning_rate": 0.00011781901534976718,
      "loss": 0.8449,
      "step": 130080
    },
    {
      "epoch": 2.283923523938271,
      "grad_norm": 0.0581410031430696,
      "learning_rate": 0.00011780803360338904,
      "loss": 0.8382,
      "step": 130090
    },
    {
      "epoch": 2.2840990888182726,
      "grad_norm": 0.04751549792219519,
      "learning_rate": 0.00011779705164404989,
      "loss": 0.8372,
      "step": 130100
    },
    {
      "epoch": 2.284274653698274,
      "grad_norm": 0.054055911529362614,
      "learning_rate": 0.00011778606947188885,
      "loss": 0.8345,
      "step": 130110
    },
    {
      "epoch": 2.2844502185782756,
      "grad_norm": 0.048129277851087934,
      "learning_rate": 0.00011777508708704514,
      "loss": 0.8483,
      "step": 130120
    },
    {
      "epoch": 2.284625783458277,
      "grad_norm": 0.10154387850153107,
      "learning_rate": 0.00011776410448965782,
      "loss": 0.8441,
      "step": 130130
    },
    {
      "epoch": 2.2848013483382785,
      "grad_norm": 0.06942838827598717,
      "learning_rate": 0.0001177531216798661,
      "loss": 0.8475,
      "step": 130140
    },
    {
      "epoch": 2.28497691321828,
      "grad_norm": 0.041660276723956065,
      "learning_rate": 0.0001177421386578091,
      "loss": 0.8493,
      "step": 130150
    },
    {
      "epoch": 2.285152478098281,
      "grad_norm": 0.04971913168920996,
      "learning_rate": 0.00011773115542362602,
      "loss": 0.8495,
      "step": 130160
    },
    {
      "epoch": 2.2853280429782825,
      "grad_norm": 0.0540441073756219,
      "learning_rate": 0.00011772017197745598,
      "loss": 0.8399,
      "step": 130170
    },
    {
      "epoch": 2.285503607858284,
      "grad_norm": 0.05517951699811016,
      "learning_rate": 0.00011770918831943814,
      "loss": 0.8491,
      "step": 130180
    },
    {
      "epoch": 2.2856791727382855,
      "grad_norm": 0.0594662062307073,
      "learning_rate": 0.00011769820444971171,
      "loss": 0.845,
      "step": 130190
    },
    {
      "epoch": 2.285854737618287,
      "grad_norm": 0.08453504144678099,
      "learning_rate": 0.00011768722036841582,
      "loss": 0.8412,
      "step": 130200
    },
    {
      "epoch": 2.2860303024982884,
      "grad_norm": 0.05319165690961686,
      "learning_rate": 0.00011767623607568965,
      "loss": 0.85,
      "step": 130210
    },
    {
      "epoch": 2.2862058673782895,
      "grad_norm": 0.06848429385570928,
      "learning_rate": 0.00011766525157167237,
      "loss": 0.8381,
      "step": 130220
    },
    {
      "epoch": 2.286381432258291,
      "grad_norm": 0.06365929768802772,
      "learning_rate": 0.00011765426685650316,
      "loss": 0.8441,
      "step": 130230
    },
    {
      "epoch": 2.2865569971382924,
      "grad_norm": 0.05679083123805447,
      "learning_rate": 0.00011764328193032121,
      "loss": 0.8397,
      "step": 130240
    },
    {
      "epoch": 2.286732562018294,
      "grad_norm": 0.05881170328216437,
      "learning_rate": 0.00011763229679326567,
      "loss": 0.8514,
      "step": 130250
    },
    {
      "epoch": 2.2869081268982954,
      "grad_norm": 0.04939913909688884,
      "learning_rate": 0.00011762131144547575,
      "loss": 0.8452,
      "step": 130260
    },
    {
      "epoch": 2.287083691778297,
      "grad_norm": 0.05299339532555847,
      "learning_rate": 0.00011761032588709065,
      "loss": 0.8463,
      "step": 130270
    },
    {
      "epoch": 2.287259256658298,
      "grad_norm": 0.05375639071371556,
      "learning_rate": 0.00011759934011824949,
      "loss": 0.8477,
      "step": 130280
    },
    {
      "epoch": 2.2874348215382994,
      "grad_norm": 0.05592721852949636,
      "learning_rate": 0.00011758835413909155,
      "loss": 0.8416,
      "step": 130290
    },
    {
      "epoch": 2.287610386418301,
      "grad_norm": 0.06963610907332314,
      "learning_rate": 0.00011757736794975594,
      "loss": 0.8427,
      "step": 130300
    },
    {
      "epoch": 2.2877859512983023,
      "grad_norm": 0.04847255716639059,
      "learning_rate": 0.00011756638155038193,
      "loss": 0.8446,
      "step": 130310
    },
    {
      "epoch": 2.287961516178304,
      "grad_norm": 0.050448783392512075,
      "learning_rate": 0.00011755539494110867,
      "loss": 0.8427,
      "step": 130320
    },
    {
      "epoch": 2.2881370810583053,
      "grad_norm": 0.04510595109748189,
      "learning_rate": 0.00011754440812207545,
      "loss": 0.8515,
      "step": 130330
    },
    {
      "epoch": 2.2883126459383067,
      "grad_norm": 0.05140627605238121,
      "learning_rate": 0.00011753342109342136,
      "loss": 0.8447,
      "step": 130340
    },
    {
      "epoch": 2.2884882108183078,
      "grad_norm": 0.05260233603945596,
      "learning_rate": 0.00011752243385528563,
      "loss": 0.8421,
      "step": 130350
    },
    {
      "epoch": 2.2886637756983093,
      "grad_norm": 0.06114952751444475,
      "learning_rate": 0.00011751144640780754,
      "loss": 0.8433,
      "step": 130360
    },
    {
      "epoch": 2.2888393405783107,
      "grad_norm": 0.047542308429475676,
      "learning_rate": 0.00011750045875112624,
      "loss": 0.8501,
      "step": 130370
    },
    {
      "epoch": 2.289014905458312,
      "grad_norm": 0.05044354449855565,
      "learning_rate": 0.00011748947088538095,
      "loss": 0.852,
      "step": 130380
    },
    {
      "epoch": 2.2891904703383137,
      "grad_norm": 0.044438386448040774,
      "learning_rate": 0.00011747848281071097,
      "loss": 0.8468,
      "step": 130390
    },
    {
      "epoch": 2.2893660352183147,
      "grad_norm": 0.057564139439678676,
      "learning_rate": 0.00011746749452725539,
      "loss": 0.8435,
      "step": 130400
    },
    {
      "epoch": 2.289541600098316,
      "grad_norm": 0.06920402807724939,
      "learning_rate": 0.0001174565060351535,
      "loss": 0.8501,
      "step": 130410
    },
    {
      "epoch": 2.2897171649783177,
      "grad_norm": 0.055609009664820186,
      "learning_rate": 0.00011744551733454454,
      "loss": 0.847,
      "step": 130420
    },
    {
      "epoch": 2.289892729858319,
      "grad_norm": 0.047779324544521905,
      "learning_rate": 0.00011743452842556772,
      "loss": 0.8469,
      "step": 130430
    },
    {
      "epoch": 2.2900682947383206,
      "grad_norm": 0.05124842457985409,
      "learning_rate": 0.0001174235393083623,
      "loss": 0.8449,
      "step": 130440
    },
    {
      "epoch": 2.290243859618322,
      "grad_norm": 0.08912957711361275,
      "learning_rate": 0.00011741254998306745,
      "loss": 0.8502,
      "step": 130450
    },
    {
      "epoch": 2.2904194244983236,
      "grad_norm": 0.07167276541809502,
      "learning_rate": 0.00011740156044982249,
      "loss": 0.8418,
      "step": 130460
    },
    {
      "epoch": 2.2905949893783246,
      "grad_norm": 0.06497985247101781,
      "learning_rate": 0.00011739057070876655,
      "loss": 0.8454,
      "step": 130470
    },
    {
      "epoch": 2.290770554258326,
      "grad_norm": 0.06648918959950612,
      "learning_rate": 0.00011737958076003899,
      "loss": 0.8486,
      "step": 130480
    },
    {
      "epoch": 2.2909461191383276,
      "grad_norm": 0.06772107189436849,
      "learning_rate": 0.00011736859060377901,
      "loss": 0.8417,
      "step": 130490
    },
    {
      "epoch": 2.291121684018329,
      "grad_norm": 0.053668656184205056,
      "learning_rate": 0.00011735760024012583,
      "loss": 0.8395,
      "step": 130500
    },
    {
      "epoch": 2.2912972488983305,
      "grad_norm": 0.07510604008406604,
      "learning_rate": 0.00011734660966921873,
      "loss": 0.8488,
      "step": 130510
    },
    {
      "epoch": 2.2914728137783316,
      "grad_norm": 0.05158789462442846,
      "learning_rate": 0.0001173356188911969,
      "loss": 0.8479,
      "step": 130520
    },
    {
      "epoch": 2.291648378658333,
      "grad_norm": 0.06470517682941958,
      "learning_rate": 0.00011732462790619974,
      "loss": 0.8477,
      "step": 130530
    },
    {
      "epoch": 2.2918239435383345,
      "grad_norm": 0.04863102973462872,
      "learning_rate": 0.00011731363671436635,
      "loss": 0.8444,
      "step": 130540
    },
    {
      "epoch": 2.291999508418336,
      "grad_norm": 0.0782203122657899,
      "learning_rate": 0.00011730264531583608,
      "loss": 0.8416,
      "step": 130550
    },
    {
      "epoch": 2.2921750732983375,
      "grad_norm": 0.05389007029814104,
      "learning_rate": 0.00011729165371074817,
      "loss": 0.8435,
      "step": 130560
    },
    {
      "epoch": 2.292350638178339,
      "grad_norm": 0.06152637476984627,
      "learning_rate": 0.00011728066189924187,
      "loss": 0.85,
      "step": 130570
    },
    {
      "epoch": 2.2925262030583404,
      "grad_norm": 0.063514584697635,
      "learning_rate": 0.00011726966988145646,
      "loss": 0.8487,
      "step": 130580
    },
    {
      "epoch": 2.2927017679383415,
      "grad_norm": 0.057610394805870366,
      "learning_rate": 0.00011725867765753123,
      "loss": 0.8394,
      "step": 130590
    },
    {
      "epoch": 2.292877332818343,
      "grad_norm": 0.04919133325890918,
      "learning_rate": 0.00011724768522760544,
      "loss": 0.8476,
      "step": 130600
    },
    {
      "epoch": 2.2930528976983444,
      "grad_norm": 0.06350863242915192,
      "learning_rate": 0.00011723669259181837,
      "loss": 0.8506,
      "step": 130610
    },
    {
      "epoch": 2.293228462578346,
      "grad_norm": 0.051005808380606664,
      "learning_rate": 0.00011722569975030928,
      "loss": 0.8395,
      "step": 130620
    },
    {
      "epoch": 2.2934040274583474,
      "grad_norm": 0.057763749670116396,
      "learning_rate": 0.00011721470670321748,
      "loss": 0.8471,
      "step": 130630
    },
    {
      "epoch": 2.2935795923383484,
      "grad_norm": 0.06863965411934685,
      "learning_rate": 0.0001172037134506822,
      "loss": 0.8404,
      "step": 130640
    },
    {
      "epoch": 2.29375515721835,
      "grad_norm": 0.07074613565051649,
      "learning_rate": 0.0001171927199928428,
      "loss": 0.8478,
      "step": 130650
    },
    {
      "epoch": 2.2939307220983514,
      "grad_norm": 0.0645219576206751,
      "learning_rate": 0.0001171817263298385,
      "loss": 0.8402,
      "step": 130660
    },
    {
      "epoch": 2.294106286978353,
      "grad_norm": 0.05282560305240269,
      "learning_rate": 0.0001171707324618087,
      "loss": 0.8455,
      "step": 130670
    },
    {
      "epoch": 2.2942818518583543,
      "grad_norm": 0.0462105874389878,
      "learning_rate": 0.00011715973838889256,
      "loss": 0.847,
      "step": 130680
    },
    {
      "epoch": 2.294457416738356,
      "grad_norm": 0.055937144814213864,
      "learning_rate": 0.00011714874411122943,
      "loss": 0.8407,
      "step": 130690
    },
    {
      "epoch": 2.2946329816183573,
      "grad_norm": 0.055192649657853685,
      "learning_rate": 0.00011713774962895863,
      "loss": 0.8448,
      "step": 130700
    },
    {
      "epoch": 2.2948085464983583,
      "grad_norm": 0.05254435924693565,
      "learning_rate": 0.00011712675494221945,
      "loss": 0.8447,
      "step": 130710
    },
    {
      "epoch": 2.29498411137836,
      "grad_norm": 0.05313263087382454,
      "learning_rate": 0.00011711576005115121,
      "loss": 0.845,
      "step": 130720
    },
    {
      "epoch": 2.2951596762583613,
      "grad_norm": 0.056783198574692655,
      "learning_rate": 0.00011710476495589322,
      "loss": 0.8399,
      "step": 130730
    },
    {
      "epoch": 2.2953352411383627,
      "grad_norm": 0.05676774145304584,
      "learning_rate": 0.00011709376965658472,
      "loss": 0.8504,
      "step": 130740
    },
    {
      "epoch": 2.295510806018364,
      "grad_norm": 0.039689195141928396,
      "learning_rate": 0.0001170827741533651,
      "loss": 0.851,
      "step": 130750
    },
    {
      "epoch": 2.2956863708983652,
      "grad_norm": 0.052842543127268445,
      "learning_rate": 0.00011707177844637362,
      "loss": 0.8512,
      "step": 130760
    },
    {
      "epoch": 2.2958619357783667,
      "grad_norm": 0.07604636442162552,
      "learning_rate": 0.00011706078253574968,
      "loss": 0.8476,
      "step": 130770
    },
    {
      "epoch": 2.296037500658368,
      "grad_norm": 0.04892689862736303,
      "learning_rate": 0.00011704978642163254,
      "loss": 0.8433,
      "step": 130780
    },
    {
      "epoch": 2.2962130655383697,
      "grad_norm": 0.08469173108161632,
      "learning_rate": 0.0001170387901041615,
      "loss": 0.8401,
      "step": 130790
    },
    {
      "epoch": 2.296388630418371,
      "grad_norm": 0.06784370279027023,
      "learning_rate": 0.00011702779358347593,
      "loss": 0.8447,
      "step": 130800
    },
    {
      "epoch": 2.2965641952983726,
      "grad_norm": 0.061389314729068514,
      "learning_rate": 0.00011701679685971511,
      "loss": 0.8479,
      "step": 130810
    },
    {
      "epoch": 2.296739760178374,
      "grad_norm": 0.05501315736259036,
      "learning_rate": 0.00011700579993301846,
      "loss": 0.8434,
      "step": 130820
    },
    {
      "epoch": 2.296915325058375,
      "grad_norm": 0.057655297129310154,
      "learning_rate": 0.00011699480280352522,
      "loss": 0.8524,
      "step": 130830
    },
    {
      "epoch": 2.2970908899383766,
      "grad_norm": 0.046803342049697265,
      "learning_rate": 0.0001169838054713748,
      "loss": 0.8528,
      "step": 130840
    },
    {
      "epoch": 2.297266454818378,
      "grad_norm": 0.048913943157529165,
      "learning_rate": 0.00011697280793670646,
      "loss": 0.8453,
      "step": 130850
    },
    {
      "epoch": 2.2974420196983796,
      "grad_norm": 0.07471874882566194,
      "learning_rate": 0.00011696181019965957,
      "loss": 0.8385,
      "step": 130860
    },
    {
      "epoch": 2.297617584578381,
      "grad_norm": 0.06041418554543387,
      "learning_rate": 0.00011695081226037354,
      "loss": 0.8364,
      "step": 130870
    },
    {
      "epoch": 2.2977931494583825,
      "grad_norm": 0.06086648925223601,
      "learning_rate": 0.00011693981411898762,
      "loss": 0.8448,
      "step": 130880
    },
    {
      "epoch": 2.2979687143383836,
      "grad_norm": 0.055335986676683185,
      "learning_rate": 0.00011692881577564123,
      "loss": 0.8389,
      "step": 130890
    },
    {
      "epoch": 2.298144279218385,
      "grad_norm": 0.10178989791892439,
      "learning_rate": 0.00011691781723047367,
      "loss": 0.8436,
      "step": 130900
    },
    {
      "epoch": 2.2983198440983865,
      "grad_norm": 0.059629490187756276,
      "learning_rate": 0.00011690681848362432,
      "loss": 0.8502,
      "step": 130910
    },
    {
      "epoch": 2.298495408978388,
      "grad_norm": 0.0651289303107902,
      "learning_rate": 0.00011689581953523255,
      "loss": 0.8429,
      "step": 130920
    },
    {
      "epoch": 2.2986709738583895,
      "grad_norm": 0.04374152360222973,
      "learning_rate": 0.00011688482038543767,
      "loss": 0.8497,
      "step": 130930
    },
    {
      "epoch": 2.298846538738391,
      "grad_norm": 0.05892319131229731,
      "learning_rate": 0.00011687382103437909,
      "loss": 0.8404,
      "step": 130940
    },
    {
      "epoch": 2.299022103618392,
      "grad_norm": 0.04978530310643424,
      "learning_rate": 0.00011686282148219616,
      "loss": 0.8408,
      "step": 130950
    },
    {
      "epoch": 2.2991976684983935,
      "grad_norm": 0.05195321263105229,
      "learning_rate": 0.00011685182172902822,
      "loss": 0.8499,
      "step": 130960
    },
    {
      "epoch": 2.299373233378395,
      "grad_norm": 0.07334988573597757,
      "learning_rate": 0.00011684082177501468,
      "loss": 0.838,
      "step": 130970
    },
    {
      "epoch": 2.2995487982583964,
      "grad_norm": 0.046647803663918706,
      "learning_rate": 0.00011682982162029487,
      "loss": 0.8409,
      "step": 130980
    },
    {
      "epoch": 2.299724363138398,
      "grad_norm": 0.08921207579025307,
      "learning_rate": 0.00011681882126500822,
      "loss": 0.8409,
      "step": 130990
    },
    {
      "epoch": 2.2998999280183994,
      "grad_norm": 0.05659291854804101,
      "learning_rate": 0.00011680782070929403,
      "loss": 0.847,
      "step": 131000
    },
    {
      "epoch": 2.300075492898401,
      "grad_norm": 0.053944679076715886,
      "learning_rate": 0.00011679681995329177,
      "loss": 0.8481,
      "step": 131010
    },
    {
      "epoch": 2.300251057778402,
      "grad_norm": 0.06009744222256656,
      "learning_rate": 0.00011678581899714074,
      "loss": 0.8434,
      "step": 131020
    },
    {
      "epoch": 2.3004266226584034,
      "grad_norm": 0.05006073568367983,
      "learning_rate": 0.00011677481784098037,
      "loss": 0.8455,
      "step": 131030
    },
    {
      "epoch": 2.300602187538405,
      "grad_norm": 0.05805742307286228,
      "learning_rate": 0.00011676381648495002,
      "loss": 0.8393,
      "step": 131040
    },
    {
      "epoch": 2.3007777524184063,
      "grad_norm": 0.09761942304386731,
      "learning_rate": 0.00011675281492918911,
      "loss": 0.8508,
      "step": 131050
    },
    {
      "epoch": 2.300953317298408,
      "grad_norm": 0.062304784438277416,
      "learning_rate": 0.00011674181317383704,
      "loss": 0.8448,
      "step": 131060
    },
    {
      "epoch": 2.301128882178409,
      "grad_norm": 0.04950995348021109,
      "learning_rate": 0.00011673081121903315,
      "loss": 0.8487,
      "step": 131070
    },
    {
      "epoch": 2.3013044470584103,
      "grad_norm": 0.06856512245848703,
      "learning_rate": 0.00011671980906491685,
      "loss": 0.8375,
      "step": 131080
    },
    {
      "epoch": 2.301480011938412,
      "grad_norm": 0.06311829583637261,
      "learning_rate": 0.00011670880671162759,
      "loss": 0.8464,
      "step": 131090
    },
    {
      "epoch": 2.3016555768184133,
      "grad_norm": 0.07822104045843159,
      "learning_rate": 0.00011669780415930471,
      "loss": 0.8441,
      "step": 131100
    },
    {
      "epoch": 2.3018311416984147,
      "grad_norm": 0.08341359681840273,
      "learning_rate": 0.00011668680140808765,
      "loss": 0.8473,
      "step": 131110
    },
    {
      "epoch": 2.302006706578416,
      "grad_norm": 0.061679289119548175,
      "learning_rate": 0.00011667579845811584,
      "loss": 0.8503,
      "step": 131120
    },
    {
      "epoch": 2.3021822714584177,
      "grad_norm": 0.05040790863303951,
      "learning_rate": 0.00011666479530952864,
      "loss": 0.8466,
      "step": 131130
    },
    {
      "epoch": 2.3023578363384187,
      "grad_norm": 0.05240311946518498,
      "learning_rate": 0.00011665379196246547,
      "loss": 0.8445,
      "step": 131140
    },
    {
      "epoch": 2.30253340121842,
      "grad_norm": 0.05030028796647428,
      "learning_rate": 0.00011664278841706575,
      "loss": 0.841,
      "step": 131150
    },
    {
      "epoch": 2.3027089660984217,
      "grad_norm": 0.06266554834575773,
      "learning_rate": 0.00011663178467346893,
      "loss": 0.8341,
      "step": 131160
    },
    {
      "epoch": 2.302884530978423,
      "grad_norm": 0.051124954481729795,
      "learning_rate": 0.00011662078073181437,
      "loss": 0.8532,
      "step": 131170
    },
    {
      "epoch": 2.3030600958584246,
      "grad_norm": 0.0634984620013231,
      "learning_rate": 0.00011660977659224157,
      "loss": 0.8443,
      "step": 131180
    },
    {
      "epoch": 2.3032356607384257,
      "grad_norm": 0.047587590116746965,
      "learning_rate": 0.00011659877225488989,
      "loss": 0.8479,
      "step": 131190
    },
    {
      "epoch": 2.303411225618427,
      "grad_norm": 0.04628180314081065,
      "learning_rate": 0.00011658776771989875,
      "loss": 0.84,
      "step": 131200
    },
    {
      "epoch": 2.3035867904984286,
      "grad_norm": 0.07793241271268357,
      "learning_rate": 0.00011657676298740763,
      "loss": 0.8497,
      "step": 131210
    },
    {
      "epoch": 2.30376235537843,
      "grad_norm": 0.05174461494623339,
      "learning_rate": 0.00011656575805755592,
      "loss": 0.8443,
      "step": 131220
    },
    {
      "epoch": 2.3039379202584316,
      "grad_norm": 0.053102632686541355,
      "learning_rate": 0.00011655475293048308,
      "loss": 0.8411,
      "step": 131230
    },
    {
      "epoch": 2.304113485138433,
      "grad_norm": 0.07300162656693761,
      "learning_rate": 0.00011654374760632855,
      "loss": 0.8434,
      "step": 131240
    },
    {
      "epoch": 2.3042890500184345,
      "grad_norm": 0.04976763921912537,
      "learning_rate": 0.00011653274208523175,
      "loss": 0.8446,
      "step": 131250
    },
    {
      "epoch": 2.3044646148984356,
      "grad_norm": 0.09929392818056601,
      "learning_rate": 0.00011652173636733212,
      "loss": 0.8391,
      "step": 131260
    },
    {
      "epoch": 2.304640179778437,
      "grad_norm": 0.05117932122731828,
      "learning_rate": 0.00011651073045276915,
      "loss": 0.8416,
      "step": 131270
    },
    {
      "epoch": 2.3048157446584385,
      "grad_norm": 0.05247397611054597,
      "learning_rate": 0.00011649972434168221,
      "loss": 0.8433,
      "step": 131280
    },
    {
      "epoch": 2.30499130953844,
      "grad_norm": 0.044199033580002114,
      "learning_rate": 0.00011648871803421084,
      "loss": 0.8513,
      "step": 131290
    },
    {
      "epoch": 2.3051668744184415,
      "grad_norm": 0.05644080909086009,
      "learning_rate": 0.00011647771153049443,
      "loss": 0.8456,
      "step": 131300
    },
    {
      "epoch": 2.3053424392984425,
      "grad_norm": 0.050575623546793534,
      "learning_rate": 0.00011646670483067242,
      "loss": 0.8471,
      "step": 131310
    },
    {
      "epoch": 2.305518004178444,
      "grad_norm": 0.05233496027446299,
      "learning_rate": 0.0001164556979348843,
      "loss": 0.8534,
      "step": 131320
    },
    {
      "epoch": 2.3056935690584455,
      "grad_norm": 0.05201036736760837,
      "learning_rate": 0.00011644469084326954,
      "loss": 0.8465,
      "step": 131330
    },
    {
      "epoch": 2.305869133938447,
      "grad_norm": 0.08620030261254873,
      "learning_rate": 0.00011643368355596758,
      "loss": 0.8424,
      "step": 131340
    },
    {
      "epoch": 2.3060446988184484,
      "grad_norm": 0.05145760583863141,
      "learning_rate": 0.00011642267607311792,
      "loss": 0.8439,
      "step": 131350
    },
    {
      "epoch": 2.30622026369845,
      "grad_norm": 0.05354654935663564,
      "learning_rate": 0.00011641166839485995,
      "loss": 0.8361,
      "step": 131360
    },
    {
      "epoch": 2.3063958285784514,
      "grad_norm": 0.048402006943001376,
      "learning_rate": 0.00011640066052133322,
      "loss": 0.8407,
      "step": 131370
    },
    {
      "epoch": 2.3065713934584524,
      "grad_norm": 0.07698078990030632,
      "learning_rate": 0.00011638965245267714,
      "loss": 0.8441,
      "step": 131380
    },
    {
      "epoch": 2.306746958338454,
      "grad_norm": 0.054625548257047345,
      "learning_rate": 0.00011637864418903124,
      "loss": 0.8428,
      "step": 131390
    },
    {
      "epoch": 2.3069225232184554,
      "grad_norm": 0.05450211837190775,
      "learning_rate": 0.00011636763573053498,
      "loss": 0.8414,
      "step": 131400
    },
    {
      "epoch": 2.307098088098457,
      "grad_norm": 0.05890585086277362,
      "learning_rate": 0.00011635662707732782,
      "loss": 0.8403,
      "step": 131410
    },
    {
      "epoch": 2.3072736529784583,
      "grad_norm": 0.04967149216693062,
      "learning_rate": 0.00011634561822954923,
      "loss": 0.8413,
      "step": 131420
    },
    {
      "epoch": 2.3074492178584594,
      "grad_norm": 0.05275346838447762,
      "learning_rate": 0.00011633460918733874,
      "loss": 0.8539,
      "step": 131430
    },
    {
      "epoch": 2.307624782738461,
      "grad_norm": 0.04216600427002104,
      "learning_rate": 0.00011632359995083581,
      "loss": 0.851,
      "step": 131440
    },
    {
      "epoch": 2.3078003476184623,
      "grad_norm": 0.06382032780914622,
      "learning_rate": 0.00011631259052017994,
      "loss": 0.8413,
      "step": 131450
    },
    {
      "epoch": 2.307975912498464,
      "grad_norm": 0.06372946525137596,
      "learning_rate": 0.0001163015808955106,
      "loss": 0.8487,
      "step": 131460
    },
    {
      "epoch": 2.3081514773784653,
      "grad_norm": 0.041429792205550715,
      "learning_rate": 0.00011629057107696728,
      "loss": 0.8373,
      "step": 131470
    },
    {
      "epoch": 2.3083270422584667,
      "grad_norm": 0.053559895566608194,
      "learning_rate": 0.00011627956106468953,
      "loss": 0.8461,
      "step": 131480
    },
    {
      "epoch": 2.308502607138468,
      "grad_norm": 0.05381463775422348,
      "learning_rate": 0.00011626855085881678,
      "loss": 0.8502,
      "step": 131490
    },
    {
      "epoch": 2.3086781720184693,
      "grad_norm": 0.06958233597383069,
      "learning_rate": 0.0001162575404594886,
      "loss": 0.8497,
      "step": 131500
    },
    {
      "epoch": 2.3088537368984707,
      "grad_norm": 0.06608289626758987,
      "learning_rate": 0.00011624652986684444,
      "loss": 0.8457,
      "step": 131510
    },
    {
      "epoch": 2.309029301778472,
      "grad_norm": 0.048163294898753546,
      "learning_rate": 0.00011623551908102384,
      "loss": 0.8439,
      "step": 131520
    },
    {
      "epoch": 2.3092048666584737,
      "grad_norm": 0.060141934038349915,
      "learning_rate": 0.00011622450810216625,
      "loss": 0.8416,
      "step": 131530
    },
    {
      "epoch": 2.309380431538475,
      "grad_norm": 0.05694419463382744,
      "learning_rate": 0.00011621349693041124,
      "loss": 0.8445,
      "step": 131540
    },
    {
      "epoch": 2.3095559964184766,
      "grad_norm": 0.05867368916359734,
      "learning_rate": 0.00011620248556589834,
      "loss": 0.847,
      "step": 131550
    },
    {
      "epoch": 2.3097315612984777,
      "grad_norm": 0.05300386029993695,
      "learning_rate": 0.000116191474008767,
      "loss": 0.8453,
      "step": 131560
    },
    {
      "epoch": 2.309907126178479,
      "grad_norm": 0.05899056874402319,
      "learning_rate": 0.00011618046225915679,
      "loss": 0.8508,
      "step": 131570
    },
    {
      "epoch": 2.3100826910584806,
      "grad_norm": 0.052619804211176256,
      "learning_rate": 0.00011616945031720723,
      "loss": 0.8437,
      "step": 131580
    },
    {
      "epoch": 2.310258255938482,
      "grad_norm": 0.05212674290772514,
      "learning_rate": 0.00011615843818305779,
      "loss": 0.843,
      "step": 131590
    },
    {
      "epoch": 2.3104338208184836,
      "grad_norm": 0.07493472543276285,
      "learning_rate": 0.00011614742585684807,
      "loss": 0.8409,
      "step": 131600
    },
    {
      "epoch": 2.310609385698485,
      "grad_norm": 0.0432054420620501,
      "learning_rate": 0.00011613641333871752,
      "loss": 0.8505,
      "step": 131610
    },
    {
      "epoch": 2.310784950578486,
      "grad_norm": 0.05055575162189773,
      "learning_rate": 0.00011612540062880574,
      "loss": 0.8436,
      "step": 131620
    },
    {
      "epoch": 2.3109605154584876,
      "grad_norm": 0.04942214384567626,
      "learning_rate": 0.00011611438772725225,
      "loss": 0.859,
      "step": 131630
    },
    {
      "epoch": 2.311136080338489,
      "grad_norm": 0.06055115115229979,
      "learning_rate": 0.00011610337463419653,
      "loss": 0.8484,
      "step": 131640
    },
    {
      "epoch": 2.3113116452184905,
      "grad_norm": 0.056671205739245924,
      "learning_rate": 0.00011609236134977818,
      "loss": 0.844,
      "step": 131650
    },
    {
      "epoch": 2.311487210098492,
      "grad_norm": 0.04976004681135234,
      "learning_rate": 0.00011608134787413671,
      "loss": 0.8338,
      "step": 131660
    },
    {
      "epoch": 2.3116627749784935,
      "grad_norm": 0.05068612885324807,
      "learning_rate": 0.00011607033420741171,
      "loss": 0.8488,
      "step": 131670
    },
    {
      "epoch": 2.3118383398584945,
      "grad_norm": 0.05696221287897958,
      "learning_rate": 0.00011605932034974264,
      "loss": 0.8509,
      "step": 131680
    },
    {
      "epoch": 2.312013904738496,
      "grad_norm": 0.06330975878779427,
      "learning_rate": 0.00011604830630126912,
      "loss": 0.8508,
      "step": 131690
    },
    {
      "epoch": 2.3121894696184975,
      "grad_norm": 0.0636138533330906,
      "learning_rate": 0.00011603729206213068,
      "loss": 0.8532,
      "step": 131700
    },
    {
      "epoch": 2.312365034498499,
      "grad_norm": 0.05393596369221491,
      "learning_rate": 0.00011602627763246682,
      "loss": 0.8457,
      "step": 131710
    },
    {
      "epoch": 2.3125405993785004,
      "grad_norm": 0.06471600158981274,
      "learning_rate": 0.0001160152630124172,
      "loss": 0.8572,
      "step": 131720
    },
    {
      "epoch": 2.312716164258502,
      "grad_norm": 0.05520742819875494,
      "learning_rate": 0.0001160042482021213,
      "loss": 0.8468,
      "step": 131730
    },
    {
      "epoch": 2.312891729138503,
      "grad_norm": 0.06825510427903188,
      "learning_rate": 0.0001159932332017187,
      "loss": 0.8451,
      "step": 131740
    },
    {
      "epoch": 2.3130672940185044,
      "grad_norm": 0.0491157910626316,
      "learning_rate": 0.00011598221801134897,
      "loss": 0.8412,
      "step": 131750
    },
    {
      "epoch": 2.313242858898506,
      "grad_norm": 0.051565602617125096,
      "learning_rate": 0.00011597120263115165,
      "loss": 0.8513,
      "step": 131760
    },
    {
      "epoch": 2.3134184237785074,
      "grad_norm": 0.07641030932102126,
      "learning_rate": 0.00011596018706126635,
      "loss": 0.845,
      "step": 131770
    },
    {
      "epoch": 2.313593988658509,
      "grad_norm": 0.08329072010095231,
      "learning_rate": 0.00011594917130183257,
      "loss": 0.8422,
      "step": 131780
    },
    {
      "epoch": 2.3137695535385103,
      "grad_norm": 0.0678388551972083,
      "learning_rate": 0.00011593815535298995,
      "loss": 0.8426,
      "step": 131790
    },
    {
      "epoch": 2.313945118418512,
      "grad_norm": 0.05041484969839633,
      "learning_rate": 0.00011592713921487806,
      "loss": 0.8527,
      "step": 131800
    },
    {
      "epoch": 2.314120683298513,
      "grad_norm": 0.06892421193249482,
      "learning_rate": 0.0001159161228876364,
      "loss": 0.8453,
      "step": 131810
    },
    {
      "epoch": 2.3142962481785143,
      "grad_norm": 0.04785766535020726,
      "learning_rate": 0.00011590510637140464,
      "loss": 0.8441,
      "step": 131820
    },
    {
      "epoch": 2.314471813058516,
      "grad_norm": 0.06556700112635146,
      "learning_rate": 0.00011589408966632234,
      "loss": 0.84,
      "step": 131830
    },
    {
      "epoch": 2.3146473779385173,
      "grad_norm": 0.06004423996784591,
      "learning_rate": 0.00011588307277252905,
      "loss": 0.8446,
      "step": 131840
    },
    {
      "epoch": 2.3148229428185187,
      "grad_norm": 0.04947966987789711,
      "learning_rate": 0.00011587205569016434,
      "loss": 0.8456,
      "step": 131850
    },
    {
      "epoch": 2.31499850769852,
      "grad_norm": 0.06713408611553574,
      "learning_rate": 0.00011586103841936793,
      "loss": 0.839,
      "step": 131860
    },
    {
      "epoch": 2.3151740725785213,
      "grad_norm": 0.06586212257832785,
      "learning_rate": 0.00011585002096027923,
      "loss": 0.8492,
      "step": 131870
    },
    {
      "epoch": 2.3153496374585227,
      "grad_norm": 0.06067289947614677,
      "learning_rate": 0.00011583900331303794,
      "loss": 0.8376,
      "step": 131880
    },
    {
      "epoch": 2.315525202338524,
      "grad_norm": 0.06901772268142187,
      "learning_rate": 0.00011582798547778366,
      "loss": 0.8415,
      "step": 131890
    },
    {
      "epoch": 2.3157007672185257,
      "grad_norm": 0.06982702721959523,
      "learning_rate": 0.00011581696745465593,
      "loss": 0.8341,
      "step": 131900
    },
    {
      "epoch": 2.315876332098527,
      "grad_norm": 0.071141246301235,
      "learning_rate": 0.0001158059492437944,
      "loss": 0.8442,
      "step": 131910
    },
    {
      "epoch": 2.3160518969785286,
      "grad_norm": 0.0628158616138464,
      "learning_rate": 0.00011579493084533866,
      "loss": 0.8364,
      "step": 131920
    },
    {
      "epoch": 2.3162274618585297,
      "grad_norm": 0.05774178562023637,
      "learning_rate": 0.00011578391225942826,
      "loss": 0.8404,
      "step": 131930
    },
    {
      "epoch": 2.316403026738531,
      "grad_norm": 0.05739360771457989,
      "learning_rate": 0.00011577289348620293,
      "loss": 0.8512,
      "step": 131940
    },
    {
      "epoch": 2.3165785916185326,
      "grad_norm": 0.0658199956845853,
      "learning_rate": 0.00011576187452580219,
      "loss": 0.8434,
      "step": 131950
    },
    {
      "epoch": 2.316754156498534,
      "grad_norm": 0.06106743794143059,
      "learning_rate": 0.00011575085537836565,
      "loss": 0.8417,
      "step": 131960
    },
    {
      "epoch": 2.3169297213785356,
      "grad_norm": 0.06365095463554989,
      "learning_rate": 0.00011573983604403295,
      "loss": 0.8512,
      "step": 131970
    },
    {
      "epoch": 2.3171052862585366,
      "grad_norm": 0.05166128641627272,
      "learning_rate": 0.0001157288165229437,
      "loss": 0.8423,
      "step": 131980
    },
    {
      "epoch": 2.317280851138538,
      "grad_norm": 0.07302910901052091,
      "learning_rate": 0.00011571779681523753,
      "loss": 0.844,
      "step": 131990
    },
    {
      "epoch": 2.3174564160185396,
      "grad_norm": 0.07233245301816965,
      "learning_rate": 0.00011570677692105405,
      "loss": 0.8497,
      "step": 132000
    },
    {
      "epoch": 2.317631980898541,
      "grad_norm": 0.06819798405951592,
      "learning_rate": 0.00011569575684053288,
      "loss": 0.8467,
      "step": 132010
    },
    {
      "epoch": 2.3178075457785425,
      "grad_norm": 0.05834454112901788,
      "learning_rate": 0.00011568473657381367,
      "loss": 0.8484,
      "step": 132020
    },
    {
      "epoch": 2.317983110658544,
      "grad_norm": 0.0431875702218534,
      "learning_rate": 0.00011567371612103603,
      "loss": 0.8414,
      "step": 132030
    },
    {
      "epoch": 2.3181586755385455,
      "grad_norm": 0.05557181398255899,
      "learning_rate": 0.00011566269548233958,
      "loss": 0.8386,
      "step": 132040
    },
    {
      "epoch": 2.3183342404185465,
      "grad_norm": 0.05272378529331595,
      "learning_rate": 0.00011565167465786399,
      "loss": 0.8471,
      "step": 132050
    },
    {
      "epoch": 2.318509805298548,
      "grad_norm": 0.038610718046955685,
      "learning_rate": 0.00011564065364774885,
      "loss": 0.8509,
      "step": 132060
    },
    {
      "epoch": 2.3186853701785495,
      "grad_norm": 0.05050281178718943,
      "learning_rate": 0.00011562963245213382,
      "loss": 0.8424,
      "step": 132070
    },
    {
      "epoch": 2.318860935058551,
      "grad_norm": 0.05493230583810596,
      "learning_rate": 0.00011561861107115859,
      "loss": 0.8446,
      "step": 132080
    },
    {
      "epoch": 2.3190364999385524,
      "grad_norm": 0.055919919139637696,
      "learning_rate": 0.00011560758950496271,
      "loss": 0.8446,
      "step": 132090
    },
    {
      "epoch": 2.3192120648185535,
      "grad_norm": 0.058064895337602826,
      "learning_rate": 0.00011559656775368588,
      "loss": 0.8488,
      "step": 132100
    },
    {
      "epoch": 2.319387629698555,
      "grad_norm": 0.06463800397202103,
      "learning_rate": 0.00011558554581746776,
      "loss": 0.8412,
      "step": 132110
    },
    {
      "epoch": 2.3195631945785564,
      "grad_norm": 0.06499608428821546,
      "learning_rate": 0.00011557452369644794,
      "loss": 0.836,
      "step": 132120
    },
    {
      "epoch": 2.319738759458558,
      "grad_norm": 0.05027817301085774,
      "learning_rate": 0.00011556350139076618,
      "loss": 0.8434,
      "step": 132130
    },
    {
      "epoch": 2.3199143243385594,
      "grad_norm": 0.06827940621544407,
      "learning_rate": 0.000115552478900562,
      "loss": 0.8418,
      "step": 132140
    },
    {
      "epoch": 2.320089889218561,
      "grad_norm": 0.04965600026573296,
      "learning_rate": 0.00011554145622597517,
      "loss": 0.834,
      "step": 132150
    },
    {
      "epoch": 2.3202654540985623,
      "grad_norm": 0.04652094177654587,
      "learning_rate": 0.00011553043336714528,
      "loss": 0.8444,
      "step": 132160
    },
    {
      "epoch": 2.3204410189785634,
      "grad_norm": 0.07212004975976503,
      "learning_rate": 0.00011551941032421199,
      "loss": 0.8437,
      "step": 132170
    },
    {
      "epoch": 2.320616583858565,
      "grad_norm": 0.052988721666986,
      "learning_rate": 0.00011550838709731502,
      "loss": 0.8416,
      "step": 132180
    },
    {
      "epoch": 2.3207921487385663,
      "grad_norm": 0.05455026552565,
      "learning_rate": 0.00011549736368659398,
      "loss": 0.851,
      "step": 132190
    },
    {
      "epoch": 2.320967713618568,
      "grad_norm": 0.04906657987831084,
      "learning_rate": 0.00011548634009218862,
      "loss": 0.8459,
      "step": 132200
    },
    {
      "epoch": 2.3211432784985693,
      "grad_norm": 0.05513239000971743,
      "learning_rate": 0.00011547531631423852,
      "loss": 0.8534,
      "step": 132210
    },
    {
      "epoch": 2.3213188433785703,
      "grad_norm": 0.06401149443933428,
      "learning_rate": 0.00011546429235288333,
      "loss": 0.842,
      "step": 132220
    },
    {
      "epoch": 2.321494408258572,
      "grad_norm": 0.06840071081576128,
      "learning_rate": 0.00011545326820826286,
      "loss": 0.8389,
      "step": 132230
    },
    {
      "epoch": 2.3216699731385733,
      "grad_norm": 0.05456158680588138,
      "learning_rate": 0.00011544224388051668,
      "loss": 0.846,
      "step": 132240
    },
    {
      "epoch": 2.3218455380185747,
      "grad_norm": 0.055083135178495356,
      "learning_rate": 0.00011543121936978452,
      "loss": 0.8438,
      "step": 132250
    },
    {
      "epoch": 2.322021102898576,
      "grad_norm": 0.04100271933025384,
      "learning_rate": 0.00011542019467620607,
      "loss": 0.8539,
      "step": 132260
    },
    {
      "epoch": 2.3221966677785777,
      "grad_norm": 0.07141339188047006,
      "learning_rate": 0.00011540916979992093,
      "loss": 0.8411,
      "step": 132270
    },
    {
      "epoch": 2.322372232658579,
      "grad_norm": 0.06038486541640014,
      "learning_rate": 0.0001153981447410689,
      "loss": 0.8459,
      "step": 132280
    },
    {
      "epoch": 2.32254779753858,
      "grad_norm": 0.049495341874494676,
      "learning_rate": 0.00011538711949978955,
      "loss": 0.8464,
      "step": 132290
    },
    {
      "epoch": 2.3227233624185817,
      "grad_norm": 0.0408329699430231,
      "learning_rate": 0.0001153760940762227,
      "loss": 0.8536,
      "step": 132300
    },
    {
      "epoch": 2.322898927298583,
      "grad_norm": 0.059803981773137026,
      "learning_rate": 0.00011536506847050797,
      "loss": 0.8438,
      "step": 132310
    },
    {
      "epoch": 2.3230744921785846,
      "grad_norm": 0.04543362482275578,
      "learning_rate": 0.00011535404268278508,
      "loss": 0.8433,
      "step": 132320
    },
    {
      "epoch": 2.323250057058586,
      "grad_norm": 0.04925225294921224,
      "learning_rate": 0.00011534301671319368,
      "loss": 0.8456,
      "step": 132330
    },
    {
      "epoch": 2.3234256219385876,
      "grad_norm": 0.06175679926016005,
      "learning_rate": 0.00011533199056187349,
      "loss": 0.8452,
      "step": 132340
    },
    {
      "epoch": 2.3236011868185886,
      "grad_norm": 0.056562169796854064,
      "learning_rate": 0.0001153209642289643,
      "loss": 0.852,
      "step": 132350
    },
    {
      "epoch": 2.32377675169859,
      "grad_norm": 0.061940195436966466,
      "learning_rate": 0.0001153099377146057,
      "loss": 0.8366,
      "step": 132360
    },
    {
      "epoch": 2.3239523165785916,
      "grad_norm": 0.04516669425689456,
      "learning_rate": 0.0001152989110189375,
      "loss": 0.8436,
      "step": 132370
    },
    {
      "epoch": 2.324127881458593,
      "grad_norm": 0.05769578225247036,
      "learning_rate": 0.00011528788414209932,
      "loss": 0.8474,
      "step": 132380
    },
    {
      "epoch": 2.3243034463385945,
      "grad_norm": 0.0453077633946414,
      "learning_rate": 0.00011527685708423088,
      "loss": 0.8474,
      "step": 132390
    },
    {
      "epoch": 2.324479011218596,
      "grad_norm": 0.0547638203209841,
      "learning_rate": 0.00011526582984547199,
      "loss": 0.8542,
      "step": 132400
    },
    {
      "epoch": 2.324654576098597,
      "grad_norm": 0.047717822971689285,
      "learning_rate": 0.00011525480242596226,
      "loss": 0.8436,
      "step": 132410
    },
    {
      "epoch": 2.3248301409785985,
      "grad_norm": 0.055083936102855176,
      "learning_rate": 0.00011524377482584146,
      "loss": 0.8548,
      "step": 132420
    },
    {
      "epoch": 2.3250057058586,
      "grad_norm": 0.04923469444673471,
      "learning_rate": 0.00011523274704524931,
      "loss": 0.8387,
      "step": 132430
    },
    {
      "epoch": 2.3251812707386015,
      "grad_norm": 0.052030749926143975,
      "learning_rate": 0.0001152217190843255,
      "loss": 0.8412,
      "step": 132440
    },
    {
      "epoch": 2.325356835618603,
      "grad_norm": 0.052971714579689626,
      "learning_rate": 0.0001152106909432098,
      "loss": 0.8418,
      "step": 132450
    },
    {
      "epoch": 2.3255324004986044,
      "grad_norm": 0.05968021151820387,
      "learning_rate": 0.00011519966262204193,
      "loss": 0.8473,
      "step": 132460
    },
    {
      "epoch": 2.325707965378606,
      "grad_norm": 0.06798298109683613,
      "learning_rate": 0.00011518863412096164,
      "loss": 0.8477,
      "step": 132470
    },
    {
      "epoch": 2.325883530258607,
      "grad_norm": 0.05599923491006632,
      "learning_rate": 0.00011517760544010861,
      "loss": 0.8489,
      "step": 132480
    },
    {
      "epoch": 2.3260590951386084,
      "grad_norm": 0.08324341160803578,
      "learning_rate": 0.0001151665765796226,
      "loss": 0.8362,
      "step": 132490
    },
    {
      "epoch": 2.32623466001861,
      "grad_norm": 0.0562488785905125,
      "learning_rate": 0.00011515554753964332,
      "loss": 0.8452,
      "step": 132500
    },
    {
      "epoch": 2.3264102248986114,
      "grad_norm": 0.04966065608802306,
      "learning_rate": 0.00011514451832031056,
      "loss": 0.8376,
      "step": 132510
    },
    {
      "epoch": 2.326585789778613,
      "grad_norm": 0.04804761078497362,
      "learning_rate": 0.00011513348892176407,
      "loss": 0.8461,
      "step": 132520
    },
    {
      "epoch": 2.326761354658614,
      "grad_norm": 0.05230637811647526,
      "learning_rate": 0.00011512245934414355,
      "loss": 0.8438,
      "step": 132530
    },
    {
      "epoch": 2.3269369195386154,
      "grad_norm": 0.06412457844151244,
      "learning_rate": 0.00011511142958758876,
      "loss": 0.8448,
      "step": 132540
    },
    {
      "epoch": 2.327112484418617,
      "grad_norm": 0.047413347292195596,
      "learning_rate": 0.00011510039965223946,
      "loss": 0.8502,
      "step": 132550
    },
    {
      "epoch": 2.3272880492986183,
      "grad_norm": 0.06002771062709836,
      "learning_rate": 0.00011508936953823538,
      "loss": 0.8495,
      "step": 132560
    },
    {
      "epoch": 2.32746361417862,
      "grad_norm": 0.049832368496222884,
      "learning_rate": 0.00011507833924571629,
      "loss": 0.8428,
      "step": 132570
    },
    {
      "epoch": 2.3276391790586213,
      "grad_norm": 0.05004195474144791,
      "learning_rate": 0.00011506730877482194,
      "loss": 0.8509,
      "step": 132580
    },
    {
      "epoch": 2.3278147439386228,
      "grad_norm": 0.05154846514265575,
      "learning_rate": 0.0001150562781256921,
      "loss": 0.8476,
      "step": 132590
    },
    {
      "epoch": 2.327990308818624,
      "grad_norm": 0.0619546505158028,
      "learning_rate": 0.00011504524729846656,
      "loss": 0.8508,
      "step": 132600
    },
    {
      "epoch": 2.3281658736986253,
      "grad_norm": 0.08696245302103729,
      "learning_rate": 0.000115034216293285,
      "loss": 0.8371,
      "step": 132610
    },
    {
      "epoch": 2.3283414385786267,
      "grad_norm": 0.06489835268600537,
      "learning_rate": 0.00011502318511028722,
      "loss": 0.846,
      "step": 132620
    },
    {
      "epoch": 2.328517003458628,
      "grad_norm": 0.05941380407014737,
      "learning_rate": 0.000115012153749613,
      "loss": 0.8468,
      "step": 132630
    },
    {
      "epoch": 2.3286925683386297,
      "grad_norm": 0.06344944910140864,
      "learning_rate": 0.00011500112221140214,
      "loss": 0.8426,
      "step": 132640
    },
    {
      "epoch": 2.3288681332186307,
      "grad_norm": 0.06585934896664666,
      "learning_rate": 0.00011499009049579437,
      "loss": 0.8435,
      "step": 132650
    },
    {
      "epoch": 2.329043698098632,
      "grad_norm": 0.0456630681738946,
      "learning_rate": 0.00011497905860292947,
      "loss": 0.842,
      "step": 132660
    },
    {
      "epoch": 2.3292192629786337,
      "grad_norm": 0.04903786987446931,
      "learning_rate": 0.0001149680265329472,
      "loss": 0.8457,
      "step": 132670
    },
    {
      "epoch": 2.329394827858635,
      "grad_norm": 0.049532543840952216,
      "learning_rate": 0.00011495699428598737,
      "loss": 0.8417,
      "step": 132680
    },
    {
      "epoch": 2.3295703927386366,
      "grad_norm": 0.06772521262005636,
      "learning_rate": 0.00011494596186218973,
      "loss": 0.8436,
      "step": 132690
    },
    {
      "epoch": 2.329745957618638,
      "grad_norm": 0.05811248080223062,
      "learning_rate": 0.00011493492926169409,
      "loss": 0.8424,
      "step": 132700
    },
    {
      "epoch": 2.3299215224986396,
      "grad_norm": 0.04837732320883386,
      "learning_rate": 0.00011492389648464026,
      "loss": 0.8421,
      "step": 132710
    },
    {
      "epoch": 2.3300970873786406,
      "grad_norm": 0.05100891972778473,
      "learning_rate": 0.00011491286353116795,
      "loss": 0.8495,
      "step": 132720
    },
    {
      "epoch": 2.330272652258642,
      "grad_norm": 0.0683264885333336,
      "learning_rate": 0.000114901830401417,
      "loss": 0.8477,
      "step": 132730
    },
    {
      "epoch": 2.3304482171386436,
      "grad_norm": 0.04729073252840122,
      "learning_rate": 0.00011489079709552722,
      "loss": 0.8541,
      "step": 132740
    },
    {
      "epoch": 2.330623782018645,
      "grad_norm": 0.08101548497065235,
      "learning_rate": 0.00011487976361363834,
      "loss": 0.8458,
      "step": 132750
    },
    {
      "epoch": 2.3307993468986465,
      "grad_norm": 0.047928862878028944,
      "learning_rate": 0.00011486872995589022,
      "loss": 0.8483,
      "step": 132760
    },
    {
      "epoch": 2.3309749117786476,
      "grad_norm": 0.04819535322127413,
      "learning_rate": 0.00011485769612242263,
      "loss": 0.8469,
      "step": 132770
    },
    {
      "epoch": 2.331150476658649,
      "grad_norm": 0.056666192873380754,
      "learning_rate": 0.00011484666211337536,
      "loss": 0.8364,
      "step": 132780
    },
    {
      "epoch": 2.3313260415386505,
      "grad_norm": 0.061740148220989235,
      "learning_rate": 0.00011483562792888827,
      "loss": 0.8395,
      "step": 132790
    },
    {
      "epoch": 2.331501606418652,
      "grad_norm": 0.04939167455822436,
      "learning_rate": 0.00011482459356910107,
      "loss": 0.8446,
      "step": 132800
    },
    {
      "epoch": 2.3316771712986535,
      "grad_norm": 0.0773360337866345,
      "learning_rate": 0.00011481355903415364,
      "loss": 0.8422,
      "step": 132810
    },
    {
      "epoch": 2.331852736178655,
      "grad_norm": 0.08080058169879259,
      "learning_rate": 0.00011480252432418578,
      "loss": 0.8412,
      "step": 132820
    },
    {
      "epoch": 2.3320283010586564,
      "grad_norm": 0.05344320363584502,
      "learning_rate": 0.0001147914894393373,
      "loss": 0.8422,
      "step": 132830
    },
    {
      "epoch": 2.3322038659386575,
      "grad_norm": 0.06337534477077497,
      "learning_rate": 0.00011478045437974797,
      "loss": 0.8448,
      "step": 132840
    },
    {
      "epoch": 2.332379430818659,
      "grad_norm": 0.06409298560970393,
      "learning_rate": 0.00011476941914555766,
      "loss": 0.8438,
      "step": 132850
    },
    {
      "epoch": 2.3325549956986604,
      "grad_norm": 0.08525732620868846,
      "learning_rate": 0.00011475838373690617,
      "loss": 0.8426,
      "step": 132860
    },
    {
      "epoch": 2.332730560578662,
      "grad_norm": 0.07036262246586462,
      "learning_rate": 0.00011474734815393333,
      "loss": 0.8381,
      "step": 132870
    },
    {
      "epoch": 2.3329061254586634,
      "grad_norm": 0.05940171872565286,
      "learning_rate": 0.00011473631239677897,
      "loss": 0.8372,
      "step": 132880
    },
    {
      "epoch": 2.3330816903386644,
      "grad_norm": 0.050642922778728366,
      "learning_rate": 0.00011472527646558286,
      "loss": 0.8449,
      "step": 132890
    },
    {
      "epoch": 2.333257255218666,
      "grad_norm": 0.045409123148277136,
      "learning_rate": 0.00011471424036048488,
      "loss": 0.8445,
      "step": 132900
    },
    {
      "epoch": 2.3334328200986674,
      "grad_norm": 0.07070525988953362,
      "learning_rate": 0.00011470320408162487,
      "loss": 0.841,
      "step": 132910
    },
    {
      "epoch": 2.333608384978669,
      "grad_norm": 0.05464269809761765,
      "learning_rate": 0.00011469216762914259,
      "loss": 0.8438,
      "step": 132920
    },
    {
      "epoch": 2.3337839498586703,
      "grad_norm": 0.0528767920599999,
      "learning_rate": 0.00011468113100317796,
      "loss": 0.8404,
      "step": 132930
    },
    {
      "epoch": 2.333959514738672,
      "grad_norm": 0.07308962953126398,
      "learning_rate": 0.00011467009420387078,
      "loss": 0.8511,
      "step": 132940
    },
    {
      "epoch": 2.3341350796186733,
      "grad_norm": 0.057043019355783625,
      "learning_rate": 0.00011465905723136086,
      "loss": 0.8447,
      "step": 132950
    },
    {
      "epoch": 2.3343106444986743,
      "grad_norm": 0.04589732737653315,
      "learning_rate": 0.0001146480200857881,
      "loss": 0.8385,
      "step": 132960
    },
    {
      "epoch": 2.334486209378676,
      "grad_norm": 0.05946600518766739,
      "learning_rate": 0.00011463698276729226,
      "loss": 0.8477,
      "step": 132970
    },
    {
      "epoch": 2.3346617742586773,
      "grad_norm": 0.06391689425084605,
      "learning_rate": 0.00011462594527601327,
      "loss": 0.8451,
      "step": 132980
    },
    {
      "epoch": 2.3348373391386787,
      "grad_norm": 0.0663536122153661,
      "learning_rate": 0.00011461490761209098,
      "loss": 0.8481,
      "step": 132990
    },
    {
      "epoch": 2.3350129040186802,
      "grad_norm": 0.04891507863726666,
      "learning_rate": 0.00011460386977566515,
      "loss": 0.8446,
      "step": 133000
    },
    {
      "epoch": 2.3351884688986817,
      "grad_norm": 0.05408297511992526,
      "learning_rate": 0.0001145928317668757,
      "loss": 0.8397,
      "step": 133010
    },
    {
      "epoch": 2.3353640337786827,
      "grad_norm": 0.06407483894065841,
      "learning_rate": 0.00011458179358586244,
      "loss": 0.8516,
      "step": 133020
    },
    {
      "epoch": 2.335539598658684,
      "grad_norm": 0.055193460072403645,
      "learning_rate": 0.00011457075523276528,
      "loss": 0.8312,
      "step": 133030
    },
    {
      "epoch": 2.3357151635386857,
      "grad_norm": 0.06625849641814638,
      "learning_rate": 0.00011455971670772401,
      "loss": 0.8444,
      "step": 133040
    },
    {
      "epoch": 2.335890728418687,
      "grad_norm": 0.04988375664156093,
      "learning_rate": 0.0001145486780108786,
      "loss": 0.849,
      "step": 133050
    },
    {
      "epoch": 2.3360662932986886,
      "grad_norm": 0.04917992404367163,
      "learning_rate": 0.00011453763914236881,
      "loss": 0.8374,
      "step": 133060
    },
    {
      "epoch": 2.33624185817869,
      "grad_norm": 0.09263533250985777,
      "learning_rate": 0.00011452660010233452,
      "loss": 0.8471,
      "step": 133070
    },
    {
      "epoch": 2.336417423058691,
      "grad_norm": 0.062406845558656006,
      "learning_rate": 0.00011451556089091562,
      "loss": 0.8401,
      "step": 133080
    },
    {
      "epoch": 2.3365929879386926,
      "grad_norm": 0.05727356816131072,
      "learning_rate": 0.00011450452150825197,
      "loss": 0.8437,
      "step": 133090
    },
    {
      "epoch": 2.336768552818694,
      "grad_norm": 0.05377210058162804,
      "learning_rate": 0.00011449348195448349,
      "loss": 0.8529,
      "step": 133100
    },
    {
      "epoch": 2.3369441176986956,
      "grad_norm": 0.05609071246329465,
      "learning_rate": 0.00011448244222974995,
      "loss": 0.847,
      "step": 133110
    },
    {
      "epoch": 2.337119682578697,
      "grad_norm": 0.06648338747308315,
      "learning_rate": 0.0001144714023341913,
      "loss": 0.8498,
      "step": 133120
    },
    {
      "epoch": 2.3372952474586985,
      "grad_norm": 0.05363243458670433,
      "learning_rate": 0.00011446036226794741,
      "loss": 0.8546,
      "step": 133130
    },
    {
      "epoch": 2.3374708123386996,
      "grad_norm": 0.0551868755198827,
      "learning_rate": 0.00011444932203115814,
      "loss": 0.8453,
      "step": 133140
    },
    {
      "epoch": 2.337646377218701,
      "grad_norm": 0.04560405110848379,
      "learning_rate": 0.00011443828162396341,
      "loss": 0.8512,
      "step": 133150
    },
    {
      "epoch": 2.3378219420987025,
      "grad_norm": 0.05840707024417413,
      "learning_rate": 0.00011442724104650306,
      "loss": 0.8443,
      "step": 133160
    },
    {
      "epoch": 2.337997506978704,
      "grad_norm": 0.043720938290349544,
      "learning_rate": 0.000114416200298917,
      "loss": 0.8461,
      "step": 133170
    },
    {
      "epoch": 2.3381730718587055,
      "grad_norm": 0.08459815444522543,
      "learning_rate": 0.00011440515938134509,
      "loss": 0.8445,
      "step": 133180
    },
    {
      "epoch": 2.338348636738707,
      "grad_norm": 0.06729653270687617,
      "learning_rate": 0.00011439411829392723,
      "loss": 0.8451,
      "step": 133190
    },
    {
      "epoch": 2.338524201618708,
      "grad_norm": 0.07184692595620784,
      "learning_rate": 0.00011438307703680336,
      "loss": 0.8452,
      "step": 133200
    },
    {
      "epoch": 2.3386997664987095,
      "grad_norm": 0.05164048398122927,
      "learning_rate": 0.00011437203561011331,
      "loss": 0.8483,
      "step": 133210
    },
    {
      "epoch": 2.338875331378711,
      "grad_norm": 0.05463675442979536,
      "learning_rate": 0.00011436099401399705,
      "loss": 0.8476,
      "step": 133220
    },
    {
      "epoch": 2.3390508962587124,
      "grad_norm": 0.04317434346770487,
      "learning_rate": 0.00011434995224859439,
      "loss": 0.8471,
      "step": 133230
    },
    {
      "epoch": 2.339226461138714,
      "grad_norm": 0.062031145240477176,
      "learning_rate": 0.00011433891031404528,
      "loss": 0.8356,
      "step": 133240
    },
    {
      "epoch": 2.3394020260187154,
      "grad_norm": 0.06520978768851769,
      "learning_rate": 0.00011432786821048965,
      "loss": 0.845,
      "step": 133250
    },
    {
      "epoch": 2.339577590898717,
      "grad_norm": 0.06514406993847237,
      "learning_rate": 0.00011431682593806734,
      "loss": 0.845,
      "step": 133260
    },
    {
      "epoch": 2.339753155778718,
      "grad_norm": 0.05417855864686621,
      "learning_rate": 0.00011430578349691832,
      "loss": 0.853,
      "step": 133270
    },
    {
      "epoch": 2.3399287206587194,
      "grad_norm": 0.05043245664462972,
      "learning_rate": 0.00011429474088718246,
      "loss": 0.8435,
      "step": 133280
    },
    {
      "epoch": 2.340104285538721,
      "grad_norm": 0.07489475626823756,
      "learning_rate": 0.00011428369810899968,
      "loss": 0.8387,
      "step": 133290
    },
    {
      "epoch": 2.3402798504187223,
      "grad_norm": 0.04874703184042555,
      "learning_rate": 0.00011427265516250993,
      "loss": 0.8403,
      "step": 133300
    },
    {
      "epoch": 2.340455415298724,
      "grad_norm": 0.06425544947963778,
      "learning_rate": 0.00011426161204785304,
      "loss": 0.8377,
      "step": 133310
    },
    {
      "epoch": 2.340630980178725,
      "grad_norm": 0.0578164646124323,
      "learning_rate": 0.00011425056876516903,
      "loss": 0.8435,
      "step": 133320
    },
    {
      "epoch": 2.3408065450587263,
      "grad_norm": 0.05192584398837824,
      "learning_rate": 0.00011423952531459776,
      "loss": 0.845,
      "step": 133330
    },
    {
      "epoch": 2.340982109938728,
      "grad_norm": 0.05136040476776188,
      "learning_rate": 0.00011422848169627918,
      "loss": 0.8518,
      "step": 133340
    },
    {
      "epoch": 2.3411576748187293,
      "grad_norm": 0.0717519613528112,
      "learning_rate": 0.0001142174379103532,
      "loss": 0.8485,
      "step": 133350
    },
    {
      "epoch": 2.3413332396987308,
      "grad_norm": 0.07435547980886908,
      "learning_rate": 0.00011420639395695972,
      "loss": 0.8429,
      "step": 133360
    },
    {
      "epoch": 2.3415088045787322,
      "grad_norm": 0.05402175982884939,
      "learning_rate": 0.00011419534983623874,
      "loss": 0.8483,
      "step": 133370
    },
    {
      "epoch": 2.3416843694587337,
      "grad_norm": 0.05020869648630097,
      "learning_rate": 0.0001141843055483301,
      "loss": 0.8477,
      "step": 133380
    },
    {
      "epoch": 2.3418599343387347,
      "grad_norm": 0.09396413716224337,
      "learning_rate": 0.00011417326109337383,
      "loss": 0.8545,
      "step": 133390
    },
    {
      "epoch": 2.342035499218736,
      "grad_norm": 0.04761332011705497,
      "learning_rate": 0.00011416221647150978,
      "loss": 0.8435,
      "step": 133400
    },
    {
      "epoch": 2.3422110640987377,
      "grad_norm": 0.05903690187175423,
      "learning_rate": 0.00011415117168287793,
      "loss": 0.8442,
      "step": 133410
    },
    {
      "epoch": 2.342386628978739,
      "grad_norm": 0.05478946130131419,
      "learning_rate": 0.00011414012672761824,
      "loss": 0.844,
      "step": 133420
    },
    {
      "epoch": 2.3425621938587406,
      "grad_norm": 0.057047027069992824,
      "learning_rate": 0.0001141290816058706,
      "loss": 0.8438,
      "step": 133430
    },
    {
      "epoch": 2.3427377587387417,
      "grad_norm": 0.07901534477583794,
      "learning_rate": 0.00011411803631777501,
      "loss": 0.8516,
      "step": 133440
    },
    {
      "epoch": 2.342913323618743,
      "grad_norm": 0.05150256974856414,
      "learning_rate": 0.00011410699086347136,
      "loss": 0.8451,
      "step": 133450
    },
    {
      "epoch": 2.3430888884987446,
      "grad_norm": 0.05341598865885184,
      "learning_rate": 0.00011409594524309961,
      "loss": 0.8437,
      "step": 133460
    },
    {
      "epoch": 2.343264453378746,
      "grad_norm": 0.05743868420471526,
      "learning_rate": 0.00011408489945679975,
      "loss": 0.8437,
      "step": 133470
    },
    {
      "epoch": 2.3434400182587476,
      "grad_norm": 0.05313064148828204,
      "learning_rate": 0.0001140738535047117,
      "loss": 0.8459,
      "step": 133480
    },
    {
      "epoch": 2.343615583138749,
      "grad_norm": 0.060995909593158024,
      "learning_rate": 0.00011406280738697541,
      "loss": 0.8442,
      "step": 133490
    },
    {
      "epoch": 2.3437911480187505,
      "grad_norm": 0.04991348597875841,
      "learning_rate": 0.00011405176110373088,
      "loss": 0.8389,
      "step": 133500
    },
    {
      "epoch": 2.3439667128987516,
      "grad_norm": 0.0634234182834563,
      "learning_rate": 0.00011404071465511801,
      "loss": 0.8503,
      "step": 133510
    },
    {
      "epoch": 2.344142277778753,
      "grad_norm": 0.060939383111456844,
      "learning_rate": 0.0001140296680412768,
      "loss": 0.8431,
      "step": 133520
    },
    {
      "epoch": 2.3443178426587545,
      "grad_norm": 0.04599197829761786,
      "learning_rate": 0.00011401862126234714,
      "loss": 0.8488,
      "step": 133530
    },
    {
      "epoch": 2.344493407538756,
      "grad_norm": 0.061431684835272025,
      "learning_rate": 0.0001140075743184691,
      "loss": 0.8486,
      "step": 133540
    },
    {
      "epoch": 2.3446689724187575,
      "grad_norm": 0.07871990985251791,
      "learning_rate": 0.0001139965272097826,
      "loss": 0.8443,
      "step": 133550
    },
    {
      "epoch": 2.3448445372987585,
      "grad_norm": 0.061352389774027676,
      "learning_rate": 0.00011398547993642765,
      "loss": 0.8441,
      "step": 133560
    },
    {
      "epoch": 2.34502010217876,
      "grad_norm": 0.054565289218371095,
      "learning_rate": 0.00011397443249854412,
      "loss": 0.8417,
      "step": 133570
    },
    {
      "epoch": 2.3451956670587615,
      "grad_norm": 0.0575526846466423,
      "learning_rate": 0.00011396338489627207,
      "loss": 0.851,
      "step": 133580
    },
    {
      "epoch": 2.345371231938763,
      "grad_norm": 0.0544901402390001,
      "learning_rate": 0.00011395233712975144,
      "loss": 0.8357,
      "step": 133590
    },
    {
      "epoch": 2.3455467968187644,
      "grad_norm": 0.07141727721092771,
      "learning_rate": 0.0001139412891991222,
      "loss": 0.8429,
      "step": 133600
    },
    {
      "epoch": 2.345722361698766,
      "grad_norm": 0.05651755053268674,
      "learning_rate": 0.00011393024110452437,
      "loss": 0.8479,
      "step": 133610
    },
    {
      "epoch": 2.3458979265787674,
      "grad_norm": 0.08793302104731628,
      "learning_rate": 0.00011391919284609791,
      "loss": 0.8415,
      "step": 133620
    },
    {
      "epoch": 2.3460734914587684,
      "grad_norm": 0.06297835490589034,
      "learning_rate": 0.0001139081444239828,
      "loss": 0.8411,
      "step": 133630
    },
    {
      "epoch": 2.34624905633877,
      "grad_norm": 0.05323587956893086,
      "learning_rate": 0.00011389709583831902,
      "loss": 0.8475,
      "step": 133640
    },
    {
      "epoch": 2.3464246212187714,
      "grad_norm": 0.06103952092462643,
      "learning_rate": 0.00011388604708924655,
      "loss": 0.8468,
      "step": 133650
    },
    {
      "epoch": 2.346600186098773,
      "grad_norm": 0.049469441562608706,
      "learning_rate": 0.00011387499817690542,
      "loss": 0.8448,
      "step": 133660
    },
    {
      "epoch": 2.3467757509787743,
      "grad_norm": 0.054330232568080956,
      "learning_rate": 0.0001138639491014356,
      "loss": 0.8488,
      "step": 133670
    },
    {
      "epoch": 2.3469513158587754,
      "grad_norm": 0.07424862133423164,
      "learning_rate": 0.00011385289986297707,
      "loss": 0.8508,
      "step": 133680
    },
    {
      "epoch": 2.347126880738777,
      "grad_norm": 0.07971854225167067,
      "learning_rate": 0.00011384185046166982,
      "loss": 0.8462,
      "step": 133690
    },
    {
      "epoch": 2.3473024456187783,
      "grad_norm": 0.0502736792146911,
      "learning_rate": 0.00011383080089765386,
      "loss": 0.8457,
      "step": 133700
    },
    {
      "epoch": 2.34747801049878,
      "grad_norm": 0.05815044357912154,
      "learning_rate": 0.0001138197511710692,
      "loss": 0.8389,
      "step": 133710
    },
    {
      "epoch": 2.3476535753787813,
      "grad_norm": 0.05750956747087427,
      "learning_rate": 0.00011380870128205583,
      "loss": 0.8481,
      "step": 133720
    },
    {
      "epoch": 2.3478291402587828,
      "grad_norm": 0.05447350098413739,
      "learning_rate": 0.0001137976512307538,
      "loss": 0.8397,
      "step": 133730
    },
    {
      "epoch": 2.3480047051387842,
      "grad_norm": 0.08051909154800427,
      "learning_rate": 0.00011378660101730305,
      "loss": 0.8453,
      "step": 133740
    },
    {
      "epoch": 2.3481802700187853,
      "grad_norm": 0.058172184245181495,
      "learning_rate": 0.00011377555064184358,
      "loss": 0.8482,
      "step": 133750
    },
    {
      "epoch": 2.3483558348987867,
      "grad_norm": 0.05350353354607648,
      "learning_rate": 0.00011376450010451547,
      "loss": 0.8491,
      "step": 133760
    },
    {
      "epoch": 2.348531399778788,
      "grad_norm": 0.04065558787295076,
      "learning_rate": 0.0001137534494054587,
      "loss": 0.8489,
      "step": 133770
    },
    {
      "epoch": 2.3487069646587897,
      "grad_norm": 0.07998306416919893,
      "learning_rate": 0.00011374239854481328,
      "loss": 0.8496,
      "step": 133780
    },
    {
      "epoch": 2.348882529538791,
      "grad_norm": 0.04725412985017815,
      "learning_rate": 0.00011373134752271923,
      "loss": 0.8482,
      "step": 133790
    },
    {
      "epoch": 2.3490580944187927,
      "grad_norm": 0.04589193171362352,
      "learning_rate": 0.00011372029633931656,
      "loss": 0.8436,
      "step": 133800
    },
    {
      "epoch": 2.3492336592987937,
      "grad_norm": 0.04564785441890648,
      "learning_rate": 0.0001137092449947453,
      "loss": 0.8502,
      "step": 133810
    },
    {
      "epoch": 2.349409224178795,
      "grad_norm": 0.053695174970113564,
      "learning_rate": 0.00011369819348914545,
      "loss": 0.8458,
      "step": 133820
    },
    {
      "epoch": 2.3495847890587966,
      "grad_norm": 0.047985305021954076,
      "learning_rate": 0.00011368714182265708,
      "loss": 0.8383,
      "step": 133830
    },
    {
      "epoch": 2.349760353938798,
      "grad_norm": 0.051069697973699035,
      "learning_rate": 0.00011367608999542019,
      "loss": 0.8564,
      "step": 133840
    },
    {
      "epoch": 2.3499359188187996,
      "grad_norm": 0.051280095325865055,
      "learning_rate": 0.00011366503800757482,
      "loss": 0.8386,
      "step": 133850
    },
    {
      "epoch": 2.350111483698801,
      "grad_norm": 0.05922752650071419,
      "learning_rate": 0.00011365398585926097,
      "loss": 0.8431,
      "step": 133860
    },
    {
      "epoch": 2.350287048578802,
      "grad_norm": 0.057142008276130286,
      "learning_rate": 0.0001136429335506187,
      "loss": 0.8385,
      "step": 133870
    },
    {
      "epoch": 2.3504626134588036,
      "grad_norm": 0.0701950070364163,
      "learning_rate": 0.00011363188108178804,
      "loss": 0.8516,
      "step": 133880
    },
    {
      "epoch": 2.350638178338805,
      "grad_norm": 0.04610639525322846,
      "learning_rate": 0.000113620828452909,
      "loss": 0.8421,
      "step": 133890
    },
    {
      "epoch": 2.3508137432188065,
      "grad_norm": 0.04922782959920052,
      "learning_rate": 0.0001136097756641217,
      "loss": 0.8418,
      "step": 133900
    },
    {
      "epoch": 2.350989308098808,
      "grad_norm": 0.050517958323559375,
      "learning_rate": 0.00011359872271556608,
      "loss": 0.847,
      "step": 133910
    },
    {
      "epoch": 2.3511648729788095,
      "grad_norm": 0.048691442953671664,
      "learning_rate": 0.00011358766960738223,
      "loss": 0.8453,
      "step": 133920
    },
    {
      "epoch": 2.351340437858811,
      "grad_norm": 0.06919537806019121,
      "learning_rate": 0.00011357661633971024,
      "loss": 0.8408,
      "step": 133930
    },
    {
      "epoch": 2.351516002738812,
      "grad_norm": 0.06366174153638317,
      "learning_rate": 0.00011356556291269007,
      "loss": 0.8375,
      "step": 133940
    },
    {
      "epoch": 2.3516915676188135,
      "grad_norm": 0.056672980579162195,
      "learning_rate": 0.00011355450932646182,
      "loss": 0.8399,
      "step": 133950
    },
    {
      "epoch": 2.351867132498815,
      "grad_norm": 0.06604085320127266,
      "learning_rate": 0.00011354345558116555,
      "loss": 0.8489,
      "step": 133960
    },
    {
      "epoch": 2.3520426973788164,
      "grad_norm": 0.05614710251027746,
      "learning_rate": 0.00011353240167694123,
      "loss": 0.8329,
      "step": 133970
    },
    {
      "epoch": 2.352218262258818,
      "grad_norm": 0.06348692754401203,
      "learning_rate": 0.00011352134761392906,
      "loss": 0.848,
      "step": 133980
    },
    {
      "epoch": 2.352393827138819,
      "grad_norm": 0.050907882333801,
      "learning_rate": 0.00011351029339226895,
      "loss": 0.8469,
      "step": 133990
    },
    {
      "epoch": 2.3525693920188204,
      "grad_norm": 0.05208965210778257,
      "learning_rate": 0.00011349923901210106,
      "loss": 0.8468,
      "step": 134000
    },
    {
      "epoch": 2.352744956898822,
      "grad_norm": 0.050262895089600974,
      "learning_rate": 0.00011348818447356542,
      "loss": 0.8454,
      "step": 134010
    },
    {
      "epoch": 2.3529205217788234,
      "grad_norm": 0.051559588390177775,
      "learning_rate": 0.0001134771297768021,
      "loss": 0.8444,
      "step": 134020
    },
    {
      "epoch": 2.353096086658825,
      "grad_norm": 0.052674868698447945,
      "learning_rate": 0.00011346607492195112,
      "loss": 0.8443,
      "step": 134030
    },
    {
      "epoch": 2.3532716515388263,
      "grad_norm": 0.06363751583000443,
      "learning_rate": 0.00011345501990915259,
      "loss": 0.8412,
      "step": 134040
    },
    {
      "epoch": 2.353447216418828,
      "grad_norm": 0.05299799458550487,
      "learning_rate": 0.00011344396473854657,
      "loss": 0.8431,
      "step": 134050
    },
    {
      "epoch": 2.353622781298829,
      "grad_norm": 0.049198275904631446,
      "learning_rate": 0.00011343290941027314,
      "loss": 0.8471,
      "step": 134060
    },
    {
      "epoch": 2.3537983461788303,
      "grad_norm": 0.057771010492266565,
      "learning_rate": 0.00011342185392447238,
      "loss": 0.8489,
      "step": 134070
    },
    {
      "epoch": 2.353973911058832,
      "grad_norm": 0.06354606721353578,
      "learning_rate": 0.00011341079828128431,
      "loss": 0.8375,
      "step": 134080
    },
    {
      "epoch": 2.3541494759388333,
      "grad_norm": 0.08993811277198738,
      "learning_rate": 0.00011339974248084906,
      "loss": 0.8437,
      "step": 134090
    },
    {
      "epoch": 2.3543250408188348,
      "grad_norm": 0.0424921774849628,
      "learning_rate": 0.00011338868652330672,
      "loss": 0.8483,
      "step": 134100
    },
    {
      "epoch": 2.354500605698836,
      "grad_norm": 0.05941425782265578,
      "learning_rate": 0.00011337763040879732,
      "loss": 0.8529,
      "step": 134110
    },
    {
      "epoch": 2.3546761705788373,
      "grad_norm": 0.0736690796796343,
      "learning_rate": 0.00011336657413746099,
      "loss": 0.8474,
      "step": 134120
    },
    {
      "epoch": 2.3548517354588387,
      "grad_norm": 0.06591040263933456,
      "learning_rate": 0.0001133555177094378,
      "loss": 0.8407,
      "step": 134130
    },
    {
      "epoch": 2.3550273003388402,
      "grad_norm": 0.05759538132838987,
      "learning_rate": 0.0001133444611248678,
      "loss": 0.8322,
      "step": 134140
    },
    {
      "epoch": 2.3552028652188417,
      "grad_norm": 0.06575002883440076,
      "learning_rate": 0.00011333340438389115,
      "loss": 0.8399,
      "step": 134150
    },
    {
      "epoch": 2.355378430098843,
      "grad_norm": 0.05478873016768162,
      "learning_rate": 0.00011332234748664789,
      "loss": 0.8486,
      "step": 134160
    },
    {
      "epoch": 2.3555539949788447,
      "grad_norm": 0.06005695129748368,
      "learning_rate": 0.00011331129043327815,
      "loss": 0.8516,
      "step": 134170
    },
    {
      "epoch": 2.3557295598588457,
      "grad_norm": 0.0798240069685387,
      "learning_rate": 0.00011330023322392198,
      "loss": 0.847,
      "step": 134180
    },
    {
      "epoch": 2.355905124738847,
      "grad_norm": 0.051078978658778106,
      "learning_rate": 0.00011328917585871948,
      "loss": 0.8437,
      "step": 134190
    },
    {
      "epoch": 2.3560806896188486,
      "grad_norm": 0.05103345377779297,
      "learning_rate": 0.00011327811833781079,
      "loss": 0.8481,
      "step": 134200
    },
    {
      "epoch": 2.35625625449885,
      "grad_norm": 0.07112212415029301,
      "learning_rate": 0.00011326706066133594,
      "loss": 0.8441,
      "step": 134210
    },
    {
      "epoch": 2.3564318193788516,
      "grad_norm": 0.0876135528747004,
      "learning_rate": 0.00011325600282943512,
      "loss": 0.8408,
      "step": 134220
    },
    {
      "epoch": 2.3566073842588526,
      "grad_norm": 0.059586193716567315,
      "learning_rate": 0.00011324494484224839,
      "loss": 0.8537,
      "step": 134230
    },
    {
      "epoch": 2.356782949138854,
      "grad_norm": 0.05213501446566046,
      "learning_rate": 0.0001132338866999159,
      "loss": 0.842,
      "step": 134240
    },
    {
      "epoch": 2.3569585140188556,
      "grad_norm": 0.053887887050827,
      "learning_rate": 0.00011322282840257768,
      "loss": 0.8479,
      "step": 134250
    },
    {
      "epoch": 2.357134078898857,
      "grad_norm": 0.05930323804927232,
      "learning_rate": 0.00011321176995037385,
      "loss": 0.8404,
      "step": 134260
    },
    {
      "epoch": 2.3573096437788585,
      "grad_norm": 0.0454905671134837,
      "learning_rate": 0.00011320071134344461,
      "loss": 0.8484,
      "step": 134270
    },
    {
      "epoch": 2.35748520865886,
      "grad_norm": 0.05985554956882661,
      "learning_rate": 0.00011318965258193,
      "loss": 0.8424,
      "step": 134280
    },
    {
      "epoch": 2.3576607735388615,
      "grad_norm": 0.07924237745192811,
      "learning_rate": 0.00011317859366597016,
      "loss": 0.837,
      "step": 134290
    },
    {
      "epoch": 2.3578363384188625,
      "grad_norm": 0.07205908809117192,
      "learning_rate": 0.0001131675345957052,
      "loss": 0.836,
      "step": 134300
    },
    {
      "epoch": 2.358011903298864,
      "grad_norm": 0.06187650068168415,
      "learning_rate": 0.00011315647537127523,
      "loss": 0.8416,
      "step": 134310
    },
    {
      "epoch": 2.3581874681788655,
      "grad_norm": 0.046592971578934994,
      "learning_rate": 0.00011314541599282042,
      "loss": 0.8436,
      "step": 134320
    },
    {
      "epoch": 2.358363033058867,
      "grad_norm": 0.06343894127762152,
      "learning_rate": 0.00011313435646048083,
      "loss": 0.8467,
      "step": 134330
    },
    {
      "epoch": 2.3585385979388684,
      "grad_norm": 0.056224469783179944,
      "learning_rate": 0.00011312329677439664,
      "loss": 0.8413,
      "step": 134340
    },
    {
      "epoch": 2.3587141628188695,
      "grad_norm": 0.05722784230457622,
      "learning_rate": 0.00011311223693470793,
      "loss": 0.8477,
      "step": 134350
    },
    {
      "epoch": 2.358889727698871,
      "grad_norm": 0.0662201563007302,
      "learning_rate": 0.00011310117694155491,
      "loss": 0.8407,
      "step": 134360
    },
    {
      "epoch": 2.3590652925788724,
      "grad_norm": 0.06773517247192046,
      "learning_rate": 0.00011309011679507761,
      "loss": 0.8427,
      "step": 134370
    },
    {
      "epoch": 2.359240857458874,
      "grad_norm": 0.06805427120856299,
      "learning_rate": 0.0001130790564954162,
      "loss": 0.8448,
      "step": 134380
    },
    {
      "epoch": 2.3594164223388754,
      "grad_norm": 0.05092751161665691,
      "learning_rate": 0.00011306799604271088,
      "loss": 0.847,
      "step": 134390
    },
    {
      "epoch": 2.359591987218877,
      "grad_norm": 0.061491711813521144,
      "learning_rate": 0.00011305693543710169,
      "loss": 0.8499,
      "step": 134400
    },
    {
      "epoch": 2.3597675520988783,
      "grad_norm": 0.04802451706881648,
      "learning_rate": 0.00011304587467872887,
      "loss": 0.8523,
      "step": 134410
    },
    {
      "epoch": 2.3599431169788794,
      "grad_norm": 0.042735942462558886,
      "learning_rate": 0.00011303481376773246,
      "loss": 0.8421,
      "step": 134420
    },
    {
      "epoch": 2.360118681858881,
      "grad_norm": 0.05639165736157412,
      "learning_rate": 0.00011302375270425268,
      "loss": 0.8458,
      "step": 134430
    },
    {
      "epoch": 2.3602942467388823,
      "grad_norm": 0.058342284284294735,
      "learning_rate": 0.00011301269148842966,
      "loss": 0.8475,
      "step": 134440
    },
    {
      "epoch": 2.360469811618884,
      "grad_norm": 0.05362668683256812,
      "learning_rate": 0.0001130016301204035,
      "loss": 0.8539,
      "step": 134450
    },
    {
      "epoch": 2.3606453764988853,
      "grad_norm": 0.05631504529116611,
      "learning_rate": 0.0001129905686003144,
      "loss": 0.8488,
      "step": 134460
    },
    {
      "epoch": 2.3608209413788868,
      "grad_norm": 0.08893655411120635,
      "learning_rate": 0.00011297950692830252,
      "loss": 0.841,
      "step": 134470
    },
    {
      "epoch": 2.360996506258888,
      "grad_norm": 0.04651328065396194,
      "learning_rate": 0.00011296844510450795,
      "loss": 0.8434,
      "step": 134480
    },
    {
      "epoch": 2.3611720711388893,
      "grad_norm": 0.07185462307379226,
      "learning_rate": 0.00011295738312907093,
      "loss": 0.8455,
      "step": 134490
    },
    {
      "epoch": 2.3613476360188907,
      "grad_norm": 0.07825435549185263,
      "learning_rate": 0.00011294632100213155,
      "loss": 0.845,
      "step": 134500
    },
    {
      "epoch": 2.3615232008988922,
      "grad_norm": 0.07329956798573353,
      "learning_rate": 0.00011293525872383,
      "loss": 0.835,
      "step": 134510
    },
    {
      "epoch": 2.3616987657788937,
      "grad_norm": 0.05449693927899409,
      "learning_rate": 0.00011292419629430644,
      "loss": 0.8414,
      "step": 134520
    },
    {
      "epoch": 2.361874330658895,
      "grad_norm": 0.05147985849243901,
      "learning_rate": 0.00011291313371370102,
      "loss": 0.8354,
      "step": 134530
    },
    {
      "epoch": 2.362049895538896,
      "grad_norm": 0.05542904436273784,
      "learning_rate": 0.00011290207098215391,
      "loss": 0.8484,
      "step": 134540
    },
    {
      "epoch": 2.3622254604188977,
      "grad_norm": 0.05972478654747157,
      "learning_rate": 0.00011289100809980528,
      "loss": 0.8407,
      "step": 134550
    },
    {
      "epoch": 2.362401025298899,
      "grad_norm": 0.04933241126890812,
      "learning_rate": 0.00011287994506679532,
      "loss": 0.8478,
      "step": 134560
    },
    {
      "epoch": 2.3625765901789006,
      "grad_norm": 0.058987268514871224,
      "learning_rate": 0.00011286888188326415,
      "loss": 0.8396,
      "step": 134570
    },
    {
      "epoch": 2.362752155058902,
      "grad_norm": 0.05704147932383239,
      "learning_rate": 0.000112857818549352,
      "loss": 0.8456,
      "step": 134580
    },
    {
      "epoch": 2.3629277199389036,
      "grad_norm": 0.0576600440389391,
      "learning_rate": 0.000112846755065199,
      "loss": 0.8495,
      "step": 134590
    },
    {
      "epoch": 2.3631032848189046,
      "grad_norm": 0.05295466996177986,
      "learning_rate": 0.00011283569143094532,
      "loss": 0.8495,
      "step": 134600
    },
    {
      "epoch": 2.363278849698906,
      "grad_norm": 0.056840243017410294,
      "learning_rate": 0.00011282462764673118,
      "loss": 0.844,
      "step": 134610
    },
    {
      "epoch": 2.3634544145789076,
      "grad_norm": 0.06014432289874687,
      "learning_rate": 0.00011281356371269674,
      "loss": 0.8445,
      "step": 134620
    },
    {
      "epoch": 2.363629979458909,
      "grad_norm": 0.06151625454873578,
      "learning_rate": 0.0001128024996289822,
      "loss": 0.8447,
      "step": 134630
    },
    {
      "epoch": 2.3638055443389105,
      "grad_norm": 0.04967207961854145,
      "learning_rate": 0.00011279143539572772,
      "loss": 0.8488,
      "step": 134640
    },
    {
      "epoch": 2.363981109218912,
      "grad_norm": 0.051862174218889154,
      "learning_rate": 0.00011278037101307345,
      "loss": 0.8375,
      "step": 134650
    },
    {
      "epoch": 2.364156674098913,
      "grad_norm": 0.054961183465586906,
      "learning_rate": 0.00011276930648115968,
      "loss": 0.8445,
      "step": 134660
    },
    {
      "epoch": 2.3643322389789145,
      "grad_norm": 0.056814842391382285,
      "learning_rate": 0.00011275824180012653,
      "loss": 0.8437,
      "step": 134670
    },
    {
      "epoch": 2.364507803858916,
      "grad_norm": 0.06028191984778268,
      "learning_rate": 0.00011274717697011419,
      "loss": 0.8416,
      "step": 134680
    },
    {
      "epoch": 2.3646833687389175,
      "grad_norm": 0.05017310789774405,
      "learning_rate": 0.00011273611199126289,
      "loss": 0.8461,
      "step": 134690
    },
    {
      "epoch": 2.364858933618919,
      "grad_norm": 0.047324711938525346,
      "learning_rate": 0.0001127250468637128,
      "loss": 0.8489,
      "step": 134700
    },
    {
      "epoch": 2.3650344984989204,
      "grad_norm": 0.04831003836280072,
      "learning_rate": 0.00011271398158760409,
      "loss": 0.845,
      "step": 134710
    },
    {
      "epoch": 2.365210063378922,
      "grad_norm": 0.055629309918760195,
      "learning_rate": 0.00011270291616307698,
      "loss": 0.847,
      "step": 134720
    },
    {
      "epoch": 2.365385628258923,
      "grad_norm": 0.043742760066158286,
      "learning_rate": 0.00011269185059027172,
      "loss": 0.8402,
      "step": 134730
    },
    {
      "epoch": 2.3655611931389244,
      "grad_norm": 0.04697661505185484,
      "learning_rate": 0.00011268078486932842,
      "loss": 0.8457,
      "step": 134740
    },
    {
      "epoch": 2.365736758018926,
      "grad_norm": 0.054346910638125356,
      "learning_rate": 0.00011266971900038743,
      "loss": 0.8418,
      "step": 134750
    },
    {
      "epoch": 2.3659123228989274,
      "grad_norm": 0.049495254821302605,
      "learning_rate": 0.00011265865298358878,
      "loss": 0.8369,
      "step": 134760
    },
    {
      "epoch": 2.366087887778929,
      "grad_norm": 0.046722739581489876,
      "learning_rate": 0.00011264758681907277,
      "loss": 0.8372,
      "step": 134770
    },
    {
      "epoch": 2.36626345265893,
      "grad_norm": 0.05772425001252415,
      "learning_rate": 0.00011263652050697964,
      "loss": 0.8446,
      "step": 134780
    },
    {
      "epoch": 2.3664390175389314,
      "grad_norm": 0.04685352141677186,
      "learning_rate": 0.00011262545404744953,
      "loss": 0.8463,
      "step": 134790
    },
    {
      "epoch": 2.366614582418933,
      "grad_norm": 0.05424722754100176,
      "learning_rate": 0.00011261438744062269,
      "loss": 0.8355,
      "step": 134800
    },
    {
      "epoch": 2.3667901472989343,
      "grad_norm": 0.059222465518485146,
      "learning_rate": 0.00011260332068663937,
      "loss": 0.8473,
      "step": 134810
    },
    {
      "epoch": 2.366965712178936,
      "grad_norm": 0.06532298023179649,
      "learning_rate": 0.00011259225378563972,
      "loss": 0.8491,
      "step": 134820
    },
    {
      "epoch": 2.3671412770589373,
      "grad_norm": 0.05633681457684546,
      "learning_rate": 0.00011258118673776402,
      "loss": 0.8434,
      "step": 134830
    },
    {
      "epoch": 2.3673168419389388,
      "grad_norm": 0.04725188109238191,
      "learning_rate": 0.00011257011954315244,
      "loss": 0.8493,
      "step": 134840
    },
    {
      "epoch": 2.36749240681894,
      "grad_norm": 0.066596828293837,
      "learning_rate": 0.00011255905220194524,
      "loss": 0.8387,
      "step": 134850
    },
    {
      "epoch": 2.3676679716989413,
      "grad_norm": 0.06594099624330622,
      "learning_rate": 0.00011254798471428263,
      "loss": 0.836,
      "step": 134860
    },
    {
      "epoch": 2.3678435365789428,
      "grad_norm": 0.0721137610254976,
      "learning_rate": 0.00011253691708030485,
      "loss": 0.8448,
      "step": 134870
    },
    {
      "epoch": 2.3680191014589442,
      "grad_norm": 0.044927891086740224,
      "learning_rate": 0.00011252584930015211,
      "loss": 0.8417,
      "step": 134880
    },
    {
      "epoch": 2.3681946663389457,
      "grad_norm": 0.06688511018046475,
      "learning_rate": 0.00011251478137396462,
      "loss": 0.844,
      "step": 134890
    },
    {
      "epoch": 2.3683702312189467,
      "grad_norm": 0.048054125548318526,
      "learning_rate": 0.00011250371330188269,
      "loss": 0.846,
      "step": 134900
    },
    {
      "epoch": 2.368545796098948,
      "grad_norm": 0.05024961818885813,
      "learning_rate": 0.00011249264508404649,
      "loss": 0.856,
      "step": 134910
    },
    {
      "epoch": 2.3687213609789497,
      "grad_norm": 0.051886287078572034,
      "learning_rate": 0.00011248157672059631,
      "loss": 0.8391,
      "step": 134920
    },
    {
      "epoch": 2.368896925858951,
      "grad_norm": 0.05240016887912636,
      "learning_rate": 0.00011247050821167229,
      "loss": 0.8393,
      "step": 134930
    },
    {
      "epoch": 2.3690724907389527,
      "grad_norm": 0.04852978780496837,
      "learning_rate": 0.00011245943955741476,
      "loss": 0.8503,
      "step": 134940
    },
    {
      "epoch": 2.369248055618954,
      "grad_norm": 0.0576291631114266,
      "learning_rate": 0.00011244837075796395,
      "loss": 0.8411,
      "step": 134950
    },
    {
      "epoch": 2.3694236204989556,
      "grad_norm": 0.04761097707247367,
      "learning_rate": 0.00011243730181346006,
      "loss": 0.8472,
      "step": 134960
    },
    {
      "epoch": 2.3695991853789566,
      "grad_norm": 0.06218803529343042,
      "learning_rate": 0.00011242623272404338,
      "loss": 0.8443,
      "step": 134970
    },
    {
      "epoch": 2.369774750258958,
      "grad_norm": 0.048111598485552166,
      "learning_rate": 0.00011241516348985414,
      "loss": 0.8471,
      "step": 134980
    },
    {
      "epoch": 2.3699503151389596,
      "grad_norm": 0.05231987940251522,
      "learning_rate": 0.00011240409411103261,
      "loss": 0.8514,
      "step": 134990
    },
    {
      "epoch": 2.370125880018961,
      "grad_norm": 0.04234706713217439,
      "learning_rate": 0.00011239302458771899,
      "loss": 0.8418,
      "step": 135000
    },
    {
      "epoch": 2.3703014448989626,
      "grad_norm": 0.05484713594735871,
      "learning_rate": 0.00011238195492005355,
      "loss": 0.8445,
      "step": 135010
    },
    {
      "epoch": 2.3704770097789636,
      "grad_norm": 0.0529803929181867,
      "learning_rate": 0.0001123708851081766,
      "loss": 0.8515,
      "step": 135020
    },
    {
      "epoch": 2.370652574658965,
      "grad_norm": 0.04823609707079143,
      "learning_rate": 0.00011235981515222835,
      "loss": 0.8441,
      "step": 135030
    },
    {
      "epoch": 2.3708281395389665,
      "grad_norm": 0.06279644760476553,
      "learning_rate": 0.00011234874505234907,
      "loss": 0.84,
      "step": 135040
    },
    {
      "epoch": 2.371003704418968,
      "grad_norm": 0.048406056834322544,
      "learning_rate": 0.000112337674808679,
      "loss": 0.847,
      "step": 135050
    },
    {
      "epoch": 2.3711792692989695,
      "grad_norm": 0.052854682366336836,
      "learning_rate": 0.0001123266044213584,
      "loss": 0.8446,
      "step": 135060
    },
    {
      "epoch": 2.371354834178971,
      "grad_norm": 0.05946573253358468,
      "learning_rate": 0.00011231553389052758,
      "loss": 0.8428,
      "step": 135070
    },
    {
      "epoch": 2.3715303990589724,
      "grad_norm": 0.05294775091356678,
      "learning_rate": 0.00011230446321632677,
      "loss": 0.8409,
      "step": 135080
    },
    {
      "epoch": 2.3717059639389735,
      "grad_norm": 0.05964013778116095,
      "learning_rate": 0.00011229339239889627,
      "loss": 0.8384,
      "step": 135090
    },
    {
      "epoch": 2.371881528818975,
      "grad_norm": 0.053903976505798666,
      "learning_rate": 0.00011228232143837629,
      "loss": 0.8419,
      "step": 135100
    },
    {
      "epoch": 2.3720570936989764,
      "grad_norm": 0.05012152712146011,
      "learning_rate": 0.00011227125033490712,
      "loss": 0.85,
      "step": 135110
    },
    {
      "epoch": 2.372232658578978,
      "grad_norm": 0.07881720507300939,
      "learning_rate": 0.0001122601790886291,
      "loss": 0.8629,
      "step": 135120
    },
    {
      "epoch": 2.3724082234589794,
      "grad_norm": 0.07840137925099194,
      "learning_rate": 0.0001122491076996824,
      "loss": 0.8453,
      "step": 135130
    },
    {
      "epoch": 2.3725837883389804,
      "grad_norm": 0.055860901533570526,
      "learning_rate": 0.0001122380361682074,
      "loss": 0.8386,
      "step": 135140
    },
    {
      "epoch": 2.372759353218982,
      "grad_norm": 0.05701449364322488,
      "learning_rate": 0.00011222696449434431,
      "loss": 0.8477,
      "step": 135150
    },
    {
      "epoch": 2.3729349180989834,
      "grad_norm": 0.0645531239355094,
      "learning_rate": 0.00011221589267823343,
      "loss": 0.8408,
      "step": 135160
    },
    {
      "epoch": 2.373110482978985,
      "grad_norm": 0.062226106757402366,
      "learning_rate": 0.00011220482072001502,
      "loss": 0.8418,
      "step": 135170
    },
    {
      "epoch": 2.3732860478589863,
      "grad_norm": 0.05243973295520538,
      "learning_rate": 0.00011219374861982936,
      "loss": 0.8433,
      "step": 135180
    },
    {
      "epoch": 2.373461612738988,
      "grad_norm": 0.048019556474398775,
      "learning_rate": 0.00011218267637781681,
      "loss": 0.8446,
      "step": 135190
    },
    {
      "epoch": 2.3736371776189893,
      "grad_norm": 0.05020999620157854,
      "learning_rate": 0.00011217160399411759,
      "loss": 0.8425,
      "step": 135200
    },
    {
      "epoch": 2.3738127424989903,
      "grad_norm": 0.07144757194152918,
      "learning_rate": 0.00011216053146887201,
      "loss": 0.8393,
      "step": 135210
    },
    {
      "epoch": 2.373988307378992,
      "grad_norm": 0.04287802012962726,
      "learning_rate": 0.00011214945880222036,
      "loss": 0.846,
      "step": 135220
    },
    {
      "epoch": 2.3741638722589933,
      "grad_norm": 0.052773430132874534,
      "learning_rate": 0.00011213838599430293,
      "loss": 0.8402,
      "step": 135230
    },
    {
      "epoch": 2.3743394371389948,
      "grad_norm": 0.06476892247770304,
      "learning_rate": 0.00011212731304526001,
      "loss": 0.8408,
      "step": 135240
    },
    {
      "epoch": 2.3745150020189962,
      "grad_norm": 0.05688639641141242,
      "learning_rate": 0.00011211623995523188,
      "loss": 0.8494,
      "step": 135250
    },
    {
      "epoch": 2.3746905668989977,
      "grad_norm": 0.05814186556294543,
      "learning_rate": 0.00011210516672435891,
      "loss": 0.8437,
      "step": 135260
    },
    {
      "epoch": 2.3748661317789987,
      "grad_norm": 0.05513625077471902,
      "learning_rate": 0.00011209409335278132,
      "loss": 0.8434,
      "step": 135270
    },
    {
      "epoch": 2.3750416966590002,
      "grad_norm": 0.05593403920596901,
      "learning_rate": 0.00011208301984063943,
      "loss": 0.8444,
      "step": 135280
    },
    {
      "epoch": 2.3752172615390017,
      "grad_norm": 0.05254619019419086,
      "learning_rate": 0.00011207194618807358,
      "loss": 0.84,
      "step": 135290
    },
    {
      "epoch": 2.375392826419003,
      "grad_norm": 0.07809043129805891,
      "learning_rate": 0.00011206087239522401,
      "loss": 0.8387,
      "step": 135300
    },
    {
      "epoch": 2.3755683912990047,
      "grad_norm": 0.058259908961249544,
      "learning_rate": 0.00011204979846223109,
      "loss": 0.8517,
      "step": 135310
    },
    {
      "epoch": 2.375743956179006,
      "grad_norm": 0.0556637799919183,
      "learning_rate": 0.00011203872438923513,
      "loss": 0.8436,
      "step": 135320
    },
    {
      "epoch": 2.375919521059007,
      "grad_norm": 0.08251639901100333,
      "learning_rate": 0.00011202765017637639,
      "loss": 0.8392,
      "step": 135330
    },
    {
      "epoch": 2.3760950859390086,
      "grad_norm": 0.06321009536591687,
      "learning_rate": 0.00011201657582379521,
      "loss": 0.8406,
      "step": 135340
    },
    {
      "epoch": 2.37627065081901,
      "grad_norm": 0.04797544349248897,
      "learning_rate": 0.00011200550133163188,
      "loss": 0.8468,
      "step": 135350
    },
    {
      "epoch": 2.3764462156990116,
      "grad_norm": 0.05270035445523443,
      "learning_rate": 0.00011199442670002678,
      "loss": 0.8386,
      "step": 135360
    },
    {
      "epoch": 2.376621780579013,
      "grad_norm": 0.06254608246911306,
      "learning_rate": 0.00011198335192912018,
      "loss": 0.8451,
      "step": 135370
    },
    {
      "epoch": 2.3767973454590146,
      "grad_norm": 0.057326655550731354,
      "learning_rate": 0.00011197227701905242,
      "loss": 0.8494,
      "step": 135380
    },
    {
      "epoch": 2.376972910339016,
      "grad_norm": 0.05723587240710077,
      "learning_rate": 0.00011196120196996376,
      "loss": 0.8498,
      "step": 135390
    },
    {
      "epoch": 2.377148475219017,
      "grad_norm": 0.04824233037035662,
      "learning_rate": 0.0001119501267819946,
      "loss": 0.8567,
      "step": 135400
    },
    {
      "epoch": 2.3773240400990185,
      "grad_norm": 0.04268248018070617,
      "learning_rate": 0.00011193905145528522,
      "loss": 0.8467,
      "step": 135410
    },
    {
      "epoch": 2.37749960497902,
      "grad_norm": 0.05245071595333195,
      "learning_rate": 0.00011192797598997597,
      "loss": 0.8478,
      "step": 135420
    },
    {
      "epoch": 2.3776751698590215,
      "grad_norm": 0.05590186334180159,
      "learning_rate": 0.00011191690038620719,
      "loss": 0.8406,
      "step": 135430
    },
    {
      "epoch": 2.377850734739023,
      "grad_norm": 0.05133072927163954,
      "learning_rate": 0.00011190582464411918,
      "loss": 0.8508,
      "step": 135440
    },
    {
      "epoch": 2.378026299619024,
      "grad_norm": 0.06758040666062896,
      "learning_rate": 0.00011189474876385227,
      "loss": 0.8518,
      "step": 135450
    },
    {
      "epoch": 2.3782018644990255,
      "grad_norm": 0.059446773083962824,
      "learning_rate": 0.00011188367274554681,
      "loss": 0.8444,
      "step": 135460
    },
    {
      "epoch": 2.378377429379027,
      "grad_norm": 0.05569368865105178,
      "learning_rate": 0.0001118725965893431,
      "loss": 0.8512,
      "step": 135470
    },
    {
      "epoch": 2.3785529942590284,
      "grad_norm": 0.08273207947771312,
      "learning_rate": 0.00011186152029538157,
      "loss": 0.8423,
      "step": 135480
    },
    {
      "epoch": 2.37872855913903,
      "grad_norm": 0.056110009326340936,
      "learning_rate": 0.00011185044386380245,
      "loss": 0.848,
      "step": 135490
    },
    {
      "epoch": 2.3789041240190314,
      "grad_norm": 0.04226877869068197,
      "learning_rate": 0.00011183936729474616,
      "loss": 0.8465,
      "step": 135500
    },
    {
      "epoch": 2.379079688899033,
      "grad_norm": 0.10686022403861829,
      "learning_rate": 0.00011182829058835298,
      "loss": 0.8414,
      "step": 135510
    },
    {
      "epoch": 2.379255253779034,
      "grad_norm": 0.05801992601437741,
      "learning_rate": 0.00011181721374476328,
      "loss": 0.8434,
      "step": 135520
    },
    {
      "epoch": 2.3794308186590354,
      "grad_norm": 0.06683512978949109,
      "learning_rate": 0.00011180613676411743,
      "loss": 0.8428,
      "step": 135530
    },
    {
      "epoch": 2.379606383539037,
      "grad_norm": 0.04794793299597641,
      "learning_rate": 0.00011179505964655577,
      "loss": 0.8475,
      "step": 135540
    },
    {
      "epoch": 2.3797819484190383,
      "grad_norm": 0.045299745619026485,
      "learning_rate": 0.00011178398239221861,
      "loss": 0.8518,
      "step": 135550
    },
    {
      "epoch": 2.37995751329904,
      "grad_norm": 0.045930076065542856,
      "learning_rate": 0.0001117729050012463,
      "loss": 0.8417,
      "step": 135560
    },
    {
      "epoch": 2.380133078179041,
      "grad_norm": 0.058636647618744604,
      "learning_rate": 0.00011176182747377925,
      "loss": 0.8395,
      "step": 135570
    },
    {
      "epoch": 2.3803086430590423,
      "grad_norm": 0.05825808180266637,
      "learning_rate": 0.0001117507498099578,
      "loss": 0.8377,
      "step": 135580
    },
    {
      "epoch": 2.380484207939044,
      "grad_norm": 0.060830254278643664,
      "learning_rate": 0.00011173967200992226,
      "loss": 0.8347,
      "step": 135590
    },
    {
      "epoch": 2.3806597728190453,
      "grad_norm": 0.05488064821773971,
      "learning_rate": 0.00011172859407381304,
      "loss": 0.8405,
      "step": 135600
    },
    {
      "epoch": 2.3808353376990468,
      "grad_norm": 0.06182569303487677,
      "learning_rate": 0.00011171751600177045,
      "loss": 0.8393,
      "step": 135610
    },
    {
      "epoch": 2.3810109025790482,
      "grad_norm": 0.04704812327882085,
      "learning_rate": 0.0001117064377939349,
      "loss": 0.8361,
      "step": 135620
    },
    {
      "epoch": 2.3811864674590497,
      "grad_norm": 0.04768114592647806,
      "learning_rate": 0.00011169535945044674,
      "loss": 0.8432,
      "step": 135630
    },
    {
      "epoch": 2.3813620323390507,
      "grad_norm": 0.062186801231298744,
      "learning_rate": 0.00011168428097144627,
      "loss": 0.8367,
      "step": 135640
    },
    {
      "epoch": 2.3815375972190522,
      "grad_norm": 0.05183491182797227,
      "learning_rate": 0.00011167320235707397,
      "loss": 0.8486,
      "step": 135650
    },
    {
      "epoch": 2.3817131620990537,
      "grad_norm": 0.05074625180052012,
      "learning_rate": 0.00011166212360747015,
      "loss": 0.8458,
      "step": 135660
    },
    {
      "epoch": 2.381888726979055,
      "grad_norm": 0.052918597370422996,
      "learning_rate": 0.00011165104472277518,
      "loss": 0.8411,
      "step": 135670
    },
    {
      "epoch": 2.3820642918590567,
      "grad_norm": 0.05108737378861503,
      "learning_rate": 0.0001116399657031294,
      "loss": 0.8461,
      "step": 135680
    },
    {
      "epoch": 2.3822398567390577,
      "grad_norm": 0.06748402102538224,
      "learning_rate": 0.00011162888654867323,
      "loss": 0.8439,
      "step": 135690
    },
    {
      "epoch": 2.382415421619059,
      "grad_norm": 0.04730311252440297,
      "learning_rate": 0.00011161780725954702,
      "loss": 0.8402,
      "step": 135700
    },
    {
      "epoch": 2.3825909864990606,
      "grad_norm": 0.0549193461447507,
      "learning_rate": 0.0001116067278358912,
      "loss": 0.8391,
      "step": 135710
    },
    {
      "epoch": 2.382766551379062,
      "grad_norm": 0.06783967411406316,
      "learning_rate": 0.00011159564827784608,
      "loss": 0.839,
      "step": 135720
    },
    {
      "epoch": 2.3829421162590636,
      "grad_norm": 0.06506584828169347,
      "learning_rate": 0.00011158456858555203,
      "loss": 0.8452,
      "step": 135730
    },
    {
      "epoch": 2.383117681139065,
      "grad_norm": 0.045019606675359945,
      "learning_rate": 0.00011157348875914948,
      "loss": 0.8471,
      "step": 135740
    },
    {
      "epoch": 2.3832932460190666,
      "grad_norm": 0.06654761031301956,
      "learning_rate": 0.00011156240879877885,
      "loss": 0.8398,
      "step": 135750
    },
    {
      "epoch": 2.3834688108990676,
      "grad_norm": 0.0631949990482006,
      "learning_rate": 0.0001115513287045804,
      "loss": 0.8485,
      "step": 135760
    },
    {
      "epoch": 2.383644375779069,
      "grad_norm": 0.05887973492961591,
      "learning_rate": 0.00011154024847669468,
      "loss": 0.8446,
      "step": 135770
    },
    {
      "epoch": 2.3838199406590705,
      "grad_norm": 0.06352633661408974,
      "learning_rate": 0.00011152916811526195,
      "loss": 0.8465,
      "step": 135780
    },
    {
      "epoch": 2.383995505539072,
      "grad_norm": 0.053464935753489644,
      "learning_rate": 0.00011151808762042262,
      "loss": 0.8491,
      "step": 135790
    },
    {
      "epoch": 2.3841710704190735,
      "grad_norm": 0.045295872018850596,
      "learning_rate": 0.00011150700699231713,
      "loss": 0.8472,
      "step": 135800
    },
    {
      "epoch": 2.3843466352990745,
      "grad_norm": 0.0570254518342167,
      "learning_rate": 0.00011149592623108585,
      "loss": 0.8431,
      "step": 135810
    },
    {
      "epoch": 2.384522200179076,
      "grad_norm": 0.056985715540532965,
      "learning_rate": 0.00011148484533686918,
      "loss": 0.843,
      "step": 135820
    },
    {
      "epoch": 2.3846977650590775,
      "grad_norm": 0.043382556958506735,
      "learning_rate": 0.0001114737643098075,
      "loss": 0.8473,
      "step": 135830
    },
    {
      "epoch": 2.384873329939079,
      "grad_norm": 0.0658055527439204,
      "learning_rate": 0.00011146268315004123,
      "loss": 0.8363,
      "step": 135840
    },
    {
      "epoch": 2.3850488948190804,
      "grad_norm": 0.059145224660092956,
      "learning_rate": 0.00011145160185771071,
      "loss": 0.8435,
      "step": 135850
    },
    {
      "epoch": 2.385224459699082,
      "grad_norm": 0.05485898769627161,
      "learning_rate": 0.00011144052043295643,
      "loss": 0.8442,
      "step": 135860
    },
    {
      "epoch": 2.3854000245790834,
      "grad_norm": 0.049091506410229184,
      "learning_rate": 0.00011142943887591877,
      "loss": 0.8484,
      "step": 135870
    },
    {
      "epoch": 2.3855755894590844,
      "grad_norm": 0.051303940117973915,
      "learning_rate": 0.00011141835718673812,
      "loss": 0.8487,
      "step": 135880
    },
    {
      "epoch": 2.385751154339086,
      "grad_norm": 0.06542643507614472,
      "learning_rate": 0.00011140727536555486,
      "loss": 0.8506,
      "step": 135890
    },
    {
      "epoch": 2.3859267192190874,
      "grad_norm": 0.053485592418217466,
      "learning_rate": 0.00011139619341250943,
      "loss": 0.8445,
      "step": 135900
    },
    {
      "epoch": 2.386102284099089,
      "grad_norm": 0.058424231912986,
      "learning_rate": 0.00011138511132774224,
      "loss": 0.8483,
      "step": 135910
    },
    {
      "epoch": 2.3862778489790903,
      "grad_norm": 0.053669615526938524,
      "learning_rate": 0.0001113740291113937,
      "loss": 0.8452,
      "step": 135920
    },
    {
      "epoch": 2.386453413859092,
      "grad_norm": 0.04823612403572056,
      "learning_rate": 0.00011136294676360421,
      "loss": 0.8461,
      "step": 135930
    },
    {
      "epoch": 2.386628978739093,
      "grad_norm": 0.04823412421703006,
      "learning_rate": 0.00011135186428451425,
      "loss": 0.8423,
      "step": 135940
    },
    {
      "epoch": 2.3868045436190943,
      "grad_norm": 0.04271457906703056,
      "learning_rate": 0.00011134078167426414,
      "loss": 0.8517,
      "step": 135950
    },
    {
      "epoch": 2.386980108499096,
      "grad_norm": 0.07619285718305875,
      "learning_rate": 0.00011132969893299434,
      "loss": 0.8392,
      "step": 135960
    },
    {
      "epoch": 2.3871556733790973,
      "grad_norm": 0.0553269336021089,
      "learning_rate": 0.0001113186160608453,
      "loss": 0.8503,
      "step": 135970
    },
    {
      "epoch": 2.3873312382590988,
      "grad_norm": 0.044262629778197775,
      "learning_rate": 0.0001113075330579574,
      "loss": 0.8413,
      "step": 135980
    },
    {
      "epoch": 2.3875068031391002,
      "grad_norm": 0.06209766054214853,
      "learning_rate": 0.00011129644992447108,
      "loss": 0.8404,
      "step": 135990
    },
    {
      "epoch": 2.3876823680191013,
      "grad_norm": 0.059659781265969734,
      "learning_rate": 0.0001112853666605268,
      "loss": 0.853,
      "step": 136000
    },
    {
      "epoch": 2.3878579328991028,
      "grad_norm": 0.07467084666959009,
      "learning_rate": 0.00011127428326626493,
      "loss": 0.8433,
      "step": 136010
    },
    {
      "epoch": 2.3880334977791042,
      "grad_norm": 0.04663758757175809,
      "learning_rate": 0.00011126319974182593,
      "loss": 0.8481,
      "step": 136020
    },
    {
      "epoch": 2.3882090626591057,
      "grad_norm": 0.052008819930882945,
      "learning_rate": 0.00011125211608735018,
      "loss": 0.8379,
      "step": 136030
    },
    {
      "epoch": 2.388384627539107,
      "grad_norm": 0.05452832214030298,
      "learning_rate": 0.00011124103230297818,
      "loss": 0.8384,
      "step": 136040
    },
    {
      "epoch": 2.3885601924191087,
      "grad_norm": 0.0448278083790518,
      "learning_rate": 0.00011122994838885036,
      "loss": 0.8456,
      "step": 136050
    },
    {
      "epoch": 2.3887357572991097,
      "grad_norm": 0.05311398625917719,
      "learning_rate": 0.00011121886434510713,
      "loss": 0.8386,
      "step": 136060
    },
    {
      "epoch": 2.388911322179111,
      "grad_norm": 0.06192246025170192,
      "learning_rate": 0.0001112077801718889,
      "loss": 0.8437,
      "step": 136070
    },
    {
      "epoch": 2.3890868870591127,
      "grad_norm": 0.051960664908592885,
      "learning_rate": 0.00011119669586933616,
      "loss": 0.8475,
      "step": 136080
    },
    {
      "epoch": 2.389262451939114,
      "grad_norm": 0.06887317870558848,
      "learning_rate": 0.00011118561143758932,
      "loss": 0.8545,
      "step": 136090
    },
    {
      "epoch": 2.3894380168191156,
      "grad_norm": 0.044313156588725824,
      "learning_rate": 0.00011117452687678882,
      "loss": 0.851,
      "step": 136100
    },
    {
      "epoch": 2.389613581699117,
      "grad_norm": 0.060814525457578116,
      "learning_rate": 0.00011116344218707517,
      "loss": 0.8474,
      "step": 136110
    },
    {
      "epoch": 2.389789146579118,
      "grad_norm": 0.059579182743328976,
      "learning_rate": 0.00011115235736858871,
      "loss": 0.8345,
      "step": 136120
    },
    {
      "epoch": 2.3899647114591196,
      "grad_norm": 0.04434285349498977,
      "learning_rate": 0.00011114127242146994,
      "loss": 0.8486,
      "step": 136130
    },
    {
      "epoch": 2.390140276339121,
      "grad_norm": 0.07230766594797622,
      "learning_rate": 0.00011113018734585931,
      "loss": 0.8511,
      "step": 136140
    },
    {
      "epoch": 2.3903158412191226,
      "grad_norm": 0.05463536055933053,
      "learning_rate": 0.00011111910214189727,
      "loss": 0.8506,
      "step": 136150
    },
    {
      "epoch": 2.390491406099124,
      "grad_norm": 0.07738893505539414,
      "learning_rate": 0.00011110801680972427,
      "loss": 0.8403,
      "step": 136160
    },
    {
      "epoch": 2.3906669709791255,
      "grad_norm": 0.07178219212596966,
      "learning_rate": 0.00011109693134948075,
      "loss": 0.8433,
      "step": 136170
    },
    {
      "epoch": 2.390842535859127,
      "grad_norm": 0.0635399788298119,
      "learning_rate": 0.00011108584576130718,
      "loss": 0.8476,
      "step": 136180
    },
    {
      "epoch": 2.391018100739128,
      "grad_norm": 0.06487786321358693,
      "learning_rate": 0.000111074760045344,
      "loss": 0.8382,
      "step": 136190
    },
    {
      "epoch": 2.3911936656191295,
      "grad_norm": 0.06584051477250506,
      "learning_rate": 0.00011106367420173166,
      "loss": 0.8413,
      "step": 136200
    },
    {
      "epoch": 2.391369230499131,
      "grad_norm": 0.08865838107384084,
      "learning_rate": 0.00011105258823061068,
      "loss": 0.8425,
      "step": 136210
    },
    {
      "epoch": 2.3915447953791324,
      "grad_norm": 0.05771633330524254,
      "learning_rate": 0.00011104150213212148,
      "loss": 0.84,
      "step": 136220
    },
    {
      "epoch": 2.391720360259134,
      "grad_norm": 0.04834714508519739,
      "learning_rate": 0.00011103041590640454,
      "loss": 0.8421,
      "step": 136230
    },
    {
      "epoch": 2.391895925139135,
      "grad_norm": 0.05836448116206925,
      "learning_rate": 0.00011101932955360026,
      "loss": 0.8503,
      "step": 136240
    },
    {
      "epoch": 2.3920714900191364,
      "grad_norm": 0.06738060086953708,
      "learning_rate": 0.00011100824307384917,
      "loss": 0.8372,
      "step": 136250
    },
    {
      "epoch": 2.392247054899138,
      "grad_norm": 0.04850017174378084,
      "learning_rate": 0.00011099715646729174,
      "loss": 0.8434,
      "step": 136260
    },
    {
      "epoch": 2.3924226197791394,
      "grad_norm": 0.052590387894497596,
      "learning_rate": 0.00011098606973406837,
      "loss": 0.8451,
      "step": 136270
    },
    {
      "epoch": 2.392598184659141,
      "grad_norm": 0.05616939755149267,
      "learning_rate": 0.00011097498287431966,
      "loss": 0.8412,
      "step": 136280
    },
    {
      "epoch": 2.3927737495391423,
      "grad_norm": 0.052127272722998705,
      "learning_rate": 0.00011096389588818597,
      "loss": 0.848,
      "step": 136290
    },
    {
      "epoch": 2.392949314419144,
      "grad_norm": 0.050196173911595164,
      "learning_rate": 0.00011095280877580778,
      "loss": 0.838,
      "step": 136300
    },
    {
      "epoch": 2.393124879299145,
      "grad_norm": 0.049077072351219775,
      "learning_rate": 0.00011094172153732565,
      "loss": 0.8453,
      "step": 136310
    },
    {
      "epoch": 2.3933004441791463,
      "grad_norm": 0.04655105464177969,
      "learning_rate": 0.00011093063417287994,
      "loss": 0.8477,
      "step": 136320
    },
    {
      "epoch": 2.393476009059148,
      "grad_norm": 0.0550153746685437,
      "learning_rate": 0.00011091954668261125,
      "loss": 0.8439,
      "step": 136330
    },
    {
      "epoch": 2.3936515739391493,
      "grad_norm": 0.06542052059717275,
      "learning_rate": 0.00011090845906665999,
      "loss": 0.8405,
      "step": 136340
    },
    {
      "epoch": 2.3938271388191508,
      "grad_norm": 0.042702515160168514,
      "learning_rate": 0.00011089737132516664,
      "loss": 0.8431,
      "step": 136350
    },
    {
      "epoch": 2.394002703699152,
      "grad_norm": 0.056007794907006254,
      "learning_rate": 0.00011088628345827167,
      "loss": 0.8496,
      "step": 136360
    },
    {
      "epoch": 2.3941782685791533,
      "grad_norm": 0.05551365976452953,
      "learning_rate": 0.00011087519546611565,
      "loss": 0.8398,
      "step": 136370
    },
    {
      "epoch": 2.3943538334591548,
      "grad_norm": 0.057603942189245784,
      "learning_rate": 0.00011086410734883898,
      "loss": 0.8448,
      "step": 136380
    },
    {
      "epoch": 2.3945293983391562,
      "grad_norm": 0.06844730949506629,
      "learning_rate": 0.00011085301910658219,
      "loss": 0.8438,
      "step": 136390
    },
    {
      "epoch": 2.3947049632191577,
      "grad_norm": 0.057142422192387195,
      "learning_rate": 0.00011084193073948577,
      "loss": 0.8457,
      "step": 136400
    },
    {
      "epoch": 2.394880528099159,
      "grad_norm": 0.07681417903834677,
      "learning_rate": 0.00011083084224769016,
      "loss": 0.8454,
      "step": 136410
    },
    {
      "epoch": 2.3950560929791607,
      "grad_norm": 0.055013037926149415,
      "learning_rate": 0.00011081975363133592,
      "loss": 0.8531,
      "step": 136420
    },
    {
      "epoch": 2.3952316578591617,
      "grad_norm": 0.04655535647637035,
      "learning_rate": 0.00011080866489056355,
      "loss": 0.8456,
      "step": 136430
    },
    {
      "epoch": 2.395407222739163,
      "grad_norm": 0.05021056972942238,
      "learning_rate": 0.00011079757602551344,
      "loss": 0.8415,
      "step": 136440
    },
    {
      "epoch": 2.3955827876191647,
      "grad_norm": 0.057315173659868116,
      "learning_rate": 0.00011078648703632623,
      "loss": 0.8431,
      "step": 136450
    },
    {
      "epoch": 2.395758352499166,
      "grad_norm": 0.05382027167189947,
      "learning_rate": 0.00011077539792314234,
      "loss": 0.8457,
      "step": 136460
    },
    {
      "epoch": 2.3959339173791676,
      "grad_norm": 0.061850942873844406,
      "learning_rate": 0.00011076430868610229,
      "loss": 0.842,
      "step": 136470
    },
    {
      "epoch": 2.3961094822591686,
      "grad_norm": 0.06566839080598091,
      "learning_rate": 0.00011075321932534654,
      "loss": 0.8482,
      "step": 136480
    },
    {
      "epoch": 2.39628504713917,
      "grad_norm": 0.06967714841597666,
      "learning_rate": 0.00011074212984101566,
      "loss": 0.8462,
      "step": 136490
    },
    {
      "epoch": 2.3964606120191716,
      "grad_norm": 0.04730992084410836,
      "learning_rate": 0.00011073104023325014,
      "loss": 0.84,
      "step": 136500
    },
    {
      "epoch": 2.396636176899173,
      "grad_norm": 0.06655770404838825,
      "learning_rate": 0.00011071995050219049,
      "loss": 0.8481,
      "step": 136510
    },
    {
      "epoch": 2.3968117417791746,
      "grad_norm": 0.05233028794403445,
      "learning_rate": 0.00011070886064797716,
      "loss": 0.8446,
      "step": 136520
    },
    {
      "epoch": 2.396987306659176,
      "grad_norm": 0.049101676228709146,
      "learning_rate": 0.00011069777067075071,
      "loss": 0.8444,
      "step": 136530
    },
    {
      "epoch": 2.3971628715391775,
      "grad_norm": 0.05727689898736926,
      "learning_rate": 0.00011068668057065164,
      "loss": 0.8436,
      "step": 136540
    },
    {
      "epoch": 2.3973384364191785,
      "grad_norm": 0.05926658296925544,
      "learning_rate": 0.00011067559034782051,
      "loss": 0.8461,
      "step": 136550
    },
    {
      "epoch": 2.39751400129918,
      "grad_norm": 0.052479974373013355,
      "learning_rate": 0.00011066450000239778,
      "loss": 0.8403,
      "step": 136560
    },
    {
      "epoch": 2.3976895661791815,
      "grad_norm": 0.05070836701874276,
      "learning_rate": 0.000110653409534524,
      "loss": 0.8536,
      "step": 136570
    },
    {
      "epoch": 2.397865131059183,
      "grad_norm": 0.047357121372627334,
      "learning_rate": 0.00011064231894433962,
      "loss": 0.8421,
      "step": 136580
    },
    {
      "epoch": 2.3980406959391845,
      "grad_norm": 0.05650538826352159,
      "learning_rate": 0.00011063122823198524,
      "loss": 0.8525,
      "step": 136590
    },
    {
      "epoch": 2.3982162608191855,
      "grad_norm": 0.05562685215665266,
      "learning_rate": 0.00011062013739760135,
      "loss": 0.8467,
      "step": 136600
    },
    {
      "epoch": 2.398391825699187,
      "grad_norm": 0.04480081302900883,
      "learning_rate": 0.00011060904644132846,
      "loss": 0.8424,
      "step": 136610
    },
    {
      "epoch": 2.3985673905791884,
      "grad_norm": 0.05459924163804141,
      "learning_rate": 0.00011059795536330717,
      "loss": 0.8426,
      "step": 136620
    },
    {
      "epoch": 2.39874295545919,
      "grad_norm": 0.04906143111538187,
      "learning_rate": 0.0001105868641636779,
      "loss": 0.8419,
      "step": 136630
    },
    {
      "epoch": 2.3989185203391914,
      "grad_norm": 0.06382727159224229,
      "learning_rate": 0.00011057577284258121,
      "loss": 0.8437,
      "step": 136640
    },
    {
      "epoch": 2.399094085219193,
      "grad_norm": 0.08068809661955471,
      "learning_rate": 0.00011056468140015767,
      "loss": 0.8451,
      "step": 136650
    },
    {
      "epoch": 2.3992696500991944,
      "grad_norm": 0.06978202492305034,
      "learning_rate": 0.00011055358983654775,
      "loss": 0.8428,
      "step": 136660
    },
    {
      "epoch": 2.3994452149791954,
      "grad_norm": 0.0510512904117097,
      "learning_rate": 0.00011054249815189204,
      "loss": 0.8381,
      "step": 136670
    },
    {
      "epoch": 2.399620779859197,
      "grad_norm": 0.06641707652888178,
      "learning_rate": 0.00011053140634633106,
      "loss": 0.8464,
      "step": 136680
    },
    {
      "epoch": 2.3997963447391983,
      "grad_norm": 0.058825084289296774,
      "learning_rate": 0.00011052031442000535,
      "loss": 0.8384,
      "step": 136690
    },
    {
      "epoch": 2.3999719096192,
      "grad_norm": 0.04751182454661389,
      "learning_rate": 0.00011050922237305539,
      "loss": 0.8499,
      "step": 136700
    },
    {
      "epoch": 2.4001474744992013,
      "grad_norm": 0.06659850094424706,
      "learning_rate": 0.00011049813020562177,
      "loss": 0.8462,
      "step": 136710
    },
    {
      "epoch": 2.4003230393792028,
      "grad_norm": 0.06361596328010449,
      "learning_rate": 0.00011048703791784503,
      "loss": 0.8456,
      "step": 136720
    },
    {
      "epoch": 2.400498604259204,
      "grad_norm": 0.05486384844260283,
      "learning_rate": 0.00011047594550986569,
      "loss": 0.8447,
      "step": 136730
    },
    {
      "epoch": 2.4006741691392053,
      "grad_norm": 0.05158121708528834,
      "learning_rate": 0.00011046485298182434,
      "loss": 0.8477,
      "step": 136740
    },
    {
      "epoch": 2.4008497340192068,
      "grad_norm": 0.05135555383065123,
      "learning_rate": 0.00011045376033386143,
      "loss": 0.8487,
      "step": 136750
    },
    {
      "epoch": 2.4010252988992082,
      "grad_norm": 0.06497606685303159,
      "learning_rate": 0.00011044266756611759,
      "loss": 0.8455,
      "step": 136760
    },
    {
      "epoch": 2.4012008637792097,
      "grad_norm": 0.060536165795421475,
      "learning_rate": 0.00011043157467873335,
      "loss": 0.8455,
      "step": 136770
    },
    {
      "epoch": 2.401376428659211,
      "grad_norm": 0.0625986700016328,
      "learning_rate": 0.00011042048167184925,
      "loss": 0.8492,
      "step": 136780
    },
    {
      "epoch": 2.4015519935392122,
      "grad_norm": 0.057870068424229155,
      "learning_rate": 0.00011040938854560585,
      "loss": 0.8462,
      "step": 136790
    },
    {
      "epoch": 2.4017275584192137,
      "grad_norm": 0.05418768954727669,
      "learning_rate": 0.0001103982953001437,
      "loss": 0.8471,
      "step": 136800
    },
    {
      "epoch": 2.401903123299215,
      "grad_norm": 0.06929949734898712,
      "learning_rate": 0.00011038720193560333,
      "loss": 0.8445,
      "step": 136810
    },
    {
      "epoch": 2.4020786881792167,
      "grad_norm": 0.05682090029782749,
      "learning_rate": 0.00011037610845212529,
      "loss": 0.8443,
      "step": 136820
    },
    {
      "epoch": 2.402254253059218,
      "grad_norm": 0.0856499150002833,
      "learning_rate": 0.0001103650148498502,
      "loss": 0.843,
      "step": 136830
    },
    {
      "epoch": 2.4024298179392196,
      "grad_norm": 0.05161714846057119,
      "learning_rate": 0.00011035392112891856,
      "loss": 0.8421,
      "step": 136840
    },
    {
      "epoch": 2.402605382819221,
      "grad_norm": 0.05586956691972077,
      "learning_rate": 0.00011034282728947095,
      "loss": 0.8388,
      "step": 136850
    },
    {
      "epoch": 2.402780947699222,
      "grad_norm": 0.04606846314487125,
      "learning_rate": 0.00011033173333164793,
      "loss": 0.8412,
      "step": 136860
    },
    {
      "epoch": 2.4029565125792236,
      "grad_norm": 0.04792118772978804,
      "learning_rate": 0.00011032063925559005,
      "loss": 0.8432,
      "step": 136870
    },
    {
      "epoch": 2.403132077459225,
      "grad_norm": 0.05059961272350703,
      "learning_rate": 0.00011030954506143787,
      "loss": 0.8373,
      "step": 136880
    },
    {
      "epoch": 2.4033076423392266,
      "grad_norm": 0.06662947613853239,
      "learning_rate": 0.000110298450749332,
      "loss": 0.8482,
      "step": 136890
    },
    {
      "epoch": 2.403483207219228,
      "grad_norm": 0.055967622939257604,
      "learning_rate": 0.00011028735631941298,
      "loss": 0.8401,
      "step": 136900
    },
    {
      "epoch": 2.403658772099229,
      "grad_norm": 0.06100004621575844,
      "learning_rate": 0.00011027626177182135,
      "loss": 0.8397,
      "step": 136910
    },
    {
      "epoch": 2.4038343369792305,
      "grad_norm": 0.04852207301499156,
      "learning_rate": 0.00011026516710669772,
      "loss": 0.837,
      "step": 136920
    },
    {
      "epoch": 2.404009901859232,
      "grad_norm": 0.050956225089374495,
      "learning_rate": 0.00011025407232418263,
      "loss": 0.8427,
      "step": 136930
    },
    {
      "epoch": 2.4041854667392335,
      "grad_norm": 0.05102746613803304,
      "learning_rate": 0.00011024297742441669,
      "loss": 0.8535,
      "step": 136940
    },
    {
      "epoch": 2.404361031619235,
      "grad_norm": 0.06290671038040319,
      "learning_rate": 0.00011023188240754045,
      "loss": 0.8421,
      "step": 136950
    },
    {
      "epoch": 2.4045365964992365,
      "grad_norm": 0.07390847516694621,
      "learning_rate": 0.00011022078727369452,
      "loss": 0.8494,
      "step": 136960
    },
    {
      "epoch": 2.404712161379238,
      "grad_norm": 0.06913948286751802,
      "learning_rate": 0.00011020969202301939,
      "loss": 0.8406,
      "step": 136970
    },
    {
      "epoch": 2.404887726259239,
      "grad_norm": 0.049642557094404675,
      "learning_rate": 0.0001101985966556557,
      "loss": 0.8326,
      "step": 136980
    },
    {
      "epoch": 2.4050632911392404,
      "grad_norm": 0.04764981681435323,
      "learning_rate": 0.00011018750117174407,
      "loss": 0.8456,
      "step": 136990
    },
    {
      "epoch": 2.405238856019242,
      "grad_norm": 0.06158291288151327,
      "learning_rate": 0.00011017640557142503,
      "loss": 0.8417,
      "step": 137000
    },
    {
      "epoch": 2.4054144208992434,
      "grad_norm": 0.0583774222476529,
      "learning_rate": 0.00011016530985483917,
      "loss": 0.8401,
      "step": 137010
    },
    {
      "epoch": 2.405589985779245,
      "grad_norm": 0.05400741413797872,
      "learning_rate": 0.00011015421402212709,
      "loss": 0.8469,
      "step": 137020
    },
    {
      "epoch": 2.405765550659246,
      "grad_norm": 0.06169430776024137,
      "learning_rate": 0.00011014311807342934,
      "loss": 0.8431,
      "step": 137030
    },
    {
      "epoch": 2.4059411155392474,
      "grad_norm": 0.05558256825288102,
      "learning_rate": 0.00011013202200888653,
      "loss": 0.8416,
      "step": 137040
    },
    {
      "epoch": 2.406116680419249,
      "grad_norm": 0.06164951664257011,
      "learning_rate": 0.00011012092582863924,
      "loss": 0.8433,
      "step": 137050
    },
    {
      "epoch": 2.4062922452992503,
      "grad_norm": 0.05740389578206979,
      "learning_rate": 0.00011010982953282809,
      "loss": 0.848,
      "step": 137060
    },
    {
      "epoch": 2.406467810179252,
      "grad_norm": 0.08114196919274737,
      "learning_rate": 0.0001100987331215937,
      "loss": 0.8404,
      "step": 137070
    },
    {
      "epoch": 2.4066433750592533,
      "grad_norm": 0.04513810983746019,
      "learning_rate": 0.00011008763659507654,
      "loss": 0.8491,
      "step": 137080
    },
    {
      "epoch": 2.4068189399392548,
      "grad_norm": 0.04909311888729204,
      "learning_rate": 0.00011007653995341731,
      "loss": 0.8514,
      "step": 137090
    },
    {
      "epoch": 2.406994504819256,
      "grad_norm": 0.053037940150736464,
      "learning_rate": 0.00011006544319675658,
      "loss": 0.8466,
      "step": 137100
    },
    {
      "epoch": 2.4071700696992573,
      "grad_norm": 0.06908631284091632,
      "learning_rate": 0.00011005434632523497,
      "loss": 0.8396,
      "step": 137110
    },
    {
      "epoch": 2.4073456345792588,
      "grad_norm": 0.05029290765132417,
      "learning_rate": 0.00011004324933899303,
      "loss": 0.8371,
      "step": 137120
    },
    {
      "epoch": 2.4075211994592602,
      "grad_norm": 0.059560613012841816,
      "learning_rate": 0.0001100321522381714,
      "loss": 0.8418,
      "step": 137130
    },
    {
      "epoch": 2.4076967643392617,
      "grad_norm": 0.08025014727274779,
      "learning_rate": 0.00011002105502291066,
      "loss": 0.8437,
      "step": 137140
    },
    {
      "epoch": 2.4078723292192628,
      "grad_norm": 0.050290833733350054,
      "learning_rate": 0.00011000995769335143,
      "loss": 0.8527,
      "step": 137150
    },
    {
      "epoch": 2.4080478940992642,
      "grad_norm": 0.0657409471740963,
      "learning_rate": 0.0001099988602496343,
      "loss": 0.836,
      "step": 137160
    },
    {
      "epoch": 2.4082234589792657,
      "grad_norm": 0.057643231650381994,
      "learning_rate": 0.0001099877626918999,
      "loss": 0.8474,
      "step": 137170
    },
    {
      "epoch": 2.408399023859267,
      "grad_norm": 0.06394960148228845,
      "learning_rate": 0.00010997666502028883,
      "loss": 0.8432,
      "step": 137180
    },
    {
      "epoch": 2.4085745887392687,
      "grad_norm": 0.08028610732149835,
      "learning_rate": 0.00010996556723494167,
      "loss": 0.8463,
      "step": 137190
    },
    {
      "epoch": 2.40875015361927,
      "grad_norm": 0.06624523070738446,
      "learning_rate": 0.00010995446933599911,
      "loss": 0.8479,
      "step": 137200
    },
    {
      "epoch": 2.4089257184992716,
      "grad_norm": 0.06170650310212568,
      "learning_rate": 0.00010994337132360164,
      "loss": 0.844,
      "step": 137210
    },
    {
      "epoch": 2.4091012833792727,
      "grad_norm": 0.056590215936247516,
      "learning_rate": 0.00010993227319788995,
      "loss": 0.8536,
      "step": 137220
    },
    {
      "epoch": 2.409276848259274,
      "grad_norm": 0.051927689179226515,
      "learning_rate": 0.00010992117495900467,
      "loss": 0.8523,
      "step": 137230
    },
    {
      "epoch": 2.4094524131392756,
      "grad_norm": 0.06743026950594895,
      "learning_rate": 0.0001099100766070864,
      "loss": 0.8429,
      "step": 137240
    },
    {
      "epoch": 2.409627978019277,
      "grad_norm": 0.06240069741760646,
      "learning_rate": 0.00010989897814227574,
      "loss": 0.8503,
      "step": 137250
    },
    {
      "epoch": 2.4098035428992786,
      "grad_norm": 0.05026155525266967,
      "learning_rate": 0.00010988787956471331,
      "loss": 0.8404,
      "step": 137260
    },
    {
      "epoch": 2.4099791077792796,
      "grad_norm": 0.05382389672092984,
      "learning_rate": 0.00010987678087453974,
      "loss": 0.8353,
      "step": 137270
    },
    {
      "epoch": 2.410154672659281,
      "grad_norm": 0.08530975365628353,
      "learning_rate": 0.00010986568207189567,
      "loss": 0.8408,
      "step": 137280
    },
    {
      "epoch": 2.4103302375392825,
      "grad_norm": 0.05552527509181815,
      "learning_rate": 0.00010985458315692171,
      "loss": 0.8417,
      "step": 137290
    },
    {
      "epoch": 2.410505802419284,
      "grad_norm": 0.06217909365328076,
      "learning_rate": 0.00010984348412975854,
      "loss": 0.8421,
      "step": 137300
    },
    {
      "epoch": 2.4106813672992855,
      "grad_norm": 0.0716807151383372,
      "learning_rate": 0.00010983238499054665,
      "loss": 0.8428,
      "step": 137310
    },
    {
      "epoch": 2.410856932179287,
      "grad_norm": 0.055183387015115014,
      "learning_rate": 0.00010982128573942674,
      "loss": 0.8493,
      "step": 137320
    },
    {
      "epoch": 2.4110324970592885,
      "grad_norm": 0.07877735990619243,
      "learning_rate": 0.00010981018637653948,
      "loss": 0.8495,
      "step": 137330
    },
    {
      "epoch": 2.4112080619392895,
      "grad_norm": 0.05442651164128158,
      "learning_rate": 0.00010979908690202548,
      "loss": 0.8467,
      "step": 137340
    },
    {
      "epoch": 2.411383626819291,
      "grad_norm": 0.049765836891351105,
      "learning_rate": 0.00010978798731602535,
      "loss": 0.8478,
      "step": 137350
    },
    {
      "epoch": 2.4115591916992924,
      "grad_norm": 0.06499969749807086,
      "learning_rate": 0.00010977688761867974,
      "loss": 0.8418,
      "step": 137360
    },
    {
      "epoch": 2.411734756579294,
      "grad_norm": 0.05315738955496484,
      "learning_rate": 0.00010976578781012931,
      "loss": 0.8446,
      "step": 137370
    },
    {
      "epoch": 2.4119103214592954,
      "grad_norm": 0.0683135937470344,
      "learning_rate": 0.00010975468789051462,
      "loss": 0.8467,
      "step": 137380
    },
    {
      "epoch": 2.412085886339297,
      "grad_norm": 0.09313826373203969,
      "learning_rate": 0.00010974358785997636,
      "loss": 0.8424,
      "step": 137390
    },
    {
      "epoch": 2.412261451219298,
      "grad_norm": 0.045207812227350215,
      "learning_rate": 0.0001097324877186552,
      "loss": 0.8534,
      "step": 137400
    },
    {
      "epoch": 2.4124370160992994,
      "grad_norm": 0.04863290880923658,
      "learning_rate": 0.00010972138746669176,
      "loss": 0.8451,
      "step": 137410
    },
    {
      "epoch": 2.412612580979301,
      "grad_norm": 0.05623480549713209,
      "learning_rate": 0.00010971028710422664,
      "loss": 0.8476,
      "step": 137420
    },
    {
      "epoch": 2.4127881458593023,
      "grad_norm": 0.0813771175110812,
      "learning_rate": 0.00010969918663140051,
      "loss": 0.8412,
      "step": 137430
    },
    {
      "epoch": 2.412963710739304,
      "grad_norm": 0.0594118739339428,
      "learning_rate": 0.00010968808604835399,
      "loss": 0.8513,
      "step": 137440
    },
    {
      "epoch": 2.4131392756193053,
      "grad_norm": 0.055622568287697856,
      "learning_rate": 0.00010967698535522779,
      "loss": 0.8502,
      "step": 137450
    },
    {
      "epoch": 2.4133148404993063,
      "grad_norm": 0.0706794106239812,
      "learning_rate": 0.00010966588455216253,
      "loss": 0.8471,
      "step": 137460
    },
    {
      "epoch": 2.413490405379308,
      "grad_norm": 0.06544628029330891,
      "learning_rate": 0.00010965478363929887,
      "loss": 0.8488,
      "step": 137470
    },
    {
      "epoch": 2.4136659702593093,
      "grad_norm": 0.052368819078921346,
      "learning_rate": 0.0001096436826167774,
      "loss": 0.8424,
      "step": 137480
    },
    {
      "epoch": 2.4138415351393108,
      "grad_norm": 0.055819658827603494,
      "learning_rate": 0.0001096325814847388,
      "loss": 0.8475,
      "step": 137490
    },
    {
      "epoch": 2.4140171000193122,
      "grad_norm": 0.04618167160977969,
      "learning_rate": 0.00010962148024332377,
      "loss": 0.8436,
      "step": 137500
    },
    {
      "epoch": 2.4141926648993137,
      "grad_norm": 0.05332779842155563,
      "learning_rate": 0.00010961037889267292,
      "loss": 0.8463,
      "step": 137510
    },
    {
      "epoch": 2.4143682297793148,
      "grad_norm": 0.05099415388408288,
      "learning_rate": 0.00010959927743292692,
      "loss": 0.8474,
      "step": 137520
    },
    {
      "epoch": 2.4145437946593162,
      "grad_norm": 0.047012871928420664,
      "learning_rate": 0.00010958817586422647,
      "loss": 0.8398,
      "step": 137530
    },
    {
      "epoch": 2.4147193595393177,
      "grad_norm": 0.05631453885804718,
      "learning_rate": 0.00010957707418671215,
      "loss": 0.8395,
      "step": 137540
    },
    {
      "epoch": 2.414894924419319,
      "grad_norm": 0.05167799598608747,
      "learning_rate": 0.00010956597240052467,
      "loss": 0.8456,
      "step": 137550
    },
    {
      "epoch": 2.4150704892993207,
      "grad_norm": 0.053885540220461464,
      "learning_rate": 0.00010955487050580462,
      "loss": 0.8493,
      "step": 137560
    },
    {
      "epoch": 2.415246054179322,
      "grad_norm": 0.06935714929004408,
      "learning_rate": 0.00010954376850269279,
      "loss": 0.8395,
      "step": 137570
    },
    {
      "epoch": 2.415421619059323,
      "grad_norm": 0.04899185490967768,
      "learning_rate": 0.00010953266639132978,
      "loss": 0.8423,
      "step": 137580
    },
    {
      "epoch": 2.4155971839393247,
      "grad_norm": 0.06141146262070851,
      "learning_rate": 0.00010952156417185619,
      "loss": 0.8442,
      "step": 137590
    },
    {
      "epoch": 2.415772748819326,
      "grad_norm": 0.043345794454243025,
      "learning_rate": 0.00010951046184441278,
      "loss": 0.8493,
      "step": 137600
    },
    {
      "epoch": 2.4159483136993276,
      "grad_norm": 0.06558421885374151,
      "learning_rate": 0.0001094993594091402,
      "loss": 0.8364,
      "step": 137610
    },
    {
      "epoch": 2.416123878579329,
      "grad_norm": 0.04141281539465908,
      "learning_rate": 0.00010948825686617909,
      "loss": 0.8432,
      "step": 137620
    },
    {
      "epoch": 2.4162994434593306,
      "grad_norm": 0.06430256442221606,
      "learning_rate": 0.00010947715421567015,
      "loss": 0.8393,
      "step": 137630
    },
    {
      "epoch": 2.416475008339332,
      "grad_norm": 0.061156441247239096,
      "learning_rate": 0.00010946605145775406,
      "loss": 0.842,
      "step": 137640
    },
    {
      "epoch": 2.416650573219333,
      "grad_norm": 0.06514012065562758,
      "learning_rate": 0.00010945494859257144,
      "loss": 0.8434,
      "step": 137650
    },
    {
      "epoch": 2.4168261380993346,
      "grad_norm": 0.05191169820637229,
      "learning_rate": 0.00010944384562026302,
      "loss": 0.8375,
      "step": 137660
    },
    {
      "epoch": 2.417001702979336,
      "grad_norm": 0.0589274666194637,
      "learning_rate": 0.00010943274254096942,
      "loss": 0.8394,
      "step": 137670
    },
    {
      "epoch": 2.4171772678593375,
      "grad_norm": 0.05333641059194803,
      "learning_rate": 0.00010942163935483139,
      "loss": 0.8437,
      "step": 137680
    },
    {
      "epoch": 2.417352832739339,
      "grad_norm": 0.056984507489653814,
      "learning_rate": 0.00010941053606198957,
      "loss": 0.8376,
      "step": 137690
    },
    {
      "epoch": 2.41752839761934,
      "grad_norm": 0.04220848481465628,
      "learning_rate": 0.00010939943266258467,
      "loss": 0.8493,
      "step": 137700
    },
    {
      "epoch": 2.4177039624993415,
      "grad_norm": 0.052367185771480106,
      "learning_rate": 0.00010938832915675734,
      "loss": 0.8419,
      "step": 137710
    },
    {
      "epoch": 2.417879527379343,
      "grad_norm": 0.0624269061465369,
      "learning_rate": 0.00010937722554464825,
      "loss": 0.8443,
      "step": 137720
    },
    {
      "epoch": 2.4180550922593445,
      "grad_norm": 0.04436506430294018,
      "learning_rate": 0.0001093661218263981,
      "loss": 0.8401,
      "step": 137730
    },
    {
      "epoch": 2.418230657139346,
      "grad_norm": 0.048776321787250936,
      "learning_rate": 0.0001093550180021476,
      "loss": 0.8408,
      "step": 137740
    },
    {
      "epoch": 2.4184062220193474,
      "grad_norm": 0.0598307749118762,
      "learning_rate": 0.00010934391407203743,
      "loss": 0.8439,
      "step": 137750
    },
    {
      "epoch": 2.418581786899349,
      "grad_norm": 0.05092083486211644,
      "learning_rate": 0.00010933281003620828,
      "loss": 0.8504,
      "step": 137760
    },
    {
      "epoch": 2.41875735177935,
      "grad_norm": 0.05297384948193832,
      "learning_rate": 0.0001093217058948008,
      "loss": 0.8493,
      "step": 137770
    },
    {
      "epoch": 2.4189329166593514,
      "grad_norm": 0.07823944355204755,
      "learning_rate": 0.00010931060164795568,
      "loss": 0.8433,
      "step": 137780
    },
    {
      "epoch": 2.419108481539353,
      "grad_norm": 0.058034857688149005,
      "learning_rate": 0.00010929949729581373,
      "loss": 0.8468,
      "step": 137790
    },
    {
      "epoch": 2.4192840464193544,
      "grad_norm": 0.05614482938233714,
      "learning_rate": 0.0001092883928385155,
      "loss": 0.8428,
      "step": 137800
    },
    {
      "epoch": 2.419459611299356,
      "grad_norm": 0.05093003905662201,
      "learning_rate": 0.00010927728827620176,
      "loss": 0.8477,
      "step": 137810
    },
    {
      "epoch": 2.419635176179357,
      "grad_norm": 0.044323777039300426,
      "learning_rate": 0.0001092661836090132,
      "loss": 0.8412,
      "step": 137820
    },
    {
      "epoch": 2.4198107410593583,
      "grad_norm": 0.06150770896839833,
      "learning_rate": 0.0001092550788370905,
      "loss": 0.852,
      "step": 137830
    },
    {
      "epoch": 2.41998630593936,
      "grad_norm": 0.06995214578943805,
      "learning_rate": 0.00010924397396057435,
      "loss": 0.8442,
      "step": 137840
    },
    {
      "epoch": 2.4201618708193613,
      "grad_norm": 0.07743940182229744,
      "learning_rate": 0.00010923286897960549,
      "loss": 0.8473,
      "step": 137850
    },
    {
      "epoch": 2.4203374356993628,
      "grad_norm": 0.05766960517834828,
      "learning_rate": 0.00010922176389432459,
      "loss": 0.8455,
      "step": 137860
    },
    {
      "epoch": 2.4205130005793642,
      "grad_norm": 0.057060844045429084,
      "learning_rate": 0.00010921065870487239,
      "loss": 0.8501,
      "step": 137870
    },
    {
      "epoch": 2.4206885654593657,
      "grad_norm": 0.04531984300891696,
      "learning_rate": 0.00010919955341138956,
      "loss": 0.8523,
      "step": 137880
    },
    {
      "epoch": 2.4208641303393668,
      "grad_norm": 0.07073359225489485,
      "learning_rate": 0.00010918844801401682,
      "loss": 0.8461,
      "step": 137890
    },
    {
      "epoch": 2.4210396952193682,
      "grad_norm": 0.05894874647124465,
      "learning_rate": 0.00010917734251289487,
      "loss": 0.8466,
      "step": 137900
    },
    {
      "epoch": 2.4212152600993697,
      "grad_norm": 0.054769093696915384,
      "learning_rate": 0.00010916623690816441,
      "loss": 0.8442,
      "step": 137910
    },
    {
      "epoch": 2.421390824979371,
      "grad_norm": 0.05347828472886927,
      "learning_rate": 0.00010915513119996618,
      "loss": 0.8461,
      "step": 137920
    },
    {
      "epoch": 2.4215663898593727,
      "grad_norm": 0.05602197203030771,
      "learning_rate": 0.00010914402538844089,
      "loss": 0.8495,
      "step": 137930
    },
    {
      "epoch": 2.4217419547393737,
      "grad_norm": 0.05843190664775145,
      "learning_rate": 0.00010913291947372921,
      "loss": 0.846,
      "step": 137940
    },
    {
      "epoch": 2.421917519619375,
      "grad_norm": 0.05200004006533318,
      "learning_rate": 0.00010912181345597185,
      "loss": 0.8444,
      "step": 137950
    },
    {
      "epoch": 2.4220930844993767,
      "grad_norm": 0.057482003198123736,
      "learning_rate": 0.00010911070733530962,
      "loss": 0.85,
      "step": 137960
    },
    {
      "epoch": 2.422268649379378,
      "grad_norm": 0.05038662406360784,
      "learning_rate": 0.00010909960111188315,
      "loss": 0.8553,
      "step": 137970
    },
    {
      "epoch": 2.4224442142593796,
      "grad_norm": 0.07325178370770673,
      "learning_rate": 0.00010908849478583318,
      "loss": 0.844,
      "step": 137980
    },
    {
      "epoch": 2.422619779139381,
      "grad_norm": 0.058351686368922046,
      "learning_rate": 0.00010907738835730042,
      "loss": 0.8404,
      "step": 137990
    },
    {
      "epoch": 2.4227953440193826,
      "grad_norm": 0.055485165116309715,
      "learning_rate": 0.00010906628182642564,
      "loss": 0.8472,
      "step": 138000
    },
    {
      "epoch": 2.4229709088993836,
      "grad_norm": 0.05322701634391084,
      "learning_rate": 0.00010905517519334947,
      "loss": 0.8476,
      "step": 138010
    },
    {
      "epoch": 2.423146473779385,
      "grad_norm": 0.05764926840348585,
      "learning_rate": 0.00010904406845821269,
      "loss": 0.846,
      "step": 138020
    },
    {
      "epoch": 2.4233220386593866,
      "grad_norm": 0.06240239786299041,
      "learning_rate": 0.00010903296162115609,
      "loss": 0.8506,
      "step": 138030
    },
    {
      "epoch": 2.423497603539388,
      "grad_norm": 0.054827477483044,
      "learning_rate": 0.00010902185468232026,
      "loss": 0.8424,
      "step": 138040
    },
    {
      "epoch": 2.4236731684193895,
      "grad_norm": 0.07921768295258343,
      "learning_rate": 0.00010901074764184604,
      "loss": 0.8449,
      "step": 138050
    },
    {
      "epoch": 2.4238487332993905,
      "grad_norm": 0.07812693485227101,
      "learning_rate": 0.00010899964049987408,
      "loss": 0.842,
      "step": 138060
    },
    {
      "epoch": 2.424024298179392,
      "grad_norm": 0.047490383043254945,
      "learning_rate": 0.00010898853325654512,
      "loss": 0.8432,
      "step": 138070
    },
    {
      "epoch": 2.4241998630593935,
      "grad_norm": 0.0492352167445432,
      "learning_rate": 0.00010897742591199995,
      "loss": 0.8493,
      "step": 138080
    },
    {
      "epoch": 2.424375427939395,
      "grad_norm": 0.046154596573439725,
      "learning_rate": 0.00010896631846637926,
      "loss": 0.8508,
      "step": 138090
    },
    {
      "epoch": 2.4245509928193965,
      "grad_norm": 0.04809948811839458,
      "learning_rate": 0.00010895521091982382,
      "loss": 0.8464,
      "step": 138100
    },
    {
      "epoch": 2.424726557699398,
      "grad_norm": 0.04934400603525908,
      "learning_rate": 0.00010894410327247429,
      "loss": 0.8541,
      "step": 138110
    },
    {
      "epoch": 2.4249021225793994,
      "grad_norm": 0.06639905107877984,
      "learning_rate": 0.00010893299552447143,
      "loss": 0.8404,
      "step": 138120
    },
    {
      "epoch": 2.4250776874594004,
      "grad_norm": 0.053614020770851104,
      "learning_rate": 0.00010892188767595604,
      "loss": 0.8371,
      "step": 138130
    },
    {
      "epoch": 2.425253252339402,
      "grad_norm": 0.05408450238428727,
      "learning_rate": 0.0001089107797270688,
      "loss": 0.8465,
      "step": 138140
    },
    {
      "epoch": 2.4254288172194034,
      "grad_norm": 0.060776088957839605,
      "learning_rate": 0.00010889967167795047,
      "loss": 0.8426,
      "step": 138150
    },
    {
      "epoch": 2.425604382099405,
      "grad_norm": 0.05462283389736188,
      "learning_rate": 0.00010888856352874176,
      "loss": 0.8446,
      "step": 138160
    },
    {
      "epoch": 2.4257799469794064,
      "grad_norm": 0.05321385417809529,
      "learning_rate": 0.00010887745527958346,
      "loss": 0.8476,
      "step": 138170
    },
    {
      "epoch": 2.425955511859408,
      "grad_norm": 0.05533671203082974,
      "learning_rate": 0.00010886634693061629,
      "loss": 0.8533,
      "step": 138180
    },
    {
      "epoch": 2.426131076739409,
      "grad_norm": 0.046158008498947994,
      "learning_rate": 0.00010885523848198096,
      "loss": 0.8392,
      "step": 138190
    },
    {
      "epoch": 2.4263066416194103,
      "grad_norm": 0.05628961264539469,
      "learning_rate": 0.00010884412993381832,
      "loss": 0.8445,
      "step": 138200
    },
    {
      "epoch": 2.426482206499412,
      "grad_norm": 0.05272738777060443,
      "learning_rate": 0.00010883302128626902,
      "loss": 0.8437,
      "step": 138210
    },
    {
      "epoch": 2.4266577713794133,
      "grad_norm": 0.08007867138060688,
      "learning_rate": 0.00010882191253947383,
      "loss": 0.8494,
      "step": 138220
    },
    {
      "epoch": 2.4268333362594148,
      "grad_norm": 0.0846876018925317,
      "learning_rate": 0.00010881080369357352,
      "loss": 0.8462,
      "step": 138230
    },
    {
      "epoch": 2.4270089011394163,
      "grad_norm": 0.05631248347329398,
      "learning_rate": 0.00010879969474870881,
      "loss": 0.8368,
      "step": 138240
    },
    {
      "epoch": 2.4271844660194173,
      "grad_norm": 0.056426035658709725,
      "learning_rate": 0.00010878858570502051,
      "loss": 0.8472,
      "step": 138250
    },
    {
      "epoch": 2.4273600308994188,
      "grad_norm": 0.06700693071516357,
      "learning_rate": 0.00010877747656264932,
      "loss": 0.8298,
      "step": 138260
    },
    {
      "epoch": 2.4275355957794202,
      "grad_norm": 0.046203705445673385,
      "learning_rate": 0.00010876636732173601,
      "loss": 0.8423,
      "step": 138270
    },
    {
      "epoch": 2.4277111606594217,
      "grad_norm": 0.06081266326374724,
      "learning_rate": 0.00010875525798242133,
      "loss": 0.835,
      "step": 138280
    },
    {
      "epoch": 2.427886725539423,
      "grad_norm": 0.043419630540682146,
      "learning_rate": 0.00010874414854484602,
      "loss": 0.8427,
      "step": 138290
    },
    {
      "epoch": 2.4280622904194247,
      "grad_norm": 0.06109734452442898,
      "learning_rate": 0.00010873303900915091,
      "loss": 0.8546,
      "step": 138300
    },
    {
      "epoch": 2.428237855299426,
      "grad_norm": 0.08554928424820614,
      "learning_rate": 0.00010872192937547671,
      "loss": 0.8514,
      "step": 138310
    },
    {
      "epoch": 2.428413420179427,
      "grad_norm": 0.087710344410742,
      "learning_rate": 0.00010871081964396417,
      "loss": 0.8397,
      "step": 138320
    },
    {
      "epoch": 2.4285889850594287,
      "grad_norm": 0.0694703819632762,
      "learning_rate": 0.00010869970981475406,
      "loss": 0.8472,
      "step": 138330
    },
    {
      "epoch": 2.42876454993943,
      "grad_norm": 0.053181878369814105,
      "learning_rate": 0.00010868859988798719,
      "loss": 0.8458,
      "step": 138340
    },
    {
      "epoch": 2.4289401148194316,
      "grad_norm": 0.053271564383928265,
      "learning_rate": 0.00010867748986380424,
      "loss": 0.8502,
      "step": 138350
    },
    {
      "epoch": 2.429115679699433,
      "grad_norm": 0.053743979473739385,
      "learning_rate": 0.00010866637974234601,
      "loss": 0.845,
      "step": 138360
    },
    {
      "epoch": 2.429291244579434,
      "grad_norm": 0.04902952535495593,
      "learning_rate": 0.00010865526952375333,
      "loss": 0.8333,
      "step": 138370
    },
    {
      "epoch": 2.4294668094594356,
      "grad_norm": 0.05206661213510157,
      "learning_rate": 0.00010864415920816691,
      "loss": 0.8454,
      "step": 138380
    },
    {
      "epoch": 2.429642374339437,
      "grad_norm": 0.050437745815659116,
      "learning_rate": 0.00010863304879572751,
      "loss": 0.8428,
      "step": 138390
    },
    {
      "epoch": 2.4298179392194386,
      "grad_norm": 0.052475786781606346,
      "learning_rate": 0.00010862193828657589,
      "loss": 0.8425,
      "step": 138400
    },
    {
      "epoch": 2.42999350409944,
      "grad_norm": 0.05377933442372619,
      "learning_rate": 0.00010861082768085288,
      "loss": 0.8436,
      "step": 138410
    },
    {
      "epoch": 2.4301690689794415,
      "grad_norm": 0.0600431074788126,
      "learning_rate": 0.00010859971697869927,
      "loss": 0.8506,
      "step": 138420
    },
    {
      "epoch": 2.430344633859443,
      "grad_norm": 0.0532063678979175,
      "learning_rate": 0.00010858860618025574,
      "loss": 0.8479,
      "step": 138430
    },
    {
      "epoch": 2.430520198739444,
      "grad_norm": 0.052102157859988146,
      "learning_rate": 0.00010857749528566312,
      "loss": 0.8504,
      "step": 138440
    },
    {
      "epoch": 2.4306957636194455,
      "grad_norm": 0.050785319420587165,
      "learning_rate": 0.00010856638429506218,
      "loss": 0.8506,
      "step": 138450
    },
    {
      "epoch": 2.430871328499447,
      "grad_norm": 0.056421863092653575,
      "learning_rate": 0.0001085552732085937,
      "loss": 0.8414,
      "step": 138460
    },
    {
      "epoch": 2.4310468933794485,
      "grad_norm": 0.05989873991400742,
      "learning_rate": 0.00010854416202639848,
      "loss": 0.8464,
      "step": 138470
    },
    {
      "epoch": 2.43122245825945,
      "grad_norm": 0.056071831789920874,
      "learning_rate": 0.00010853305074861728,
      "loss": 0.8449,
      "step": 138480
    },
    {
      "epoch": 2.431398023139451,
      "grad_norm": 0.0572346362888503,
      "learning_rate": 0.00010852193937539088,
      "loss": 0.8475,
      "step": 138490
    },
    {
      "epoch": 2.4315735880194524,
      "grad_norm": 0.07778943840793782,
      "learning_rate": 0.00010851082790686004,
      "loss": 0.8444,
      "step": 138500
    },
    {
      "epoch": 2.431749152899454,
      "grad_norm": 0.057627103494665204,
      "learning_rate": 0.0001084997163431656,
      "loss": 0.8511,
      "step": 138510
    },
    {
      "epoch": 2.4319247177794554,
      "grad_norm": 0.06445006049028995,
      "learning_rate": 0.00010848860468444827,
      "loss": 0.846,
      "step": 138520
    },
    {
      "epoch": 2.432100282659457,
      "grad_norm": 0.05730500456016436,
      "learning_rate": 0.00010847749293084893,
      "loss": 0.8474,
      "step": 138530
    },
    {
      "epoch": 2.4322758475394584,
      "grad_norm": 0.06199419574661729,
      "learning_rate": 0.00010846638108250832,
      "loss": 0.8364,
      "step": 138540
    },
    {
      "epoch": 2.43245141241946,
      "grad_norm": 0.05458843134993274,
      "learning_rate": 0.00010845526913956723,
      "loss": 0.8487,
      "step": 138550
    },
    {
      "epoch": 2.432626977299461,
      "grad_norm": 0.05813834066398378,
      "learning_rate": 0.00010844415710216645,
      "loss": 0.846,
      "step": 138560
    },
    {
      "epoch": 2.4328025421794623,
      "grad_norm": 0.044287339589768764,
      "learning_rate": 0.00010843304497044677,
      "loss": 0.8471,
      "step": 138570
    },
    {
      "epoch": 2.432978107059464,
      "grad_norm": 0.07440425960765305,
      "learning_rate": 0.00010842193274454895,
      "loss": 0.8444,
      "step": 138580
    },
    {
      "epoch": 2.4331536719394653,
      "grad_norm": 0.06391839244061702,
      "learning_rate": 0.00010841082042461388,
      "loss": 0.8488,
      "step": 138590
    },
    {
      "epoch": 2.433329236819467,
      "grad_norm": 0.05113292306584992,
      "learning_rate": 0.00010839970801078225,
      "loss": 0.8466,
      "step": 138600
    },
    {
      "epoch": 2.433504801699468,
      "grad_norm": 0.06527200167768382,
      "learning_rate": 0.00010838859550319494,
      "loss": 0.8421,
      "step": 138610
    },
    {
      "epoch": 2.4336803665794693,
      "grad_norm": 0.05870857530578415,
      "learning_rate": 0.00010837748290199268,
      "loss": 0.8423,
      "step": 138620
    },
    {
      "epoch": 2.4338559314594708,
      "grad_norm": 0.05305169036511596,
      "learning_rate": 0.00010836637020731629,
      "loss": 0.8397,
      "step": 138630
    },
    {
      "epoch": 2.4340314963394722,
      "grad_norm": 0.06642253974267923,
      "learning_rate": 0.00010835525741930662,
      "loss": 0.8485,
      "step": 138640
    },
    {
      "epoch": 2.4342070612194737,
      "grad_norm": 0.05057135410593122,
      "learning_rate": 0.0001083441445381044,
      "loss": 0.8539,
      "step": 138650
    },
    {
      "epoch": 2.434382626099475,
      "grad_norm": 0.049702915419726185,
      "learning_rate": 0.00010833303156385046,
      "loss": 0.8475,
      "step": 138660
    },
    {
      "epoch": 2.4345581909794767,
      "grad_norm": 0.0602290539510732,
      "learning_rate": 0.0001083219184966856,
      "loss": 0.8462,
      "step": 138670
    },
    {
      "epoch": 2.4347337558594777,
      "grad_norm": 0.05529993230809688,
      "learning_rate": 0.00010831080533675064,
      "loss": 0.853,
      "step": 138680
    },
    {
      "epoch": 2.434909320739479,
      "grad_norm": 0.06519048696358372,
      "learning_rate": 0.00010829969208418636,
      "loss": 0.8483,
      "step": 138690
    },
    {
      "epoch": 2.4350848856194807,
      "grad_norm": 0.06266686083880009,
      "learning_rate": 0.00010828857873913358,
      "loss": 0.8329,
      "step": 138700
    },
    {
      "epoch": 2.435260450499482,
      "grad_norm": 0.058928798039583584,
      "learning_rate": 0.00010827746530173313,
      "loss": 0.846,
      "step": 138710
    },
    {
      "epoch": 2.4354360153794836,
      "grad_norm": 0.05565180161374259,
      "learning_rate": 0.0001082663517721258,
      "loss": 0.8368,
      "step": 138720
    },
    {
      "epoch": 2.4356115802594847,
      "grad_norm": 0.07189555444597007,
      "learning_rate": 0.0001082552381504524,
      "loss": 0.8453,
      "step": 138730
    },
    {
      "epoch": 2.435787145139486,
      "grad_norm": 0.0616172225787174,
      "learning_rate": 0.00010824412443685374,
      "loss": 0.8448,
      "step": 138740
    },
    {
      "epoch": 2.4359627100194876,
      "grad_norm": 0.05349570470519258,
      "learning_rate": 0.00010823301063147062,
      "loss": 0.8463,
      "step": 138750
    },
    {
      "epoch": 2.436138274899489,
      "grad_norm": 0.04581866055354856,
      "learning_rate": 0.00010822189673444387,
      "loss": 0.8439,
      "step": 138760
    },
    {
      "epoch": 2.4363138397794906,
      "grad_norm": 0.04258477454204033,
      "learning_rate": 0.00010821078274591435,
      "loss": 0.8418,
      "step": 138770
    },
    {
      "epoch": 2.436489404659492,
      "grad_norm": 0.053990462507345606,
      "learning_rate": 0.0001081996686660228,
      "loss": 0.8359,
      "step": 138780
    },
    {
      "epoch": 2.4366649695394935,
      "grad_norm": 0.09678792780292192,
      "learning_rate": 0.00010818855449491003,
      "loss": 0.8431,
      "step": 138790
    },
    {
      "epoch": 2.4368405344194946,
      "grad_norm": 0.05567909323027837,
      "learning_rate": 0.00010817744023271694,
      "loss": 0.8443,
      "step": 138800
    },
    {
      "epoch": 2.437016099299496,
      "grad_norm": 0.04944176214765995,
      "learning_rate": 0.00010816632587958431,
      "loss": 0.8424,
      "step": 138810
    },
    {
      "epoch": 2.4371916641794975,
      "grad_norm": 0.05211883478441271,
      "learning_rate": 0.00010815521143565295,
      "loss": 0.8507,
      "step": 138820
    },
    {
      "epoch": 2.437367229059499,
      "grad_norm": 0.045807717631084474,
      "learning_rate": 0.00010814409690106371,
      "loss": 0.8419,
      "step": 138830
    },
    {
      "epoch": 2.4375427939395005,
      "grad_norm": 0.09739799324175519,
      "learning_rate": 0.00010813298227595734,
      "loss": 0.8514,
      "step": 138840
    },
    {
      "epoch": 2.437718358819502,
      "grad_norm": 0.04707167684635818,
      "learning_rate": 0.00010812186756047476,
      "loss": 0.843,
      "step": 138850
    },
    {
      "epoch": 2.437893923699503,
      "grad_norm": 0.04184921716950721,
      "learning_rate": 0.0001081107527547567,
      "loss": 0.8459,
      "step": 138860
    },
    {
      "epoch": 2.4380694885795045,
      "grad_norm": 0.05181462061405836,
      "learning_rate": 0.00010809963785894409,
      "loss": 0.8452,
      "step": 138870
    },
    {
      "epoch": 2.438245053459506,
      "grad_norm": 0.0620588740858952,
      "learning_rate": 0.0001080885228731777,
      "loss": 0.8434,
      "step": 138880
    },
    {
      "epoch": 2.4384206183395074,
      "grad_norm": 0.05403583369822361,
      "learning_rate": 0.00010807740779759839,
      "loss": 0.8459,
      "step": 138890
    },
    {
      "epoch": 2.438596183219509,
      "grad_norm": 0.061560158492213364,
      "learning_rate": 0.00010806629263234693,
      "loss": 0.8411,
      "step": 138900
    },
    {
      "epoch": 2.4387717480995104,
      "grad_norm": 0.04772474756899504,
      "learning_rate": 0.00010805517737756418,
      "loss": 0.8483,
      "step": 138910
    },
    {
      "epoch": 2.4389473129795114,
      "grad_norm": 0.06488691608735811,
      "learning_rate": 0.000108044062033391,
      "loss": 0.8422,
      "step": 138920
    },
    {
      "epoch": 2.439122877859513,
      "grad_norm": 0.06994295193812318,
      "learning_rate": 0.00010803294659996821,
      "loss": 0.8362,
      "step": 138930
    },
    {
      "epoch": 2.4392984427395143,
      "grad_norm": 0.058258748373833054,
      "learning_rate": 0.00010802183107743665,
      "loss": 0.8428,
      "step": 138940
    },
    {
      "epoch": 2.439474007619516,
      "grad_norm": 0.06301556021352936,
      "learning_rate": 0.00010801071546593712,
      "loss": 0.8354,
      "step": 138950
    },
    {
      "epoch": 2.4396495724995173,
      "grad_norm": 0.059281168686661945,
      "learning_rate": 0.00010799959976561047,
      "loss": 0.8423,
      "step": 138960
    },
    {
      "epoch": 2.439825137379519,
      "grad_norm": 0.0857164381142839,
      "learning_rate": 0.00010798848397659755,
      "loss": 0.8394,
      "step": 138970
    },
    {
      "epoch": 2.44000070225952,
      "grad_norm": 0.0608753016102415,
      "learning_rate": 0.00010797736809903925,
      "loss": 0.8465,
      "step": 138980
    },
    {
      "epoch": 2.4401762671395213,
      "grad_norm": 0.047531584864686637,
      "learning_rate": 0.00010796625213307632,
      "loss": 0.8377,
      "step": 138990
    },
    {
      "epoch": 2.4403518320195228,
      "grad_norm": 0.06432272759229787,
      "learning_rate": 0.00010795513607884964,
      "loss": 0.8348,
      "step": 139000
    },
    {
      "epoch": 2.4405273968995242,
      "grad_norm": 0.07825750149159091,
      "learning_rate": 0.00010794401993650008,
      "loss": 0.8465,
      "step": 139010
    },
    {
      "epoch": 2.4407029617795257,
      "grad_norm": 0.04858011688216637,
      "learning_rate": 0.00010793290370616845,
      "loss": 0.8459,
      "step": 139020
    },
    {
      "epoch": 2.440878526659527,
      "grad_norm": 0.06753189955099899,
      "learning_rate": 0.00010792178738799554,
      "loss": 0.8415,
      "step": 139030
    },
    {
      "epoch": 2.4410540915395282,
      "grad_norm": 0.05209288799463472,
      "learning_rate": 0.00010791067098212228,
      "loss": 0.8484,
      "step": 139040
    },
    {
      "epoch": 2.4412296564195297,
      "grad_norm": 0.059128543351422376,
      "learning_rate": 0.00010789955448868953,
      "loss": 0.8471,
      "step": 139050
    },
    {
      "epoch": 2.441405221299531,
      "grad_norm": 0.05691539585431973,
      "learning_rate": 0.00010788843790783812,
      "loss": 0.8415,
      "step": 139060
    },
    {
      "epoch": 2.4415807861795327,
      "grad_norm": 0.05521125928816694,
      "learning_rate": 0.00010787732123970886,
      "loss": 0.8439,
      "step": 139070
    },
    {
      "epoch": 2.441756351059534,
      "grad_norm": 0.05256589261484946,
      "learning_rate": 0.0001078662044844426,
      "loss": 0.8407,
      "step": 139080
    },
    {
      "epoch": 2.4419319159395356,
      "grad_norm": 0.046783484466326526,
      "learning_rate": 0.00010785508764218022,
      "loss": 0.8399,
      "step": 139090
    },
    {
      "epoch": 2.442107480819537,
      "grad_norm": 0.05762469981618379,
      "learning_rate": 0.00010784397071306257,
      "loss": 0.8409,
      "step": 139100
    },
    {
      "epoch": 2.442283045699538,
      "grad_norm": 0.05040480210817086,
      "learning_rate": 0.00010783285369723051,
      "loss": 0.843,
      "step": 139110
    },
    {
      "epoch": 2.4424586105795396,
      "grad_norm": 0.05027694423494546,
      "learning_rate": 0.00010782173659482489,
      "loss": 0.8459,
      "step": 139120
    },
    {
      "epoch": 2.442634175459541,
      "grad_norm": 0.07476427061334603,
      "learning_rate": 0.00010781061940598653,
      "loss": 0.8423,
      "step": 139130
    },
    {
      "epoch": 2.4428097403395426,
      "grad_norm": 0.06867658105582809,
      "learning_rate": 0.00010779950213085632,
      "loss": 0.8433,
      "step": 139140
    },
    {
      "epoch": 2.442985305219544,
      "grad_norm": 0.04846270962573446,
      "learning_rate": 0.00010778838476957513,
      "loss": 0.8406,
      "step": 139150
    },
    {
      "epoch": 2.443160870099545,
      "grad_norm": 0.059248974462294525,
      "learning_rate": 0.00010777726732228381,
      "loss": 0.8457,
      "step": 139160
    },
    {
      "epoch": 2.4433364349795466,
      "grad_norm": 0.056016148580251904,
      "learning_rate": 0.00010776614978912323,
      "loss": 0.8593,
      "step": 139170
    },
    {
      "epoch": 2.443511999859548,
      "grad_norm": 0.06072270956571663,
      "learning_rate": 0.0001077550321702342,
      "loss": 0.8384,
      "step": 139180
    },
    {
      "epoch": 2.4436875647395495,
      "grad_norm": 0.057867928940824265,
      "learning_rate": 0.00010774391446575762,
      "loss": 0.8404,
      "step": 139190
    },
    {
      "epoch": 2.443863129619551,
      "grad_norm": 0.05580598281208407,
      "learning_rate": 0.00010773279667583433,
      "loss": 0.8378,
      "step": 139200
    },
    {
      "epoch": 2.4440386944995525,
      "grad_norm": 0.06671792313698893,
      "learning_rate": 0.00010772167880060524,
      "loss": 0.839,
      "step": 139210
    },
    {
      "epoch": 2.444214259379554,
      "grad_norm": 0.061139927499415256,
      "learning_rate": 0.00010771056084021121,
      "loss": 0.8467,
      "step": 139220
    },
    {
      "epoch": 2.444389824259555,
      "grad_norm": 0.07874127769888122,
      "learning_rate": 0.00010769944279479309,
      "loss": 0.8455,
      "step": 139230
    },
    {
      "epoch": 2.4445653891395565,
      "grad_norm": 0.0503824821463882,
      "learning_rate": 0.00010768832466449175,
      "loss": 0.8455,
      "step": 139240
    },
    {
      "epoch": 2.444740954019558,
      "grad_norm": 0.05877565760474133,
      "learning_rate": 0.00010767720644944804,
      "loss": 0.8466,
      "step": 139250
    },
    {
      "epoch": 2.4449165188995594,
      "grad_norm": 0.07637042788921777,
      "learning_rate": 0.00010766608814980281,
      "loss": 0.845,
      "step": 139260
    },
    {
      "epoch": 2.445092083779561,
      "grad_norm": 0.08255299071877574,
      "learning_rate": 0.000107654969765697,
      "loss": 0.8379,
      "step": 139270
    },
    {
      "epoch": 2.445267648659562,
      "grad_norm": 0.07344950242724003,
      "learning_rate": 0.00010764385129727146,
      "loss": 0.8472,
      "step": 139280
    },
    {
      "epoch": 2.4454432135395634,
      "grad_norm": 0.05031710871691274,
      "learning_rate": 0.00010763273274466703,
      "loss": 0.8478,
      "step": 139290
    },
    {
      "epoch": 2.445618778419565,
      "grad_norm": 0.06988529647182527,
      "learning_rate": 0.00010762161410802461,
      "loss": 0.8459,
      "step": 139300
    },
    {
      "epoch": 2.4457943432995664,
      "grad_norm": 0.06309485769650092,
      "learning_rate": 0.00010761049538748504,
      "loss": 0.8421,
      "step": 139310
    },
    {
      "epoch": 2.445969908179568,
      "grad_norm": 0.05746105316261229,
      "learning_rate": 0.00010759937658318929,
      "loss": 0.8397,
      "step": 139320
    },
    {
      "epoch": 2.4461454730595693,
      "grad_norm": 0.06613815557750234,
      "learning_rate": 0.00010758825769527814,
      "loss": 0.8376,
      "step": 139330
    },
    {
      "epoch": 2.446321037939571,
      "grad_norm": 0.04629324957719398,
      "learning_rate": 0.0001075771387238925,
      "loss": 0.8369,
      "step": 139340
    },
    {
      "epoch": 2.446496602819572,
      "grad_norm": 0.059800457752016964,
      "learning_rate": 0.00010756601966917325,
      "loss": 0.8503,
      "step": 139350
    },
    {
      "epoch": 2.4466721676995733,
      "grad_norm": 0.06894743527251784,
      "learning_rate": 0.00010755490053126129,
      "loss": 0.8402,
      "step": 139360
    },
    {
      "epoch": 2.4468477325795748,
      "grad_norm": 0.05899519968451426,
      "learning_rate": 0.00010754378131029745,
      "loss": 0.8431,
      "step": 139370
    },
    {
      "epoch": 2.4470232974595763,
      "grad_norm": 0.04389934922132425,
      "learning_rate": 0.00010753266200642266,
      "loss": 0.8407,
      "step": 139380
    },
    {
      "epoch": 2.4471988623395777,
      "grad_norm": 0.07835992249271169,
      "learning_rate": 0.00010752154261977782,
      "loss": 0.8445,
      "step": 139390
    },
    {
      "epoch": 2.4473744272195788,
      "grad_norm": 0.06022488531460472,
      "learning_rate": 0.00010751042315050376,
      "loss": 0.8481,
      "step": 139400
    },
    {
      "epoch": 2.4475499920995802,
      "grad_norm": 0.05457302462371137,
      "learning_rate": 0.00010749930359874141,
      "loss": 0.8436,
      "step": 139410
    },
    {
      "epoch": 2.4477255569795817,
      "grad_norm": 0.057151945453919675,
      "learning_rate": 0.00010748818396463161,
      "loss": 0.8409,
      "step": 139420
    },
    {
      "epoch": 2.447901121859583,
      "grad_norm": 0.0663709911507082,
      "learning_rate": 0.00010747706424831527,
      "loss": 0.8446,
      "step": 139430
    },
    {
      "epoch": 2.4480766867395847,
      "grad_norm": 0.07443112275834046,
      "learning_rate": 0.00010746594444993331,
      "loss": 0.8482,
      "step": 139440
    },
    {
      "epoch": 2.448252251619586,
      "grad_norm": 0.04682379529695092,
      "learning_rate": 0.00010745482456962661,
      "loss": 0.8474,
      "step": 139450
    },
    {
      "epoch": 2.4484278164995876,
      "grad_norm": 0.054350496034104696,
      "learning_rate": 0.00010744370460753604,
      "loss": 0.8431,
      "step": 139460
    },
    {
      "epoch": 2.4486033813795887,
      "grad_norm": 0.04889709139767008,
      "learning_rate": 0.00010743258456380246,
      "loss": 0.8436,
      "step": 139470
    },
    {
      "epoch": 2.44877894625959,
      "grad_norm": 0.06502153926833601,
      "learning_rate": 0.0001074214644385668,
      "loss": 0.8443,
      "step": 139480
    },
    {
      "epoch": 2.4489545111395916,
      "grad_norm": 0.05635381529994116,
      "learning_rate": 0.00010741034423196999,
      "loss": 0.8557,
      "step": 139490
    },
    {
      "epoch": 2.449130076019593,
      "grad_norm": 0.08438780419698313,
      "learning_rate": 0.00010739922394415287,
      "loss": 0.839,
      "step": 139500
    },
    {
      "epoch": 2.4493056408995946,
      "grad_norm": 0.0571624825043205,
      "learning_rate": 0.00010738810357525637,
      "loss": 0.8422,
      "step": 139510
    },
    {
      "epoch": 2.4494812057795956,
      "grad_norm": 0.05011519750369978,
      "learning_rate": 0.00010737698312542135,
      "loss": 0.8472,
      "step": 139520
    },
    {
      "epoch": 2.449656770659597,
      "grad_norm": 0.04631353702673392,
      "learning_rate": 0.00010736586259478876,
      "loss": 0.8442,
      "step": 139530
    },
    {
      "epoch": 2.4498323355395986,
      "grad_norm": 0.06406254991843116,
      "learning_rate": 0.00010735474198349946,
      "loss": 0.8364,
      "step": 139540
    },
    {
      "epoch": 2.4500079004196,
      "grad_norm": 0.0486210269555218,
      "learning_rate": 0.00010734362129169435,
      "loss": 0.8464,
      "step": 139550
    },
    {
      "epoch": 2.4501834652996015,
      "grad_norm": 0.05889301449051244,
      "learning_rate": 0.00010733250051951435,
      "loss": 0.844,
      "step": 139560
    },
    {
      "epoch": 2.450359030179603,
      "grad_norm": 0.048333387941989346,
      "learning_rate": 0.00010732137966710037,
      "loss": 0.84,
      "step": 139570
    },
    {
      "epoch": 2.4505345950596045,
      "grad_norm": 0.04713292081938731,
      "learning_rate": 0.00010731025873459328,
      "loss": 0.8451,
      "step": 139580
    },
    {
      "epoch": 2.4507101599396055,
      "grad_norm": 0.056924708073910736,
      "learning_rate": 0.00010729913772213401,
      "loss": 0.839,
      "step": 139590
    },
    {
      "epoch": 2.450885724819607,
      "grad_norm": 0.06018799719979223,
      "learning_rate": 0.00010728801662986343,
      "loss": 0.8451,
      "step": 139600
    },
    {
      "epoch": 2.4510612896996085,
      "grad_norm": 0.0408115932220194,
      "learning_rate": 0.0001072768954579225,
      "loss": 0.8402,
      "step": 139610
    },
    {
      "epoch": 2.45123685457961,
      "grad_norm": 0.04433150980285789,
      "learning_rate": 0.00010726577420645215,
      "loss": 0.851,
      "step": 139620
    },
    {
      "epoch": 2.4514124194596114,
      "grad_norm": 0.04736806717660565,
      "learning_rate": 0.00010725465287559317,
      "loss": 0.8488,
      "step": 139630
    },
    {
      "epoch": 2.451587984339613,
      "grad_norm": 0.06159540430011085,
      "learning_rate": 0.00010724353146548656,
      "loss": 0.8448,
      "step": 139640
    },
    {
      "epoch": 2.451763549219614,
      "grad_norm": 0.062310198753876334,
      "learning_rate": 0.00010723240997627319,
      "loss": 0.8424,
      "step": 139650
    },
    {
      "epoch": 2.4519391140996154,
      "grad_norm": 0.04915562763322431,
      "learning_rate": 0.00010722128840809402,
      "loss": 0.8491,
      "step": 139660
    },
    {
      "epoch": 2.452114678979617,
      "grad_norm": 0.04448373855444157,
      "learning_rate": 0.00010721016676108992,
      "loss": 0.8504,
      "step": 139670
    },
    {
      "epoch": 2.4522902438596184,
      "grad_norm": 0.0497512621092882,
      "learning_rate": 0.00010719904503540186,
      "loss": 0.8433,
      "step": 139680
    },
    {
      "epoch": 2.45246580873962,
      "grad_norm": 0.05908373833454216,
      "learning_rate": 0.00010718792323117063,
      "loss": 0.8479,
      "step": 139690
    },
    {
      "epoch": 2.4526413736196213,
      "grad_norm": 0.05822968605100334,
      "learning_rate": 0.00010717680134853731,
      "loss": 0.8424,
      "step": 139700
    },
    {
      "epoch": 2.4528169384996223,
      "grad_norm": 0.07430365054800442,
      "learning_rate": 0.00010716567938764268,
      "loss": 0.8467,
      "step": 139710
    },
    {
      "epoch": 2.452992503379624,
      "grad_norm": 0.05164788271239653,
      "learning_rate": 0.0001071545573486277,
      "loss": 0.845,
      "step": 139720
    },
    {
      "epoch": 2.4531680682596253,
      "grad_norm": 0.052325976726875355,
      "learning_rate": 0.00010714343523163333,
      "loss": 0.8399,
      "step": 139730
    },
    {
      "epoch": 2.453343633139627,
      "grad_norm": 0.044152130008538966,
      "learning_rate": 0.00010713231303680045,
      "loss": 0.8472,
      "step": 139740
    },
    {
      "epoch": 2.4535191980196283,
      "grad_norm": 0.067258984639649,
      "learning_rate": 0.00010712119076427,
      "loss": 0.8453,
      "step": 139750
    },
    {
      "epoch": 2.4536947628996297,
      "grad_norm": 0.06618982590503078,
      "learning_rate": 0.00010711006841418287,
      "loss": 0.8379,
      "step": 139760
    },
    {
      "epoch": 2.453870327779631,
      "grad_norm": 0.07709192464954004,
      "learning_rate": 0.00010709894598668,
      "loss": 0.8421,
      "step": 139770
    },
    {
      "epoch": 2.4540458926596322,
      "grad_norm": 0.05057556913178446,
      "learning_rate": 0.00010708782348190231,
      "loss": 0.8491,
      "step": 139780
    },
    {
      "epoch": 2.4542214575396337,
      "grad_norm": 0.048908212356533864,
      "learning_rate": 0.00010707670089999077,
      "loss": 0.8443,
      "step": 139790
    },
    {
      "epoch": 2.454397022419635,
      "grad_norm": 0.04795269682403395,
      "learning_rate": 0.00010706557824108627,
      "loss": 0.8391,
      "step": 139800
    },
    {
      "epoch": 2.4545725872996367,
      "grad_norm": 0.05172809334854762,
      "learning_rate": 0.00010705445550532969,
      "loss": 0.8458,
      "step": 139810
    },
    {
      "epoch": 2.454748152179638,
      "grad_norm": 0.07085929201686694,
      "learning_rate": 0.000107043332692862,
      "loss": 0.844,
      "step": 139820
    },
    {
      "epoch": 2.454923717059639,
      "grad_norm": 0.06059013155531974,
      "learning_rate": 0.00010703220980382414,
      "loss": 0.8456,
      "step": 139830
    },
    {
      "epoch": 2.4550992819396407,
      "grad_norm": 0.06529095887593535,
      "learning_rate": 0.00010702108683835703,
      "loss": 0.8441,
      "step": 139840
    },
    {
      "epoch": 2.455274846819642,
      "grad_norm": 0.0690165900313399,
      "learning_rate": 0.00010700996379660161,
      "loss": 0.8414,
      "step": 139850
    },
    {
      "epoch": 2.4554504116996436,
      "grad_norm": 0.06595988284417453,
      "learning_rate": 0.00010699884067869874,
      "loss": 0.8431,
      "step": 139860
    },
    {
      "epoch": 2.455625976579645,
      "grad_norm": 0.06283731092037313,
      "learning_rate": 0.00010698771748478949,
      "loss": 0.8493,
      "step": 139870
    },
    {
      "epoch": 2.4558015414596466,
      "grad_norm": 0.08066593643948584,
      "learning_rate": 0.00010697659421501462,
      "loss": 0.8393,
      "step": 139880
    },
    {
      "epoch": 2.455977106339648,
      "grad_norm": 0.052301542163509955,
      "learning_rate": 0.0001069654708695152,
      "loss": 0.8472,
      "step": 139890
    },
    {
      "epoch": 2.456152671219649,
      "grad_norm": 0.04623195786201849,
      "learning_rate": 0.00010695434744843213,
      "loss": 0.8402,
      "step": 139900
    },
    {
      "epoch": 2.4563282360996506,
      "grad_norm": 0.0805115226138945,
      "learning_rate": 0.00010694322395190635,
      "loss": 0.8438,
      "step": 139910
    },
    {
      "epoch": 2.456503800979652,
      "grad_norm": 0.06026044303460531,
      "learning_rate": 0.00010693210038007877,
      "loss": 0.8475,
      "step": 139920
    },
    {
      "epoch": 2.4566793658596535,
      "grad_norm": 0.05078631746966634,
      "learning_rate": 0.00010692097673309034,
      "loss": 0.8502,
      "step": 139930
    },
    {
      "epoch": 2.456854930739655,
      "grad_norm": 0.0643617932772678,
      "learning_rate": 0.00010690985301108196,
      "loss": 0.8466,
      "step": 139940
    },
    {
      "epoch": 2.457030495619656,
      "grad_norm": 0.06556317712456211,
      "learning_rate": 0.00010689872921419466,
      "loss": 0.8417,
      "step": 139950
    },
    {
      "epoch": 2.4572060604996575,
      "grad_norm": 0.07345260139864503,
      "learning_rate": 0.00010688760534256932,
      "loss": 0.8424,
      "step": 139960
    },
    {
      "epoch": 2.457381625379659,
      "grad_norm": 0.05073824527104381,
      "learning_rate": 0.00010687648139634692,
      "loss": 0.8392,
      "step": 139970
    },
    {
      "epoch": 2.4575571902596605,
      "grad_norm": 0.06503089392318616,
      "learning_rate": 0.00010686535737566831,
      "loss": 0.8454,
      "step": 139980
    },
    {
      "epoch": 2.457732755139662,
      "grad_norm": 0.048056896387641136,
      "learning_rate": 0.00010685423328067453,
      "loss": 0.8402,
      "step": 139990
    },
    {
      "epoch": 2.4579083200196634,
      "grad_norm": 0.04976786769663579,
      "learning_rate": 0.0001068431091115065,
      "loss": 0.8443,
      "step": 140000
    },
    {
      "epoch": 2.458083884899665,
      "grad_norm": 0.07369271714501902,
      "learning_rate": 0.00010683198486830515,
      "loss": 0.8433,
      "step": 140010
    },
    {
      "epoch": 2.458259449779666,
      "grad_norm": 0.06083022212511011,
      "learning_rate": 0.00010682086055121144,
      "loss": 0.8459,
      "step": 140020
    },
    {
      "epoch": 2.4584350146596674,
      "grad_norm": 0.04824501791334105,
      "learning_rate": 0.0001068097361603663,
      "loss": 0.8382,
      "step": 140030
    },
    {
      "epoch": 2.458610579539669,
      "grad_norm": 0.07646471778210112,
      "learning_rate": 0.00010679861169591068,
      "loss": 0.8445,
      "step": 140040
    },
    {
      "epoch": 2.4587861444196704,
      "grad_norm": 0.051661833613632316,
      "learning_rate": 0.00010678748715798556,
      "loss": 0.8414,
      "step": 140050
    },
    {
      "epoch": 2.458961709299672,
      "grad_norm": 0.04688285571207312,
      "learning_rate": 0.00010677636254673185,
      "loss": 0.8429,
      "step": 140060
    },
    {
      "epoch": 2.459137274179673,
      "grad_norm": 0.06336325371876984,
      "learning_rate": 0.00010676523786229053,
      "loss": 0.8517,
      "step": 140070
    },
    {
      "epoch": 2.4593128390596743,
      "grad_norm": 0.05001639058052521,
      "learning_rate": 0.00010675411310480254,
      "loss": 0.8328,
      "step": 140080
    },
    {
      "epoch": 2.459488403939676,
      "grad_norm": 0.06906854364577189,
      "learning_rate": 0.00010674298827440884,
      "loss": 0.8481,
      "step": 140090
    },
    {
      "epoch": 2.4596639688196773,
      "grad_norm": 0.056591379712198964,
      "learning_rate": 0.00010673186337125038,
      "loss": 0.8475,
      "step": 140100
    },
    {
      "epoch": 2.459839533699679,
      "grad_norm": 0.06364305838206809,
      "learning_rate": 0.0001067207383954681,
      "loss": 0.8457,
      "step": 140110
    },
    {
      "epoch": 2.4600150985796803,
      "grad_norm": 0.056824893808396566,
      "learning_rate": 0.000106709613347203,
      "loss": 0.8441,
      "step": 140120
    },
    {
      "epoch": 2.4601906634596817,
      "grad_norm": 0.05906092460069395,
      "learning_rate": 0.00010669848822659597,
      "loss": 0.8375,
      "step": 140130
    },
    {
      "epoch": 2.4603662283396828,
      "grad_norm": 0.06818137287497983,
      "learning_rate": 0.00010668736303378801,
      "loss": 0.8326,
      "step": 140140
    },
    {
      "epoch": 2.4605417932196842,
      "grad_norm": 0.05768943424607159,
      "learning_rate": 0.00010667623776892008,
      "loss": 0.8406,
      "step": 140150
    },
    {
      "epoch": 2.4607173580996857,
      "grad_norm": 0.058497698834913466,
      "learning_rate": 0.00010666511243213311,
      "loss": 0.8469,
      "step": 140160
    },
    {
      "epoch": 2.460892922979687,
      "grad_norm": 0.05264638796787191,
      "learning_rate": 0.00010665398702356809,
      "loss": 0.8477,
      "step": 140170
    },
    {
      "epoch": 2.4610684878596887,
      "grad_norm": 0.0456203242374375,
      "learning_rate": 0.00010664286154336599,
      "loss": 0.8468,
      "step": 140180
    },
    {
      "epoch": 2.4612440527396897,
      "grad_norm": 0.05136684409329226,
      "learning_rate": 0.00010663173599166773,
      "loss": 0.8481,
      "step": 140190
    },
    {
      "epoch": 2.461419617619691,
      "grad_norm": 0.057581485687536765,
      "learning_rate": 0.0001066206103686143,
      "loss": 0.8435,
      "step": 140200
    },
    {
      "epoch": 2.4615951824996927,
      "grad_norm": 0.05018572870773924,
      "learning_rate": 0.00010660948467434669,
      "loss": 0.8426,
      "step": 140210
    },
    {
      "epoch": 2.461770747379694,
      "grad_norm": 0.04593932196783975,
      "learning_rate": 0.00010659835890900579,
      "loss": 0.8423,
      "step": 140220
    },
    {
      "epoch": 2.4619463122596956,
      "grad_norm": 0.04627495889242524,
      "learning_rate": 0.0001065872330727326,
      "loss": 0.8381,
      "step": 140230
    },
    {
      "epoch": 2.462121877139697,
      "grad_norm": 0.04915499944820688,
      "learning_rate": 0.00010657610716566812,
      "loss": 0.8445,
      "step": 140240
    },
    {
      "epoch": 2.4622974420196986,
      "grad_norm": 0.05943464754398215,
      "learning_rate": 0.0001065649811879533,
      "loss": 0.846,
      "step": 140250
    },
    {
      "epoch": 2.4624730068996996,
      "grad_norm": 0.07062770142553829,
      "learning_rate": 0.00010655385513972914,
      "loss": 0.8496,
      "step": 140260
    },
    {
      "epoch": 2.462648571779701,
      "grad_norm": 0.040124996450011524,
      "learning_rate": 0.00010654272902113649,
      "loss": 0.8463,
      "step": 140270
    },
    {
      "epoch": 2.4628241366597026,
      "grad_norm": 0.05792174367063797,
      "learning_rate": 0.00010653160283231643,
      "loss": 0.8491,
      "step": 140280
    },
    {
      "epoch": 2.462999701539704,
      "grad_norm": 0.05505362131202561,
      "learning_rate": 0.00010652047657340992,
      "loss": 0.8424,
      "step": 140290
    },
    {
      "epoch": 2.4631752664197055,
      "grad_norm": 0.061239062437517686,
      "learning_rate": 0.0001065093502445579,
      "loss": 0.8469,
      "step": 140300
    },
    {
      "epoch": 2.463350831299707,
      "grad_norm": 0.0631183649163558,
      "learning_rate": 0.00010649822384590138,
      "loss": 0.84,
      "step": 140310
    },
    {
      "epoch": 2.463526396179708,
      "grad_norm": 0.056628513637462685,
      "learning_rate": 0.00010648709737758127,
      "loss": 0.8448,
      "step": 140320
    },
    {
      "epoch": 2.4637019610597095,
      "grad_norm": 0.06305001375193917,
      "learning_rate": 0.00010647597083973861,
      "loss": 0.8427,
      "step": 140330
    },
    {
      "epoch": 2.463877525939711,
      "grad_norm": 0.05824184638462682,
      "learning_rate": 0.00010646484423251433,
      "loss": 0.8372,
      "step": 140340
    },
    {
      "epoch": 2.4640530908197125,
      "grad_norm": 0.051284903487539106,
      "learning_rate": 0.00010645371755604944,
      "loss": 0.8361,
      "step": 140350
    },
    {
      "epoch": 2.464228655699714,
      "grad_norm": 0.03929949098560273,
      "learning_rate": 0.00010644259081048493,
      "loss": 0.8459,
      "step": 140360
    },
    {
      "epoch": 2.4644042205797154,
      "grad_norm": 0.04718904480172594,
      "learning_rate": 0.00010643146399596171,
      "loss": 0.8485,
      "step": 140370
    },
    {
      "epoch": 2.4645797854597165,
      "grad_norm": 0.06753123780624587,
      "learning_rate": 0.00010642033711262083,
      "loss": 0.8463,
      "step": 140380
    },
    {
      "epoch": 2.464755350339718,
      "grad_norm": 0.05679989376863633,
      "learning_rate": 0.00010640921016060321,
      "loss": 0.8397,
      "step": 140390
    },
    {
      "epoch": 2.4649309152197194,
      "grad_norm": 0.05774899346995416,
      "learning_rate": 0.00010639808314004989,
      "loss": 0.8348,
      "step": 140400
    },
    {
      "epoch": 2.465106480099721,
      "grad_norm": 0.057653930387040975,
      "learning_rate": 0.0001063869560511018,
      "loss": 0.8455,
      "step": 140410
    },
    {
      "epoch": 2.4652820449797224,
      "grad_norm": 0.060708465117583224,
      "learning_rate": 0.00010637582889389997,
      "loss": 0.849,
      "step": 140420
    },
    {
      "epoch": 2.465457609859724,
      "grad_norm": 0.06102384998283003,
      "learning_rate": 0.00010636470166858534,
      "loss": 0.8351,
      "step": 140430
    },
    {
      "epoch": 2.465633174739725,
      "grad_norm": 0.07209260420656376,
      "learning_rate": 0.00010635357437529891,
      "loss": 0.8386,
      "step": 140440
    },
    {
      "epoch": 2.4658087396197264,
      "grad_norm": 0.05268788964748818,
      "learning_rate": 0.00010634244701418166,
      "loss": 0.8445,
      "step": 140450
    },
    {
      "epoch": 2.465984304499728,
      "grad_norm": 0.05736672720473146,
      "learning_rate": 0.00010633131958537458,
      "loss": 0.8414,
      "step": 140460
    },
    {
      "epoch": 2.4661598693797293,
      "grad_norm": 0.04747307448489311,
      "learning_rate": 0.0001063201920890187,
      "loss": 0.8511,
      "step": 140470
    },
    {
      "epoch": 2.466335434259731,
      "grad_norm": 0.05398558295394328,
      "learning_rate": 0.00010630906452525493,
      "loss": 0.8415,
      "step": 140480
    },
    {
      "epoch": 2.4665109991397323,
      "grad_norm": 0.06298292879931912,
      "learning_rate": 0.00010629793689422429,
      "loss": 0.8489,
      "step": 140490
    },
    {
      "epoch": 2.4666865640197333,
      "grad_norm": 0.06067057305070187,
      "learning_rate": 0.0001062868091960678,
      "loss": 0.8488,
      "step": 140500
    },
    {
      "epoch": 2.4668621288997348,
      "grad_norm": 0.05360498779794087,
      "learning_rate": 0.00010627568143092638,
      "loss": 0.8486,
      "step": 140510
    },
    {
      "epoch": 2.4670376937797363,
      "grad_norm": 0.04867349810354595,
      "learning_rate": 0.00010626455359894111,
      "loss": 0.8436,
      "step": 140520
    },
    {
      "epoch": 2.4672132586597377,
      "grad_norm": 0.06381058610771635,
      "learning_rate": 0.00010625342570025294,
      "loss": 0.8513,
      "step": 140530
    },
    {
      "epoch": 2.467388823539739,
      "grad_norm": 0.055628125595396545,
      "learning_rate": 0.00010624229773500285,
      "loss": 0.8488,
      "step": 140540
    },
    {
      "epoch": 2.4675643884197407,
      "grad_norm": 0.04716913633532676,
      "learning_rate": 0.00010623116970333182,
      "loss": 0.8488,
      "step": 140550
    },
    {
      "epoch": 2.467739953299742,
      "grad_norm": 0.06320933744723864,
      "learning_rate": 0.00010622004160538087,
      "loss": 0.8522,
      "step": 140560
    },
    {
      "epoch": 2.467915518179743,
      "grad_norm": 0.045697785736836555,
      "learning_rate": 0.000106208913441291,
      "loss": 0.8417,
      "step": 140570
    },
    {
      "epoch": 2.4680910830597447,
      "grad_norm": 0.04705346592468497,
      "learning_rate": 0.0001061977852112032,
      "loss": 0.838,
      "step": 140580
    },
    {
      "epoch": 2.468266647939746,
      "grad_norm": 0.05529540547491532,
      "learning_rate": 0.00010618665691525851,
      "loss": 0.8418,
      "step": 140590
    },
    {
      "epoch": 2.4684422128197476,
      "grad_norm": 0.06581325951316432,
      "learning_rate": 0.00010617552855359783,
      "loss": 0.8384,
      "step": 140600
    },
    {
      "epoch": 2.468617777699749,
      "grad_norm": 0.054684438943840355,
      "learning_rate": 0.00010616440012636221,
      "loss": 0.8449,
      "step": 140610
    },
    {
      "epoch": 2.46879334257975,
      "grad_norm": 0.08115973180295742,
      "learning_rate": 0.00010615327163369267,
      "loss": 0.8439,
      "step": 140620
    },
    {
      "epoch": 2.4689689074597516,
      "grad_norm": 0.061246944943384306,
      "learning_rate": 0.00010614214307573018,
      "loss": 0.8409,
      "step": 140630
    },
    {
      "epoch": 2.469144472339753,
      "grad_norm": 0.07187081484673698,
      "learning_rate": 0.00010613101445261577,
      "loss": 0.845,
      "step": 140640
    },
    {
      "epoch": 2.4693200372197546,
      "grad_norm": 0.05275786540244002,
      "learning_rate": 0.0001061198857644904,
      "loss": 0.8425,
      "step": 140650
    },
    {
      "epoch": 2.469495602099756,
      "grad_norm": 0.0582187346093273,
      "learning_rate": 0.00010610875701149512,
      "loss": 0.8487,
      "step": 140660
    },
    {
      "epoch": 2.4696711669797575,
      "grad_norm": 0.10384059066644377,
      "learning_rate": 0.00010609762819377091,
      "loss": 0.839,
      "step": 140670
    },
    {
      "epoch": 2.469846731859759,
      "grad_norm": 0.049092745960341404,
      "learning_rate": 0.00010608649931145873,
      "loss": 0.8462,
      "step": 140680
    },
    {
      "epoch": 2.47002229673976,
      "grad_norm": 0.0461885101880256,
      "learning_rate": 0.00010607537036469967,
      "loss": 0.8449,
      "step": 140690
    },
    {
      "epoch": 2.4701978616197615,
      "grad_norm": 0.0671182150078732,
      "learning_rate": 0.0001060642413536347,
      "loss": 0.8389,
      "step": 140700
    },
    {
      "epoch": 2.470373426499763,
      "grad_norm": 0.056659454820161055,
      "learning_rate": 0.00010605311227840478,
      "loss": 0.843,
      "step": 140710
    },
    {
      "epoch": 2.4705489913797645,
      "grad_norm": 0.04209201533564868,
      "learning_rate": 0.00010604198313915102,
      "loss": 0.8356,
      "step": 140720
    },
    {
      "epoch": 2.470724556259766,
      "grad_norm": 0.05953341310483423,
      "learning_rate": 0.00010603085393601428,
      "loss": 0.8449,
      "step": 140730
    },
    {
      "epoch": 2.470900121139767,
      "grad_norm": 0.06480318883827578,
      "learning_rate": 0.00010601972466913572,
      "loss": 0.8397,
      "step": 140740
    },
    {
      "epoch": 2.4710756860197685,
      "grad_norm": 0.052237210709336644,
      "learning_rate": 0.00010600859533865628,
      "loss": 0.8442,
      "step": 140750
    },
    {
      "epoch": 2.47125125089977,
      "grad_norm": 0.05479178833347863,
      "learning_rate": 0.00010599746594471698,
      "loss": 0.8387,
      "step": 140760
    },
    {
      "epoch": 2.4714268157797714,
      "grad_norm": 0.05997774240027584,
      "learning_rate": 0.00010598633648745884,
      "loss": 0.8503,
      "step": 140770
    },
    {
      "epoch": 2.471602380659773,
      "grad_norm": 0.06467788365725892,
      "learning_rate": 0.00010597520696702281,
      "loss": 0.8477,
      "step": 140780
    },
    {
      "epoch": 2.4717779455397744,
      "grad_norm": 0.05573733461502403,
      "learning_rate": 0.00010596407738354998,
      "loss": 0.8419,
      "step": 140790
    },
    {
      "epoch": 2.471953510419776,
      "grad_norm": 0.05514512088645426,
      "learning_rate": 0.00010595294773718134,
      "loss": 0.8466,
      "step": 140800
    },
    {
      "epoch": 2.472129075299777,
      "grad_norm": 0.05504376676220845,
      "learning_rate": 0.00010594181802805794,
      "loss": 0.8388,
      "step": 140810
    },
    {
      "epoch": 2.4723046401797784,
      "grad_norm": 0.04338916309511015,
      "learning_rate": 0.00010593068825632071,
      "loss": 0.8396,
      "step": 140820
    },
    {
      "epoch": 2.47248020505978,
      "grad_norm": 0.07260980626060104,
      "learning_rate": 0.00010591955842211072,
      "loss": 0.8493,
      "step": 140830
    },
    {
      "epoch": 2.4726557699397813,
      "grad_norm": 0.06604089493741674,
      "learning_rate": 0.000105908428525569,
      "loss": 0.8462,
      "step": 140840
    },
    {
      "epoch": 2.472831334819783,
      "grad_norm": 0.04499487087626009,
      "learning_rate": 0.00010589729856683652,
      "loss": 0.8488,
      "step": 140850
    },
    {
      "epoch": 2.473006899699784,
      "grad_norm": 0.053754727000872,
      "learning_rate": 0.00010588616854605437,
      "loss": 0.8415,
      "step": 140860
    },
    {
      "epoch": 2.4731824645797853,
      "grad_norm": 0.065711081364765,
      "learning_rate": 0.00010587503846336353,
      "loss": 0.8436,
      "step": 140870
    },
    {
      "epoch": 2.4733580294597868,
      "grad_norm": 0.07720108047035099,
      "learning_rate": 0.00010586390831890498,
      "loss": 0.8484,
      "step": 140880
    },
    {
      "epoch": 2.4735335943397883,
      "grad_norm": 0.0846952594508885,
      "learning_rate": 0.0001058527781128198,
      "loss": 0.8451,
      "step": 140890
    },
    {
      "epoch": 2.4737091592197897,
      "grad_norm": 0.06282758156617851,
      "learning_rate": 0.00010584164784524897,
      "loss": 0.8444,
      "step": 140900
    },
    {
      "epoch": 2.473884724099791,
      "grad_norm": 0.06111602007508314,
      "learning_rate": 0.00010583051751633355,
      "loss": 0.8507,
      "step": 140910
    },
    {
      "epoch": 2.4740602889797927,
      "grad_norm": 0.06329279421229739,
      "learning_rate": 0.00010581938712621456,
      "loss": 0.8522,
      "step": 140920
    },
    {
      "epoch": 2.4742358538597937,
      "grad_norm": 0.057021633765020965,
      "learning_rate": 0.00010580825667503301,
      "loss": 0.8442,
      "step": 140930
    },
    {
      "epoch": 2.474411418739795,
      "grad_norm": 0.05589839633214291,
      "learning_rate": 0.00010579712616292992,
      "loss": 0.8436,
      "step": 140940
    },
    {
      "epoch": 2.4745869836197967,
      "grad_norm": 0.06467841074617604,
      "learning_rate": 0.0001057859955900463,
      "loss": 0.8425,
      "step": 140950
    },
    {
      "epoch": 2.474762548499798,
      "grad_norm": 0.05650440382613081,
      "learning_rate": 0.0001057748649565232,
      "loss": 0.8409,
      "step": 140960
    },
    {
      "epoch": 2.4749381133797996,
      "grad_norm": 0.06435604600945766,
      "learning_rate": 0.00010576373426250168,
      "loss": 0.8414,
      "step": 140970
    },
    {
      "epoch": 2.4751136782598007,
      "grad_norm": 0.0556899292667784,
      "learning_rate": 0.00010575260350812271,
      "loss": 0.8381,
      "step": 140980
    },
    {
      "epoch": 2.475289243139802,
      "grad_norm": 0.05651885265866803,
      "learning_rate": 0.00010574147269352736,
      "loss": 0.8462,
      "step": 140990
    },
    {
      "epoch": 2.4754648080198036,
      "grad_norm": 0.05427047039667237,
      "learning_rate": 0.00010573034181885663,
      "loss": 0.8417,
      "step": 141000
    },
    {
      "epoch": 2.475640372899805,
      "grad_norm": 0.059441529982651814,
      "learning_rate": 0.00010571921088425158,
      "loss": 0.8536,
      "step": 141010
    },
    {
      "epoch": 2.4758159377798066,
      "grad_norm": 0.048262839928189256,
      "learning_rate": 0.00010570807988985321,
      "loss": 0.8447,
      "step": 141020
    },
    {
      "epoch": 2.475991502659808,
      "grad_norm": 0.07994261724134998,
      "learning_rate": 0.00010569694883580257,
      "loss": 0.8441,
      "step": 141030
    },
    {
      "epoch": 2.4761670675398095,
      "grad_norm": 0.04298441126560902,
      "learning_rate": 0.00010568581772224068,
      "loss": 0.8437,
      "step": 141040
    },
    {
      "epoch": 2.4763426324198106,
      "grad_norm": 0.05987272350899147,
      "learning_rate": 0.00010567468654930857,
      "loss": 0.849,
      "step": 141050
    },
    {
      "epoch": 2.476518197299812,
      "grad_norm": 0.061944943996116414,
      "learning_rate": 0.0001056635553171473,
      "loss": 0.8512,
      "step": 141060
    },
    {
      "epoch": 2.4766937621798135,
      "grad_norm": 0.054952410326743556,
      "learning_rate": 0.00010565242402589787,
      "loss": 0.8312,
      "step": 141070
    },
    {
      "epoch": 2.476869327059815,
      "grad_norm": 0.056759187413332704,
      "learning_rate": 0.00010564129267570136,
      "loss": 0.8505,
      "step": 141080
    },
    {
      "epoch": 2.4770448919398165,
      "grad_norm": 0.07770775299613315,
      "learning_rate": 0.00010563016126669878,
      "loss": 0.8462,
      "step": 141090
    },
    {
      "epoch": 2.477220456819818,
      "grad_norm": 0.049751009824943945,
      "learning_rate": 0.00010561902979903118,
      "loss": 0.8413,
      "step": 141100
    },
    {
      "epoch": 2.477396021699819,
      "grad_norm": 0.043105301079509224,
      "learning_rate": 0.00010560789827283957,
      "loss": 0.8377,
      "step": 141110
    },
    {
      "epoch": 2.4775715865798205,
      "grad_norm": 0.0467546724067557,
      "learning_rate": 0.000105596766688265,
      "loss": 0.8399,
      "step": 141120
    },
    {
      "epoch": 2.477747151459822,
      "grad_norm": 0.05514519582483944,
      "learning_rate": 0.00010558563504544852,
      "loss": 0.847,
      "step": 141130
    },
    {
      "epoch": 2.4779227163398234,
      "grad_norm": 0.06830067806174461,
      "learning_rate": 0.00010557450334453118,
      "loss": 0.8504,
      "step": 141140
    },
    {
      "epoch": 2.478098281219825,
      "grad_norm": 0.061114882959954674,
      "learning_rate": 0.00010556337158565399,
      "loss": 0.8412,
      "step": 141150
    },
    {
      "epoch": 2.4782738460998264,
      "grad_norm": 0.053193481611393784,
      "learning_rate": 0.00010555223976895803,
      "loss": 0.848,
      "step": 141160
    },
    {
      "epoch": 2.4784494109798274,
      "grad_norm": 0.04315212452047698,
      "learning_rate": 0.0001055411078945843,
      "loss": 0.8511,
      "step": 141170
    },
    {
      "epoch": 2.478624975859829,
      "grad_norm": 0.04978606686372905,
      "learning_rate": 0.00010552997596267386,
      "loss": 0.8476,
      "step": 141180
    },
    {
      "epoch": 2.4788005407398304,
      "grad_norm": 0.0515367665537428,
      "learning_rate": 0.00010551884397336777,
      "loss": 0.845,
      "step": 141190
    },
    {
      "epoch": 2.478976105619832,
      "grad_norm": 0.05362264688854677,
      "learning_rate": 0.00010550771192680703,
      "loss": 0.8497,
      "step": 141200
    },
    {
      "epoch": 2.4791516704998333,
      "grad_norm": 0.05370879271061718,
      "learning_rate": 0.00010549657982313277,
      "loss": 0.8455,
      "step": 141210
    },
    {
      "epoch": 2.479327235379835,
      "grad_norm": 0.05048036647910976,
      "learning_rate": 0.00010548544766248596,
      "loss": 0.8435,
      "step": 141220
    },
    {
      "epoch": 2.4795028002598363,
      "grad_norm": 0.04682822215102697,
      "learning_rate": 0.00010547431544500763,
      "loss": 0.8406,
      "step": 141230
    },
    {
      "epoch": 2.4796783651398373,
      "grad_norm": 0.07127245120496307,
      "learning_rate": 0.00010546318317083889,
      "loss": 0.8385,
      "step": 141240
    },
    {
      "epoch": 2.479853930019839,
      "grad_norm": 0.053332583429102985,
      "learning_rate": 0.00010545205084012078,
      "loss": 0.8346,
      "step": 141250
    },
    {
      "epoch": 2.4800294948998403,
      "grad_norm": 0.06590022288037752,
      "learning_rate": 0.0001054409184529943,
      "loss": 0.8404,
      "step": 141260
    },
    {
      "epoch": 2.4802050597798417,
      "grad_norm": 0.05448751764938702,
      "learning_rate": 0.00010542978600960058,
      "loss": 0.8429,
      "step": 141270
    },
    {
      "epoch": 2.480380624659843,
      "grad_norm": 0.054737463745841715,
      "learning_rate": 0.0001054186535100806,
      "loss": 0.8437,
      "step": 141280
    },
    {
      "epoch": 2.4805561895398442,
      "grad_norm": 0.05178535694330406,
      "learning_rate": 0.00010540752095457542,
      "loss": 0.8446,
      "step": 141290
    },
    {
      "epoch": 2.4807317544198457,
      "grad_norm": 0.04882629512268577,
      "learning_rate": 0.00010539638834322611,
      "loss": 0.8346,
      "step": 141300
    },
    {
      "epoch": 2.480907319299847,
      "grad_norm": 0.049036567381972956,
      "learning_rate": 0.00010538525567617372,
      "loss": 0.8392,
      "step": 141310
    },
    {
      "epoch": 2.4810828841798487,
      "grad_norm": 0.04670606965632068,
      "learning_rate": 0.00010537412295355933,
      "loss": 0.8472,
      "step": 141320
    },
    {
      "epoch": 2.48125844905985,
      "grad_norm": 0.05167211926943529,
      "learning_rate": 0.00010536299017552393,
      "loss": 0.8407,
      "step": 141330
    },
    {
      "epoch": 2.4814340139398516,
      "grad_norm": 0.04943696673748868,
      "learning_rate": 0.00010535185734220861,
      "loss": 0.8429,
      "step": 141340
    },
    {
      "epoch": 2.481609578819853,
      "grad_norm": 0.06851660649862971,
      "learning_rate": 0.00010534072445375442,
      "loss": 0.8487,
      "step": 141350
    },
    {
      "epoch": 2.481785143699854,
      "grad_norm": 0.058449668045541456,
      "learning_rate": 0.00010532959151030244,
      "loss": 0.8452,
      "step": 141360
    },
    {
      "epoch": 2.4819607085798556,
      "grad_norm": 0.048919557138296925,
      "learning_rate": 0.00010531845851199367,
      "loss": 0.8408,
      "step": 141370
    },
    {
      "epoch": 2.482136273459857,
      "grad_norm": 0.07144592670136457,
      "learning_rate": 0.00010530732545896926,
      "loss": 0.847,
      "step": 141380
    },
    {
      "epoch": 2.4823118383398586,
      "grad_norm": 0.058942689103166794,
      "learning_rate": 0.00010529619235137013,
      "loss": 0.8412,
      "step": 141390
    },
    {
      "epoch": 2.48248740321986,
      "grad_norm": 0.05648432069405592,
      "learning_rate": 0.0001052850591893375,
      "loss": 0.8443,
      "step": 141400
    },
    {
      "epoch": 2.482662968099861,
      "grad_norm": 0.05877827191470621,
      "learning_rate": 0.00010527392597301229,
      "loss": 0.8421,
      "step": 141410
    },
    {
      "epoch": 2.4828385329798626,
      "grad_norm": 0.07137333275941961,
      "learning_rate": 0.00010526279270253563,
      "loss": 0.8471,
      "step": 141420
    },
    {
      "epoch": 2.483014097859864,
      "grad_norm": 0.053825859202638854,
      "learning_rate": 0.0001052516593780486,
      "loss": 0.8395,
      "step": 141430
    },
    {
      "epoch": 2.4831896627398655,
      "grad_norm": 0.05977659088501342,
      "learning_rate": 0.00010524052599969221,
      "loss": 0.8378,
      "step": 141440
    },
    {
      "epoch": 2.483365227619867,
      "grad_norm": 0.048555321826256215,
      "learning_rate": 0.00010522939256760758,
      "loss": 0.8465,
      "step": 141450
    },
    {
      "epoch": 2.4835407924998685,
      "grad_norm": 0.058678793088739134,
      "learning_rate": 0.00010521825908193568,
      "loss": 0.8495,
      "step": 141460
    },
    {
      "epoch": 2.48371635737987,
      "grad_norm": 0.07031408733843651,
      "learning_rate": 0.00010520712554281765,
      "loss": 0.8538,
      "step": 141470
    },
    {
      "epoch": 2.483891922259871,
      "grad_norm": 0.05297499022720643,
      "learning_rate": 0.00010519599195039453,
      "loss": 0.8393,
      "step": 141480
    },
    {
      "epoch": 2.4840674871398725,
      "grad_norm": 0.061894619964784,
      "learning_rate": 0.0001051848583048074,
      "loss": 0.8454,
      "step": 141490
    },
    {
      "epoch": 2.484243052019874,
      "grad_norm": 0.1058851132553602,
      "learning_rate": 0.0001051737246061973,
      "loss": 0.8397,
      "step": 141500
    },
    {
      "epoch": 2.4844186168998754,
      "grad_norm": 0.05070932117733203,
      "learning_rate": 0.00010516259085470533,
      "loss": 0.8395,
      "step": 141510
    },
    {
      "epoch": 2.484594181779877,
      "grad_norm": 0.05252368029975073,
      "learning_rate": 0.00010515145705047254,
      "loss": 0.8384,
      "step": 141520
    },
    {
      "epoch": 2.484769746659878,
      "grad_norm": 0.04977653668167214,
      "learning_rate": 0.00010514032319363993,
      "loss": 0.8486,
      "step": 141530
    },
    {
      "epoch": 2.4849453115398794,
      "grad_norm": 0.05886931517272005,
      "learning_rate": 0.00010512918928434867,
      "loss": 0.8478,
      "step": 141540
    },
    {
      "epoch": 2.485120876419881,
      "grad_norm": 0.058903552687155455,
      "learning_rate": 0.00010511805532273984,
      "loss": 0.8442,
      "step": 141550
    },
    {
      "epoch": 2.4852964412998824,
      "grad_norm": 0.051702075964993995,
      "learning_rate": 0.00010510692130895439,
      "loss": 0.8498,
      "step": 141560
    },
    {
      "epoch": 2.485472006179884,
      "grad_norm": 0.061092203925222764,
      "learning_rate": 0.0001050957872431335,
      "loss": 0.8435,
      "step": 141570
    },
    {
      "epoch": 2.4856475710598853,
      "grad_norm": 0.056700786199733164,
      "learning_rate": 0.00010508465312541818,
      "loss": 0.8431,
      "step": 141580
    },
    {
      "epoch": 2.485823135939887,
      "grad_norm": 0.07285358608374154,
      "learning_rate": 0.00010507351895594952,
      "loss": 0.844,
      "step": 141590
    },
    {
      "epoch": 2.485998700819888,
      "grad_norm": 0.04829673219344612,
      "learning_rate": 0.00010506238473486858,
      "loss": 0.842,
      "step": 141600
    },
    {
      "epoch": 2.4861742656998893,
      "grad_norm": 0.05536771182411283,
      "learning_rate": 0.0001050512504623165,
      "loss": 0.8481,
      "step": 141610
    },
    {
      "epoch": 2.486349830579891,
      "grad_norm": 0.07831034003997578,
      "learning_rate": 0.00010504011613843428,
      "loss": 0.8443,
      "step": 141620
    },
    {
      "epoch": 2.4865253954598923,
      "grad_norm": 0.06102218312137498,
      "learning_rate": 0.00010502898176336297,
      "loss": 0.8464,
      "step": 141630
    },
    {
      "epoch": 2.4867009603398937,
      "grad_norm": 0.04953658750584508,
      "learning_rate": 0.00010501784733724373,
      "loss": 0.8529,
      "step": 141640
    },
    {
      "epoch": 2.4868765252198948,
      "grad_norm": 0.05240776350286182,
      "learning_rate": 0.0001050067128602176,
      "loss": 0.8476,
      "step": 141650
    },
    {
      "epoch": 2.4870520900998963,
      "grad_norm": 0.05163337646103805,
      "learning_rate": 0.00010499557833242564,
      "loss": 0.8472,
      "step": 141660
    },
    {
      "epoch": 2.4872276549798977,
      "grad_norm": 0.05749833432983265,
      "learning_rate": 0.00010498444375400893,
      "loss": 0.8506,
      "step": 141670
    },
    {
      "epoch": 2.487403219859899,
      "grad_norm": 0.051480024393017104,
      "learning_rate": 0.00010497330912510857,
      "loss": 0.8451,
      "step": 141680
    },
    {
      "epoch": 2.4875787847399007,
      "grad_norm": 0.06899398663785598,
      "learning_rate": 0.00010496217444586562,
      "loss": 0.8417,
      "step": 141690
    },
    {
      "epoch": 2.487754349619902,
      "grad_norm": 0.057326765551918804,
      "learning_rate": 0.00010495103971642113,
      "loss": 0.843,
      "step": 141700
    },
    {
      "epoch": 2.4879299144999036,
      "grad_norm": 0.07227280375462652,
      "learning_rate": 0.00010493990493691626,
      "loss": 0.8496,
      "step": 141710
    },
    {
      "epoch": 2.4881054793799047,
      "grad_norm": 0.053331212678263164,
      "learning_rate": 0.00010492877010749199,
      "loss": 0.8422,
      "step": 141720
    },
    {
      "epoch": 2.488281044259906,
      "grad_norm": 0.04690911920317232,
      "learning_rate": 0.00010491763522828948,
      "loss": 0.8473,
      "step": 141730
    },
    {
      "epoch": 2.4884566091399076,
      "grad_norm": 0.05467108831596971,
      "learning_rate": 0.00010490650029944975,
      "loss": 0.8449,
      "step": 141740
    },
    {
      "epoch": 2.488632174019909,
      "grad_norm": 0.046484146120907004,
      "learning_rate": 0.00010489536532111393,
      "loss": 0.8456,
      "step": 141750
    },
    {
      "epoch": 2.4888077388999106,
      "grad_norm": 0.05034361195285806,
      "learning_rate": 0.00010488423029342306,
      "loss": 0.8394,
      "step": 141760
    },
    {
      "epoch": 2.488983303779912,
      "grad_norm": 0.057409100848781994,
      "learning_rate": 0.00010487309521651829,
      "loss": 0.8379,
      "step": 141770
    },
    {
      "epoch": 2.489158868659913,
      "grad_norm": 0.0508968444999565,
      "learning_rate": 0.00010486196009054068,
      "loss": 0.8476,
      "step": 141780
    },
    {
      "epoch": 2.4893344335399146,
      "grad_norm": 0.0742680209460331,
      "learning_rate": 0.00010485082491563127,
      "loss": 0.846,
      "step": 141790
    },
    {
      "epoch": 2.489509998419916,
      "grad_norm": 0.1304869517551756,
      "learning_rate": 0.00010483968969193115,
      "loss": 0.847,
      "step": 141800
    },
    {
      "epoch": 2.4896855632999175,
      "grad_norm": 0.09357684290269677,
      "learning_rate": 0.00010482855441958143,
      "loss": 0.8425,
      "step": 141810
    },
    {
      "epoch": 2.489861128179919,
      "grad_norm": 0.044414598500096596,
      "learning_rate": 0.0001048174190987232,
      "loss": 0.8394,
      "step": 141820
    },
    {
      "epoch": 2.4900366930599205,
      "grad_norm": 0.06088374474657002,
      "learning_rate": 0.00010480628372949755,
      "loss": 0.8491,
      "step": 141830
    },
    {
      "epoch": 2.4902122579399215,
      "grad_norm": 0.06421947097970253,
      "learning_rate": 0.00010479514831204555,
      "loss": 0.8406,
      "step": 141840
    },
    {
      "epoch": 2.490387822819923,
      "grad_norm": 0.057555405683508296,
      "learning_rate": 0.00010478401284650827,
      "loss": 0.8409,
      "step": 141850
    },
    {
      "epoch": 2.4905633876999245,
      "grad_norm": 0.056023966110684165,
      "learning_rate": 0.00010477287733302686,
      "loss": 0.839,
      "step": 141860
    },
    {
      "epoch": 2.490738952579926,
      "grad_norm": 0.06268569362521197,
      "learning_rate": 0.00010476174177174232,
      "loss": 0.8451,
      "step": 141870
    },
    {
      "epoch": 2.4909145174599274,
      "grad_norm": 0.04949981440752606,
      "learning_rate": 0.00010475060616279585,
      "loss": 0.8474,
      "step": 141880
    },
    {
      "epoch": 2.491090082339929,
      "grad_norm": 0.07426592546754071,
      "learning_rate": 0.00010473947050632847,
      "loss": 0.8467,
      "step": 141890
    },
    {
      "epoch": 2.49126564721993,
      "grad_norm": 0.05736928363324326,
      "learning_rate": 0.00010472833480248125,
      "loss": 0.8461,
      "step": 141900
    },
    {
      "epoch": 2.4914412120999314,
      "grad_norm": 0.05464208418813337,
      "learning_rate": 0.00010471719905139534,
      "loss": 0.8472,
      "step": 141910
    },
    {
      "epoch": 2.491616776979933,
      "grad_norm": 0.05309799711449186,
      "learning_rate": 0.00010470606325321177,
      "loss": 0.843,
      "step": 141920
    },
    {
      "epoch": 2.4917923418599344,
      "grad_norm": 0.05290438307295485,
      "learning_rate": 0.00010469492740807167,
      "loss": 0.8453,
      "step": 141930
    },
    {
      "epoch": 2.491967906739936,
      "grad_norm": 0.05835679355655987,
      "learning_rate": 0.00010468379151611616,
      "loss": 0.8439,
      "step": 141940
    },
    {
      "epoch": 2.4921434716199373,
      "grad_norm": 0.058849418620705383,
      "learning_rate": 0.00010467265557748632,
      "loss": 0.8538,
      "step": 141950
    },
    {
      "epoch": 2.4923190364999384,
      "grad_norm": 0.07309043041961152,
      "learning_rate": 0.00010466151959232322,
      "loss": 0.839,
      "step": 141960
    },
    {
      "epoch": 2.49249460137994,
      "grad_norm": 0.041047442527093234,
      "learning_rate": 0.00010465038356076795,
      "loss": 0.8457,
      "step": 141970
    },
    {
      "epoch": 2.4926701662599413,
      "grad_norm": 0.058291872669000946,
      "learning_rate": 0.00010463924748296161,
      "loss": 0.8461,
      "step": 141980
    },
    {
      "epoch": 2.492845731139943,
      "grad_norm": 0.06546787291502906,
      "learning_rate": 0.0001046281113590453,
      "loss": 0.8435,
      "step": 141990
    },
    {
      "epoch": 2.4930212960199443,
      "grad_norm": 0.05267869889204819,
      "learning_rate": 0.00010461697518916018,
      "loss": 0.8368,
      "step": 142000
    },
    {
      "epoch": 2.4931968608999457,
      "grad_norm": 0.05474825289738824,
      "learning_rate": 0.00010460583897344727,
      "loss": 0.8501,
      "step": 142010
    },
    {
      "epoch": 2.493372425779947,
      "grad_norm": 0.04925234422680641,
      "learning_rate": 0.00010459470271204766,
      "loss": 0.8531,
      "step": 142020
    },
    {
      "epoch": 2.4935479906599483,
      "grad_norm": 0.04651521177890632,
      "learning_rate": 0.00010458356640510252,
      "loss": 0.8457,
      "step": 142030
    },
    {
      "epoch": 2.4937235555399497,
      "grad_norm": 0.05699343007551018,
      "learning_rate": 0.00010457243005275285,
      "loss": 0.8399,
      "step": 142040
    },
    {
      "epoch": 2.493899120419951,
      "grad_norm": 0.06505259454110476,
      "learning_rate": 0.00010456129365513984,
      "loss": 0.84,
      "step": 142050
    },
    {
      "epoch": 2.4940746852999527,
      "grad_norm": 0.05673080852467737,
      "learning_rate": 0.00010455015721240456,
      "loss": 0.8378,
      "step": 142060
    },
    {
      "epoch": 2.494250250179954,
      "grad_norm": 0.052708267239213805,
      "learning_rate": 0.00010453902072468809,
      "loss": 0.8412,
      "step": 142070
    },
    {
      "epoch": 2.494425815059955,
      "grad_norm": 0.060724540356128384,
      "learning_rate": 0.00010452788419213158,
      "loss": 0.8454,
      "step": 142080
    },
    {
      "epoch": 2.4946013799399567,
      "grad_norm": 0.07733627610369524,
      "learning_rate": 0.00010451674761487607,
      "loss": 0.8378,
      "step": 142090
    },
    {
      "epoch": 2.494776944819958,
      "grad_norm": 0.0645278853624601,
      "learning_rate": 0.00010450561099306268,
      "loss": 0.8496,
      "step": 142100
    },
    {
      "epoch": 2.4949525096999596,
      "grad_norm": 0.05421089871574251,
      "learning_rate": 0.00010449447432683256,
      "loss": 0.8451,
      "step": 142110
    },
    {
      "epoch": 2.495128074579961,
      "grad_norm": 0.05614927869641993,
      "learning_rate": 0.0001044833376163268,
      "loss": 0.8399,
      "step": 142120
    },
    {
      "epoch": 2.4953036394599626,
      "grad_norm": 0.06364964037771509,
      "learning_rate": 0.00010447220086168645,
      "loss": 0.8424,
      "step": 142130
    },
    {
      "epoch": 2.495479204339964,
      "grad_norm": 0.06680214731601943,
      "learning_rate": 0.00010446106406305265,
      "loss": 0.8474,
      "step": 142140
    },
    {
      "epoch": 2.495654769219965,
      "grad_norm": 0.04942739663474548,
      "learning_rate": 0.00010444992722056652,
      "loss": 0.8452,
      "step": 142150
    },
    {
      "epoch": 2.4958303340999666,
      "grad_norm": 0.05675780846077037,
      "learning_rate": 0.00010443879033436914,
      "loss": 0.8514,
      "step": 142160
    },
    {
      "epoch": 2.496005898979968,
      "grad_norm": 0.06864216851061107,
      "learning_rate": 0.00010442765340460163,
      "loss": 0.8456,
      "step": 142170
    },
    {
      "epoch": 2.4961814638599695,
      "grad_norm": 0.050860233743889025,
      "learning_rate": 0.00010441651643140511,
      "loss": 0.8395,
      "step": 142180
    },
    {
      "epoch": 2.496357028739971,
      "grad_norm": 0.06389000823511705,
      "learning_rate": 0.00010440537941492064,
      "loss": 0.8506,
      "step": 142190
    },
    {
      "epoch": 2.496532593619972,
      "grad_norm": 0.05647017971916099,
      "learning_rate": 0.00010439424235528938,
      "loss": 0.8432,
      "step": 142200
    },
    {
      "epoch": 2.4967081584999735,
      "grad_norm": 0.05111701388427384,
      "learning_rate": 0.00010438310525265243,
      "loss": 0.8433,
      "step": 142210
    },
    {
      "epoch": 2.496883723379975,
      "grad_norm": 0.053502953813700814,
      "learning_rate": 0.00010437196810715086,
      "loss": 0.8463,
      "step": 142220
    },
    {
      "epoch": 2.4970592882599765,
      "grad_norm": 0.05337880583213217,
      "learning_rate": 0.00010436083091892585,
      "loss": 0.8494,
      "step": 142230
    },
    {
      "epoch": 2.497234853139978,
      "grad_norm": 0.04092303648713773,
      "learning_rate": 0.00010434969368811842,
      "loss": 0.8492,
      "step": 142240
    },
    {
      "epoch": 2.4974104180199794,
      "grad_norm": 0.045957268689319125,
      "learning_rate": 0.00010433855641486976,
      "loss": 0.8394,
      "step": 142250
    },
    {
      "epoch": 2.497585982899981,
      "grad_norm": 0.04907324937328052,
      "learning_rate": 0.00010432741909932093,
      "loss": 0.8441,
      "step": 142260
    },
    {
      "epoch": 2.497761547779982,
      "grad_norm": 0.04700952308354452,
      "learning_rate": 0.00010431628174161305,
      "loss": 0.842,
      "step": 142270
    },
    {
      "epoch": 2.4979371126599834,
      "grad_norm": 0.08220551903276717,
      "learning_rate": 0.00010430514434188728,
      "loss": 0.84,
      "step": 142280
    },
    {
      "epoch": 2.498112677539985,
      "grad_norm": 0.08191950755686028,
      "learning_rate": 0.00010429400690028468,
      "loss": 0.84,
      "step": 142290
    },
    {
      "epoch": 2.4982882424199864,
      "grad_norm": 0.052410491442479196,
      "learning_rate": 0.00010428286941694642,
      "loss": 0.8431,
      "step": 142300
    },
    {
      "epoch": 2.498463807299988,
      "grad_norm": 0.06969151232710846,
      "learning_rate": 0.00010427173189201354,
      "loss": 0.8376,
      "step": 142310
    },
    {
      "epoch": 2.498639372179989,
      "grad_norm": 0.06032139152171676,
      "learning_rate": 0.00010426059432562716,
      "loss": 0.8494,
      "step": 142320
    },
    {
      "epoch": 2.4988149370599904,
      "grad_norm": 0.050807264956958885,
      "learning_rate": 0.00010424945671792848,
      "loss": 0.835,
      "step": 142330
    },
    {
      "epoch": 2.498990501939992,
      "grad_norm": 0.051323001522319306,
      "learning_rate": 0.00010423831906905851,
      "loss": 0.8405,
      "step": 142340
    },
    {
      "epoch": 2.4991660668199933,
      "grad_norm": 0.060196482442892454,
      "learning_rate": 0.00010422718137915848,
      "loss": 0.8389,
      "step": 142350
    },
    {
      "epoch": 2.499341631699995,
      "grad_norm": 0.060034011378967314,
      "learning_rate": 0.00010421604364836939,
      "loss": 0.8451,
      "step": 142360
    },
    {
      "epoch": 2.4995171965799963,
      "grad_norm": 0.05381261118122093,
      "learning_rate": 0.00010420490587683244,
      "loss": 0.8424,
      "step": 142370
    },
    {
      "epoch": 2.4996927614599977,
      "grad_norm": 0.06091959945379747,
      "learning_rate": 0.00010419376806468868,
      "loss": 0.8421,
      "step": 142380
    },
    {
      "epoch": 2.499868326339999,
      "grad_norm": 0.05239617074176352,
      "learning_rate": 0.0001041826302120793,
      "loss": 0.8405,
      "step": 142390
    },
    {
      "epoch": 2.5000438912200003,
      "grad_norm": 0.055557674307425864,
      "learning_rate": 0.0001041714923191454,
      "loss": 0.8474,
      "step": 142400
    },
    {
      "epoch": 2.5002194561000017,
      "grad_norm": 0.0482929592092805,
      "learning_rate": 0.00010416035438602804,
      "loss": 0.8502,
      "step": 142410
    },
    {
      "epoch": 2.500395020980003,
      "grad_norm": 0.05050958514265602,
      "learning_rate": 0.00010414921641286841,
      "loss": 0.8463,
      "step": 142420
    },
    {
      "epoch": 2.5005705858600047,
      "grad_norm": 0.048200336744174906,
      "learning_rate": 0.0001041380783998076,
      "loss": 0.8446,
      "step": 142430
    },
    {
      "epoch": 2.5007461507400057,
      "grad_norm": 0.058710630735546866,
      "learning_rate": 0.00010412694034698673,
      "loss": 0.8425,
      "step": 142440
    },
    {
      "epoch": 2.5009217156200076,
      "grad_norm": 0.058655749321967945,
      "learning_rate": 0.00010411580225454694,
      "loss": 0.8547,
      "step": 142450
    },
    {
      "epoch": 2.5010972805000087,
      "grad_norm": 0.067835974452207,
      "learning_rate": 0.00010410466412262933,
      "loss": 0.8429,
      "step": 142460
    },
    {
      "epoch": 2.50127284538001,
      "grad_norm": 0.049072308401928146,
      "learning_rate": 0.00010409352595137507,
      "loss": 0.8511,
      "step": 142470
    },
    {
      "epoch": 2.5014484102600116,
      "grad_norm": 0.048639282001247924,
      "learning_rate": 0.00010408238774092518,
      "loss": 0.8536,
      "step": 142480
    },
    {
      "epoch": 2.501623975140013,
      "grad_norm": 0.05043195079806146,
      "learning_rate": 0.00010407124949142086,
      "loss": 0.8485,
      "step": 142490
    },
    {
      "epoch": 2.5017995400200146,
      "grad_norm": 0.06178616927742885,
      "learning_rate": 0.00010406011120300326,
      "loss": 0.8524,
      "step": 142500
    },
    {
      "epoch": 2.5019751049000156,
      "grad_norm": 0.06873577295154092,
      "learning_rate": 0.00010404897287581343,
      "loss": 0.8448,
      "step": 142510
    },
    {
      "epoch": 2.502150669780017,
      "grad_norm": 0.0630658956141933,
      "learning_rate": 0.00010403783450999254,
      "loss": 0.8454,
      "step": 142520
    },
    {
      "epoch": 2.5023262346600186,
      "grad_norm": 0.05107508657042763,
      "learning_rate": 0.00010402669610568169,
      "loss": 0.8373,
      "step": 142530
    },
    {
      "epoch": 2.50250179954002,
      "grad_norm": 0.050393969471772326,
      "learning_rate": 0.00010401555766302204,
      "loss": 0.8428,
      "step": 142540
    },
    {
      "epoch": 2.5026773644200215,
      "grad_norm": 0.05525439178992741,
      "learning_rate": 0.00010400441918215469,
      "loss": 0.8347,
      "step": 142550
    },
    {
      "epoch": 2.5028529293000226,
      "grad_norm": 0.05034115191450633,
      "learning_rate": 0.00010399328066322078,
      "loss": 0.8455,
      "step": 142560
    },
    {
      "epoch": 2.5030284941800245,
      "grad_norm": 0.057166155164746754,
      "learning_rate": 0.00010398214210636141,
      "loss": 0.8475,
      "step": 142570
    },
    {
      "epoch": 2.5032040590600255,
      "grad_norm": 0.04144004981974555,
      "learning_rate": 0.00010397100351171773,
      "loss": 0.8505,
      "step": 142580
    },
    {
      "epoch": 2.503379623940027,
      "grad_norm": 0.06444361298871397,
      "learning_rate": 0.0001039598648794309,
      "loss": 0.8462,
      "step": 142590
    },
    {
      "epoch": 2.5035551888200285,
      "grad_norm": 0.055629641149774194,
      "learning_rate": 0.00010394872620964196,
      "loss": 0.8494,
      "step": 142600
    },
    {
      "epoch": 2.50373075370003,
      "grad_norm": 0.0657218175800107,
      "learning_rate": 0.00010393758750249212,
      "loss": 0.8438,
      "step": 142610
    },
    {
      "epoch": 2.5039063185800314,
      "grad_norm": 0.058003185508603984,
      "learning_rate": 0.0001039264487581225,
      "loss": 0.8365,
      "step": 142620
    },
    {
      "epoch": 2.5040818834600325,
      "grad_norm": 0.04714232775417791,
      "learning_rate": 0.00010391530997667421,
      "loss": 0.8401,
      "step": 142630
    },
    {
      "epoch": 2.504257448340034,
      "grad_norm": 0.04799103250452571,
      "learning_rate": 0.00010390417115828839,
      "loss": 0.8345,
      "step": 142640
    },
    {
      "epoch": 2.5044330132200354,
      "grad_norm": 0.04788733917392599,
      "learning_rate": 0.00010389303230310612,
      "loss": 0.8532,
      "step": 142650
    },
    {
      "epoch": 2.504608578100037,
      "grad_norm": 0.04465443008171607,
      "learning_rate": 0.0001038818934112686,
      "loss": 0.8479,
      "step": 142660
    },
    {
      "epoch": 2.5047841429800384,
      "grad_norm": 0.06597121954923378,
      "learning_rate": 0.00010387075448291695,
      "loss": 0.8442,
      "step": 142670
    },
    {
      "epoch": 2.5049597078600394,
      "grad_norm": 0.054370616025878726,
      "learning_rate": 0.0001038596155181923,
      "loss": 0.8432,
      "step": 142680
    },
    {
      "epoch": 2.5051352727400413,
      "grad_norm": 0.056732381270401276,
      "learning_rate": 0.00010384847651723574,
      "loss": 0.8465,
      "step": 142690
    },
    {
      "epoch": 2.5053108376200424,
      "grad_norm": 0.054813611673800215,
      "learning_rate": 0.00010383733748018846,
      "loss": 0.8397,
      "step": 142700
    },
    {
      "epoch": 2.505486402500044,
      "grad_norm": 0.05237390028097457,
      "learning_rate": 0.00010382619840719156,
      "loss": 0.8467,
      "step": 142710
    },
    {
      "epoch": 2.5056619673800453,
      "grad_norm": 0.05332096390885668,
      "learning_rate": 0.0001038150592983862,
      "loss": 0.8459,
      "step": 142720
    },
    {
      "epoch": 2.505837532260047,
      "grad_norm": 0.05529552762971052,
      "learning_rate": 0.0001038039201539135,
      "loss": 0.8535,
      "step": 142730
    },
    {
      "epoch": 2.5060130971400483,
      "grad_norm": 0.05174451113559161,
      "learning_rate": 0.00010379278097391458,
      "loss": 0.8446,
      "step": 142740
    },
    {
      "epoch": 2.5061886620200493,
      "grad_norm": 0.05249996742814569,
      "learning_rate": 0.00010378164175853058,
      "loss": 0.8452,
      "step": 142750
    },
    {
      "epoch": 2.506364226900051,
      "grad_norm": 0.06305566057914178,
      "learning_rate": 0.00010377050250790268,
      "loss": 0.8477,
      "step": 142760
    },
    {
      "epoch": 2.5065397917800523,
      "grad_norm": 0.056534575274087694,
      "learning_rate": 0.00010375936322217192,
      "loss": 0.844,
      "step": 142770
    },
    {
      "epoch": 2.5067153566600537,
      "grad_norm": 0.04844733368167266,
      "learning_rate": 0.00010374822390147956,
      "loss": 0.8394,
      "step": 142780
    },
    {
      "epoch": 2.506890921540055,
      "grad_norm": 0.05920005241109342,
      "learning_rate": 0.00010373708454596665,
      "loss": 0.8408,
      "step": 142790
    },
    {
      "epoch": 2.5070664864200567,
      "grad_norm": 0.06286544601064697,
      "learning_rate": 0.00010372594515577437,
      "loss": 0.8348,
      "step": 142800
    },
    {
      "epoch": 2.507242051300058,
      "grad_norm": 0.06218715112525819,
      "learning_rate": 0.00010371480573104385,
      "loss": 0.8354,
      "step": 142810
    },
    {
      "epoch": 2.507417616180059,
      "grad_norm": 0.04384301886611398,
      "learning_rate": 0.00010370366627191617,
      "loss": 0.8495,
      "step": 142820
    },
    {
      "epoch": 2.5075931810600607,
      "grad_norm": 0.05464613872097064,
      "learning_rate": 0.00010369252677853253,
      "loss": 0.8461,
      "step": 142830
    },
    {
      "epoch": 2.507768745940062,
      "grad_norm": 0.050536930753246084,
      "learning_rate": 0.00010368138725103411,
      "loss": 0.8438,
      "step": 142840
    },
    {
      "epoch": 2.5079443108200636,
      "grad_norm": 0.06704454584497295,
      "learning_rate": 0.00010367024768956199,
      "loss": 0.8416,
      "step": 142850
    },
    {
      "epoch": 2.508119875700065,
      "grad_norm": 0.05479580488975961,
      "learning_rate": 0.00010365910809425729,
      "loss": 0.8372,
      "step": 142860
    },
    {
      "epoch": 2.508295440580066,
      "grad_norm": 0.061197145274567,
      "learning_rate": 0.00010364796846526118,
      "loss": 0.8465,
      "step": 142870
    },
    {
      "epoch": 2.5084710054600676,
      "grad_norm": 0.07195592349859847,
      "learning_rate": 0.00010363682880271486,
      "loss": 0.8369,
      "step": 142880
    },
    {
      "epoch": 2.508646570340069,
      "grad_norm": 0.06480400746900229,
      "learning_rate": 0.00010362568910675934,
      "loss": 0.8451,
      "step": 142890
    },
    {
      "epoch": 2.5088221352200706,
      "grad_norm": 0.0552972896075506,
      "learning_rate": 0.00010361454937753587,
      "loss": 0.8511,
      "step": 142900
    },
    {
      "epoch": 2.508997700100072,
      "grad_norm": 0.051880569818987775,
      "learning_rate": 0.00010360340961518558,
      "loss": 0.8485,
      "step": 142910
    },
    {
      "epoch": 2.5091732649800735,
      "grad_norm": 0.04872346645911148,
      "learning_rate": 0.00010359226981984955,
      "loss": 0.8448,
      "step": 142920
    },
    {
      "epoch": 2.509348829860075,
      "grad_norm": 0.053363997554426564,
      "learning_rate": 0.00010358112999166898,
      "loss": 0.8441,
      "step": 142930
    },
    {
      "epoch": 2.509524394740076,
      "grad_norm": 0.05260189444263742,
      "learning_rate": 0.00010356999013078499,
      "loss": 0.8464,
      "step": 142940
    },
    {
      "epoch": 2.5096999596200775,
      "grad_norm": 0.04745895206458539,
      "learning_rate": 0.0001035588502373387,
      "loss": 0.8472,
      "step": 142950
    },
    {
      "epoch": 2.509875524500079,
      "grad_norm": 0.05397405849271323,
      "learning_rate": 0.00010354771031147137,
      "loss": 0.8478,
      "step": 142960
    },
    {
      "epoch": 2.5100510893800805,
      "grad_norm": 0.056935908980208644,
      "learning_rate": 0.00010353657035332402,
      "loss": 0.8495,
      "step": 142970
    },
    {
      "epoch": 2.510226654260082,
      "grad_norm": 0.047590178594552515,
      "learning_rate": 0.00010352543036303786,
      "loss": 0.8403,
      "step": 142980
    },
    {
      "epoch": 2.510402219140083,
      "grad_norm": 0.06080154464338497,
      "learning_rate": 0.00010351429034075395,
      "loss": 0.8442,
      "step": 142990
    },
    {
      "epoch": 2.5105777840200845,
      "grad_norm": 0.07800170021664385,
      "learning_rate": 0.00010350315028661355,
      "loss": 0.8478,
      "step": 143000
    },
    {
      "epoch": 2.510753348900086,
      "grad_norm": 0.05202677419223641,
      "learning_rate": 0.00010349201020075776,
      "loss": 0.8394,
      "step": 143010
    },
    {
      "epoch": 2.5109289137800874,
      "grad_norm": 0.047149188386531975,
      "learning_rate": 0.0001034808700833277,
      "loss": 0.8485,
      "step": 143020
    },
    {
      "epoch": 2.511104478660089,
      "grad_norm": 0.04972712760529804,
      "learning_rate": 0.00010346972993446458,
      "loss": 0.8435,
      "step": 143030
    },
    {
      "epoch": 2.5112800435400904,
      "grad_norm": 0.06667301327830123,
      "learning_rate": 0.00010345858975430945,
      "loss": 0.8382,
      "step": 143040
    },
    {
      "epoch": 2.511455608420092,
      "grad_norm": 0.05446105374794501,
      "learning_rate": 0.00010344744954300356,
      "loss": 0.8493,
      "step": 143050
    },
    {
      "epoch": 2.511631173300093,
      "grad_norm": 0.04810330425553091,
      "learning_rate": 0.000103436309300688,
      "loss": 0.8486,
      "step": 143060
    },
    {
      "epoch": 2.5118067381800944,
      "grad_norm": 0.04639368740059439,
      "learning_rate": 0.00010342516902750391,
      "loss": 0.8449,
      "step": 143070
    },
    {
      "epoch": 2.511982303060096,
      "grad_norm": 0.07705661593047594,
      "learning_rate": 0.00010341402872359254,
      "loss": 0.844,
      "step": 143080
    },
    {
      "epoch": 2.5121578679400973,
      "grad_norm": 0.047487684482404854,
      "learning_rate": 0.00010340288838909488,
      "loss": 0.8462,
      "step": 143090
    },
    {
      "epoch": 2.512333432820099,
      "grad_norm": 0.05002069853802224,
      "learning_rate": 0.0001033917480241522,
      "loss": 0.8458,
      "step": 143100
    },
    {
      "epoch": 2.5125089977001,
      "grad_norm": 0.05818478345926438,
      "learning_rate": 0.0001033806076289056,
      "loss": 0.8409,
      "step": 143110
    },
    {
      "epoch": 2.5126845625801013,
      "grad_norm": 0.05757689159003821,
      "learning_rate": 0.00010336946720349625,
      "loss": 0.844,
      "step": 143120
    },
    {
      "epoch": 2.512860127460103,
      "grad_norm": 0.05057550846153971,
      "learning_rate": 0.00010335832674806529,
      "loss": 0.838,
      "step": 143130
    },
    {
      "epoch": 2.5130356923401043,
      "grad_norm": 0.050480831780058855,
      "learning_rate": 0.0001033471862627539,
      "loss": 0.8427,
      "step": 143140
    },
    {
      "epoch": 2.5132112572201057,
      "grad_norm": 0.050954800323316805,
      "learning_rate": 0.00010333604574770323,
      "loss": 0.8499,
      "step": 143150
    },
    {
      "epoch": 2.513386822100107,
      "grad_norm": 0.053307985974010166,
      "learning_rate": 0.00010332490520305436,
      "loss": 0.8434,
      "step": 143160
    },
    {
      "epoch": 2.5135623869801087,
      "grad_norm": 0.054141337067350435,
      "learning_rate": 0.00010331376462894853,
      "loss": 0.8456,
      "step": 143170
    },
    {
      "epoch": 2.5137379518601097,
      "grad_norm": 0.05742022412315701,
      "learning_rate": 0.00010330262402552682,
      "loss": 0.8502,
      "step": 143180
    },
    {
      "epoch": 2.513913516740111,
      "grad_norm": 0.06033206272016542,
      "learning_rate": 0.00010329148339293047,
      "loss": 0.85,
      "step": 143190
    },
    {
      "epoch": 2.5140890816201127,
      "grad_norm": 0.05197275287615295,
      "learning_rate": 0.00010328034273130059,
      "loss": 0.8418,
      "step": 143200
    },
    {
      "epoch": 2.514264646500114,
      "grad_norm": 0.042280136978048805,
      "learning_rate": 0.00010326920204077828,
      "loss": 0.8421,
      "step": 143210
    },
    {
      "epoch": 2.5144402113801156,
      "grad_norm": 0.06394313887351781,
      "learning_rate": 0.00010325806132150478,
      "loss": 0.8494,
      "step": 143220
    },
    {
      "epoch": 2.5146157762601167,
      "grad_norm": 0.04909531656881485,
      "learning_rate": 0.00010324692057362118,
      "loss": 0.853,
      "step": 143230
    },
    {
      "epoch": 2.5147913411401186,
      "grad_norm": 0.047220374289755514,
      "learning_rate": 0.0001032357797972687,
      "loss": 0.8473,
      "step": 143240
    },
    {
      "epoch": 2.5149669060201196,
      "grad_norm": 0.05023423412730481,
      "learning_rate": 0.00010322463899258846,
      "loss": 0.8492,
      "step": 143250
    },
    {
      "epoch": 2.515142470900121,
      "grad_norm": 0.06281718691099095,
      "learning_rate": 0.00010321349815972159,
      "loss": 0.8474,
      "step": 143260
    },
    {
      "epoch": 2.5153180357801226,
      "grad_norm": 0.04412149621324071,
      "learning_rate": 0.00010320235729880929,
      "loss": 0.843,
      "step": 143270
    },
    {
      "epoch": 2.515493600660124,
      "grad_norm": 0.05260416439622253,
      "learning_rate": 0.00010319121640999269,
      "loss": 0.8413,
      "step": 143280
    },
    {
      "epoch": 2.5156691655401255,
      "grad_norm": 0.04531566356763744,
      "learning_rate": 0.00010318007549341295,
      "loss": 0.8536,
      "step": 143290
    },
    {
      "epoch": 2.5158447304201266,
      "grad_norm": 0.05063641109357206,
      "learning_rate": 0.00010316893454921126,
      "loss": 0.8505,
      "step": 143300
    },
    {
      "epoch": 2.516020295300128,
      "grad_norm": 0.06587380659479843,
      "learning_rate": 0.00010315779357752877,
      "loss": 0.8394,
      "step": 143310
    },
    {
      "epoch": 2.5161958601801295,
      "grad_norm": 0.04969717168841306,
      "learning_rate": 0.00010314665257850659,
      "loss": 0.848,
      "step": 143320
    },
    {
      "epoch": 2.516371425060131,
      "grad_norm": 0.06130554251609734,
      "learning_rate": 0.00010313551155228593,
      "loss": 0.8475,
      "step": 143330
    },
    {
      "epoch": 2.5165469899401325,
      "grad_norm": 0.059024783862386354,
      "learning_rate": 0.00010312437049900792,
      "loss": 0.8404,
      "step": 143340
    },
    {
      "epoch": 2.5167225548201335,
      "grad_norm": 0.05513863385961283,
      "learning_rate": 0.00010311322941881371,
      "loss": 0.8329,
      "step": 143350
    },
    {
      "epoch": 2.5168981197001354,
      "grad_norm": 0.0771009684609572,
      "learning_rate": 0.00010310208831184451,
      "loss": 0.8507,
      "step": 143360
    },
    {
      "epoch": 2.5170736845801365,
      "grad_norm": 0.07427337771390935,
      "learning_rate": 0.00010309094717824144,
      "loss": 0.847,
      "step": 143370
    },
    {
      "epoch": 2.517249249460138,
      "grad_norm": 0.047207835039594415,
      "learning_rate": 0.00010307980601814567,
      "loss": 0.8481,
      "step": 143380
    },
    {
      "epoch": 2.5174248143401394,
      "grad_norm": 0.05030302750316479,
      "learning_rate": 0.00010306866483169837,
      "loss": 0.8361,
      "step": 143390
    },
    {
      "epoch": 2.517600379220141,
      "grad_norm": 0.04286832227402328,
      "learning_rate": 0.00010305752361904064,
      "loss": 0.8352,
      "step": 143400
    },
    {
      "epoch": 2.5177759441001424,
      "grad_norm": 0.06418985086182974,
      "learning_rate": 0.00010304638238031375,
      "loss": 0.8468,
      "step": 143410
    },
    {
      "epoch": 2.5179515089801434,
      "grad_norm": 0.05056812838781507,
      "learning_rate": 0.00010303524111565876,
      "loss": 0.8455,
      "step": 143420
    },
    {
      "epoch": 2.518127073860145,
      "grad_norm": 0.0586320259523311,
      "learning_rate": 0.00010302409982521688,
      "loss": 0.8394,
      "step": 143430
    },
    {
      "epoch": 2.5183026387401464,
      "grad_norm": 0.055120060448774134,
      "learning_rate": 0.00010301295850912931,
      "loss": 0.8427,
      "step": 143440
    },
    {
      "epoch": 2.518478203620148,
      "grad_norm": 0.048793231668800924,
      "learning_rate": 0.00010300181716753711,
      "loss": 0.8418,
      "step": 143450
    },
    {
      "epoch": 2.5186537685001493,
      "grad_norm": 0.04439354475546193,
      "learning_rate": 0.00010299067580058153,
      "loss": 0.8398,
      "step": 143460
    },
    {
      "epoch": 2.5188293333801504,
      "grad_norm": 0.07418661259855411,
      "learning_rate": 0.00010297953440840371,
      "loss": 0.8438,
      "step": 143470
    },
    {
      "epoch": 2.5190048982601523,
      "grad_norm": 0.05646826331049359,
      "learning_rate": 0.00010296839299114483,
      "loss": 0.8514,
      "step": 143480
    },
    {
      "epoch": 2.5191804631401533,
      "grad_norm": 0.04712102428632186,
      "learning_rate": 0.000102957251548946,
      "loss": 0.8407,
      "step": 143490
    },
    {
      "epoch": 2.519356028020155,
      "grad_norm": 0.07152034074391402,
      "learning_rate": 0.00010294611008194843,
      "loss": 0.8387,
      "step": 143500
    },
    {
      "epoch": 2.5195315929001563,
      "grad_norm": 0.07374617627746383,
      "learning_rate": 0.00010293496859029329,
      "loss": 0.8417,
      "step": 143510
    },
    {
      "epoch": 2.5197071577801577,
      "grad_norm": 0.04223256039630638,
      "learning_rate": 0.00010292382707412169,
      "loss": 0.8461,
      "step": 143520
    },
    {
      "epoch": 2.5198827226601592,
      "grad_norm": 0.04689582242625291,
      "learning_rate": 0.00010291268553357487,
      "loss": 0.8395,
      "step": 143530
    },
    {
      "epoch": 2.5200582875401603,
      "grad_norm": 0.05075645851690045,
      "learning_rate": 0.00010290154396879393,
      "loss": 0.8305,
      "step": 143540
    },
    {
      "epoch": 2.5202338524201617,
      "grad_norm": 0.06366515165136206,
      "learning_rate": 0.00010289040237992007,
      "loss": 0.8454,
      "step": 143550
    },
    {
      "epoch": 2.520409417300163,
      "grad_norm": 0.058699963349016344,
      "learning_rate": 0.00010287926076709447,
      "loss": 0.8413,
      "step": 143560
    },
    {
      "epoch": 2.5205849821801647,
      "grad_norm": 0.05038195763213132,
      "learning_rate": 0.00010286811913045822,
      "loss": 0.8436,
      "step": 143570
    },
    {
      "epoch": 2.520760547060166,
      "grad_norm": 0.08045375640577782,
      "learning_rate": 0.00010285697747015262,
      "loss": 0.8484,
      "step": 143580
    },
    {
      "epoch": 2.5209361119401676,
      "grad_norm": 0.051158523151698096,
      "learning_rate": 0.0001028458357863187,
      "loss": 0.8425,
      "step": 143590
    },
    {
      "epoch": 2.521111676820169,
      "grad_norm": 0.05618380014742749,
      "learning_rate": 0.00010283469407909769,
      "loss": 0.8384,
      "step": 143600
    },
    {
      "epoch": 2.52128724170017,
      "grad_norm": 0.055280703522958774,
      "learning_rate": 0.00010282355234863081,
      "loss": 0.838,
      "step": 143610
    },
    {
      "epoch": 2.5214628065801716,
      "grad_norm": 0.056053489934573854,
      "learning_rate": 0.00010281241059505911,
      "loss": 0.8374,
      "step": 143620
    },
    {
      "epoch": 2.521638371460173,
      "grad_norm": 0.047892457124697946,
      "learning_rate": 0.00010280126881852383,
      "loss": 0.8422,
      "step": 143630
    },
    {
      "epoch": 2.5218139363401746,
      "grad_norm": 0.06246235642706999,
      "learning_rate": 0.00010279012701916615,
      "loss": 0.8462,
      "step": 143640
    },
    {
      "epoch": 2.521989501220176,
      "grad_norm": 0.059865640418197565,
      "learning_rate": 0.0001027789851971272,
      "loss": 0.8519,
      "step": 143650
    },
    {
      "epoch": 2.522165066100177,
      "grad_norm": 0.04524054412480435,
      "learning_rate": 0.00010276784335254821,
      "loss": 0.8452,
      "step": 143660
    },
    {
      "epoch": 2.5223406309801786,
      "grad_norm": 0.05014890722281261,
      "learning_rate": 0.00010275670148557026,
      "loss": 0.8492,
      "step": 143670
    },
    {
      "epoch": 2.52251619586018,
      "grad_norm": 0.06285568492259515,
      "learning_rate": 0.00010274555959633458,
      "loss": 0.8384,
      "step": 143680
    },
    {
      "epoch": 2.5226917607401815,
      "grad_norm": 0.060456780860240804,
      "learning_rate": 0.00010273441768498231,
      "loss": 0.8503,
      "step": 143690
    },
    {
      "epoch": 2.522867325620183,
      "grad_norm": 0.047242234917190734,
      "learning_rate": 0.00010272327575165466,
      "loss": 0.8461,
      "step": 143700
    },
    {
      "epoch": 2.5230428905001845,
      "grad_norm": 0.04502981961272794,
      "learning_rate": 0.00010271213379649275,
      "loss": 0.8406,
      "step": 143710
    },
    {
      "epoch": 2.523218455380186,
      "grad_norm": 0.049124791540028086,
      "learning_rate": 0.00010270099181963781,
      "loss": 0.8387,
      "step": 143720
    },
    {
      "epoch": 2.523394020260187,
      "grad_norm": 0.050553661365431955,
      "learning_rate": 0.00010268984982123098,
      "loss": 0.8479,
      "step": 143730
    },
    {
      "epoch": 2.5235695851401885,
      "grad_norm": 0.055505030502217736,
      "learning_rate": 0.00010267870780141338,
      "loss": 0.8352,
      "step": 143740
    },
    {
      "epoch": 2.52374515002019,
      "grad_norm": 0.05633584035815997,
      "learning_rate": 0.00010266756576032628,
      "loss": 0.8379,
      "step": 143750
    },
    {
      "epoch": 2.5239207149001914,
      "grad_norm": 0.05404121847433988,
      "learning_rate": 0.00010265642369811078,
      "loss": 0.8331,
      "step": 143760
    },
    {
      "epoch": 2.524096279780193,
      "grad_norm": 0.06611752726945576,
      "learning_rate": 0.00010264528161490805,
      "loss": 0.8418,
      "step": 143770
    },
    {
      "epoch": 2.524271844660194,
      "grad_norm": 0.04590745348889938,
      "learning_rate": 0.00010263413951085932,
      "loss": 0.8412,
      "step": 143780
    },
    {
      "epoch": 2.5244474095401954,
      "grad_norm": 0.05572260867829112,
      "learning_rate": 0.0001026229973861057,
      "loss": 0.8419,
      "step": 143790
    },
    {
      "epoch": 2.524622974420197,
      "grad_norm": 0.07500880332763435,
      "learning_rate": 0.00010261185524078842,
      "loss": 0.8444,
      "step": 143800
    },
    {
      "epoch": 2.5247985393001984,
      "grad_norm": 0.054865828010991345,
      "learning_rate": 0.00010260071307504862,
      "loss": 0.8424,
      "step": 143810
    },
    {
      "epoch": 2.5249741041802,
      "grad_norm": 0.052831401316494775,
      "learning_rate": 0.00010258957088902747,
      "loss": 0.8404,
      "step": 143820
    },
    {
      "epoch": 2.5251496690602013,
      "grad_norm": 0.05013736606394887,
      "learning_rate": 0.0001025784286828662,
      "loss": 0.8418,
      "step": 143830
    },
    {
      "epoch": 2.525325233940203,
      "grad_norm": 0.04933393561334326,
      "learning_rate": 0.00010256728645670586,
      "loss": 0.8383,
      "step": 143840
    },
    {
      "epoch": 2.525500798820204,
      "grad_norm": 0.08843762484670437,
      "learning_rate": 0.00010255614421068776,
      "loss": 0.8478,
      "step": 143850
    },
    {
      "epoch": 2.5256763637002053,
      "grad_norm": 0.04230827413399932,
      "learning_rate": 0.00010254500194495298,
      "loss": 0.842,
      "step": 143860
    },
    {
      "epoch": 2.525851928580207,
      "grad_norm": 0.052849253128135515,
      "learning_rate": 0.00010253385965964275,
      "loss": 0.8462,
      "step": 143870
    },
    {
      "epoch": 2.5260274934602083,
      "grad_norm": 0.06650163105911105,
      "learning_rate": 0.00010252271735489823,
      "loss": 0.8466,
      "step": 143880
    },
    {
      "epoch": 2.5262030583402098,
      "grad_norm": 0.05522858290699449,
      "learning_rate": 0.00010251157503086056,
      "loss": 0.8368,
      "step": 143890
    },
    {
      "epoch": 2.526378623220211,
      "grad_norm": 0.05281469347519334,
      "learning_rate": 0.00010250043268767096,
      "loss": 0.8427,
      "step": 143900
    },
    {
      "epoch": 2.5265541881002127,
      "grad_norm": 0.04011224673159607,
      "learning_rate": 0.00010248929032547056,
      "loss": 0.8381,
      "step": 143910
    },
    {
      "epoch": 2.5267297529802137,
      "grad_norm": 0.05999450994954265,
      "learning_rate": 0.0001024781479444006,
      "loss": 0.8506,
      "step": 143920
    },
    {
      "epoch": 2.526905317860215,
      "grad_norm": 0.04188012957177769,
      "learning_rate": 0.00010246700554460225,
      "loss": 0.8512,
      "step": 143930
    },
    {
      "epoch": 2.5270808827402167,
      "grad_norm": 0.06000569626031392,
      "learning_rate": 0.0001024558631262166,
      "loss": 0.8412,
      "step": 143940
    },
    {
      "epoch": 2.527256447620218,
      "grad_norm": 0.048639245334721685,
      "learning_rate": 0.00010244472068938493,
      "loss": 0.8405,
      "step": 143950
    },
    {
      "epoch": 2.5274320125002196,
      "grad_norm": 0.046443631883778716,
      "learning_rate": 0.00010243357823424832,
      "loss": 0.8507,
      "step": 143960
    },
    {
      "epoch": 2.5276075773802207,
      "grad_norm": 0.04861461288076238,
      "learning_rate": 0.00010242243576094803,
      "loss": 0.8421,
      "step": 143970
    },
    {
      "epoch": 2.527783142260222,
      "grad_norm": 0.047273022154055065,
      "learning_rate": 0.00010241129326962522,
      "loss": 0.8474,
      "step": 143980
    },
    {
      "epoch": 2.5279587071402236,
      "grad_norm": 0.047189964789937706,
      "learning_rate": 0.00010240015076042104,
      "loss": 0.8528,
      "step": 143990
    },
    {
      "epoch": 2.528134272020225,
      "grad_norm": 0.0558005030110785,
      "learning_rate": 0.0001023890082334767,
      "loss": 0.8421,
      "step": 144000
    },
    {
      "epoch": 2.5283098369002266,
      "grad_norm": 0.04880202261107336,
      "learning_rate": 0.00010237786568893331,
      "loss": 0.8426,
      "step": 144010
    },
    {
      "epoch": 2.5284854017802276,
      "grad_norm": 0.04943245530534555,
      "learning_rate": 0.00010236672312693214,
      "loss": 0.8431,
      "step": 144020
    },
    {
      "epoch": 2.5286609666602295,
      "grad_norm": 0.053766068110709934,
      "learning_rate": 0.00010235558054761431,
      "loss": 0.841,
      "step": 144030
    },
    {
      "epoch": 2.5288365315402306,
      "grad_norm": 0.05953189811720162,
      "learning_rate": 0.00010234443795112106,
      "loss": 0.8443,
      "step": 144040
    },
    {
      "epoch": 2.529012096420232,
      "grad_norm": 0.06341358409027839,
      "learning_rate": 0.00010233329533759346,
      "loss": 0.8426,
      "step": 144050
    },
    {
      "epoch": 2.5291876613002335,
      "grad_norm": 0.05503407426221624,
      "learning_rate": 0.00010232215270717278,
      "loss": 0.8469,
      "step": 144060
    },
    {
      "epoch": 2.529363226180235,
      "grad_norm": 0.04594711287537838,
      "learning_rate": 0.00010231101006000017,
      "loss": 0.8415,
      "step": 144070
    },
    {
      "epoch": 2.5295387910602365,
      "grad_norm": 0.04537270047287433,
      "learning_rate": 0.0001022998673962168,
      "loss": 0.8444,
      "step": 144080
    },
    {
      "epoch": 2.5297143559402375,
      "grad_norm": 0.06079361127125157,
      "learning_rate": 0.0001022887247159639,
      "loss": 0.8464,
      "step": 144090
    },
    {
      "epoch": 2.529889920820239,
      "grad_norm": 0.053222530632800365,
      "learning_rate": 0.00010227758201938256,
      "loss": 0.8413,
      "step": 144100
    },
    {
      "epoch": 2.5300654857002405,
      "grad_norm": 0.047679746815606834,
      "learning_rate": 0.00010226643930661404,
      "loss": 0.8465,
      "step": 144110
    },
    {
      "epoch": 2.530241050580242,
      "grad_norm": 0.05505141574667855,
      "learning_rate": 0.00010225529657779946,
      "loss": 0.8433,
      "step": 144120
    },
    {
      "epoch": 2.5304166154602434,
      "grad_norm": 0.06708201657668068,
      "learning_rate": 0.00010224415383308008,
      "loss": 0.8418,
      "step": 144130
    },
    {
      "epoch": 2.5305921803402445,
      "grad_norm": 0.06965235478992829,
      "learning_rate": 0.00010223301107259699,
      "loss": 0.8376,
      "step": 144140
    },
    {
      "epoch": 2.5307677452202464,
      "grad_norm": 0.044920926809689676,
      "learning_rate": 0.00010222186829649143,
      "loss": 0.8357,
      "step": 144150
    },
    {
      "epoch": 2.5309433101002474,
      "grad_norm": 0.06288929959794745,
      "learning_rate": 0.00010221072550490456,
      "loss": 0.8481,
      "step": 144160
    },
    {
      "epoch": 2.531118874980249,
      "grad_norm": 0.05597092464521089,
      "learning_rate": 0.0001021995826979776,
      "loss": 0.8453,
      "step": 144170
    },
    {
      "epoch": 2.5312944398602504,
      "grad_norm": 0.06173046836415393,
      "learning_rate": 0.00010218843987585165,
      "loss": 0.837,
      "step": 144180
    },
    {
      "epoch": 2.531470004740252,
      "grad_norm": 0.06481432620981036,
      "learning_rate": 0.00010217729703866796,
      "loss": 0.8442,
      "step": 144190
    },
    {
      "epoch": 2.5316455696202533,
      "grad_norm": 0.059688425957187735,
      "learning_rate": 0.0001021661541865677,
      "loss": 0.8406,
      "step": 144200
    },
    {
      "epoch": 2.5318211345002544,
      "grad_norm": 0.07596246309292436,
      "learning_rate": 0.00010215501131969202,
      "loss": 0.8329,
      "step": 144210
    },
    {
      "epoch": 2.531996699380256,
      "grad_norm": 0.042385445910712666,
      "learning_rate": 0.00010214386843818215,
      "loss": 0.8397,
      "step": 144220
    },
    {
      "epoch": 2.5321722642602573,
      "grad_norm": 0.06918010427965082,
      "learning_rate": 0.00010213272554217922,
      "loss": 0.8458,
      "step": 144230
    },
    {
      "epoch": 2.532347829140259,
      "grad_norm": 0.052288260099418685,
      "learning_rate": 0.00010212158263182448,
      "loss": 0.8461,
      "step": 144240
    },
    {
      "epoch": 2.5325233940202603,
      "grad_norm": 0.047955869550531806,
      "learning_rate": 0.00010211043970725904,
      "loss": 0.8431,
      "step": 144250
    },
    {
      "epoch": 2.5326989589002618,
      "grad_norm": 0.05594957274267593,
      "learning_rate": 0.00010209929676862413,
      "loss": 0.8445,
      "step": 144260
    },
    {
      "epoch": 2.5328745237802632,
      "grad_norm": 0.046785056958387444,
      "learning_rate": 0.00010208815381606091,
      "loss": 0.849,
      "step": 144270
    },
    {
      "epoch": 2.5330500886602643,
      "grad_norm": 0.053502871926243784,
      "learning_rate": 0.00010207701084971058,
      "loss": 0.8462,
      "step": 144280
    },
    {
      "epoch": 2.5332256535402657,
      "grad_norm": 0.06364130437395492,
      "learning_rate": 0.00010206586786971433,
      "loss": 0.8363,
      "step": 144290
    },
    {
      "epoch": 2.533401218420267,
      "grad_norm": 0.05683247775459558,
      "learning_rate": 0.00010205472487621328,
      "loss": 0.8423,
      "step": 144300
    },
    {
      "epoch": 2.5335767833002687,
      "grad_norm": 0.0519159611717314,
      "learning_rate": 0.00010204358186934868,
      "loss": 0.8419,
      "step": 144310
    },
    {
      "epoch": 2.53375234818027,
      "grad_norm": 0.07093728518189547,
      "learning_rate": 0.0001020324388492617,
      "loss": 0.841,
      "step": 144320
    },
    {
      "epoch": 2.533927913060271,
      "grad_norm": 0.0685843281271463,
      "learning_rate": 0.00010202129581609359,
      "loss": 0.8351,
      "step": 144330
    },
    {
      "epoch": 2.5341034779402727,
      "grad_norm": 0.05906505409707912,
      "learning_rate": 0.0001020101527699854,
      "loss": 0.8358,
      "step": 144340
    },
    {
      "epoch": 2.534279042820274,
      "grad_norm": 0.048824994034561246,
      "learning_rate": 0.00010199900971107839,
      "loss": 0.8492,
      "step": 144350
    },
    {
      "epoch": 2.5344546077002756,
      "grad_norm": 0.05821740554587821,
      "learning_rate": 0.00010198786663951374,
      "loss": 0.8491,
      "step": 144360
    },
    {
      "epoch": 2.534630172580277,
      "grad_norm": 0.05377262560185397,
      "learning_rate": 0.00010197672355543262,
      "loss": 0.8453,
      "step": 144370
    },
    {
      "epoch": 2.5348057374602786,
      "grad_norm": 0.05582314628441169,
      "learning_rate": 0.00010196558045897622,
      "loss": 0.8462,
      "step": 144380
    },
    {
      "epoch": 2.53498130234028,
      "grad_norm": 0.049937731389901895,
      "learning_rate": 0.00010195443735028575,
      "loss": 0.8513,
      "step": 144390
    },
    {
      "epoch": 2.535156867220281,
      "grad_norm": 0.046747617082913966,
      "learning_rate": 0.00010194329422950237,
      "loss": 0.8448,
      "step": 144400
    },
    {
      "epoch": 2.5353324321002826,
      "grad_norm": 0.045071129301062066,
      "learning_rate": 0.00010193215109676729,
      "loss": 0.8432,
      "step": 144410
    },
    {
      "epoch": 2.535507996980284,
      "grad_norm": 0.061590295420515175,
      "learning_rate": 0.00010192100795222163,
      "loss": 0.8419,
      "step": 144420
    },
    {
      "epoch": 2.5356835618602855,
      "grad_norm": 0.054864349427898924,
      "learning_rate": 0.00010190986479600666,
      "loss": 0.8468,
      "step": 144430
    },
    {
      "epoch": 2.535859126740287,
      "grad_norm": 0.04997988179934848,
      "learning_rate": 0.00010189872162826353,
      "loss": 0.8502,
      "step": 144440
    },
    {
      "epoch": 2.536034691620288,
      "grad_norm": 0.06149772076252384,
      "learning_rate": 0.00010188757844913341,
      "loss": 0.8435,
      "step": 144450
    },
    {
      "epoch": 2.5362102565002895,
      "grad_norm": 0.08083725828211547,
      "learning_rate": 0.00010187643525875751,
      "loss": 0.852,
      "step": 144460
    },
    {
      "epoch": 2.536385821380291,
      "grad_norm": 0.06488596642745485,
      "learning_rate": 0.00010186529205727699,
      "loss": 0.8443,
      "step": 144470
    },
    {
      "epoch": 2.5365613862602925,
      "grad_norm": 0.05996403690704777,
      "learning_rate": 0.00010185414884483304,
      "loss": 0.8456,
      "step": 144480
    },
    {
      "epoch": 2.536736951140294,
      "grad_norm": 0.05914132578562475,
      "learning_rate": 0.00010184300562156693,
      "loss": 0.849,
      "step": 144490
    },
    {
      "epoch": 2.5369125160202954,
      "grad_norm": 0.05471811567553182,
      "learning_rate": 0.00010183186238761972,
      "loss": 0.8475,
      "step": 144500
    },
    {
      "epoch": 2.537088080900297,
      "grad_norm": 0.06303871161348656,
      "learning_rate": 0.0001018207191431327,
      "loss": 0.844,
      "step": 144510
    },
    {
      "epoch": 2.537263645780298,
      "grad_norm": 0.058058800540748065,
      "learning_rate": 0.00010180957588824693,
      "loss": 0.844,
      "step": 144520
    },
    {
      "epoch": 2.5374392106602994,
      "grad_norm": 0.04793444768644256,
      "learning_rate": 0.00010179843262310375,
      "loss": 0.8468,
      "step": 144530
    },
    {
      "epoch": 2.537614775540301,
      "grad_norm": 0.05785365742343034,
      "learning_rate": 0.00010178728934784425,
      "loss": 0.8422,
      "step": 144540
    },
    {
      "epoch": 2.5377903404203024,
      "grad_norm": 0.05845609424828943,
      "learning_rate": 0.00010177614606260966,
      "loss": 0.8473,
      "step": 144550
    },
    {
      "epoch": 2.537965905300304,
      "grad_norm": 0.06079905654199603,
      "learning_rate": 0.00010176500276754115,
      "loss": 0.8373,
      "step": 144560
    },
    {
      "epoch": 2.538141470180305,
      "grad_norm": 0.06554006279206685,
      "learning_rate": 0.00010175385946277986,
      "loss": 0.8396,
      "step": 144570
    },
    {
      "epoch": 2.5383170350603064,
      "grad_norm": 0.05718782101564981,
      "learning_rate": 0.00010174271614846707,
      "loss": 0.8384,
      "step": 144580
    },
    {
      "epoch": 2.538492599940308,
      "grad_norm": 0.056149209630912134,
      "learning_rate": 0.00010173157282474393,
      "loss": 0.8442,
      "step": 144590
    },
    {
      "epoch": 2.5386681648203093,
      "grad_norm": 0.055232908549785846,
      "learning_rate": 0.00010172042949175161,
      "loss": 0.8419,
      "step": 144600
    },
    {
      "epoch": 2.538843729700311,
      "grad_norm": 0.061850051139595436,
      "learning_rate": 0.0001017092861496313,
      "loss": 0.845,
      "step": 144610
    },
    {
      "epoch": 2.5390192945803123,
      "grad_norm": 0.04867572995144214,
      "learning_rate": 0.00010169814279852421,
      "loss": 0.8477,
      "step": 144620
    },
    {
      "epoch": 2.5391948594603138,
      "grad_norm": 0.08973743334317331,
      "learning_rate": 0.00010168699943857152,
      "loss": 0.8471,
      "step": 144630
    },
    {
      "epoch": 2.539370424340315,
      "grad_norm": 0.06956218533716965,
      "learning_rate": 0.00010167585606991438,
      "loss": 0.8416,
      "step": 144640
    },
    {
      "epoch": 2.5395459892203163,
      "grad_norm": 0.0658479959181583,
      "learning_rate": 0.00010166471269269404,
      "loss": 0.8472,
      "step": 144650
    },
    {
      "epoch": 2.5397215541003177,
      "grad_norm": 0.058252494935241504,
      "learning_rate": 0.00010165356930705168,
      "loss": 0.8488,
      "step": 144660
    },
    {
      "epoch": 2.5398971189803192,
      "grad_norm": 0.07004803287577482,
      "learning_rate": 0.00010164242591312847,
      "loss": 0.8413,
      "step": 144670
    },
    {
      "epoch": 2.5400726838603207,
      "grad_norm": 0.055773906507143395,
      "learning_rate": 0.0001016312825110656,
      "loss": 0.8525,
      "step": 144680
    },
    {
      "epoch": 2.5402482487403217,
      "grad_norm": 0.056751392129289396,
      "learning_rate": 0.00010162013910100423,
      "loss": 0.8439,
      "step": 144690
    },
    {
      "epoch": 2.5404238136203237,
      "grad_norm": 0.0621924878510073,
      "learning_rate": 0.00010160899568308562,
      "loss": 0.8448,
      "step": 144700
    },
    {
      "epoch": 2.5405993785003247,
      "grad_norm": 0.04592756572800951,
      "learning_rate": 0.00010159785225745086,
      "loss": 0.8445,
      "step": 144710
    },
    {
      "epoch": 2.540774943380326,
      "grad_norm": 0.05489442169792734,
      "learning_rate": 0.00010158670882424125,
      "loss": 0.8431,
      "step": 144720
    },
    {
      "epoch": 2.5409505082603276,
      "grad_norm": 0.045538711917369794,
      "learning_rate": 0.00010157556538359792,
      "loss": 0.849,
      "step": 144730
    },
    {
      "epoch": 2.541126073140329,
      "grad_norm": 0.052661033324916014,
      "learning_rate": 0.00010156442193566205,
      "loss": 0.8377,
      "step": 144740
    },
    {
      "epoch": 2.5413016380203306,
      "grad_norm": 0.04705759945810261,
      "learning_rate": 0.00010155327848057486,
      "loss": 0.8527,
      "step": 144750
    },
    {
      "epoch": 2.5414772029003316,
      "grad_norm": 0.06587685768727086,
      "learning_rate": 0.00010154213501847749,
      "loss": 0.8516,
      "step": 144760
    },
    {
      "epoch": 2.541652767780333,
      "grad_norm": 0.04693078142782494,
      "learning_rate": 0.0001015309915495112,
      "loss": 0.8443,
      "step": 144770
    },
    {
      "epoch": 2.5418283326603346,
      "grad_norm": 0.0727219481334051,
      "learning_rate": 0.00010151984807381713,
      "loss": 0.8395,
      "step": 144780
    },
    {
      "epoch": 2.542003897540336,
      "grad_norm": 0.05949711413971895,
      "learning_rate": 0.0001015087045915365,
      "loss": 0.8443,
      "step": 144790
    },
    {
      "epoch": 2.5421794624203375,
      "grad_norm": 0.04297050088318855,
      "learning_rate": 0.00010149756110281048,
      "loss": 0.8421,
      "step": 144800
    },
    {
      "epoch": 2.5423550273003386,
      "grad_norm": 0.053398916101025436,
      "learning_rate": 0.00010148641760778026,
      "loss": 0.8426,
      "step": 144810
    },
    {
      "epoch": 2.5425305921803405,
      "grad_norm": 0.06893171259816591,
      "learning_rate": 0.00010147527410658704,
      "loss": 0.8391,
      "step": 144820
    },
    {
      "epoch": 2.5427061570603415,
      "grad_norm": 0.05259010420760769,
      "learning_rate": 0.00010146413059937199,
      "loss": 0.8417,
      "step": 144830
    },
    {
      "epoch": 2.542881721940343,
      "grad_norm": 0.04647543262511668,
      "learning_rate": 0.00010145298708627633,
      "loss": 0.8496,
      "step": 144840
    },
    {
      "epoch": 2.5430572868203445,
      "grad_norm": 0.06046248625407634,
      "learning_rate": 0.00010144184356744126,
      "loss": 0.8439,
      "step": 144850
    },
    {
      "epoch": 2.543232851700346,
      "grad_norm": 0.05221153291169053,
      "learning_rate": 0.00010143070004300792,
      "loss": 0.8425,
      "step": 144860
    },
    {
      "epoch": 2.5434084165803474,
      "grad_norm": 0.04515723361578262,
      "learning_rate": 0.00010141955651311752,
      "loss": 0.846,
      "step": 144870
    },
    {
      "epoch": 2.5435839814603485,
      "grad_norm": 0.07598957018794181,
      "learning_rate": 0.00010140841297791128,
      "loss": 0.8501,
      "step": 144880
    },
    {
      "epoch": 2.54375954634035,
      "grad_norm": 0.043723290304553304,
      "learning_rate": 0.00010139726943753036,
      "loss": 0.8481,
      "step": 144890
    },
    {
      "epoch": 2.5439351112203514,
      "grad_norm": 0.06921229023281936,
      "learning_rate": 0.00010138612589211598,
      "loss": 0.8484,
      "step": 144900
    },
    {
      "epoch": 2.544110676100353,
      "grad_norm": 0.04743897426046057,
      "learning_rate": 0.00010137498234180926,
      "loss": 0.8532,
      "step": 144910
    },
    {
      "epoch": 2.5442862409803544,
      "grad_norm": 0.04090631601278316,
      "learning_rate": 0.00010136383878675151,
      "loss": 0.8536,
      "step": 144920
    },
    {
      "epoch": 2.5444618058603554,
      "grad_norm": 0.06315272754525823,
      "learning_rate": 0.00010135269522708379,
      "loss": 0.8409,
      "step": 144930
    },
    {
      "epoch": 2.5446373707403573,
      "grad_norm": 0.05290937526904948,
      "learning_rate": 0.00010134155166294741,
      "loss": 0.842,
      "step": 144940
    },
    {
      "epoch": 2.5448129356203584,
      "grad_norm": 0.05194938221519553,
      "learning_rate": 0.00010133040809448349,
      "loss": 0.843,
      "step": 144950
    },
    {
      "epoch": 2.54498850050036,
      "grad_norm": 0.05176302302682112,
      "learning_rate": 0.00010131926452183319,
      "loss": 0.8386,
      "step": 144960
    },
    {
      "epoch": 2.5451640653803613,
      "grad_norm": 0.047515023886895866,
      "learning_rate": 0.0001013081209451378,
      "loss": 0.8417,
      "step": 144970
    },
    {
      "epoch": 2.545339630260363,
      "grad_norm": 0.0578206852867643,
      "learning_rate": 0.00010129697736453841,
      "loss": 0.8504,
      "step": 144980
    },
    {
      "epoch": 2.5455151951403643,
      "grad_norm": 0.05066047032591141,
      "learning_rate": 0.00010128583378017627,
      "loss": 0.8384,
      "step": 144990
    },
    {
      "epoch": 2.5456907600203653,
      "grad_norm": 0.0643579102414522,
      "learning_rate": 0.0001012746901921926,
      "loss": 0.8488,
      "step": 145000
    },
    {
      "epoch": 2.545866324900367,
      "grad_norm": 0.05057623385838652,
      "learning_rate": 0.00010126354660072854,
      "loss": 0.844,
      "step": 145010
    },
    {
      "epoch": 2.5460418897803683,
      "grad_norm": 0.060996196798083165,
      "learning_rate": 0.00010125240300592529,
      "loss": 0.842,
      "step": 145020
    },
    {
      "epoch": 2.5462174546603698,
      "grad_norm": 0.0657188275066196,
      "learning_rate": 0.00010124125940792404,
      "loss": 0.8478,
      "step": 145030
    },
    {
      "epoch": 2.5463930195403712,
      "grad_norm": 0.04714476592435689,
      "learning_rate": 0.00010123011580686599,
      "loss": 0.843,
      "step": 145040
    },
    {
      "epoch": 2.5465685844203727,
      "grad_norm": 0.07733053067239935,
      "learning_rate": 0.00010121897220289232,
      "loss": 0.8329,
      "step": 145050
    },
    {
      "epoch": 2.546744149300374,
      "grad_norm": 0.06416761378815437,
      "learning_rate": 0.00010120782859614422,
      "loss": 0.8482,
      "step": 145060
    },
    {
      "epoch": 2.546919714180375,
      "grad_norm": 0.04802477667600119,
      "learning_rate": 0.00010119668498676293,
      "loss": 0.8481,
      "step": 145070
    },
    {
      "epoch": 2.5470952790603767,
      "grad_norm": 0.06169772444817597,
      "learning_rate": 0.00010118554137488955,
      "loss": 0.8414,
      "step": 145080
    },
    {
      "epoch": 2.547270843940378,
      "grad_norm": 0.06337408088746518,
      "learning_rate": 0.00010117439776066539,
      "loss": 0.8281,
      "step": 145090
    },
    {
      "epoch": 2.5474464088203796,
      "grad_norm": 0.053185567360471864,
      "learning_rate": 0.00010116325414423153,
      "loss": 0.8422,
      "step": 145100
    },
    {
      "epoch": 2.547621973700381,
      "grad_norm": 0.04621873375386792,
      "learning_rate": 0.00010115211052572924,
      "loss": 0.8373,
      "step": 145110
    },
    {
      "epoch": 2.547797538580382,
      "grad_norm": 0.049372102716025955,
      "learning_rate": 0.00010114096690529968,
      "loss": 0.8427,
      "step": 145120
    },
    {
      "epoch": 2.5479731034603836,
      "grad_norm": 0.05278620942892563,
      "learning_rate": 0.000101129823283084,
      "loss": 0.8393,
      "step": 145130
    },
    {
      "epoch": 2.548148668340385,
      "grad_norm": 0.06623980307345328,
      "learning_rate": 0.00010111867965922346,
      "loss": 0.8388,
      "step": 145140
    },
    {
      "epoch": 2.5483242332203866,
      "grad_norm": 0.055319029118582996,
      "learning_rate": 0.0001011075360338592,
      "loss": 0.8379,
      "step": 145150
    },
    {
      "epoch": 2.548499798100388,
      "grad_norm": 0.04736109967925576,
      "learning_rate": 0.00010109639240713247,
      "loss": 0.8437,
      "step": 145160
    },
    {
      "epoch": 2.5486753629803895,
      "grad_norm": 0.04428945695276519,
      "learning_rate": 0.00010108524877918444,
      "loss": 0.8464,
      "step": 145170
    },
    {
      "epoch": 2.548850927860391,
      "grad_norm": 0.05136090746388668,
      "learning_rate": 0.00010107410515015627,
      "loss": 0.8418,
      "step": 145180
    },
    {
      "epoch": 2.549026492740392,
      "grad_norm": 0.054464202599446856,
      "learning_rate": 0.0001010629615201892,
      "loss": 0.8474,
      "step": 145190
    },
    {
      "epoch": 2.5492020576203935,
      "grad_norm": 0.05040663039267084,
      "learning_rate": 0.00010105181788942435,
      "loss": 0.8413,
      "step": 145200
    },
    {
      "epoch": 2.549377622500395,
      "grad_norm": 0.06452319515705326,
      "learning_rate": 0.00010104067425800298,
      "loss": 0.8394,
      "step": 145210
    },
    {
      "epoch": 2.5495531873803965,
      "grad_norm": 0.04912930771701553,
      "learning_rate": 0.00010102953062606626,
      "loss": 0.8497,
      "step": 145220
    },
    {
      "epoch": 2.549728752260398,
      "grad_norm": 0.052677713748579914,
      "learning_rate": 0.00010101838699375539,
      "loss": 0.8518,
      "step": 145230
    },
    {
      "epoch": 2.549904317140399,
      "grad_norm": 0.05156330544295093,
      "learning_rate": 0.00010100724336121157,
      "loss": 0.8416,
      "step": 145240
    },
    {
      "epoch": 2.5500798820204005,
      "grad_norm": 0.07347596393768058,
      "learning_rate": 0.00010099609972857595,
      "loss": 0.8468,
      "step": 145250
    },
    {
      "epoch": 2.550255446900402,
      "grad_norm": 0.07107528092463682,
      "learning_rate": 0.00010098495609598973,
      "loss": 0.8385,
      "step": 145260
    },
    {
      "epoch": 2.5504310117804034,
      "grad_norm": 0.10097798218705023,
      "learning_rate": 0.00010097381246359415,
      "loss": 0.8516,
      "step": 145270
    },
    {
      "epoch": 2.550606576660405,
      "grad_norm": 0.07881473901503924,
      "learning_rate": 0.00010096266883153036,
      "loss": 0.8415,
      "step": 145280
    },
    {
      "epoch": 2.5507821415404064,
      "grad_norm": 0.04919089705435014,
      "learning_rate": 0.00010095152519993953,
      "loss": 0.8477,
      "step": 145290
    },
    {
      "epoch": 2.550957706420408,
      "grad_norm": 0.046771163228420454,
      "learning_rate": 0.00010094038156896293,
      "loss": 0.8455,
      "step": 145300
    },
    {
      "epoch": 2.551133271300409,
      "grad_norm": 0.058449524156204524,
      "learning_rate": 0.00010092923793874172,
      "loss": 0.8347,
      "step": 145310
    },
    {
      "epoch": 2.5513088361804104,
      "grad_norm": 0.042984196668141204,
      "learning_rate": 0.00010091809430941706,
      "loss": 0.8484,
      "step": 145320
    },
    {
      "epoch": 2.551484401060412,
      "grad_norm": 0.054942757970958704,
      "learning_rate": 0.00010090695068113009,
      "loss": 0.8449,
      "step": 145330
    },
    {
      "epoch": 2.5516599659404133,
      "grad_norm": 0.057068652608588705,
      "learning_rate": 0.00010089580705402215,
      "loss": 0.8381,
      "step": 145340
    },
    {
      "epoch": 2.551835530820415,
      "grad_norm": 0.04752293066358601,
      "learning_rate": 0.00010088466342823439,
      "loss": 0.8405,
      "step": 145350
    },
    {
      "epoch": 2.552011095700416,
      "grad_norm": 0.05978421852128778,
      "learning_rate": 0.00010087351980390787,
      "loss": 0.8448,
      "step": 145360
    },
    {
      "epoch": 2.5521866605804178,
      "grad_norm": 0.0481399387560657,
      "learning_rate": 0.00010086237618118395,
      "loss": 0.8463,
      "step": 145370
    },
    {
      "epoch": 2.552362225460419,
      "grad_norm": 0.06554957336424094,
      "learning_rate": 0.00010085123256020372,
      "loss": 0.848,
      "step": 145380
    },
    {
      "epoch": 2.5525377903404203,
      "grad_norm": 0.059838806875171235,
      "learning_rate": 0.00010084008894110839,
      "loss": 0.8446,
      "step": 145390
    },
    {
      "epoch": 2.5527133552204218,
      "grad_norm": 0.0908204614349178,
      "learning_rate": 0.0001008289453240392,
      "loss": 0.8414,
      "step": 145400
    },
    {
      "epoch": 2.5528889201004232,
      "grad_norm": 0.0428732652389429,
      "learning_rate": 0.00010081780170913729,
      "loss": 0.8386,
      "step": 145410
    },
    {
      "epoch": 2.5530644849804247,
      "grad_norm": 0.047394351095253716,
      "learning_rate": 0.00010080665809654388,
      "loss": 0.8455,
      "step": 145420
    },
    {
      "epoch": 2.5532400498604257,
      "grad_norm": 0.04656822576067342,
      "learning_rate": 0.0001007955144864001,
      "loss": 0.8497,
      "step": 145430
    },
    {
      "epoch": 2.553415614740427,
      "grad_norm": 0.04957968522549764,
      "learning_rate": 0.00010078437087884722,
      "loss": 0.8397,
      "step": 145440
    },
    {
      "epoch": 2.5535911796204287,
      "grad_norm": 0.043468376605640444,
      "learning_rate": 0.00010077322727402643,
      "loss": 0.8505,
      "step": 145450
    },
    {
      "epoch": 2.55376674450043,
      "grad_norm": 0.07847624441052135,
      "learning_rate": 0.00010076208367207885,
      "loss": 0.8469,
      "step": 145460
    },
    {
      "epoch": 2.5539423093804317,
      "grad_norm": 0.051330513576354736,
      "learning_rate": 0.00010075094007314573,
      "loss": 0.8478,
      "step": 145470
    },
    {
      "epoch": 2.5541178742604327,
      "grad_norm": 0.059310758878030785,
      "learning_rate": 0.00010073979647736828,
      "loss": 0.8338,
      "step": 145480
    },
    {
      "epoch": 2.5542934391404346,
      "grad_norm": 0.05595662786267769,
      "learning_rate": 0.00010072865288488762,
      "loss": 0.8457,
      "step": 145490
    },
    {
      "epoch": 2.5544690040204356,
      "grad_norm": 0.07012873411959922,
      "learning_rate": 0.00010071750929584496,
      "loss": 0.8416,
      "step": 145500
    },
    {
      "epoch": 2.554644568900437,
      "grad_norm": 0.04342683563765623,
      "learning_rate": 0.00010070636571038157,
      "loss": 0.8485,
      "step": 145510
    },
    {
      "epoch": 2.5548201337804386,
      "grad_norm": 0.06252645124893601,
      "learning_rate": 0.00010069522212863857,
      "loss": 0.8479,
      "step": 145520
    },
    {
      "epoch": 2.55499569866044,
      "grad_norm": 0.04538783878507589,
      "learning_rate": 0.00010068407855075713,
      "loss": 0.8462,
      "step": 145530
    },
    {
      "epoch": 2.5551712635404416,
      "grad_norm": 0.07186351028806649,
      "learning_rate": 0.00010067293497687849,
      "loss": 0.8468,
      "step": 145540
    },
    {
      "epoch": 2.5553468284204426,
      "grad_norm": 0.057958288808702345,
      "learning_rate": 0.00010066179140714384,
      "loss": 0.8417,
      "step": 145550
    },
    {
      "epoch": 2.555522393300444,
      "grad_norm": 0.05493087336234765,
      "learning_rate": 0.00010065064784169433,
      "loss": 0.8431,
      "step": 145560
    },
    {
      "epoch": 2.5556979581804455,
      "grad_norm": 0.04912410926654572,
      "learning_rate": 0.00010063950428067121,
      "loss": 0.8441,
      "step": 145570
    },
    {
      "epoch": 2.555873523060447,
      "grad_norm": 0.08780644148743749,
      "learning_rate": 0.00010062836072421565,
      "loss": 0.8393,
      "step": 145580
    },
    {
      "epoch": 2.5560490879404485,
      "grad_norm": 0.05624489549246763,
      "learning_rate": 0.00010061721717246881,
      "loss": 0.8458,
      "step": 145590
    },
    {
      "epoch": 2.5562246528204495,
      "grad_norm": 0.056588942492452464,
      "learning_rate": 0.00010060607362557186,
      "loss": 0.8499,
      "step": 145600
    },
    {
      "epoch": 2.5564002177004514,
      "grad_norm": 0.06800144440336547,
      "learning_rate": 0.00010059493008366607,
      "loss": 0.8498,
      "step": 145610
    },
    {
      "epoch": 2.5565757825804525,
      "grad_norm": 0.05594840231280971,
      "learning_rate": 0.00010058378654689262,
      "loss": 0.8402,
      "step": 145620
    },
    {
      "epoch": 2.556751347460454,
      "grad_norm": 0.0499615620064173,
      "learning_rate": 0.00010057264301539264,
      "loss": 0.846,
      "step": 145630
    },
    {
      "epoch": 2.5569269123404554,
      "grad_norm": 0.07878863223548171,
      "learning_rate": 0.00010056149948930738,
      "loss": 0.8425,
      "step": 145640
    },
    {
      "epoch": 2.557102477220457,
      "grad_norm": 0.04836017826252143,
      "learning_rate": 0.00010055035596877798,
      "loss": 0.8474,
      "step": 145650
    },
    {
      "epoch": 2.5572780421004584,
      "grad_norm": 0.05077379369849083,
      "learning_rate": 0.00010053921245394569,
      "loss": 0.8456,
      "step": 145660
    },
    {
      "epoch": 2.5574536069804594,
      "grad_norm": 0.058326395586401736,
      "learning_rate": 0.00010052806894495159,
      "loss": 0.8522,
      "step": 145670
    },
    {
      "epoch": 2.557629171860461,
      "grad_norm": 0.08557586024156495,
      "learning_rate": 0.00010051692544193703,
      "loss": 0.8378,
      "step": 145680
    },
    {
      "epoch": 2.5578047367404624,
      "grad_norm": 0.0694625651725857,
      "learning_rate": 0.00010050578194504313,
      "loss": 0.8372,
      "step": 145690
    },
    {
      "epoch": 2.557980301620464,
      "grad_norm": 0.05246393789326766,
      "learning_rate": 0.00010049463845441097,
      "loss": 0.8411,
      "step": 145700
    },
    {
      "epoch": 2.5581558665004653,
      "grad_norm": 0.04911594044564899,
      "learning_rate": 0.0001004834949701819,
      "loss": 0.847,
      "step": 145710
    },
    {
      "epoch": 2.558331431380467,
      "grad_norm": 0.04639987024437049,
      "learning_rate": 0.00010047235149249705,
      "loss": 0.8425,
      "step": 145720
    },
    {
      "epoch": 2.5585069962604683,
      "grad_norm": 0.05198394782596791,
      "learning_rate": 0.00010046120802149758,
      "loss": 0.8473,
      "step": 145730
    },
    {
      "epoch": 2.5586825611404693,
      "grad_norm": 0.08651410570655242,
      "learning_rate": 0.00010045006455732473,
      "loss": 0.8519,
      "step": 145740
    },
    {
      "epoch": 2.558858126020471,
      "grad_norm": 0.0741389680905672,
      "learning_rate": 0.00010043892110011967,
      "loss": 0.8494,
      "step": 145750
    },
    {
      "epoch": 2.5590336909004723,
      "grad_norm": 0.06813705236142924,
      "learning_rate": 0.0001004277776500236,
      "loss": 0.8505,
      "step": 145760
    },
    {
      "epoch": 2.5592092557804738,
      "grad_norm": 0.05110025789740185,
      "learning_rate": 0.00010041663420717764,
      "loss": 0.8414,
      "step": 145770
    },
    {
      "epoch": 2.5593848206604752,
      "grad_norm": 0.04963426523117195,
      "learning_rate": 0.00010040549077172308,
      "loss": 0.8482,
      "step": 145780
    },
    {
      "epoch": 2.5595603855404763,
      "grad_norm": 0.050690998094462196,
      "learning_rate": 0.00010039434734380106,
      "loss": 0.8399,
      "step": 145790
    },
    {
      "epoch": 2.5597359504204777,
      "grad_norm": 0.05761201805093799,
      "learning_rate": 0.00010038320392355276,
      "loss": 0.8497,
      "step": 145800
    },
    {
      "epoch": 2.5599115153004792,
      "grad_norm": 0.04710625332333758,
      "learning_rate": 0.00010037206051111942,
      "loss": 0.8421,
      "step": 145810
    },
    {
      "epoch": 2.5600870801804807,
      "grad_norm": 0.04157857919789309,
      "learning_rate": 0.0001003609171066422,
      "loss": 0.8496,
      "step": 145820
    },
    {
      "epoch": 2.560262645060482,
      "grad_norm": 0.04011833263340493,
      "learning_rate": 0.00010034977371026224,
      "loss": 0.8489,
      "step": 145830
    },
    {
      "epoch": 2.5604382099404837,
      "grad_norm": 0.040614760461466594,
      "learning_rate": 0.00010033863032212077,
      "loss": 0.8472,
      "step": 145840
    },
    {
      "epoch": 2.560613774820485,
      "grad_norm": 0.06816586913378796,
      "learning_rate": 0.00010032748694235898,
      "loss": 0.8555,
      "step": 145850
    },
    {
      "epoch": 2.560789339700486,
      "grad_norm": 0.0435392572235769,
      "learning_rate": 0.00010031634357111812,
      "loss": 0.8423,
      "step": 145860
    },
    {
      "epoch": 2.5609649045804876,
      "grad_norm": 0.043835891591403754,
      "learning_rate": 0.00010030520020853924,
      "loss": 0.8517,
      "step": 145870
    },
    {
      "epoch": 2.561140469460489,
      "grad_norm": 0.05576117389087177,
      "learning_rate": 0.00010029405685476365,
      "loss": 0.8428,
      "step": 145880
    },
    {
      "epoch": 2.5613160343404906,
      "grad_norm": 0.04820290473246229,
      "learning_rate": 0.00010028291350993248,
      "loss": 0.8499,
      "step": 145890
    },
    {
      "epoch": 2.561491599220492,
      "grad_norm": 0.0689576723216291,
      "learning_rate": 0.00010027177017418695,
      "loss": 0.8474,
      "step": 145900
    },
    {
      "epoch": 2.561667164100493,
      "grad_norm": 0.040884522783742466,
      "learning_rate": 0.00010026062684766822,
      "loss": 0.8524,
      "step": 145910
    },
    {
      "epoch": 2.5618427289804946,
      "grad_norm": 0.058052004220503814,
      "learning_rate": 0.00010024948353051751,
      "loss": 0.8462,
      "step": 145920
    },
    {
      "epoch": 2.562018293860496,
      "grad_norm": 0.08634507108634117,
      "learning_rate": 0.00010023834022287597,
      "loss": 0.8458,
      "step": 145930
    },
    {
      "epoch": 2.5621938587404975,
      "grad_norm": 0.06060628719761223,
      "learning_rate": 0.00010022719692488475,
      "loss": 0.842,
      "step": 145940
    },
    {
      "epoch": 2.562369423620499,
      "grad_norm": 0.04711331136288181,
      "learning_rate": 0.00010021605363668519,
      "loss": 0.8458,
      "step": 145950
    },
    {
      "epoch": 2.5625449885005005,
      "grad_norm": 0.07205084655469775,
      "learning_rate": 0.00010020491035841832,
      "loss": 0.8442,
      "step": 145960
    },
    {
      "epoch": 2.562720553380502,
      "grad_norm": 0.053156267061501596,
      "learning_rate": 0.00010019376709022539,
      "loss": 0.8458,
      "step": 145970
    },
    {
      "epoch": 2.562896118260503,
      "grad_norm": 0.057277012810616236,
      "learning_rate": 0.00010018262383224762,
      "loss": 0.8452,
      "step": 145980
    },
    {
      "epoch": 2.5630716831405045,
      "grad_norm": 0.058252548250647115,
      "learning_rate": 0.00010017148058462616,
      "loss": 0.8349,
      "step": 145990
    },
    {
      "epoch": 2.563247248020506,
      "grad_norm": 0.048497540432814096,
      "learning_rate": 0.00010016033734750219,
      "loss": 0.8445,
      "step": 146000
    },
    {
      "epoch": 2.5634228129005074,
      "grad_norm": 0.05898967218380269,
      "learning_rate": 0.00010014919412101686,
      "loss": 0.8461,
      "step": 146010
    },
    {
      "epoch": 2.563598377780509,
      "grad_norm": 0.051075791943676646,
      "learning_rate": 0.00010013805090531146,
      "loss": 0.8505,
      "step": 146020
    },
    {
      "epoch": 2.56377394266051,
      "grad_norm": 0.06490550773787757,
      "learning_rate": 0.00010012690770052712,
      "loss": 0.8463,
      "step": 146030
    },
    {
      "epoch": 2.5639495075405114,
      "grad_norm": 0.07285058734529765,
      "learning_rate": 0.00010011576450680498,
      "loss": 0.8458,
      "step": 146040
    },
    {
      "epoch": 2.564125072420513,
      "grad_norm": 0.043682219164542437,
      "learning_rate": 0.00010010462132428631,
      "loss": 0.835,
      "step": 146050
    },
    {
      "epoch": 2.5643006373005144,
      "grad_norm": 0.05490195993816553,
      "learning_rate": 0.00010009347815311226,
      "loss": 0.8439,
      "step": 146060
    },
    {
      "epoch": 2.564476202180516,
      "grad_norm": 0.05672823267219942,
      "learning_rate": 0.00010008233499342401,
      "loss": 0.8442,
      "step": 146070
    },
    {
      "epoch": 2.5646517670605173,
      "grad_norm": 0.07290858444767635,
      "learning_rate": 0.00010007119184536275,
      "loss": 0.8387,
      "step": 146080
    },
    {
      "epoch": 2.564827331940519,
      "grad_norm": 0.044706142305627375,
      "learning_rate": 0.00010006004870906968,
      "loss": 0.8421,
      "step": 146090
    },
    {
      "epoch": 2.56500289682052,
      "grad_norm": 0.055519367290141716,
      "learning_rate": 0.00010004890558468596,
      "loss": 0.8451,
      "step": 146100
    },
    {
      "epoch": 2.5651784617005213,
      "grad_norm": 0.05209146995553215,
      "learning_rate": 0.00010003776247235274,
      "loss": 0.8518,
      "step": 146110
    },
    {
      "epoch": 2.565354026580523,
      "grad_norm": 0.06306903738286265,
      "learning_rate": 0.00010002661937221134,
      "loss": 0.8424,
      "step": 146120
    },
    {
      "epoch": 2.5655295914605243,
      "grad_norm": 0.05934446811937853,
      "learning_rate": 0.00010001547628440282,
      "loss": 0.8426,
      "step": 146130
    },
    {
      "epoch": 2.5657051563405258,
      "grad_norm": 0.06041377447219753,
      "learning_rate": 0.0001000043332090684,
      "loss": 0.8484,
      "step": 146140
    },
    {
      "epoch": 2.565880721220527,
      "grad_norm": 0.06262626094541089,
      "learning_rate": 9.999319014634928e-05,
      "loss": 0.8421,
      "step": 146150
    },
    {
      "epoch": 2.5660562861005287,
      "grad_norm": 0.06411044017580053,
      "learning_rate": 9.998204709638664e-05,
      "loss": 0.8363,
      "step": 146160
    },
    {
      "epoch": 2.5662318509805297,
      "grad_norm": 0.07691173482900454,
      "learning_rate": 9.997090405932165e-05,
      "loss": 0.8519,
      "step": 146170
    },
    {
      "epoch": 2.5664074158605312,
      "grad_norm": 0.0621556625099902,
      "learning_rate": 9.995976103529547e-05,
      "loss": 0.8449,
      "step": 146180
    },
    {
      "epoch": 2.5665829807405327,
      "grad_norm": 0.0572128418196605,
      "learning_rate": 9.994861802444937e-05,
      "loss": 0.8421,
      "step": 146190
    },
    {
      "epoch": 2.566758545620534,
      "grad_norm": 0.06843379169498615,
      "learning_rate": 9.993747502692448e-05,
      "loss": 0.8495,
      "step": 146200
    },
    {
      "epoch": 2.5669341105005357,
      "grad_norm": 0.057601706552095684,
      "learning_rate": 9.992633204286192e-05,
      "loss": 0.8451,
      "step": 146210
    },
    {
      "epoch": 2.5671096753805367,
      "grad_norm": 0.04885623182275867,
      "learning_rate": 9.9915189072403e-05,
      "loss": 0.8457,
      "step": 146220
    },
    {
      "epoch": 2.567285240260538,
      "grad_norm": 0.06002304203262183,
      "learning_rate": 9.990404611568883e-05,
      "loss": 0.8452,
      "step": 146230
    },
    {
      "epoch": 2.5674608051405396,
      "grad_norm": 0.04814948853304108,
      "learning_rate": 9.989290317286058e-05,
      "loss": 0.8445,
      "step": 146240
    },
    {
      "epoch": 2.567636370020541,
      "grad_norm": 0.048274489983868005,
      "learning_rate": 9.988176024405949e-05,
      "loss": 0.8466,
      "step": 146250
    },
    {
      "epoch": 2.5678119349005426,
      "grad_norm": 0.06738956276566968,
      "learning_rate": 9.987061732942672e-05,
      "loss": 0.8464,
      "step": 146260
    },
    {
      "epoch": 2.5679874997805436,
      "grad_norm": 0.04433818650317746,
      "learning_rate": 9.985947442910343e-05,
      "loss": 0.8473,
      "step": 146270
    },
    {
      "epoch": 2.5681630646605456,
      "grad_norm": 0.05861951733924947,
      "learning_rate": 9.984833154323077e-05,
      "loss": 0.8446,
      "step": 146280
    },
    {
      "epoch": 2.5683386295405466,
      "grad_norm": 0.06823972124733521,
      "learning_rate": 9.983718867195004e-05,
      "loss": 0.8516,
      "step": 146290
    },
    {
      "epoch": 2.568514194420548,
      "grad_norm": 0.09128356034849289,
      "learning_rate": 9.982604581540232e-05,
      "loss": 0.843,
      "step": 146300
    },
    {
      "epoch": 2.5686897593005495,
      "grad_norm": 0.04691055569511221,
      "learning_rate": 9.981490297372883e-05,
      "loss": 0.8413,
      "step": 146310
    },
    {
      "epoch": 2.568865324180551,
      "grad_norm": 0.045782214712630394,
      "learning_rate": 9.980376014707074e-05,
      "loss": 0.8535,
      "step": 146320
    },
    {
      "epoch": 2.5690408890605525,
      "grad_norm": 0.04617833688510015,
      "learning_rate": 9.979261733556926e-05,
      "loss": 0.8433,
      "step": 146330
    },
    {
      "epoch": 2.5692164539405535,
      "grad_norm": 0.054647884435286,
      "learning_rate": 9.978147453936555e-05,
      "loss": 0.8429,
      "step": 146340
    },
    {
      "epoch": 2.569392018820555,
      "grad_norm": 0.04570658946414179,
      "learning_rate": 9.977033175860075e-05,
      "loss": 0.8404,
      "step": 146350
    },
    {
      "epoch": 2.5695675837005565,
      "grad_norm": 0.052241941521705124,
      "learning_rate": 9.975918899341611e-05,
      "loss": 0.8344,
      "step": 146360
    },
    {
      "epoch": 2.569743148580558,
      "grad_norm": 0.050142087843227934,
      "learning_rate": 9.974804624395281e-05,
      "loss": 0.8408,
      "step": 146370
    },
    {
      "epoch": 2.5699187134605594,
      "grad_norm": 0.06420525371674392,
      "learning_rate": 9.973690351035193e-05,
      "loss": 0.8461,
      "step": 146380
    },
    {
      "epoch": 2.5700942783405605,
      "grad_norm": 0.053281292810661435,
      "learning_rate": 9.972576079275479e-05,
      "loss": 0.8457,
      "step": 146390
    },
    {
      "epoch": 2.5702698432205624,
      "grad_norm": 0.056630073054151936,
      "learning_rate": 9.971461809130249e-05,
      "loss": 0.8449,
      "step": 146400
    },
    {
      "epoch": 2.5704454081005634,
      "grad_norm": 0.04935767124659955,
      "learning_rate": 9.97034754061362e-05,
      "loss": 0.852,
      "step": 146410
    },
    {
      "epoch": 2.570620972980565,
      "grad_norm": 0.04004155627914177,
      "learning_rate": 9.969233273739716e-05,
      "loss": 0.8421,
      "step": 146420
    },
    {
      "epoch": 2.5707965378605664,
      "grad_norm": 0.04995995451069953,
      "learning_rate": 9.968119008522651e-05,
      "loss": 0.8488,
      "step": 146430
    },
    {
      "epoch": 2.570972102740568,
      "grad_norm": 0.05175085594660993,
      "learning_rate": 9.967004744976543e-05,
      "loss": 0.8485,
      "step": 146440
    },
    {
      "epoch": 2.5711476676205693,
      "grad_norm": 0.05503301134090389,
      "learning_rate": 9.965890483115507e-05,
      "loss": 0.8435,
      "step": 146450
    },
    {
      "epoch": 2.5713232325005704,
      "grad_norm": 0.05187030000384985,
      "learning_rate": 9.964776222953668e-05,
      "loss": 0.8476,
      "step": 146460
    },
    {
      "epoch": 2.571498797380572,
      "grad_norm": 0.07195463168495114,
      "learning_rate": 9.96366196450514e-05,
      "loss": 0.849,
      "step": 146470
    },
    {
      "epoch": 2.5716743622605733,
      "grad_norm": 0.058179058236588796,
      "learning_rate": 9.962547707784039e-05,
      "loss": 0.844,
      "step": 146480
    },
    {
      "epoch": 2.571849927140575,
      "grad_norm": 0.058948297848585025,
      "learning_rate": 9.961433452804486e-05,
      "loss": 0.8453,
      "step": 146490
    },
    {
      "epoch": 2.5720254920205763,
      "grad_norm": 0.051401258154649104,
      "learning_rate": 9.9603191995806e-05,
      "loss": 0.8398,
      "step": 146500
    },
    {
      "epoch": 2.5722010569005778,
      "grad_norm": 0.049725620164639564,
      "learning_rate": 9.959204948126494e-05,
      "loss": 0.8463,
      "step": 146510
    },
    {
      "epoch": 2.5723766217805792,
      "grad_norm": 0.05110050481661468,
      "learning_rate": 9.958090698456286e-05,
      "loss": 0.8404,
      "step": 146520
    },
    {
      "epoch": 2.5725521866605803,
      "grad_norm": 0.054169003711872034,
      "learning_rate": 9.9569764505841e-05,
      "loss": 0.843,
      "step": 146530
    },
    {
      "epoch": 2.5727277515405818,
      "grad_norm": 0.07868375675623045,
      "learning_rate": 9.955862204524051e-05,
      "loss": 0.8455,
      "step": 146540
    },
    {
      "epoch": 2.5729033164205832,
      "grad_norm": 0.051763139387078834,
      "learning_rate": 9.95474796029025e-05,
      "loss": 0.8494,
      "step": 146550
    },
    {
      "epoch": 2.5730788813005847,
      "grad_norm": 0.08367336332910742,
      "learning_rate": 9.953633717896825e-05,
      "loss": 0.8442,
      "step": 146560
    },
    {
      "epoch": 2.573254446180586,
      "grad_norm": 0.0486173153768386,
      "learning_rate": 9.952519477357886e-05,
      "loss": 0.8514,
      "step": 146570
    },
    {
      "epoch": 2.573430011060587,
      "grad_norm": 0.06418235208669962,
      "learning_rate": 9.951405238687554e-05,
      "loss": 0.8488,
      "step": 146580
    },
    {
      "epoch": 2.5736055759405887,
      "grad_norm": 0.06468097487080823,
      "learning_rate": 9.950291001899947e-05,
      "loss": 0.8472,
      "step": 146590
    },
    {
      "epoch": 2.57378114082059,
      "grad_norm": 0.05406673332022471,
      "learning_rate": 9.949176767009184e-05,
      "loss": 0.8465,
      "step": 146600
    },
    {
      "epoch": 2.5739567057005917,
      "grad_norm": 0.0784857714825616,
      "learning_rate": 9.948062534029378e-05,
      "loss": 0.8418,
      "step": 146610
    },
    {
      "epoch": 2.574132270580593,
      "grad_norm": 0.060074209029519966,
      "learning_rate": 9.946948302974644e-05,
      "loss": 0.8512,
      "step": 146620
    },
    {
      "epoch": 2.5743078354605946,
      "grad_norm": 0.05852566639122022,
      "learning_rate": 9.94583407385911e-05,
      "loss": 0.8508,
      "step": 146630
    },
    {
      "epoch": 2.574483400340596,
      "grad_norm": 0.060288112310218626,
      "learning_rate": 9.944719846696888e-05,
      "loss": 0.8483,
      "step": 146640
    },
    {
      "epoch": 2.574658965220597,
      "grad_norm": 0.05522370146116506,
      "learning_rate": 9.943605621502092e-05,
      "loss": 0.8484,
      "step": 146650
    },
    {
      "epoch": 2.5748345301005986,
      "grad_norm": 0.05605921153263445,
      "learning_rate": 9.942491398288845e-05,
      "loss": 0.8505,
      "step": 146660
    },
    {
      "epoch": 2.5750100949806,
      "grad_norm": 0.049999839028364156,
      "learning_rate": 9.941377177071264e-05,
      "loss": 0.8396,
      "step": 146670
    },
    {
      "epoch": 2.5751856598606016,
      "grad_norm": 0.06368342068855877,
      "learning_rate": 9.940262957863464e-05,
      "loss": 0.8512,
      "step": 146680
    },
    {
      "epoch": 2.575361224740603,
      "grad_norm": 0.04605098255555727,
      "learning_rate": 9.939148740679557e-05,
      "loss": 0.847,
      "step": 146690
    },
    {
      "epoch": 2.575536789620604,
      "grad_norm": 0.05755265984221994,
      "learning_rate": 9.93803452553367e-05,
      "loss": 0.846,
      "step": 146700
    },
    {
      "epoch": 2.5757123545006055,
      "grad_norm": 0.0501450403001465,
      "learning_rate": 9.936920312439922e-05,
      "loss": 0.8441,
      "step": 146710
    },
    {
      "epoch": 2.575887919380607,
      "grad_norm": 0.07431845130723996,
      "learning_rate": 9.935806101412417e-05,
      "loss": 0.8452,
      "step": 146720
    },
    {
      "epoch": 2.5760634842606085,
      "grad_norm": 0.053549406383956485,
      "learning_rate": 9.934691892465284e-05,
      "loss": 0.8453,
      "step": 146730
    },
    {
      "epoch": 2.57623904914061,
      "grad_norm": 0.04132751255295063,
      "learning_rate": 9.933577685612635e-05,
      "loss": 0.8431,
      "step": 146740
    },
    {
      "epoch": 2.5764146140206114,
      "grad_norm": 0.05488265057348745,
      "learning_rate": 9.93246348086859e-05,
      "loss": 0.839,
      "step": 146750
    },
    {
      "epoch": 2.576590178900613,
      "grad_norm": 0.05501389852167699,
      "learning_rate": 9.931349278247264e-05,
      "loss": 0.8495,
      "step": 146760
    },
    {
      "epoch": 2.576765743780614,
      "grad_norm": 0.044275427210355114,
      "learning_rate": 9.930235077762776e-05,
      "loss": 0.8469,
      "step": 146770
    },
    {
      "epoch": 2.5769413086606154,
      "grad_norm": 0.04419997773267886,
      "learning_rate": 9.929120879429241e-05,
      "loss": 0.8431,
      "step": 146780
    },
    {
      "epoch": 2.577116873540617,
      "grad_norm": 0.13742601762718246,
      "learning_rate": 9.928006683260774e-05,
      "loss": 0.8455,
      "step": 146790
    },
    {
      "epoch": 2.5772924384206184,
      "grad_norm": 0.07062651950502362,
      "learning_rate": 9.926892489271501e-05,
      "loss": 0.8451,
      "step": 146800
    },
    {
      "epoch": 2.57746800330062,
      "grad_norm": 0.07030458373503558,
      "learning_rate": 9.92577829747553e-05,
      "loss": 0.8417,
      "step": 146810
    },
    {
      "epoch": 2.577643568180621,
      "grad_norm": 0.04772812792552056,
      "learning_rate": 9.924664107886982e-05,
      "loss": 0.8503,
      "step": 146820
    },
    {
      "epoch": 2.577819133060623,
      "grad_norm": 0.06807692920424081,
      "learning_rate": 9.923549920519975e-05,
      "loss": 0.8515,
      "step": 146830
    },
    {
      "epoch": 2.577994697940624,
      "grad_norm": 0.047433431841640374,
      "learning_rate": 9.922435735388624e-05,
      "loss": 0.8466,
      "step": 146840
    },
    {
      "epoch": 2.5781702628206253,
      "grad_norm": 0.042899147158382445,
      "learning_rate": 9.921321552507046e-05,
      "loss": 0.8445,
      "step": 146850
    },
    {
      "epoch": 2.578345827700627,
      "grad_norm": 0.05178923122160359,
      "learning_rate": 9.920207371889354e-05,
      "loss": 0.8447,
      "step": 146860
    },
    {
      "epoch": 2.5785213925806283,
      "grad_norm": 0.06289465386682991,
      "learning_rate": 9.919093193549673e-05,
      "loss": 0.8452,
      "step": 146870
    },
    {
      "epoch": 2.5786969574606298,
      "grad_norm": 0.05687207959644087,
      "learning_rate": 9.917979017502119e-05,
      "loss": 0.8382,
      "step": 146880
    },
    {
      "epoch": 2.578872522340631,
      "grad_norm": 0.057020995779003276,
      "learning_rate": 9.916864843760798e-05,
      "loss": 0.8451,
      "step": 146890
    },
    {
      "epoch": 2.5790480872206323,
      "grad_norm": 0.055882517786806224,
      "learning_rate": 9.91575067233984e-05,
      "loss": 0.8483,
      "step": 146900
    },
    {
      "epoch": 2.5792236521006338,
      "grad_norm": 0.04074806932181728,
      "learning_rate": 9.914636503253357e-05,
      "loss": 0.842,
      "step": 146910
    },
    {
      "epoch": 2.5793992169806352,
      "grad_norm": 0.05098606334725768,
      "learning_rate": 9.913522336515461e-05,
      "loss": 0.8482,
      "step": 146920
    },
    {
      "epoch": 2.5795747818606367,
      "grad_norm": 0.05140615670859901,
      "learning_rate": 9.912408172140275e-05,
      "loss": 0.8467,
      "step": 146930
    },
    {
      "epoch": 2.5797503467406377,
      "grad_norm": 0.053680007212813644,
      "learning_rate": 9.911294010141914e-05,
      "loss": 0.8453,
      "step": 146940
    },
    {
      "epoch": 2.5799259116206397,
      "grad_norm": 0.057842901393784465,
      "learning_rate": 9.910179850534495e-05,
      "loss": 0.8392,
      "step": 146950
    },
    {
      "epoch": 2.5801014765006407,
      "grad_norm": 0.05418440999961439,
      "learning_rate": 9.909065693332127e-05,
      "loss": 0.853,
      "step": 146960
    },
    {
      "epoch": 2.580277041380642,
      "grad_norm": 0.04857181384640481,
      "learning_rate": 9.907951538548939e-05,
      "loss": 0.8382,
      "step": 146970
    },
    {
      "epoch": 2.5804526062606437,
      "grad_norm": 0.059653497572903946,
      "learning_rate": 9.906837386199043e-05,
      "loss": 0.8407,
      "step": 146980
    },
    {
      "epoch": 2.580628171140645,
      "grad_norm": 0.06510525268681691,
      "learning_rate": 9.905723236296551e-05,
      "loss": 0.8463,
      "step": 146990
    },
    {
      "epoch": 2.5808037360206466,
      "grad_norm": 0.05585266174190998,
      "learning_rate": 9.904609088855584e-05,
      "loss": 0.8497,
      "step": 147000
    },
    {
      "epoch": 2.5809793009006476,
      "grad_norm": 0.05450246813805095,
      "learning_rate": 9.90349494389026e-05,
      "loss": 0.8414,
      "step": 147010
    },
    {
      "epoch": 2.581154865780649,
      "grad_norm": 0.05930219351235824,
      "learning_rate": 9.902380801414689e-05,
      "loss": 0.8448,
      "step": 147020
    },
    {
      "epoch": 2.5813304306606506,
      "grad_norm": 0.07791015329286377,
      "learning_rate": 9.90126666144299e-05,
      "loss": 0.8436,
      "step": 147030
    },
    {
      "epoch": 2.581505995540652,
      "grad_norm": 0.05423156103069297,
      "learning_rate": 9.900152523989284e-05,
      "loss": 0.8487,
      "step": 147040
    },
    {
      "epoch": 2.5816815604206536,
      "grad_norm": 0.054649649782138265,
      "learning_rate": 9.899038389067686e-05,
      "loss": 0.8525,
      "step": 147050
    },
    {
      "epoch": 2.5818571253006546,
      "grad_norm": 0.04645690101660406,
      "learning_rate": 9.897924256692303e-05,
      "loss": 0.8479,
      "step": 147060
    },
    {
      "epoch": 2.5820326901806565,
      "grad_norm": 0.060930932662461655,
      "learning_rate": 9.896810126877262e-05,
      "loss": 0.8434,
      "step": 147070
    },
    {
      "epoch": 2.5822082550606575,
      "grad_norm": 0.06339822495109591,
      "learning_rate": 9.895695999636676e-05,
      "loss": 0.8382,
      "step": 147080
    },
    {
      "epoch": 2.582383819940659,
      "grad_norm": 0.10079415538372968,
      "learning_rate": 9.89458187498466e-05,
      "loss": 0.8427,
      "step": 147090
    },
    {
      "epoch": 2.5825593848206605,
      "grad_norm": 0.08176275764383532,
      "learning_rate": 9.893467752935331e-05,
      "loss": 0.8386,
      "step": 147100
    },
    {
      "epoch": 2.582734949700662,
      "grad_norm": 0.08027292849750255,
      "learning_rate": 9.892353633502808e-05,
      "loss": 0.8444,
      "step": 147110
    },
    {
      "epoch": 2.5829105145806635,
      "grad_norm": 0.07872449391987432,
      "learning_rate": 9.891239516701203e-05,
      "loss": 0.841,
      "step": 147120
    },
    {
      "epoch": 2.5830860794606645,
      "grad_norm": 0.07268053572640772,
      "learning_rate": 9.890125402544629e-05,
      "loss": 0.85,
      "step": 147130
    },
    {
      "epoch": 2.583261644340666,
      "grad_norm": 0.05088040697868812,
      "learning_rate": 9.889011291047209e-05,
      "loss": 0.8442,
      "step": 147140
    },
    {
      "epoch": 2.5834372092206674,
      "grad_norm": 0.0542506736941002,
      "learning_rate": 9.887897182223058e-05,
      "loss": 0.8424,
      "step": 147150
    },
    {
      "epoch": 2.583612774100669,
      "grad_norm": 0.04946359346567688,
      "learning_rate": 9.886783076086288e-05,
      "loss": 0.8471,
      "step": 147160
    },
    {
      "epoch": 2.5837883389806704,
      "grad_norm": 0.050728259705896624,
      "learning_rate": 9.88566897265102e-05,
      "loss": 0.8468,
      "step": 147170
    },
    {
      "epoch": 2.583963903860672,
      "grad_norm": 0.055829763616227705,
      "learning_rate": 9.884554871931367e-05,
      "loss": 0.8373,
      "step": 147180
    },
    {
      "epoch": 2.5841394687406734,
      "grad_norm": 0.05485295717195287,
      "learning_rate": 9.883440773941445e-05,
      "loss": 0.8428,
      "step": 147190
    },
    {
      "epoch": 2.5843150336206744,
      "grad_norm": 0.05086155268313271,
      "learning_rate": 9.882326678695365e-05,
      "loss": 0.8381,
      "step": 147200
    },
    {
      "epoch": 2.584490598500676,
      "grad_norm": 0.04475620358525788,
      "learning_rate": 9.881212586207253e-05,
      "loss": 0.8366,
      "step": 147210
    },
    {
      "epoch": 2.5846661633806773,
      "grad_norm": 0.0467774506776065,
      "learning_rate": 9.88009849649122e-05,
      "loss": 0.848,
      "step": 147220
    },
    {
      "epoch": 2.584841728260679,
      "grad_norm": 0.062065717656144104,
      "learning_rate": 9.878984409561376e-05,
      "loss": 0.8497,
      "step": 147230
    },
    {
      "epoch": 2.5850172931406803,
      "grad_norm": 0.050966388968865586,
      "learning_rate": 9.877870325431846e-05,
      "loss": 0.848,
      "step": 147240
    },
    {
      "epoch": 2.5851928580206813,
      "grad_norm": 0.07225026952756464,
      "learning_rate": 9.876756244116742e-05,
      "loss": 0.8391,
      "step": 147250
    },
    {
      "epoch": 2.585368422900683,
      "grad_norm": 0.04772454106130492,
      "learning_rate": 9.87564216563018e-05,
      "loss": 0.8475,
      "step": 147260
    },
    {
      "epoch": 2.5855439877806843,
      "grad_norm": 0.057434879090166334,
      "learning_rate": 9.874528089986269e-05,
      "loss": 0.8507,
      "step": 147270
    },
    {
      "epoch": 2.5857195526606858,
      "grad_norm": 0.05472384060991604,
      "learning_rate": 9.873414017199136e-05,
      "loss": 0.8476,
      "step": 147280
    },
    {
      "epoch": 2.5858951175406872,
      "grad_norm": 0.07035272721761722,
      "learning_rate": 9.872299947282889e-05,
      "loss": 0.8423,
      "step": 147290
    },
    {
      "epoch": 2.5860706824206887,
      "grad_norm": 0.06449830045253888,
      "learning_rate": 9.871185880251642e-05,
      "loss": 0.8463,
      "step": 147300
    },
    {
      "epoch": 2.58624624730069,
      "grad_norm": 0.048455860455401734,
      "learning_rate": 9.870071816119517e-05,
      "loss": 0.8461,
      "step": 147310
    },
    {
      "epoch": 2.5864218121806912,
      "grad_norm": 0.05796943927000105,
      "learning_rate": 9.868957754900628e-05,
      "loss": 0.8409,
      "step": 147320
    },
    {
      "epoch": 2.5865973770606927,
      "grad_norm": 0.04478078236577288,
      "learning_rate": 9.867843696609084e-05,
      "loss": 0.85,
      "step": 147330
    },
    {
      "epoch": 2.586772941940694,
      "grad_norm": 0.048662074837266485,
      "learning_rate": 9.866729641259008e-05,
      "loss": 0.8509,
      "step": 147340
    },
    {
      "epoch": 2.5869485068206957,
      "grad_norm": 0.06708825880288484,
      "learning_rate": 9.865615588864513e-05,
      "loss": 0.8443,
      "step": 147350
    },
    {
      "epoch": 2.587124071700697,
      "grad_norm": 0.06655146176009022,
      "learning_rate": 9.864501539439712e-05,
      "loss": 0.8456,
      "step": 147360
    },
    {
      "epoch": 2.587299636580698,
      "grad_norm": 0.06314165800050596,
      "learning_rate": 9.863387492998719e-05,
      "loss": 0.849,
      "step": 147370
    },
    {
      "epoch": 2.5874752014606996,
      "grad_norm": 0.059545518689901905,
      "learning_rate": 9.862273449555654e-05,
      "loss": 0.8447,
      "step": 147380
    },
    {
      "epoch": 2.587650766340701,
      "grad_norm": 0.050097119446454265,
      "learning_rate": 9.861159409124634e-05,
      "loss": 0.8541,
      "step": 147390
    },
    {
      "epoch": 2.5878263312207026,
      "grad_norm": 0.06473059704513145,
      "learning_rate": 9.860045371719763e-05,
      "loss": 0.8515,
      "step": 147400
    },
    {
      "epoch": 2.588001896100704,
      "grad_norm": 0.05314135392610301,
      "learning_rate": 9.858931337355168e-05,
      "loss": 0.8494,
      "step": 147410
    },
    {
      "epoch": 2.5881774609807056,
      "grad_norm": 0.05391444701996537,
      "learning_rate": 9.857817306044957e-05,
      "loss": 0.8502,
      "step": 147420
    },
    {
      "epoch": 2.588353025860707,
      "grad_norm": 0.04853816537706338,
      "learning_rate": 9.856703277803249e-05,
      "loss": 0.8456,
      "step": 147430
    },
    {
      "epoch": 2.588528590740708,
      "grad_norm": 0.06121508391020887,
      "learning_rate": 9.855589252644156e-05,
      "loss": 0.8414,
      "step": 147440
    },
    {
      "epoch": 2.5887041556207095,
      "grad_norm": 0.04437211078803025,
      "learning_rate": 9.854475230581794e-05,
      "loss": 0.8339,
      "step": 147450
    },
    {
      "epoch": 2.588879720500711,
      "grad_norm": 0.07079537428045025,
      "learning_rate": 9.853361211630278e-05,
      "loss": 0.8447,
      "step": 147460
    },
    {
      "epoch": 2.5890552853807125,
      "grad_norm": 0.05458759267848393,
      "learning_rate": 9.852247195803719e-05,
      "loss": 0.8484,
      "step": 147470
    },
    {
      "epoch": 2.589230850260714,
      "grad_norm": 0.048226157971619735,
      "learning_rate": 9.851133183116241e-05,
      "loss": 0.8439,
      "step": 147480
    },
    {
      "epoch": 2.589406415140715,
      "grad_norm": 0.05465416180096269,
      "learning_rate": 9.85001917358195e-05,
      "loss": 0.8452,
      "step": 147490
    },
    {
      "epoch": 2.5895819800207165,
      "grad_norm": 0.058749663987442406,
      "learning_rate": 9.848905167214964e-05,
      "loss": 0.8469,
      "step": 147500
    },
    {
      "epoch": 2.589757544900718,
      "grad_norm": 0.0740518944977814,
      "learning_rate": 9.8477911640294e-05,
      "loss": 0.8375,
      "step": 147510
    },
    {
      "epoch": 2.5899331097807194,
      "grad_norm": 0.05514573478608806,
      "learning_rate": 9.84667716403937e-05,
      "loss": 0.8482,
      "step": 147520
    },
    {
      "epoch": 2.590108674660721,
      "grad_norm": 0.0856644123462127,
      "learning_rate": 9.84556316725899e-05,
      "loss": 0.8496,
      "step": 147530
    },
    {
      "epoch": 2.5902842395407224,
      "grad_norm": 0.05269314385537763,
      "learning_rate": 9.844449173702368e-05,
      "loss": 0.8417,
      "step": 147540
    },
    {
      "epoch": 2.590459804420724,
      "grad_norm": 0.04941095457436785,
      "learning_rate": 9.843335183383627e-05,
      "loss": 0.8407,
      "step": 147550
    },
    {
      "epoch": 2.590635369300725,
      "grad_norm": 0.05950228320757808,
      "learning_rate": 9.842221196316883e-05,
      "loss": 0.8409,
      "step": 147560
    },
    {
      "epoch": 2.5908109341807264,
      "grad_norm": 0.050507856740222803,
      "learning_rate": 9.841107212516239e-05,
      "loss": 0.8404,
      "step": 147570
    },
    {
      "epoch": 2.590986499060728,
      "grad_norm": 0.04948154635320294,
      "learning_rate": 9.839993231995819e-05,
      "loss": 0.8459,
      "step": 147580
    },
    {
      "epoch": 2.5911620639407293,
      "grad_norm": 0.04823070699373961,
      "learning_rate": 9.838879254769735e-05,
      "loss": 0.8426,
      "step": 147590
    },
    {
      "epoch": 2.591337628820731,
      "grad_norm": 0.04937029494831054,
      "learning_rate": 9.837765280852101e-05,
      "loss": 0.8472,
      "step": 147600
    },
    {
      "epoch": 2.591513193700732,
      "grad_norm": 0.05753203884437324,
      "learning_rate": 9.83665131025703e-05,
      "loss": 0.8454,
      "step": 147610
    },
    {
      "epoch": 2.5916887585807338,
      "grad_norm": 0.06463382112193573,
      "learning_rate": 9.835537342998639e-05,
      "loss": 0.8493,
      "step": 147620
    },
    {
      "epoch": 2.591864323460735,
      "grad_norm": 0.058510801296406456,
      "learning_rate": 9.834423379091038e-05,
      "loss": 0.8376,
      "step": 147630
    },
    {
      "epoch": 2.5920398883407363,
      "grad_norm": 0.058895582633066774,
      "learning_rate": 9.833309418548344e-05,
      "loss": 0.845,
      "step": 147640
    },
    {
      "epoch": 2.5922154532207378,
      "grad_norm": 0.057129868886500686,
      "learning_rate": 9.832195461384672e-05,
      "loss": 0.8498,
      "step": 147650
    },
    {
      "epoch": 2.5923910181007392,
      "grad_norm": 0.05135041684763619,
      "learning_rate": 9.831081507614136e-05,
      "loss": 0.8399,
      "step": 147660
    },
    {
      "epoch": 2.5925665829807407,
      "grad_norm": 0.04862838960195055,
      "learning_rate": 9.829967557250846e-05,
      "loss": 0.848,
      "step": 147670
    },
    {
      "epoch": 2.5927421478607418,
      "grad_norm": 0.046818654094750174,
      "learning_rate": 9.828853610308921e-05,
      "loss": 0.8526,
      "step": 147680
    },
    {
      "epoch": 2.5929177127407432,
      "grad_norm": 0.07216657285794623,
      "learning_rate": 9.827739666802473e-05,
      "loss": 0.8433,
      "step": 147690
    },
    {
      "epoch": 2.5930932776207447,
      "grad_norm": 0.05258533587038273,
      "learning_rate": 9.826625726745616e-05,
      "loss": 0.8427,
      "step": 147700
    },
    {
      "epoch": 2.593268842500746,
      "grad_norm": 0.06267471533086195,
      "learning_rate": 9.825511790152459e-05,
      "loss": 0.8451,
      "step": 147710
    },
    {
      "epoch": 2.5934444073807477,
      "grad_norm": 0.058913777732563906,
      "learning_rate": 9.824397857037123e-05,
      "loss": 0.8502,
      "step": 147720
    },
    {
      "epoch": 2.5936199722607487,
      "grad_norm": 0.0985323161287128,
      "learning_rate": 9.823283927413724e-05,
      "loss": 0.8398,
      "step": 147730
    },
    {
      "epoch": 2.5937955371407506,
      "grad_norm": 0.0591748145872972,
      "learning_rate": 9.822170001296362e-05,
      "loss": 0.8416,
      "step": 147740
    },
    {
      "epoch": 2.5939711020207517,
      "grad_norm": 0.048353865559411326,
      "learning_rate": 9.821056078699163e-05,
      "loss": 0.8433,
      "step": 147750
    },
    {
      "epoch": 2.594146666900753,
      "grad_norm": 0.05181039510104841,
      "learning_rate": 9.819942159636238e-05,
      "loss": 0.8448,
      "step": 147760
    },
    {
      "epoch": 2.5943222317807546,
      "grad_norm": 0.05081592328824589,
      "learning_rate": 9.818828244121701e-05,
      "loss": 0.8488,
      "step": 147770
    },
    {
      "epoch": 2.594497796660756,
      "grad_norm": 0.06100357490359965,
      "learning_rate": 9.81771433216966e-05,
      "loss": 0.8468,
      "step": 147780
    },
    {
      "epoch": 2.5946733615407576,
      "grad_norm": 0.06202976605167687,
      "learning_rate": 9.816600423794235e-05,
      "loss": 0.8389,
      "step": 147790
    },
    {
      "epoch": 2.5948489264207586,
      "grad_norm": 0.06234411669547329,
      "learning_rate": 9.815486519009537e-05,
      "loss": 0.8576,
      "step": 147800
    },
    {
      "epoch": 2.59502449130076,
      "grad_norm": 0.06452239213985769,
      "learning_rate": 9.814372617829675e-05,
      "loss": 0.8406,
      "step": 147810
    },
    {
      "epoch": 2.5952000561807615,
      "grad_norm": 0.08223888370568902,
      "learning_rate": 9.813258720268772e-05,
      "loss": 0.8406,
      "step": 147820
    },
    {
      "epoch": 2.595375621060763,
      "grad_norm": 0.05999489348770206,
      "learning_rate": 9.812144826340935e-05,
      "loss": 0.8419,
      "step": 147830
    },
    {
      "epoch": 2.5955511859407645,
      "grad_norm": 0.07512940446740234,
      "learning_rate": 9.811030936060276e-05,
      "loss": 0.8422,
      "step": 147840
    },
    {
      "epoch": 2.5957267508207655,
      "grad_norm": 0.051881934186704934,
      "learning_rate": 9.809917049440913e-05,
      "loss": 0.843,
      "step": 147850
    },
    {
      "epoch": 2.5959023157007675,
      "grad_norm": 0.05858031966936865,
      "learning_rate": 9.808803166496956e-05,
      "loss": 0.843,
      "step": 147860
    },
    {
      "epoch": 2.5960778805807685,
      "grad_norm": 0.06454072851231737,
      "learning_rate": 9.807689287242518e-05,
      "loss": 0.8481,
      "step": 147870
    },
    {
      "epoch": 2.59625344546077,
      "grad_norm": 0.05290184040292287,
      "learning_rate": 9.806575411691711e-05,
      "loss": 0.8339,
      "step": 147880
    },
    {
      "epoch": 2.5964290103407714,
      "grad_norm": 0.06573344261943087,
      "learning_rate": 9.805461539858652e-05,
      "loss": 0.838,
      "step": 147890
    },
    {
      "epoch": 2.596604575220773,
      "grad_norm": 0.05216866014009823,
      "learning_rate": 9.804347671757455e-05,
      "loss": 0.8402,
      "step": 147900
    },
    {
      "epoch": 2.5967801401007744,
      "grad_norm": 0.06078928904316723,
      "learning_rate": 9.803233807402221e-05,
      "loss": 0.8389,
      "step": 147910
    },
    {
      "epoch": 2.5969557049807754,
      "grad_norm": 0.05936017588007188,
      "learning_rate": 9.802119946807079e-05,
      "loss": 0.8422,
      "step": 147920
    },
    {
      "epoch": 2.597131269860777,
      "grad_norm": 0.06610648339368014,
      "learning_rate": 9.801006089986132e-05,
      "loss": 0.8435,
      "step": 147930
    },
    {
      "epoch": 2.5973068347407784,
      "grad_norm": 0.05516239202311649,
      "learning_rate": 9.799892236953496e-05,
      "loss": 0.8456,
      "step": 147940
    },
    {
      "epoch": 2.59748239962078,
      "grad_norm": 0.05530881850135712,
      "learning_rate": 9.798778387723281e-05,
      "loss": 0.8483,
      "step": 147950
    },
    {
      "epoch": 2.5976579645007813,
      "grad_norm": 0.07090014561902151,
      "learning_rate": 9.797664542309603e-05,
      "loss": 0.851,
      "step": 147960
    },
    {
      "epoch": 2.597833529380783,
      "grad_norm": 0.05371929783575964,
      "learning_rate": 9.796550700726575e-05,
      "loss": 0.8383,
      "step": 147970
    },
    {
      "epoch": 2.5980090942607843,
      "grad_norm": 0.06557406484698074,
      "learning_rate": 9.795436862988301e-05,
      "loss": 0.844,
      "step": 147980
    },
    {
      "epoch": 2.5981846591407853,
      "grad_norm": 0.04401152068852839,
      "learning_rate": 9.794323029108906e-05,
      "loss": 0.8409,
      "step": 147990
    },
    {
      "epoch": 2.598360224020787,
      "grad_norm": 0.04878131381248311,
      "learning_rate": 9.793209199102496e-05,
      "loss": 0.846,
      "step": 148000
    },
    {
      "epoch": 2.5985357889007883,
      "grad_norm": 0.051657241136799946,
      "learning_rate": 9.792095372983182e-05,
      "loss": 0.845,
      "step": 148010
    },
    {
      "epoch": 2.5987113537807898,
      "grad_norm": 0.07040866577279255,
      "learning_rate": 9.790981550765082e-05,
      "loss": 0.8388,
      "step": 148020
    },
    {
      "epoch": 2.5988869186607912,
      "grad_norm": 0.05485597771308122,
      "learning_rate": 9.789867732462303e-05,
      "loss": 0.8396,
      "step": 148030
    },
    {
      "epoch": 2.5990624835407923,
      "grad_norm": 0.05379423769304707,
      "learning_rate": 9.788753918088961e-05,
      "loss": 0.837,
      "step": 148040
    },
    {
      "epoch": 2.5992380484207938,
      "grad_norm": 0.07001354579574294,
      "learning_rate": 9.787640107659161e-05,
      "loss": 0.8328,
      "step": 148050
    },
    {
      "epoch": 2.5994136133007952,
      "grad_norm": 0.05009994280985529,
      "learning_rate": 9.786526301187028e-05,
      "loss": 0.8482,
      "step": 148060
    },
    {
      "epoch": 2.5995891781807967,
      "grad_norm": 0.07006126651661213,
      "learning_rate": 9.785412498686667e-05,
      "loss": 0.8429,
      "step": 148070
    },
    {
      "epoch": 2.599764743060798,
      "grad_norm": 0.04711642015508274,
      "learning_rate": 9.784298700172182e-05,
      "loss": 0.8449,
      "step": 148080
    },
    {
      "epoch": 2.5999403079407997,
      "grad_norm": 0.04468429688079152,
      "learning_rate": 9.783184905657699e-05,
      "loss": 0.8406,
      "step": 148090
    },
    {
      "epoch": 2.600115872820801,
      "grad_norm": 0.04941276705293672,
      "learning_rate": 9.782071115157324e-05,
      "loss": 0.8409,
      "step": 148100
    },
    {
      "epoch": 2.600291437700802,
      "grad_norm": 0.06808846627379002,
      "learning_rate": 9.780957328685169e-05,
      "loss": 0.8392,
      "step": 148110
    },
    {
      "epoch": 2.6004670025808037,
      "grad_norm": 0.06596324553933017,
      "learning_rate": 9.779843546255343e-05,
      "loss": 0.8493,
      "step": 148120
    },
    {
      "epoch": 2.600642567460805,
      "grad_norm": 0.05818817958491927,
      "learning_rate": 9.778729767881961e-05,
      "loss": 0.8337,
      "step": 148130
    },
    {
      "epoch": 2.6008181323408066,
      "grad_norm": 0.04987284966226184,
      "learning_rate": 9.777615993579137e-05,
      "loss": 0.8414,
      "step": 148140
    },
    {
      "epoch": 2.600993697220808,
      "grad_norm": 0.06169261766583526,
      "learning_rate": 9.776502223360977e-05,
      "loss": 0.8445,
      "step": 148150
    },
    {
      "epoch": 2.601169262100809,
      "grad_norm": 0.05056205358891741,
      "learning_rate": 9.775388457241596e-05,
      "loss": 0.8423,
      "step": 148160
    },
    {
      "epoch": 2.6013448269808106,
      "grad_norm": 0.054912617720560566,
      "learning_rate": 9.774274695235109e-05,
      "loss": 0.8386,
      "step": 148170
    },
    {
      "epoch": 2.601520391860812,
      "grad_norm": 0.05911986632084129,
      "learning_rate": 9.77316093735562e-05,
      "loss": 0.8458,
      "step": 148180
    },
    {
      "epoch": 2.6016959567408136,
      "grad_norm": 0.054224299966651957,
      "learning_rate": 9.772047183617248e-05,
      "loss": 0.8389,
      "step": 148190
    },
    {
      "epoch": 2.601871521620815,
      "grad_norm": 0.08677448777445855,
      "learning_rate": 9.7709334340341e-05,
      "loss": 0.841,
      "step": 148200
    },
    {
      "epoch": 2.6020470865008165,
      "grad_norm": 0.07499928057974105,
      "learning_rate": 9.769819688620287e-05,
      "loss": 0.8386,
      "step": 148210
    },
    {
      "epoch": 2.602222651380818,
      "grad_norm": 0.05566648680345528,
      "learning_rate": 9.76870594738992e-05,
      "loss": 0.8454,
      "step": 148220
    },
    {
      "epoch": 2.602398216260819,
      "grad_norm": 0.05643924788458931,
      "learning_rate": 9.767592210357115e-05,
      "loss": 0.8481,
      "step": 148230
    },
    {
      "epoch": 2.6025737811408205,
      "grad_norm": 0.05671698263214561,
      "learning_rate": 9.766478477535984e-05,
      "loss": 0.8484,
      "step": 148240
    },
    {
      "epoch": 2.602749346020822,
      "grad_norm": 0.04914611743769659,
      "learning_rate": 9.765364748940627e-05,
      "loss": 0.8473,
      "step": 148250
    },
    {
      "epoch": 2.6029249109008235,
      "grad_norm": 0.05484894128136698,
      "learning_rate": 9.764251024585169e-05,
      "loss": 0.8421,
      "step": 148260
    },
    {
      "epoch": 2.603100475780825,
      "grad_norm": 0.05236877720333318,
      "learning_rate": 9.763137304483713e-05,
      "loss": 0.846,
      "step": 148270
    },
    {
      "epoch": 2.603276040660826,
      "grad_norm": 0.06072420547956174,
      "learning_rate": 9.762023588650372e-05,
      "loss": 0.839,
      "step": 148280
    },
    {
      "epoch": 2.603451605540828,
      "grad_norm": 0.04361133423246296,
      "learning_rate": 9.760909877099254e-05,
      "loss": 0.8461,
      "step": 148290
    },
    {
      "epoch": 2.603627170420829,
      "grad_norm": 0.0445805762939854,
      "learning_rate": 9.759796169844478e-05,
      "loss": 0.8418,
      "step": 148300
    },
    {
      "epoch": 2.6038027353008304,
      "grad_norm": 0.04898657118516653,
      "learning_rate": 9.758682466900146e-05,
      "loss": 0.8436,
      "step": 148310
    },
    {
      "epoch": 2.603978300180832,
      "grad_norm": 0.0580903403959444,
      "learning_rate": 9.75756876828037e-05,
      "loss": 0.8378,
      "step": 148320
    },
    {
      "epoch": 2.6041538650608334,
      "grad_norm": 0.06518444350154497,
      "learning_rate": 9.756455073999268e-05,
      "loss": 0.8558,
      "step": 148330
    },
    {
      "epoch": 2.604329429940835,
      "grad_norm": 0.05110578587179342,
      "learning_rate": 9.755341384070946e-05,
      "loss": 0.8447,
      "step": 148340
    },
    {
      "epoch": 2.604504994820836,
      "grad_norm": 0.05095964823622443,
      "learning_rate": 9.754227698509512e-05,
      "loss": 0.8367,
      "step": 148350
    },
    {
      "epoch": 2.6046805597008373,
      "grad_norm": 0.0518030373832241,
      "learning_rate": 9.753114017329082e-05,
      "loss": 0.8398,
      "step": 148360
    },
    {
      "epoch": 2.604856124580839,
      "grad_norm": 0.06351048006430467,
      "learning_rate": 9.752000340543765e-05,
      "loss": 0.8465,
      "step": 148370
    },
    {
      "epoch": 2.6050316894608403,
      "grad_norm": 0.05398249659584223,
      "learning_rate": 9.750886668167668e-05,
      "loss": 0.8508,
      "step": 148380
    },
    {
      "epoch": 2.6052072543408418,
      "grad_norm": 0.06688135161491413,
      "learning_rate": 9.7497730002149e-05,
      "loss": 0.8431,
      "step": 148390
    },
    {
      "epoch": 2.605382819220843,
      "grad_norm": 0.045836521170116704,
      "learning_rate": 9.748659336699579e-05,
      "loss": 0.8393,
      "step": 148400
    },
    {
      "epoch": 2.6055583841008447,
      "grad_norm": 0.0668292468147587,
      "learning_rate": 9.747545677635815e-05,
      "loss": 0.8374,
      "step": 148410
    },
    {
      "epoch": 2.6057339489808458,
      "grad_norm": 0.06950155121043203,
      "learning_rate": 9.746432023037709e-05,
      "loss": 0.8474,
      "step": 148420
    },
    {
      "epoch": 2.6059095138608472,
      "grad_norm": 0.05780111044925125,
      "learning_rate": 9.74531837291938e-05,
      "loss": 0.8435,
      "step": 148430
    },
    {
      "epoch": 2.6060850787408487,
      "grad_norm": 0.05202676176255151,
      "learning_rate": 9.744204727294936e-05,
      "loss": 0.8451,
      "step": 148440
    },
    {
      "epoch": 2.60626064362085,
      "grad_norm": 0.05076461741091999,
      "learning_rate": 9.743091086178486e-05,
      "loss": 0.8432,
      "step": 148450
    },
    {
      "epoch": 2.6064362085008517,
      "grad_norm": 0.0856210307720876,
      "learning_rate": 9.741977449584137e-05,
      "loss": 0.8386,
      "step": 148460
    },
    {
      "epoch": 2.6066117733808527,
      "grad_norm": 0.09024446301937361,
      "learning_rate": 9.740863817526006e-05,
      "loss": 0.8451,
      "step": 148470
    },
    {
      "epoch": 2.606787338260854,
      "grad_norm": 0.06889585343774948,
      "learning_rate": 9.7397501900182e-05,
      "loss": 0.8444,
      "step": 148480
    },
    {
      "epoch": 2.6069629031408557,
      "grad_norm": 0.050642845654905635,
      "learning_rate": 9.738636567074822e-05,
      "loss": 0.8434,
      "step": 148490
    },
    {
      "epoch": 2.607138468020857,
      "grad_norm": 0.05199152864105722,
      "learning_rate": 9.737522948709993e-05,
      "loss": 0.8373,
      "step": 148500
    },
    {
      "epoch": 2.6073140329008586,
      "grad_norm": 0.05818131506435347,
      "learning_rate": 9.736409334937817e-05,
      "loss": 0.8449,
      "step": 148510
    },
    {
      "epoch": 2.6074895977808596,
      "grad_norm": 0.104043923644471,
      "learning_rate": 9.735295725772403e-05,
      "loss": 0.8446,
      "step": 148520
    },
    {
      "epoch": 2.6076651626608616,
      "grad_norm": 0.04754962112220924,
      "learning_rate": 9.734182121227865e-05,
      "loss": 0.8493,
      "step": 148530
    },
    {
      "epoch": 2.6078407275408626,
      "grad_norm": 0.07218904838363353,
      "learning_rate": 9.733068521318309e-05,
      "loss": 0.8439,
      "step": 148540
    },
    {
      "epoch": 2.608016292420864,
      "grad_norm": 0.05577658443820227,
      "learning_rate": 9.731954926057843e-05,
      "loss": 0.8444,
      "step": 148550
    },
    {
      "epoch": 2.6081918573008656,
      "grad_norm": 0.05330034014568891,
      "learning_rate": 9.730841335460576e-05,
      "loss": 0.846,
      "step": 148560
    },
    {
      "epoch": 2.608367422180867,
      "grad_norm": 0.0709464335160026,
      "learning_rate": 9.729727749540625e-05,
      "loss": 0.8414,
      "step": 148570
    },
    {
      "epoch": 2.6085429870608685,
      "grad_norm": 0.04605868413401664,
      "learning_rate": 9.728614168312097e-05,
      "loss": 0.8484,
      "step": 148580
    },
    {
      "epoch": 2.6087185519408695,
      "grad_norm": 0.07649715417571938,
      "learning_rate": 9.72750059178909e-05,
      "loss": 0.8498,
      "step": 148590
    },
    {
      "epoch": 2.608894116820871,
      "grad_norm": 0.05506703025058585,
      "learning_rate": 9.726387019985729e-05,
      "loss": 0.8565,
      "step": 148600
    },
    {
      "epoch": 2.6090696817008725,
      "grad_norm": 0.047689529500663096,
      "learning_rate": 9.725273452916112e-05,
      "loss": 0.8349,
      "step": 148610
    },
    {
      "epoch": 2.609245246580874,
      "grad_norm": 0.06589454659922182,
      "learning_rate": 9.724159890594356e-05,
      "loss": 0.8379,
      "step": 148620
    },
    {
      "epoch": 2.6094208114608755,
      "grad_norm": 0.04737852614201036,
      "learning_rate": 9.723046333034562e-05,
      "loss": 0.8453,
      "step": 148630
    },
    {
      "epoch": 2.609596376340877,
      "grad_norm": 0.06260414369602472,
      "learning_rate": 9.721932780250846e-05,
      "loss": 0.8416,
      "step": 148640
    },
    {
      "epoch": 2.6097719412208784,
      "grad_norm": 0.04487099986055861,
      "learning_rate": 9.720819232257315e-05,
      "loss": 0.8444,
      "step": 148650
    },
    {
      "epoch": 2.6099475061008794,
      "grad_norm": 0.06605675306595325,
      "learning_rate": 9.719705689068072e-05,
      "loss": 0.8488,
      "step": 148660
    },
    {
      "epoch": 2.610123070980881,
      "grad_norm": 0.052495614480569136,
      "learning_rate": 9.718592150697234e-05,
      "loss": 0.845,
      "step": 148670
    },
    {
      "epoch": 2.6102986358608824,
      "grad_norm": 0.04494260050418276,
      "learning_rate": 9.71747861715891e-05,
      "loss": 0.8403,
      "step": 148680
    },
    {
      "epoch": 2.610474200740884,
      "grad_norm": 0.05254935338264068,
      "learning_rate": 9.716365088467198e-05,
      "loss": 0.839,
      "step": 148690
    },
    {
      "epoch": 2.6106497656208854,
      "grad_norm": 0.08301725609118399,
      "learning_rate": 9.715251564636219e-05,
      "loss": 0.8437,
      "step": 148700
    },
    {
      "epoch": 2.6108253305008864,
      "grad_norm": 0.07827085311437729,
      "learning_rate": 9.714138045680077e-05,
      "loss": 0.8556,
      "step": 148710
    },
    {
      "epoch": 2.611000895380888,
      "grad_norm": 0.048498017735527664,
      "learning_rate": 9.713024531612879e-05,
      "loss": 0.8411,
      "step": 148720
    },
    {
      "epoch": 2.6111764602608893,
      "grad_norm": 0.05386358688663874,
      "learning_rate": 9.71191102244873e-05,
      "loss": 0.8513,
      "step": 148730
    },
    {
      "epoch": 2.611352025140891,
      "grad_norm": 0.0431439081703766,
      "learning_rate": 9.710797518201747e-05,
      "loss": 0.8487,
      "step": 148740
    },
    {
      "epoch": 2.6115275900208923,
      "grad_norm": 0.05590396130367564,
      "learning_rate": 9.709684018886036e-05,
      "loss": 0.8319,
      "step": 148750
    },
    {
      "epoch": 2.6117031549008938,
      "grad_norm": 0.06788566502568791,
      "learning_rate": 9.708570524515697e-05,
      "loss": 0.8407,
      "step": 148760
    },
    {
      "epoch": 2.6118787197808953,
      "grad_norm": 0.05651921240583665,
      "learning_rate": 9.70745703510485e-05,
      "loss": 0.8456,
      "step": 148770
    },
    {
      "epoch": 2.6120542846608963,
      "grad_norm": 0.04944660496525395,
      "learning_rate": 9.706343550667596e-05,
      "loss": 0.8487,
      "step": 148780
    },
    {
      "epoch": 2.6122298495408978,
      "grad_norm": 0.06052856477628428,
      "learning_rate": 9.705230071218045e-05,
      "loss": 0.8434,
      "step": 148790
    },
    {
      "epoch": 2.6124054144208992,
      "grad_norm": 0.04323274399365831,
      "learning_rate": 9.704116596770304e-05,
      "loss": 0.8408,
      "step": 148800
    },
    {
      "epoch": 2.6125809793009007,
      "grad_norm": 0.04717051804950489,
      "learning_rate": 9.70300312733848e-05,
      "loss": 0.8432,
      "step": 148810
    },
    {
      "epoch": 2.612756544180902,
      "grad_norm": 0.05747838920389176,
      "learning_rate": 9.701889662936685e-05,
      "loss": 0.8413,
      "step": 148820
    },
    {
      "epoch": 2.6129321090609032,
      "grad_norm": 0.05291222122754614,
      "learning_rate": 9.70077620357902e-05,
      "loss": 0.8394,
      "step": 148830
    },
    {
      "epoch": 2.6131076739409047,
      "grad_norm": 0.05333152286896178,
      "learning_rate": 9.699662749279601e-05,
      "loss": 0.8437,
      "step": 148840
    },
    {
      "epoch": 2.613283238820906,
      "grad_norm": 0.056449483987345055,
      "learning_rate": 9.698549300052531e-05,
      "loss": 0.84,
      "step": 148850
    },
    {
      "epoch": 2.6134588037009077,
      "grad_norm": 0.056425295251929124,
      "learning_rate": 9.697435855911917e-05,
      "loss": 0.8513,
      "step": 148860
    },
    {
      "epoch": 2.613634368580909,
      "grad_norm": 0.05992418870810401,
      "learning_rate": 9.696322416871872e-05,
      "loss": 0.8493,
      "step": 148870
    },
    {
      "epoch": 2.6138099334609106,
      "grad_norm": 0.05400688067172821,
      "learning_rate": 9.695208982946495e-05,
      "loss": 0.8359,
      "step": 148880
    },
    {
      "epoch": 2.613985498340912,
      "grad_norm": 0.04912363729480335,
      "learning_rate": 9.694095554149901e-05,
      "loss": 0.8433,
      "step": 148890
    },
    {
      "epoch": 2.614161063220913,
      "grad_norm": 0.05642845369998292,
      "learning_rate": 9.692982130496189e-05,
      "loss": 0.8375,
      "step": 148900
    },
    {
      "epoch": 2.6143366281009146,
      "grad_norm": 0.057219470030742424,
      "learning_rate": 9.691868711999476e-05,
      "loss": 0.8412,
      "step": 148910
    },
    {
      "epoch": 2.614512192980916,
      "grad_norm": 0.05936441481125335,
      "learning_rate": 9.690755298673866e-05,
      "loss": 0.841,
      "step": 148920
    },
    {
      "epoch": 2.6146877578609176,
      "grad_norm": 0.051039809484219024,
      "learning_rate": 9.689641890533459e-05,
      "loss": 0.8432,
      "step": 148930
    },
    {
      "epoch": 2.614863322740919,
      "grad_norm": 0.04557116494351116,
      "learning_rate": 9.688528487592372e-05,
      "loss": 0.8522,
      "step": 148940
    },
    {
      "epoch": 2.61503888762092,
      "grad_norm": 0.05211617273327291,
      "learning_rate": 9.687415089864707e-05,
      "loss": 0.8407,
      "step": 148950
    },
    {
      "epoch": 2.6152144525009215,
      "grad_norm": 0.05548170093340263,
      "learning_rate": 9.686301697364573e-05,
      "loss": 0.8427,
      "step": 148960
    },
    {
      "epoch": 2.615390017380923,
      "grad_norm": 0.06847118055815275,
      "learning_rate": 9.685188310106073e-05,
      "loss": 0.8413,
      "step": 148970
    },
    {
      "epoch": 2.6155655822609245,
      "grad_norm": 0.08072311061194068,
      "learning_rate": 9.684074928103319e-05,
      "loss": 0.836,
      "step": 148980
    },
    {
      "epoch": 2.615741147140926,
      "grad_norm": 0.0632325503496503,
      "learning_rate": 9.682961551370415e-05,
      "loss": 0.8418,
      "step": 148990
    },
    {
      "epoch": 2.6159167120209275,
      "grad_norm": 0.04135816947972696,
      "learning_rate": 9.681848179921464e-05,
      "loss": 0.8457,
      "step": 149000
    },
    {
      "epoch": 2.616092276900929,
      "grad_norm": 0.0532048371229542,
      "learning_rate": 9.680734813770582e-05,
      "loss": 0.8402,
      "step": 149010
    },
    {
      "epoch": 2.61626784178093,
      "grad_norm": 0.05285669322215486,
      "learning_rate": 9.67962145293187e-05,
      "loss": 0.836,
      "step": 149020
    },
    {
      "epoch": 2.6164434066609314,
      "grad_norm": 0.09505545557798047,
      "learning_rate": 9.678508097419433e-05,
      "loss": 0.8449,
      "step": 149030
    },
    {
      "epoch": 2.616618971540933,
      "grad_norm": 0.05381000512369598,
      "learning_rate": 9.677394747247379e-05,
      "loss": 0.8398,
      "step": 149040
    },
    {
      "epoch": 2.6167945364209344,
      "grad_norm": 0.04500362958014994,
      "learning_rate": 9.676281402429817e-05,
      "loss": 0.8482,
      "step": 149050
    },
    {
      "epoch": 2.616970101300936,
      "grad_norm": 0.08531100218193165,
      "learning_rate": 9.675168062980849e-05,
      "loss": 0.8441,
      "step": 149060
    },
    {
      "epoch": 2.617145666180937,
      "grad_norm": 0.04890135908927714,
      "learning_rate": 9.674054728914579e-05,
      "loss": 0.8511,
      "step": 149070
    },
    {
      "epoch": 2.617321231060939,
      "grad_norm": 0.054678410601242705,
      "learning_rate": 9.672941400245122e-05,
      "loss": 0.8357,
      "step": 149080
    },
    {
      "epoch": 2.61749679594094,
      "grad_norm": 0.04978985577691709,
      "learning_rate": 9.67182807698658e-05,
      "loss": 0.8408,
      "step": 149090
    },
    {
      "epoch": 2.6176723608209413,
      "grad_norm": 0.05378695151685628,
      "learning_rate": 9.670714759153056e-05,
      "loss": 0.8408,
      "step": 149100
    },
    {
      "epoch": 2.617847925700943,
      "grad_norm": 0.05405855527018492,
      "learning_rate": 9.669601446758659e-05,
      "loss": 0.84,
      "step": 149110
    },
    {
      "epoch": 2.6180234905809443,
      "grad_norm": 0.048510827612217054,
      "learning_rate": 9.668488139817496e-05,
      "loss": 0.8454,
      "step": 149120
    },
    {
      "epoch": 2.618199055460946,
      "grad_norm": 0.04675100446415186,
      "learning_rate": 9.667374838343668e-05,
      "loss": 0.8383,
      "step": 149130
    },
    {
      "epoch": 2.618374620340947,
      "grad_norm": 0.05401832624829268,
      "learning_rate": 9.666261542351285e-05,
      "loss": 0.8432,
      "step": 149140
    },
    {
      "epoch": 2.6185501852209483,
      "grad_norm": 0.042931441195958084,
      "learning_rate": 9.665148251854452e-05,
      "loss": 0.843,
      "step": 149150
    },
    {
      "epoch": 2.6187257501009498,
      "grad_norm": 0.046612459582340095,
      "learning_rate": 9.664034966867273e-05,
      "loss": 0.8429,
      "step": 149160
    },
    {
      "epoch": 2.6189013149809512,
      "grad_norm": 0.05229556015044999,
      "learning_rate": 9.662921687403853e-05,
      "loss": 0.8478,
      "step": 149170
    },
    {
      "epoch": 2.6190768798609527,
      "grad_norm": 0.038333504747604684,
      "learning_rate": 9.661808413478302e-05,
      "loss": 0.846,
      "step": 149180
    },
    {
      "epoch": 2.6192524447409538,
      "grad_norm": 0.041494572761358206,
      "learning_rate": 9.660695145104721e-05,
      "loss": 0.8372,
      "step": 149190
    },
    {
      "epoch": 2.6194280096209557,
      "grad_norm": 0.05760031269854195,
      "learning_rate": 9.659581882297216e-05,
      "loss": 0.8492,
      "step": 149200
    },
    {
      "epoch": 2.6196035745009567,
      "grad_norm": 0.06922416803801278,
      "learning_rate": 9.658468625069895e-05,
      "loss": 0.8495,
      "step": 149210
    },
    {
      "epoch": 2.619779139380958,
      "grad_norm": 0.05868728444926621,
      "learning_rate": 9.657355373436862e-05,
      "loss": 0.8407,
      "step": 149220
    },
    {
      "epoch": 2.6199547042609597,
      "grad_norm": 0.0528337708505043,
      "learning_rate": 9.656242127412218e-05,
      "loss": 0.8479,
      "step": 149230
    },
    {
      "epoch": 2.620130269140961,
      "grad_norm": 0.07178368285549026,
      "learning_rate": 9.65512888701007e-05,
      "loss": 0.8409,
      "step": 149240
    },
    {
      "epoch": 2.6203058340209626,
      "grad_norm": 0.05137993523324375,
      "learning_rate": 9.654015652244527e-05,
      "loss": 0.8397,
      "step": 149250
    },
    {
      "epoch": 2.6204813989009637,
      "grad_norm": 0.0564926233076178,
      "learning_rate": 9.652902423129694e-05,
      "loss": 0.8358,
      "step": 149260
    },
    {
      "epoch": 2.620656963780965,
      "grad_norm": 0.06161737915464768,
      "learning_rate": 9.651789199679667e-05,
      "loss": 0.848,
      "step": 149270
    },
    {
      "epoch": 2.6208325286609666,
      "grad_norm": 0.056535459949189425,
      "learning_rate": 9.650675981908561e-05,
      "loss": 0.8376,
      "step": 149280
    },
    {
      "epoch": 2.621008093540968,
      "grad_norm": 0.05489480689404442,
      "learning_rate": 9.649562769830475e-05,
      "loss": 0.8488,
      "step": 149290
    },
    {
      "epoch": 2.6211836584209696,
      "grad_norm": 0.05544587426793913,
      "learning_rate": 9.648449563459517e-05,
      "loss": 0.8436,
      "step": 149300
    },
    {
      "epoch": 2.6213592233009706,
      "grad_norm": 0.049120120833366705,
      "learning_rate": 9.647336362809787e-05,
      "loss": 0.8451,
      "step": 149310
    },
    {
      "epoch": 2.6215347881809725,
      "grad_norm": 0.054385955676251284,
      "learning_rate": 9.646223167895393e-05,
      "loss": 0.8473,
      "step": 149320
    },
    {
      "epoch": 2.6217103530609736,
      "grad_norm": 0.0437482921435556,
      "learning_rate": 9.645109978730438e-05,
      "loss": 0.8396,
      "step": 149330
    },
    {
      "epoch": 2.621885917940975,
      "grad_norm": 0.06544402391853625,
      "learning_rate": 9.643996795329025e-05,
      "loss": 0.8423,
      "step": 149340
    },
    {
      "epoch": 2.6220614828209765,
      "grad_norm": 0.07954463741373206,
      "learning_rate": 9.642883617705263e-05,
      "loss": 0.842,
      "step": 149350
    },
    {
      "epoch": 2.622237047700978,
      "grad_norm": 0.045045171361275324,
      "learning_rate": 9.641770445873251e-05,
      "loss": 0.8411,
      "step": 149360
    },
    {
      "epoch": 2.6224126125809795,
      "grad_norm": 0.06386724070672639,
      "learning_rate": 9.640657279847097e-05,
      "loss": 0.8503,
      "step": 149370
    },
    {
      "epoch": 2.6225881774609805,
      "grad_norm": 0.04117305832466428,
      "learning_rate": 9.639544119640902e-05,
      "loss": 0.8451,
      "step": 149380
    },
    {
      "epoch": 2.622763742340982,
      "grad_norm": 0.047621512098159416,
      "learning_rate": 9.638430965268773e-05,
      "loss": 0.8353,
      "step": 149390
    },
    {
      "epoch": 2.6229393072209835,
      "grad_norm": 0.05618669379975567,
      "learning_rate": 9.637317816744813e-05,
      "loss": 0.843,
      "step": 149400
    },
    {
      "epoch": 2.623114872100985,
      "grad_norm": 0.06571374184535919,
      "learning_rate": 9.636204674083119e-05,
      "loss": 0.8478,
      "step": 149410
    },
    {
      "epoch": 2.6232904369809864,
      "grad_norm": 0.06583888023747694,
      "learning_rate": 9.635091537297806e-05,
      "loss": 0.8505,
      "step": 149420
    },
    {
      "epoch": 2.623466001860988,
      "grad_norm": 0.05615538819905938,
      "learning_rate": 9.633978406402974e-05,
      "loss": 0.8426,
      "step": 149430
    },
    {
      "epoch": 2.6236415667409894,
      "grad_norm": 0.06280151972513157,
      "learning_rate": 9.632865281412717e-05,
      "loss": 0.8421,
      "step": 149440
    },
    {
      "epoch": 2.6238171316209904,
      "grad_norm": 0.058037472999262645,
      "learning_rate": 9.631752162341153e-05,
      "loss": 0.8408,
      "step": 149450
    },
    {
      "epoch": 2.623992696500992,
      "grad_norm": 0.06247469062795787,
      "learning_rate": 9.630639049202378e-05,
      "loss": 0.8459,
      "step": 149460
    },
    {
      "epoch": 2.6241682613809933,
      "grad_norm": 0.0530914612154189,
      "learning_rate": 9.629525942010497e-05,
      "loss": 0.8358,
      "step": 149470
    },
    {
      "epoch": 2.624343826260995,
      "grad_norm": 0.060116164971770164,
      "learning_rate": 9.628412840779608e-05,
      "loss": 0.8487,
      "step": 149480
    },
    {
      "epoch": 2.6245193911409963,
      "grad_norm": 0.0546875792433196,
      "learning_rate": 9.627299745523824e-05,
      "loss": 0.8454,
      "step": 149490
    },
    {
      "epoch": 2.6246949560209973,
      "grad_norm": 0.06231463811300292,
      "learning_rate": 9.62618665625724e-05,
      "loss": 0.8446,
      "step": 149500
    },
    {
      "epoch": 2.624870520900999,
      "grad_norm": 0.05270711930962299,
      "learning_rate": 9.625073572993959e-05,
      "loss": 0.8527,
      "step": 149510
    },
    {
      "epoch": 2.6250460857810003,
      "grad_norm": 0.06735497566652919,
      "learning_rate": 9.623960495748092e-05,
      "loss": 0.8404,
      "step": 149520
    },
    {
      "epoch": 2.6252216506610018,
      "grad_norm": 0.07296505654996063,
      "learning_rate": 9.622847424533735e-05,
      "loss": 0.8432,
      "step": 149530
    },
    {
      "epoch": 2.6253972155410032,
      "grad_norm": 0.04971465606045532,
      "learning_rate": 9.62173435936499e-05,
      "loss": 0.8511,
      "step": 149540
    },
    {
      "epoch": 2.6255727804210047,
      "grad_norm": 0.052105548510371705,
      "learning_rate": 9.620621300255966e-05,
      "loss": 0.8387,
      "step": 149550
    },
    {
      "epoch": 2.625748345301006,
      "grad_norm": 0.07696174329201198,
      "learning_rate": 9.619508247220762e-05,
      "loss": 0.8398,
      "step": 149560
    },
    {
      "epoch": 2.6259239101810072,
      "grad_norm": 0.061746581082822834,
      "learning_rate": 9.618395200273479e-05,
      "loss": 0.8479,
      "step": 149570
    },
    {
      "epoch": 2.6260994750610087,
      "grad_norm": 0.0599390640265801,
      "learning_rate": 9.617282159428219e-05,
      "loss": 0.8432,
      "step": 149580
    },
    {
      "epoch": 2.62627503994101,
      "grad_norm": 0.05538749639008634,
      "learning_rate": 9.61616912469909e-05,
      "loss": 0.8327,
      "step": 149590
    },
    {
      "epoch": 2.6264506048210117,
      "grad_norm": 0.05995095263015516,
      "learning_rate": 9.615056096100193e-05,
      "loss": 0.8426,
      "step": 149600
    },
    {
      "epoch": 2.626626169701013,
      "grad_norm": 0.0643223055449359,
      "learning_rate": 9.613943073645621e-05,
      "loss": 0.8444,
      "step": 149610
    },
    {
      "epoch": 2.626801734581014,
      "grad_norm": 0.046671137611403936,
      "learning_rate": 9.612830057349488e-05,
      "loss": 0.8457,
      "step": 149620
    },
    {
      "epoch": 2.6269772994610157,
      "grad_norm": 0.06342235335686192,
      "learning_rate": 9.611717047225892e-05,
      "loss": 0.8355,
      "step": 149630
    },
    {
      "epoch": 2.627152864341017,
      "grad_norm": 0.07119294691488533,
      "learning_rate": 9.610604043288934e-05,
      "loss": 0.849,
      "step": 149640
    },
    {
      "epoch": 2.6273284292210186,
      "grad_norm": 0.05549860431141439,
      "learning_rate": 9.609491045552715e-05,
      "loss": 0.8361,
      "step": 149650
    },
    {
      "epoch": 2.62750399410102,
      "grad_norm": 0.06273018250271999,
      "learning_rate": 9.608378054031342e-05,
      "loss": 0.8321,
      "step": 149660
    },
    {
      "epoch": 2.6276795589810216,
      "grad_norm": 0.05137761371207923,
      "learning_rate": 9.60726506873891e-05,
      "loss": 0.831,
      "step": 149670
    },
    {
      "epoch": 2.627855123861023,
      "grad_norm": 0.05174266619529791,
      "learning_rate": 9.606152089689521e-05,
      "loss": 0.8403,
      "step": 149680
    },
    {
      "epoch": 2.628030688741024,
      "grad_norm": 0.052993014047188834,
      "learning_rate": 9.605039116897285e-05,
      "loss": 0.8409,
      "step": 149690
    },
    {
      "epoch": 2.6282062536210256,
      "grad_norm": 0.07247590198152491,
      "learning_rate": 9.603926150376297e-05,
      "loss": 0.8474,
      "step": 149700
    },
    {
      "epoch": 2.628381818501027,
      "grad_norm": 0.06675279973902536,
      "learning_rate": 9.60281319014066e-05,
      "loss": 0.8458,
      "step": 149710
    },
    {
      "epoch": 2.6285573833810285,
      "grad_norm": 0.055002063965730055,
      "learning_rate": 9.601700236204475e-05,
      "loss": 0.8429,
      "step": 149720
    },
    {
      "epoch": 2.62873294826103,
      "grad_norm": 0.05933811094845235,
      "learning_rate": 9.600587288581844e-05,
      "loss": 0.8426,
      "step": 149730
    },
    {
      "epoch": 2.628908513141031,
      "grad_norm": 0.05216756203591894,
      "learning_rate": 9.599474347286868e-05,
      "loss": 0.8338,
      "step": 149740
    },
    {
      "epoch": 2.629084078021033,
      "grad_norm": 0.04579029953303335,
      "learning_rate": 9.598361412333645e-05,
      "loss": 0.8417,
      "step": 149750
    },
    {
      "epoch": 2.629259642901034,
      "grad_norm": 0.050382687607705944,
      "learning_rate": 9.59724848373628e-05,
      "loss": 0.8499,
      "step": 149760
    },
    {
      "epoch": 2.6294352077810355,
      "grad_norm": 0.06520660662691584,
      "learning_rate": 9.596135561508878e-05,
      "loss": 0.8454,
      "step": 149770
    },
    {
      "epoch": 2.629610772661037,
      "grad_norm": 0.052319997823392986,
      "learning_rate": 9.595022645665527e-05,
      "loss": 0.8372,
      "step": 149780
    },
    {
      "epoch": 2.6297863375410384,
      "grad_norm": 0.07419277194970694,
      "learning_rate": 9.593909736220341e-05,
      "loss": 0.8428,
      "step": 149790
    },
    {
      "epoch": 2.62996190242104,
      "grad_norm": 0.06114662177023278,
      "learning_rate": 9.592796833187415e-05,
      "loss": 0.845,
      "step": 149800
    },
    {
      "epoch": 2.630137467301041,
      "grad_norm": 0.05312003594657317,
      "learning_rate": 9.591683936580851e-05,
      "loss": 0.834,
      "step": 149810
    },
    {
      "epoch": 2.6303130321810424,
      "grad_norm": 0.06303346912270143,
      "learning_rate": 9.590571046414746e-05,
      "loss": 0.8414,
      "step": 149820
    },
    {
      "epoch": 2.630488597061044,
      "grad_norm": 0.06963264108001856,
      "learning_rate": 9.589458162703206e-05,
      "loss": 0.8418,
      "step": 149830
    },
    {
      "epoch": 2.6306641619410454,
      "grad_norm": 0.0588116277803903,
      "learning_rate": 9.58834528546033e-05,
      "loss": 0.843,
      "step": 149840
    },
    {
      "epoch": 2.630839726821047,
      "grad_norm": 0.0496700951633619,
      "learning_rate": 9.587232414700212e-05,
      "loss": 0.8399,
      "step": 149850
    },
    {
      "epoch": 2.631015291701048,
      "grad_norm": 0.05341327007339182,
      "learning_rate": 9.586119550436962e-05,
      "loss": 0.8452,
      "step": 149860
    },
    {
      "epoch": 2.63119085658105,
      "grad_norm": 0.052570623509831355,
      "learning_rate": 9.585006692684675e-05,
      "loss": 0.8419,
      "step": 149870
    },
    {
      "epoch": 2.631366421461051,
      "grad_norm": 0.05201961260347382,
      "learning_rate": 9.583893841457452e-05,
      "loss": 0.835,
      "step": 149880
    },
    {
      "epoch": 2.6315419863410523,
      "grad_norm": 0.05518315917118054,
      "learning_rate": 9.582780996769392e-05,
      "loss": 0.8471,
      "step": 149890
    },
    {
      "epoch": 2.6317175512210538,
      "grad_norm": 0.04397449845489029,
      "learning_rate": 9.581668158634598e-05,
      "loss": 0.8444,
      "step": 149900
    },
    {
      "epoch": 2.6318931161010553,
      "grad_norm": 0.05877241156014018,
      "learning_rate": 9.580555327067168e-05,
      "loss": 0.8476,
      "step": 149910
    },
    {
      "epoch": 2.6320686809810567,
      "grad_norm": 0.08117107516217152,
      "learning_rate": 9.579442502081195e-05,
      "loss": 0.8437,
      "step": 149920
    },
    {
      "epoch": 2.6322442458610578,
      "grad_norm": 0.03978691457218288,
      "learning_rate": 9.578329683690792e-05,
      "loss": 0.8484,
      "step": 149930
    },
    {
      "epoch": 2.6324198107410592,
      "grad_norm": 0.055799186216497425,
      "learning_rate": 9.577216871910053e-05,
      "loss": 0.8409,
      "step": 149940
    },
    {
      "epoch": 2.6325953756210607,
      "grad_norm": 0.04217624050796169,
      "learning_rate": 9.57610406675307e-05,
      "loss": 0.8425,
      "step": 149950
    },
    {
      "epoch": 2.632770940501062,
      "grad_norm": 0.08480092748997428,
      "learning_rate": 9.574991268233954e-05,
      "loss": 0.8484,
      "step": 149960
    },
    {
      "epoch": 2.6329465053810637,
      "grad_norm": 0.07004760676745236,
      "learning_rate": 9.573878476366797e-05,
      "loss": 0.8454,
      "step": 149970
    },
    {
      "epoch": 2.6331220702610647,
      "grad_norm": 0.05245658945745399,
      "learning_rate": 9.572765691165703e-05,
      "loss": 0.8483,
      "step": 149980
    },
    {
      "epoch": 2.6332976351410666,
      "grad_norm": 0.05734165641079432,
      "learning_rate": 9.571652912644766e-05,
      "loss": 0.8464,
      "step": 149990
    },
    {
      "epoch": 2.6334732000210677,
      "grad_norm": 0.08623257968528705,
      "learning_rate": 9.57054014081809e-05,
      "loss": 0.8393,
      "step": 150000
    },
    {
      "epoch": 2.633648764901069,
      "grad_norm": 0.05172833227882661,
      "learning_rate": 9.56942737569977e-05,
      "loss": 0.8419,
      "step": 150010
    },
    {
      "epoch": 2.6338243297810706,
      "grad_norm": 0.059653317620287326,
      "learning_rate": 9.568314617303904e-05,
      "loss": 0.8407,
      "step": 150020
    },
    {
      "epoch": 2.633999894661072,
      "grad_norm": 0.04823312227221882,
      "learning_rate": 9.567201865644598e-05,
      "loss": 0.8485,
      "step": 150030
    },
    {
      "epoch": 2.6341754595410736,
      "grad_norm": 0.05560223177884161,
      "learning_rate": 9.566089120735947e-05,
      "loss": 0.8412,
      "step": 150040
    },
    {
      "epoch": 2.6343510244210746,
      "grad_norm": 0.04882396167813291,
      "learning_rate": 9.564976382592046e-05,
      "loss": 0.8418,
      "step": 150050
    },
    {
      "epoch": 2.634526589301076,
      "grad_norm": 0.05273956610736683,
      "learning_rate": 9.563863651227e-05,
      "loss": 0.8475,
      "step": 150060
    },
    {
      "epoch": 2.6347021541810776,
      "grad_norm": 0.06200737151092648,
      "learning_rate": 9.562750926654903e-05,
      "loss": 0.8451,
      "step": 150070
    },
    {
      "epoch": 2.634877719061079,
      "grad_norm": 0.04604065798280388,
      "learning_rate": 9.561638208889855e-05,
      "loss": 0.8396,
      "step": 150080
    },
    {
      "epoch": 2.6350532839410805,
      "grad_norm": 0.04840571911142227,
      "learning_rate": 9.56052549794595e-05,
      "loss": 0.8475,
      "step": 150090
    },
    {
      "epoch": 2.635228848821082,
      "grad_norm": 0.058463661838908755,
      "learning_rate": 9.559412793837296e-05,
      "loss": 0.8422,
      "step": 150100
    },
    {
      "epoch": 2.6354044137010835,
      "grad_norm": 0.04698610385035131,
      "learning_rate": 9.558300096577986e-05,
      "loss": 0.8426,
      "step": 150110
    },
    {
      "epoch": 2.6355799785810845,
      "grad_norm": 0.05143380089026764,
      "learning_rate": 9.557187406182111e-05,
      "loss": 0.8491,
      "step": 150120
    },
    {
      "epoch": 2.635755543461086,
      "grad_norm": 0.044676957125037234,
      "learning_rate": 9.556074722663781e-05,
      "loss": 0.8403,
      "step": 150130
    },
    {
      "epoch": 2.6359311083410875,
      "grad_norm": 0.04627369274367202,
      "learning_rate": 9.55496204603709e-05,
      "loss": 0.8456,
      "step": 150140
    },
    {
      "epoch": 2.636106673221089,
      "grad_norm": 0.07031860346117061,
      "learning_rate": 9.55384937631613e-05,
      "loss": 0.8477,
      "step": 150150
    },
    {
      "epoch": 2.6362822381010904,
      "grad_norm": 0.0754247601965701,
      "learning_rate": 9.552736713515005e-05,
      "loss": 0.8328,
      "step": 150160
    },
    {
      "epoch": 2.6364578029810914,
      "grad_norm": 0.0572978431648883,
      "learning_rate": 9.551624057647812e-05,
      "loss": 0.8443,
      "step": 150170
    },
    {
      "epoch": 2.636633367861093,
      "grad_norm": 0.05646921307797406,
      "learning_rate": 9.550511408728645e-05,
      "loss": 0.8376,
      "step": 150180
    },
    {
      "epoch": 2.6368089327410944,
      "grad_norm": 0.058780056165529324,
      "learning_rate": 9.5493987667716e-05,
      "loss": 0.8456,
      "step": 150190
    },
    {
      "epoch": 2.636984497621096,
      "grad_norm": 0.04702439103919158,
      "learning_rate": 9.548286131790784e-05,
      "loss": 0.8413,
      "step": 150200
    },
    {
      "epoch": 2.6371600625010974,
      "grad_norm": 0.05230596264548972,
      "learning_rate": 9.547173503800288e-05,
      "loss": 0.8401,
      "step": 150210
    },
    {
      "epoch": 2.637335627381099,
      "grad_norm": 0.05964142926476374,
      "learning_rate": 9.546060882814207e-05,
      "loss": 0.8423,
      "step": 150220
    },
    {
      "epoch": 2.6375111922611003,
      "grad_norm": 0.05303576302906718,
      "learning_rate": 9.544948268846643e-05,
      "loss": 0.8453,
      "step": 150230
    },
    {
      "epoch": 2.6376867571411013,
      "grad_norm": 0.05713849963915812,
      "learning_rate": 9.543835661911692e-05,
      "loss": 0.8414,
      "step": 150240
    },
    {
      "epoch": 2.637862322021103,
      "grad_norm": 0.0508317730430756,
      "learning_rate": 9.542723062023449e-05,
      "loss": 0.8462,
      "step": 150250
    },
    {
      "epoch": 2.6380378869011043,
      "grad_norm": 0.06132367381294787,
      "learning_rate": 9.541610469196007e-05,
      "loss": 0.8437,
      "step": 150260
    },
    {
      "epoch": 2.638213451781106,
      "grad_norm": 0.0572479006303145,
      "learning_rate": 9.540497883443471e-05,
      "loss": 0.8458,
      "step": 150270
    },
    {
      "epoch": 2.6383890166611073,
      "grad_norm": 0.04298453043445209,
      "learning_rate": 9.539385304779938e-05,
      "loss": 0.8406,
      "step": 150280
    },
    {
      "epoch": 2.6385645815411083,
      "grad_norm": 0.05035699886809428,
      "learning_rate": 9.538272733219492e-05,
      "loss": 0.853,
      "step": 150290
    },
    {
      "epoch": 2.6387401464211098,
      "grad_norm": 0.05894702343350641,
      "learning_rate": 9.537160168776243e-05,
      "loss": 0.8453,
      "step": 150300
    },
    {
      "epoch": 2.6389157113011112,
      "grad_norm": 0.05512447883790953,
      "learning_rate": 9.536047611464284e-05,
      "loss": 0.8447,
      "step": 150310
    },
    {
      "epoch": 2.6390912761811127,
      "grad_norm": 0.04538804017774835,
      "learning_rate": 9.53493506129771e-05,
      "loss": 0.8386,
      "step": 150320
    },
    {
      "epoch": 2.639266841061114,
      "grad_norm": 0.04689891449628058,
      "learning_rate": 9.533822518290612e-05,
      "loss": 0.8432,
      "step": 150330
    },
    {
      "epoch": 2.6394424059411157,
      "grad_norm": 0.038317179834365604,
      "learning_rate": 9.532709982457095e-05,
      "loss": 0.8478,
      "step": 150340
    },
    {
      "epoch": 2.639617970821117,
      "grad_norm": 0.050098136592777696,
      "learning_rate": 9.531597453811253e-05,
      "loss": 0.8519,
      "step": 150350
    },
    {
      "epoch": 2.639793535701118,
      "grad_norm": 0.04697156669859706,
      "learning_rate": 9.530484932367174e-05,
      "loss": 0.8431,
      "step": 150360
    },
    {
      "epoch": 2.6399691005811197,
      "grad_norm": 0.05872813814427585,
      "learning_rate": 9.529372418138964e-05,
      "loss": 0.8446,
      "step": 150370
    },
    {
      "epoch": 2.640144665461121,
      "grad_norm": 0.054238965058305916,
      "learning_rate": 9.528259911140717e-05,
      "loss": 0.845,
      "step": 150380
    },
    {
      "epoch": 2.6403202303411226,
      "grad_norm": 0.050426372711101536,
      "learning_rate": 9.527147411386522e-05,
      "loss": 0.8472,
      "step": 150390
    },
    {
      "epoch": 2.640495795221124,
      "grad_norm": 0.06484215891855358,
      "learning_rate": 9.526034918890482e-05,
      "loss": 0.8316,
      "step": 150400
    },
    {
      "epoch": 2.640671360101125,
      "grad_norm": 0.051327459712313984,
      "learning_rate": 9.52492243366669e-05,
      "loss": 0.8439,
      "step": 150410
    },
    {
      "epoch": 2.640846924981127,
      "grad_norm": 0.061182362156691684,
      "learning_rate": 9.523809955729241e-05,
      "loss": 0.8427,
      "step": 150420
    },
    {
      "epoch": 2.641022489861128,
      "grad_norm": 0.05237457871344223,
      "learning_rate": 9.522697485092226e-05,
      "loss": 0.8454,
      "step": 150430
    },
    {
      "epoch": 2.6411980547411296,
      "grad_norm": 0.07222364206760126,
      "learning_rate": 9.521585021769748e-05,
      "loss": 0.8387,
      "step": 150440
    },
    {
      "epoch": 2.641373619621131,
      "grad_norm": 0.07468505434274399,
      "learning_rate": 9.520472565775903e-05,
      "loss": 0.8441,
      "step": 150450
    },
    {
      "epoch": 2.6415491845011325,
      "grad_norm": 0.05546983227222983,
      "learning_rate": 9.519360117124772e-05,
      "loss": 0.8435,
      "step": 150460
    },
    {
      "epoch": 2.641724749381134,
      "grad_norm": 0.06564430056741517,
      "learning_rate": 9.518247675830465e-05,
      "loss": 0.8356,
      "step": 150470
    },
    {
      "epoch": 2.641900314261135,
      "grad_norm": 0.06153798422847794,
      "learning_rate": 9.517135241907072e-05,
      "loss": 0.8334,
      "step": 150480
    },
    {
      "epoch": 2.6420758791411365,
      "grad_norm": 0.05261457958447237,
      "learning_rate": 9.516022815368686e-05,
      "loss": 0.8518,
      "step": 150490
    },
    {
      "epoch": 2.642251444021138,
      "grad_norm": 0.040705986598762164,
      "learning_rate": 9.514910396229402e-05,
      "loss": 0.843,
      "step": 150500
    },
    {
      "epoch": 2.6424270089011395,
      "grad_norm": 0.051413937153309,
      "learning_rate": 9.513797984503318e-05,
      "loss": 0.8474,
      "step": 150510
    },
    {
      "epoch": 2.642602573781141,
      "grad_norm": 0.04755823817389561,
      "learning_rate": 9.512685580204525e-05,
      "loss": 0.8351,
      "step": 150520
    },
    {
      "epoch": 2.642778138661142,
      "grad_norm": 0.05195692816767245,
      "learning_rate": 9.511573183347113e-05,
      "loss": 0.8458,
      "step": 150530
    },
    {
      "epoch": 2.642953703541144,
      "grad_norm": 0.048141615807225355,
      "learning_rate": 9.510460793945188e-05,
      "loss": 0.8522,
      "step": 150540
    },
    {
      "epoch": 2.643129268421145,
      "grad_norm": 0.04870337479498233,
      "learning_rate": 9.509348412012836e-05,
      "loss": 0.8388,
      "step": 150550
    },
    {
      "epoch": 2.6433048333011464,
      "grad_norm": 0.0465642536417802,
      "learning_rate": 9.508236037564151e-05,
      "loss": 0.8422,
      "step": 150560
    },
    {
      "epoch": 2.643480398181148,
      "grad_norm": 0.049972029226784564,
      "learning_rate": 9.507123670613231e-05,
      "loss": 0.8477,
      "step": 150570
    },
    {
      "epoch": 2.6436559630611494,
      "grad_norm": 0.06405183027133467,
      "learning_rate": 9.50601131117417e-05,
      "loss": 0.8443,
      "step": 150580
    },
    {
      "epoch": 2.643831527941151,
      "grad_norm": 0.05369390083517443,
      "learning_rate": 9.504898959261056e-05,
      "loss": 0.852,
      "step": 150590
    },
    {
      "epoch": 2.644007092821152,
      "grad_norm": 0.06278608932150226,
      "learning_rate": 9.503786614887982e-05,
      "loss": 0.8311,
      "step": 150600
    },
    {
      "epoch": 2.6441826577011533,
      "grad_norm": 0.06342700372586185,
      "learning_rate": 9.502674278069053e-05,
      "loss": 0.8492,
      "step": 150610
    },
    {
      "epoch": 2.644358222581155,
      "grad_norm": 0.054480674683152396,
      "learning_rate": 9.501561948818357e-05,
      "loss": 0.8395,
      "step": 150620
    },
    {
      "epoch": 2.6445337874611563,
      "grad_norm": 0.05606492596299416,
      "learning_rate": 9.500449627149979e-05,
      "loss": 0.84,
      "step": 150630
    },
    {
      "epoch": 2.644709352341158,
      "grad_norm": 0.05627660239459318,
      "learning_rate": 9.499337313078021e-05,
      "loss": 0.8542,
      "step": 150640
    },
    {
      "epoch": 2.644884917221159,
      "grad_norm": 0.04950510795614886,
      "learning_rate": 9.498225006616577e-05,
      "loss": 0.8424,
      "step": 150650
    },
    {
      "epoch": 2.6450604821011607,
      "grad_norm": 0.044928179431561376,
      "learning_rate": 9.497112707779738e-05,
      "loss": 0.8485,
      "step": 150660
    },
    {
      "epoch": 2.6452360469811618,
      "grad_norm": 0.05786539789038271,
      "learning_rate": 9.496000416581593e-05,
      "loss": 0.8405,
      "step": 150670
    },
    {
      "epoch": 2.6454116118611632,
      "grad_norm": 0.04054334536572059,
      "learning_rate": 9.49488813303624e-05,
      "loss": 0.8382,
      "step": 150680
    },
    {
      "epoch": 2.6455871767411647,
      "grad_norm": 0.06368528938182326,
      "learning_rate": 9.493775857157771e-05,
      "loss": 0.8465,
      "step": 150690
    },
    {
      "epoch": 2.645762741621166,
      "grad_norm": 0.0678491071923747,
      "learning_rate": 9.492663588960274e-05,
      "loss": 0.8447,
      "step": 150700
    },
    {
      "epoch": 2.6459383065011677,
      "grad_norm": 0.05476480735595069,
      "learning_rate": 9.49155132845785e-05,
      "loss": 0.8401,
      "step": 150710
    },
    {
      "epoch": 2.6461138713811687,
      "grad_norm": 0.05206330859401549,
      "learning_rate": 9.490439075664587e-05,
      "loss": 0.8424,
      "step": 150720
    },
    {
      "epoch": 2.64628943626117,
      "grad_norm": 0.05355659043853604,
      "learning_rate": 9.489326830594575e-05,
      "loss": 0.8495,
      "step": 150730
    },
    {
      "epoch": 2.6464650011411717,
      "grad_norm": 0.06367552687603881,
      "learning_rate": 9.488214593261912e-05,
      "loss": 0.8473,
      "step": 150740
    },
    {
      "epoch": 2.646640566021173,
      "grad_norm": 0.0554972288038264,
      "learning_rate": 9.487102363680687e-05,
      "loss": 0.8495,
      "step": 150750
    },
    {
      "epoch": 2.6468161309011746,
      "grad_norm": 0.05469454165470387,
      "learning_rate": 9.485990141864995e-05,
      "loss": 0.8458,
      "step": 150760
    },
    {
      "epoch": 2.6469916957811757,
      "grad_norm": 0.053677430698421714,
      "learning_rate": 9.484877927828917e-05,
      "loss": 0.8488,
      "step": 150770
    },
    {
      "epoch": 2.6471672606611776,
      "grad_norm": 0.04847487908525251,
      "learning_rate": 9.483765721586563e-05,
      "loss": 0.8479,
      "step": 150780
    },
    {
      "epoch": 2.6473428255411786,
      "grad_norm": 0.05069205204478219,
      "learning_rate": 9.482653523152013e-05,
      "loss": 0.8421,
      "step": 150790
    },
    {
      "epoch": 2.64751839042118,
      "grad_norm": 0.05410371853226987,
      "learning_rate": 9.481541332539356e-05,
      "loss": 0.8475,
      "step": 150800
    },
    {
      "epoch": 2.6476939553011816,
      "grad_norm": 0.049412515720492545,
      "learning_rate": 9.480429149762694e-05,
      "loss": 0.8496,
      "step": 150810
    },
    {
      "epoch": 2.647869520181183,
      "grad_norm": 0.04387684302463949,
      "learning_rate": 9.47931697483611e-05,
      "loss": 0.8373,
      "step": 150820
    },
    {
      "epoch": 2.6480450850611845,
      "grad_norm": 0.0707093570397422,
      "learning_rate": 9.478204807773704e-05,
      "loss": 0.8415,
      "step": 150830
    },
    {
      "epoch": 2.6482206499411856,
      "grad_norm": 0.06002975443612054,
      "learning_rate": 9.477092648589557e-05,
      "loss": 0.851,
      "step": 150840
    },
    {
      "epoch": 2.648396214821187,
      "grad_norm": 0.07288038346538545,
      "learning_rate": 9.47598049729777e-05,
      "loss": 0.8427,
      "step": 150850
    },
    {
      "epoch": 2.6485717797011885,
      "grad_norm": 0.057554526768286506,
      "learning_rate": 9.474868353912426e-05,
      "loss": 0.8444,
      "step": 150860
    },
    {
      "epoch": 2.64874734458119,
      "grad_norm": 0.05276667004389696,
      "learning_rate": 9.473756218447617e-05,
      "loss": 0.8484,
      "step": 150870
    },
    {
      "epoch": 2.6489229094611915,
      "grad_norm": 0.04314839841287447,
      "learning_rate": 9.472644090917443e-05,
      "loss": 0.8434,
      "step": 150880
    },
    {
      "epoch": 2.649098474341193,
      "grad_norm": 0.04618779202788762,
      "learning_rate": 9.471531971335988e-05,
      "loss": 0.8441,
      "step": 150890
    },
    {
      "epoch": 2.6492740392211944,
      "grad_norm": 0.061340445781124284,
      "learning_rate": 9.47041985971734e-05,
      "loss": 0.8456,
      "step": 150900
    },
    {
      "epoch": 2.6494496041011955,
      "grad_norm": 0.05526981902758779,
      "learning_rate": 9.469307756075595e-05,
      "loss": 0.8432,
      "step": 150910
    },
    {
      "epoch": 2.649625168981197,
      "grad_norm": 0.07571979196422328,
      "learning_rate": 9.468195660424842e-05,
      "loss": 0.8508,
      "step": 150920
    },
    {
      "epoch": 2.6498007338611984,
      "grad_norm": 0.07213875787355617,
      "learning_rate": 9.467083572779171e-05,
      "loss": 0.8461,
      "step": 150930
    },
    {
      "epoch": 2.6499762987412,
      "grad_norm": 0.07596694955474238,
      "learning_rate": 9.465971493152669e-05,
      "loss": 0.8462,
      "step": 150940
    },
    {
      "epoch": 2.6501518636212014,
      "grad_norm": 0.08149923190433367,
      "learning_rate": 9.464859421559437e-05,
      "loss": 0.843,
      "step": 150950
    },
    {
      "epoch": 2.6503274285012024,
      "grad_norm": 0.0769680022576516,
      "learning_rate": 9.463747358013554e-05,
      "loss": 0.8442,
      "step": 150960
    },
    {
      "epoch": 2.650502993381204,
      "grad_norm": 0.06621862852788316,
      "learning_rate": 9.462635302529109e-05,
      "loss": 0.8421,
      "step": 150970
    },
    {
      "epoch": 2.6506785582612054,
      "grad_norm": 0.05845632533427299,
      "learning_rate": 9.461523255120201e-05,
      "loss": 0.8451,
      "step": 150980
    },
    {
      "epoch": 2.650854123141207,
      "grad_norm": 0.05820433824000859,
      "learning_rate": 9.460411215800919e-05,
      "loss": 0.8422,
      "step": 150990
    },
    {
      "epoch": 2.6510296880212083,
      "grad_norm": 0.04097402836856973,
      "learning_rate": 9.459299184585347e-05,
      "loss": 0.8417,
      "step": 151000
    },
    {
      "epoch": 2.65120525290121,
      "grad_norm": 0.04742844584043897,
      "learning_rate": 9.458187161487576e-05,
      "loss": 0.8425,
      "step": 151010
    },
    {
      "epoch": 2.6513808177812113,
      "grad_norm": 0.08824869893495693,
      "learning_rate": 9.457075146521697e-05,
      "loss": 0.8382,
      "step": 151020
    },
    {
      "epoch": 2.6515563826612123,
      "grad_norm": 0.05574136162049148,
      "learning_rate": 9.4559631397018e-05,
      "loss": 0.8423,
      "step": 151030
    },
    {
      "epoch": 2.6517319475412138,
      "grad_norm": 0.0946649835908488,
      "learning_rate": 9.45485114104197e-05,
      "loss": 0.8409,
      "step": 151040
    },
    {
      "epoch": 2.6519075124212153,
      "grad_norm": 0.0554722735733332,
      "learning_rate": 9.453739150556303e-05,
      "loss": 0.8423,
      "step": 151050
    },
    {
      "epoch": 2.6520830773012167,
      "grad_norm": 0.06274020004140252,
      "learning_rate": 9.452627168258886e-05,
      "loss": 0.8491,
      "step": 151060
    },
    {
      "epoch": 2.652258642181218,
      "grad_norm": 0.05094417001924733,
      "learning_rate": 9.451515194163802e-05,
      "loss": 0.8405,
      "step": 151070
    },
    {
      "epoch": 2.6524342070612192,
      "grad_norm": 0.057749377664201865,
      "learning_rate": 9.45040322828515e-05,
      "loss": 0.8451,
      "step": 151080
    },
    {
      "epoch": 2.6526097719412207,
      "grad_norm": 0.05148488744647607,
      "learning_rate": 9.449291270637012e-05,
      "loss": 0.8418,
      "step": 151090
    },
    {
      "epoch": 2.652785336821222,
      "grad_norm": 0.08033013339787454,
      "learning_rate": 9.448179321233478e-05,
      "loss": 0.8359,
      "step": 151100
    },
    {
      "epoch": 2.6529609017012237,
      "grad_norm": 0.05269473334408509,
      "learning_rate": 9.447067380088634e-05,
      "loss": 0.8486,
      "step": 151110
    },
    {
      "epoch": 2.653136466581225,
      "grad_norm": 0.05246529695522727,
      "learning_rate": 9.445955447216577e-05,
      "loss": 0.8405,
      "step": 151120
    },
    {
      "epoch": 2.6533120314612266,
      "grad_norm": 0.057093226020914406,
      "learning_rate": 9.444843522631388e-05,
      "loss": 0.8479,
      "step": 151130
    },
    {
      "epoch": 2.653487596341228,
      "grad_norm": 0.04953418434989364,
      "learning_rate": 9.443731606347153e-05,
      "loss": 0.8431,
      "step": 151140
    },
    {
      "epoch": 2.653663161221229,
      "grad_norm": 0.04191530354351525,
      "learning_rate": 9.442619698377966e-05,
      "loss": 0.8412,
      "step": 151150
    },
    {
      "epoch": 2.6538387261012306,
      "grad_norm": 0.06184632968300676,
      "learning_rate": 9.441507798737918e-05,
      "loss": 0.8428,
      "step": 151160
    },
    {
      "epoch": 2.654014290981232,
      "grad_norm": 0.04110614335997871,
      "learning_rate": 9.44039590744109e-05,
      "loss": 0.8452,
      "step": 151170
    },
    {
      "epoch": 2.6541898558612336,
      "grad_norm": 0.041830347520999976,
      "learning_rate": 9.43928402450157e-05,
      "loss": 0.8474,
      "step": 151180
    },
    {
      "epoch": 2.654365420741235,
      "grad_norm": 0.05155225076603792,
      "learning_rate": 9.43817214993345e-05,
      "loss": 0.8537,
      "step": 151190
    },
    {
      "epoch": 2.654540985621236,
      "grad_norm": 0.061040070144417424,
      "learning_rate": 9.437060283750816e-05,
      "loss": 0.8406,
      "step": 151200
    },
    {
      "epoch": 2.654716550501238,
      "grad_norm": 0.057920294088537674,
      "learning_rate": 9.43594842596775e-05,
      "loss": 0.8499,
      "step": 151210
    },
    {
      "epoch": 2.654892115381239,
      "grad_norm": 0.0531599741658295,
      "learning_rate": 9.43483657659835e-05,
      "loss": 0.8456,
      "step": 151220
    },
    {
      "epoch": 2.6550676802612405,
      "grad_norm": 0.058628426846472556,
      "learning_rate": 9.433724735656698e-05,
      "loss": 0.8469,
      "step": 151230
    },
    {
      "epoch": 2.655243245141242,
      "grad_norm": 0.057773137961478004,
      "learning_rate": 9.43261290315688e-05,
      "loss": 0.8545,
      "step": 151240
    },
    {
      "epoch": 2.6554188100212435,
      "grad_norm": 0.05383486865127076,
      "learning_rate": 9.431501079112985e-05,
      "loss": 0.8429,
      "step": 151250
    },
    {
      "epoch": 2.655594374901245,
      "grad_norm": 0.07155345968666717,
      "learning_rate": 9.430389263539101e-05,
      "loss": 0.8471,
      "step": 151260
    },
    {
      "epoch": 2.655769939781246,
      "grad_norm": 0.06401494276052641,
      "learning_rate": 9.429277456449312e-05,
      "loss": 0.847,
      "step": 151270
    },
    {
      "epoch": 2.6559455046612475,
      "grad_norm": 0.06836103238995506,
      "learning_rate": 9.428165657857703e-05,
      "loss": 0.8379,
      "step": 151280
    },
    {
      "epoch": 2.656121069541249,
      "grad_norm": 0.050722043682443686,
      "learning_rate": 9.42705386777837e-05,
      "loss": 0.8414,
      "step": 151290
    },
    {
      "epoch": 2.6562966344212504,
      "grad_norm": 0.046982396014335154,
      "learning_rate": 9.425942086225391e-05,
      "loss": 0.8469,
      "step": 151300
    },
    {
      "epoch": 2.656472199301252,
      "grad_norm": 0.06211801086673411,
      "learning_rate": 9.42483031321285e-05,
      "loss": 0.8432,
      "step": 151310
    },
    {
      "epoch": 2.656647764181253,
      "grad_norm": 0.043922380497226625,
      "learning_rate": 9.423718548754844e-05,
      "loss": 0.851,
      "step": 151320
    },
    {
      "epoch": 2.656823329061255,
      "grad_norm": 0.0651496574472913,
      "learning_rate": 9.422606792865454e-05,
      "loss": 0.853,
      "step": 151330
    },
    {
      "epoch": 2.656998893941256,
      "grad_norm": 0.048014259969741284,
      "learning_rate": 9.421495045558764e-05,
      "loss": 0.847,
      "step": 151340
    },
    {
      "epoch": 2.6571744588212574,
      "grad_norm": 0.048519125269346604,
      "learning_rate": 9.42038330684886e-05,
      "loss": 0.8406,
      "step": 151350
    },
    {
      "epoch": 2.657350023701259,
      "grad_norm": 0.06595264218138389,
      "learning_rate": 9.419271576749833e-05,
      "loss": 0.8469,
      "step": 151360
    },
    {
      "epoch": 2.6575255885812603,
      "grad_norm": 0.0662592467128317,
      "learning_rate": 9.418159855275765e-05,
      "loss": 0.8396,
      "step": 151370
    },
    {
      "epoch": 2.657701153461262,
      "grad_norm": 0.05470432833496833,
      "learning_rate": 9.417048142440738e-05,
      "loss": 0.8431,
      "step": 151380
    },
    {
      "epoch": 2.657876718341263,
      "grad_norm": 0.07262935633043825,
      "learning_rate": 9.415936438258845e-05,
      "loss": 0.8432,
      "step": 151390
    },
    {
      "epoch": 2.6580522832212643,
      "grad_norm": 0.043635201799264986,
      "learning_rate": 9.41482474274417e-05,
      "loss": 0.8497,
      "step": 151400
    },
    {
      "epoch": 2.6582278481012658,
      "grad_norm": 0.06110140088084929,
      "learning_rate": 9.413713055910794e-05,
      "loss": 0.8323,
      "step": 151410
    },
    {
      "epoch": 2.6584034129812673,
      "grad_norm": 0.04326757093664401,
      "learning_rate": 9.412601377772808e-05,
      "loss": 0.8419,
      "step": 151420
    },
    {
      "epoch": 2.6585789778612687,
      "grad_norm": 0.060329512671560934,
      "learning_rate": 9.411489708344294e-05,
      "loss": 0.841,
      "step": 151430
    },
    {
      "epoch": 2.6587545427412698,
      "grad_norm": 0.07697380559197292,
      "learning_rate": 9.410378047639337e-05,
      "loss": 0.8399,
      "step": 151440
    },
    {
      "epoch": 2.6589301076212717,
      "grad_norm": 0.04562904078584801,
      "learning_rate": 9.409266395672019e-05,
      "loss": 0.8437,
      "step": 151450
    },
    {
      "epoch": 2.6591056725012727,
      "grad_norm": 0.059182396275992324,
      "learning_rate": 9.408154752456431e-05,
      "loss": 0.8433,
      "step": 151460
    },
    {
      "epoch": 2.659281237381274,
      "grad_norm": 0.06586524458301399,
      "learning_rate": 9.407043118006657e-05,
      "loss": 0.8468,
      "step": 151470
    },
    {
      "epoch": 2.6594568022612757,
      "grad_norm": 0.05802966359183291,
      "learning_rate": 9.405931492336774e-05,
      "loss": 0.8421,
      "step": 151480
    },
    {
      "epoch": 2.659632367141277,
      "grad_norm": 0.053052986961727375,
      "learning_rate": 9.404819875460876e-05,
      "loss": 0.8426,
      "step": 151490
    },
    {
      "epoch": 2.6598079320212786,
      "grad_norm": 0.0674717105196779,
      "learning_rate": 9.403708267393043e-05,
      "loss": 0.8467,
      "step": 151500
    },
    {
      "epoch": 2.6599834969012797,
      "grad_norm": 0.06253556499112238,
      "learning_rate": 9.402596668147359e-05,
      "loss": 0.8413,
      "step": 151510
    },
    {
      "epoch": 2.660159061781281,
      "grad_norm": 0.06045174406338209,
      "learning_rate": 9.401485077737909e-05,
      "loss": 0.8448,
      "step": 151520
    },
    {
      "epoch": 2.6603346266612826,
      "grad_norm": 0.07009732419609538,
      "learning_rate": 9.400373496178779e-05,
      "loss": 0.8437,
      "step": 151530
    },
    {
      "epoch": 2.660510191541284,
      "grad_norm": 0.06112488252201541,
      "learning_rate": 9.399261923484048e-05,
      "loss": 0.8439,
      "step": 151540
    },
    {
      "epoch": 2.6606857564212856,
      "grad_norm": 0.053243130790166356,
      "learning_rate": 9.398150359667801e-05,
      "loss": 0.8492,
      "step": 151550
    },
    {
      "epoch": 2.660861321301287,
      "grad_norm": 0.06897967455432359,
      "learning_rate": 9.397038804744126e-05,
      "loss": 0.8436,
      "step": 151560
    },
    {
      "epoch": 2.6610368861812885,
      "grad_norm": 0.07174911565265259,
      "learning_rate": 9.395927258727105e-05,
      "loss": 0.8442,
      "step": 151570
    },
    {
      "epoch": 2.6612124510612896,
      "grad_norm": 0.058736910478789525,
      "learning_rate": 9.394815721630818e-05,
      "loss": 0.8359,
      "step": 151580
    },
    {
      "epoch": 2.661388015941291,
      "grad_norm": 0.05316290620424622,
      "learning_rate": 9.393704193469354e-05,
      "loss": 0.8525,
      "step": 151590
    },
    {
      "epoch": 2.6615635808212925,
      "grad_norm": 0.04695926389173282,
      "learning_rate": 9.392592674256793e-05,
      "loss": 0.8462,
      "step": 151600
    },
    {
      "epoch": 2.661739145701294,
      "grad_norm": 0.04239226957676117,
      "learning_rate": 9.391481164007218e-05,
      "loss": 0.8372,
      "step": 151610
    },
    {
      "epoch": 2.6619147105812955,
      "grad_norm": 0.08845960197999383,
      "learning_rate": 9.390369662734709e-05,
      "loss": 0.843,
      "step": 151620
    },
    {
      "epoch": 2.6620902754612965,
      "grad_norm": 0.05008142472934135,
      "learning_rate": 9.389258170453357e-05,
      "loss": 0.8434,
      "step": 151630
    },
    {
      "epoch": 2.662265840341298,
      "grad_norm": 0.0546049519020161,
      "learning_rate": 9.38814668717724e-05,
      "loss": 0.8461,
      "step": 151640
    },
    {
      "epoch": 2.6624414052212995,
      "grad_norm": 0.06679228146694062,
      "learning_rate": 9.387035212920436e-05,
      "loss": 0.8451,
      "step": 151650
    },
    {
      "epoch": 2.662616970101301,
      "grad_norm": 0.059704940822562245,
      "learning_rate": 9.385923747697035e-05,
      "loss": 0.8479,
      "step": 151660
    },
    {
      "epoch": 2.6627925349813024,
      "grad_norm": 0.048580153425783174,
      "learning_rate": 9.384812291521119e-05,
      "loss": 0.8333,
      "step": 151670
    },
    {
      "epoch": 2.662968099861304,
      "grad_norm": 0.04601129438607629,
      "learning_rate": 9.383700844406769e-05,
      "loss": 0.851,
      "step": 151680
    },
    {
      "epoch": 2.6631436647413054,
      "grad_norm": 0.05491009124172223,
      "learning_rate": 9.382589406368062e-05,
      "loss": 0.8489,
      "step": 151690
    },
    {
      "epoch": 2.6633192296213064,
      "grad_norm": 0.05515910480758032,
      "learning_rate": 9.381477977419089e-05,
      "loss": 0.849,
      "step": 151700
    },
    {
      "epoch": 2.663494794501308,
      "grad_norm": 0.051289890581998336,
      "learning_rate": 9.380366557573927e-05,
      "loss": 0.84,
      "step": 151710
    },
    {
      "epoch": 2.6636703593813094,
      "grad_norm": 0.05430308707308694,
      "learning_rate": 9.379255146846654e-05,
      "loss": 0.8542,
      "step": 151720
    },
    {
      "epoch": 2.663845924261311,
      "grad_norm": 0.07004799457353754,
      "learning_rate": 9.37814374525136e-05,
      "loss": 0.8441,
      "step": 151730
    },
    {
      "epoch": 2.6640214891413123,
      "grad_norm": 0.05440321153917079,
      "learning_rate": 9.377032352802125e-05,
      "loss": 0.843,
      "step": 151740
    },
    {
      "epoch": 2.6641970540213133,
      "grad_norm": 0.055904892436044344,
      "learning_rate": 9.375920969513025e-05,
      "loss": 0.8403,
      "step": 151750
    },
    {
      "epoch": 2.664372618901315,
      "grad_norm": 0.05904186190022445,
      "learning_rate": 9.374809595398148e-05,
      "loss": 0.8488,
      "step": 151760
    },
    {
      "epoch": 2.6645481837813163,
      "grad_norm": 0.06199559719261382,
      "learning_rate": 9.373698230471572e-05,
      "loss": 0.8463,
      "step": 151770
    },
    {
      "epoch": 2.664723748661318,
      "grad_norm": 0.053340251714137386,
      "learning_rate": 9.372586874747378e-05,
      "loss": 0.8386,
      "step": 151780
    },
    {
      "epoch": 2.6648993135413193,
      "grad_norm": 0.0775580153289824,
      "learning_rate": 9.371475528239644e-05,
      "loss": 0.8444,
      "step": 151790
    },
    {
      "epoch": 2.6650748784213207,
      "grad_norm": 0.07194891174385153,
      "learning_rate": 9.370364190962461e-05,
      "loss": 0.8394,
      "step": 151800
    },
    {
      "epoch": 2.665250443301322,
      "grad_norm": 0.06135079128105748,
      "learning_rate": 9.369252862929902e-05,
      "loss": 0.8438,
      "step": 151810
    },
    {
      "epoch": 2.6654260081813232,
      "grad_norm": 0.04790855297114918,
      "learning_rate": 9.368141544156045e-05,
      "loss": 0.8416,
      "step": 151820
    },
    {
      "epoch": 2.6656015730613247,
      "grad_norm": 0.04914342030497403,
      "learning_rate": 9.367030234654979e-05,
      "loss": 0.8369,
      "step": 151830
    },
    {
      "epoch": 2.665777137941326,
      "grad_norm": 0.060655517535139404,
      "learning_rate": 9.365918934440781e-05,
      "loss": 0.8424,
      "step": 151840
    },
    {
      "epoch": 2.6659527028213277,
      "grad_norm": 0.046011456212257135,
      "learning_rate": 9.364807643527529e-05,
      "loss": 0.8423,
      "step": 151850
    },
    {
      "epoch": 2.666128267701329,
      "grad_norm": 0.05518479107555286,
      "learning_rate": 9.363696361929304e-05,
      "loss": 0.8435,
      "step": 151860
    },
    {
      "epoch": 2.66630383258133,
      "grad_norm": 0.05790758054613443,
      "learning_rate": 9.36258508966019e-05,
      "loss": 0.8375,
      "step": 151870
    },
    {
      "epoch": 2.666479397461332,
      "grad_norm": 0.054673538196099145,
      "learning_rate": 9.361473826734263e-05,
      "loss": 0.8468,
      "step": 151880
    },
    {
      "epoch": 2.666654962341333,
      "grad_norm": 0.06077453022078411,
      "learning_rate": 9.3603625731656e-05,
      "loss": 0.8491,
      "step": 151890
    },
    {
      "epoch": 2.6668305272213346,
      "grad_norm": 0.05617187864035669,
      "learning_rate": 9.359251328968289e-05,
      "loss": 0.8405,
      "step": 151900
    },
    {
      "epoch": 2.667006092101336,
      "grad_norm": 0.07229341166488415,
      "learning_rate": 9.358140094156407e-05,
      "loss": 0.8431,
      "step": 151910
    },
    {
      "epoch": 2.6671816569813376,
      "grad_norm": 0.0750717279055238,
      "learning_rate": 9.357028868744029e-05,
      "loss": 0.8451,
      "step": 151920
    },
    {
      "epoch": 2.667357221861339,
      "grad_norm": 0.06670722389305597,
      "learning_rate": 9.355917652745241e-05,
      "loss": 0.843,
      "step": 151930
    },
    {
      "epoch": 2.66753278674134,
      "grad_norm": 0.059290926293185116,
      "learning_rate": 9.354806446174117e-05,
      "loss": 0.8431,
      "step": 151940
    },
    {
      "epoch": 2.6677083516213416,
      "grad_norm": 0.058769299962455626,
      "learning_rate": 9.353695249044739e-05,
      "loss": 0.8424,
      "step": 151950
    },
    {
      "epoch": 2.667883916501343,
      "grad_norm": 0.04637076245513428,
      "learning_rate": 9.352584061371181e-05,
      "loss": 0.8425,
      "step": 151960
    },
    {
      "epoch": 2.6680594813813445,
      "grad_norm": 0.07310013270795306,
      "learning_rate": 9.351472883167534e-05,
      "loss": 0.8374,
      "step": 151970
    },
    {
      "epoch": 2.668235046261346,
      "grad_norm": 0.05701322261005276,
      "learning_rate": 9.350361714447864e-05,
      "loss": 0.851,
      "step": 151980
    },
    {
      "epoch": 2.668410611141347,
      "grad_norm": 0.08405645229184856,
      "learning_rate": 9.349250555226252e-05,
      "loss": 0.8452,
      "step": 151990
    },
    {
      "epoch": 2.668586176021349,
      "grad_norm": 0.047082007582514865,
      "learning_rate": 9.348139405516784e-05,
      "loss": 0.8417,
      "step": 152000
    },
    {
      "epoch": 2.66876174090135,
      "grad_norm": 0.06321431018641786,
      "learning_rate": 9.347028265333533e-05,
      "loss": 0.8387,
      "step": 152010
    },
    {
      "epoch": 2.6689373057813515,
      "grad_norm": 0.040195656497357464,
      "learning_rate": 9.34591713469058e-05,
      "loss": 0.8472,
      "step": 152020
    },
    {
      "epoch": 2.669112870661353,
      "grad_norm": 0.04249472151596116,
      "learning_rate": 9.344806013601996e-05,
      "loss": 0.8525,
      "step": 152030
    },
    {
      "epoch": 2.6692884355413544,
      "grad_norm": 0.06487958946452857,
      "learning_rate": 9.34369490208187e-05,
      "loss": 0.8427,
      "step": 152040
    },
    {
      "epoch": 2.669464000421356,
      "grad_norm": 0.06126634624242237,
      "learning_rate": 9.342583800144273e-05,
      "loss": 0.8496,
      "step": 152050
    },
    {
      "epoch": 2.669639565301357,
      "grad_norm": 0.0679857249555314,
      "learning_rate": 9.34147270780328e-05,
      "loss": 0.8409,
      "step": 152060
    },
    {
      "epoch": 2.6698151301813584,
      "grad_norm": 0.05596262746697922,
      "learning_rate": 9.340361625072978e-05,
      "loss": 0.8394,
      "step": 152070
    },
    {
      "epoch": 2.66999069506136,
      "grad_norm": 0.05510650338451066,
      "learning_rate": 9.33925055196744e-05,
      "loss": 0.8445,
      "step": 152080
    },
    {
      "epoch": 2.6701662599413614,
      "grad_norm": 0.046744292785754205,
      "learning_rate": 9.33813948850074e-05,
      "loss": 0.8406,
      "step": 152090
    },
    {
      "epoch": 2.670341824821363,
      "grad_norm": 0.06375653077670541,
      "learning_rate": 9.337028434686961e-05,
      "loss": 0.8518,
      "step": 152100
    },
    {
      "epoch": 2.670517389701364,
      "grad_norm": 0.07576922465788263,
      "learning_rate": 9.33591739054018e-05,
      "loss": 0.8388,
      "step": 152110
    },
    {
      "epoch": 2.670692954581366,
      "grad_norm": 0.06424313931645602,
      "learning_rate": 9.334806356074471e-05,
      "loss": 0.8491,
      "step": 152120
    },
    {
      "epoch": 2.670868519461367,
      "grad_norm": 0.049950357614262995,
      "learning_rate": 9.333695331303907e-05,
      "loss": 0.8419,
      "step": 152130
    },
    {
      "epoch": 2.6710440843413683,
      "grad_norm": 0.045742338660962355,
      "learning_rate": 9.332584316242576e-05,
      "loss": 0.8424,
      "step": 152140
    },
    {
      "epoch": 2.67121964922137,
      "grad_norm": 0.05841628490872988,
      "learning_rate": 9.331473310904548e-05,
      "loss": 0.8515,
      "step": 152150
    },
    {
      "epoch": 2.6713952141013713,
      "grad_norm": 0.05566140485016339,
      "learning_rate": 9.330362315303896e-05,
      "loss": 0.8454,
      "step": 152160
    },
    {
      "epoch": 2.6715707789813727,
      "grad_norm": 0.04597940736744022,
      "learning_rate": 9.329251329454704e-05,
      "loss": 0.8408,
      "step": 152170
    },
    {
      "epoch": 2.6717463438613738,
      "grad_norm": 0.053627364603004254,
      "learning_rate": 9.328140353371047e-05,
      "loss": 0.844,
      "step": 152180
    },
    {
      "epoch": 2.6719219087413753,
      "grad_norm": 0.04590096516072428,
      "learning_rate": 9.327029387066999e-05,
      "loss": 0.8469,
      "step": 152190
    },
    {
      "epoch": 2.6720974736213767,
      "grad_norm": 0.05582242311212083,
      "learning_rate": 9.325918430556634e-05,
      "loss": 0.8444,
      "step": 152200
    },
    {
      "epoch": 2.672273038501378,
      "grad_norm": 0.054299658120264514,
      "learning_rate": 9.324807483854034e-05,
      "loss": 0.8484,
      "step": 152210
    },
    {
      "epoch": 2.6724486033813797,
      "grad_norm": 0.058437884344375864,
      "learning_rate": 9.323696546973271e-05,
      "loss": 0.8483,
      "step": 152220
    },
    {
      "epoch": 2.6726241682613807,
      "grad_norm": 0.07101100283848001,
      "learning_rate": 9.322585619928418e-05,
      "loss": 0.8434,
      "step": 152230
    },
    {
      "epoch": 2.6727997331413826,
      "grad_norm": 0.05738182968370937,
      "learning_rate": 9.321474702733559e-05,
      "loss": 0.8389,
      "step": 152240
    },
    {
      "epoch": 2.6729752980213837,
      "grad_norm": 0.045923186677970296,
      "learning_rate": 9.320363795402763e-05,
      "loss": 0.8495,
      "step": 152250
    },
    {
      "epoch": 2.673150862901385,
      "grad_norm": 0.06105469594235696,
      "learning_rate": 9.319252897950104e-05,
      "loss": 0.8379,
      "step": 152260
    },
    {
      "epoch": 2.6733264277813866,
      "grad_norm": 0.05680237358663035,
      "learning_rate": 9.318142010389666e-05,
      "loss": 0.8407,
      "step": 152270
    },
    {
      "epoch": 2.673501992661388,
      "grad_norm": 0.08997906213932236,
      "learning_rate": 9.317031132735515e-05,
      "loss": 0.8449,
      "step": 152280
    },
    {
      "epoch": 2.6736775575413896,
      "grad_norm": 0.049770088237063526,
      "learning_rate": 9.315920265001733e-05,
      "loss": 0.8311,
      "step": 152290
    },
    {
      "epoch": 2.6738531224213906,
      "grad_norm": 0.0770280048078519,
      "learning_rate": 9.314809407202385e-05,
      "loss": 0.8435,
      "step": 152300
    },
    {
      "epoch": 2.674028687301392,
      "grad_norm": 0.04722590106936225,
      "learning_rate": 9.313698559351558e-05,
      "loss": 0.8395,
      "step": 152310
    },
    {
      "epoch": 2.6742042521813936,
      "grad_norm": 0.06355255071593646,
      "learning_rate": 9.312587721463322e-05,
      "loss": 0.8383,
      "step": 152320
    },
    {
      "epoch": 2.674379817061395,
      "grad_norm": 0.05112094531701857,
      "learning_rate": 9.311476893551744e-05,
      "loss": 0.84,
      "step": 152330
    },
    {
      "epoch": 2.6745553819413965,
      "grad_norm": 0.07340276187826353,
      "learning_rate": 9.31036607563091e-05,
      "loss": 0.8453,
      "step": 152340
    },
    {
      "epoch": 2.674730946821398,
      "grad_norm": 0.0525384378728642,
      "learning_rate": 9.309255267714887e-05,
      "loss": 0.8499,
      "step": 152350
    },
    {
      "epoch": 2.6749065117013995,
      "grad_norm": 0.05629619751554896,
      "learning_rate": 9.308144469817755e-05,
      "loss": 0.8416,
      "step": 152360
    },
    {
      "epoch": 2.6750820765814005,
      "grad_norm": 0.05136062711568219,
      "learning_rate": 9.307033681953579e-05,
      "loss": 0.8465,
      "step": 152370
    },
    {
      "epoch": 2.675257641461402,
      "grad_norm": 0.04589431471619701,
      "learning_rate": 9.305922904136442e-05,
      "loss": 0.8521,
      "step": 152380
    },
    {
      "epoch": 2.6754332063414035,
      "grad_norm": 0.08167448146657148,
      "learning_rate": 9.304812136380415e-05,
      "loss": 0.8408,
      "step": 152390
    },
    {
      "epoch": 2.675608771221405,
      "grad_norm": 0.058154736362337346,
      "learning_rate": 9.303701378699564e-05,
      "loss": 0.8422,
      "step": 152400
    },
    {
      "epoch": 2.6757843361014064,
      "grad_norm": 0.0577268815333584,
      "learning_rate": 9.302590631107973e-05,
      "loss": 0.8434,
      "step": 152410
    },
    {
      "epoch": 2.6759599009814075,
      "grad_norm": 0.07425919604586005,
      "learning_rate": 9.301479893619713e-05,
      "loss": 0.8495,
      "step": 152420
    },
    {
      "epoch": 2.676135465861409,
      "grad_norm": 0.07776117096970407,
      "learning_rate": 9.300369166248854e-05,
      "loss": 0.8431,
      "step": 152430
    },
    {
      "epoch": 2.6763110307414104,
      "grad_norm": 0.0615278036862135,
      "learning_rate": 9.299258449009473e-05,
      "loss": 0.8449,
      "step": 152440
    },
    {
      "epoch": 2.676486595621412,
      "grad_norm": 0.06394509945224608,
      "learning_rate": 9.298147741915641e-05,
      "loss": 0.8469,
      "step": 152450
    },
    {
      "epoch": 2.6766621605014134,
      "grad_norm": 0.08245693775963499,
      "learning_rate": 9.29703704498143e-05,
      "loss": 0.8429,
      "step": 152460
    },
    {
      "epoch": 2.676837725381415,
      "grad_norm": 0.054530249223794916,
      "learning_rate": 9.29592635822091e-05,
      "loss": 0.8358,
      "step": 152470
    },
    {
      "epoch": 2.6770132902614163,
      "grad_norm": 0.05399329890527645,
      "learning_rate": 9.294815681648163e-05,
      "loss": 0.8423,
      "step": 152480
    },
    {
      "epoch": 2.6771888551414174,
      "grad_norm": 0.044577209272264995,
      "learning_rate": 9.293705015277251e-05,
      "loss": 0.8401,
      "step": 152490
    },
    {
      "epoch": 2.677364420021419,
      "grad_norm": 0.05659341228061812,
      "learning_rate": 9.292594359122251e-05,
      "loss": 0.8377,
      "step": 152500
    },
    {
      "epoch": 2.6775399849014203,
      "grad_norm": 0.10072920592133022,
      "learning_rate": 9.291483713197237e-05,
      "loss": 0.8464,
      "step": 152510
    },
    {
      "epoch": 2.677715549781422,
      "grad_norm": 0.0588887859597484,
      "learning_rate": 9.29037307751628e-05,
      "loss": 0.8465,
      "step": 152520
    },
    {
      "epoch": 2.6778911146614233,
      "grad_norm": 0.07134665574745754,
      "learning_rate": 9.289262452093451e-05,
      "loss": 0.8477,
      "step": 152530
    },
    {
      "epoch": 2.6780666795414243,
      "grad_norm": 0.04715900935713044,
      "learning_rate": 9.288151836942821e-05,
      "loss": 0.8354,
      "step": 152540
    },
    {
      "epoch": 2.6782422444214258,
      "grad_norm": 0.06731306263124895,
      "learning_rate": 9.287041232078465e-05,
      "loss": 0.8522,
      "step": 152550
    },
    {
      "epoch": 2.6784178093014273,
      "grad_norm": 0.04837418460872785,
      "learning_rate": 9.28593063751445e-05,
      "loss": 0.8431,
      "step": 152560
    },
    {
      "epoch": 2.6785933741814287,
      "grad_norm": 0.07649811518163255,
      "learning_rate": 9.284820053264847e-05,
      "loss": 0.8455,
      "step": 152570
    },
    {
      "epoch": 2.67876893906143,
      "grad_norm": 0.052845130564030285,
      "learning_rate": 9.283709479343735e-05,
      "loss": 0.8413,
      "step": 152580
    },
    {
      "epoch": 2.6789445039414317,
      "grad_norm": 0.05721379556546547,
      "learning_rate": 9.28259891576518e-05,
      "loss": 0.8412,
      "step": 152590
    },
    {
      "epoch": 2.679120068821433,
      "grad_norm": 0.05338628692536719,
      "learning_rate": 9.281488362543252e-05,
      "loss": 0.8471,
      "step": 152600
    },
    {
      "epoch": 2.679295633701434,
      "grad_norm": 0.048449545287269136,
      "learning_rate": 9.280377819692025e-05,
      "loss": 0.8436,
      "step": 152610
    },
    {
      "epoch": 2.6794711985814357,
      "grad_norm": 0.06101197794163572,
      "learning_rate": 9.279267287225566e-05,
      "loss": 0.8449,
      "step": 152620
    },
    {
      "epoch": 2.679646763461437,
      "grad_norm": 0.049279091553564115,
      "learning_rate": 9.27815676515795e-05,
      "loss": 0.8446,
      "step": 152630
    },
    {
      "epoch": 2.6798223283414386,
      "grad_norm": 0.04948952130134291,
      "learning_rate": 9.277046253503242e-05,
      "loss": 0.8443,
      "step": 152640
    },
    {
      "epoch": 2.67999789322144,
      "grad_norm": 0.0571028413126792,
      "learning_rate": 9.27593575227552e-05,
      "loss": 0.8334,
      "step": 152650
    },
    {
      "epoch": 2.680173458101441,
      "grad_norm": 0.054860868213983234,
      "learning_rate": 9.274825261488848e-05,
      "loss": 0.8439,
      "step": 152660
    },
    {
      "epoch": 2.680349022981443,
      "grad_norm": 0.058544555806549714,
      "learning_rate": 9.273714781157295e-05,
      "loss": 0.8461,
      "step": 152670
    },
    {
      "epoch": 2.680524587861444,
      "grad_norm": 0.04527271464649203,
      "learning_rate": 9.272604311294937e-05,
      "loss": 0.8475,
      "step": 152680
    },
    {
      "epoch": 2.6807001527414456,
      "grad_norm": 0.04802682328970565,
      "learning_rate": 9.271493851915842e-05,
      "loss": 0.8498,
      "step": 152690
    },
    {
      "epoch": 2.680875717621447,
      "grad_norm": 0.04704125637769525,
      "learning_rate": 9.270383403034079e-05,
      "loss": 0.8522,
      "step": 152700
    },
    {
      "epoch": 2.6810512825014485,
      "grad_norm": 0.05214428265206426,
      "learning_rate": 9.269272964663714e-05,
      "loss": 0.8365,
      "step": 152710
    },
    {
      "epoch": 2.68122684738145,
      "grad_norm": 0.063973721938654,
      "learning_rate": 9.268162536818821e-05,
      "loss": 0.8453,
      "step": 152720
    },
    {
      "epoch": 2.681402412261451,
      "grad_norm": 0.062142365247881734,
      "learning_rate": 9.267052119513469e-05,
      "loss": 0.8466,
      "step": 152730
    },
    {
      "epoch": 2.6815779771414525,
      "grad_norm": 0.06605197930079247,
      "learning_rate": 9.265941712761723e-05,
      "loss": 0.8446,
      "step": 152740
    },
    {
      "epoch": 2.681753542021454,
      "grad_norm": 0.1240452914172638,
      "learning_rate": 9.26483131657766e-05,
      "loss": 0.843,
      "step": 152750
    },
    {
      "epoch": 2.6819291069014555,
      "grad_norm": 0.0476867999128541,
      "learning_rate": 9.263720930975342e-05,
      "loss": 0.8395,
      "step": 152760
    },
    {
      "epoch": 2.682104671781457,
      "grad_norm": 0.05701071207977274,
      "learning_rate": 9.262610555968841e-05,
      "loss": 0.8386,
      "step": 152770
    },
    {
      "epoch": 2.682280236661458,
      "grad_norm": 0.05165825994183015,
      "learning_rate": 9.261500191572224e-05,
      "loss": 0.8479,
      "step": 152780
    },
    {
      "epoch": 2.68245580154146,
      "grad_norm": 0.058579349653477185,
      "learning_rate": 9.26038983779956e-05,
      "loss": 0.8468,
      "step": 152790
    },
    {
      "epoch": 2.682631366421461,
      "grad_norm": 0.05513710021594284,
      "learning_rate": 9.25927949466492e-05,
      "loss": 0.8461,
      "step": 152800
    },
    {
      "epoch": 2.6828069313014624,
      "grad_norm": 0.05072131288114822,
      "learning_rate": 9.258169162182364e-05,
      "loss": 0.8391,
      "step": 152810
    },
    {
      "epoch": 2.682982496181464,
      "grad_norm": 0.054049144612205204,
      "learning_rate": 9.257058840365973e-05,
      "loss": 0.8323,
      "step": 152820
    },
    {
      "epoch": 2.6831580610614654,
      "grad_norm": 0.04978363878546913,
      "learning_rate": 9.255948529229807e-05,
      "loss": 0.8428,
      "step": 152830
    },
    {
      "epoch": 2.683333625941467,
      "grad_norm": 0.05273420883704863,
      "learning_rate": 9.254838228787929e-05,
      "loss": 0.8385,
      "step": 152840
    },
    {
      "epoch": 2.683509190821468,
      "grad_norm": 0.052685064740719184,
      "learning_rate": 9.253727939054419e-05,
      "loss": 0.85,
      "step": 152850
    },
    {
      "epoch": 2.6836847557014694,
      "grad_norm": 0.05680890757034855,
      "learning_rate": 9.252617660043336e-05,
      "loss": 0.8451,
      "step": 152860
    },
    {
      "epoch": 2.683860320581471,
      "grad_norm": 0.05575378143869414,
      "learning_rate": 9.25150739176875e-05,
      "loss": 0.8394,
      "step": 152870
    },
    {
      "epoch": 2.6840358854614723,
      "grad_norm": 0.06162964861425104,
      "learning_rate": 9.250397134244726e-05,
      "loss": 0.8488,
      "step": 152880
    },
    {
      "epoch": 2.684211450341474,
      "grad_norm": 0.04986582545277134,
      "learning_rate": 9.249286887485335e-05,
      "loss": 0.851,
      "step": 152890
    },
    {
      "epoch": 2.684387015221475,
      "grad_norm": 0.036661447678279696,
      "learning_rate": 9.248176651504643e-05,
      "loss": 0.8373,
      "step": 152900
    },
    {
      "epoch": 2.6845625801014767,
      "grad_norm": 0.051883593238355834,
      "learning_rate": 9.247066426316712e-05,
      "loss": 0.8481,
      "step": 152910
    },
    {
      "epoch": 2.684738144981478,
      "grad_norm": 0.05706996063708991,
      "learning_rate": 9.245956211935617e-05,
      "loss": 0.8455,
      "step": 152920
    },
    {
      "epoch": 2.6849137098614793,
      "grad_norm": 0.052864794094417945,
      "learning_rate": 9.24484600837542e-05,
      "loss": 0.8514,
      "step": 152930
    },
    {
      "epoch": 2.6850892747414807,
      "grad_norm": 0.09781838510003275,
      "learning_rate": 9.243735815650186e-05,
      "loss": 0.8422,
      "step": 152940
    },
    {
      "epoch": 2.685264839621482,
      "grad_norm": 0.063510635387078,
      "learning_rate": 9.242625633773985e-05,
      "loss": 0.8418,
      "step": 152950
    },
    {
      "epoch": 2.6854404045014837,
      "grad_norm": 0.057780057768587495,
      "learning_rate": 9.241515462760881e-05,
      "loss": 0.8523,
      "step": 152960
    },
    {
      "epoch": 2.6856159693814847,
      "grad_norm": 0.06530525194478863,
      "learning_rate": 9.240405302624941e-05,
      "loss": 0.8328,
      "step": 152970
    },
    {
      "epoch": 2.685791534261486,
      "grad_norm": 0.05464484202693666,
      "learning_rate": 9.239295153380228e-05,
      "loss": 0.8437,
      "step": 152980
    },
    {
      "epoch": 2.6859670991414877,
      "grad_norm": 0.04909995707885833,
      "learning_rate": 9.238185015040814e-05,
      "loss": 0.8426,
      "step": 152990
    },
    {
      "epoch": 2.686142664021489,
      "grad_norm": 0.047379316586592356,
      "learning_rate": 9.237074887620761e-05,
      "loss": 0.8469,
      "step": 153000
    },
    {
      "epoch": 2.6863182289014906,
      "grad_norm": 0.06670203320123567,
      "learning_rate": 9.23596477113413e-05,
      "loss": 0.8397,
      "step": 153010
    },
    {
      "epoch": 2.686493793781492,
      "grad_norm": 0.059993525853792314,
      "learning_rate": 9.234854665594994e-05,
      "loss": 0.8358,
      "step": 153020
    },
    {
      "epoch": 2.6866693586614936,
      "grad_norm": 0.060422412492559904,
      "learning_rate": 9.233744571017418e-05,
      "loss": 0.8414,
      "step": 153030
    },
    {
      "epoch": 2.6868449235414946,
      "grad_norm": 0.06027954086393374,
      "learning_rate": 9.232634487415461e-05,
      "loss": 0.8421,
      "step": 153040
    },
    {
      "epoch": 2.687020488421496,
      "grad_norm": 0.06557348287116291,
      "learning_rate": 9.231524414803191e-05,
      "loss": 0.8435,
      "step": 153050
    },
    {
      "epoch": 2.6871960533014976,
      "grad_norm": 0.057888168681653814,
      "learning_rate": 9.230414353194675e-05,
      "loss": 0.8461,
      "step": 153060
    },
    {
      "epoch": 2.687371618181499,
      "grad_norm": 0.05171648842308461,
      "learning_rate": 9.229304302603975e-05,
      "loss": 0.8455,
      "step": 153070
    },
    {
      "epoch": 2.6875471830615005,
      "grad_norm": 0.05724761196840583,
      "learning_rate": 9.228194263045153e-05,
      "loss": 0.8417,
      "step": 153080
    },
    {
      "epoch": 2.6877227479415016,
      "grad_norm": 0.05442034676275662,
      "learning_rate": 9.22708423453228e-05,
      "loss": 0.8439,
      "step": 153090
    },
    {
      "epoch": 2.687898312821503,
      "grad_norm": 0.05721631493794233,
      "learning_rate": 9.225974217079418e-05,
      "loss": 0.8511,
      "step": 153100
    },
    {
      "epoch": 2.6880738777015045,
      "grad_norm": 0.04872197576270406,
      "learning_rate": 9.22486421070063e-05,
      "loss": 0.8492,
      "step": 153110
    },
    {
      "epoch": 2.688249442581506,
      "grad_norm": 0.09018986461301493,
      "learning_rate": 9.223754215409978e-05,
      "loss": 0.8423,
      "step": 153120
    },
    {
      "epoch": 2.6884250074615075,
      "grad_norm": 0.053292042048418246,
      "learning_rate": 9.22264423122153e-05,
      "loss": 0.839,
      "step": 153130
    },
    {
      "epoch": 2.688600572341509,
      "grad_norm": 0.05093580223636449,
      "learning_rate": 9.221534258149348e-05,
      "loss": 0.8379,
      "step": 153140
    },
    {
      "epoch": 2.6887761372215104,
      "grad_norm": 0.0618175551769233,
      "learning_rate": 9.220424296207491e-05,
      "loss": 0.8404,
      "step": 153150
    },
    {
      "epoch": 2.6889517021015115,
      "grad_norm": 0.06750134360588075,
      "learning_rate": 9.219314345410033e-05,
      "loss": 0.8516,
      "step": 153160
    },
    {
      "epoch": 2.689127266981513,
      "grad_norm": 0.05308286021982019,
      "learning_rate": 9.21820440577103e-05,
      "loss": 0.8513,
      "step": 153170
    },
    {
      "epoch": 2.6893028318615144,
      "grad_norm": 0.05170722186165824,
      "learning_rate": 9.21709447730454e-05,
      "loss": 0.8459,
      "step": 153180
    },
    {
      "epoch": 2.689478396741516,
      "grad_norm": 0.05559983936481459,
      "learning_rate": 9.215984560024637e-05,
      "loss": 0.8442,
      "step": 153190
    },
    {
      "epoch": 2.6896539616215174,
      "grad_norm": 0.05506347681803979,
      "learning_rate": 9.214874653945379e-05,
      "loss": 0.8468,
      "step": 153200
    },
    {
      "epoch": 2.6898295265015184,
      "grad_norm": 0.06456457858920216,
      "learning_rate": 9.213764759080829e-05,
      "loss": 0.8459,
      "step": 153210
    },
    {
      "epoch": 2.69000509138152,
      "grad_norm": 0.06547391528723716,
      "learning_rate": 9.212654875445047e-05,
      "loss": 0.8418,
      "step": 153220
    },
    {
      "epoch": 2.6901806562615214,
      "grad_norm": 0.06254885356101456,
      "learning_rate": 9.211545003052101e-05,
      "loss": 0.8437,
      "step": 153230
    },
    {
      "epoch": 2.690356221141523,
      "grad_norm": 0.05929870777963771,
      "learning_rate": 9.210435141916049e-05,
      "loss": 0.8445,
      "step": 153240
    },
    {
      "epoch": 2.6905317860215243,
      "grad_norm": 0.05911177851363282,
      "learning_rate": 9.20932529205095e-05,
      "loss": 0.8476,
      "step": 153250
    },
    {
      "epoch": 2.690707350901526,
      "grad_norm": 0.05018336492926591,
      "learning_rate": 9.208215453470875e-05,
      "loss": 0.8483,
      "step": 153260
    },
    {
      "epoch": 2.6908829157815273,
      "grad_norm": 0.06641238102218287,
      "learning_rate": 9.207105626189882e-05,
      "loss": 0.8412,
      "step": 153270
    },
    {
      "epoch": 2.6910584806615283,
      "grad_norm": 0.051386439055780976,
      "learning_rate": 9.20599581022203e-05,
      "loss": 0.8446,
      "step": 153280
    },
    {
      "epoch": 2.69123404554153,
      "grad_norm": 0.04711846172737254,
      "learning_rate": 9.204886005581381e-05,
      "loss": 0.8362,
      "step": 153290
    },
    {
      "epoch": 2.6914096104215313,
      "grad_norm": 0.06187678090730837,
      "learning_rate": 9.203776212282e-05,
      "loss": 0.8462,
      "step": 153300
    },
    {
      "epoch": 2.6915851753015327,
      "grad_norm": 0.04873260961512266,
      "learning_rate": 9.202666430337947e-05,
      "loss": 0.8467,
      "step": 153310
    },
    {
      "epoch": 2.691760740181534,
      "grad_norm": 0.10651066365578471,
      "learning_rate": 9.201556659763276e-05,
      "loss": 0.8439,
      "step": 153320
    },
    {
      "epoch": 2.6919363050615353,
      "grad_norm": 0.06242569013089358,
      "learning_rate": 9.200446900572062e-05,
      "loss": 0.8394,
      "step": 153330
    },
    {
      "epoch": 2.692111869941537,
      "grad_norm": 0.05534419357814975,
      "learning_rate": 9.199337152778357e-05,
      "loss": 0.8447,
      "step": 153340
    },
    {
      "epoch": 2.692287434821538,
      "grad_norm": 0.06538293376870602,
      "learning_rate": 9.198227416396216e-05,
      "loss": 0.8462,
      "step": 153350
    },
    {
      "epoch": 2.6924629997015397,
      "grad_norm": 0.0530169205770271,
      "learning_rate": 9.197117691439713e-05,
      "loss": 0.8431,
      "step": 153360
    },
    {
      "epoch": 2.692638564581541,
      "grad_norm": 0.045660905166294175,
      "learning_rate": 9.196007977922902e-05,
      "loss": 0.8432,
      "step": 153370
    },
    {
      "epoch": 2.6928141294615426,
      "grad_norm": 0.052563008604315724,
      "learning_rate": 9.194898275859841e-05,
      "loss": 0.842,
      "step": 153380
    },
    {
      "epoch": 2.692989694341544,
      "grad_norm": 0.05196975902888274,
      "learning_rate": 9.193788585264591e-05,
      "loss": 0.8412,
      "step": 153390
    },
    {
      "epoch": 2.693165259221545,
      "grad_norm": 0.07110607277942092,
      "learning_rate": 9.192678906151216e-05,
      "loss": 0.844,
      "step": 153400
    },
    {
      "epoch": 2.6933408241015466,
      "grad_norm": 0.05035163265839753,
      "learning_rate": 9.191569238533773e-05,
      "loss": 0.8377,
      "step": 153410
    },
    {
      "epoch": 2.693516388981548,
      "grad_norm": 0.06549385743011141,
      "learning_rate": 9.190459582426318e-05,
      "loss": 0.8429,
      "step": 153420
    },
    {
      "epoch": 2.6936919538615496,
      "grad_norm": 0.06503531568155219,
      "learning_rate": 9.189349937842919e-05,
      "loss": 0.8517,
      "step": 153430
    },
    {
      "epoch": 2.693867518741551,
      "grad_norm": 0.05598572994059235,
      "learning_rate": 9.18824030479763e-05,
      "loss": 0.8329,
      "step": 153440
    },
    {
      "epoch": 2.694043083621552,
      "grad_norm": 0.0514939793722406,
      "learning_rate": 9.187130683304511e-05,
      "loss": 0.8436,
      "step": 153450
    },
    {
      "epoch": 2.694218648501554,
      "grad_norm": 0.053348599832058384,
      "learning_rate": 9.186021073377619e-05,
      "loss": 0.8374,
      "step": 153460
    },
    {
      "epoch": 2.694394213381555,
      "grad_norm": 0.0529937935921469,
      "learning_rate": 9.184911475031019e-05,
      "loss": 0.8492,
      "step": 153470
    },
    {
      "epoch": 2.6945697782615565,
      "grad_norm": 0.04955633215574851,
      "learning_rate": 9.183801888278764e-05,
      "loss": 0.8443,
      "step": 153480
    },
    {
      "epoch": 2.694745343141558,
      "grad_norm": 0.05057965255667922,
      "learning_rate": 9.182692313134911e-05,
      "loss": 0.8396,
      "step": 153490
    },
    {
      "epoch": 2.6949209080215595,
      "grad_norm": 0.05475695338081478,
      "learning_rate": 9.181582749613528e-05,
      "loss": 0.8414,
      "step": 153500
    },
    {
      "epoch": 2.695096472901561,
      "grad_norm": 0.04393662105109547,
      "learning_rate": 9.180473197728666e-05,
      "loss": 0.8397,
      "step": 153510
    },
    {
      "epoch": 2.695272037781562,
      "grad_norm": 0.049039913766967876,
      "learning_rate": 9.17936365749438e-05,
      "loss": 0.8434,
      "step": 153520
    },
    {
      "epoch": 2.6954476026615635,
      "grad_norm": 0.04907528828096971,
      "learning_rate": 9.178254128924738e-05,
      "loss": 0.8341,
      "step": 153530
    },
    {
      "epoch": 2.695623167541565,
      "grad_norm": 0.053067124460201155,
      "learning_rate": 9.177144612033794e-05,
      "loss": 0.842,
      "step": 153540
    },
    {
      "epoch": 2.6957987324215664,
      "grad_norm": 0.046878226473837645,
      "learning_rate": 9.176035106835602e-05,
      "loss": 0.8465,
      "step": 153550
    },
    {
      "epoch": 2.695974297301568,
      "grad_norm": 0.08530757232935116,
      "learning_rate": 9.17492561334422e-05,
      "loss": 0.8449,
      "step": 153560
    },
    {
      "epoch": 2.696149862181569,
      "grad_norm": 0.06088777119722741,
      "learning_rate": 9.173816131573713e-05,
      "loss": 0.8421,
      "step": 153570
    },
    {
      "epoch": 2.696325427061571,
      "grad_norm": 0.050517659489040856,
      "learning_rate": 9.172706661538131e-05,
      "loss": 0.84,
      "step": 153580
    },
    {
      "epoch": 2.696500991941572,
      "grad_norm": 0.06578177206286466,
      "learning_rate": 9.17159720325153e-05,
      "loss": 0.8447,
      "step": 153590
    },
    {
      "epoch": 2.6966765568215734,
      "grad_norm": 0.06962017313922292,
      "learning_rate": 9.170487756727974e-05,
      "loss": 0.8388,
      "step": 153600
    },
    {
      "epoch": 2.696852121701575,
      "grad_norm": 0.07793316376677074,
      "learning_rate": 9.169378321981515e-05,
      "loss": 0.8365,
      "step": 153610
    },
    {
      "epoch": 2.6970276865815763,
      "grad_norm": 0.06866811149347585,
      "learning_rate": 9.168268899026212e-05,
      "loss": 0.8445,
      "step": 153620
    },
    {
      "epoch": 2.697203251461578,
      "grad_norm": 0.048853286756278994,
      "learning_rate": 9.16715948787612e-05,
      "loss": 0.8395,
      "step": 153630
    },
    {
      "epoch": 2.697378816341579,
      "grad_norm": 0.06996334532792522,
      "learning_rate": 9.166050088545296e-05,
      "loss": 0.8413,
      "step": 153640
    },
    {
      "epoch": 2.6975543812215803,
      "grad_norm": 0.04943857005628,
      "learning_rate": 9.164940701047798e-05,
      "loss": 0.8343,
      "step": 153650
    },
    {
      "epoch": 2.697729946101582,
      "grad_norm": 0.06516058592031486,
      "learning_rate": 9.163831325397676e-05,
      "loss": 0.8473,
      "step": 153660
    },
    {
      "epoch": 2.6979055109815833,
      "grad_norm": 0.05561851446192938,
      "learning_rate": 9.162721961608997e-05,
      "loss": 0.8395,
      "step": 153670
    },
    {
      "epoch": 2.6980810758615847,
      "grad_norm": 0.07455233133567626,
      "learning_rate": 9.161612609695807e-05,
      "loss": 0.8339,
      "step": 153680
    },
    {
      "epoch": 2.6982566407415858,
      "grad_norm": 0.07876281509117587,
      "learning_rate": 9.160503269672162e-05,
      "loss": 0.8388,
      "step": 153690
    },
    {
      "epoch": 2.6984322056215877,
      "grad_norm": 0.07047235925325852,
      "learning_rate": 9.159393941552123e-05,
      "loss": 0.836,
      "step": 153700
    },
    {
      "epoch": 2.6986077705015887,
      "grad_norm": 0.05342458521997134,
      "learning_rate": 9.158284625349746e-05,
      "loss": 0.8401,
      "step": 153710
    },
    {
      "epoch": 2.69878333538159,
      "grad_norm": 0.07257108756036125,
      "learning_rate": 9.157175321079081e-05,
      "loss": 0.8528,
      "step": 153720
    },
    {
      "epoch": 2.6989589002615917,
      "grad_norm": 0.0721117561253353,
      "learning_rate": 9.156066028754184e-05,
      "loss": 0.8447,
      "step": 153730
    },
    {
      "epoch": 2.699134465141593,
      "grad_norm": 0.06233727174805513,
      "learning_rate": 9.154956748389112e-05,
      "loss": 0.8331,
      "step": 153740
    },
    {
      "epoch": 2.6993100300215946,
      "grad_norm": 0.05722367628238072,
      "learning_rate": 9.153847479997923e-05,
      "loss": 0.8376,
      "step": 153750
    },
    {
      "epoch": 2.6994855949015957,
      "grad_norm": 0.06261171647119032,
      "learning_rate": 9.15273822359466e-05,
      "loss": 0.847,
      "step": 153760
    },
    {
      "epoch": 2.699661159781597,
      "grad_norm": 0.0556007451751719,
      "learning_rate": 9.151628979193391e-05,
      "loss": 0.8412,
      "step": 153770
    },
    {
      "epoch": 2.6998367246615986,
      "grad_norm": 0.053886841567105215,
      "learning_rate": 9.150519746808164e-05,
      "loss": 0.8418,
      "step": 153780
    },
    {
      "epoch": 2.7000122895416,
      "grad_norm": 0.05779219394679433,
      "learning_rate": 9.149410526453034e-05,
      "loss": 0.8449,
      "step": 153790
    },
    {
      "epoch": 2.7001878544216016,
      "grad_norm": 0.05190953529280323,
      "learning_rate": 9.148301318142052e-05,
      "loss": 0.8456,
      "step": 153800
    },
    {
      "epoch": 2.700363419301603,
      "grad_norm": 0.04780340208951375,
      "learning_rate": 9.147192121889277e-05,
      "loss": 0.8444,
      "step": 153810
    },
    {
      "epoch": 2.7005389841816045,
      "grad_norm": 0.06215655355941395,
      "learning_rate": 9.146082937708762e-05,
      "loss": 0.8454,
      "step": 153820
    },
    {
      "epoch": 2.7007145490616056,
      "grad_norm": 0.06527664480600504,
      "learning_rate": 9.144973765614553e-05,
      "loss": 0.8459,
      "step": 153830
    },
    {
      "epoch": 2.700890113941607,
      "grad_norm": 0.05796336598298572,
      "learning_rate": 9.143864605620718e-05,
      "loss": 0.8417,
      "step": 153840
    },
    {
      "epoch": 2.7010656788216085,
      "grad_norm": 0.07369426273021604,
      "learning_rate": 9.142755457741298e-05,
      "loss": 0.8406,
      "step": 153850
    },
    {
      "epoch": 2.70124124370161,
      "grad_norm": 0.0471214403386902,
      "learning_rate": 9.141646321990344e-05,
      "loss": 0.8385,
      "step": 153860
    },
    {
      "epoch": 2.7014168085816115,
      "grad_norm": 0.06278374196428196,
      "learning_rate": 9.14053719838192e-05,
      "loss": 0.8455,
      "step": 153870
    },
    {
      "epoch": 2.7015923734616125,
      "grad_norm": 0.05743047792616827,
      "learning_rate": 9.139428086930076e-05,
      "loss": 0.8481,
      "step": 153880
    },
    {
      "epoch": 2.701767938341614,
      "grad_norm": 0.06436924970980892,
      "learning_rate": 9.13831898764886e-05,
      "loss": 0.835,
      "step": 153890
    },
    {
      "epoch": 2.7019435032216155,
      "grad_norm": 0.051876956911328514,
      "learning_rate": 9.137209900552325e-05,
      "loss": 0.8415,
      "step": 153900
    },
    {
      "epoch": 2.702119068101617,
      "grad_norm": 0.058844055478789764,
      "learning_rate": 9.136100825654526e-05,
      "loss": 0.8446,
      "step": 153910
    },
    {
      "epoch": 2.7022946329816184,
      "grad_norm": 0.05207096601907242,
      "learning_rate": 9.134991762969515e-05,
      "loss": 0.8422,
      "step": 153920
    },
    {
      "epoch": 2.70247019786162,
      "grad_norm": 0.044200884685514225,
      "learning_rate": 9.133882712511341e-05,
      "loss": 0.8397,
      "step": 153930
    },
    {
      "epoch": 2.7026457627416214,
      "grad_norm": 0.04316051261683632,
      "learning_rate": 9.13277367429406e-05,
      "loss": 0.8487,
      "step": 153940
    },
    {
      "epoch": 2.7028213276216224,
      "grad_norm": 0.07076930291847372,
      "learning_rate": 9.131664648331724e-05,
      "loss": 0.8445,
      "step": 153950
    },
    {
      "epoch": 2.702996892501624,
      "grad_norm": 0.044837474175757376,
      "learning_rate": 9.13055563463838e-05,
      "loss": 0.8436,
      "step": 153960
    },
    {
      "epoch": 2.7031724573816254,
      "grad_norm": 0.05500627010720617,
      "learning_rate": 9.129446633228083e-05,
      "loss": 0.8472,
      "step": 153970
    },
    {
      "epoch": 2.703348022261627,
      "grad_norm": 0.04603671719646018,
      "learning_rate": 9.128337644114881e-05,
      "loss": 0.8512,
      "step": 153980
    },
    {
      "epoch": 2.7035235871416283,
      "grad_norm": 0.04999783791144092,
      "learning_rate": 9.12722866731283e-05,
      "loss": 0.8421,
      "step": 153990
    },
    {
      "epoch": 2.7036991520216294,
      "grad_norm": 0.0525331583989158,
      "learning_rate": 9.126119702835975e-05,
      "loss": 0.8382,
      "step": 154000
    },
    {
      "epoch": 2.703874716901631,
      "grad_norm": 0.044358415572394096,
      "learning_rate": 9.125010750698375e-05,
      "loss": 0.8454,
      "step": 154010
    },
    {
      "epoch": 2.7040502817816323,
      "grad_norm": 0.06310074062970107,
      "learning_rate": 9.123901810914073e-05,
      "loss": 0.8495,
      "step": 154020
    },
    {
      "epoch": 2.704225846661634,
      "grad_norm": 0.056123534307589096,
      "learning_rate": 9.122792883497119e-05,
      "loss": 0.8393,
      "step": 154030
    },
    {
      "epoch": 2.7044014115416353,
      "grad_norm": 0.05940787139523218,
      "learning_rate": 9.121683968461569e-05,
      "loss": 0.8468,
      "step": 154040
    },
    {
      "epoch": 2.7045769764216367,
      "grad_norm": 0.06157938857338847,
      "learning_rate": 9.120575065821471e-05,
      "loss": 0.8435,
      "step": 154050
    },
    {
      "epoch": 2.7047525413016382,
      "grad_norm": 0.05517299525637102,
      "learning_rate": 9.119466175590876e-05,
      "loss": 0.8462,
      "step": 154060
    },
    {
      "epoch": 2.7049281061816393,
      "grad_norm": 0.06666601081659004,
      "learning_rate": 9.11835729778383e-05,
      "loss": 0.8379,
      "step": 154070
    },
    {
      "epoch": 2.7051036710616407,
      "grad_norm": 0.048283391591432495,
      "learning_rate": 9.117248432414387e-05,
      "loss": 0.8406,
      "step": 154080
    },
    {
      "epoch": 2.705279235941642,
      "grad_norm": 0.05275829574982348,
      "learning_rate": 9.116139579496595e-05,
      "loss": 0.8488,
      "step": 154090
    },
    {
      "epoch": 2.7054548008216437,
      "grad_norm": 0.0455556000816163,
      "learning_rate": 9.115030739044497e-05,
      "loss": 0.8545,
      "step": 154100
    },
    {
      "epoch": 2.705630365701645,
      "grad_norm": 0.05361970074250794,
      "learning_rate": 9.113921911072155e-05,
      "loss": 0.8462,
      "step": 154110
    },
    {
      "epoch": 2.705805930581646,
      "grad_norm": 0.062136363167253454,
      "learning_rate": 9.11281309559361e-05,
      "loss": 0.8429,
      "step": 154120
    },
    {
      "epoch": 2.705981495461648,
      "grad_norm": 0.051722349490947825,
      "learning_rate": 9.111704292622913e-05,
      "loss": 0.8332,
      "step": 154130
    },
    {
      "epoch": 2.706157060341649,
      "grad_norm": 0.05756188599949267,
      "learning_rate": 9.11059550217411e-05,
      "loss": 0.8449,
      "step": 154140
    },
    {
      "epoch": 2.7063326252216506,
      "grad_norm": 0.05775776207357769,
      "learning_rate": 9.109486724261254e-05,
      "loss": 0.8443,
      "step": 154150
    },
    {
      "epoch": 2.706508190101652,
      "grad_norm": 0.07700578540503504,
      "learning_rate": 9.10837795889839e-05,
      "loss": 0.8346,
      "step": 154160
    },
    {
      "epoch": 2.7066837549816536,
      "grad_norm": 0.05221660804497351,
      "learning_rate": 9.107269206099563e-05,
      "loss": 0.8392,
      "step": 154170
    },
    {
      "epoch": 2.706859319861655,
      "grad_norm": 0.04980954410128407,
      "learning_rate": 9.106160465878832e-05,
      "loss": 0.8367,
      "step": 154180
    },
    {
      "epoch": 2.707034884741656,
      "grad_norm": 0.056611677114973495,
      "learning_rate": 9.105051738250236e-05,
      "loss": 0.8443,
      "step": 154190
    },
    {
      "epoch": 2.7072104496216576,
      "grad_norm": 0.05638718061439771,
      "learning_rate": 9.103943023227822e-05,
      "loss": 0.8407,
      "step": 154200
    },
    {
      "epoch": 2.707386014501659,
      "grad_norm": 0.04058570117672139,
      "learning_rate": 9.102834320825643e-05,
      "loss": 0.8506,
      "step": 154210
    },
    {
      "epoch": 2.7075615793816605,
      "grad_norm": 0.054873678756784304,
      "learning_rate": 9.101725631057747e-05,
      "loss": 0.837,
      "step": 154220
    },
    {
      "epoch": 2.707737144261662,
      "grad_norm": 0.05080532146899716,
      "learning_rate": 9.100616953938176e-05,
      "loss": 0.8479,
      "step": 154230
    },
    {
      "epoch": 2.707912709141663,
      "grad_norm": 0.04676975465942692,
      "learning_rate": 9.09950828948098e-05,
      "loss": 0.8404,
      "step": 154240
    },
    {
      "epoch": 2.708088274021665,
      "grad_norm": 0.058767114667575754,
      "learning_rate": 9.098399637700207e-05,
      "loss": 0.85,
      "step": 154250
    },
    {
      "epoch": 2.708263838901666,
      "grad_norm": 0.06564695790522658,
      "learning_rate": 9.097290998609902e-05,
      "loss": 0.8439,
      "step": 154260
    },
    {
      "epoch": 2.7084394037816675,
      "grad_norm": 0.044204615132739,
      "learning_rate": 9.096182372224108e-05,
      "loss": 0.8471,
      "step": 154270
    },
    {
      "epoch": 2.708614968661669,
      "grad_norm": 0.052009669332128555,
      "learning_rate": 9.09507375855688e-05,
      "loss": 0.8393,
      "step": 154280
    },
    {
      "epoch": 2.7087905335416704,
      "grad_norm": 0.07015435611322605,
      "learning_rate": 9.09396515762226e-05,
      "loss": 0.8459,
      "step": 154290
    },
    {
      "epoch": 2.708966098421672,
      "grad_norm": 0.04896178396777164,
      "learning_rate": 9.092856569434295e-05,
      "loss": 0.8447,
      "step": 154300
    },
    {
      "epoch": 2.709141663301673,
      "grad_norm": 0.054790024006316315,
      "learning_rate": 9.091747994007028e-05,
      "loss": 0.8509,
      "step": 154310
    },
    {
      "epoch": 2.7093172281816744,
      "grad_norm": 0.05624928379824036,
      "learning_rate": 9.090639431354509e-05,
      "loss": 0.8482,
      "step": 154320
    },
    {
      "epoch": 2.709492793061676,
      "grad_norm": 0.04987514822911432,
      "learning_rate": 9.089530881490781e-05,
      "loss": 0.8381,
      "step": 154330
    },
    {
      "epoch": 2.7096683579416774,
      "grad_norm": 0.05808821692306947,
      "learning_rate": 9.088422344429888e-05,
      "loss": 0.8375,
      "step": 154340
    },
    {
      "epoch": 2.709843922821679,
      "grad_norm": 0.06316372777602167,
      "learning_rate": 9.087313820185882e-05,
      "loss": 0.8444,
      "step": 154350
    },
    {
      "epoch": 2.71001948770168,
      "grad_norm": 0.04807315940573827,
      "learning_rate": 9.086205308772803e-05,
      "loss": 0.8508,
      "step": 154360
    },
    {
      "epoch": 2.710195052581682,
      "grad_norm": 0.05509055168015543,
      "learning_rate": 9.085096810204693e-05,
      "loss": 0.8375,
      "step": 154370
    },
    {
      "epoch": 2.710370617461683,
      "grad_norm": 0.05441390840489102,
      "learning_rate": 9.083988324495604e-05,
      "loss": 0.8341,
      "step": 154380
    },
    {
      "epoch": 2.7105461823416843,
      "grad_norm": 0.07348029507797692,
      "learning_rate": 9.082879851659579e-05,
      "loss": 0.8504,
      "step": 154390
    },
    {
      "epoch": 2.710721747221686,
      "grad_norm": 0.0532918310212392,
      "learning_rate": 9.081771391710662e-05,
      "loss": 0.8469,
      "step": 154400
    },
    {
      "epoch": 2.7108973121016873,
      "grad_norm": 0.04244320706785365,
      "learning_rate": 9.080662944662892e-05,
      "loss": 0.8443,
      "step": 154410
    },
    {
      "epoch": 2.7110728769816888,
      "grad_norm": 0.07469473152827645,
      "learning_rate": 9.079554510530321e-05,
      "loss": 0.8469,
      "step": 154420
    },
    {
      "epoch": 2.71124844186169,
      "grad_norm": 0.05150949478395944,
      "learning_rate": 9.07844608932699e-05,
      "loss": 0.8472,
      "step": 154430
    },
    {
      "epoch": 2.7114240067416913,
      "grad_norm": 0.07386180806879152,
      "learning_rate": 9.07733768106694e-05,
      "loss": 0.8467,
      "step": 154440
    },
    {
      "epoch": 2.7115995716216927,
      "grad_norm": 0.05055191450667085,
      "learning_rate": 9.07622928576422e-05,
      "loss": 0.8474,
      "step": 154450
    },
    {
      "epoch": 2.711775136501694,
      "grad_norm": 0.046952911724529145,
      "learning_rate": 9.07512090343287e-05,
      "loss": 0.8484,
      "step": 154460
    },
    {
      "epoch": 2.7119507013816957,
      "grad_norm": 0.04894912833141031,
      "learning_rate": 9.074012534086937e-05,
      "loss": 0.8351,
      "step": 154470
    },
    {
      "epoch": 2.712126266261697,
      "grad_norm": 0.057365381454763505,
      "learning_rate": 9.072904177740459e-05,
      "loss": 0.8443,
      "step": 154480
    },
    {
      "epoch": 2.7123018311416986,
      "grad_norm": 0.05471033307262754,
      "learning_rate": 9.071795834407484e-05,
      "loss": 0.848,
      "step": 154490
    },
    {
      "epoch": 2.7124773960216997,
      "grad_norm": 0.05752915237665505,
      "learning_rate": 9.070687504102053e-05,
      "loss": 0.8498,
      "step": 154500
    },
    {
      "epoch": 2.712652960901701,
      "grad_norm": 0.04352731607760854,
      "learning_rate": 9.069579186838204e-05,
      "loss": 0.8467,
      "step": 154510
    },
    {
      "epoch": 2.7128285257817026,
      "grad_norm": 0.04287803718277755,
      "learning_rate": 9.06847088262999e-05,
      "loss": 0.8428,
      "step": 154520
    },
    {
      "epoch": 2.713004090661704,
      "grad_norm": 0.05810708627929148,
      "learning_rate": 9.067362591491447e-05,
      "loss": 0.8379,
      "step": 154530
    },
    {
      "epoch": 2.7131796555417056,
      "grad_norm": 0.05588771275889898,
      "learning_rate": 9.066254313436613e-05,
      "loss": 0.843,
      "step": 154540
    },
    {
      "epoch": 2.7133552204217066,
      "grad_norm": 0.06212812533486526,
      "learning_rate": 9.065146048479539e-05,
      "loss": 0.8513,
      "step": 154550
    },
    {
      "epoch": 2.713530785301708,
      "grad_norm": 0.05174527078240542,
      "learning_rate": 9.064037796634263e-05,
      "loss": 0.8391,
      "step": 154560
    },
    {
      "epoch": 2.7137063501817096,
      "grad_norm": 0.06842215993733804,
      "learning_rate": 9.062929557914825e-05,
      "loss": 0.8495,
      "step": 154570
    },
    {
      "epoch": 2.713881915061711,
      "grad_norm": 0.04609208194836482,
      "learning_rate": 9.061821332335268e-05,
      "loss": 0.8478,
      "step": 154580
    },
    {
      "epoch": 2.7140574799417125,
      "grad_norm": 0.053084821688936586,
      "learning_rate": 9.060713119909637e-05,
      "loss": 0.8481,
      "step": 154590
    },
    {
      "epoch": 2.714233044821714,
      "grad_norm": 0.05086414747581364,
      "learning_rate": 9.059604920651968e-05,
      "loss": 0.8405,
      "step": 154600
    },
    {
      "epoch": 2.7144086097017155,
      "grad_norm": 0.04840270268910114,
      "learning_rate": 9.0584967345763e-05,
      "loss": 0.8483,
      "step": 154610
    },
    {
      "epoch": 2.7145841745817165,
      "grad_norm": 0.06728584570453013,
      "learning_rate": 9.057388561696681e-05,
      "loss": 0.8339,
      "step": 154620
    },
    {
      "epoch": 2.714759739461718,
      "grad_norm": 0.060944344093464005,
      "learning_rate": 9.056280402027151e-05,
      "loss": 0.8473,
      "step": 154630
    },
    {
      "epoch": 2.7149353043417195,
      "grad_norm": 0.07926794592147861,
      "learning_rate": 9.055172255581747e-05,
      "loss": 0.8424,
      "step": 154640
    },
    {
      "epoch": 2.715110869221721,
      "grad_norm": 0.05761421930222924,
      "learning_rate": 9.054064122374509e-05,
      "loss": 0.8446,
      "step": 154650
    },
    {
      "epoch": 2.7152864341017224,
      "grad_norm": 0.06798534591052041,
      "learning_rate": 9.052956002419482e-05,
      "loss": 0.843,
      "step": 154660
    },
    {
      "epoch": 2.7154619989817235,
      "grad_norm": 0.0650564766198692,
      "learning_rate": 9.051847895730703e-05,
      "loss": 0.831,
      "step": 154670
    },
    {
      "epoch": 2.715637563861725,
      "grad_norm": 0.07578954677868643,
      "learning_rate": 9.050739802322207e-05,
      "loss": 0.8381,
      "step": 154680
    },
    {
      "epoch": 2.7158131287417264,
      "grad_norm": 0.08638958128584583,
      "learning_rate": 9.049631722208045e-05,
      "loss": 0.8482,
      "step": 154690
    },
    {
      "epoch": 2.715988693621728,
      "grad_norm": 0.09309691517575748,
      "learning_rate": 9.048523655402249e-05,
      "loss": 0.8435,
      "step": 154700
    },
    {
      "epoch": 2.7161642585017294,
      "grad_norm": 0.05527126490152272,
      "learning_rate": 9.047415601918855e-05,
      "loss": 0.8418,
      "step": 154710
    },
    {
      "epoch": 2.716339823381731,
      "grad_norm": 0.051983791280825455,
      "learning_rate": 9.046307561771912e-05,
      "loss": 0.8377,
      "step": 154720
    },
    {
      "epoch": 2.7165153882617323,
      "grad_norm": 0.05951340077736277,
      "learning_rate": 9.045199534975453e-05,
      "loss": 0.8477,
      "step": 154730
    },
    {
      "epoch": 2.7166909531417334,
      "grad_norm": 0.057038106176347886,
      "learning_rate": 9.04409152154352e-05,
      "loss": 0.8399,
      "step": 154740
    },
    {
      "epoch": 2.716866518021735,
      "grad_norm": 0.05871381566064623,
      "learning_rate": 9.042983521490144e-05,
      "loss": 0.8399,
      "step": 154750
    },
    {
      "epoch": 2.7170420829017363,
      "grad_norm": 0.06928297804635102,
      "learning_rate": 9.041875534829375e-05,
      "loss": 0.8425,
      "step": 154760
    },
    {
      "epoch": 2.717217647781738,
      "grad_norm": 0.05589264409308278,
      "learning_rate": 9.040767561575243e-05,
      "loss": 0.84,
      "step": 154770
    },
    {
      "epoch": 2.7173932126617393,
      "grad_norm": 0.05220679495985041,
      "learning_rate": 9.039659601741788e-05,
      "loss": 0.8434,
      "step": 154780
    },
    {
      "epoch": 2.7175687775417403,
      "grad_norm": 0.04341720738806322,
      "learning_rate": 9.038551655343052e-05,
      "loss": 0.8452,
      "step": 154790
    },
    {
      "epoch": 2.7177443424217422,
      "grad_norm": 0.04685448440804894,
      "learning_rate": 9.037443722393068e-05,
      "loss": 0.8488,
      "step": 154800
    },
    {
      "epoch": 2.7179199073017433,
      "grad_norm": 0.05113302605684361,
      "learning_rate": 9.036335802905878e-05,
      "loss": 0.8475,
      "step": 154810
    },
    {
      "epoch": 2.7180954721817447,
      "grad_norm": 0.06468493185968716,
      "learning_rate": 9.035227896895515e-05,
      "loss": 0.8491,
      "step": 154820
    },
    {
      "epoch": 2.718271037061746,
      "grad_norm": 0.058565900532168674,
      "learning_rate": 9.034120004376018e-05,
      "loss": 0.8451,
      "step": 154830
    },
    {
      "epoch": 2.7184466019417477,
      "grad_norm": 0.051395010155118895,
      "learning_rate": 9.033012125361428e-05,
      "loss": 0.8348,
      "step": 154840
    },
    {
      "epoch": 2.718622166821749,
      "grad_norm": 0.05079786089220619,
      "learning_rate": 9.031904259865773e-05,
      "loss": 0.8381,
      "step": 154850
    },
    {
      "epoch": 2.71879773170175,
      "grad_norm": 0.06225748197008897,
      "learning_rate": 9.030796407903103e-05,
      "loss": 0.844,
      "step": 154860
    },
    {
      "epoch": 2.7189732965817517,
      "grad_norm": 0.07903351960489867,
      "learning_rate": 9.029688569487446e-05,
      "loss": 0.8371,
      "step": 154870
    },
    {
      "epoch": 2.719148861461753,
      "grad_norm": 0.06985732252741639,
      "learning_rate": 9.028580744632834e-05,
      "loss": 0.8406,
      "step": 154880
    },
    {
      "epoch": 2.7193244263417546,
      "grad_norm": 0.05579558246732594,
      "learning_rate": 9.027472933353314e-05,
      "loss": 0.8514,
      "step": 154890
    },
    {
      "epoch": 2.719499991221756,
      "grad_norm": 0.05597785421429863,
      "learning_rate": 9.026365135662918e-05,
      "loss": 0.8373,
      "step": 154900
    },
    {
      "epoch": 2.719675556101757,
      "grad_norm": 0.04847088340863627,
      "learning_rate": 9.02525735157568e-05,
      "loss": 0.8479,
      "step": 154910
    },
    {
      "epoch": 2.719851120981759,
      "grad_norm": 0.05033729156039539,
      "learning_rate": 9.024149581105637e-05,
      "loss": 0.8466,
      "step": 154920
    },
    {
      "epoch": 2.72002668586176,
      "grad_norm": 0.05889381389640331,
      "learning_rate": 9.023041824266826e-05,
      "loss": 0.8491,
      "step": 154930
    },
    {
      "epoch": 2.7202022507417616,
      "grad_norm": 0.048972518779509644,
      "learning_rate": 9.021934081073284e-05,
      "loss": 0.8462,
      "step": 154940
    },
    {
      "epoch": 2.720377815621763,
      "grad_norm": 0.06892656764609656,
      "learning_rate": 9.020826351539036e-05,
      "loss": 0.8357,
      "step": 154950
    },
    {
      "epoch": 2.7205533805017645,
      "grad_norm": 0.05836119327936013,
      "learning_rate": 9.01971863567813e-05,
      "loss": 0.8442,
      "step": 154960
    },
    {
      "epoch": 2.720728945381766,
      "grad_norm": 0.050887022714182044,
      "learning_rate": 9.018610933504597e-05,
      "loss": 0.8478,
      "step": 154970
    },
    {
      "epoch": 2.720904510261767,
      "grad_norm": 0.0664802956001212,
      "learning_rate": 9.017503245032472e-05,
      "loss": 0.8401,
      "step": 154980
    },
    {
      "epoch": 2.7210800751417685,
      "grad_norm": 0.043442031432604514,
      "learning_rate": 9.016395570275783e-05,
      "loss": 0.8412,
      "step": 154990
    },
    {
      "epoch": 2.72125564002177,
      "grad_norm": 0.06546213748836832,
      "learning_rate": 9.015287909248574e-05,
      "loss": 0.8484,
      "step": 155000
    },
    {
      "epoch": 2.7214312049017715,
      "grad_norm": 0.06110184837742583,
      "learning_rate": 9.014180261964874e-05,
      "loss": 0.8394,
      "step": 155010
    },
    {
      "epoch": 2.721606769781773,
      "grad_norm": 0.050507732489943155,
      "learning_rate": 9.013072628438714e-05,
      "loss": 0.8448,
      "step": 155020
    },
    {
      "epoch": 2.721782334661774,
      "grad_norm": 0.0581979095215319,
      "learning_rate": 9.01196500868414e-05,
      "loss": 0.8426,
      "step": 155030
    },
    {
      "epoch": 2.721957899541776,
      "grad_norm": 0.05624933325167508,
      "learning_rate": 9.010857402715175e-05,
      "loss": 0.8402,
      "step": 155040
    },
    {
      "epoch": 2.722133464421777,
      "grad_norm": 0.049470777589182516,
      "learning_rate": 9.00974981054585e-05,
      "loss": 0.8402,
      "step": 155050
    },
    {
      "epoch": 2.7223090293017784,
      "grad_norm": 0.052497273487474294,
      "learning_rate": 9.00864223219021e-05,
      "loss": 0.8449,
      "step": 155060
    },
    {
      "epoch": 2.72248459418178,
      "grad_norm": 0.07556388436896447,
      "learning_rate": 9.00753466766228e-05,
      "loss": 0.8481,
      "step": 155070
    },
    {
      "epoch": 2.7226601590617814,
      "grad_norm": 0.05991658900482496,
      "learning_rate": 9.006427116976096e-05,
      "loss": 0.8405,
      "step": 155080
    },
    {
      "epoch": 2.722835723941783,
      "grad_norm": 0.058074178570444945,
      "learning_rate": 9.005319580145689e-05,
      "loss": 0.8426,
      "step": 155090
    },
    {
      "epoch": 2.723011288821784,
      "grad_norm": 0.07190598762725503,
      "learning_rate": 9.004212057185094e-05,
      "loss": 0.843,
      "step": 155100
    },
    {
      "epoch": 2.7231868537017854,
      "grad_norm": 0.05507424892635152,
      "learning_rate": 9.003104548108342e-05,
      "loss": 0.8407,
      "step": 155110
    },
    {
      "epoch": 2.723362418581787,
      "grad_norm": 0.05173279103608182,
      "learning_rate": 9.001997052929462e-05,
      "loss": 0.8372,
      "step": 155120
    },
    {
      "epoch": 2.7235379834617883,
      "grad_norm": 0.061055240524716045,
      "learning_rate": 9.000889571662492e-05,
      "loss": 0.8406,
      "step": 155130
    },
    {
      "epoch": 2.72371354834179,
      "grad_norm": 0.043254969432970296,
      "learning_rate": 8.999782104321464e-05,
      "loss": 0.8473,
      "step": 155140
    },
    {
      "epoch": 2.723889113221791,
      "grad_norm": 0.060919102171730195,
      "learning_rate": 8.998674650920406e-05,
      "loss": 0.842,
      "step": 155150
    },
    {
      "epoch": 2.7240646781017928,
      "grad_norm": 0.08242381794614954,
      "learning_rate": 8.997567211473352e-05,
      "loss": 0.8401,
      "step": 155160
    },
    {
      "epoch": 2.724240242981794,
      "grad_norm": 0.0653946971094182,
      "learning_rate": 8.996459785994331e-05,
      "loss": 0.8475,
      "step": 155170
    },
    {
      "epoch": 2.7244158078617953,
      "grad_norm": 0.055461899277198655,
      "learning_rate": 8.995352374497379e-05,
      "loss": 0.8474,
      "step": 155180
    },
    {
      "epoch": 2.7245913727417967,
      "grad_norm": 0.05745037770802562,
      "learning_rate": 8.994244976996518e-05,
      "loss": 0.8461,
      "step": 155190
    },
    {
      "epoch": 2.7247669376217982,
      "grad_norm": 0.05625404812217857,
      "learning_rate": 8.993137593505792e-05,
      "loss": 0.8436,
      "step": 155200
    },
    {
      "epoch": 2.7249425025017997,
      "grad_norm": 0.054623190461124266,
      "learning_rate": 8.992030224039223e-05,
      "loss": 0.8433,
      "step": 155210
    },
    {
      "epoch": 2.7251180673818007,
      "grad_norm": 0.06151421648745013,
      "learning_rate": 8.990922868610839e-05,
      "loss": 0.8524,
      "step": 155220
    },
    {
      "epoch": 2.725293632261802,
      "grad_norm": 0.04503915917187927,
      "learning_rate": 8.989815527234679e-05,
      "loss": 0.84,
      "step": 155230
    },
    {
      "epoch": 2.7254691971418037,
      "grad_norm": 0.05607517504323106,
      "learning_rate": 8.988708199924769e-05,
      "loss": 0.8419,
      "step": 155240
    },
    {
      "epoch": 2.725644762021805,
      "grad_norm": 0.06294600607308953,
      "learning_rate": 8.98760088669514e-05,
      "loss": 0.8427,
      "step": 155250
    },
    {
      "epoch": 2.7258203269018066,
      "grad_norm": 0.05452832958601964,
      "learning_rate": 8.986493587559817e-05,
      "loss": 0.8497,
      "step": 155260
    },
    {
      "epoch": 2.725995891781808,
      "grad_norm": 0.04397536402781362,
      "learning_rate": 8.985386302532837e-05,
      "loss": 0.8503,
      "step": 155270
    },
    {
      "epoch": 2.7261714566618096,
      "grad_norm": 0.05458062418410865,
      "learning_rate": 8.984279031628226e-05,
      "loss": 0.8415,
      "step": 155280
    },
    {
      "epoch": 2.7263470215418106,
      "grad_norm": 0.048099606204296076,
      "learning_rate": 8.98317177486001e-05,
      "loss": 0.8466,
      "step": 155290
    },
    {
      "epoch": 2.726522586421812,
      "grad_norm": 0.07007090732710797,
      "learning_rate": 8.982064532242225e-05,
      "loss": 0.835,
      "step": 155300
    },
    {
      "epoch": 2.7266981513018136,
      "grad_norm": 0.05171950020902155,
      "learning_rate": 8.980957303788896e-05,
      "loss": 0.8416,
      "step": 155310
    },
    {
      "epoch": 2.726873716181815,
      "grad_norm": 0.06161040294120233,
      "learning_rate": 8.979850089514053e-05,
      "loss": 0.8622,
      "step": 155320
    },
    {
      "epoch": 2.7270492810618165,
      "grad_norm": 0.05070599793677813,
      "learning_rate": 8.978742889431722e-05,
      "loss": 0.8421,
      "step": 155330
    },
    {
      "epoch": 2.7272248459418176,
      "grad_norm": 0.048389361413623565,
      "learning_rate": 8.977635703555936e-05,
      "loss": 0.8414,
      "step": 155340
    },
    {
      "epoch": 2.727400410821819,
      "grad_norm": 0.05464806202077278,
      "learning_rate": 8.976528531900721e-05,
      "loss": 0.8466,
      "step": 155350
    },
    {
      "epoch": 2.7275759757018205,
      "grad_norm": 0.05289754476206841,
      "learning_rate": 8.9754213744801e-05,
      "loss": 0.8439,
      "step": 155360
    },
    {
      "epoch": 2.727751540581822,
      "grad_norm": 0.05558533668228626,
      "learning_rate": 8.974314231308112e-05,
      "loss": 0.8498,
      "step": 155370
    },
    {
      "epoch": 2.7279271054618235,
      "grad_norm": 0.04818793051177853,
      "learning_rate": 8.973207102398776e-05,
      "loss": 0.837,
      "step": 155380
    },
    {
      "epoch": 2.728102670341825,
      "grad_norm": 0.0649043875182662,
      "learning_rate": 8.972099987766118e-05,
      "loss": 0.8411,
      "step": 155390
    },
    {
      "epoch": 2.7282782352218264,
      "grad_norm": 0.06483816744848885,
      "learning_rate": 8.970992887424174e-05,
      "loss": 0.8465,
      "step": 155400
    },
    {
      "epoch": 2.7284538001018275,
      "grad_norm": 0.055823270132392,
      "learning_rate": 8.969885801386966e-05,
      "loss": 0.8444,
      "step": 155410
    },
    {
      "epoch": 2.728629364981829,
      "grad_norm": 0.054368723734964285,
      "learning_rate": 8.968778729668523e-05,
      "loss": 0.839,
      "step": 155420
    },
    {
      "epoch": 2.7288049298618304,
      "grad_norm": 0.05663005169252125,
      "learning_rate": 8.967671672282864e-05,
      "loss": 0.8449,
      "step": 155430
    },
    {
      "epoch": 2.728980494741832,
      "grad_norm": 0.06895950734263263,
      "learning_rate": 8.966564629244029e-05,
      "loss": 0.845,
      "step": 155440
    },
    {
      "epoch": 2.7291560596218334,
      "grad_norm": 0.055672556718884526,
      "learning_rate": 8.965457600566034e-05,
      "loss": 0.8445,
      "step": 155450
    },
    {
      "epoch": 2.7293316245018344,
      "grad_norm": 0.08338140227290301,
      "learning_rate": 8.964350586262906e-05,
      "loss": 0.8459,
      "step": 155460
    },
    {
      "epoch": 2.729507189381836,
      "grad_norm": 0.059259240157909156,
      "learning_rate": 8.963243586348678e-05,
      "loss": 0.8458,
      "step": 155470
    },
    {
      "epoch": 2.7296827542618374,
      "grad_norm": 0.042847228616197254,
      "learning_rate": 8.96213660083737e-05,
      "loss": 0.8457,
      "step": 155480
    },
    {
      "epoch": 2.729858319141839,
      "grad_norm": 0.05738965180078845,
      "learning_rate": 8.96102962974301e-05,
      "loss": 0.836,
      "step": 155490
    },
    {
      "epoch": 2.7300338840218403,
      "grad_norm": 0.05370951231062569,
      "learning_rate": 8.959922673079623e-05,
      "loss": 0.837,
      "step": 155500
    },
    {
      "epoch": 2.730209448901842,
      "grad_norm": 0.06208879139771057,
      "learning_rate": 8.958815730861234e-05,
      "loss": 0.8437,
      "step": 155510
    },
    {
      "epoch": 2.7303850137818433,
      "grad_norm": 0.0584387217890538,
      "learning_rate": 8.957708803101869e-05,
      "loss": 0.8373,
      "step": 155520
    },
    {
      "epoch": 2.7305605786618443,
      "grad_norm": 0.04399152521178906,
      "learning_rate": 8.956601889815548e-05,
      "loss": 0.8342,
      "step": 155530
    },
    {
      "epoch": 2.730736143541846,
      "grad_norm": 0.04686451625124202,
      "learning_rate": 8.955494991016306e-05,
      "loss": 0.8476,
      "step": 155540
    },
    {
      "epoch": 2.7309117084218473,
      "grad_norm": 0.046393777717331136,
      "learning_rate": 8.954388106718161e-05,
      "loss": 0.8456,
      "step": 155550
    },
    {
      "epoch": 2.7310872733018488,
      "grad_norm": 0.04912554402379017,
      "learning_rate": 8.953281236935134e-05,
      "loss": 0.84,
      "step": 155560
    },
    {
      "epoch": 2.7312628381818502,
      "grad_norm": 0.053350228866521625,
      "learning_rate": 8.952174381681256e-05,
      "loss": 0.8485,
      "step": 155570
    },
    {
      "epoch": 2.7314384030618513,
      "grad_norm": 0.06533487343853546,
      "learning_rate": 8.951067540970549e-05,
      "loss": 0.8421,
      "step": 155580
    },
    {
      "epoch": 2.731613967941853,
      "grad_norm": 0.0700306932837539,
      "learning_rate": 8.949960714817038e-05,
      "loss": 0.8499,
      "step": 155590
    },
    {
      "epoch": 2.731789532821854,
      "grad_norm": 0.05104891259343752,
      "learning_rate": 8.948853903234742e-05,
      "loss": 0.8305,
      "step": 155600
    },
    {
      "epoch": 2.7319650977018557,
      "grad_norm": 0.047832144949749494,
      "learning_rate": 8.94774710623769e-05,
      "loss": 0.8447,
      "step": 155610
    },
    {
      "epoch": 2.732140662581857,
      "grad_norm": 0.055310994109528805,
      "learning_rate": 8.946640323839904e-05,
      "loss": 0.8409,
      "step": 155620
    },
    {
      "epoch": 2.7323162274618586,
      "grad_norm": 0.07734682866456732,
      "learning_rate": 8.9455335560554e-05,
      "loss": 0.8338,
      "step": 155630
    },
    {
      "epoch": 2.73249179234186,
      "grad_norm": 0.053537042591836254,
      "learning_rate": 8.944426802898211e-05,
      "loss": 0.8433,
      "step": 155640
    },
    {
      "epoch": 2.732667357221861,
      "grad_norm": 0.058533522415928796,
      "learning_rate": 8.943320064382359e-05,
      "loss": 0.8455,
      "step": 155650
    },
    {
      "epoch": 2.7328429221018626,
      "grad_norm": 0.05015474357318129,
      "learning_rate": 8.94221334052186e-05,
      "loss": 0.8456,
      "step": 155660
    },
    {
      "epoch": 2.733018486981864,
      "grad_norm": 0.05569900684171495,
      "learning_rate": 8.941106631330739e-05,
      "loss": 0.8467,
      "step": 155670
    },
    {
      "epoch": 2.7331940518618656,
      "grad_norm": 0.05182997150490459,
      "learning_rate": 8.939999936823022e-05,
      "loss": 0.847,
      "step": 155680
    },
    {
      "epoch": 2.733369616741867,
      "grad_norm": 0.07281966369835097,
      "learning_rate": 8.938893257012729e-05,
      "loss": 0.8502,
      "step": 155690
    },
    {
      "epoch": 2.733545181621868,
      "grad_norm": 0.06346215987883515,
      "learning_rate": 8.937786591913874e-05,
      "loss": 0.8436,
      "step": 155700
    },
    {
      "epoch": 2.73372074650187,
      "grad_norm": 0.049231608894537066,
      "learning_rate": 8.936679941540494e-05,
      "loss": 0.8435,
      "step": 155710
    },
    {
      "epoch": 2.733896311381871,
      "grad_norm": 0.04847594238885205,
      "learning_rate": 8.935573305906598e-05,
      "loss": 0.842,
      "step": 155720
    },
    {
      "epoch": 2.7340718762618725,
      "grad_norm": 0.0809754880959855,
      "learning_rate": 8.93446668502621e-05,
      "loss": 0.8486,
      "step": 155730
    },
    {
      "epoch": 2.734247441141874,
      "grad_norm": 0.05156526571454091,
      "learning_rate": 8.933360078913354e-05,
      "loss": 0.8497,
      "step": 155740
    },
    {
      "epoch": 2.7344230060218755,
      "grad_norm": 0.05607785227017324,
      "learning_rate": 8.932253487582051e-05,
      "loss": 0.8447,
      "step": 155750
    },
    {
      "epoch": 2.734598570901877,
      "grad_norm": 0.05735395948213819,
      "learning_rate": 8.93114691104632e-05,
      "loss": 0.8431,
      "step": 155760
    },
    {
      "epoch": 2.734774135781878,
      "grad_norm": 0.05506475823682268,
      "learning_rate": 8.930040349320178e-05,
      "loss": 0.8423,
      "step": 155770
    },
    {
      "epoch": 2.7349497006618795,
      "grad_norm": 0.0815044412086397,
      "learning_rate": 8.928933802417653e-05,
      "loss": 0.8408,
      "step": 155780
    },
    {
      "epoch": 2.735125265541881,
      "grad_norm": 0.055793140469159834,
      "learning_rate": 8.92782727035276e-05,
      "loss": 0.8442,
      "step": 155790
    },
    {
      "epoch": 2.7353008304218824,
      "grad_norm": 0.052129389497533526,
      "learning_rate": 8.926720753139516e-05,
      "loss": 0.8412,
      "step": 155800
    },
    {
      "epoch": 2.735476395301884,
      "grad_norm": 0.06381819806493876,
      "learning_rate": 8.925614250791949e-05,
      "loss": 0.8441,
      "step": 155810
    },
    {
      "epoch": 2.735651960181885,
      "grad_norm": 0.05829510955535627,
      "learning_rate": 8.924507763324075e-05,
      "loss": 0.8318,
      "step": 155820
    },
    {
      "epoch": 2.735827525061887,
      "grad_norm": 0.06877609919058732,
      "learning_rate": 8.923401290749913e-05,
      "loss": 0.8423,
      "step": 155830
    },
    {
      "epoch": 2.736003089941888,
      "grad_norm": 0.04495651209178678,
      "learning_rate": 8.922294833083479e-05,
      "loss": 0.8472,
      "step": 155840
    },
    {
      "epoch": 2.7361786548218894,
      "grad_norm": 0.05446901243709873,
      "learning_rate": 8.921188390338798e-05,
      "loss": 0.8461,
      "step": 155850
    },
    {
      "epoch": 2.736354219701891,
      "grad_norm": 0.06843397846559066,
      "learning_rate": 8.920081962529888e-05,
      "loss": 0.8423,
      "step": 155860
    },
    {
      "epoch": 2.7365297845818923,
      "grad_norm": 0.054148378863888166,
      "learning_rate": 8.918975549670758e-05,
      "loss": 0.8422,
      "step": 155870
    },
    {
      "epoch": 2.736705349461894,
      "grad_norm": 0.05106995086345871,
      "learning_rate": 8.917869151775443e-05,
      "loss": 0.846,
      "step": 155880
    },
    {
      "epoch": 2.736880914341895,
      "grad_norm": 0.05542072775592845,
      "learning_rate": 8.91676276885795e-05,
      "loss": 0.8459,
      "step": 155890
    },
    {
      "epoch": 2.7370564792218963,
      "grad_norm": 0.05720098583548352,
      "learning_rate": 8.915656400932294e-05,
      "loss": 0.8378,
      "step": 155900
    },
    {
      "epoch": 2.737232044101898,
      "grad_norm": 0.05682918280022831,
      "learning_rate": 8.914550048012504e-05,
      "loss": 0.8408,
      "step": 155910
    },
    {
      "epoch": 2.7374076089818993,
      "grad_norm": 0.0611119123030408,
      "learning_rate": 8.91344371011259e-05,
      "loss": 0.8345,
      "step": 155920
    },
    {
      "epoch": 2.7375831738619008,
      "grad_norm": 0.07237699474408447,
      "learning_rate": 8.912337387246574e-05,
      "loss": 0.8423,
      "step": 155930
    },
    {
      "epoch": 2.7377587387419022,
      "grad_norm": 0.06357118415718814,
      "learning_rate": 8.911231079428467e-05,
      "loss": 0.8408,
      "step": 155940
    },
    {
      "epoch": 2.7379343036219037,
      "grad_norm": 0.06597580049400058,
      "learning_rate": 8.910124786672293e-05,
      "loss": 0.8363,
      "step": 155950
    },
    {
      "epoch": 2.7381098685019047,
      "grad_norm": 0.049843542283602875,
      "learning_rate": 8.909018508992065e-05,
      "loss": 0.8378,
      "step": 155960
    },
    {
      "epoch": 2.738285433381906,
      "grad_norm": 0.05604466003006176,
      "learning_rate": 8.907912246401798e-05,
      "loss": 0.8486,
      "step": 155970
    },
    {
      "epoch": 2.7384609982619077,
      "grad_norm": 0.05751373619544122,
      "learning_rate": 8.906805998915514e-05,
      "loss": 0.8455,
      "step": 155980
    },
    {
      "epoch": 2.738636563141909,
      "grad_norm": 0.054044567011016516,
      "learning_rate": 8.905699766547229e-05,
      "loss": 0.8473,
      "step": 155990
    },
    {
      "epoch": 2.7388121280219107,
      "grad_norm": 0.05910728866014906,
      "learning_rate": 8.904593549310954e-05,
      "loss": 0.8403,
      "step": 156000
    },
    {
      "epoch": 2.7389876929019117,
      "grad_norm": 0.05947345267166262,
      "learning_rate": 8.903487347220708e-05,
      "loss": 0.8399,
      "step": 156010
    },
    {
      "epoch": 2.739163257781913,
      "grad_norm": 0.06426089564717849,
      "learning_rate": 8.902381160290507e-05,
      "loss": 0.8327,
      "step": 156020
    },
    {
      "epoch": 2.7393388226619146,
      "grad_norm": 0.07578030329053079,
      "learning_rate": 8.901274988534369e-05,
      "loss": 0.842,
      "step": 156030
    },
    {
      "epoch": 2.739514387541916,
      "grad_norm": 0.06627332719167481,
      "learning_rate": 8.9001688319663e-05,
      "loss": 0.841,
      "step": 156040
    },
    {
      "epoch": 2.7396899524219176,
      "grad_norm": 0.05394894911142923,
      "learning_rate": 8.899062690600327e-05,
      "loss": 0.8404,
      "step": 156050
    },
    {
      "epoch": 2.739865517301919,
      "grad_norm": 0.05625012983921044,
      "learning_rate": 8.89795656445046e-05,
      "loss": 0.8429,
      "step": 156060
    },
    {
      "epoch": 2.7400410821819206,
      "grad_norm": 0.05269143304418486,
      "learning_rate": 8.896850453530708e-05,
      "loss": 0.847,
      "step": 156070
    },
    {
      "epoch": 2.7402166470619216,
      "grad_norm": 0.06191090381131492,
      "learning_rate": 8.895744357855097e-05,
      "loss": 0.8461,
      "step": 156080
    },
    {
      "epoch": 2.740392211941923,
      "grad_norm": 0.07081879408244118,
      "learning_rate": 8.894638277437636e-05,
      "loss": 0.8443,
      "step": 156090
    },
    {
      "epoch": 2.7405677768219245,
      "grad_norm": 0.0541432659240325,
      "learning_rate": 8.893532212292338e-05,
      "loss": 0.8481,
      "step": 156100
    },
    {
      "epoch": 2.740743341701926,
      "grad_norm": 0.0668375699844316,
      "learning_rate": 8.892426162433216e-05,
      "loss": 0.8465,
      "step": 156110
    },
    {
      "epoch": 2.7409189065819275,
      "grad_norm": 0.047143997207747486,
      "learning_rate": 8.891320127874288e-05,
      "loss": 0.8526,
      "step": 156120
    },
    {
      "epoch": 2.7410944714619285,
      "grad_norm": 0.05284006097686649,
      "learning_rate": 8.890214108629567e-05,
      "loss": 0.8354,
      "step": 156130
    },
    {
      "epoch": 2.74127003634193,
      "grad_norm": 0.04994521888271643,
      "learning_rate": 8.88910810471306e-05,
      "loss": 0.8391,
      "step": 156140
    },
    {
      "epoch": 2.7414456012219315,
      "grad_norm": 0.03830859375189048,
      "learning_rate": 8.88800211613879e-05,
      "loss": 0.8447,
      "step": 156150
    },
    {
      "epoch": 2.741621166101933,
      "grad_norm": 0.05176914136916239,
      "learning_rate": 8.886896142920764e-05,
      "loss": 0.8476,
      "step": 156160
    },
    {
      "epoch": 2.7417967309819344,
      "grad_norm": 0.06959860141322372,
      "learning_rate": 8.885790185072998e-05,
      "loss": 0.837,
      "step": 156170
    },
    {
      "epoch": 2.741972295861936,
      "grad_norm": 0.050606326381222005,
      "learning_rate": 8.884684242609498e-05,
      "loss": 0.8484,
      "step": 156180
    },
    {
      "epoch": 2.7421478607419374,
      "grad_norm": 0.04627675237209581,
      "learning_rate": 8.883578315544288e-05,
      "loss": 0.843,
      "step": 156190
    },
    {
      "epoch": 2.7423234256219384,
      "grad_norm": 0.04876315799200678,
      "learning_rate": 8.882472403891369e-05,
      "loss": 0.842,
      "step": 156200
    },
    {
      "epoch": 2.74249899050194,
      "grad_norm": 0.05987493070261072,
      "learning_rate": 8.88136650766476e-05,
      "loss": 0.8409,
      "step": 156210
    },
    {
      "epoch": 2.7426745553819414,
      "grad_norm": 0.05420106014359952,
      "learning_rate": 8.880260626878472e-05,
      "loss": 0.8502,
      "step": 156220
    },
    {
      "epoch": 2.742850120261943,
      "grad_norm": 0.06001126912379015,
      "learning_rate": 8.879154761546515e-05,
      "loss": 0.8508,
      "step": 156230
    },
    {
      "epoch": 2.7430256851419443,
      "grad_norm": 0.06678362699181668,
      "learning_rate": 8.878048911682896e-05,
      "loss": 0.8398,
      "step": 156240
    },
    {
      "epoch": 2.7432012500219454,
      "grad_norm": 0.07126025953455313,
      "learning_rate": 8.876943077301638e-05,
      "loss": 0.8459,
      "step": 156250
    },
    {
      "epoch": 2.7433768149019473,
      "grad_norm": 0.1056210565719668,
      "learning_rate": 8.875837258416745e-05,
      "loss": 0.8433,
      "step": 156260
    },
    {
      "epoch": 2.7435523797819483,
      "grad_norm": 0.04534477382793864,
      "learning_rate": 8.874731455042227e-05,
      "loss": 0.845,
      "step": 156270
    },
    {
      "epoch": 2.74372794466195,
      "grad_norm": 0.04418312841552144,
      "learning_rate": 8.873625667192095e-05,
      "loss": 0.8475,
      "step": 156280
    },
    {
      "epoch": 2.7439035095419513,
      "grad_norm": 0.05737372577090714,
      "learning_rate": 8.872519894880363e-05,
      "loss": 0.8423,
      "step": 156290
    },
    {
      "epoch": 2.7440790744219528,
      "grad_norm": 0.055396026150413516,
      "learning_rate": 8.87141413812104e-05,
      "loss": 0.8354,
      "step": 156300
    },
    {
      "epoch": 2.7442546393019542,
      "grad_norm": 0.06676750622489885,
      "learning_rate": 8.870308396928131e-05,
      "loss": 0.8437,
      "step": 156310
    },
    {
      "epoch": 2.7444302041819553,
      "grad_norm": 0.05805953865870403,
      "learning_rate": 8.869202671315655e-05,
      "loss": 0.8519,
      "step": 156320
    },
    {
      "epoch": 2.7446057690619567,
      "grad_norm": 0.0514987843601286,
      "learning_rate": 8.868096961297616e-05,
      "loss": 0.8422,
      "step": 156330
    },
    {
      "epoch": 2.7447813339419582,
      "grad_norm": 0.06266746436985629,
      "learning_rate": 8.866991266888022e-05,
      "loss": 0.8368,
      "step": 156340
    },
    {
      "epoch": 2.7449568988219597,
      "grad_norm": 0.047973759082011166,
      "learning_rate": 8.865885588100886e-05,
      "loss": 0.8418,
      "step": 156350
    },
    {
      "epoch": 2.745132463701961,
      "grad_norm": 0.04782082359751911,
      "learning_rate": 8.864779924950219e-05,
      "loss": 0.84,
      "step": 156360
    },
    {
      "epoch": 2.745308028581962,
      "grad_norm": 0.05674484740447128,
      "learning_rate": 8.863674277450024e-05,
      "loss": 0.8456,
      "step": 156370
    },
    {
      "epoch": 2.745483593461964,
      "grad_norm": 0.0569024146546244,
      "learning_rate": 8.862568645614311e-05,
      "loss": 0.8436,
      "step": 156380
    },
    {
      "epoch": 2.745659158341965,
      "grad_norm": 0.05027813097255785,
      "learning_rate": 8.861463029457096e-05,
      "loss": 0.8348,
      "step": 156390
    },
    {
      "epoch": 2.7458347232219666,
      "grad_norm": 0.060073598166479066,
      "learning_rate": 8.86035742899238e-05,
      "loss": 0.8456,
      "step": 156400
    },
    {
      "epoch": 2.746010288101968,
      "grad_norm": 0.0596709515455734,
      "learning_rate": 8.859251844234168e-05,
      "loss": 0.8471,
      "step": 156410
    },
    {
      "epoch": 2.7461858529819696,
      "grad_norm": 0.04947402463879879,
      "learning_rate": 8.858146275196478e-05,
      "loss": 0.8446,
      "step": 156420
    },
    {
      "epoch": 2.746361417861971,
      "grad_norm": 0.06210672041204684,
      "learning_rate": 8.857040721893312e-05,
      "loss": 0.838,
      "step": 156430
    },
    {
      "epoch": 2.746536982741972,
      "grad_norm": 0.052808262221882826,
      "learning_rate": 8.855935184338679e-05,
      "loss": 0.8377,
      "step": 156440
    },
    {
      "epoch": 2.7467125476219736,
      "grad_norm": 0.05880272516152189,
      "learning_rate": 8.854829662546581e-05,
      "loss": 0.8465,
      "step": 156450
    },
    {
      "epoch": 2.746888112501975,
      "grad_norm": 0.052288304081444525,
      "learning_rate": 8.853724156531035e-05,
      "loss": 0.8484,
      "step": 156460
    },
    {
      "epoch": 2.7470636773819765,
      "grad_norm": 0.058251935977457314,
      "learning_rate": 8.852618666306041e-05,
      "loss": 0.8412,
      "step": 156470
    },
    {
      "epoch": 2.747239242261978,
      "grad_norm": 0.053060096089677,
      "learning_rate": 8.851513191885605e-05,
      "loss": 0.8457,
      "step": 156480
    },
    {
      "epoch": 2.747414807141979,
      "grad_norm": 0.06494868513058978,
      "learning_rate": 8.85040773328374e-05,
      "loss": 0.8395,
      "step": 156490
    },
    {
      "epoch": 2.747590372021981,
      "grad_norm": 0.04466637518146537,
      "learning_rate": 8.849302290514448e-05,
      "loss": 0.8411,
      "step": 156500
    },
    {
      "epoch": 2.747765936901982,
      "grad_norm": 0.05370407867000349,
      "learning_rate": 8.848196863591736e-05,
      "loss": 0.8382,
      "step": 156510
    },
    {
      "epoch": 2.7479415017819835,
      "grad_norm": 0.06004801133990986,
      "learning_rate": 8.847091452529607e-05,
      "loss": 0.8451,
      "step": 156520
    },
    {
      "epoch": 2.748117066661985,
      "grad_norm": 0.04445700802535055,
      "learning_rate": 8.845986057342071e-05,
      "loss": 0.8409,
      "step": 156530
    },
    {
      "epoch": 2.7482926315419864,
      "grad_norm": 0.04991079496011218,
      "learning_rate": 8.844880678043132e-05,
      "loss": 0.8421,
      "step": 156540
    },
    {
      "epoch": 2.748468196421988,
      "grad_norm": 0.07338296781473577,
      "learning_rate": 8.843775314646793e-05,
      "loss": 0.8543,
      "step": 156550
    },
    {
      "epoch": 2.748643761301989,
      "grad_norm": 0.04806306409552457,
      "learning_rate": 8.842669967167067e-05,
      "loss": 0.8414,
      "step": 156560
    },
    {
      "epoch": 2.7488193261819904,
      "grad_norm": 0.051368725857169145,
      "learning_rate": 8.84156463561795e-05,
      "loss": 0.849,
      "step": 156570
    },
    {
      "epoch": 2.748994891061992,
      "grad_norm": 0.0470505692629954,
      "learning_rate": 8.840459320013448e-05,
      "loss": 0.8395,
      "step": 156580
    },
    {
      "epoch": 2.7491704559419934,
      "grad_norm": 0.050394502009126846,
      "learning_rate": 8.83935402036757e-05,
      "loss": 0.8435,
      "step": 156590
    },
    {
      "epoch": 2.749346020821995,
      "grad_norm": 0.05864527602745477,
      "learning_rate": 8.83824873669432e-05,
      "loss": 0.8402,
      "step": 156600
    },
    {
      "epoch": 2.749521585701996,
      "grad_norm": 0.04859245680460384,
      "learning_rate": 8.837143469007701e-05,
      "loss": 0.8467,
      "step": 156610
    },
    {
      "epoch": 2.749697150581998,
      "grad_norm": 0.045091030493305234,
      "learning_rate": 8.83603821732171e-05,
      "loss": 0.8425,
      "step": 156620
    },
    {
      "epoch": 2.749872715461999,
      "grad_norm": 0.047902014995531324,
      "learning_rate": 8.834932981650362e-05,
      "loss": 0.8453,
      "step": 156630
    },
    {
      "epoch": 2.7500482803420003,
      "grad_norm": 0.05369132229194304,
      "learning_rate": 8.833827762007657e-05,
      "loss": 0.8397,
      "step": 156640
    },
    {
      "epoch": 2.750223845222002,
      "grad_norm": 0.05334721045150262,
      "learning_rate": 8.83272255840759e-05,
      "loss": 0.8424,
      "step": 156650
    },
    {
      "epoch": 2.7503994101020033,
      "grad_norm": 0.04688574357041273,
      "learning_rate": 8.831617370864176e-05,
      "loss": 0.8401,
      "step": 156660
    },
    {
      "epoch": 2.7505749749820048,
      "grad_norm": 0.08905608579105294,
      "learning_rate": 8.830512199391415e-05,
      "loss": 0.8543,
      "step": 156670
    },
    {
      "epoch": 2.750750539862006,
      "grad_norm": 0.06278228671775822,
      "learning_rate": 8.829407044003304e-05,
      "loss": 0.8422,
      "step": 156680
    },
    {
      "epoch": 2.7509261047420073,
      "grad_norm": 0.047434056972916534,
      "learning_rate": 8.828301904713851e-05,
      "loss": 0.8375,
      "step": 156690
    },
    {
      "epoch": 2.7511016696220087,
      "grad_norm": 0.07986100122579445,
      "learning_rate": 8.827196781537057e-05,
      "loss": 0.8375,
      "step": 156700
    },
    {
      "epoch": 2.7512772345020102,
      "grad_norm": 0.06616294229888202,
      "learning_rate": 8.826091674486924e-05,
      "loss": 0.8452,
      "step": 156710
    },
    {
      "epoch": 2.7514527993820117,
      "grad_norm": 0.057503881866973694,
      "learning_rate": 8.824986583577448e-05,
      "loss": 0.8355,
      "step": 156720
    },
    {
      "epoch": 2.751628364262013,
      "grad_norm": 0.051803196903713045,
      "learning_rate": 8.823881508822645e-05,
      "loss": 0.8392,
      "step": 156730
    },
    {
      "epoch": 2.7518039291420147,
      "grad_norm": 0.05285531775128753,
      "learning_rate": 8.822776450236504e-05,
      "loss": 0.8471,
      "step": 156740
    },
    {
      "epoch": 2.7519794940220157,
      "grad_norm": 0.07578620787157585,
      "learning_rate": 8.821671407833026e-05,
      "loss": 0.845,
      "step": 156750
    },
    {
      "epoch": 2.752155058902017,
      "grad_norm": 0.05122256202584284,
      "learning_rate": 8.820566381626221e-05,
      "loss": 0.8402,
      "step": 156760
    },
    {
      "epoch": 2.7523306237820186,
      "grad_norm": 0.0633493029032213,
      "learning_rate": 8.819461371630086e-05,
      "loss": 0.8456,
      "step": 156770
    },
    {
      "epoch": 2.75250618866202,
      "grad_norm": 0.04622485553071532,
      "learning_rate": 8.81835637785862e-05,
      "loss": 0.8481,
      "step": 156780
    },
    {
      "epoch": 2.7526817535420216,
      "grad_norm": 0.04610272840477818,
      "learning_rate": 8.817251400325822e-05,
      "loss": 0.84,
      "step": 156790
    },
    {
      "epoch": 2.7528573184220226,
      "grad_norm": 0.04906004463388337,
      "learning_rate": 8.816146439045695e-05,
      "loss": 0.8369,
      "step": 156800
    },
    {
      "epoch": 2.753032883302024,
      "grad_norm": 0.06653329032289927,
      "learning_rate": 8.815041494032242e-05,
      "loss": 0.8436,
      "step": 156810
    },
    {
      "epoch": 2.7532084481820256,
      "grad_norm": 0.050302624468321545,
      "learning_rate": 8.813936565299453e-05,
      "loss": 0.8474,
      "step": 156820
    },
    {
      "epoch": 2.753384013062027,
      "grad_norm": 0.051971886103538985,
      "learning_rate": 8.81283165286134e-05,
      "loss": 0.8442,
      "step": 156830
    },
    {
      "epoch": 2.7535595779420285,
      "grad_norm": 0.07088303134988598,
      "learning_rate": 8.811726756731897e-05,
      "loss": 0.8439,
      "step": 156840
    },
    {
      "epoch": 2.75373514282203,
      "grad_norm": 0.06503991474092136,
      "learning_rate": 8.810621876925121e-05,
      "loss": 0.8467,
      "step": 156850
    },
    {
      "epoch": 2.7539107077020315,
      "grad_norm": 0.05244304440489128,
      "learning_rate": 8.809517013455011e-05,
      "loss": 0.8502,
      "step": 156860
    },
    {
      "epoch": 2.7540862725820325,
      "grad_norm": 0.05436677783201485,
      "learning_rate": 8.80841216633557e-05,
      "loss": 0.838,
      "step": 156870
    },
    {
      "epoch": 2.754261837462034,
      "grad_norm": 0.07584943053362546,
      "learning_rate": 8.807307335580795e-05,
      "loss": 0.8452,
      "step": 156880
    },
    {
      "epoch": 2.7544374023420355,
      "grad_norm": 0.05538641589503444,
      "learning_rate": 8.806202521204679e-05,
      "loss": 0.8413,
      "step": 156890
    },
    {
      "epoch": 2.754612967222037,
      "grad_norm": 0.06514015941031055,
      "learning_rate": 8.805097723221232e-05,
      "loss": 0.8413,
      "step": 156900
    },
    {
      "epoch": 2.7547885321020384,
      "grad_norm": 0.06654063200553595,
      "learning_rate": 8.803992941644443e-05,
      "loss": 0.8432,
      "step": 156910
    },
    {
      "epoch": 2.7549640969820395,
      "grad_norm": 0.068562819275101,
      "learning_rate": 8.802888176488307e-05,
      "loss": 0.8491,
      "step": 156920
    },
    {
      "epoch": 2.755139661862041,
      "grad_norm": 0.08178585288875613,
      "learning_rate": 8.80178342776683e-05,
      "loss": 0.8402,
      "step": 156930
    },
    {
      "epoch": 2.7553152267420424,
      "grad_norm": 0.0525003098354295,
      "learning_rate": 8.800678695494008e-05,
      "loss": 0.8472,
      "step": 156940
    },
    {
      "epoch": 2.755490791622044,
      "grad_norm": 0.04093625254925667,
      "learning_rate": 8.799573979683833e-05,
      "loss": 0.8402,
      "step": 156950
    },
    {
      "epoch": 2.7556663565020454,
      "grad_norm": 0.06955406499657853,
      "learning_rate": 8.798469280350306e-05,
      "loss": 0.8436,
      "step": 156960
    },
    {
      "epoch": 2.755841921382047,
      "grad_norm": 0.06647004199709922,
      "learning_rate": 8.797364597507422e-05,
      "loss": 0.8446,
      "step": 156970
    },
    {
      "epoch": 2.7560174862620483,
      "grad_norm": 0.11430824667153548,
      "learning_rate": 8.79625993116918e-05,
      "loss": 0.8476,
      "step": 156980
    },
    {
      "epoch": 2.7561930511420494,
      "grad_norm": 0.07522935996864957,
      "learning_rate": 8.79515528134957e-05,
      "loss": 0.8408,
      "step": 156990
    },
    {
      "epoch": 2.756368616022051,
      "grad_norm": 0.073714792330851,
      "learning_rate": 8.794050648062595e-05,
      "loss": 0.8404,
      "step": 157000
    },
    {
      "epoch": 2.7565441809020523,
      "grad_norm": 0.05562373078891054,
      "learning_rate": 8.792946031322251e-05,
      "loss": 0.847,
      "step": 157010
    },
    {
      "epoch": 2.756719745782054,
      "grad_norm": 0.05551150272161376,
      "learning_rate": 8.79184143114253e-05,
      "loss": 0.8412,
      "step": 157020
    },
    {
      "epoch": 2.7568953106620553,
      "grad_norm": 0.05464405378540042,
      "learning_rate": 8.790736847537427e-05,
      "loss": 0.8406,
      "step": 157030
    },
    {
      "epoch": 2.7570708755420563,
      "grad_norm": 0.04482463921532193,
      "learning_rate": 8.789632280520942e-05,
      "loss": 0.8362,
      "step": 157040
    },
    {
      "epoch": 2.7572464404220582,
      "grad_norm": 0.06976320978903298,
      "learning_rate": 8.788527730107066e-05,
      "loss": 0.8454,
      "step": 157050
    },
    {
      "epoch": 2.7574220053020593,
      "grad_norm": 0.0583763613773101,
      "learning_rate": 8.78742319630979e-05,
      "loss": 0.8437,
      "step": 157060
    },
    {
      "epoch": 2.7575975701820608,
      "grad_norm": 0.04892111168315456,
      "learning_rate": 8.786318679143122e-05,
      "loss": 0.8388,
      "step": 157070
    },
    {
      "epoch": 2.7577731350620622,
      "grad_norm": 0.047853255734508104,
      "learning_rate": 8.785214178621046e-05,
      "loss": 0.8438,
      "step": 157080
    },
    {
      "epoch": 2.7579486999420637,
      "grad_norm": 0.06860882352644158,
      "learning_rate": 8.784109694757553e-05,
      "loss": 0.8439,
      "step": 157090
    },
    {
      "epoch": 2.758124264822065,
      "grad_norm": 0.04504324988952119,
      "learning_rate": 8.783005227566647e-05,
      "loss": 0.8383,
      "step": 157100
    },
    {
      "epoch": 2.758299829702066,
      "grad_norm": 0.053814224615869904,
      "learning_rate": 8.781900777062317e-05,
      "loss": 0.8425,
      "step": 157110
    },
    {
      "epoch": 2.7584753945820677,
      "grad_norm": 0.07253287428485908,
      "learning_rate": 8.780796343258556e-05,
      "loss": 0.8498,
      "step": 157120
    },
    {
      "epoch": 2.758650959462069,
      "grad_norm": 0.05968525914165523,
      "learning_rate": 8.779691926169356e-05,
      "loss": 0.8403,
      "step": 157130
    },
    {
      "epoch": 2.7588265243420707,
      "grad_norm": 0.0669612503730494,
      "learning_rate": 8.778587525808716e-05,
      "loss": 0.8369,
      "step": 157140
    },
    {
      "epoch": 2.759002089222072,
      "grad_norm": 0.0669846889969934,
      "learning_rate": 8.777483142190625e-05,
      "loss": 0.8388,
      "step": 157150
    },
    {
      "epoch": 2.759177654102073,
      "grad_norm": 0.055318561655830184,
      "learning_rate": 8.77637877532907e-05,
      "loss": 0.8492,
      "step": 157160
    },
    {
      "epoch": 2.759353218982075,
      "grad_norm": 0.05446823527924726,
      "learning_rate": 8.775274425238056e-05,
      "loss": 0.8359,
      "step": 157170
    },
    {
      "epoch": 2.759528783862076,
      "grad_norm": 0.05643128989196942,
      "learning_rate": 8.774170091931568e-05,
      "loss": 0.849,
      "step": 157180
    },
    {
      "epoch": 2.7597043487420776,
      "grad_norm": 0.044843689263744775,
      "learning_rate": 8.7730657754236e-05,
      "loss": 0.8426,
      "step": 157190
    },
    {
      "epoch": 2.759879913622079,
      "grad_norm": 0.05628389102890671,
      "learning_rate": 8.771961475728138e-05,
      "loss": 0.8507,
      "step": 157200
    },
    {
      "epoch": 2.7600554785020806,
      "grad_norm": 0.06972056383216203,
      "learning_rate": 8.770857192859183e-05,
      "loss": 0.847,
      "step": 157210
    },
    {
      "epoch": 2.760231043382082,
      "grad_norm": 0.04893008537970542,
      "learning_rate": 8.769752926830722e-05,
      "loss": 0.8432,
      "step": 157220
    },
    {
      "epoch": 2.760406608262083,
      "grad_norm": 0.06530181723082469,
      "learning_rate": 8.768648677656742e-05,
      "loss": 0.8489,
      "step": 157230
    },
    {
      "epoch": 2.7605821731420845,
      "grad_norm": 0.058839535662014816,
      "learning_rate": 8.767544445351245e-05,
      "loss": 0.843,
      "step": 157240
    },
    {
      "epoch": 2.760757738022086,
      "grad_norm": 0.08069468431956432,
      "learning_rate": 8.766440229928212e-05,
      "loss": 0.8556,
      "step": 157250
    },
    {
      "epoch": 2.7609333029020875,
      "grad_norm": 0.0491133514494684,
      "learning_rate": 8.765336031401634e-05,
      "loss": 0.8434,
      "step": 157260
    },
    {
      "epoch": 2.761108867782089,
      "grad_norm": 0.052770297142934804,
      "learning_rate": 8.764231849785507e-05,
      "loss": 0.8442,
      "step": 157270
    },
    {
      "epoch": 2.76128443266209,
      "grad_norm": 0.04449842375501373,
      "learning_rate": 8.76312768509382e-05,
      "loss": 0.8433,
      "step": 157280
    },
    {
      "epoch": 2.761459997542092,
      "grad_norm": 0.060483888501226206,
      "learning_rate": 8.76202353734056e-05,
      "loss": 0.8428,
      "step": 157290
    },
    {
      "epoch": 2.761635562422093,
      "grad_norm": 0.0590747816071487,
      "learning_rate": 8.760919406539717e-05,
      "loss": 0.8501,
      "step": 157300
    },
    {
      "epoch": 2.7618111273020944,
      "grad_norm": 0.06594742633754799,
      "learning_rate": 8.759815292705284e-05,
      "loss": 0.8402,
      "step": 157310
    },
    {
      "epoch": 2.761986692182096,
      "grad_norm": 0.07461783835545312,
      "learning_rate": 8.758711195851248e-05,
      "loss": 0.8476,
      "step": 157320
    },
    {
      "epoch": 2.7621622570620974,
      "grad_norm": 0.05364486655179432,
      "learning_rate": 8.757607115991593e-05,
      "loss": 0.8359,
      "step": 157330
    },
    {
      "epoch": 2.762337821942099,
      "grad_norm": 0.05024527387530762,
      "learning_rate": 8.756503053140319e-05,
      "loss": 0.8483,
      "step": 157340
    },
    {
      "epoch": 2.7625133868221,
      "grad_norm": 0.052916735615289964,
      "learning_rate": 8.755399007311408e-05,
      "loss": 0.8359,
      "step": 157350
    },
    {
      "epoch": 2.7626889517021014,
      "grad_norm": 0.04926095648069745,
      "learning_rate": 8.754294978518848e-05,
      "loss": 0.8428,
      "step": 157360
    },
    {
      "epoch": 2.762864516582103,
      "grad_norm": 0.050011426679076196,
      "learning_rate": 8.753190966776628e-05,
      "loss": 0.8442,
      "step": 157370
    },
    {
      "epoch": 2.7630400814621043,
      "grad_norm": 0.05024618282544578,
      "learning_rate": 8.752086972098738e-05,
      "loss": 0.8437,
      "step": 157380
    },
    {
      "epoch": 2.763215646342106,
      "grad_norm": 0.06277252324111474,
      "learning_rate": 8.750982994499166e-05,
      "loss": 0.847,
      "step": 157390
    },
    {
      "epoch": 2.7633912112221073,
      "grad_norm": 0.05170288840330667,
      "learning_rate": 8.749879033991893e-05,
      "loss": 0.8341,
      "step": 157400
    },
    {
      "epoch": 2.7635667761021088,
      "grad_norm": 0.05377042858591191,
      "learning_rate": 8.748775090590916e-05,
      "loss": 0.8407,
      "step": 157410
    },
    {
      "epoch": 2.76374234098211,
      "grad_norm": 0.08202363213430908,
      "learning_rate": 8.747671164310215e-05,
      "loss": 0.8434,
      "step": 157420
    },
    {
      "epoch": 2.7639179058621113,
      "grad_norm": 0.04649054107016494,
      "learning_rate": 8.746567255163778e-05,
      "loss": 0.8488,
      "step": 157430
    },
    {
      "epoch": 2.7640934707421128,
      "grad_norm": 0.0605360250064372,
      "learning_rate": 8.745463363165597e-05,
      "loss": 0.8428,
      "step": 157440
    },
    {
      "epoch": 2.7642690356221142,
      "grad_norm": 0.07152242211938314,
      "learning_rate": 8.744359488329654e-05,
      "loss": 0.8411,
      "step": 157450
    },
    {
      "epoch": 2.7644446005021157,
      "grad_norm": 0.05281659293114526,
      "learning_rate": 8.743255630669937e-05,
      "loss": 0.8421,
      "step": 157460
    },
    {
      "epoch": 2.7646201653821167,
      "grad_norm": 0.05494862875335562,
      "learning_rate": 8.742151790200428e-05,
      "loss": 0.841,
      "step": 157470
    },
    {
      "epoch": 2.7647957302621182,
      "grad_norm": 0.06206497033131458,
      "learning_rate": 8.741047966935119e-05,
      "loss": 0.8491,
      "step": 157480
    },
    {
      "epoch": 2.7649712951421197,
      "grad_norm": 0.07548319024096652,
      "learning_rate": 8.739944160887991e-05,
      "loss": 0.8473,
      "step": 157490
    },
    {
      "epoch": 2.765146860022121,
      "grad_norm": 0.057634523609308166,
      "learning_rate": 8.738840372073028e-05,
      "loss": 0.8477,
      "step": 157500
    },
    {
      "epoch": 2.7653224249021227,
      "grad_norm": 0.04532186700354208,
      "learning_rate": 8.737736600504223e-05,
      "loss": 0.8392,
      "step": 157510
    },
    {
      "epoch": 2.765497989782124,
      "grad_norm": 0.06886100925686674,
      "learning_rate": 8.736632846195555e-05,
      "loss": 0.8386,
      "step": 157520
    },
    {
      "epoch": 2.7656735546621256,
      "grad_norm": 0.04977441439542064,
      "learning_rate": 8.735529109161012e-05,
      "loss": 0.8501,
      "step": 157530
    },
    {
      "epoch": 2.7658491195421266,
      "grad_norm": 0.07685475688445494,
      "learning_rate": 8.734425389414573e-05,
      "loss": 0.8402,
      "step": 157540
    },
    {
      "epoch": 2.766024684422128,
      "grad_norm": 0.05175423923634387,
      "learning_rate": 8.733321686970229e-05,
      "loss": 0.8388,
      "step": 157550
    },
    {
      "epoch": 2.7662002493021296,
      "grad_norm": 0.10521635639912774,
      "learning_rate": 8.732218001841963e-05,
      "loss": 0.8296,
      "step": 157560
    },
    {
      "epoch": 2.766375814182131,
      "grad_norm": 0.053390976242546635,
      "learning_rate": 8.73111433404375e-05,
      "loss": 0.844,
      "step": 157570
    },
    {
      "epoch": 2.7665513790621326,
      "grad_norm": 0.0565352904652092,
      "learning_rate": 8.730010683589588e-05,
      "loss": 0.8497,
      "step": 157580
    },
    {
      "epoch": 2.7667269439421336,
      "grad_norm": 0.06787607211849983,
      "learning_rate": 8.728907050493452e-05,
      "loss": 0.8488,
      "step": 157590
    },
    {
      "epoch": 2.766902508822135,
      "grad_norm": 0.04916681965352802,
      "learning_rate": 8.72780343476932e-05,
      "loss": 0.8376,
      "step": 157600
    },
    {
      "epoch": 2.7670780737021365,
      "grad_norm": 0.05663280667059593,
      "learning_rate": 8.726699836431186e-05,
      "loss": 0.8393,
      "step": 157610
    },
    {
      "epoch": 2.767253638582138,
      "grad_norm": 0.05316591462715608,
      "learning_rate": 8.725596255493031e-05,
      "loss": 0.8396,
      "step": 157620
    },
    {
      "epoch": 2.7674292034621395,
      "grad_norm": 0.05387106125260006,
      "learning_rate": 8.724492691968834e-05,
      "loss": 0.8463,
      "step": 157630
    },
    {
      "epoch": 2.767604768342141,
      "grad_norm": 0.04306219140455374,
      "learning_rate": 8.723389145872572e-05,
      "loss": 0.8434,
      "step": 157640
    },
    {
      "epoch": 2.7677803332221425,
      "grad_norm": 0.054590018581044994,
      "learning_rate": 8.722285617218239e-05,
      "loss": 0.8438,
      "step": 157650
    },
    {
      "epoch": 2.7679558981021435,
      "grad_norm": 0.04035791602368389,
      "learning_rate": 8.721182106019812e-05,
      "loss": 0.84,
      "step": 157660
    },
    {
      "epoch": 2.768131462982145,
      "grad_norm": 0.048682997848577644,
      "learning_rate": 8.720078612291266e-05,
      "loss": 0.8329,
      "step": 157670
    },
    {
      "epoch": 2.7683070278621464,
      "grad_norm": 0.05092098623382221,
      "learning_rate": 8.718975136046593e-05,
      "loss": 0.8346,
      "step": 157680
    },
    {
      "epoch": 2.768482592742148,
      "grad_norm": 0.07207592967554345,
      "learning_rate": 8.717871677299771e-05,
      "loss": 0.8388,
      "step": 157690
    },
    {
      "epoch": 2.7686581576221494,
      "grad_norm": 0.0560342206135456,
      "learning_rate": 8.716768236064778e-05,
      "loss": 0.8358,
      "step": 157700
    },
    {
      "epoch": 2.7688337225021504,
      "grad_norm": 0.06303645839217051,
      "learning_rate": 8.715664812355595e-05,
      "loss": 0.8419,
      "step": 157710
    },
    {
      "epoch": 2.7690092873821524,
      "grad_norm": 0.04731197077963036,
      "learning_rate": 8.714561406186206e-05,
      "loss": 0.8441,
      "step": 157720
    },
    {
      "epoch": 2.7691848522621534,
      "grad_norm": 0.057166184081497706,
      "learning_rate": 8.713458017570591e-05,
      "loss": 0.8487,
      "step": 157730
    },
    {
      "epoch": 2.769360417142155,
      "grad_norm": 0.04614596029528185,
      "learning_rate": 8.712354646522722e-05,
      "loss": 0.8493,
      "step": 157740
    },
    {
      "epoch": 2.7695359820221563,
      "grad_norm": 0.05682703239556557,
      "learning_rate": 8.711251293056594e-05,
      "loss": 0.843,
      "step": 157750
    },
    {
      "epoch": 2.769711546902158,
      "grad_norm": 0.0638358810620907,
      "learning_rate": 8.710147957186176e-05,
      "loss": 0.8332,
      "step": 157760
    },
    {
      "epoch": 2.7698871117821593,
      "grad_norm": 0.06717455383836266,
      "learning_rate": 8.709044638925443e-05,
      "loss": 0.8433,
      "step": 157770
    },
    {
      "epoch": 2.7700626766621603,
      "grad_norm": 0.06178642902593861,
      "learning_rate": 8.707941338288388e-05,
      "loss": 0.843,
      "step": 157780
    },
    {
      "epoch": 2.770238241542162,
      "grad_norm": 0.05202276328891303,
      "learning_rate": 8.706838055288983e-05,
      "loss": 0.8389,
      "step": 157790
    },
    {
      "epoch": 2.7704138064221633,
      "grad_norm": 0.06118544534581905,
      "learning_rate": 8.705734789941205e-05,
      "loss": 0.8421,
      "step": 157800
    },
    {
      "epoch": 2.7705893713021648,
      "grad_norm": 0.0479883874439823,
      "learning_rate": 8.704631542259033e-05,
      "loss": 0.8507,
      "step": 157810
    },
    {
      "epoch": 2.7707649361821662,
      "grad_norm": 0.06252791121061436,
      "learning_rate": 8.70352831225645e-05,
      "loss": 0.847,
      "step": 157820
    },
    {
      "epoch": 2.7709405010621673,
      "grad_norm": 0.06868476888808146,
      "learning_rate": 8.70242509994743e-05,
      "loss": 0.843,
      "step": 157830
    },
    {
      "epoch": 2.771116065942169,
      "grad_norm": 0.06370421750646033,
      "learning_rate": 8.701321905345947e-05,
      "loss": 0.8463,
      "step": 157840
    },
    {
      "epoch": 2.7712916308221702,
      "grad_norm": 0.0483724568502102,
      "learning_rate": 8.700218728465989e-05,
      "loss": 0.8379,
      "step": 157850
    },
    {
      "epoch": 2.7714671957021717,
      "grad_norm": 0.04462636167297381,
      "learning_rate": 8.699115569321526e-05,
      "loss": 0.837,
      "step": 157860
    },
    {
      "epoch": 2.771642760582173,
      "grad_norm": 0.060898790616022776,
      "learning_rate": 8.69801242792654e-05,
      "loss": 0.8424,
      "step": 157870
    },
    {
      "epoch": 2.7718183254621747,
      "grad_norm": 0.06995987591454805,
      "learning_rate": 8.696909304295001e-05,
      "loss": 0.8429,
      "step": 157880
    },
    {
      "epoch": 2.771993890342176,
      "grad_norm": 0.057444182390312466,
      "learning_rate": 8.695806198440893e-05,
      "loss": 0.8394,
      "step": 157890
    },
    {
      "epoch": 2.772169455222177,
      "grad_norm": 0.04450752316775776,
      "learning_rate": 8.694703110378191e-05,
      "loss": 0.8425,
      "step": 157900
    },
    {
      "epoch": 2.7723450201021786,
      "grad_norm": 0.05448558014265505,
      "learning_rate": 8.693600040120863e-05,
      "loss": 0.8397,
      "step": 157910
    },
    {
      "epoch": 2.77252058498218,
      "grad_norm": 0.04644179176854391,
      "learning_rate": 8.6924969876829e-05,
      "loss": 0.8433,
      "step": 157920
    },
    {
      "epoch": 2.7726961498621816,
      "grad_norm": 0.062290138087552085,
      "learning_rate": 8.691393953078267e-05,
      "loss": 0.8425,
      "step": 157930
    },
    {
      "epoch": 2.772871714742183,
      "grad_norm": 0.05636090416898908,
      "learning_rate": 8.690290936320937e-05,
      "loss": 0.8481,
      "step": 157940
    },
    {
      "epoch": 2.773047279622184,
      "grad_norm": 0.05149563146390951,
      "learning_rate": 8.689187937424899e-05,
      "loss": 0.8479,
      "step": 157950
    },
    {
      "epoch": 2.773222844502186,
      "grad_norm": 0.04347969033298756,
      "learning_rate": 8.688084956404117e-05,
      "loss": 0.8382,
      "step": 157960
    },
    {
      "epoch": 2.773398409382187,
      "grad_norm": 0.05143316475387085,
      "learning_rate": 8.686981993272572e-05,
      "loss": 0.8414,
      "step": 157970
    },
    {
      "epoch": 2.7735739742621885,
      "grad_norm": 0.05669151626306484,
      "learning_rate": 8.68587904804423e-05,
      "loss": 0.8444,
      "step": 157980
    },
    {
      "epoch": 2.77374953914219,
      "grad_norm": 0.05396448463268302,
      "learning_rate": 8.684776120733076e-05,
      "loss": 0.8429,
      "step": 157990
    },
    {
      "epoch": 2.7739251040221915,
      "grad_norm": 0.04641172818122232,
      "learning_rate": 8.683673211353079e-05,
      "loss": 0.8538,
      "step": 158000
    },
    {
      "epoch": 2.774100668902193,
      "grad_norm": 0.04572721519495913,
      "learning_rate": 8.682570319918211e-05,
      "loss": 0.841,
      "step": 158010
    },
    {
      "epoch": 2.774276233782194,
      "grad_norm": 0.07184679260116335,
      "learning_rate": 8.681467446442452e-05,
      "loss": 0.8437,
      "step": 158020
    },
    {
      "epoch": 2.7744517986621955,
      "grad_norm": 0.049338966478491036,
      "learning_rate": 8.680364590939772e-05,
      "loss": 0.8438,
      "step": 158030
    },
    {
      "epoch": 2.774627363542197,
      "grad_norm": 0.0572081364549409,
      "learning_rate": 8.679261753424145e-05,
      "loss": 0.8456,
      "step": 158040
    },
    {
      "epoch": 2.7748029284221984,
      "grad_norm": 0.04951340304904494,
      "learning_rate": 8.678158933909541e-05,
      "loss": 0.8462,
      "step": 158050
    },
    {
      "epoch": 2.7749784933022,
      "grad_norm": 0.05335713292404227,
      "learning_rate": 8.67705613240994e-05,
      "loss": 0.8401,
      "step": 158060
    },
    {
      "epoch": 2.775154058182201,
      "grad_norm": 0.07799416115840617,
      "learning_rate": 8.675953348939308e-05,
      "loss": 0.8363,
      "step": 158070
    },
    {
      "epoch": 2.775329623062203,
      "grad_norm": 0.05062751986449493,
      "learning_rate": 8.674850583511617e-05,
      "loss": 0.8399,
      "step": 158080
    },
    {
      "epoch": 2.775505187942204,
      "grad_norm": 0.04947107079062952,
      "learning_rate": 8.67374783614085e-05,
      "loss": 0.8391,
      "step": 158090
    },
    {
      "epoch": 2.7756807528222054,
      "grad_norm": 0.06667438723927128,
      "learning_rate": 8.672645106840967e-05,
      "loss": 0.8386,
      "step": 158100
    },
    {
      "epoch": 2.775856317702207,
      "grad_norm": 0.06208108880711767,
      "learning_rate": 8.671542395625941e-05,
      "loss": 0.8459,
      "step": 158110
    },
    {
      "epoch": 2.7760318825822083,
      "grad_norm": 0.048090918426966334,
      "learning_rate": 8.670439702509749e-05,
      "loss": 0.8495,
      "step": 158120
    },
    {
      "epoch": 2.77620744746221,
      "grad_norm": 0.07152244259270604,
      "learning_rate": 8.669337027506363e-05,
      "loss": 0.8371,
      "step": 158130
    },
    {
      "epoch": 2.776383012342211,
      "grad_norm": 0.06166699841861759,
      "learning_rate": 8.668234370629748e-05,
      "loss": 0.847,
      "step": 158140
    },
    {
      "epoch": 2.7765585772222123,
      "grad_norm": 0.04878371050190883,
      "learning_rate": 8.667131731893877e-05,
      "loss": 0.8383,
      "step": 158150
    },
    {
      "epoch": 2.776734142102214,
      "grad_norm": 0.06049321938653896,
      "learning_rate": 8.666029111312724e-05,
      "loss": 0.8447,
      "step": 158160
    },
    {
      "epoch": 2.7769097069822153,
      "grad_norm": 0.04687975523956496,
      "learning_rate": 8.664926508900258e-05,
      "loss": 0.8453,
      "step": 158170
    },
    {
      "epoch": 2.7770852718622168,
      "grad_norm": 0.044803268807973244,
      "learning_rate": 8.663823924670442e-05,
      "loss": 0.8432,
      "step": 158180
    },
    {
      "epoch": 2.7772608367422182,
      "grad_norm": 0.060195486639835205,
      "learning_rate": 8.662721358637257e-05,
      "loss": 0.8502,
      "step": 158190
    },
    {
      "epoch": 2.7774364016222197,
      "grad_norm": 0.04596803590194419,
      "learning_rate": 8.661618810814668e-05,
      "loss": 0.8445,
      "step": 158200
    },
    {
      "epoch": 2.7776119665022208,
      "grad_norm": 0.06729081329942413,
      "learning_rate": 8.660516281216645e-05,
      "loss": 0.8378,
      "step": 158210
    },
    {
      "epoch": 2.7777875313822222,
      "grad_norm": 0.04696980261365667,
      "learning_rate": 8.659413769857153e-05,
      "loss": 0.8496,
      "step": 158220
    },
    {
      "epoch": 2.7779630962622237,
      "grad_norm": 0.05131835651328704,
      "learning_rate": 8.658311276750167e-05,
      "loss": 0.8442,
      "step": 158230
    },
    {
      "epoch": 2.778138661142225,
      "grad_norm": 0.058271109449735604,
      "learning_rate": 8.657208801909654e-05,
      "loss": 0.8388,
      "step": 158240
    },
    {
      "epoch": 2.7783142260222267,
      "grad_norm": 0.04635449966942686,
      "learning_rate": 8.656106345349577e-05,
      "loss": 0.8386,
      "step": 158250
    },
    {
      "epoch": 2.7784897909022277,
      "grad_norm": 0.056211429093765475,
      "learning_rate": 8.655003907083918e-05,
      "loss": 0.8469,
      "step": 158260
    },
    {
      "epoch": 2.778665355782229,
      "grad_norm": 0.04893357825554613,
      "learning_rate": 8.653901487126632e-05,
      "loss": 0.8465,
      "step": 158270
    },
    {
      "epoch": 2.7788409206622307,
      "grad_norm": 0.05198798882084262,
      "learning_rate": 8.652799085491692e-05,
      "loss": 0.8496,
      "step": 158280
    },
    {
      "epoch": 2.779016485542232,
      "grad_norm": 0.06326205860011798,
      "learning_rate": 8.65169670219306e-05,
      "loss": 0.8486,
      "step": 158290
    },
    {
      "epoch": 2.7791920504222336,
      "grad_norm": 0.11329436860621157,
      "learning_rate": 8.650594337244713e-05,
      "loss": 0.8427,
      "step": 158300
    },
    {
      "epoch": 2.779367615302235,
      "grad_norm": 0.043525768791299324,
      "learning_rate": 8.649491990660615e-05,
      "loss": 0.8361,
      "step": 158310
    },
    {
      "epoch": 2.7795431801822366,
      "grad_norm": 0.06493938779800938,
      "learning_rate": 8.648389662454728e-05,
      "loss": 0.8521,
      "step": 158320
    },
    {
      "epoch": 2.7797187450622376,
      "grad_norm": 0.06690131278318882,
      "learning_rate": 8.647287352641023e-05,
      "loss": 0.8448,
      "step": 158330
    },
    {
      "epoch": 2.779894309942239,
      "grad_norm": 0.0525095478718427,
      "learning_rate": 8.646185061233469e-05,
      "loss": 0.844,
      "step": 158340
    },
    {
      "epoch": 2.7800698748222405,
      "grad_norm": 0.0639953191584054,
      "learning_rate": 8.645082788246022e-05,
      "loss": 0.8403,
      "step": 158350
    },
    {
      "epoch": 2.780245439702242,
      "grad_norm": 0.054570928249641645,
      "learning_rate": 8.64398053369266e-05,
      "loss": 0.8453,
      "step": 158360
    },
    {
      "epoch": 2.7804210045822435,
      "grad_norm": 0.048841201085192125,
      "learning_rate": 8.642878297587343e-05,
      "loss": 0.8419,
      "step": 158370
    },
    {
      "epoch": 2.7805965694622445,
      "grad_norm": 0.10443432113410675,
      "learning_rate": 8.641776079944038e-05,
      "loss": 0.8407,
      "step": 158380
    },
    {
      "epoch": 2.780772134342246,
      "grad_norm": 0.05549723210141879,
      "learning_rate": 8.640673880776708e-05,
      "loss": 0.8478,
      "step": 158390
    },
    {
      "epoch": 2.7809476992222475,
      "grad_norm": 0.05519627587518345,
      "learning_rate": 8.639571700099321e-05,
      "loss": 0.8459,
      "step": 158400
    },
    {
      "epoch": 2.781123264102249,
      "grad_norm": 0.05143219693949454,
      "learning_rate": 8.63846953792584e-05,
      "loss": 0.8367,
      "step": 158410
    },
    {
      "epoch": 2.7812988289822504,
      "grad_norm": 0.0449010734983184,
      "learning_rate": 8.637367394270226e-05,
      "loss": 0.8395,
      "step": 158420
    },
    {
      "epoch": 2.781474393862252,
      "grad_norm": 0.0477941032902872,
      "learning_rate": 8.636265269146454e-05,
      "loss": 0.8408,
      "step": 158430
    },
    {
      "epoch": 2.7816499587422534,
      "grad_norm": 0.05579431535867657,
      "learning_rate": 8.635163162568478e-05,
      "loss": 0.8456,
      "step": 158440
    },
    {
      "epoch": 2.7818255236222544,
      "grad_norm": 0.05455066420252598,
      "learning_rate": 8.634061074550268e-05,
      "loss": 0.8365,
      "step": 158450
    },
    {
      "epoch": 2.782001088502256,
      "grad_norm": 0.04363808512331947,
      "learning_rate": 8.63295900510578e-05,
      "loss": 0.8501,
      "step": 158460
    },
    {
      "epoch": 2.7821766533822574,
      "grad_norm": 0.05616773680683007,
      "learning_rate": 8.631856954248985e-05,
      "loss": 0.8409,
      "step": 158470
    },
    {
      "epoch": 2.782352218262259,
      "grad_norm": 0.06686073303482128,
      "learning_rate": 8.630754921993846e-05,
      "loss": 0.8372,
      "step": 158480
    },
    {
      "epoch": 2.7825277831422603,
      "grad_norm": 0.05069042331209733,
      "learning_rate": 8.629652908354321e-05,
      "loss": 0.8375,
      "step": 158490
    },
    {
      "epoch": 2.7827033480222614,
      "grad_norm": 0.057597028543256976,
      "learning_rate": 8.628550913344378e-05,
      "loss": 0.8405,
      "step": 158500
    },
    {
      "epoch": 2.7828789129022633,
      "grad_norm": 0.07186710768314743,
      "learning_rate": 8.627448936977976e-05,
      "loss": 0.8423,
      "step": 158510
    },
    {
      "epoch": 2.7830544777822643,
      "grad_norm": 0.047845336660489984,
      "learning_rate": 8.626346979269075e-05,
      "loss": 0.8371,
      "step": 158520
    },
    {
      "epoch": 2.783230042662266,
      "grad_norm": 0.05201955868370502,
      "learning_rate": 8.625245040231644e-05,
      "loss": 0.8405,
      "step": 158530
    },
    {
      "epoch": 2.7834056075422673,
      "grad_norm": 0.044581402378895144,
      "learning_rate": 8.624143119879641e-05,
      "loss": 0.8416,
      "step": 158540
    },
    {
      "epoch": 2.7835811724222688,
      "grad_norm": 0.06382787286265847,
      "learning_rate": 8.623041218227027e-05,
      "loss": 0.8419,
      "step": 158550
    },
    {
      "epoch": 2.7837567373022702,
      "grad_norm": 0.0488743912975959,
      "learning_rate": 8.621939335287762e-05,
      "loss": 0.8362,
      "step": 158560
    },
    {
      "epoch": 2.7839323021822713,
      "grad_norm": 0.05093683996168448,
      "learning_rate": 8.620837471075811e-05,
      "loss": 0.8532,
      "step": 158570
    },
    {
      "epoch": 2.7841078670622728,
      "grad_norm": 0.06218276900639767,
      "learning_rate": 8.619735625605133e-05,
      "loss": 0.8448,
      "step": 158580
    },
    {
      "epoch": 2.7842834319422742,
      "grad_norm": 0.04854696670169906,
      "learning_rate": 8.618633798889683e-05,
      "loss": 0.8472,
      "step": 158590
    },
    {
      "epoch": 2.7844589968222757,
      "grad_norm": 0.04605396199715163,
      "learning_rate": 8.617531990943433e-05,
      "loss": 0.8416,
      "step": 158600
    },
    {
      "epoch": 2.784634561702277,
      "grad_norm": 0.07836526005636074,
      "learning_rate": 8.616430201780337e-05,
      "loss": 0.8432,
      "step": 158610
    },
    {
      "epoch": 2.7848101265822782,
      "grad_norm": 0.05268586318815114,
      "learning_rate": 8.615328431414352e-05,
      "loss": 0.8424,
      "step": 158620
    },
    {
      "epoch": 2.78498569146228,
      "grad_norm": 0.06865943463411125,
      "learning_rate": 8.614226679859438e-05,
      "loss": 0.8417,
      "step": 158630
    },
    {
      "epoch": 2.785161256342281,
      "grad_norm": 0.04725660904191454,
      "learning_rate": 8.613124947129561e-05,
      "loss": 0.8419,
      "step": 158640
    },
    {
      "epoch": 2.7853368212222827,
      "grad_norm": 0.05352157562236789,
      "learning_rate": 8.612023233238674e-05,
      "loss": 0.84,
      "step": 158650
    },
    {
      "epoch": 2.785512386102284,
      "grad_norm": 0.055172886636973166,
      "learning_rate": 8.610921538200737e-05,
      "loss": 0.8375,
      "step": 158660
    },
    {
      "epoch": 2.7856879509822856,
      "grad_norm": 0.04920089970144094,
      "learning_rate": 8.60981986202971e-05,
      "loss": 0.8471,
      "step": 158670
    },
    {
      "epoch": 2.785863515862287,
      "grad_norm": 0.06024649074027634,
      "learning_rate": 8.608718204739554e-05,
      "loss": 0.8404,
      "step": 158680
    },
    {
      "epoch": 2.786039080742288,
      "grad_norm": 0.06036381666778544,
      "learning_rate": 8.607616566344217e-05,
      "loss": 0.8472,
      "step": 158690
    },
    {
      "epoch": 2.7862146456222896,
      "grad_norm": 0.05807791863694737,
      "learning_rate": 8.606514946857671e-05,
      "loss": 0.8496,
      "step": 158700
    },
    {
      "epoch": 2.786390210502291,
      "grad_norm": 0.051859188052313925,
      "learning_rate": 8.605413346293864e-05,
      "loss": 0.8422,
      "step": 158710
    },
    {
      "epoch": 2.7865657753822926,
      "grad_norm": 0.05603217176067365,
      "learning_rate": 8.604311764666758e-05,
      "loss": 0.8451,
      "step": 158720
    },
    {
      "epoch": 2.786741340262294,
      "grad_norm": 0.04973074231568203,
      "learning_rate": 8.603210201990306e-05,
      "loss": 0.8426,
      "step": 158730
    },
    {
      "epoch": 2.786916905142295,
      "grad_norm": 0.06104351631040734,
      "learning_rate": 8.60210865827847e-05,
      "loss": 0.8436,
      "step": 158740
    },
    {
      "epoch": 2.787092470022297,
      "grad_norm": 0.05970773404756986,
      "learning_rate": 8.601007133545204e-05,
      "loss": 0.8494,
      "step": 158750
    },
    {
      "epoch": 2.787268034902298,
      "grad_norm": 0.05296430330061749,
      "learning_rate": 8.599905627804462e-05,
      "loss": 0.8463,
      "step": 158760
    },
    {
      "epoch": 2.7874435997822995,
      "grad_norm": 0.059176164419269456,
      "learning_rate": 8.598804141070207e-05,
      "loss": 0.8431,
      "step": 158770
    },
    {
      "epoch": 2.787619164662301,
      "grad_norm": 0.06591104176025231,
      "learning_rate": 8.597702673356391e-05,
      "loss": 0.8388,
      "step": 158780
    },
    {
      "epoch": 2.7877947295423025,
      "grad_norm": 0.04986061205472818,
      "learning_rate": 8.596601224676972e-05,
      "loss": 0.8426,
      "step": 158790
    },
    {
      "epoch": 2.787970294422304,
      "grad_norm": 0.06610617950020466,
      "learning_rate": 8.595499795045896e-05,
      "loss": 0.8395,
      "step": 158800
    },
    {
      "epoch": 2.788145859302305,
      "grad_norm": 0.05314232025389221,
      "learning_rate": 8.594398384477132e-05,
      "loss": 0.8478,
      "step": 158810
    },
    {
      "epoch": 2.7883214241823064,
      "grad_norm": 0.04653120537964994,
      "learning_rate": 8.593296992984627e-05,
      "loss": 0.8349,
      "step": 158820
    },
    {
      "epoch": 2.788496989062308,
      "grad_norm": 0.05706595515811312,
      "learning_rate": 8.592195620582336e-05,
      "loss": 0.8446,
      "step": 158830
    },
    {
      "epoch": 2.7886725539423094,
      "grad_norm": 0.05553761594945306,
      "learning_rate": 8.591094267284219e-05,
      "loss": 0.8357,
      "step": 158840
    },
    {
      "epoch": 2.788848118822311,
      "grad_norm": 0.05860022512185123,
      "learning_rate": 8.589992933104227e-05,
      "loss": 0.8394,
      "step": 158850
    },
    {
      "epoch": 2.7890236837023124,
      "grad_norm": 0.049281395507239485,
      "learning_rate": 8.58889161805631e-05,
      "loss": 0.8442,
      "step": 158860
    },
    {
      "epoch": 2.789199248582314,
      "grad_norm": 0.04880175818854498,
      "learning_rate": 8.587790322154429e-05,
      "loss": 0.8386,
      "step": 158870
    },
    {
      "epoch": 2.789374813462315,
      "grad_norm": 0.06105512038976036,
      "learning_rate": 8.586689045412533e-05,
      "loss": 0.8419,
      "step": 158880
    },
    {
      "epoch": 2.7895503783423163,
      "grad_norm": 0.055084473067207415,
      "learning_rate": 8.585587787844578e-05,
      "loss": 0.8394,
      "step": 158890
    },
    {
      "epoch": 2.789725943222318,
      "grad_norm": 0.06160682739639107,
      "learning_rate": 8.584486549464515e-05,
      "loss": 0.8409,
      "step": 158900
    },
    {
      "epoch": 2.7899015081023193,
      "grad_norm": 0.05128942998158076,
      "learning_rate": 8.583385330286298e-05,
      "loss": 0.8395,
      "step": 158910
    },
    {
      "epoch": 2.7900770729823208,
      "grad_norm": 0.07787642794463664,
      "learning_rate": 8.582284130323882e-05,
      "loss": 0.8419,
      "step": 158920
    },
    {
      "epoch": 2.790252637862322,
      "grad_norm": 0.05207960146624838,
      "learning_rate": 8.581182949591211e-05,
      "loss": 0.8468,
      "step": 158930
    },
    {
      "epoch": 2.7904282027423233,
      "grad_norm": 0.05407801650758095,
      "learning_rate": 8.580081788102251e-05,
      "loss": 0.8426,
      "step": 158940
    },
    {
      "epoch": 2.7906037676223248,
      "grad_norm": 0.05350693271373002,
      "learning_rate": 8.578980645870942e-05,
      "loss": 0.8363,
      "step": 158950
    },
    {
      "epoch": 2.7907793325023262,
      "grad_norm": 0.0521973306336239,
      "learning_rate": 8.577879522911241e-05,
      "loss": 0.8475,
      "step": 158960
    },
    {
      "epoch": 2.7909548973823277,
      "grad_norm": 0.050346091585339016,
      "learning_rate": 8.576778419237093e-05,
      "loss": 0.8485,
      "step": 158970
    },
    {
      "epoch": 2.791130462262329,
      "grad_norm": 0.05300812578281723,
      "learning_rate": 8.57567733486246e-05,
      "loss": 0.8409,
      "step": 158980
    },
    {
      "epoch": 2.7913060271423307,
      "grad_norm": 0.05532956063875565,
      "learning_rate": 8.574576269801286e-05,
      "loss": 0.84,
      "step": 158990
    },
    {
      "epoch": 2.7914815920223317,
      "grad_norm": 0.052333398373000356,
      "learning_rate": 8.57347522406752e-05,
      "loss": 0.8435,
      "step": 159000
    },
    {
      "epoch": 2.791657156902333,
      "grad_norm": 0.042597584738477426,
      "learning_rate": 8.572374197675119e-05,
      "loss": 0.8383,
      "step": 159010
    },
    {
      "epoch": 2.7918327217823347,
      "grad_norm": 0.044470674889723394,
      "learning_rate": 8.57127319063803e-05,
      "loss": 0.8425,
      "step": 159020
    },
    {
      "epoch": 2.792008286662336,
      "grad_norm": 0.06084240265873605,
      "learning_rate": 8.570172202970198e-05,
      "loss": 0.837,
      "step": 159030
    },
    {
      "epoch": 2.7921838515423376,
      "grad_norm": 0.05085402855286349,
      "learning_rate": 8.569071234685582e-05,
      "loss": 0.8427,
      "step": 159040
    },
    {
      "epoch": 2.7923594164223386,
      "grad_norm": 0.05624618692124495,
      "learning_rate": 8.567970285798125e-05,
      "loss": 0.8474,
      "step": 159050
    },
    {
      "epoch": 2.79253498130234,
      "grad_norm": 0.05093056436742468,
      "learning_rate": 8.566869356321781e-05,
      "loss": 0.8442,
      "step": 159060
    },
    {
      "epoch": 2.7927105461823416,
      "grad_norm": 0.05958375153675825,
      "learning_rate": 8.565768446270491e-05,
      "loss": 0.8417,
      "step": 159070
    },
    {
      "epoch": 2.792886111062343,
      "grad_norm": 0.06138728443283651,
      "learning_rate": 8.564667555658214e-05,
      "loss": 0.8466,
      "step": 159080
    },
    {
      "epoch": 2.7930616759423446,
      "grad_norm": 0.0554213553814544,
      "learning_rate": 8.56356668449889e-05,
      "loss": 0.8375,
      "step": 159090
    },
    {
      "epoch": 2.793237240822346,
      "grad_norm": 0.05128291532401258,
      "learning_rate": 8.562465832806469e-05,
      "loss": 0.8467,
      "step": 159100
    },
    {
      "epoch": 2.7934128057023475,
      "grad_norm": 0.06604643071122562,
      "learning_rate": 8.561365000594906e-05,
      "loss": 0.8363,
      "step": 159110
    },
    {
      "epoch": 2.7935883705823485,
      "grad_norm": 0.051331133465648916,
      "learning_rate": 8.560264187878141e-05,
      "loss": 0.8471,
      "step": 159120
    },
    {
      "epoch": 2.79376393546235,
      "grad_norm": 0.05278182983648806,
      "learning_rate": 8.559163394670125e-05,
      "loss": 0.8415,
      "step": 159130
    },
    {
      "epoch": 2.7939395003423515,
      "grad_norm": 0.08589667770226175,
      "learning_rate": 8.558062620984798e-05,
      "loss": 0.8382,
      "step": 159140
    },
    {
      "epoch": 2.794115065222353,
      "grad_norm": 0.04716796796982661,
      "learning_rate": 8.556961866836118e-05,
      "loss": 0.8409,
      "step": 159150
    },
    {
      "epoch": 2.7942906301023545,
      "grad_norm": 0.05081963981492891,
      "learning_rate": 8.555861132238026e-05,
      "loss": 0.8383,
      "step": 159160
    },
    {
      "epoch": 2.7944661949823555,
      "grad_norm": 0.05022615085480669,
      "learning_rate": 8.554760417204468e-05,
      "loss": 0.8449,
      "step": 159170
    },
    {
      "epoch": 2.7946417598623574,
      "grad_norm": 0.07193317320925817,
      "learning_rate": 8.553659721749394e-05,
      "loss": 0.8345,
      "step": 159180
    },
    {
      "epoch": 2.7948173247423584,
      "grad_norm": 0.051188041024636935,
      "learning_rate": 8.552559045886749e-05,
      "loss": 0.8309,
      "step": 159190
    },
    {
      "epoch": 2.79499288962236,
      "grad_norm": 0.05771250632597885,
      "learning_rate": 8.551458389630471e-05,
      "loss": 0.8429,
      "step": 159200
    },
    {
      "epoch": 2.7951684545023614,
      "grad_norm": 0.049492463621484434,
      "learning_rate": 8.550357752994517e-05,
      "loss": 0.8454,
      "step": 159210
    },
    {
      "epoch": 2.795344019382363,
      "grad_norm": 0.05395801458060741,
      "learning_rate": 8.549257135992827e-05,
      "loss": 0.8426,
      "step": 159220
    },
    {
      "epoch": 2.7955195842623644,
      "grad_norm": 0.05515268161196609,
      "learning_rate": 8.548156538639347e-05,
      "loss": 0.8428,
      "step": 159230
    },
    {
      "epoch": 2.7956951491423654,
      "grad_norm": 0.0717171683345448,
      "learning_rate": 8.54705596094802e-05,
      "loss": 0.8401,
      "step": 159240
    },
    {
      "epoch": 2.795870714022367,
      "grad_norm": 0.052781533002429,
      "learning_rate": 8.545955402932792e-05,
      "loss": 0.851,
      "step": 159250
    },
    {
      "epoch": 2.7960462789023683,
      "grad_norm": 0.0549593198434756,
      "learning_rate": 8.544854864607607e-05,
      "loss": 0.8407,
      "step": 159260
    },
    {
      "epoch": 2.79622184378237,
      "grad_norm": 0.05355986374197193,
      "learning_rate": 8.543754345986405e-05,
      "loss": 0.8455,
      "step": 159270
    },
    {
      "epoch": 2.7963974086623713,
      "grad_norm": 0.04388339799737148,
      "learning_rate": 8.542653847083139e-05,
      "loss": 0.8397,
      "step": 159280
    },
    {
      "epoch": 2.7965729735423723,
      "grad_norm": 0.053226627880271445,
      "learning_rate": 8.541553367911748e-05,
      "loss": 0.8494,
      "step": 159290
    },
    {
      "epoch": 2.7967485384223743,
      "grad_norm": 0.0494812997689779,
      "learning_rate": 8.540452908486172e-05,
      "loss": 0.8472,
      "step": 159300
    },
    {
      "epoch": 2.7969241033023753,
      "grad_norm": 0.07640198276218552,
      "learning_rate": 8.539352468820352e-05,
      "loss": 0.8372,
      "step": 159310
    },
    {
      "epoch": 2.7970996681823768,
      "grad_norm": 0.04713261500811805,
      "learning_rate": 8.53825204892824e-05,
      "loss": 0.8469,
      "step": 159320
    },
    {
      "epoch": 2.7972752330623782,
      "grad_norm": 0.07569580350041524,
      "learning_rate": 8.537151648823775e-05,
      "loss": 0.8392,
      "step": 159330
    },
    {
      "epoch": 2.7974507979423797,
      "grad_norm": 0.05107531632923652,
      "learning_rate": 8.536051268520895e-05,
      "loss": 0.8496,
      "step": 159340
    },
    {
      "epoch": 2.797626362822381,
      "grad_norm": 0.04918029626961618,
      "learning_rate": 8.534950908033548e-05,
      "loss": 0.8413,
      "step": 159350
    },
    {
      "epoch": 2.7978019277023822,
      "grad_norm": 0.05055961600921949,
      "learning_rate": 8.533850567375672e-05,
      "loss": 0.8463,
      "step": 159360
    },
    {
      "epoch": 2.7979774925823837,
      "grad_norm": 0.05119570020531404,
      "learning_rate": 8.532750246561206e-05,
      "loss": 0.8483,
      "step": 159370
    },
    {
      "epoch": 2.798153057462385,
      "grad_norm": 0.056599074651058304,
      "learning_rate": 8.531649945604099e-05,
      "loss": 0.8493,
      "step": 159380
    },
    {
      "epoch": 2.7983286223423867,
      "grad_norm": 0.050220836376803406,
      "learning_rate": 8.530549664518288e-05,
      "loss": 0.8315,
      "step": 159390
    },
    {
      "epoch": 2.798504187222388,
      "grad_norm": 0.06863333639428916,
      "learning_rate": 8.529449403317713e-05,
      "loss": 0.8414,
      "step": 159400
    },
    {
      "epoch": 2.798679752102389,
      "grad_norm": 0.0952199321845148,
      "learning_rate": 8.528349162016312e-05,
      "loss": 0.8438,
      "step": 159410
    },
    {
      "epoch": 2.798855316982391,
      "grad_norm": 0.05864172388304522,
      "learning_rate": 8.527248940628032e-05,
      "loss": 0.8427,
      "step": 159420
    },
    {
      "epoch": 2.799030881862392,
      "grad_norm": 0.07012663186760426,
      "learning_rate": 8.526148739166808e-05,
      "loss": 0.8494,
      "step": 159430
    },
    {
      "epoch": 2.7992064467423936,
      "grad_norm": 0.04946192938762748,
      "learning_rate": 8.525048557646578e-05,
      "loss": 0.8362,
      "step": 159440
    },
    {
      "epoch": 2.799382011622395,
      "grad_norm": 0.04955313043542959,
      "learning_rate": 8.52394839608129e-05,
      "loss": 0.8437,
      "step": 159450
    },
    {
      "epoch": 2.7995575765023966,
      "grad_norm": 0.05789980965811333,
      "learning_rate": 8.522848254484877e-05,
      "loss": 0.8431,
      "step": 159460
    },
    {
      "epoch": 2.799733141382398,
      "grad_norm": 0.05677707531491286,
      "learning_rate": 8.521748132871278e-05,
      "loss": 0.84,
      "step": 159470
    },
    {
      "epoch": 2.799908706262399,
      "grad_norm": 0.05874476914172145,
      "learning_rate": 8.520648031254428e-05,
      "loss": 0.8476,
      "step": 159480
    },
    {
      "epoch": 2.8000842711424005,
      "grad_norm": 0.05211770171996852,
      "learning_rate": 8.519547949648276e-05,
      "loss": 0.8424,
      "step": 159490
    },
    {
      "epoch": 2.800259836022402,
      "grad_norm": 0.0472693846047477,
      "learning_rate": 8.518447888066752e-05,
      "loss": 0.8394,
      "step": 159500
    },
    {
      "epoch": 2.8004354009024035,
      "grad_norm": 0.05930359456210888,
      "learning_rate": 8.517347846523795e-05,
      "loss": 0.8474,
      "step": 159510
    },
    {
      "epoch": 2.800610965782405,
      "grad_norm": 0.0661985779369035,
      "learning_rate": 8.516247825033348e-05,
      "loss": 0.8503,
      "step": 159520
    },
    {
      "epoch": 2.800786530662406,
      "grad_norm": 0.07132274988183425,
      "learning_rate": 8.515147823609342e-05,
      "loss": 0.8428,
      "step": 159530
    },
    {
      "epoch": 2.800962095542408,
      "grad_norm": 0.07163047668333795,
      "learning_rate": 8.514047842265714e-05,
      "loss": 0.8431,
      "step": 159540
    },
    {
      "epoch": 2.801137660422409,
      "grad_norm": 0.056944741211752484,
      "learning_rate": 8.512947881016407e-05,
      "loss": 0.8416,
      "step": 159550
    },
    {
      "epoch": 2.8013132253024104,
      "grad_norm": 0.048927663520118035,
      "learning_rate": 8.511847939875357e-05,
      "loss": 0.8505,
      "step": 159560
    },
    {
      "epoch": 2.801488790182412,
      "grad_norm": 0.05575729156511879,
      "learning_rate": 8.510748018856496e-05,
      "loss": 0.8391,
      "step": 159570
    },
    {
      "epoch": 2.8016643550624134,
      "grad_norm": 0.06199330793792168,
      "learning_rate": 8.50964811797376e-05,
      "loss": 0.841,
      "step": 159580
    },
    {
      "epoch": 2.801839919942415,
      "grad_norm": 0.05684838541316369,
      "learning_rate": 8.508548237241091e-05,
      "loss": 0.849,
      "step": 159590
    },
    {
      "epoch": 2.802015484822416,
      "grad_norm": 0.08586333865159287,
      "learning_rate": 8.50744837667242e-05,
      "loss": 0.8395,
      "step": 159600
    },
    {
      "epoch": 2.8021910497024174,
      "grad_norm": 0.057259831093895035,
      "learning_rate": 8.506348536281677e-05,
      "loss": 0.8456,
      "step": 159610
    },
    {
      "epoch": 2.802366614582419,
      "grad_norm": 0.1657633234467731,
      "learning_rate": 8.505248716082811e-05,
      "loss": 0.8301,
      "step": 159620
    },
    {
      "epoch": 2.8025421794624203,
      "grad_norm": 0.048835635706618696,
      "learning_rate": 8.50414891608975e-05,
      "loss": 0.8489,
      "step": 159630
    },
    {
      "epoch": 2.802717744342422,
      "grad_norm": 0.05605573010585747,
      "learning_rate": 8.503049136316425e-05,
      "loss": 0.8451,
      "step": 159640
    },
    {
      "epoch": 2.8028933092224233,
      "grad_norm": 0.04641556803912904,
      "learning_rate": 8.50194937677677e-05,
      "loss": 0.8347,
      "step": 159650
    },
    {
      "epoch": 2.803068874102425,
      "grad_norm": 0.04618997714988377,
      "learning_rate": 8.500849637484727e-05,
      "loss": 0.8313,
      "step": 159660
    },
    {
      "epoch": 2.803244438982426,
      "grad_norm": 0.05050587143416418,
      "learning_rate": 8.499749918454225e-05,
      "loss": 0.8467,
      "step": 159670
    },
    {
      "epoch": 2.8034200038624273,
      "grad_norm": 0.06033682875467223,
      "learning_rate": 8.498650219699197e-05,
      "loss": 0.8477,
      "step": 159680
    },
    {
      "epoch": 2.8035955687424288,
      "grad_norm": 0.05144665609336715,
      "learning_rate": 8.497550541233578e-05,
      "loss": 0.8526,
      "step": 159690
    },
    {
      "epoch": 2.8037711336224302,
      "grad_norm": 0.07491240359445099,
      "learning_rate": 8.496450883071302e-05,
      "loss": 0.8469,
      "step": 159700
    },
    {
      "epoch": 2.8039466985024317,
      "grad_norm": 0.07489346537225301,
      "learning_rate": 8.495351245226297e-05,
      "loss": 0.8449,
      "step": 159710
    },
    {
      "epoch": 2.8041222633824328,
      "grad_norm": 0.058331713247399795,
      "learning_rate": 8.494251627712503e-05,
      "loss": 0.8526,
      "step": 159720
    },
    {
      "epoch": 2.8042978282624342,
      "grad_norm": 0.06254350588900517,
      "learning_rate": 8.493152030543848e-05,
      "loss": 0.8465,
      "step": 159730
    },
    {
      "epoch": 2.8044733931424357,
      "grad_norm": 0.05309013627845153,
      "learning_rate": 8.492052453734266e-05,
      "loss": 0.8423,
      "step": 159740
    },
    {
      "epoch": 2.804648958022437,
      "grad_norm": 0.042902265312203355,
      "learning_rate": 8.490952897297685e-05,
      "loss": 0.8387,
      "step": 159750
    },
    {
      "epoch": 2.8048245229024387,
      "grad_norm": 0.04501293243603132,
      "learning_rate": 8.48985336124804e-05,
      "loss": 0.8434,
      "step": 159760
    },
    {
      "epoch": 2.80500008778244,
      "grad_norm": 0.05562700824700469,
      "learning_rate": 8.488753845599263e-05,
      "loss": 0.8427,
      "step": 159770
    },
    {
      "epoch": 2.8051756526624416,
      "grad_norm": 0.04717435511897762,
      "learning_rate": 8.48765435036528e-05,
      "loss": 0.852,
      "step": 159780
    },
    {
      "epoch": 2.8053512175424427,
      "grad_norm": 0.07038818041343443,
      "learning_rate": 8.486554875560031e-05,
      "loss": 0.8436,
      "step": 159790
    },
    {
      "epoch": 2.805526782422444,
      "grad_norm": 0.06692863432221277,
      "learning_rate": 8.485455421197438e-05,
      "loss": 0.8381,
      "step": 159800
    },
    {
      "epoch": 2.8057023473024456,
      "grad_norm": 0.05204373705291223,
      "learning_rate": 8.484355987291435e-05,
      "loss": 0.8425,
      "step": 159810
    },
    {
      "epoch": 2.805877912182447,
      "grad_norm": 0.04236628705575725,
      "learning_rate": 8.483256573855946e-05,
      "loss": 0.8467,
      "step": 159820
    },
    {
      "epoch": 2.8060534770624486,
      "grad_norm": 0.06488416632436568,
      "learning_rate": 8.48215718090491e-05,
      "loss": 0.8507,
      "step": 159830
    },
    {
      "epoch": 2.8062290419424496,
      "grad_norm": 0.05642276851051412,
      "learning_rate": 8.481057808452254e-05,
      "loss": 0.8467,
      "step": 159840
    },
    {
      "epoch": 2.806404606822451,
      "grad_norm": 0.04980231602289356,
      "learning_rate": 8.479958456511903e-05,
      "loss": 0.8353,
      "step": 159850
    },
    {
      "epoch": 2.8065801717024526,
      "grad_norm": 0.09598246136587105,
      "learning_rate": 8.478859125097792e-05,
      "loss": 0.8384,
      "step": 159860
    },
    {
      "epoch": 2.806755736582454,
      "grad_norm": 0.05564655975606287,
      "learning_rate": 8.477759814223846e-05,
      "loss": 0.8269,
      "step": 159870
    },
    {
      "epoch": 2.8069313014624555,
      "grad_norm": 0.061936184111871566,
      "learning_rate": 8.47666052390399e-05,
      "loss": 0.8447,
      "step": 159880
    },
    {
      "epoch": 2.807106866342457,
      "grad_norm": 0.05362869037056806,
      "learning_rate": 8.47556125415216e-05,
      "loss": 0.841,
      "step": 159890
    },
    {
      "epoch": 2.8072824312224585,
      "grad_norm": 0.04205825136860857,
      "learning_rate": 8.474462004982279e-05,
      "loss": 0.8373,
      "step": 159900
    },
    {
      "epoch": 2.8074579961024595,
      "grad_norm": 0.0539424428029717,
      "learning_rate": 8.473362776408279e-05,
      "loss": 0.8445,
      "step": 159910
    },
    {
      "epoch": 2.807633560982461,
      "grad_norm": 0.06256823357144665,
      "learning_rate": 8.47226356844408e-05,
      "loss": 0.8436,
      "step": 159920
    },
    {
      "epoch": 2.8078091258624625,
      "grad_norm": 0.053094663794082224,
      "learning_rate": 8.471164381103616e-05,
      "loss": 0.8448,
      "step": 159930
    },
    {
      "epoch": 2.807984690742464,
      "grad_norm": 0.052275358924130134,
      "learning_rate": 8.470065214400812e-05,
      "loss": 0.836,
      "step": 159940
    },
    {
      "epoch": 2.8081602556224654,
      "grad_norm": 0.055402022163586015,
      "learning_rate": 8.468966068349588e-05,
      "loss": 0.856,
      "step": 159950
    },
    {
      "epoch": 2.8083358205024664,
      "grad_norm": 0.056982509726870884,
      "learning_rate": 8.467866942963886e-05,
      "loss": 0.8425,
      "step": 159960
    },
    {
      "epoch": 2.8085113853824684,
      "grad_norm": 0.06113078799450122,
      "learning_rate": 8.466767838257618e-05,
      "loss": 0.8403,
      "step": 159970
    },
    {
      "epoch": 2.8086869502624694,
      "grad_norm": 0.0559447521449295,
      "learning_rate": 8.465668754244715e-05,
      "loss": 0.8428,
      "step": 159980
    },
    {
      "epoch": 2.808862515142471,
      "grad_norm": 0.062118545851345595,
      "learning_rate": 8.4645696909391e-05,
      "loss": 0.8382,
      "step": 159990
    },
    {
      "epoch": 2.8090380800224724,
      "grad_norm": 0.04044311778708643,
      "learning_rate": 8.463470648354703e-05,
      "loss": 0.8439,
      "step": 160000
    },
    {
      "epoch": 2.809213644902474,
      "grad_norm": 0.06728042689674879,
      "learning_rate": 8.462371626505446e-05,
      "loss": 0.8409,
      "step": 160010
    },
    {
      "epoch": 2.8093892097824753,
      "grad_norm": 0.06076037449401969,
      "learning_rate": 8.461272625405252e-05,
      "loss": 0.8518,
      "step": 160020
    },
    {
      "epoch": 2.8095647746624763,
      "grad_norm": 0.0556081199606742,
      "learning_rate": 8.460173645068051e-05,
      "loss": 0.8402,
      "step": 160030
    },
    {
      "epoch": 2.809740339542478,
      "grad_norm": 0.056883802526134015,
      "learning_rate": 8.459074685507765e-05,
      "loss": 0.8416,
      "step": 160040
    },
    {
      "epoch": 2.8099159044224793,
      "grad_norm": 0.08659864001329541,
      "learning_rate": 8.457975746738312e-05,
      "loss": 0.8391,
      "step": 160050
    },
    {
      "epoch": 2.8100914693024808,
      "grad_norm": 0.04711294224348655,
      "learning_rate": 8.456876828773624e-05,
      "loss": 0.8427,
      "step": 160060
    },
    {
      "epoch": 2.8102670341824822,
      "grad_norm": 0.04987986496807564,
      "learning_rate": 8.455777931627622e-05,
      "loss": 0.8435,
      "step": 160070
    },
    {
      "epoch": 2.8104425990624833,
      "grad_norm": 0.06643118471009697,
      "learning_rate": 8.454679055314229e-05,
      "loss": 0.8443,
      "step": 160080
    },
    {
      "epoch": 2.810618163942485,
      "grad_norm": 0.06026034238001293,
      "learning_rate": 8.453580199847366e-05,
      "loss": 0.8421,
      "step": 160090
    },
    {
      "epoch": 2.8107937288224862,
      "grad_norm": 0.05086473232537924,
      "learning_rate": 8.45248136524096e-05,
      "loss": 0.8497,
      "step": 160100
    },
    {
      "epoch": 2.8109692937024877,
      "grad_norm": 0.04317276033245278,
      "learning_rate": 8.45138255150893e-05,
      "loss": 0.8492,
      "step": 160110
    },
    {
      "epoch": 2.811144858582489,
      "grad_norm": 0.062194531812848244,
      "learning_rate": 8.450283758665194e-05,
      "loss": 0.8328,
      "step": 160120
    },
    {
      "epoch": 2.8113204234624907,
      "grad_norm": 0.049008940008715654,
      "learning_rate": 8.449184986723689e-05,
      "loss": 0.8327,
      "step": 160130
    },
    {
      "epoch": 2.811495988342492,
      "grad_norm": 0.05845368572181331,
      "learning_rate": 8.448086235698322e-05,
      "loss": 0.8464,
      "step": 160140
    },
    {
      "epoch": 2.811671553222493,
      "grad_norm": 0.046765418424337374,
      "learning_rate": 8.446987505603018e-05,
      "loss": 0.8509,
      "step": 160150
    },
    {
      "epoch": 2.8118471181024947,
      "grad_norm": 0.0525753534101565,
      "learning_rate": 8.445888796451695e-05,
      "loss": 0.8373,
      "step": 160160
    },
    {
      "epoch": 2.812022682982496,
      "grad_norm": 0.046712304923504346,
      "learning_rate": 8.444790108258283e-05,
      "loss": 0.8472,
      "step": 160170
    },
    {
      "epoch": 2.8121982478624976,
      "grad_norm": 0.05638136878664301,
      "learning_rate": 8.443691441036698e-05,
      "loss": 0.8474,
      "step": 160180
    },
    {
      "epoch": 2.812373812742499,
      "grad_norm": 0.0484868188051071,
      "learning_rate": 8.442592794800856e-05,
      "loss": 0.8451,
      "step": 160190
    },
    {
      "epoch": 2.8125493776225,
      "grad_norm": 0.0575641705197477,
      "learning_rate": 8.441494169564686e-05,
      "loss": 0.834,
      "step": 160200
    },
    {
      "epoch": 2.812724942502502,
      "grad_norm": 0.0745061800363056,
      "learning_rate": 8.440395565342099e-05,
      "loss": 0.8451,
      "step": 160210
    },
    {
      "epoch": 2.812900507382503,
      "grad_norm": 0.06997591147753346,
      "learning_rate": 8.439296982147016e-05,
      "loss": 0.8437,
      "step": 160220
    },
    {
      "epoch": 2.8130760722625046,
      "grad_norm": 0.04496761981868282,
      "learning_rate": 8.438198419993365e-05,
      "loss": 0.8355,
      "step": 160230
    },
    {
      "epoch": 2.813251637142506,
      "grad_norm": 0.041772632620062866,
      "learning_rate": 8.437099878895055e-05,
      "loss": 0.8443,
      "step": 160240
    },
    {
      "epoch": 2.8134272020225075,
      "grad_norm": 0.051967683876474394,
      "learning_rate": 8.436001358866009e-05,
      "loss": 0.8404,
      "step": 160250
    },
    {
      "epoch": 2.813602766902509,
      "grad_norm": 0.04373002151529893,
      "learning_rate": 8.434902859920143e-05,
      "loss": 0.846,
      "step": 160260
    },
    {
      "epoch": 2.81377833178251,
      "grad_norm": 0.059664531802383486,
      "learning_rate": 8.433804382071379e-05,
      "loss": 0.8422,
      "step": 160270
    },
    {
      "epoch": 2.8139538966625115,
      "grad_norm": 0.055966461313222846,
      "learning_rate": 8.432705925333634e-05,
      "loss": 0.841,
      "step": 160280
    },
    {
      "epoch": 2.814129461542513,
      "grad_norm": 0.11555489421721625,
      "learning_rate": 8.431607489720817e-05,
      "loss": 0.8306,
      "step": 160290
    },
    {
      "epoch": 2.8143050264225145,
      "grad_norm": 0.050466665874079795,
      "learning_rate": 8.430509075246862e-05,
      "loss": 0.84,
      "step": 160300
    },
    {
      "epoch": 2.814480591302516,
      "grad_norm": 0.053472563371376354,
      "learning_rate": 8.429410681925674e-05,
      "loss": 0.8426,
      "step": 160310
    },
    {
      "epoch": 2.8146561561825174,
      "grad_norm": 0.0570417131838305,
      "learning_rate": 8.428312309771173e-05,
      "loss": 0.8409,
      "step": 160320
    },
    {
      "epoch": 2.814831721062519,
      "grad_norm": 0.07247368502504879,
      "learning_rate": 8.427213958797271e-05,
      "loss": 0.8361,
      "step": 160330
    },
    {
      "epoch": 2.81500728594252,
      "grad_norm": 0.059485401650077596,
      "learning_rate": 8.426115629017893e-05,
      "loss": 0.8367,
      "step": 160340
    },
    {
      "epoch": 2.8151828508225214,
      "grad_norm": 0.04759677880549446,
      "learning_rate": 8.425017320446951e-05,
      "loss": 0.8443,
      "step": 160350
    },
    {
      "epoch": 2.815358415702523,
      "grad_norm": 0.0558809618167103,
      "learning_rate": 8.423919033098358e-05,
      "loss": 0.8419,
      "step": 160360
    },
    {
      "epoch": 2.8155339805825244,
      "grad_norm": 0.050405992100773664,
      "learning_rate": 8.422820766986035e-05,
      "loss": 0.8421,
      "step": 160370
    },
    {
      "epoch": 2.815709545462526,
      "grad_norm": 0.05105944068409853,
      "learning_rate": 8.421722522123894e-05,
      "loss": 0.8427,
      "step": 160380
    },
    {
      "epoch": 2.815885110342527,
      "grad_norm": 0.051126352641093366,
      "learning_rate": 8.420624298525846e-05,
      "loss": 0.8429,
      "step": 160390
    },
    {
      "epoch": 2.8160606752225283,
      "grad_norm": 0.06597960371656657,
      "learning_rate": 8.419526096205815e-05,
      "loss": 0.8428,
      "step": 160400
    },
    {
      "epoch": 2.81623624010253,
      "grad_norm": 0.05903723913568439,
      "learning_rate": 8.418427915177708e-05,
      "loss": 0.8458,
      "step": 160410
    },
    {
      "epoch": 2.8164118049825313,
      "grad_norm": 0.04878496246932129,
      "learning_rate": 8.417329755455444e-05,
      "loss": 0.8572,
      "step": 160420
    },
    {
      "epoch": 2.8165873698625328,
      "grad_norm": 0.05781724038648827,
      "learning_rate": 8.41623161705293e-05,
      "loss": 0.8342,
      "step": 160430
    },
    {
      "epoch": 2.8167629347425343,
      "grad_norm": 0.06317435376389822,
      "learning_rate": 8.415133499984088e-05,
      "loss": 0.8467,
      "step": 160440
    },
    {
      "epoch": 2.8169384996225357,
      "grad_norm": 0.04822215327793413,
      "learning_rate": 8.414035404262827e-05,
      "loss": 0.8415,
      "step": 160450
    },
    {
      "epoch": 2.8171140645025368,
      "grad_norm": 0.06179800779636427,
      "learning_rate": 8.412937329903057e-05,
      "loss": 0.8394,
      "step": 160460
    },
    {
      "epoch": 2.8172896293825382,
      "grad_norm": 0.06918561973666568,
      "learning_rate": 8.4118392769187e-05,
      "loss": 0.8354,
      "step": 160470
    },
    {
      "epoch": 2.8174651942625397,
      "grad_norm": 0.050660734829735814,
      "learning_rate": 8.41074124532366e-05,
      "loss": 0.8402,
      "step": 160480
    },
    {
      "epoch": 2.817640759142541,
      "grad_norm": 0.04159047899303103,
      "learning_rate": 8.409643235131854e-05,
      "loss": 0.8346,
      "step": 160490
    },
    {
      "epoch": 2.8178163240225427,
      "grad_norm": 0.05672225986691627,
      "learning_rate": 8.408545246357185e-05,
      "loss": 0.8403,
      "step": 160500
    },
    {
      "epoch": 2.8179918889025437,
      "grad_norm": 0.05574511987506748,
      "learning_rate": 8.40744727901358e-05,
      "loss": 0.8385,
      "step": 160510
    },
    {
      "epoch": 2.818167453782545,
      "grad_norm": 0.0688718831689664,
      "learning_rate": 8.406349333114941e-05,
      "loss": 0.8402,
      "step": 160520
    },
    {
      "epoch": 2.8183430186625467,
      "grad_norm": 0.045608467189816974,
      "learning_rate": 8.405251408675177e-05,
      "loss": 0.8379,
      "step": 160530
    },
    {
      "epoch": 2.818518583542548,
      "grad_norm": 0.05085523997679157,
      "learning_rate": 8.404153505708206e-05,
      "loss": 0.846,
      "step": 160540
    },
    {
      "epoch": 2.8186941484225496,
      "grad_norm": 0.05727174233175065,
      "learning_rate": 8.403055624227935e-05,
      "loss": 0.8405,
      "step": 160550
    },
    {
      "epoch": 2.818869713302551,
      "grad_norm": 0.05380111863542939,
      "learning_rate": 8.401957764248268e-05,
      "loss": 0.8464,
      "step": 160560
    },
    {
      "epoch": 2.8190452781825526,
      "grad_norm": 0.07387734321479636,
      "learning_rate": 8.400859925783129e-05,
      "loss": 0.8466,
      "step": 160570
    },
    {
      "epoch": 2.8192208430625536,
      "grad_norm": 0.05277215242188288,
      "learning_rate": 8.399762108846419e-05,
      "loss": 0.8431,
      "step": 160580
    },
    {
      "epoch": 2.819396407942555,
      "grad_norm": 0.06575437184227235,
      "learning_rate": 8.398664313452048e-05,
      "loss": 0.844,
      "step": 160590
    },
    {
      "epoch": 2.8195719728225566,
      "grad_norm": 0.0647803792997902,
      "learning_rate": 8.397566539613925e-05,
      "loss": 0.8385,
      "step": 160600
    },
    {
      "epoch": 2.819747537702558,
      "grad_norm": 0.07371792463609031,
      "learning_rate": 8.396468787345962e-05,
      "loss": 0.84,
      "step": 160610
    },
    {
      "epoch": 2.8199231025825595,
      "grad_norm": 0.06459258822254475,
      "learning_rate": 8.395371056662065e-05,
      "loss": 0.8506,
      "step": 160620
    },
    {
      "epoch": 2.8200986674625605,
      "grad_norm": 0.06244756732189531,
      "learning_rate": 8.394273347576141e-05,
      "loss": 0.8484,
      "step": 160630
    },
    {
      "epoch": 2.8202742323425625,
      "grad_norm": 0.04252185642260427,
      "learning_rate": 8.393175660102107e-05,
      "loss": 0.8362,
      "step": 160640
    },
    {
      "epoch": 2.8204497972225635,
      "grad_norm": 0.052231924657637645,
      "learning_rate": 8.392077994253862e-05,
      "loss": 0.8495,
      "step": 160650
    },
    {
      "epoch": 2.820625362102565,
      "grad_norm": 0.043214377114657705,
      "learning_rate": 8.390980350045317e-05,
      "loss": 0.8423,
      "step": 160660
    },
    {
      "epoch": 2.8208009269825665,
      "grad_norm": 0.05601303219764068,
      "learning_rate": 8.389882727490372e-05,
      "loss": 0.8435,
      "step": 160670
    },
    {
      "epoch": 2.820976491862568,
      "grad_norm": 0.06155243052001449,
      "learning_rate": 8.388785126602946e-05,
      "loss": 0.8485,
      "step": 160680
    },
    {
      "epoch": 2.8211520567425694,
      "grad_norm": 0.05992365579731497,
      "learning_rate": 8.38768754739694e-05,
      "loss": 0.8431,
      "step": 160690
    },
    {
      "epoch": 2.8213276216225704,
      "grad_norm": 0.06370609179867602,
      "learning_rate": 8.38658998988626e-05,
      "loss": 0.8362,
      "step": 160700
    },
    {
      "epoch": 2.821503186502572,
      "grad_norm": 0.056898203181916095,
      "learning_rate": 8.385492454084814e-05,
      "loss": 0.852,
      "step": 160710
    },
    {
      "epoch": 2.8216787513825734,
      "grad_norm": 0.059579429617853744,
      "learning_rate": 8.384394940006508e-05,
      "loss": 0.847,
      "step": 160720
    },
    {
      "epoch": 2.821854316262575,
      "grad_norm": 0.060412728889174795,
      "learning_rate": 8.383297447665241e-05,
      "loss": 0.8426,
      "step": 160730
    },
    {
      "epoch": 2.8220298811425764,
      "grad_norm": 0.05067039840667377,
      "learning_rate": 8.382199977074931e-05,
      "loss": 0.8401,
      "step": 160740
    },
    {
      "epoch": 2.8222054460225774,
      "grad_norm": 0.0481578999753639,
      "learning_rate": 8.381102528249475e-05,
      "loss": 0.8471,
      "step": 160750
    },
    {
      "epoch": 2.8223810109025793,
      "grad_norm": 0.05291318320412409,
      "learning_rate": 8.38000510120278e-05,
      "loss": 0.8459,
      "step": 160760
    },
    {
      "epoch": 2.8225565757825803,
      "grad_norm": 0.04984936321336054,
      "learning_rate": 8.378907695948748e-05,
      "loss": 0.8389,
      "step": 160770
    },
    {
      "epoch": 2.822732140662582,
      "grad_norm": 0.04695901713969476,
      "learning_rate": 8.377810312501287e-05,
      "loss": 0.8424,
      "step": 160780
    },
    {
      "epoch": 2.8229077055425833,
      "grad_norm": 0.06951030595853698,
      "learning_rate": 8.376712950874299e-05,
      "loss": 0.8402,
      "step": 160790
    },
    {
      "epoch": 2.823083270422585,
      "grad_norm": 0.07047893547175117,
      "learning_rate": 8.375615611081685e-05,
      "loss": 0.8535,
      "step": 160800
    },
    {
      "epoch": 2.8232588353025863,
      "grad_norm": 0.05202334726848883,
      "learning_rate": 8.374518293137357e-05,
      "loss": 0.8423,
      "step": 160810
    },
    {
      "epoch": 2.8234344001825873,
      "grad_norm": 0.056331920665803295,
      "learning_rate": 8.373420997055213e-05,
      "loss": 0.8404,
      "step": 160820
    },
    {
      "epoch": 2.8236099650625888,
      "grad_norm": 0.05494773620617497,
      "learning_rate": 8.372323722849154e-05,
      "loss": 0.8439,
      "step": 160830
    },
    {
      "epoch": 2.8237855299425902,
      "grad_norm": 0.049669524179407676,
      "learning_rate": 8.37122647053308e-05,
      "loss": 0.8445,
      "step": 160840
    },
    {
      "epoch": 2.8239610948225917,
      "grad_norm": 0.06594467178127901,
      "learning_rate": 8.370129240120902e-05,
      "loss": 0.8425,
      "step": 160850
    },
    {
      "epoch": 2.824136659702593,
      "grad_norm": 0.06672538171476199,
      "learning_rate": 8.369032031626519e-05,
      "loss": 0.8413,
      "step": 160860
    },
    {
      "epoch": 2.8243122245825942,
      "grad_norm": 0.04329497888010286,
      "learning_rate": 8.36793484506383e-05,
      "loss": 0.8496,
      "step": 160870
    },
    {
      "epoch": 2.824487789462596,
      "grad_norm": 0.0563431323062391,
      "learning_rate": 8.366837680446739e-05,
      "loss": 0.8421,
      "step": 160880
    },
    {
      "epoch": 2.824663354342597,
      "grad_norm": 0.06598508460957991,
      "learning_rate": 8.365740537789148e-05,
      "loss": 0.8457,
      "step": 160890
    },
    {
      "epoch": 2.8248389192225987,
      "grad_norm": 0.06338171391397843,
      "learning_rate": 8.364643417104953e-05,
      "loss": 0.843,
      "step": 160900
    },
    {
      "epoch": 2.8250144841026,
      "grad_norm": 0.053133144676679275,
      "learning_rate": 8.36354631840806e-05,
      "loss": 0.8532,
      "step": 160910
    },
    {
      "epoch": 2.8251900489826016,
      "grad_norm": 0.06638971560757652,
      "learning_rate": 8.362449241712372e-05,
      "loss": 0.8475,
      "step": 160920
    },
    {
      "epoch": 2.825365613862603,
      "grad_norm": 0.049766892012850415,
      "learning_rate": 8.361352187031783e-05,
      "loss": 0.8443,
      "step": 160930
    },
    {
      "epoch": 2.825541178742604,
      "grad_norm": 0.05312968151168612,
      "learning_rate": 8.360255154380193e-05,
      "loss": 0.8423,
      "step": 160940
    },
    {
      "epoch": 2.8257167436226056,
      "grad_norm": 0.050932177349733786,
      "learning_rate": 8.359158143771505e-05,
      "loss": 0.848,
      "step": 160950
    },
    {
      "epoch": 2.825892308502607,
      "grad_norm": 0.05383920867891556,
      "learning_rate": 8.358061155219618e-05,
      "loss": 0.8433,
      "step": 160960
    },
    {
      "epoch": 2.8260678733826086,
      "grad_norm": 0.04994403400387315,
      "learning_rate": 8.356964188738425e-05,
      "loss": 0.8465,
      "step": 160970
    },
    {
      "epoch": 2.82624343826261,
      "grad_norm": 0.0713052962373993,
      "learning_rate": 8.355867244341837e-05,
      "loss": 0.8341,
      "step": 160980
    },
    {
      "epoch": 2.826419003142611,
      "grad_norm": 0.046884375364308874,
      "learning_rate": 8.354770322043744e-05,
      "loss": 0.8499,
      "step": 160990
    },
    {
      "epoch": 2.826594568022613,
      "grad_norm": 0.0901140477719118,
      "learning_rate": 8.353673421858043e-05,
      "loss": 0.8422,
      "step": 161000
    },
    {
      "epoch": 2.826770132902614,
      "grad_norm": 0.0464544426222421,
      "learning_rate": 8.352576543798632e-05,
      "loss": 0.8485,
      "step": 161010
    },
    {
      "epoch": 2.8269456977826155,
      "grad_norm": 0.05938897461276247,
      "learning_rate": 8.351479687879414e-05,
      "loss": 0.8429,
      "step": 161020
    },
    {
      "epoch": 2.827121262662617,
      "grad_norm": 0.05233920933364172,
      "learning_rate": 8.350382854114284e-05,
      "loss": 0.8376,
      "step": 161030
    },
    {
      "epoch": 2.8272968275426185,
      "grad_norm": 0.058200493850367865,
      "learning_rate": 8.349286042517137e-05,
      "loss": 0.8475,
      "step": 161040
    },
    {
      "epoch": 2.82747239242262,
      "grad_norm": 0.05315796691585697,
      "learning_rate": 8.348189253101873e-05,
      "loss": 0.8484,
      "step": 161050
    },
    {
      "epoch": 2.827647957302621,
      "grad_norm": 0.05280917913489597,
      "learning_rate": 8.347092485882387e-05,
      "loss": 0.8432,
      "step": 161060
    },
    {
      "epoch": 2.8278235221826225,
      "grad_norm": 0.057598187507443895,
      "learning_rate": 8.34599574087257e-05,
      "loss": 0.8459,
      "step": 161070
    },
    {
      "epoch": 2.827999087062624,
      "grad_norm": 0.05708825997686406,
      "learning_rate": 8.344899018086329e-05,
      "loss": 0.8427,
      "step": 161080
    },
    {
      "epoch": 2.8281746519426254,
      "grad_norm": 0.05298927480658464,
      "learning_rate": 8.343802317537553e-05,
      "loss": 0.8415,
      "step": 161090
    },
    {
      "epoch": 2.828350216822627,
      "grad_norm": 0.05091246725504582,
      "learning_rate": 8.34270563924014e-05,
      "loss": 0.8506,
      "step": 161100
    },
    {
      "epoch": 2.8285257817026284,
      "grad_norm": 0.075943728040945,
      "learning_rate": 8.341608983207978e-05,
      "loss": 0.8405,
      "step": 161110
    },
    {
      "epoch": 2.82870134658263,
      "grad_norm": 0.07144087086614326,
      "learning_rate": 8.34051234945497e-05,
      "loss": 0.8382,
      "step": 161120
    },
    {
      "epoch": 2.828876911462631,
      "grad_norm": 0.049003354466764945,
      "learning_rate": 8.33941573799501e-05,
      "loss": 0.8476,
      "step": 161130
    },
    {
      "epoch": 2.8290524763426323,
      "grad_norm": 0.06304019585611115,
      "learning_rate": 8.338319148841984e-05,
      "loss": 0.8359,
      "step": 161140
    },
    {
      "epoch": 2.829228041222634,
      "grad_norm": 0.04821670338258324,
      "learning_rate": 8.337222582009799e-05,
      "loss": 0.8478,
      "step": 161150
    },
    {
      "epoch": 2.8294036061026353,
      "grad_norm": 0.04222361871720223,
      "learning_rate": 8.336126037512339e-05,
      "loss": 0.834,
      "step": 161160
    },
    {
      "epoch": 2.829579170982637,
      "grad_norm": 0.0641254436980975,
      "learning_rate": 8.3350295153635e-05,
      "loss": 0.8408,
      "step": 161170
    },
    {
      "epoch": 2.829754735862638,
      "grad_norm": 0.09679599544049684,
      "learning_rate": 8.333933015577172e-05,
      "loss": 0.8412,
      "step": 161180
    },
    {
      "epoch": 2.8299303007426393,
      "grad_norm": 0.05031183709937867,
      "learning_rate": 8.332836538167255e-05,
      "loss": 0.8408,
      "step": 161190
    },
    {
      "epoch": 2.8301058656226408,
      "grad_norm": 0.04880848014918155,
      "learning_rate": 8.331740083147638e-05,
      "loss": 0.8386,
      "step": 161200
    },
    {
      "epoch": 2.8302814305026422,
      "grad_norm": 0.0546806108383019,
      "learning_rate": 8.33064365053221e-05,
      "loss": 0.8396,
      "step": 161210
    },
    {
      "epoch": 2.8304569953826437,
      "grad_norm": 0.04944813041339866,
      "learning_rate": 8.329547240334869e-05,
      "loss": 0.8416,
      "step": 161220
    },
    {
      "epoch": 2.830632560262645,
      "grad_norm": 0.07125638484630656,
      "learning_rate": 8.328450852569503e-05,
      "loss": 0.8385,
      "step": 161230
    },
    {
      "epoch": 2.8308081251426467,
      "grad_norm": 0.056433657817796426,
      "learning_rate": 8.32735448725e-05,
      "loss": 0.8411,
      "step": 161240
    },
    {
      "epoch": 2.8309836900226477,
      "grad_norm": 0.06300921345140825,
      "learning_rate": 8.32625814439026e-05,
      "loss": 0.8426,
      "step": 161250
    },
    {
      "epoch": 2.831159254902649,
      "grad_norm": 0.05106785909013939,
      "learning_rate": 8.32516182400417e-05,
      "loss": 0.8423,
      "step": 161260
    },
    {
      "epoch": 2.8313348197826507,
      "grad_norm": 0.05989199990080894,
      "learning_rate": 8.324065526105619e-05,
      "loss": 0.8405,
      "step": 161270
    },
    {
      "epoch": 2.831510384662652,
      "grad_norm": 0.05336664831684142,
      "learning_rate": 8.322969250708497e-05,
      "loss": 0.8386,
      "step": 161280
    },
    {
      "epoch": 2.8316859495426536,
      "grad_norm": 0.054615930934829755,
      "learning_rate": 8.321872997826696e-05,
      "loss": 0.8497,
      "step": 161290
    },
    {
      "epoch": 2.8318615144226547,
      "grad_norm": 0.04819567333642558,
      "learning_rate": 8.320776767474105e-05,
      "loss": 0.8455,
      "step": 161300
    },
    {
      "epoch": 2.832037079302656,
      "grad_norm": 0.04633640998796185,
      "learning_rate": 8.31968055966461e-05,
      "loss": 0.8405,
      "step": 161310
    },
    {
      "epoch": 2.8322126441826576,
      "grad_norm": 0.04818204050425656,
      "learning_rate": 8.318584374412109e-05,
      "loss": 0.8443,
      "step": 161320
    },
    {
      "epoch": 2.832388209062659,
      "grad_norm": 0.04548776383061095,
      "learning_rate": 8.317488211730483e-05,
      "loss": 0.8473,
      "step": 161330
    },
    {
      "epoch": 2.8325637739426606,
      "grad_norm": 0.055742034853920513,
      "learning_rate": 8.316392071633624e-05,
      "loss": 0.8441,
      "step": 161340
    },
    {
      "epoch": 2.832739338822662,
      "grad_norm": 0.04621637588001382,
      "learning_rate": 8.315295954135415e-05,
      "loss": 0.8437,
      "step": 161350
    },
    {
      "epoch": 2.8329149037026635,
      "grad_norm": 0.05214606693068586,
      "learning_rate": 8.314199859249752e-05,
      "loss": 0.8488,
      "step": 161360
    },
    {
      "epoch": 2.8330904685826646,
      "grad_norm": 0.05342567359676524,
      "learning_rate": 8.313103786990518e-05,
      "loss": 0.8477,
      "step": 161370
    },
    {
      "epoch": 2.833266033462666,
      "grad_norm": 0.05732042329931968,
      "learning_rate": 8.312007737371601e-05,
      "loss": 0.838,
      "step": 161380
    },
    {
      "epoch": 2.8334415983426675,
      "grad_norm": 0.042875719247276387,
      "learning_rate": 8.310911710406891e-05,
      "loss": 0.844,
      "step": 161390
    },
    {
      "epoch": 2.833617163222669,
      "grad_norm": 0.05629729384410281,
      "learning_rate": 8.309815706110271e-05,
      "loss": 0.8379,
      "step": 161400
    },
    {
      "epoch": 2.8337927281026705,
      "grad_norm": 0.045939402641018305,
      "learning_rate": 8.308719724495625e-05,
      "loss": 0.8451,
      "step": 161410
    },
    {
      "epoch": 2.8339682929826715,
      "grad_norm": 0.054227105442147844,
      "learning_rate": 8.30762376557685e-05,
      "loss": 0.8528,
      "step": 161420
    },
    {
      "epoch": 2.8341438578626734,
      "grad_norm": 0.05159681739641528,
      "learning_rate": 8.306527829367821e-05,
      "loss": 0.835,
      "step": 161430
    },
    {
      "epoch": 2.8343194227426745,
      "grad_norm": 0.048048390966518056,
      "learning_rate": 8.305431915882432e-05,
      "loss": 0.8349,
      "step": 161440
    },
    {
      "epoch": 2.834494987622676,
      "grad_norm": 0.06703813625824395,
      "learning_rate": 8.304336025134561e-05,
      "loss": 0.8408,
      "step": 161450
    },
    {
      "epoch": 2.8346705525026774,
      "grad_norm": 0.07120008713807575,
      "learning_rate": 8.3032401571381e-05,
      "loss": 0.8393,
      "step": 161460
    },
    {
      "epoch": 2.834846117382679,
      "grad_norm": 0.059988475697522324,
      "learning_rate": 8.302144311906929e-05,
      "loss": 0.839,
      "step": 161470
    },
    {
      "epoch": 2.8350216822626804,
      "grad_norm": 0.05178536138157026,
      "learning_rate": 8.301048489454931e-05,
      "loss": 0.845,
      "step": 161480
    },
    {
      "epoch": 2.8351972471426814,
      "grad_norm": 0.042932550769479556,
      "learning_rate": 8.299952689795999e-05,
      "loss": 0.8503,
      "step": 161490
    },
    {
      "epoch": 2.835372812022683,
      "grad_norm": 0.051540638972817215,
      "learning_rate": 8.29885691294401e-05,
      "loss": 0.8396,
      "step": 161500
    },
    {
      "epoch": 2.8355483769026844,
      "grad_norm": 0.057603776614736237,
      "learning_rate": 8.297761158912846e-05,
      "loss": 0.8523,
      "step": 161510
    },
    {
      "epoch": 2.835723941782686,
      "grad_norm": 0.054455439314500365,
      "learning_rate": 8.296665427716393e-05,
      "loss": 0.8425,
      "step": 161520
    },
    {
      "epoch": 2.8358995066626873,
      "grad_norm": 0.05497401025630339,
      "learning_rate": 8.295569719368537e-05,
      "loss": 0.8375,
      "step": 161530
    },
    {
      "epoch": 2.8360750715426883,
      "grad_norm": 0.04273922057684765,
      "learning_rate": 8.29447403388316e-05,
      "loss": 0.8477,
      "step": 161540
    },
    {
      "epoch": 2.8362506364226903,
      "grad_norm": 0.045899771442003974,
      "learning_rate": 8.29337837127414e-05,
      "loss": 0.8365,
      "step": 161550
    },
    {
      "epoch": 2.8364262013026913,
      "grad_norm": 0.05099709320510753,
      "learning_rate": 8.292282731555365e-05,
      "loss": 0.8426,
      "step": 161560
    },
    {
      "epoch": 2.8366017661826928,
      "grad_norm": 0.05440906384108579,
      "learning_rate": 8.291187114740713e-05,
      "loss": 0.8423,
      "step": 161570
    },
    {
      "epoch": 2.8367773310626943,
      "grad_norm": 0.0489468974967421,
      "learning_rate": 8.290091520844063e-05,
      "loss": 0.8435,
      "step": 161580
    },
    {
      "epoch": 2.8369528959426957,
      "grad_norm": 0.07604436633873468,
      "learning_rate": 8.288995949879307e-05,
      "loss": 0.8422,
      "step": 161590
    },
    {
      "epoch": 2.837128460822697,
      "grad_norm": 0.05191956805488617,
      "learning_rate": 8.287900401860316e-05,
      "loss": 0.8522,
      "step": 161600
    },
    {
      "epoch": 2.8373040257026982,
      "grad_norm": 0.06790531270237672,
      "learning_rate": 8.286804876800977e-05,
      "loss": 0.8424,
      "step": 161610
    },
    {
      "epoch": 2.8374795905826997,
      "grad_norm": 0.05377815333362147,
      "learning_rate": 8.285709374715163e-05,
      "loss": 0.8381,
      "step": 161620
    },
    {
      "epoch": 2.837655155462701,
      "grad_norm": 0.04891461672490368,
      "learning_rate": 8.284613895616763e-05,
      "loss": 0.8398,
      "step": 161630
    },
    {
      "epoch": 2.8378307203427027,
      "grad_norm": 0.05765333970875278,
      "learning_rate": 8.28351843951965e-05,
      "loss": 0.8339,
      "step": 161640
    },
    {
      "epoch": 2.838006285222704,
      "grad_norm": 0.05066845573257661,
      "learning_rate": 8.282423006437706e-05,
      "loss": 0.8461,
      "step": 161650
    },
    {
      "epoch": 2.838181850102705,
      "grad_norm": 0.05164358603181113,
      "learning_rate": 8.281327596384816e-05,
      "loss": 0.8389,
      "step": 161660
    },
    {
      "epoch": 2.838357414982707,
      "grad_norm": 0.05853827701392623,
      "learning_rate": 8.280232209374851e-05,
      "loss": 0.8405,
      "step": 161670
    },
    {
      "epoch": 2.838532979862708,
      "grad_norm": 0.057493381565370254,
      "learning_rate": 8.279136845421694e-05,
      "loss": 0.8501,
      "step": 161680
    },
    {
      "epoch": 2.8387085447427096,
      "grad_norm": 0.0621057140740637,
      "learning_rate": 8.278041504539216e-05,
      "loss": 0.8334,
      "step": 161690
    },
    {
      "epoch": 2.838884109622711,
      "grad_norm": 0.07335919981819886,
      "learning_rate": 8.276946186741306e-05,
      "loss": 0.8445,
      "step": 161700
    },
    {
      "epoch": 2.8390596745027126,
      "grad_norm": 0.050989536919732584,
      "learning_rate": 8.275850892041838e-05,
      "loss": 0.8403,
      "step": 161710
    },
    {
      "epoch": 2.839235239382714,
      "grad_norm": 0.07052997982731116,
      "learning_rate": 8.274755620454686e-05,
      "loss": 0.8407,
      "step": 161720
    },
    {
      "epoch": 2.839410804262715,
      "grad_norm": 0.05714342611283918,
      "learning_rate": 8.273660371993733e-05,
      "loss": 0.8434,
      "step": 161730
    },
    {
      "epoch": 2.8395863691427166,
      "grad_norm": 0.04943065958952161,
      "learning_rate": 8.272565146672851e-05,
      "loss": 0.8476,
      "step": 161740
    },
    {
      "epoch": 2.839761934022718,
      "grad_norm": 0.05001586954645202,
      "learning_rate": 8.271469944505917e-05,
      "loss": 0.8462,
      "step": 161750
    },
    {
      "epoch": 2.8399374989027195,
      "grad_norm": 0.06562455709767095,
      "learning_rate": 8.270374765506813e-05,
      "loss": 0.8372,
      "step": 161760
    },
    {
      "epoch": 2.840113063782721,
      "grad_norm": 0.04571430575003914,
      "learning_rate": 8.269279609689411e-05,
      "loss": 0.8455,
      "step": 161770
    },
    {
      "epoch": 2.8402886286627225,
      "grad_norm": 0.05348535368597109,
      "learning_rate": 8.268184477067586e-05,
      "loss": 0.8391,
      "step": 161780
    },
    {
      "epoch": 2.840464193542724,
      "grad_norm": 0.06045641150486199,
      "learning_rate": 8.267089367655215e-05,
      "loss": 0.8376,
      "step": 161790
    },
    {
      "epoch": 2.840639758422725,
      "grad_norm": 0.053890183527500614,
      "learning_rate": 8.265994281466173e-05,
      "loss": 0.8457,
      "step": 161800
    },
    {
      "epoch": 2.8408153233027265,
      "grad_norm": 0.04276363932334596,
      "learning_rate": 8.264899218514336e-05,
      "loss": 0.8426,
      "step": 161810
    },
    {
      "epoch": 2.840990888182728,
      "grad_norm": 0.05370237270740511,
      "learning_rate": 8.263804178813574e-05,
      "loss": 0.8383,
      "step": 161820
    },
    {
      "epoch": 2.8411664530627294,
      "grad_norm": 0.06349509882973133,
      "learning_rate": 8.262709162377769e-05,
      "loss": 0.848,
      "step": 161830
    },
    {
      "epoch": 2.841342017942731,
      "grad_norm": 0.06307543752785787,
      "learning_rate": 8.261614169220793e-05,
      "loss": 0.837,
      "step": 161840
    },
    {
      "epoch": 2.841517582822732,
      "grad_norm": 0.055538084809664094,
      "learning_rate": 8.260519199356513e-05,
      "loss": 0.8417,
      "step": 161850
    },
    {
      "epoch": 2.8416931477027334,
      "grad_norm": 0.06963862828543421,
      "learning_rate": 8.259424252798806e-05,
      "loss": 0.8452,
      "step": 161860
    },
    {
      "epoch": 2.841868712582735,
      "grad_norm": 0.04981898396865155,
      "learning_rate": 8.258329329561549e-05,
      "loss": 0.8394,
      "step": 161870
    },
    {
      "epoch": 2.8420442774627364,
      "grad_norm": 0.06577278843000728,
      "learning_rate": 8.257234429658616e-05,
      "loss": 0.8421,
      "step": 161880
    },
    {
      "epoch": 2.842219842342738,
      "grad_norm": 0.06773542502098678,
      "learning_rate": 8.25613955310387e-05,
      "loss": 0.8421,
      "step": 161890
    },
    {
      "epoch": 2.8423954072227393,
      "grad_norm": 0.047970059472273216,
      "learning_rate": 8.255044699911192e-05,
      "loss": 0.8511,
      "step": 161900
    },
    {
      "epoch": 2.842570972102741,
      "grad_norm": 0.060730390263423244,
      "learning_rate": 8.253949870094453e-05,
      "loss": 0.8386,
      "step": 161910
    },
    {
      "epoch": 2.842746536982742,
      "grad_norm": 0.04989283469599602,
      "learning_rate": 8.252855063667517e-05,
      "loss": 0.8478,
      "step": 161920
    },
    {
      "epoch": 2.8429221018627433,
      "grad_norm": 0.050557443333727935,
      "learning_rate": 8.251760280644266e-05,
      "loss": 0.8416,
      "step": 161930
    },
    {
      "epoch": 2.8430976667427448,
      "grad_norm": 0.06063688754617192,
      "learning_rate": 8.250665521038566e-05,
      "loss": 0.8464,
      "step": 161940
    },
    {
      "epoch": 2.8432732316227463,
      "grad_norm": 0.07570491864730164,
      "learning_rate": 8.249570784864289e-05,
      "loss": 0.8478,
      "step": 161950
    },
    {
      "epoch": 2.8434487965027477,
      "grad_norm": 0.06903430478739418,
      "learning_rate": 8.248476072135302e-05,
      "loss": 0.8382,
      "step": 161960
    },
    {
      "epoch": 2.8436243613827488,
      "grad_norm": 0.057449822209230006,
      "learning_rate": 8.247381382865481e-05,
      "loss": 0.8384,
      "step": 161970
    },
    {
      "epoch": 2.8437999262627502,
      "grad_norm": 0.04790329099495463,
      "learning_rate": 8.246286717068693e-05,
      "loss": 0.844,
      "step": 161980
    },
    {
      "epoch": 2.8439754911427517,
      "grad_norm": 0.05456021561389492,
      "learning_rate": 8.245192074758802e-05,
      "loss": 0.8323,
      "step": 161990
    },
    {
      "epoch": 2.844151056022753,
      "grad_norm": 0.044150882708937134,
      "learning_rate": 8.244097455949691e-05,
      "loss": 0.8345,
      "step": 162000
    },
    {
      "epoch": 2.8443266209027547,
      "grad_norm": 0.061365970101624485,
      "learning_rate": 8.243002860655217e-05,
      "loss": 0.8435,
      "step": 162010
    },
    {
      "epoch": 2.844502185782756,
      "grad_norm": 0.049108490992043154,
      "learning_rate": 8.241908288889254e-05,
      "loss": 0.8466,
      "step": 162020
    },
    {
      "epoch": 2.8446777506627576,
      "grad_norm": 0.04934576339567541,
      "learning_rate": 8.240813740665665e-05,
      "loss": 0.8391,
      "step": 162030
    },
    {
      "epoch": 2.8448533155427587,
      "grad_norm": 0.0481653560124774,
      "learning_rate": 8.239719215998326e-05,
      "loss": 0.8442,
      "step": 162040
    },
    {
      "epoch": 2.84502888042276,
      "grad_norm": 0.04840076522613405,
      "learning_rate": 8.2386247149011e-05,
      "loss": 0.8406,
      "step": 162050
    },
    {
      "epoch": 2.8452044453027616,
      "grad_norm": 0.061527903275703684,
      "learning_rate": 8.237530237387854e-05,
      "loss": 0.8385,
      "step": 162060
    },
    {
      "epoch": 2.845380010182763,
      "grad_norm": 0.05804447228935628,
      "learning_rate": 8.236435783472459e-05,
      "loss": 0.8405,
      "step": 162070
    },
    {
      "epoch": 2.8455555750627646,
      "grad_norm": 0.048735179238661865,
      "learning_rate": 8.23534135316878e-05,
      "loss": 0.8447,
      "step": 162080
    },
    {
      "epoch": 2.8457311399427656,
      "grad_norm": 0.06534405839571197,
      "learning_rate": 8.234246946490679e-05,
      "loss": 0.832,
      "step": 162090
    },
    {
      "epoch": 2.8459067048227675,
      "grad_norm": 0.04767505760847162,
      "learning_rate": 8.233152563452032e-05,
      "loss": 0.8382,
      "step": 162100
    },
    {
      "epoch": 2.8460822697027686,
      "grad_norm": 0.05740396988213716,
      "learning_rate": 8.232058204066699e-05,
      "loss": 0.8471,
      "step": 162110
    },
    {
      "epoch": 2.84625783458277,
      "grad_norm": 0.05784867177467993,
      "learning_rate": 8.230963868348547e-05,
      "loss": 0.8397,
      "step": 162120
    },
    {
      "epoch": 2.8464333994627715,
      "grad_norm": 0.048418751412068,
      "learning_rate": 8.229869556311438e-05,
      "loss": 0.8409,
      "step": 162130
    },
    {
      "epoch": 2.846608964342773,
      "grad_norm": 0.04359847623543304,
      "learning_rate": 8.228775267969244e-05,
      "loss": 0.8458,
      "step": 162140
    },
    {
      "epoch": 2.8467845292227745,
      "grad_norm": 0.05461070574660009,
      "learning_rate": 8.227681003335824e-05,
      "loss": 0.8416,
      "step": 162150
    },
    {
      "epoch": 2.8469600941027755,
      "grad_norm": 0.04721975327462989,
      "learning_rate": 8.226586762425041e-05,
      "loss": 0.8453,
      "step": 162160
    },
    {
      "epoch": 2.847135658982777,
      "grad_norm": 0.0529950570799471,
      "learning_rate": 8.22549254525077e-05,
      "loss": 0.8462,
      "step": 162170
    },
    {
      "epoch": 2.8473112238627785,
      "grad_norm": 0.04584626356035472,
      "learning_rate": 8.224398351826864e-05,
      "loss": 0.8467,
      "step": 162180
    },
    {
      "epoch": 2.84748678874278,
      "grad_norm": 0.06754019899847151,
      "learning_rate": 8.223304182167191e-05,
      "loss": 0.8437,
      "step": 162190
    },
    {
      "epoch": 2.8476623536227814,
      "grad_norm": 0.0564330543232566,
      "learning_rate": 8.222210036285609e-05,
      "loss": 0.8446,
      "step": 162200
    },
    {
      "epoch": 2.8478379185027825,
      "grad_norm": 0.03901778955614253,
      "learning_rate": 8.221115914195991e-05,
      "loss": 0.8471,
      "step": 162210
    },
    {
      "epoch": 2.8480134833827844,
      "grad_norm": 0.04232536870274305,
      "learning_rate": 8.220021815912193e-05,
      "loss": 0.8463,
      "step": 162220
    },
    {
      "epoch": 2.8481890482627854,
      "grad_norm": 0.04309869886852857,
      "learning_rate": 8.218927741448079e-05,
      "loss": 0.8401,
      "step": 162230
    },
    {
      "epoch": 2.848364613142787,
      "grad_norm": 0.06819919043603538,
      "learning_rate": 8.21783369081751e-05,
      "loss": 0.8427,
      "step": 162240
    },
    {
      "epoch": 2.8485401780227884,
      "grad_norm": 0.05102213578603289,
      "learning_rate": 8.216739664034352e-05,
      "loss": 0.842,
      "step": 162250
    },
    {
      "epoch": 2.84871574290279,
      "grad_norm": 0.05593126484714346,
      "learning_rate": 8.215645661112457e-05,
      "loss": 0.8423,
      "step": 162260
    },
    {
      "epoch": 2.8488913077827913,
      "grad_norm": 0.05405296895021275,
      "learning_rate": 8.214551682065697e-05,
      "loss": 0.8446,
      "step": 162270
    },
    {
      "epoch": 2.8490668726627923,
      "grad_norm": 0.055243858955163545,
      "learning_rate": 8.21345772690793e-05,
      "loss": 0.8337,
      "step": 162280
    },
    {
      "epoch": 2.849242437542794,
      "grad_norm": 0.05704205494363902,
      "learning_rate": 8.212363795653015e-05,
      "loss": 0.8361,
      "step": 162290
    },
    {
      "epoch": 2.8494180024227953,
      "grad_norm": 0.059050960496171485,
      "learning_rate": 8.211269888314809e-05,
      "loss": 0.8432,
      "step": 162300
    },
    {
      "epoch": 2.849593567302797,
      "grad_norm": 0.058757388581464204,
      "learning_rate": 8.210176004907178e-05,
      "loss": 0.8458,
      "step": 162310
    },
    {
      "epoch": 2.8497691321827983,
      "grad_norm": 0.045474684258182636,
      "learning_rate": 8.20908214544398e-05,
      "loss": 0.8344,
      "step": 162320
    },
    {
      "epoch": 2.8499446970627993,
      "grad_norm": 0.051263066291495366,
      "learning_rate": 8.207988309939068e-05,
      "loss": 0.837,
      "step": 162330
    },
    {
      "epoch": 2.850120261942801,
      "grad_norm": 0.05158000043847049,
      "learning_rate": 8.206894498406314e-05,
      "loss": 0.8434,
      "step": 162340
    },
    {
      "epoch": 2.8502958268228022,
      "grad_norm": 0.057232470353888246,
      "learning_rate": 8.205800710859567e-05,
      "loss": 0.8458,
      "step": 162350
    },
    {
      "epoch": 2.8504713917028037,
      "grad_norm": 0.05043335886361356,
      "learning_rate": 8.204706947312689e-05,
      "loss": 0.846,
      "step": 162360
    },
    {
      "epoch": 2.850646956582805,
      "grad_norm": 0.048233414457965654,
      "learning_rate": 8.203613207779531e-05,
      "loss": 0.8432,
      "step": 162370
    },
    {
      "epoch": 2.8508225214628067,
      "grad_norm": 0.04325784782450015,
      "learning_rate": 8.202519492273964e-05,
      "loss": 0.8422,
      "step": 162380
    },
    {
      "epoch": 2.850998086342808,
      "grad_norm": 0.04340542349814386,
      "learning_rate": 8.201425800809837e-05,
      "loss": 0.842,
      "step": 162390
    },
    {
      "epoch": 2.851173651222809,
      "grad_norm": 0.06657601465237896,
      "learning_rate": 8.200332133401008e-05,
      "loss": 0.8496,
      "step": 162400
    },
    {
      "epoch": 2.8513492161028107,
      "grad_norm": 0.05181655095759343,
      "learning_rate": 8.199238490061335e-05,
      "loss": 0.8425,
      "step": 162410
    },
    {
      "epoch": 2.851524780982812,
      "grad_norm": 0.06377610841499992,
      "learning_rate": 8.198144870804677e-05,
      "loss": 0.8478,
      "step": 162420
    },
    {
      "epoch": 2.8517003458628136,
      "grad_norm": 0.06373956813880655,
      "learning_rate": 8.197051275644881e-05,
      "loss": 0.8454,
      "step": 162430
    },
    {
      "epoch": 2.851875910742815,
      "grad_norm": 0.04798020461218606,
      "learning_rate": 8.195957704595816e-05,
      "loss": 0.8455,
      "step": 162440
    },
    {
      "epoch": 2.852051475622816,
      "grad_norm": 0.06165376899498941,
      "learning_rate": 8.194864157671333e-05,
      "loss": 0.842,
      "step": 162450
    },
    {
      "epoch": 2.852227040502818,
      "grad_norm": 0.05027855299825563,
      "learning_rate": 8.193770634885283e-05,
      "loss": 0.8385,
      "step": 162460
    },
    {
      "epoch": 2.852402605382819,
      "grad_norm": 0.06189387396670984,
      "learning_rate": 8.192677136251524e-05,
      "loss": 0.8407,
      "step": 162470
    },
    {
      "epoch": 2.8525781702628206,
      "grad_norm": 0.05852379800571084,
      "learning_rate": 8.191583661783915e-05,
      "loss": 0.8422,
      "step": 162480
    },
    {
      "epoch": 2.852753735142822,
      "grad_norm": 0.0668579467830791,
      "learning_rate": 8.190490211496306e-05,
      "loss": 0.8424,
      "step": 162490
    },
    {
      "epoch": 2.8529293000228235,
      "grad_norm": 0.06084699433088085,
      "learning_rate": 8.189396785402545e-05,
      "loss": 0.8504,
      "step": 162500
    },
    {
      "epoch": 2.853104864902825,
      "grad_norm": 0.04643516722405198,
      "learning_rate": 8.188303383516502e-05,
      "loss": 0.8419,
      "step": 162510
    },
    {
      "epoch": 2.853280429782826,
      "grad_norm": 0.04427016925593285,
      "learning_rate": 8.187210005852019e-05,
      "loss": 0.8408,
      "step": 162520
    },
    {
      "epoch": 2.8534559946628275,
      "grad_norm": 0.04654805850499281,
      "learning_rate": 8.18611665242295e-05,
      "loss": 0.8416,
      "step": 162530
    },
    {
      "epoch": 2.853631559542829,
      "grad_norm": 0.04931074456185344,
      "learning_rate": 8.185023323243148e-05,
      "loss": 0.8461,
      "step": 162540
    },
    {
      "epoch": 2.8538071244228305,
      "grad_norm": 0.059975398576798115,
      "learning_rate": 8.18393001832647e-05,
      "loss": 0.8433,
      "step": 162550
    },
    {
      "epoch": 2.853982689302832,
      "grad_norm": 0.047896083186611524,
      "learning_rate": 8.182836737686766e-05,
      "loss": 0.8437,
      "step": 162560
    },
    {
      "epoch": 2.8541582541828334,
      "grad_norm": 0.03806441206115155,
      "learning_rate": 8.181743481337888e-05,
      "loss": 0.8418,
      "step": 162570
    },
    {
      "epoch": 2.854333819062835,
      "grad_norm": 0.05583155728995068,
      "learning_rate": 8.180650249293688e-05,
      "loss": 0.8391,
      "step": 162580
    },
    {
      "epoch": 2.854509383942836,
      "grad_norm": 0.06705698091978694,
      "learning_rate": 8.179557041568018e-05,
      "loss": 0.8424,
      "step": 162590
    },
    {
      "epoch": 2.8546849488228374,
      "grad_norm": 0.061874443607374485,
      "learning_rate": 8.178463858174723e-05,
      "loss": 0.8368,
      "step": 162600
    },
    {
      "epoch": 2.854860513702839,
      "grad_norm": 0.05068795317931121,
      "learning_rate": 8.177370699127663e-05,
      "loss": 0.845,
      "step": 162610
    },
    {
      "epoch": 2.8550360785828404,
      "grad_norm": 0.05220231853010955,
      "learning_rate": 8.176277564440686e-05,
      "loss": 0.8437,
      "step": 162620
    },
    {
      "epoch": 2.855211643462842,
      "grad_norm": 0.06299384803963559,
      "learning_rate": 8.175184454127641e-05,
      "loss": 0.8479,
      "step": 162630
    },
    {
      "epoch": 2.855387208342843,
      "grad_norm": 0.05265288454350242,
      "learning_rate": 8.174091368202376e-05,
      "loss": 0.8409,
      "step": 162640
    },
    {
      "epoch": 2.8555627732228444,
      "grad_norm": 0.05068697016570266,
      "learning_rate": 8.172998306678743e-05,
      "loss": 0.8431,
      "step": 162650
    },
    {
      "epoch": 2.855738338102846,
      "grad_norm": 0.05939998687945758,
      "learning_rate": 8.171905269570592e-05,
      "loss": 0.8428,
      "step": 162660
    },
    {
      "epoch": 2.8559139029828473,
      "grad_norm": 0.05572139825493476,
      "learning_rate": 8.170812256891766e-05,
      "loss": 0.8452,
      "step": 162670
    },
    {
      "epoch": 2.856089467862849,
      "grad_norm": 0.0504388906431784,
      "learning_rate": 8.169719268656126e-05,
      "loss": 0.8389,
      "step": 162680
    },
    {
      "epoch": 2.8562650327428503,
      "grad_norm": 0.05848367196557318,
      "learning_rate": 8.168626304877508e-05,
      "loss": 0.8333,
      "step": 162690
    },
    {
      "epoch": 2.8564405976228517,
      "grad_norm": 0.06876380835322635,
      "learning_rate": 8.167533365569768e-05,
      "loss": 0.8362,
      "step": 162700
    },
    {
      "epoch": 2.8566161625028528,
      "grad_norm": 0.049126938089931405,
      "learning_rate": 8.166440450746745e-05,
      "loss": 0.8497,
      "step": 162710
    },
    {
      "epoch": 2.8567917273828543,
      "grad_norm": 0.052511292376052156,
      "learning_rate": 8.165347560422296e-05,
      "loss": 0.8446,
      "step": 162720
    },
    {
      "epoch": 2.8569672922628557,
      "grad_norm": 0.06256450476223914,
      "learning_rate": 8.164254694610264e-05,
      "loss": 0.8303,
      "step": 162730
    },
    {
      "epoch": 2.857142857142857,
      "grad_norm": 0.0549613630081144,
      "learning_rate": 8.163161853324495e-05,
      "loss": 0.8492,
      "step": 162740
    },
    {
      "epoch": 2.8573184220228587,
      "grad_norm": 0.05192751145455247,
      "learning_rate": 8.162069036578837e-05,
      "loss": 0.8372,
      "step": 162750
    },
    {
      "epoch": 2.8574939869028597,
      "grad_norm": 0.08167884062093288,
      "learning_rate": 8.160976244387137e-05,
      "loss": 0.848,
      "step": 162760
    },
    {
      "epoch": 2.857669551782861,
      "grad_norm": 0.057468442838540065,
      "learning_rate": 8.159883476763237e-05,
      "loss": 0.8375,
      "step": 162770
    },
    {
      "epoch": 2.8578451166628627,
      "grad_norm": 0.04778123679699133,
      "learning_rate": 8.158790733720986e-05,
      "loss": 0.8448,
      "step": 162780
    },
    {
      "epoch": 2.858020681542864,
      "grad_norm": 0.06323057607837164,
      "learning_rate": 8.157698015274229e-05,
      "loss": 0.8336,
      "step": 162790
    },
    {
      "epoch": 2.8581962464228656,
      "grad_norm": 0.0500788293298188,
      "learning_rate": 8.15660532143681e-05,
      "loss": 0.8392,
      "step": 162800
    },
    {
      "epoch": 2.858371811302867,
      "grad_norm": 0.07453984394980245,
      "learning_rate": 8.155512652222576e-05,
      "loss": 0.8476,
      "step": 162810
    },
    {
      "epoch": 2.8585473761828686,
      "grad_norm": 0.04918525062631145,
      "learning_rate": 8.154420007645367e-05,
      "loss": 0.8446,
      "step": 162820
    },
    {
      "epoch": 2.8587229410628696,
      "grad_norm": 0.04869536779840181,
      "learning_rate": 8.153327387719032e-05,
      "loss": 0.8451,
      "step": 162830
    },
    {
      "epoch": 2.858898505942871,
      "grad_norm": 0.057795593043404724,
      "learning_rate": 8.152234792457406e-05,
      "loss": 0.846,
      "step": 162840
    },
    {
      "epoch": 2.8590740708228726,
      "grad_norm": 0.0450387358080175,
      "learning_rate": 8.151142221874346e-05,
      "loss": 0.8337,
      "step": 162850
    },
    {
      "epoch": 2.859249635702874,
      "grad_norm": 0.0626789043935756,
      "learning_rate": 8.150049675983684e-05,
      "loss": 0.8383,
      "step": 162860
    },
    {
      "epoch": 2.8594252005828755,
      "grad_norm": 0.05638264363038742,
      "learning_rate": 8.148957154799268e-05,
      "loss": 0.8411,
      "step": 162870
    },
    {
      "epoch": 2.8596007654628766,
      "grad_norm": 0.04698034709406854,
      "learning_rate": 8.147864658334933e-05,
      "loss": 0.838,
      "step": 162880
    },
    {
      "epoch": 2.8597763303428785,
      "grad_norm": 0.0722142086377913,
      "learning_rate": 8.146772186604532e-05,
      "loss": 0.8414,
      "step": 162890
    },
    {
      "epoch": 2.8599518952228795,
      "grad_norm": 0.05024195956629383,
      "learning_rate": 8.145679739621903e-05,
      "loss": 0.8385,
      "step": 162900
    },
    {
      "epoch": 2.860127460102881,
      "grad_norm": 0.054024809805604444,
      "learning_rate": 8.144587317400882e-05,
      "loss": 0.8423,
      "step": 162910
    },
    {
      "epoch": 2.8603030249828825,
      "grad_norm": 0.06405160886122989,
      "learning_rate": 8.143494919955319e-05,
      "loss": 0.8426,
      "step": 162920
    },
    {
      "epoch": 2.860478589862884,
      "grad_norm": 0.07151645960102027,
      "learning_rate": 8.142402547299047e-05,
      "loss": 0.85,
      "step": 162930
    },
    {
      "epoch": 2.8606541547428854,
      "grad_norm": 0.059535665381651566,
      "learning_rate": 8.141310199445908e-05,
      "loss": 0.8363,
      "step": 162940
    },
    {
      "epoch": 2.8608297196228865,
      "grad_norm": 0.05272226771435376,
      "learning_rate": 8.140217876409748e-05,
      "loss": 0.846,
      "step": 162950
    },
    {
      "epoch": 2.861005284502888,
      "grad_norm": 0.052183812097720925,
      "learning_rate": 8.139125578204403e-05,
      "loss": 0.8449,
      "step": 162960
    },
    {
      "epoch": 2.8611808493828894,
      "grad_norm": 0.06616577647812245,
      "learning_rate": 8.138033304843712e-05,
      "loss": 0.8378,
      "step": 162970
    },
    {
      "epoch": 2.861356414262891,
      "grad_norm": 0.048023629795010056,
      "learning_rate": 8.136941056341513e-05,
      "loss": 0.8415,
      "step": 162980
    },
    {
      "epoch": 2.8615319791428924,
      "grad_norm": 0.0431467722183038,
      "learning_rate": 8.13584883271165e-05,
      "loss": 0.8381,
      "step": 162990
    },
    {
      "epoch": 2.8617075440228934,
      "grad_norm": 0.05246154289572494,
      "learning_rate": 8.13475663396796e-05,
      "loss": 0.8421,
      "step": 163000
    },
    {
      "epoch": 2.8618831089028953,
      "grad_norm": 0.06361487137104242,
      "learning_rate": 8.133664460124273e-05,
      "loss": 0.8434,
      "step": 163010
    },
    {
      "epoch": 2.8620586737828964,
      "grad_norm": 0.0708313659868788,
      "learning_rate": 8.132572311194441e-05,
      "loss": 0.8426,
      "step": 163020
    },
    {
      "epoch": 2.862234238662898,
      "grad_norm": 0.04956245823427152,
      "learning_rate": 8.131480187192295e-05,
      "loss": 0.8425,
      "step": 163030
    },
    {
      "epoch": 2.8624098035428993,
      "grad_norm": 0.057041988982654404,
      "learning_rate": 8.13038808813167e-05,
      "loss": 0.8416,
      "step": 163040
    },
    {
      "epoch": 2.862585368422901,
      "grad_norm": 0.06379251884335417,
      "learning_rate": 8.129296014026401e-05,
      "loss": 0.8442,
      "step": 163050
    },
    {
      "epoch": 2.8627609333029023,
      "grad_norm": 0.04755739622755285,
      "learning_rate": 8.128203964890337e-05,
      "loss": 0.84,
      "step": 163060
    },
    {
      "epoch": 2.8629364981829033,
      "grad_norm": 0.05182591020423575,
      "learning_rate": 8.127111940737303e-05,
      "loss": 0.8458,
      "step": 163070
    },
    {
      "epoch": 2.8631120630629048,
      "grad_norm": 0.05334332728264644,
      "learning_rate": 8.126019941581138e-05,
      "loss": 0.8351,
      "step": 163080
    },
    {
      "epoch": 2.8632876279429063,
      "grad_norm": 0.04600240054878718,
      "learning_rate": 8.12492796743568e-05,
      "loss": 0.8376,
      "step": 163090
    },
    {
      "epoch": 2.8634631928229077,
      "grad_norm": 0.05348331849959821,
      "learning_rate": 8.123836018314764e-05,
      "loss": 0.8395,
      "step": 163100
    },
    {
      "epoch": 2.863638757702909,
      "grad_norm": 0.06064516718761185,
      "learning_rate": 8.12274409423222e-05,
      "loss": 0.8426,
      "step": 163110
    },
    {
      "epoch": 2.8638143225829102,
      "grad_norm": 0.07285671059213056,
      "learning_rate": 8.121652195201892e-05,
      "loss": 0.8464,
      "step": 163120
    },
    {
      "epoch": 2.863989887462912,
      "grad_norm": 0.05525504850931629,
      "learning_rate": 8.120560321237608e-05,
      "loss": 0.8433,
      "step": 163130
    },
    {
      "epoch": 2.864165452342913,
      "grad_norm": 0.06398093264421183,
      "learning_rate": 8.119468472353206e-05,
      "loss": 0.8434,
      "step": 163140
    },
    {
      "epoch": 2.8643410172229147,
      "grad_norm": 0.05594982036241478,
      "learning_rate": 8.118376648562514e-05,
      "loss": 0.8447,
      "step": 163150
    },
    {
      "epoch": 2.864516582102916,
      "grad_norm": 0.05859826752301311,
      "learning_rate": 8.117284849879373e-05,
      "loss": 0.8433,
      "step": 163160
    },
    {
      "epoch": 2.8646921469829176,
      "grad_norm": 0.10678554876006183,
      "learning_rate": 8.116193076317613e-05,
      "loss": 0.8459,
      "step": 163170
    },
    {
      "epoch": 2.864867711862919,
      "grad_norm": 0.05813428486022015,
      "learning_rate": 8.115101327891063e-05,
      "loss": 0.8384,
      "step": 163180
    },
    {
      "epoch": 2.86504327674292,
      "grad_norm": 0.046226082818739696,
      "learning_rate": 8.114009604613564e-05,
      "loss": 0.8456,
      "step": 163190
    },
    {
      "epoch": 2.8652188416229216,
      "grad_norm": 0.05741284212655288,
      "learning_rate": 8.112917906498945e-05,
      "loss": 0.8518,
      "step": 163200
    },
    {
      "epoch": 2.865394406502923,
      "grad_norm": 0.048346226437310874,
      "learning_rate": 8.111826233561035e-05,
      "loss": 0.8505,
      "step": 163210
    },
    {
      "epoch": 2.8655699713829246,
      "grad_norm": 0.06683782646608724,
      "learning_rate": 8.110734585813664e-05,
      "loss": 0.8414,
      "step": 163220
    },
    {
      "epoch": 2.865745536262926,
      "grad_norm": 0.045426886313238976,
      "learning_rate": 8.10964296327067e-05,
      "loss": 0.8408,
      "step": 163230
    },
    {
      "epoch": 2.8659211011429275,
      "grad_norm": 0.06033805810401283,
      "learning_rate": 8.108551365945884e-05,
      "loss": 0.8562,
      "step": 163240
    },
    {
      "epoch": 2.866096666022929,
      "grad_norm": 0.05266856444507684,
      "learning_rate": 8.107459793853129e-05,
      "loss": 0.8488,
      "step": 163250
    },
    {
      "epoch": 2.86627223090293,
      "grad_norm": 0.08339892545292522,
      "learning_rate": 8.106368247006243e-05,
      "loss": 0.8341,
      "step": 163260
    },
    {
      "epoch": 2.8664477957829315,
      "grad_norm": 0.05614205030508958,
      "learning_rate": 8.105276725419053e-05,
      "loss": 0.8394,
      "step": 163270
    },
    {
      "epoch": 2.866623360662933,
      "grad_norm": 0.049346279324845366,
      "learning_rate": 8.104185229105385e-05,
      "loss": 0.8434,
      "step": 163280
    },
    {
      "epoch": 2.8667989255429345,
      "grad_norm": 0.05092970896104869,
      "learning_rate": 8.103093758079078e-05,
      "loss": 0.8476,
      "step": 163290
    },
    {
      "epoch": 2.866974490422936,
      "grad_norm": 0.052788375025162246,
      "learning_rate": 8.102002312353956e-05,
      "loss": 0.8449,
      "step": 163300
    },
    {
      "epoch": 2.867150055302937,
      "grad_norm": 0.07002463118425746,
      "learning_rate": 8.100910891943847e-05,
      "loss": 0.8405,
      "step": 163310
    },
    {
      "epoch": 2.8673256201829385,
      "grad_norm": 0.05196463676732262,
      "learning_rate": 8.099819496862576e-05,
      "loss": 0.8407,
      "step": 163320
    },
    {
      "epoch": 2.86750118506294,
      "grad_norm": 0.05830463412388009,
      "learning_rate": 8.09872812712398e-05,
      "loss": 0.8401,
      "step": 163330
    },
    {
      "epoch": 2.8676767499429414,
      "grad_norm": 0.04965013704044104,
      "learning_rate": 8.097636782741881e-05,
      "loss": 0.8428,
      "step": 163340
    },
    {
      "epoch": 2.867852314822943,
      "grad_norm": 0.05435896288585626,
      "learning_rate": 8.096545463730107e-05,
      "loss": 0.8504,
      "step": 163350
    },
    {
      "epoch": 2.8680278797029444,
      "grad_norm": 0.05605073974557314,
      "learning_rate": 8.095454170102487e-05,
      "loss": 0.8469,
      "step": 163360
    },
    {
      "epoch": 2.868203444582946,
      "grad_norm": 0.054974432845167336,
      "learning_rate": 8.094362901872846e-05,
      "loss": 0.833,
      "step": 163370
    },
    {
      "epoch": 2.868379009462947,
      "grad_norm": 0.04597222943413021,
      "learning_rate": 8.093271659055013e-05,
      "loss": 0.8416,
      "step": 163380
    },
    {
      "epoch": 2.8685545743429484,
      "grad_norm": 0.06056814201321721,
      "learning_rate": 8.092180441662808e-05,
      "loss": 0.8456,
      "step": 163390
    },
    {
      "epoch": 2.86873013922295,
      "grad_norm": 0.049990943562034706,
      "learning_rate": 8.091089249710065e-05,
      "loss": 0.8372,
      "step": 163400
    },
    {
      "epoch": 2.8689057041029513,
      "grad_norm": 0.051530910982830226,
      "learning_rate": 8.089998083210607e-05,
      "loss": 0.8436,
      "step": 163410
    },
    {
      "epoch": 2.869081268982953,
      "grad_norm": 0.057118034433227945,
      "learning_rate": 8.088906942178256e-05,
      "loss": 0.8393,
      "step": 163420
    },
    {
      "epoch": 2.869256833862954,
      "grad_norm": 0.04717821686598869,
      "learning_rate": 8.087815826626841e-05,
      "loss": 0.8359,
      "step": 163430
    },
    {
      "epoch": 2.8694323987429553,
      "grad_norm": 0.05591925332329214,
      "learning_rate": 8.086724736570185e-05,
      "loss": 0.8472,
      "step": 163440
    },
    {
      "epoch": 2.869607963622957,
      "grad_norm": 0.054229705204863234,
      "learning_rate": 8.085633672022111e-05,
      "loss": 0.8389,
      "step": 163450
    },
    {
      "epoch": 2.8697835285029583,
      "grad_norm": 0.04940609969856351,
      "learning_rate": 8.084542632996446e-05,
      "loss": 0.8433,
      "step": 163460
    },
    {
      "epoch": 2.8699590933829597,
      "grad_norm": 0.044581899690309415,
      "learning_rate": 8.083451619507013e-05,
      "loss": 0.8403,
      "step": 163470
    },
    {
      "epoch": 2.870134658262961,
      "grad_norm": 0.06504490892970459,
      "learning_rate": 8.082360631567634e-05,
      "loss": 0.8477,
      "step": 163480
    },
    {
      "epoch": 2.8703102231429627,
      "grad_norm": 0.062387831742608424,
      "learning_rate": 8.08126966919213e-05,
      "loss": 0.842,
      "step": 163490
    },
    {
      "epoch": 2.8704857880229637,
      "grad_norm": 0.059434905620544674,
      "learning_rate": 8.08017873239433e-05,
      "loss": 0.8323,
      "step": 163500
    },
    {
      "epoch": 2.870661352902965,
      "grad_norm": 0.05444103492686895,
      "learning_rate": 8.079087821188051e-05,
      "loss": 0.8471,
      "step": 163510
    },
    {
      "epoch": 2.8708369177829667,
      "grad_norm": 0.06124977438126976,
      "learning_rate": 8.077996935587116e-05,
      "loss": 0.8437,
      "step": 163520
    },
    {
      "epoch": 2.871012482662968,
      "grad_norm": 0.08425522350222941,
      "learning_rate": 8.076906075605347e-05,
      "loss": 0.8367,
      "step": 163530
    },
    {
      "epoch": 2.8711880475429696,
      "grad_norm": 0.05391009515541444,
      "learning_rate": 8.075815241256568e-05,
      "loss": 0.8476,
      "step": 163540
    },
    {
      "epoch": 2.8713636124229707,
      "grad_norm": 0.056470803750990596,
      "learning_rate": 8.074724432554596e-05,
      "loss": 0.8453,
      "step": 163550
    },
    {
      "epoch": 2.8715391773029726,
      "grad_norm": 0.05005768544533669,
      "learning_rate": 8.07363364951325e-05,
      "loss": 0.8473,
      "step": 163560
    },
    {
      "epoch": 2.8717147421829736,
      "grad_norm": 0.05020742185704079,
      "learning_rate": 8.07254289214636e-05,
      "loss": 0.844,
      "step": 163570
    },
    {
      "epoch": 2.871890307062975,
      "grad_norm": 0.04940352338640399,
      "learning_rate": 8.071452160467738e-05,
      "loss": 0.8383,
      "step": 163580
    },
    {
      "epoch": 2.8720658719429766,
      "grad_norm": 0.06140483222707047,
      "learning_rate": 8.070361454491205e-05,
      "loss": 0.8476,
      "step": 163590
    },
    {
      "epoch": 2.872241436822978,
      "grad_norm": 0.05612778976012454,
      "learning_rate": 8.069270774230582e-05,
      "loss": 0.8455,
      "step": 163600
    },
    {
      "epoch": 2.8724170017029795,
      "grad_norm": 0.06041962494337837,
      "learning_rate": 8.068180119699689e-05,
      "loss": 0.8422,
      "step": 163610
    },
    {
      "epoch": 2.8725925665829806,
      "grad_norm": 0.05754898274970149,
      "learning_rate": 8.067089490912343e-05,
      "loss": 0.8452,
      "step": 163620
    },
    {
      "epoch": 2.872768131462982,
      "grad_norm": 0.05176380510696448,
      "learning_rate": 8.065998887882358e-05,
      "loss": 0.8461,
      "step": 163630
    },
    {
      "epoch": 2.8729436963429835,
      "grad_norm": 0.05555859871069693,
      "learning_rate": 8.064908310623562e-05,
      "loss": 0.8401,
      "step": 163640
    },
    {
      "epoch": 2.873119261222985,
      "grad_norm": 0.0509709137396716,
      "learning_rate": 8.063817759149767e-05,
      "loss": 0.8476,
      "step": 163650
    },
    {
      "epoch": 2.8732948261029865,
      "grad_norm": 0.06647058964496766,
      "learning_rate": 8.062727233474788e-05,
      "loss": 0.8382,
      "step": 163660
    },
    {
      "epoch": 2.8734703909829875,
      "grad_norm": 0.06073905037657786,
      "learning_rate": 8.061636733612449e-05,
      "loss": 0.8512,
      "step": 163670
    },
    {
      "epoch": 2.8736459558629894,
      "grad_norm": 0.04360878303563667,
      "learning_rate": 8.060546259576564e-05,
      "loss": 0.8352,
      "step": 163680
    },
    {
      "epoch": 2.8738215207429905,
      "grad_norm": 0.049097171467727345,
      "learning_rate": 8.059455811380945e-05,
      "loss": 0.8353,
      "step": 163690
    },
    {
      "epoch": 2.873997085622992,
      "grad_norm": 0.05003799599546377,
      "learning_rate": 8.058365389039415e-05,
      "loss": 0.8425,
      "step": 163700
    },
    {
      "epoch": 2.8741726505029934,
      "grad_norm": 0.05288302778477292,
      "learning_rate": 8.057274992565787e-05,
      "loss": 0.8535,
      "step": 163710
    },
    {
      "epoch": 2.874348215382995,
      "grad_norm": 0.06503624905195027,
      "learning_rate": 8.056184621973875e-05,
      "loss": 0.8437,
      "step": 163720
    },
    {
      "epoch": 2.8745237802629964,
      "grad_norm": 0.06120406167658384,
      "learning_rate": 8.055094277277492e-05,
      "loss": 0.8404,
      "step": 163730
    },
    {
      "epoch": 2.8746993451429974,
      "grad_norm": 0.04874130447915148,
      "learning_rate": 8.05400395849046e-05,
      "loss": 0.8323,
      "step": 163740
    },
    {
      "epoch": 2.874874910022999,
      "grad_norm": 0.052575885676632704,
      "learning_rate": 8.052913665626591e-05,
      "loss": 0.8417,
      "step": 163750
    },
    {
      "epoch": 2.8750504749030004,
      "grad_norm": 0.057728949374124934,
      "learning_rate": 8.051823398699695e-05,
      "loss": 0.8459,
      "step": 163760
    },
    {
      "epoch": 2.875226039783002,
      "grad_norm": 0.056920732376910185,
      "learning_rate": 8.050733157723592e-05,
      "loss": 0.8461,
      "step": 163770
    },
    {
      "epoch": 2.8754016046630033,
      "grad_norm": 0.06928049898829447,
      "learning_rate": 8.049642942712093e-05,
      "loss": 0.8394,
      "step": 163780
    },
    {
      "epoch": 2.8755771695430044,
      "grad_norm": 0.0538632900877508,
      "learning_rate": 8.04855275367901e-05,
      "loss": 0.8398,
      "step": 163790
    },
    {
      "epoch": 2.8757527344230063,
      "grad_norm": 0.07533421882953373,
      "learning_rate": 8.047462590638154e-05,
      "loss": 0.8456,
      "step": 163800
    },
    {
      "epoch": 2.8759282993030073,
      "grad_norm": 0.06490630220946883,
      "learning_rate": 8.046372453603343e-05,
      "loss": 0.8345,
      "step": 163810
    },
    {
      "epoch": 2.876103864183009,
      "grad_norm": 0.06790292604353296,
      "learning_rate": 8.045282342588387e-05,
      "loss": 0.8441,
      "step": 163820
    },
    {
      "epoch": 2.8762794290630103,
      "grad_norm": 0.06781292671873275,
      "learning_rate": 8.044192257607094e-05,
      "loss": 0.8287,
      "step": 163830
    },
    {
      "epoch": 2.8764549939430117,
      "grad_norm": 0.05614155479539132,
      "learning_rate": 8.043102198673283e-05,
      "loss": 0.8436,
      "step": 163840
    },
    {
      "epoch": 2.876630558823013,
      "grad_norm": 0.051163105613065835,
      "learning_rate": 8.042012165800763e-05,
      "loss": 0.8463,
      "step": 163850
    },
    {
      "epoch": 2.8768061237030143,
      "grad_norm": 0.04492574894355648,
      "learning_rate": 8.040922159003339e-05,
      "loss": 0.8424,
      "step": 163860
    },
    {
      "epoch": 2.8769816885830157,
      "grad_norm": 0.057647834773502035,
      "learning_rate": 8.039832178294828e-05,
      "loss": 0.8444,
      "step": 163870
    },
    {
      "epoch": 2.877157253463017,
      "grad_norm": 0.046084739090995315,
      "learning_rate": 8.038742223689038e-05,
      "loss": 0.8382,
      "step": 163880
    },
    {
      "epoch": 2.8773328183430187,
      "grad_norm": 0.05420621571128835,
      "learning_rate": 8.03765229519978e-05,
      "loss": 0.8374,
      "step": 163890
    },
    {
      "epoch": 2.87750838322302,
      "grad_norm": 0.053182929121883736,
      "learning_rate": 8.036562392840858e-05,
      "loss": 0.8458,
      "step": 163900
    },
    {
      "epoch": 2.877683948103021,
      "grad_norm": 0.044492395965217056,
      "learning_rate": 8.035472516626089e-05,
      "loss": 0.8416,
      "step": 163910
    },
    {
      "epoch": 2.877859512983023,
      "grad_norm": 0.05213957798619921,
      "learning_rate": 8.03438266656928e-05,
      "loss": 0.8452,
      "step": 163920
    },
    {
      "epoch": 2.878035077863024,
      "grad_norm": 0.04565662268389031,
      "learning_rate": 8.033292842684237e-05,
      "loss": 0.8473,
      "step": 163930
    },
    {
      "epoch": 2.8782106427430256,
      "grad_norm": 0.04857716028988213,
      "learning_rate": 8.03220304498477e-05,
      "loss": 0.8342,
      "step": 163940
    },
    {
      "epoch": 2.878386207623027,
      "grad_norm": 0.0716437913526914,
      "learning_rate": 8.031113273484689e-05,
      "loss": 0.8439,
      "step": 163950
    },
    {
      "epoch": 2.8785617725030286,
      "grad_norm": 0.05451283660295179,
      "learning_rate": 8.030023528197797e-05,
      "loss": 0.8442,
      "step": 163960
    },
    {
      "epoch": 2.87873733738303,
      "grad_norm": 0.06507232167358691,
      "learning_rate": 8.0289338091379e-05,
      "loss": 0.8465,
      "step": 163970
    },
    {
      "epoch": 2.878912902263031,
      "grad_norm": 0.05702443535893696,
      "learning_rate": 8.027844116318814e-05,
      "loss": 0.8395,
      "step": 163980
    },
    {
      "epoch": 2.8790884671430326,
      "grad_norm": 0.04772749741698408,
      "learning_rate": 8.02675444975434e-05,
      "loss": 0.8424,
      "step": 163990
    },
    {
      "epoch": 2.879264032023034,
      "grad_norm": 0.05862817848584737,
      "learning_rate": 8.025664809458281e-05,
      "loss": 0.8404,
      "step": 164000
    },
    {
      "epoch": 2.8794395969030355,
      "grad_norm": 0.06464179080571206,
      "learning_rate": 8.02457519544445e-05,
      "loss": 0.8413,
      "step": 164010
    },
    {
      "epoch": 2.879615161783037,
      "grad_norm": 0.0525854322260668,
      "learning_rate": 8.023485607726648e-05,
      "loss": 0.8347,
      "step": 164020
    },
    {
      "epoch": 2.8797907266630385,
      "grad_norm": 0.049699587562264544,
      "learning_rate": 8.02239604631868e-05,
      "loss": 0.8476,
      "step": 164030
    },
    {
      "epoch": 2.87996629154304,
      "grad_norm": 0.050489485106475246,
      "learning_rate": 8.021306511234354e-05,
      "loss": 0.8411,
      "step": 164040
    },
    {
      "epoch": 2.880141856423041,
      "grad_norm": 0.05773654814520366,
      "learning_rate": 8.020217002487473e-05,
      "loss": 0.8442,
      "step": 164050
    },
    {
      "epoch": 2.8803174213030425,
      "grad_norm": 0.049342532508366344,
      "learning_rate": 8.019127520091841e-05,
      "loss": 0.8464,
      "step": 164060
    },
    {
      "epoch": 2.880492986183044,
      "grad_norm": 0.05063845328365805,
      "learning_rate": 8.018038064061258e-05,
      "loss": 0.8413,
      "step": 164070
    },
    {
      "epoch": 2.8806685510630454,
      "grad_norm": 0.05464963244407542,
      "learning_rate": 8.016948634409535e-05,
      "loss": 0.8449,
      "step": 164080
    },
    {
      "epoch": 2.880844115943047,
      "grad_norm": 0.05716379777508978,
      "learning_rate": 8.015859231150474e-05,
      "loss": 0.8444,
      "step": 164090
    },
    {
      "epoch": 2.881019680823048,
      "grad_norm": 0.06119433129876756,
      "learning_rate": 8.014769854297873e-05,
      "loss": 0.8509,
      "step": 164100
    },
    {
      "epoch": 2.8811952457030494,
      "grad_norm": 0.05181638914098098,
      "learning_rate": 8.013680503865539e-05,
      "loss": 0.8444,
      "step": 164110
    },
    {
      "epoch": 2.881370810583051,
      "grad_norm": 0.0554875437189586,
      "learning_rate": 8.012591179867272e-05,
      "loss": 0.8414,
      "step": 164120
    },
    {
      "epoch": 2.8815463754630524,
      "grad_norm": 0.046560522920289116,
      "learning_rate": 8.011501882316877e-05,
      "loss": 0.8388,
      "step": 164130
    },
    {
      "epoch": 2.881721940343054,
      "grad_norm": 0.054509868678674474,
      "learning_rate": 8.010412611228148e-05,
      "loss": 0.8453,
      "step": 164140
    },
    {
      "epoch": 2.8818975052230553,
      "grad_norm": 0.07593517343067131,
      "learning_rate": 8.009323366614896e-05,
      "loss": 0.8505,
      "step": 164150
    },
    {
      "epoch": 2.882073070103057,
      "grad_norm": 0.09009387524857683,
      "learning_rate": 8.008234148490918e-05,
      "loss": 0.8398,
      "step": 164160
    },
    {
      "epoch": 2.882248634983058,
      "grad_norm": 0.06679378239060768,
      "learning_rate": 8.007144956870012e-05,
      "loss": 0.8408,
      "step": 164170
    },
    {
      "epoch": 2.8824241998630593,
      "grad_norm": 0.050529121174976675,
      "learning_rate": 8.006055791765983e-05,
      "loss": 0.8483,
      "step": 164180
    },
    {
      "epoch": 2.882599764743061,
      "grad_norm": 0.050952368641644485,
      "learning_rate": 8.004966653192628e-05,
      "loss": 0.8353,
      "step": 164190
    },
    {
      "epoch": 2.8827753296230623,
      "grad_norm": 0.050381294388273594,
      "learning_rate": 8.003877541163744e-05,
      "loss": 0.8443,
      "step": 164200
    },
    {
      "epoch": 2.8829508945030637,
      "grad_norm": 0.04964273682025577,
      "learning_rate": 8.002788455693136e-05,
      "loss": 0.8455,
      "step": 164210
    },
    {
      "epoch": 2.8831264593830648,
      "grad_norm": 0.06301153018468897,
      "learning_rate": 8.001699396794601e-05,
      "loss": 0.8444,
      "step": 164220
    },
    {
      "epoch": 2.8833020242630663,
      "grad_norm": 0.04969106786133441,
      "learning_rate": 8.000610364481936e-05,
      "loss": 0.8471,
      "step": 164230
    },
    {
      "epoch": 2.8834775891430677,
      "grad_norm": 0.050907621357371556,
      "learning_rate": 7.999521358768936e-05,
      "loss": 0.8385,
      "step": 164240
    },
    {
      "epoch": 2.883653154023069,
      "grad_norm": 0.11008279542007941,
      "learning_rate": 7.998432379669407e-05,
      "loss": 0.8453,
      "step": 164250
    },
    {
      "epoch": 2.8838287189030707,
      "grad_norm": 0.053294209910177394,
      "learning_rate": 7.997343427197143e-05,
      "loss": 0.843,
      "step": 164260
    },
    {
      "epoch": 2.884004283783072,
      "grad_norm": 0.05525098234896964,
      "learning_rate": 7.996254501365938e-05,
      "loss": 0.8443,
      "step": 164270
    },
    {
      "epoch": 2.8841798486630736,
      "grad_norm": 0.06409245107831955,
      "learning_rate": 7.995165602189595e-05,
      "loss": 0.8404,
      "step": 164280
    },
    {
      "epoch": 2.8843554135430747,
      "grad_norm": 0.04288998488917623,
      "learning_rate": 7.994076729681907e-05,
      "loss": 0.8468,
      "step": 164290
    },
    {
      "epoch": 2.884530978423076,
      "grad_norm": 0.044541377236568376,
      "learning_rate": 7.992987883856669e-05,
      "loss": 0.847,
      "step": 164300
    },
    {
      "epoch": 2.8847065433030776,
      "grad_norm": 0.04797457344656801,
      "learning_rate": 7.991899064727675e-05,
      "loss": 0.8401,
      "step": 164310
    },
    {
      "epoch": 2.884882108183079,
      "grad_norm": 0.04809825096581969,
      "learning_rate": 7.990810272308726e-05,
      "loss": 0.8388,
      "step": 164320
    },
    {
      "epoch": 2.8850576730630806,
      "grad_norm": 0.06618928809774713,
      "learning_rate": 7.989721506613618e-05,
      "loss": 0.8439,
      "step": 164330
    },
    {
      "epoch": 2.8852332379430816,
      "grad_norm": 0.052838659786869364,
      "learning_rate": 7.98863276765614e-05,
      "loss": 0.8513,
      "step": 164340
    },
    {
      "epoch": 2.8854088028230835,
      "grad_norm": 0.052577542515685846,
      "learning_rate": 7.987544055450091e-05,
      "loss": 0.8449,
      "step": 164350
    },
    {
      "epoch": 2.8855843677030846,
      "grad_norm": 0.04481390825164697,
      "learning_rate": 7.986455370009265e-05,
      "loss": 0.8399,
      "step": 164360
    },
    {
      "epoch": 2.885759932583086,
      "grad_norm": 0.04113300278070752,
      "learning_rate": 7.985366711347451e-05,
      "loss": 0.8374,
      "step": 164370
    },
    {
      "epoch": 2.8859354974630875,
      "grad_norm": 0.042794710202876854,
      "learning_rate": 7.984278079478448e-05,
      "loss": 0.8432,
      "step": 164380
    },
    {
      "epoch": 2.886111062343089,
      "grad_norm": 0.06162722694628333,
      "learning_rate": 7.983189474416048e-05,
      "loss": 0.838,
      "step": 164390
    },
    {
      "epoch": 2.8862866272230905,
      "grad_norm": 0.05913161594201383,
      "learning_rate": 7.982100896174043e-05,
      "loss": 0.8385,
      "step": 164400
    },
    {
      "epoch": 2.8864621921030915,
      "grad_norm": 0.052233208898119485,
      "learning_rate": 7.981012344766222e-05,
      "loss": 0.8327,
      "step": 164410
    },
    {
      "epoch": 2.886637756983093,
      "grad_norm": 0.044276303062170626,
      "learning_rate": 7.979923820206383e-05,
      "loss": 0.8402,
      "step": 164420
    },
    {
      "epoch": 2.8868133218630945,
      "grad_norm": 0.052317884569435306,
      "learning_rate": 7.978835322508316e-05,
      "loss": 0.8353,
      "step": 164430
    },
    {
      "epoch": 2.886988886743096,
      "grad_norm": 0.06436234771140761,
      "learning_rate": 7.977746851685811e-05,
      "loss": 0.8448,
      "step": 164440
    },
    {
      "epoch": 2.8871644516230974,
      "grad_norm": 0.048543116616244546,
      "learning_rate": 7.976658407752661e-05,
      "loss": 0.8498,
      "step": 164450
    },
    {
      "epoch": 2.8873400165030985,
      "grad_norm": 0.04556062547954166,
      "learning_rate": 7.975569990722657e-05,
      "loss": 0.8423,
      "step": 164460
    },
    {
      "epoch": 2.8875155813831004,
      "grad_norm": 0.045746954669165556,
      "learning_rate": 7.974481600609589e-05,
      "loss": 0.8372,
      "step": 164470
    },
    {
      "epoch": 2.8876911462631014,
      "grad_norm": 0.06848181654166628,
      "learning_rate": 7.97339323742724e-05,
      "loss": 0.8434,
      "step": 164480
    },
    {
      "epoch": 2.887866711143103,
      "grad_norm": 0.04428072130418775,
      "learning_rate": 7.972304901189412e-05,
      "loss": 0.8457,
      "step": 164490
    },
    {
      "epoch": 2.8880422760231044,
      "grad_norm": 0.0660548780984429,
      "learning_rate": 7.971216591909889e-05,
      "loss": 0.8483,
      "step": 164500
    },
    {
      "epoch": 2.888217840903106,
      "grad_norm": 0.049371849817551695,
      "learning_rate": 7.970128309602458e-05,
      "loss": 0.8338,
      "step": 164510
    },
    {
      "epoch": 2.8883934057831073,
      "grad_norm": 0.057561847424960365,
      "learning_rate": 7.96904005428091e-05,
      "loss": 0.8418,
      "step": 164520
    },
    {
      "epoch": 2.8885689706631084,
      "grad_norm": 0.04417383830317901,
      "learning_rate": 7.967951825959033e-05,
      "loss": 0.8446,
      "step": 164530
    },
    {
      "epoch": 2.88874453554311,
      "grad_norm": 0.04841890504133837,
      "learning_rate": 7.966863624650615e-05,
      "loss": 0.8431,
      "step": 164540
    },
    {
      "epoch": 2.8889201004231113,
      "grad_norm": 0.057932762906252495,
      "learning_rate": 7.965775450369444e-05,
      "loss": 0.8431,
      "step": 164550
    },
    {
      "epoch": 2.889095665303113,
      "grad_norm": 0.04487903632591741,
      "learning_rate": 7.964687303129305e-05,
      "loss": 0.8417,
      "step": 164560
    },
    {
      "epoch": 2.8892712301831143,
      "grad_norm": 0.045707941855346525,
      "learning_rate": 7.96359918294399e-05,
      "loss": 0.8387,
      "step": 164570
    },
    {
      "epoch": 2.8894467950631153,
      "grad_norm": 0.1779684039108651,
      "learning_rate": 7.96251108982728e-05,
      "loss": 0.8374,
      "step": 164580
    },
    {
      "epoch": 2.8896223599431172,
      "grad_norm": 0.06519482879720584,
      "learning_rate": 7.961423023792965e-05,
      "loss": 0.8375,
      "step": 164590
    },
    {
      "epoch": 2.8897979248231183,
      "grad_norm": 0.07128934392250127,
      "learning_rate": 7.960334984854831e-05,
      "loss": 0.8442,
      "step": 164600
    },
    {
      "epoch": 2.8899734897031197,
      "grad_norm": 0.042749673591414114,
      "learning_rate": 7.959246973026661e-05,
      "loss": 0.8464,
      "step": 164610
    },
    {
      "epoch": 2.890149054583121,
      "grad_norm": 0.04893387632849866,
      "learning_rate": 7.958158988322245e-05,
      "loss": 0.8295,
      "step": 164620
    },
    {
      "epoch": 2.8903246194631227,
      "grad_norm": 0.06155614682734781,
      "learning_rate": 7.957071030755366e-05,
      "loss": 0.8356,
      "step": 164630
    },
    {
      "epoch": 2.890500184343124,
      "grad_norm": 0.07078123724694628,
      "learning_rate": 7.955983100339804e-05,
      "loss": 0.8317,
      "step": 164640
    },
    {
      "epoch": 2.890675749223125,
      "grad_norm": 0.06126176235016261,
      "learning_rate": 7.954895197089345e-05,
      "loss": 0.8438,
      "step": 164650
    },
    {
      "epoch": 2.8908513141031267,
      "grad_norm": 0.06533619459505607,
      "learning_rate": 7.95380732101778e-05,
      "loss": 0.8383,
      "step": 164660
    },
    {
      "epoch": 2.891026878983128,
      "grad_norm": 0.05487284975067578,
      "learning_rate": 7.952719472138886e-05,
      "loss": 0.8407,
      "step": 164670
    },
    {
      "epoch": 2.8912024438631296,
      "grad_norm": 0.04537778914844278,
      "learning_rate": 7.951631650466447e-05,
      "loss": 0.8461,
      "step": 164680
    },
    {
      "epoch": 2.891378008743131,
      "grad_norm": 0.04703510405876707,
      "learning_rate": 7.950543856014248e-05,
      "loss": 0.8425,
      "step": 164690
    },
    {
      "epoch": 2.8915535736231326,
      "grad_norm": 0.04825765736566031,
      "learning_rate": 7.94945608879607e-05,
      "loss": 0.8354,
      "step": 164700
    },
    {
      "epoch": 2.891729138503134,
      "grad_norm": 0.04541220782718501,
      "learning_rate": 7.948368348825696e-05,
      "loss": 0.8383,
      "step": 164710
    },
    {
      "epoch": 2.891904703383135,
      "grad_norm": 0.04527739055181864,
      "learning_rate": 7.947280636116907e-05,
      "loss": 0.8461,
      "step": 164720
    },
    {
      "epoch": 2.8920802682631366,
      "grad_norm": 0.049083591717485095,
      "learning_rate": 7.946192950683485e-05,
      "loss": 0.8475,
      "step": 164730
    },
    {
      "epoch": 2.892255833143138,
      "grad_norm": 0.04790645419383023,
      "learning_rate": 7.945105292539212e-05,
      "loss": 0.8477,
      "step": 164740
    },
    {
      "epoch": 2.8924313980231395,
      "grad_norm": 0.061216821428596045,
      "learning_rate": 7.944017661697862e-05,
      "loss": 0.8341,
      "step": 164750
    },
    {
      "epoch": 2.892606962903141,
      "grad_norm": 0.0638818083442616,
      "learning_rate": 7.942930058173226e-05,
      "loss": 0.8373,
      "step": 164760
    },
    {
      "epoch": 2.892782527783142,
      "grad_norm": 0.047463435283314204,
      "learning_rate": 7.941842481979081e-05,
      "loss": 0.844,
      "step": 164770
    },
    {
      "epoch": 2.8929580926631435,
      "grad_norm": 0.0751329154988648,
      "learning_rate": 7.940754933129204e-05,
      "loss": 0.8386,
      "step": 164780
    },
    {
      "epoch": 2.893133657543145,
      "grad_norm": 0.05049052898937298,
      "learning_rate": 7.939667411637375e-05,
      "loss": 0.847,
      "step": 164790
    },
    {
      "epoch": 2.8933092224231465,
      "grad_norm": 0.047994846908921474,
      "learning_rate": 7.938579917517376e-05,
      "loss": 0.8427,
      "step": 164800
    },
    {
      "epoch": 2.893484787303148,
      "grad_norm": 0.07289777191896038,
      "learning_rate": 7.937492450782985e-05,
      "loss": 0.8428,
      "step": 164810
    },
    {
      "epoch": 2.8936603521831494,
      "grad_norm": 0.05601936831815173,
      "learning_rate": 7.93640501144797e-05,
      "loss": 0.84,
      "step": 164820
    },
    {
      "epoch": 2.893835917063151,
      "grad_norm": 0.045425698523703845,
      "learning_rate": 7.935317599526126e-05,
      "loss": 0.8432,
      "step": 164830
    },
    {
      "epoch": 2.894011481943152,
      "grad_norm": 0.04651845258407664,
      "learning_rate": 7.934230215031223e-05,
      "loss": 0.8414,
      "step": 164840
    },
    {
      "epoch": 2.8941870468231534,
      "grad_norm": 0.04728872284449045,
      "learning_rate": 7.933142857977035e-05,
      "loss": 0.8504,
      "step": 164850
    },
    {
      "epoch": 2.894362611703155,
      "grad_norm": 0.05601568586353126,
      "learning_rate": 7.932055528377345e-05,
      "loss": 0.8451,
      "step": 164860
    },
    {
      "epoch": 2.8945381765831564,
      "grad_norm": 0.054008424216360026,
      "learning_rate": 7.930968226245927e-05,
      "loss": 0.8407,
      "step": 164870
    },
    {
      "epoch": 2.894713741463158,
      "grad_norm": 0.0532644645711892,
      "learning_rate": 7.929880951596554e-05,
      "loss": 0.8432,
      "step": 164880
    },
    {
      "epoch": 2.894889306343159,
      "grad_norm": 0.0507627907228056,
      "learning_rate": 7.928793704443009e-05,
      "loss": 0.8508,
      "step": 164890
    },
    {
      "epoch": 2.8950648712231604,
      "grad_norm": 0.053066995405196854,
      "learning_rate": 7.927706484799062e-05,
      "loss": 0.8379,
      "step": 164900
    },
    {
      "epoch": 2.895240436103162,
      "grad_norm": 0.06185318189538912,
      "learning_rate": 7.926619292678492e-05,
      "loss": 0.8389,
      "step": 164910
    },
    {
      "epoch": 2.8954160009831633,
      "grad_norm": 0.047291741811332375,
      "learning_rate": 7.925532128095067e-05,
      "loss": 0.8398,
      "step": 164920
    },
    {
      "epoch": 2.895591565863165,
      "grad_norm": 0.04449707001193786,
      "learning_rate": 7.924444991062571e-05,
      "loss": 0.8396,
      "step": 164930
    },
    {
      "epoch": 2.8957671307431663,
      "grad_norm": 0.058939464900428264,
      "learning_rate": 7.923357881594775e-05,
      "loss": 0.8431,
      "step": 164940
    },
    {
      "epoch": 2.8959426956231678,
      "grad_norm": 0.054200014118717915,
      "learning_rate": 7.922270799705448e-05,
      "loss": 0.8418,
      "step": 164950
    },
    {
      "epoch": 2.896118260503169,
      "grad_norm": 0.05837151621951188,
      "learning_rate": 7.92118374540837e-05,
      "loss": 0.8412,
      "step": 164960
    },
    {
      "epoch": 2.8962938253831703,
      "grad_norm": 0.06107232668243562,
      "learning_rate": 7.920096718717311e-05,
      "loss": 0.8521,
      "step": 164970
    },
    {
      "epoch": 2.8964693902631717,
      "grad_norm": 0.048827021047823396,
      "learning_rate": 7.919009719646045e-05,
      "loss": 0.8323,
      "step": 164980
    },
    {
      "epoch": 2.896644955143173,
      "grad_norm": 0.05161919529471505,
      "learning_rate": 7.917922748208339e-05,
      "loss": 0.843,
      "step": 164990
    },
    {
      "epoch": 2.8968205200231747,
      "grad_norm": 0.06033225716024112,
      "learning_rate": 7.916835804417975e-05,
      "loss": 0.8486,
      "step": 165000
    },
    {
      "epoch": 2.8969960849031757,
      "grad_norm": 0.05739356371418361,
      "learning_rate": 7.91574888828872e-05,
      "loss": 0.8508,
      "step": 165010
    },
    {
      "epoch": 2.8971716497831776,
      "grad_norm": 0.0597487299890794,
      "learning_rate": 7.914661999834341e-05,
      "loss": 0.8369,
      "step": 165020
    },
    {
      "epoch": 2.8973472146631787,
      "grad_norm": 0.05735902193027662,
      "learning_rate": 7.913575139068618e-05,
      "loss": 0.8468,
      "step": 165030
    },
    {
      "epoch": 2.89752277954318,
      "grad_norm": 0.04900408101297695,
      "learning_rate": 7.912488306005315e-05,
      "loss": 0.8473,
      "step": 165040
    },
    {
      "epoch": 2.8976983444231816,
      "grad_norm": 0.05548607137323572,
      "learning_rate": 7.911401500658204e-05,
      "loss": 0.8378,
      "step": 165050
    },
    {
      "epoch": 2.897873909303183,
      "grad_norm": 0.05598183534115939,
      "learning_rate": 7.910314723041058e-05,
      "loss": 0.8357,
      "step": 165060
    },
    {
      "epoch": 2.8980494741831846,
      "grad_norm": 0.047376233464168076,
      "learning_rate": 7.909227973167642e-05,
      "loss": 0.8385,
      "step": 165070
    },
    {
      "epoch": 2.8982250390631856,
      "grad_norm": 0.06127089571989709,
      "learning_rate": 7.908141251051728e-05,
      "loss": 0.842,
      "step": 165080
    },
    {
      "epoch": 2.898400603943187,
      "grad_norm": 0.05207108200068786,
      "learning_rate": 7.907054556707082e-05,
      "loss": 0.8389,
      "step": 165090
    },
    {
      "epoch": 2.8985761688231886,
      "grad_norm": 0.05810174549344937,
      "learning_rate": 7.905967890147478e-05,
      "loss": 0.8392,
      "step": 165100
    },
    {
      "epoch": 2.89875173370319,
      "grad_norm": 0.05308136493040261,
      "learning_rate": 7.90488125138668e-05,
      "loss": 0.8461,
      "step": 165110
    },
    {
      "epoch": 2.8989272985831915,
      "grad_norm": 0.05499762315284451,
      "learning_rate": 7.903794640438457e-05,
      "loss": 0.8418,
      "step": 165120
    },
    {
      "epoch": 2.8991028634631926,
      "grad_norm": 0.07884283117068985,
      "learning_rate": 7.902708057316578e-05,
      "loss": 0.8413,
      "step": 165130
    },
    {
      "epoch": 2.8992784283431945,
      "grad_norm": 0.06470154533631292,
      "learning_rate": 7.901621502034809e-05,
      "loss": 0.8444,
      "step": 165140
    },
    {
      "epoch": 2.8994539932231955,
      "grad_norm": 0.05180766531293872,
      "learning_rate": 7.900534974606916e-05,
      "loss": 0.838,
      "step": 165150
    },
    {
      "epoch": 2.899629558103197,
      "grad_norm": 0.06182557993704867,
      "learning_rate": 7.899448475046664e-05,
      "loss": 0.8378,
      "step": 165160
    },
    {
      "epoch": 2.8998051229831985,
      "grad_norm": 0.04653752642248593,
      "learning_rate": 7.898362003367825e-05,
      "loss": 0.8443,
      "step": 165170
    },
    {
      "epoch": 2.8999806878632,
      "grad_norm": 0.05409006350120545,
      "learning_rate": 7.897275559584161e-05,
      "loss": 0.8381,
      "step": 165180
    },
    {
      "epoch": 2.9001562527432014,
      "grad_norm": 0.052743204096247595,
      "learning_rate": 7.896189143709437e-05,
      "loss": 0.8412,
      "step": 165190
    },
    {
      "epoch": 2.9003318176232025,
      "grad_norm": 0.04892632007477496,
      "learning_rate": 7.895102755757418e-05,
      "loss": 0.836,
      "step": 165200
    },
    {
      "epoch": 2.900507382503204,
      "grad_norm": 0.07529828766102573,
      "learning_rate": 7.894016395741873e-05,
      "loss": 0.8383,
      "step": 165210
    },
    {
      "epoch": 2.9006829473832054,
      "grad_norm": 0.06498285067366927,
      "learning_rate": 7.892930063676561e-05,
      "loss": 0.8377,
      "step": 165220
    },
    {
      "epoch": 2.900858512263207,
      "grad_norm": 0.05387818699544917,
      "learning_rate": 7.891843759575249e-05,
      "loss": 0.8486,
      "step": 165230
    },
    {
      "epoch": 2.9010340771432084,
      "grad_norm": 0.06230406054209931,
      "learning_rate": 7.890757483451699e-05,
      "loss": 0.8428,
      "step": 165240
    },
    {
      "epoch": 2.9012096420232094,
      "grad_norm": 0.06493074229677168,
      "learning_rate": 7.889671235319676e-05,
      "loss": 0.8401,
      "step": 165250
    },
    {
      "epoch": 2.9013852069032113,
      "grad_norm": 0.04902074329873495,
      "learning_rate": 7.888585015192937e-05,
      "loss": 0.8494,
      "step": 165260
    },
    {
      "epoch": 2.9015607717832124,
      "grad_norm": 0.061237183083524556,
      "learning_rate": 7.887498823085254e-05,
      "loss": 0.8438,
      "step": 165270
    },
    {
      "epoch": 2.901736336663214,
      "grad_norm": 0.06198898877944065,
      "learning_rate": 7.886412659010386e-05,
      "loss": 0.8428,
      "step": 165280
    },
    {
      "epoch": 2.9019119015432153,
      "grad_norm": 0.04848012839689154,
      "learning_rate": 7.885326522982092e-05,
      "loss": 0.8433,
      "step": 165290
    },
    {
      "epoch": 2.902087466423217,
      "grad_norm": 0.061282570575668896,
      "learning_rate": 7.884240415014136e-05,
      "loss": 0.8438,
      "step": 165300
    },
    {
      "epoch": 2.9022630313032183,
      "grad_norm": 0.04579474890976249,
      "learning_rate": 7.88315433512028e-05,
      "loss": 0.8445,
      "step": 165310
    },
    {
      "epoch": 2.9024385961832193,
      "grad_norm": 0.0872503554614973,
      "learning_rate": 7.88206828331428e-05,
      "loss": 0.8456,
      "step": 165320
    },
    {
      "epoch": 2.902614161063221,
      "grad_norm": 0.05990792300212447,
      "learning_rate": 7.880982259609901e-05,
      "loss": 0.844,
      "step": 165330
    },
    {
      "epoch": 2.9027897259432223,
      "grad_norm": 0.054948028245571885,
      "learning_rate": 7.879896264020903e-05,
      "loss": 0.841,
      "step": 165340
    },
    {
      "epoch": 2.9029652908232237,
      "grad_norm": 0.04383429433797701,
      "learning_rate": 7.878810296561045e-05,
      "loss": 0.8436,
      "step": 165350
    },
    {
      "epoch": 2.903140855703225,
      "grad_norm": 0.049726534044412844,
      "learning_rate": 7.877724357244084e-05,
      "loss": 0.8368,
      "step": 165360
    },
    {
      "epoch": 2.9033164205832263,
      "grad_norm": 0.06542284634239183,
      "learning_rate": 7.876638446083784e-05,
      "loss": 0.8402,
      "step": 165370
    },
    {
      "epoch": 2.903491985463228,
      "grad_norm": 0.05349988315403465,
      "learning_rate": 7.8755525630939e-05,
      "loss": 0.8542,
      "step": 165380
    },
    {
      "epoch": 2.903667550343229,
      "grad_norm": 0.04880275663184911,
      "learning_rate": 7.874466708288191e-05,
      "loss": 0.8429,
      "step": 165390
    },
    {
      "epoch": 2.9038431152232307,
      "grad_norm": 0.058924372443803025,
      "learning_rate": 7.873380881680416e-05,
      "loss": 0.8376,
      "step": 165400
    },
    {
      "epoch": 2.904018680103232,
      "grad_norm": 0.06039927239169922,
      "learning_rate": 7.872295083284334e-05,
      "loss": 0.8456,
      "step": 165410
    },
    {
      "epoch": 2.9041942449832336,
      "grad_norm": 0.046543890782821125,
      "learning_rate": 7.871209313113699e-05,
      "loss": 0.8444,
      "step": 165420
    },
    {
      "epoch": 2.904369809863235,
      "grad_norm": 0.05398777329956209,
      "learning_rate": 7.870123571182264e-05,
      "loss": 0.8413,
      "step": 165430
    },
    {
      "epoch": 2.904545374743236,
      "grad_norm": 0.05696068515737202,
      "learning_rate": 7.869037857503796e-05,
      "loss": 0.8394,
      "step": 165440
    },
    {
      "epoch": 2.9047209396232376,
      "grad_norm": 0.046054667280499326,
      "learning_rate": 7.867952172092045e-05,
      "loss": 0.8462,
      "step": 165450
    },
    {
      "epoch": 2.904896504503239,
      "grad_norm": 0.04569259050513103,
      "learning_rate": 7.866866514960766e-05,
      "loss": 0.8439,
      "step": 165460
    },
    {
      "epoch": 2.9050720693832406,
      "grad_norm": 0.0524147587428973,
      "learning_rate": 7.865780886123719e-05,
      "loss": 0.8251,
      "step": 165470
    },
    {
      "epoch": 2.905247634263242,
      "grad_norm": 0.042754109970196075,
      "learning_rate": 7.864695285594655e-05,
      "loss": 0.8406,
      "step": 165480
    },
    {
      "epoch": 2.9054231991432435,
      "grad_norm": 0.04896570510736805,
      "learning_rate": 7.863609713387332e-05,
      "loss": 0.8362,
      "step": 165490
    },
    {
      "epoch": 2.905598764023245,
      "grad_norm": 0.044885225661702076,
      "learning_rate": 7.862524169515497e-05,
      "loss": 0.8363,
      "step": 165500
    },
    {
      "epoch": 2.905774328903246,
      "grad_norm": 0.056392219419541215,
      "learning_rate": 7.861438653992914e-05,
      "loss": 0.852,
      "step": 165510
    },
    {
      "epoch": 2.9059498937832475,
      "grad_norm": 0.037288573692969375,
      "learning_rate": 7.860353166833333e-05,
      "loss": 0.8431,
      "step": 165520
    },
    {
      "epoch": 2.906125458663249,
      "grad_norm": 0.052324638957516684,
      "learning_rate": 7.859267708050503e-05,
      "loss": 0.8471,
      "step": 165530
    },
    {
      "epoch": 2.9063010235432505,
      "grad_norm": 0.05192738990188919,
      "learning_rate": 7.858182277658184e-05,
      "loss": 0.8534,
      "step": 165540
    },
    {
      "epoch": 2.906476588423252,
      "grad_norm": 0.04237851082354374,
      "learning_rate": 7.857096875670123e-05,
      "loss": 0.8378,
      "step": 165550
    },
    {
      "epoch": 2.906652153303253,
      "grad_norm": 0.04767650915085455,
      "learning_rate": 7.856011502100075e-05,
      "loss": 0.8432,
      "step": 165560
    },
    {
      "epoch": 2.9068277181832545,
      "grad_norm": 0.05300932904915909,
      "learning_rate": 7.854926156961793e-05,
      "loss": 0.8421,
      "step": 165570
    },
    {
      "epoch": 2.907003283063256,
      "grad_norm": 0.06475118633688669,
      "learning_rate": 7.853840840269027e-05,
      "loss": 0.8375,
      "step": 165580
    },
    {
      "epoch": 2.9071788479432574,
      "grad_norm": 0.04989486812906141,
      "learning_rate": 7.852755552035526e-05,
      "loss": 0.8437,
      "step": 165590
    },
    {
      "epoch": 2.907354412823259,
      "grad_norm": 0.061831890170899816,
      "learning_rate": 7.851670292275041e-05,
      "loss": 0.846,
      "step": 165600
    },
    {
      "epoch": 2.9075299777032604,
      "grad_norm": 0.05452765804967601,
      "learning_rate": 7.850585061001327e-05,
      "loss": 0.8395,
      "step": 165610
    },
    {
      "epoch": 2.907705542583262,
      "grad_norm": 0.05869293199194346,
      "learning_rate": 7.849499858228131e-05,
      "loss": 0.8447,
      "step": 165620
    },
    {
      "epoch": 2.907881107463263,
      "grad_norm": 0.06832213540372248,
      "learning_rate": 7.848414683969205e-05,
      "loss": 0.8448,
      "step": 165630
    },
    {
      "epoch": 2.9080566723432644,
      "grad_norm": 0.06248404787433853,
      "learning_rate": 7.847329538238293e-05,
      "loss": 0.8428,
      "step": 165640
    },
    {
      "epoch": 2.908232237223266,
      "grad_norm": 0.06917811180229637,
      "learning_rate": 7.84624442104915e-05,
      "loss": 0.8414,
      "step": 165650
    },
    {
      "epoch": 2.9084078021032673,
      "grad_norm": 0.07265304295679542,
      "learning_rate": 7.845159332415524e-05,
      "loss": 0.8408,
      "step": 165660
    },
    {
      "epoch": 2.908583366983269,
      "grad_norm": 0.04973933192077872,
      "learning_rate": 7.844074272351155e-05,
      "loss": 0.8403,
      "step": 165670
    },
    {
      "epoch": 2.90875893186327,
      "grad_norm": 0.05226555193730388,
      "learning_rate": 7.842989240869801e-05,
      "loss": 0.8467,
      "step": 165680
    },
    {
      "epoch": 2.9089344967432713,
      "grad_norm": 0.05395302213528715,
      "learning_rate": 7.841904237985206e-05,
      "loss": 0.8427,
      "step": 165690
    },
    {
      "epoch": 2.909110061623273,
      "grad_norm": 0.04368203364067192,
      "learning_rate": 7.840819263711116e-05,
      "loss": 0.8461,
      "step": 165700
    },
    {
      "epoch": 2.9092856265032743,
      "grad_norm": 0.04726870642649374,
      "learning_rate": 7.839734318061281e-05,
      "loss": 0.8446,
      "step": 165710
    },
    {
      "epoch": 2.9094611913832757,
      "grad_norm": 0.047443523137579394,
      "learning_rate": 7.838649401049444e-05,
      "loss": 0.8491,
      "step": 165720
    },
    {
      "epoch": 2.9096367562632772,
      "grad_norm": 0.05771561329891542,
      "learning_rate": 7.837564512689351e-05,
      "loss": 0.8401,
      "step": 165730
    },
    {
      "epoch": 2.9098123211432787,
      "grad_norm": 0.04635264701108506,
      "learning_rate": 7.836479652994752e-05,
      "loss": 0.8396,
      "step": 165740
    },
    {
      "epoch": 2.9099878860232797,
      "grad_norm": 0.04792438454035386,
      "learning_rate": 7.835394821979389e-05,
      "loss": 0.8398,
      "step": 165750
    },
    {
      "epoch": 2.910163450903281,
      "grad_norm": 0.0503070454850688,
      "learning_rate": 7.83431001965701e-05,
      "loss": 0.8533,
      "step": 165760
    },
    {
      "epoch": 2.9103390157832827,
      "grad_norm": 0.04593015625466987,
      "learning_rate": 7.833225246041349e-05,
      "loss": 0.8426,
      "step": 165770
    },
    {
      "epoch": 2.910514580663284,
      "grad_norm": 0.07494906546536324,
      "learning_rate": 7.832140501146166e-05,
      "loss": 0.8512,
      "step": 165780
    },
    {
      "epoch": 2.9106901455432856,
      "grad_norm": 0.06921750099233032,
      "learning_rate": 7.831055784985195e-05,
      "loss": 0.8331,
      "step": 165790
    },
    {
      "epoch": 2.9108657104232867,
      "grad_norm": 0.053961657216221544,
      "learning_rate": 7.829971097572182e-05,
      "loss": 0.8378,
      "step": 165800
    },
    {
      "epoch": 2.9110412753032886,
      "grad_norm": 0.06649835885607418,
      "learning_rate": 7.828886438920872e-05,
      "loss": 0.841,
      "step": 165810
    },
    {
      "epoch": 2.9112168401832896,
      "grad_norm": 0.05172761002991576,
      "learning_rate": 7.827801809045005e-05,
      "loss": 0.8397,
      "step": 165820
    },
    {
      "epoch": 2.911392405063291,
      "grad_norm": 0.04543963249207889,
      "learning_rate": 7.826717207958325e-05,
      "loss": 0.8375,
      "step": 165830
    },
    {
      "epoch": 2.9115679699432926,
      "grad_norm": 0.05095610716050716,
      "learning_rate": 7.82563263567457e-05,
      "loss": 0.8427,
      "step": 165840
    },
    {
      "epoch": 2.911743534823294,
      "grad_norm": 0.05273616924121215,
      "learning_rate": 7.824548092207489e-05,
      "loss": 0.8481,
      "step": 165850
    },
    {
      "epoch": 2.9119190997032955,
      "grad_norm": 0.055370374817857254,
      "learning_rate": 7.823463577570819e-05,
      "loss": 0.8406,
      "step": 165860
    },
    {
      "epoch": 2.9120946645832966,
      "grad_norm": 0.043660072190548016,
      "learning_rate": 7.822379091778301e-05,
      "loss": 0.8488,
      "step": 165870
    },
    {
      "epoch": 2.912270229463298,
      "grad_norm": 0.0652923709862853,
      "learning_rate": 7.821294634843678e-05,
      "loss": 0.8363,
      "step": 165880
    },
    {
      "epoch": 2.9124457943432995,
      "grad_norm": 0.06107177071902202,
      "learning_rate": 7.820210206780689e-05,
      "loss": 0.8437,
      "step": 165890
    },
    {
      "epoch": 2.912621359223301,
      "grad_norm": 0.0627301351450824,
      "learning_rate": 7.819125807603072e-05,
      "loss": 0.8361,
      "step": 165900
    },
    {
      "epoch": 2.9127969241033025,
      "grad_norm": 0.04557597049071098,
      "learning_rate": 7.818041437324568e-05,
      "loss": 0.8434,
      "step": 165910
    },
    {
      "epoch": 2.9129724889833035,
      "grad_norm": 0.05133549025355707,
      "learning_rate": 7.81695709595892e-05,
      "loss": 0.8424,
      "step": 165920
    },
    {
      "epoch": 2.9131480538633054,
      "grad_norm": 0.05197887747822707,
      "learning_rate": 7.81587278351986e-05,
      "loss": 0.8417,
      "step": 165930
    },
    {
      "epoch": 2.9133236187433065,
      "grad_norm": 0.05435068826125067,
      "learning_rate": 7.814788500021127e-05,
      "loss": 0.8409,
      "step": 165940
    },
    {
      "epoch": 2.913499183623308,
      "grad_norm": 0.1021368473580131,
      "learning_rate": 7.813704245476464e-05,
      "loss": 0.8468,
      "step": 165950
    },
    {
      "epoch": 2.9136747485033094,
      "grad_norm": 0.054436145222125004,
      "learning_rate": 7.812620019899608e-05,
      "loss": 0.8447,
      "step": 165960
    },
    {
      "epoch": 2.913850313383311,
      "grad_norm": 0.05270005599282899,
      "learning_rate": 7.811535823304293e-05,
      "loss": 0.8424,
      "step": 165970
    },
    {
      "epoch": 2.9140258782633124,
      "grad_norm": 0.05959146098223891,
      "learning_rate": 7.810451655704258e-05,
      "loss": 0.8459,
      "step": 165980
    },
    {
      "epoch": 2.9142014431433134,
      "grad_norm": 0.04508258371363016,
      "learning_rate": 7.809367517113241e-05,
      "loss": 0.8406,
      "step": 165990
    },
    {
      "epoch": 2.914377008023315,
      "grad_norm": 0.0727531640824531,
      "learning_rate": 7.808283407544975e-05,
      "loss": 0.8441,
      "step": 166000
    },
    {
      "epoch": 2.9145525729033164,
      "grad_norm": 0.058758635191030834,
      "learning_rate": 7.807199327013193e-05,
      "loss": 0.8458,
      "step": 166010
    },
    {
      "epoch": 2.914728137783318,
      "grad_norm": 0.0483319645919758,
      "learning_rate": 7.806115275531641e-05,
      "loss": 0.8396,
      "step": 166020
    },
    {
      "epoch": 2.9149037026633193,
      "grad_norm": 0.039960011856797865,
      "learning_rate": 7.805031253114048e-05,
      "loss": 0.8444,
      "step": 166030
    },
    {
      "epoch": 2.9150792675433204,
      "grad_norm": 0.06125641982138102,
      "learning_rate": 7.803947259774146e-05,
      "loss": 0.8505,
      "step": 166040
    },
    {
      "epoch": 2.9152548324233223,
      "grad_norm": 0.057504935479887685,
      "learning_rate": 7.802863295525674e-05,
      "loss": 0.8397,
      "step": 166050
    },
    {
      "epoch": 2.9154303973033233,
      "grad_norm": 0.04777985684367588,
      "learning_rate": 7.801779360382366e-05,
      "loss": 0.8417,
      "step": 166060
    },
    {
      "epoch": 2.915605962183325,
      "grad_norm": 0.04840367022976688,
      "learning_rate": 7.800695454357949e-05,
      "loss": 0.8264,
      "step": 166070
    },
    {
      "epoch": 2.9157815270633263,
      "grad_norm": 0.04174615101537317,
      "learning_rate": 7.799611577466164e-05,
      "loss": 0.8502,
      "step": 166080
    },
    {
      "epoch": 2.9159570919433278,
      "grad_norm": 0.06330348248941466,
      "learning_rate": 7.798527729720743e-05,
      "loss": 0.849,
      "step": 166090
    },
    {
      "epoch": 2.9161326568233292,
      "grad_norm": 0.052422612086748484,
      "learning_rate": 7.797443911135417e-05,
      "loss": 0.8413,
      "step": 166100
    },
    {
      "epoch": 2.9163082217033303,
      "grad_norm": 0.0487058399948113,
      "learning_rate": 7.796360121723912e-05,
      "loss": 0.8419,
      "step": 166110
    },
    {
      "epoch": 2.9164837865833317,
      "grad_norm": 0.050827726709818534,
      "learning_rate": 7.79527636149997e-05,
      "loss": 0.8465,
      "step": 166120
    },
    {
      "epoch": 2.916659351463333,
      "grad_norm": 0.05868166042161775,
      "learning_rate": 7.794192630477318e-05,
      "loss": 0.8505,
      "step": 166130
    },
    {
      "epoch": 2.9168349163433347,
      "grad_norm": 0.07103298814228204,
      "learning_rate": 7.793108928669686e-05,
      "loss": 0.8493,
      "step": 166140
    },
    {
      "epoch": 2.917010481223336,
      "grad_norm": 0.047186532602645416,
      "learning_rate": 7.792025256090809e-05,
      "loss": 0.8418,
      "step": 166150
    },
    {
      "epoch": 2.9171860461033376,
      "grad_norm": 0.0483025508335643,
      "learning_rate": 7.790941612754413e-05,
      "loss": 0.8377,
      "step": 166160
    },
    {
      "epoch": 2.917361610983339,
      "grad_norm": 0.050754332218129784,
      "learning_rate": 7.78985799867423e-05,
      "loss": 0.8422,
      "step": 166170
    },
    {
      "epoch": 2.91753717586334,
      "grad_norm": 0.05738464852893005,
      "learning_rate": 7.788774413863985e-05,
      "loss": 0.8357,
      "step": 166180
    },
    {
      "epoch": 2.9177127407433416,
      "grad_norm": 0.06868846570291848,
      "learning_rate": 7.787690858337414e-05,
      "loss": 0.8486,
      "step": 166190
    },
    {
      "epoch": 2.917888305623343,
      "grad_norm": 0.049933501666357474,
      "learning_rate": 7.786607332108243e-05,
      "loss": 0.8445,
      "step": 166200
    },
    {
      "epoch": 2.9180638705033446,
      "grad_norm": 0.04403981815529309,
      "learning_rate": 7.7855238351902e-05,
      "loss": 0.8469,
      "step": 166210
    },
    {
      "epoch": 2.918239435383346,
      "grad_norm": 0.07208491033598806,
      "learning_rate": 7.784440367597014e-05,
      "loss": 0.8425,
      "step": 166220
    },
    {
      "epoch": 2.918415000263347,
      "grad_norm": 0.05846288186459382,
      "learning_rate": 7.783356929342413e-05,
      "loss": 0.8395,
      "step": 166230
    },
    {
      "epoch": 2.9185905651433486,
      "grad_norm": 0.04911420198667091,
      "learning_rate": 7.782273520440123e-05,
      "loss": 0.8476,
      "step": 166240
    },
    {
      "epoch": 2.91876613002335,
      "grad_norm": 0.048189081147507314,
      "learning_rate": 7.781190140903871e-05,
      "loss": 0.8478,
      "step": 166250
    },
    {
      "epoch": 2.9189416949033515,
      "grad_norm": 0.04810481457151729,
      "learning_rate": 7.780106790747386e-05,
      "loss": 0.8487,
      "step": 166260
    },
    {
      "epoch": 2.919117259783353,
      "grad_norm": 0.052769526215993415,
      "learning_rate": 7.779023469984392e-05,
      "loss": 0.845,
      "step": 166270
    },
    {
      "epoch": 2.9192928246633545,
      "grad_norm": 0.0632756804635532,
      "learning_rate": 7.777940178628611e-05,
      "loss": 0.8424,
      "step": 166280
    },
    {
      "epoch": 2.919468389543356,
      "grad_norm": 0.06497860831767845,
      "learning_rate": 7.776856916693777e-05,
      "loss": 0.8387,
      "step": 166290
    },
    {
      "epoch": 2.919643954423357,
      "grad_norm": 0.05688427138092298,
      "learning_rate": 7.775773684193612e-05,
      "loss": 0.8464,
      "step": 166300
    },
    {
      "epoch": 2.9198195193033585,
      "grad_norm": 0.05641670204015309,
      "learning_rate": 7.774690481141839e-05,
      "loss": 0.8431,
      "step": 166310
    },
    {
      "epoch": 2.91999508418336,
      "grad_norm": 0.048705652882684705,
      "learning_rate": 7.773607307552183e-05,
      "loss": 0.8413,
      "step": 166320
    },
    {
      "epoch": 2.9201706490633614,
      "grad_norm": 0.07836644853790159,
      "learning_rate": 7.772524163438369e-05,
      "loss": 0.8459,
      "step": 166330
    },
    {
      "epoch": 2.920346213943363,
      "grad_norm": 0.056259123268985294,
      "learning_rate": 7.77144104881412e-05,
      "loss": 0.8363,
      "step": 166340
    },
    {
      "epoch": 2.920521778823364,
      "grad_norm": 0.050450250684245766,
      "learning_rate": 7.770357963693155e-05,
      "loss": 0.8419,
      "step": 166350
    },
    {
      "epoch": 2.9206973437033654,
      "grad_norm": 0.05907650952413917,
      "learning_rate": 7.769274908089206e-05,
      "loss": 0.8441,
      "step": 166360
    },
    {
      "epoch": 2.920872908583367,
      "grad_norm": 0.05909657468510653,
      "learning_rate": 7.76819188201599e-05,
      "loss": 0.835,
      "step": 166370
    },
    {
      "epoch": 2.9210484734633684,
      "grad_norm": 0.05551637686590776,
      "learning_rate": 7.767108885487227e-05,
      "loss": 0.8408,
      "step": 166380
    },
    {
      "epoch": 2.92122403834337,
      "grad_norm": 0.06469456838254341,
      "learning_rate": 7.766025918516644e-05,
      "loss": 0.842,
      "step": 166390
    },
    {
      "epoch": 2.9213996032233713,
      "grad_norm": 0.0846292278031828,
      "learning_rate": 7.76494298111796e-05,
      "loss": 0.8484,
      "step": 166400
    },
    {
      "epoch": 2.921575168103373,
      "grad_norm": 0.05668584676318539,
      "learning_rate": 7.763860073304895e-05,
      "loss": 0.8431,
      "step": 166410
    },
    {
      "epoch": 2.921750732983374,
      "grad_norm": 0.0515852774591227,
      "learning_rate": 7.762777195091172e-05,
      "loss": 0.8446,
      "step": 166420
    },
    {
      "epoch": 2.9219262978633753,
      "grad_norm": 0.060572032865394025,
      "learning_rate": 7.761694346490507e-05,
      "loss": 0.8446,
      "step": 166430
    },
    {
      "epoch": 2.922101862743377,
      "grad_norm": 0.05912812394225451,
      "learning_rate": 7.760611527516626e-05,
      "loss": 0.8345,
      "step": 166440
    },
    {
      "epoch": 2.9222774276233783,
      "grad_norm": 0.05604543423936016,
      "learning_rate": 7.759528738183241e-05,
      "loss": 0.8382,
      "step": 166450
    },
    {
      "epoch": 2.9224529925033798,
      "grad_norm": 0.05040745375908166,
      "learning_rate": 7.758445978504079e-05,
      "loss": 0.8485,
      "step": 166460
    },
    {
      "epoch": 2.922628557383381,
      "grad_norm": 0.07513519436534286,
      "learning_rate": 7.757363248492855e-05,
      "loss": 0.8446,
      "step": 166470
    },
    {
      "epoch": 2.9228041222633827,
      "grad_norm": 0.05736943771844863,
      "learning_rate": 7.756280548163285e-05,
      "loss": 0.848,
      "step": 166480
    },
    {
      "epoch": 2.9229796871433837,
      "grad_norm": 0.04710239753109713,
      "learning_rate": 7.755197877529094e-05,
      "loss": 0.8312,
      "step": 166490
    },
    {
      "epoch": 2.923155252023385,
      "grad_norm": 0.04653874204967587,
      "learning_rate": 7.754115236603992e-05,
      "loss": 0.8419,
      "step": 166500
    },
    {
      "epoch": 2.9233308169033867,
      "grad_norm": 0.03955193040174793,
      "learning_rate": 7.753032625401702e-05,
      "loss": 0.8501,
      "step": 166510
    },
    {
      "epoch": 2.923506381783388,
      "grad_norm": 0.049023372795033465,
      "learning_rate": 7.751950043935933e-05,
      "loss": 0.8412,
      "step": 166520
    },
    {
      "epoch": 2.9236819466633897,
      "grad_norm": 0.051459062197955124,
      "learning_rate": 7.750867492220411e-05,
      "loss": 0.8442,
      "step": 166530
    },
    {
      "epoch": 2.9238575115433907,
      "grad_norm": 0.0485287147939274,
      "learning_rate": 7.749784970268848e-05,
      "loss": 0.8502,
      "step": 166540
    },
    {
      "epoch": 2.924033076423392,
      "grad_norm": 0.04345010354523265,
      "learning_rate": 7.748702478094959e-05,
      "loss": 0.8492,
      "step": 166550
    },
    {
      "epoch": 2.9242086413033936,
      "grad_norm": 0.06657553596398272,
      "learning_rate": 7.747620015712461e-05,
      "loss": 0.847,
      "step": 166560
    },
    {
      "epoch": 2.924384206183395,
      "grad_norm": 0.059315573667107824,
      "learning_rate": 7.746537583135069e-05,
      "loss": 0.8407,
      "step": 166570
    },
    {
      "epoch": 2.9245597710633966,
      "grad_norm": 0.060749478477166066,
      "learning_rate": 7.745455180376495e-05,
      "loss": 0.8421,
      "step": 166580
    },
    {
      "epoch": 2.9247353359433976,
      "grad_norm": 0.06506337191750973,
      "learning_rate": 7.744372807450457e-05,
      "loss": 0.8474,
      "step": 166590
    },
    {
      "epoch": 2.9249109008233996,
      "grad_norm": 0.04640862234833421,
      "learning_rate": 7.743290464370668e-05,
      "loss": 0.8417,
      "step": 166600
    },
    {
      "epoch": 2.9250864657034006,
      "grad_norm": 0.062418411281024884,
      "learning_rate": 7.742208151150838e-05,
      "loss": 0.8345,
      "step": 166610
    },
    {
      "epoch": 2.925262030583402,
      "grad_norm": 0.05400079145610111,
      "learning_rate": 7.74112586780468e-05,
      "loss": 0.8497,
      "step": 166620
    },
    {
      "epoch": 2.9254375954634035,
      "grad_norm": 0.06213927992558956,
      "learning_rate": 7.740043614345913e-05,
      "loss": 0.8409,
      "step": 166630
    },
    {
      "epoch": 2.925613160343405,
      "grad_norm": 0.05906238946124199,
      "learning_rate": 7.738961390788247e-05,
      "loss": 0.8427,
      "step": 166640
    },
    {
      "epoch": 2.9257887252234065,
      "grad_norm": 0.05148532741739543,
      "learning_rate": 7.737879197145389e-05,
      "loss": 0.8486,
      "step": 166650
    },
    {
      "epoch": 2.9259642901034075,
      "grad_norm": 0.04730806105614654,
      "learning_rate": 7.736797033431058e-05,
      "loss": 0.8415,
      "step": 166660
    },
    {
      "epoch": 2.926139854983409,
      "grad_norm": 0.0607120728418609,
      "learning_rate": 7.735714899658959e-05,
      "loss": 0.8454,
      "step": 166670
    },
    {
      "epoch": 2.9263154198634105,
      "grad_norm": 0.0513334301788394,
      "learning_rate": 7.734632795842806e-05,
      "loss": 0.8409,
      "step": 166680
    },
    {
      "epoch": 2.926490984743412,
      "grad_norm": 0.042953905908412554,
      "learning_rate": 7.733550721996307e-05,
      "loss": 0.8404,
      "step": 166690
    },
    {
      "epoch": 2.9266665496234134,
      "grad_norm": 0.04763087114728016,
      "learning_rate": 7.732468678133176e-05,
      "loss": 0.8446,
      "step": 166700
    },
    {
      "epoch": 2.9268421145034145,
      "grad_norm": 0.051061153826170894,
      "learning_rate": 7.731386664267121e-05,
      "loss": 0.85,
      "step": 166710
    },
    {
      "epoch": 2.9270176793834164,
      "grad_norm": 0.058634711522613375,
      "learning_rate": 7.730304680411848e-05,
      "loss": 0.8438,
      "step": 166720
    },
    {
      "epoch": 2.9271932442634174,
      "grad_norm": 0.05197016540511245,
      "learning_rate": 7.729222726581072e-05,
      "loss": 0.8403,
      "step": 166730
    },
    {
      "epoch": 2.927368809143419,
      "grad_norm": 0.051707658920558244,
      "learning_rate": 7.728140802788499e-05,
      "loss": 0.8515,
      "step": 166740
    },
    {
      "epoch": 2.9275443740234204,
      "grad_norm": 0.043830458310189044,
      "learning_rate": 7.727058909047834e-05,
      "loss": 0.8463,
      "step": 166750
    },
    {
      "epoch": 2.927719938903422,
      "grad_norm": 0.07002949592372561,
      "learning_rate": 7.725977045372788e-05,
      "loss": 0.8413,
      "step": 166760
    },
    {
      "epoch": 2.9278955037834233,
      "grad_norm": 0.047407768386804744,
      "learning_rate": 7.72489521177707e-05,
      "loss": 0.8385,
      "step": 166770
    },
    {
      "epoch": 2.9280710686634244,
      "grad_norm": 0.0460042786531674,
      "learning_rate": 7.723813408274383e-05,
      "loss": 0.8399,
      "step": 166780
    },
    {
      "epoch": 2.928246633543426,
      "grad_norm": 0.058379525170610624,
      "learning_rate": 7.722731634878432e-05,
      "loss": 0.8425,
      "step": 166790
    },
    {
      "epoch": 2.9284221984234273,
      "grad_norm": 0.047737995343531514,
      "learning_rate": 7.721649891602931e-05,
      "loss": 0.8416,
      "step": 166800
    },
    {
      "epoch": 2.928597763303429,
      "grad_norm": 0.045241707418165446,
      "learning_rate": 7.720568178461583e-05,
      "loss": 0.8429,
      "step": 166810
    },
    {
      "epoch": 2.9287733281834303,
      "grad_norm": 0.051490168222875554,
      "learning_rate": 7.719486495468088e-05,
      "loss": 0.8419,
      "step": 166820
    },
    {
      "epoch": 2.9289488930634318,
      "grad_norm": 0.04828833016454676,
      "learning_rate": 7.718404842636159e-05,
      "loss": 0.8402,
      "step": 166830
    },
    {
      "epoch": 2.9291244579434332,
      "grad_norm": 0.06240246129451937,
      "learning_rate": 7.717323219979497e-05,
      "loss": 0.8404,
      "step": 166840
    },
    {
      "epoch": 2.9293000228234343,
      "grad_norm": 0.05862251942122953,
      "learning_rate": 7.716241627511806e-05,
      "loss": 0.8422,
      "step": 166850
    },
    {
      "epoch": 2.9294755877034357,
      "grad_norm": 0.08241112629659283,
      "learning_rate": 7.715160065246783e-05,
      "loss": 0.8345,
      "step": 166860
    },
    {
      "epoch": 2.9296511525834372,
      "grad_norm": 0.07896378187248665,
      "learning_rate": 7.714078533198148e-05,
      "loss": 0.8381,
      "step": 166870
    },
    {
      "epoch": 2.9298267174634387,
      "grad_norm": 0.0657651100751414,
      "learning_rate": 7.712997031379595e-05,
      "loss": 0.8372,
      "step": 166880
    },
    {
      "epoch": 2.93000228234344,
      "grad_norm": 0.057413882201555265,
      "learning_rate": 7.711915559804822e-05,
      "loss": 0.8434,
      "step": 166890
    },
    {
      "epoch": 2.930177847223441,
      "grad_norm": 0.05328435909952621,
      "learning_rate": 7.71083411848754e-05,
      "loss": 0.8469,
      "step": 166900
    },
    {
      "epoch": 2.9303534121034427,
      "grad_norm": 0.047516414734544904,
      "learning_rate": 7.709752707441449e-05,
      "loss": 0.8429,
      "step": 166910
    },
    {
      "epoch": 2.930528976983444,
      "grad_norm": 0.0710805986910066,
      "learning_rate": 7.708671326680246e-05,
      "loss": 0.8507,
      "step": 166920
    },
    {
      "epoch": 2.9307045418634456,
      "grad_norm": 0.062334487088238154,
      "learning_rate": 7.707589976217638e-05,
      "loss": 0.847,
      "step": 166930
    },
    {
      "epoch": 2.930880106743447,
      "grad_norm": 0.04636308537665379,
      "learning_rate": 7.706508656067322e-05,
      "loss": 0.8457,
      "step": 166940
    },
    {
      "epoch": 2.9310556716234486,
      "grad_norm": 0.05185100728605853,
      "learning_rate": 7.705427366243001e-05,
      "loss": 0.8419,
      "step": 166950
    },
    {
      "epoch": 2.93123123650345,
      "grad_norm": 0.050931917173152236,
      "learning_rate": 7.70434610675837e-05,
      "loss": 0.8525,
      "step": 166960
    },
    {
      "epoch": 2.931406801383451,
      "grad_norm": 0.043980088508648145,
      "learning_rate": 7.703264877627137e-05,
      "loss": 0.8502,
      "step": 166970
    },
    {
      "epoch": 2.9315823662634526,
      "grad_norm": 0.059125002572070026,
      "learning_rate": 7.702183678862999e-05,
      "loss": 0.843,
      "step": 166980
    },
    {
      "epoch": 2.931757931143454,
      "grad_norm": 0.0663576935900154,
      "learning_rate": 7.70110251047965e-05,
      "loss": 0.8437,
      "step": 166990
    },
    {
      "epoch": 2.9319334960234555,
      "grad_norm": 0.04882193263292663,
      "learning_rate": 7.700021372490793e-05,
      "loss": 0.8404,
      "step": 167000
    },
    {
      "epoch": 2.932109060903457,
      "grad_norm": 0.06144883620233042,
      "learning_rate": 7.698940264910127e-05,
      "loss": 0.8432,
      "step": 167010
    },
    {
      "epoch": 2.932284625783458,
      "grad_norm": 0.06782261950577774,
      "learning_rate": 7.697859187751345e-05,
      "loss": 0.835,
      "step": 167020
    },
    {
      "epoch": 2.9324601906634595,
      "grad_norm": 0.05457665092446601,
      "learning_rate": 7.696778141028146e-05,
      "loss": 0.8409,
      "step": 167030
    },
    {
      "epoch": 2.932635755543461,
      "grad_norm": 0.05260217311915117,
      "learning_rate": 7.695697124754233e-05,
      "loss": 0.8428,
      "step": 167040
    },
    {
      "epoch": 2.9328113204234625,
      "grad_norm": 0.061349883719698926,
      "learning_rate": 7.694616138943296e-05,
      "loss": 0.8366,
      "step": 167050
    },
    {
      "epoch": 2.932986885303464,
      "grad_norm": 0.05346974844369348,
      "learning_rate": 7.693535183609033e-05,
      "loss": 0.8386,
      "step": 167060
    },
    {
      "epoch": 2.9331624501834654,
      "grad_norm": 0.053809009652835575,
      "learning_rate": 7.692454258765141e-05,
      "loss": 0.8431,
      "step": 167070
    },
    {
      "epoch": 2.933338015063467,
      "grad_norm": 0.06176942948813524,
      "learning_rate": 7.691373364425318e-05,
      "loss": 0.8385,
      "step": 167080
    },
    {
      "epoch": 2.933513579943468,
      "grad_norm": 0.05946631390305561,
      "learning_rate": 7.690292500603252e-05,
      "loss": 0.8397,
      "step": 167090
    },
    {
      "epoch": 2.9336891448234694,
      "grad_norm": 0.07324454811440743,
      "learning_rate": 7.689211667312644e-05,
      "loss": 0.8412,
      "step": 167100
    },
    {
      "epoch": 2.933864709703471,
      "grad_norm": 0.05704678130269038,
      "learning_rate": 7.688130864567186e-05,
      "loss": 0.847,
      "step": 167110
    },
    {
      "epoch": 2.9340402745834724,
      "grad_norm": 0.0716168039490432,
      "learning_rate": 7.687050092380573e-05,
      "loss": 0.8446,
      "step": 167120
    },
    {
      "epoch": 2.934215839463474,
      "grad_norm": 0.05770671084572743,
      "learning_rate": 7.685969350766491e-05,
      "loss": 0.8326,
      "step": 167130
    },
    {
      "epoch": 2.934391404343475,
      "grad_norm": 0.045246851124167986,
      "learning_rate": 7.684888639738647e-05,
      "loss": 0.8454,
      "step": 167140
    },
    {
      "epoch": 2.9345669692234764,
      "grad_norm": 0.04458263834904529,
      "learning_rate": 7.683807959310726e-05,
      "loss": 0.841,
      "step": 167150
    },
    {
      "epoch": 2.934742534103478,
      "grad_norm": 0.053847472729427415,
      "learning_rate": 7.68272730949642e-05,
      "loss": 0.8404,
      "step": 167160
    },
    {
      "epoch": 2.9349180989834793,
      "grad_norm": 0.05483141185636448,
      "learning_rate": 7.681646690309424e-05,
      "loss": 0.8413,
      "step": 167170
    },
    {
      "epoch": 2.935093663863481,
      "grad_norm": 0.05075179863469563,
      "learning_rate": 7.680566101763427e-05,
      "loss": 0.8416,
      "step": 167180
    },
    {
      "epoch": 2.9352692287434823,
      "grad_norm": 0.048571346454187626,
      "learning_rate": 7.67948554387212e-05,
      "loss": 0.8357,
      "step": 167190
    },
    {
      "epoch": 2.9354447936234838,
      "grad_norm": 0.0510030599497992,
      "learning_rate": 7.678405016649193e-05,
      "loss": 0.8343,
      "step": 167200
    },
    {
      "epoch": 2.935620358503485,
      "grad_norm": 0.052926795884638424,
      "learning_rate": 7.677324520108343e-05,
      "loss": 0.8405,
      "step": 167210
    },
    {
      "epoch": 2.9357959233834863,
      "grad_norm": 0.07167274464752811,
      "learning_rate": 7.676244054263255e-05,
      "loss": 0.8491,
      "step": 167220
    },
    {
      "epoch": 2.9359714882634877,
      "grad_norm": 0.051856992041895415,
      "learning_rate": 7.675163619127617e-05,
      "loss": 0.8449,
      "step": 167230
    },
    {
      "epoch": 2.9361470531434892,
      "grad_norm": 0.05258186355520015,
      "learning_rate": 7.674083214715122e-05,
      "loss": 0.8443,
      "step": 167240
    },
    {
      "epoch": 2.9363226180234907,
      "grad_norm": 0.051522462079176316,
      "learning_rate": 7.673002841039458e-05,
      "loss": 0.8474,
      "step": 167250
    },
    {
      "epoch": 2.9364981829034917,
      "grad_norm": 0.04765901651546467,
      "learning_rate": 7.671922498114313e-05,
      "loss": 0.8371,
      "step": 167260
    },
    {
      "epoch": 2.9366737477834937,
      "grad_norm": 0.05453116348929225,
      "learning_rate": 7.670842185953374e-05,
      "loss": 0.8396,
      "step": 167270
    },
    {
      "epoch": 2.9368493126634947,
      "grad_norm": 0.07337771944900864,
      "learning_rate": 7.669761904570333e-05,
      "loss": 0.8369,
      "step": 167280
    },
    {
      "epoch": 2.937024877543496,
      "grad_norm": 0.05465935556667515,
      "learning_rate": 7.668681653978874e-05,
      "loss": 0.8471,
      "step": 167290
    },
    {
      "epoch": 2.9372004424234976,
      "grad_norm": 0.0416806972932303,
      "learning_rate": 7.667601434192678e-05,
      "loss": 0.8448,
      "step": 167300
    },
    {
      "epoch": 2.937376007303499,
      "grad_norm": 0.04974689739762725,
      "learning_rate": 7.666521245225444e-05,
      "loss": 0.8406,
      "step": 167310
    },
    {
      "epoch": 2.9375515721835006,
      "grad_norm": 0.057115272496869124,
      "learning_rate": 7.665441087090852e-05,
      "loss": 0.8459,
      "step": 167320
    },
    {
      "epoch": 2.9377271370635016,
      "grad_norm": 0.05542625134889735,
      "learning_rate": 7.664360959802586e-05,
      "loss": 0.8469,
      "step": 167330
    },
    {
      "epoch": 2.937902701943503,
      "grad_norm": 0.05470042737506913,
      "learning_rate": 7.663280863374335e-05,
      "loss": 0.8362,
      "step": 167340
    },
    {
      "epoch": 2.9380782668235046,
      "grad_norm": 0.04903613653776323,
      "learning_rate": 7.662200797819781e-05,
      "loss": 0.8424,
      "step": 167350
    },
    {
      "epoch": 2.938253831703506,
      "grad_norm": 0.06690396571934669,
      "learning_rate": 7.661120763152614e-05,
      "loss": 0.8381,
      "step": 167360
    },
    {
      "epoch": 2.9384293965835075,
      "grad_norm": 0.042971797690049156,
      "learning_rate": 7.660040759386508e-05,
      "loss": 0.8372,
      "step": 167370
    },
    {
      "epoch": 2.9386049614635086,
      "grad_norm": 0.06032063780906177,
      "learning_rate": 7.658960786535157e-05,
      "loss": 0.8432,
      "step": 167380
    },
    {
      "epoch": 2.9387805263435105,
      "grad_norm": 0.06048670555213265,
      "learning_rate": 7.657880844612241e-05,
      "loss": 0.8412,
      "step": 167390
    },
    {
      "epoch": 2.9389560912235115,
      "grad_norm": 0.04817859239173368,
      "learning_rate": 7.65680093363144e-05,
      "loss": 0.8467,
      "step": 167400
    },
    {
      "epoch": 2.939131656103513,
      "grad_norm": 0.05575193527077909,
      "learning_rate": 7.655721053606442e-05,
      "loss": 0.8444,
      "step": 167410
    },
    {
      "epoch": 2.9393072209835145,
      "grad_norm": 0.05483906984470944,
      "learning_rate": 7.654641204550928e-05,
      "loss": 0.8404,
      "step": 167420
    },
    {
      "epoch": 2.939482785863516,
      "grad_norm": 0.045493182766710674,
      "learning_rate": 7.653561386478574e-05,
      "loss": 0.8449,
      "step": 167430
    },
    {
      "epoch": 2.9396583507435174,
      "grad_norm": 0.05603244855533851,
      "learning_rate": 7.65248159940307e-05,
      "loss": 0.8474,
      "step": 167440
    },
    {
      "epoch": 2.9398339156235185,
      "grad_norm": 0.05853603231741939,
      "learning_rate": 7.651401843338093e-05,
      "loss": 0.8342,
      "step": 167450
    },
    {
      "epoch": 2.94000948050352,
      "grad_norm": 0.05566164231456087,
      "learning_rate": 7.650322118297323e-05,
      "loss": 0.8402,
      "step": 167460
    },
    {
      "epoch": 2.9401850453835214,
      "grad_norm": 0.051713883259529914,
      "learning_rate": 7.649242424294439e-05,
      "loss": 0.8395,
      "step": 167470
    },
    {
      "epoch": 2.940360610263523,
      "grad_norm": 0.047275120689416236,
      "learning_rate": 7.648162761343125e-05,
      "loss": 0.845,
      "step": 167480
    },
    {
      "epoch": 2.9405361751435244,
      "grad_norm": 0.08252658073536305,
      "learning_rate": 7.64708312945706e-05,
      "loss": 0.8464,
      "step": 167490
    },
    {
      "epoch": 2.9407117400235254,
      "grad_norm": 0.04987927584632699,
      "learning_rate": 7.646003528649918e-05,
      "loss": 0.838,
      "step": 167500
    },
    {
      "epoch": 2.9408873049035273,
      "grad_norm": 0.054047057749771595,
      "learning_rate": 7.644923958935388e-05,
      "loss": 0.8257,
      "step": 167510
    },
    {
      "epoch": 2.9410628697835284,
      "grad_norm": 0.059793250962775664,
      "learning_rate": 7.643844420327138e-05,
      "loss": 0.8393,
      "step": 167520
    },
    {
      "epoch": 2.94123843466353,
      "grad_norm": 0.062234001157796606,
      "learning_rate": 7.642764912838851e-05,
      "loss": 0.8448,
      "step": 167530
    },
    {
      "epoch": 2.9414139995435313,
      "grad_norm": 0.04659344276845347,
      "learning_rate": 7.641685436484198e-05,
      "loss": 0.842,
      "step": 167540
    },
    {
      "epoch": 2.941589564423533,
      "grad_norm": 0.05611422703837808,
      "learning_rate": 7.640605991276869e-05,
      "loss": 0.8406,
      "step": 167550
    },
    {
      "epoch": 2.9417651293035343,
      "grad_norm": 0.04028619650795887,
      "learning_rate": 7.639526577230532e-05,
      "loss": 0.8457,
      "step": 167560
    },
    {
      "epoch": 2.9419406941835353,
      "grad_norm": 0.05894137818495544,
      "learning_rate": 7.638447194358861e-05,
      "loss": 0.8435,
      "step": 167570
    },
    {
      "epoch": 2.942116259063537,
      "grad_norm": 0.07160573090231345,
      "learning_rate": 7.637367842675539e-05,
      "loss": 0.8359,
      "step": 167580
    },
    {
      "epoch": 2.9422918239435383,
      "grad_norm": 0.053497757981359736,
      "learning_rate": 7.63628852219424e-05,
      "loss": 0.8494,
      "step": 167590
    },
    {
      "epoch": 2.9424673888235398,
      "grad_norm": 0.06012371987072552,
      "learning_rate": 7.635209232928634e-05,
      "loss": 0.8436,
      "step": 167600
    },
    {
      "epoch": 2.9426429537035412,
      "grad_norm": 0.07031234370640112,
      "learning_rate": 7.634129974892401e-05,
      "loss": 0.8467,
      "step": 167610
    },
    {
      "epoch": 2.9428185185835427,
      "grad_norm": 0.06270809944207574,
      "learning_rate": 7.633050748099215e-05,
      "loss": 0.8533,
      "step": 167620
    },
    {
      "epoch": 2.942994083463544,
      "grad_norm": 0.04683057850854764,
      "learning_rate": 7.631971552562748e-05,
      "loss": 0.8372,
      "step": 167630
    },
    {
      "epoch": 2.943169648343545,
      "grad_norm": 0.046914694623841936,
      "learning_rate": 7.630892388296671e-05,
      "loss": 0.8452,
      "step": 167640
    },
    {
      "epoch": 2.9433452132235467,
      "grad_norm": 0.05148026107676541,
      "learning_rate": 7.629813255314663e-05,
      "loss": 0.8417,
      "step": 167650
    },
    {
      "epoch": 2.943520778103548,
      "grad_norm": 0.05738833491848509,
      "learning_rate": 7.628734153630395e-05,
      "loss": 0.84,
      "step": 167660
    },
    {
      "epoch": 2.9436963429835497,
      "grad_norm": 0.045503100913919174,
      "learning_rate": 7.627655083257539e-05,
      "loss": 0.8423,
      "step": 167670
    },
    {
      "epoch": 2.943871907863551,
      "grad_norm": 0.05465802373924811,
      "learning_rate": 7.626576044209768e-05,
      "loss": 0.8419,
      "step": 167680
    },
    {
      "epoch": 2.944047472743552,
      "grad_norm": 0.052361837595991584,
      "learning_rate": 7.625497036500752e-05,
      "loss": 0.8377,
      "step": 167690
    },
    {
      "epoch": 2.9442230376235536,
      "grad_norm": 0.060158176810609434,
      "learning_rate": 7.624418060144162e-05,
      "loss": 0.8445,
      "step": 167700
    },
    {
      "epoch": 2.944398602503555,
      "grad_norm": 0.057607989439654075,
      "learning_rate": 7.623339115153666e-05,
      "loss": 0.8406,
      "step": 167710
    },
    {
      "epoch": 2.9445741673835566,
      "grad_norm": 0.05505641573742459,
      "learning_rate": 7.622260201542944e-05,
      "loss": 0.8391,
      "step": 167720
    },
    {
      "epoch": 2.944749732263558,
      "grad_norm": 0.048266621179283695,
      "learning_rate": 7.621181319325658e-05,
      "loss": 0.8453,
      "step": 167730
    },
    {
      "epoch": 2.9449252971435596,
      "grad_norm": 0.06087906420928808,
      "learning_rate": 7.620102468515477e-05,
      "loss": 0.8543,
      "step": 167740
    },
    {
      "epoch": 2.945100862023561,
      "grad_norm": 0.0569085967411823,
      "learning_rate": 7.619023649126078e-05,
      "loss": 0.8486,
      "step": 167750
    },
    {
      "epoch": 2.945276426903562,
      "grad_norm": 0.11252056267459147,
      "learning_rate": 7.617944861171123e-05,
      "loss": 0.8449,
      "step": 167760
    },
    {
      "epoch": 2.9454519917835635,
      "grad_norm": 0.05788285861172505,
      "learning_rate": 7.61686610466428e-05,
      "loss": 0.8369,
      "step": 167770
    },
    {
      "epoch": 2.945627556663565,
      "grad_norm": 0.04373757203857897,
      "learning_rate": 7.615787379619223e-05,
      "loss": 0.8356,
      "step": 167780
    },
    {
      "epoch": 2.9458031215435665,
      "grad_norm": 0.05427874840646249,
      "learning_rate": 7.614708686049615e-05,
      "loss": 0.8455,
      "step": 167790
    },
    {
      "epoch": 2.945978686423568,
      "grad_norm": 0.06652827925563219,
      "learning_rate": 7.613630023969124e-05,
      "loss": 0.841,
      "step": 167800
    },
    {
      "epoch": 2.946154251303569,
      "grad_norm": 0.057269712352646336,
      "learning_rate": 7.612551393391412e-05,
      "loss": 0.8444,
      "step": 167810
    },
    {
      "epoch": 2.9463298161835705,
      "grad_norm": 0.04681576222835532,
      "learning_rate": 7.611472794330157e-05,
      "loss": 0.8429,
      "step": 167820
    },
    {
      "epoch": 2.946505381063572,
      "grad_norm": 0.06720197087811926,
      "learning_rate": 7.610394226799017e-05,
      "loss": 0.8349,
      "step": 167830
    },
    {
      "epoch": 2.9466809459435734,
      "grad_norm": 0.06340060315270866,
      "learning_rate": 7.609315690811658e-05,
      "loss": 0.8469,
      "step": 167840
    },
    {
      "epoch": 2.946856510823575,
      "grad_norm": 0.056064844517107304,
      "learning_rate": 7.608237186381748e-05,
      "loss": 0.8426,
      "step": 167850
    },
    {
      "epoch": 2.9470320757035764,
      "grad_norm": 0.05897118214110981,
      "learning_rate": 7.607158713522951e-05,
      "loss": 0.8376,
      "step": 167860
    },
    {
      "epoch": 2.947207640583578,
      "grad_norm": 0.05875696078994256,
      "learning_rate": 7.60608027224893e-05,
      "loss": 0.8449,
      "step": 167870
    },
    {
      "epoch": 2.947383205463579,
      "grad_norm": 0.053709030571638394,
      "learning_rate": 7.605001862573346e-05,
      "loss": 0.8467,
      "step": 167880
    },
    {
      "epoch": 2.9475587703435804,
      "grad_norm": 0.05475490886281946,
      "learning_rate": 7.603923484509871e-05,
      "loss": 0.8447,
      "step": 167890
    },
    {
      "epoch": 2.947734335223582,
      "grad_norm": 0.06743129615930185,
      "learning_rate": 7.602845138072164e-05,
      "loss": 0.8415,
      "step": 167900
    },
    {
      "epoch": 2.9479099001035833,
      "grad_norm": 0.05086987174609467,
      "learning_rate": 7.601766823273886e-05,
      "loss": 0.8437,
      "step": 167910
    },
    {
      "epoch": 2.948085464983585,
      "grad_norm": 0.0587244108503237,
      "learning_rate": 7.600688540128701e-05,
      "loss": 0.8382,
      "step": 167920
    },
    {
      "epoch": 2.948261029863586,
      "grad_norm": 0.05286717918118161,
      "learning_rate": 7.599610288650274e-05,
      "loss": 0.843,
      "step": 167930
    },
    {
      "epoch": 2.9484365947435878,
      "grad_norm": 0.05086535414070614,
      "learning_rate": 7.598532068852261e-05,
      "loss": 0.8474,
      "step": 167940
    },
    {
      "epoch": 2.948612159623589,
      "grad_norm": 0.0405764007780522,
      "learning_rate": 7.597453880748328e-05,
      "loss": 0.8493,
      "step": 167950
    },
    {
      "epoch": 2.9487877245035903,
      "grad_norm": 0.04896297165384287,
      "learning_rate": 7.596375724352134e-05,
      "loss": 0.8431,
      "step": 167960
    },
    {
      "epoch": 2.9489632893835918,
      "grad_norm": 0.043128019717402453,
      "learning_rate": 7.595297599677338e-05,
      "loss": 0.8475,
      "step": 167970
    },
    {
      "epoch": 2.9491388542635932,
      "grad_norm": 0.046912508085332004,
      "learning_rate": 7.594219506737599e-05,
      "loss": 0.8461,
      "step": 167980
    },
    {
      "epoch": 2.9493144191435947,
      "grad_norm": 0.04703778561653673,
      "learning_rate": 7.593141445546583e-05,
      "loss": 0.8408,
      "step": 167990
    },
    {
      "epoch": 2.9494899840235957,
      "grad_norm": 0.045342409411449194,
      "learning_rate": 7.592063416117944e-05,
      "loss": 0.839,
      "step": 168000
    },
    {
      "epoch": 2.9496655489035972,
      "grad_norm": 0.05593102131841939,
      "learning_rate": 7.590985418465341e-05,
      "loss": 0.8474,
      "step": 168010
    },
    {
      "epoch": 2.9498411137835987,
      "grad_norm": 0.059301676448780705,
      "learning_rate": 7.589907452602436e-05,
      "loss": 0.841,
      "step": 168020
    },
    {
      "epoch": 2.9500166786636,
      "grad_norm": 0.05194284643239799,
      "learning_rate": 7.588829518542884e-05,
      "loss": 0.8465,
      "step": 168030
    },
    {
      "epoch": 2.9501922435436017,
      "grad_norm": 0.05529019323184446,
      "learning_rate": 7.587751616300343e-05,
      "loss": 0.8428,
      "step": 168040
    },
    {
      "epoch": 2.9503678084236027,
      "grad_norm": 0.04808647361527978,
      "learning_rate": 7.586673745888468e-05,
      "loss": 0.8488,
      "step": 168050
    },
    {
      "epoch": 2.9505433733036046,
      "grad_norm": 0.04975819342008567,
      "learning_rate": 7.585595907320921e-05,
      "loss": 0.8402,
      "step": 168060
    },
    {
      "epoch": 2.9507189381836056,
      "grad_norm": 0.0586336065148446,
      "learning_rate": 7.584518100611356e-05,
      "loss": 0.8375,
      "step": 168070
    },
    {
      "epoch": 2.950894503063607,
      "grad_norm": 0.07539323878946506,
      "learning_rate": 7.583440325773427e-05,
      "loss": 0.8394,
      "step": 168080
    },
    {
      "epoch": 2.9510700679436086,
      "grad_norm": 0.039633913099940576,
      "learning_rate": 7.582362582820794e-05,
      "loss": 0.845,
      "step": 168090
    },
    {
      "epoch": 2.95124563282361,
      "grad_norm": 0.06677743474199084,
      "learning_rate": 7.581284871767107e-05,
      "loss": 0.8482,
      "step": 168100
    },
    {
      "epoch": 2.9514211977036116,
      "grad_norm": 0.0467693023524509,
      "learning_rate": 7.580207192626023e-05,
      "loss": 0.8389,
      "step": 168110
    },
    {
      "epoch": 2.9515967625836126,
      "grad_norm": 0.05723427041565265,
      "learning_rate": 7.5791295454112e-05,
      "loss": 0.8421,
      "step": 168120
    },
    {
      "epoch": 2.951772327463614,
      "grad_norm": 0.0468938007776844,
      "learning_rate": 7.578051930136287e-05,
      "loss": 0.8367,
      "step": 168130
    },
    {
      "epoch": 2.9519478923436155,
      "grad_norm": 0.061955339527713794,
      "learning_rate": 7.57697434681494e-05,
      "loss": 0.8411,
      "step": 168140
    },
    {
      "epoch": 2.952123457223617,
      "grad_norm": 0.050350681248804804,
      "learning_rate": 7.575896795460807e-05,
      "loss": 0.8433,
      "step": 168150
    },
    {
      "epoch": 2.9522990221036185,
      "grad_norm": 0.051714408132565635,
      "learning_rate": 7.574819276087552e-05,
      "loss": 0.8385,
      "step": 168160
    },
    {
      "epoch": 2.9524745869836195,
      "grad_norm": 0.07381364169206747,
      "learning_rate": 7.573741788708818e-05,
      "loss": 0.8419,
      "step": 168170
    },
    {
      "epoch": 2.9526501518636215,
      "grad_norm": 0.05870476025688064,
      "learning_rate": 7.572664333338259e-05,
      "loss": 0.8387,
      "step": 168180
    },
    {
      "epoch": 2.9528257167436225,
      "grad_norm": 0.04114187498376406,
      "learning_rate": 7.571586909989527e-05,
      "loss": 0.8391,
      "step": 168190
    },
    {
      "epoch": 2.953001281623624,
      "grad_norm": 0.03838024391843325,
      "learning_rate": 7.570509518676276e-05,
      "loss": 0.8355,
      "step": 168200
    },
    {
      "epoch": 2.9531768465036254,
      "grad_norm": 0.03969178182245593,
      "learning_rate": 7.569432159412154e-05,
      "loss": 0.854,
      "step": 168210
    },
    {
      "epoch": 2.953352411383627,
      "grad_norm": 0.052199144822701424,
      "learning_rate": 7.568354832210807e-05,
      "loss": 0.8408,
      "step": 168220
    },
    {
      "epoch": 2.9535279762636284,
      "grad_norm": 0.055050452920475314,
      "learning_rate": 7.567277537085893e-05,
      "loss": 0.8422,
      "step": 168230
    },
    {
      "epoch": 2.9537035411436294,
      "grad_norm": 0.06902453854859493,
      "learning_rate": 7.566200274051059e-05,
      "loss": 0.8463,
      "step": 168240
    },
    {
      "epoch": 2.953879106023631,
      "grad_norm": 0.048937878770431036,
      "learning_rate": 7.56512304311995e-05,
      "loss": 0.8474,
      "step": 168250
    },
    {
      "epoch": 2.9540546709036324,
      "grad_norm": 0.12478754849888977,
      "learning_rate": 7.564045844306219e-05,
      "loss": 0.8434,
      "step": 168260
    },
    {
      "epoch": 2.954230235783634,
      "grad_norm": 0.08442486199663052,
      "learning_rate": 7.562968677623516e-05,
      "loss": 0.8358,
      "step": 168270
    },
    {
      "epoch": 2.9544058006636353,
      "grad_norm": 0.04656447768474815,
      "learning_rate": 7.561891543085481e-05,
      "loss": 0.8503,
      "step": 168280
    },
    {
      "epoch": 2.954581365543637,
      "grad_norm": 0.05944589916537408,
      "learning_rate": 7.560814440705771e-05,
      "loss": 0.8357,
      "step": 168290
    },
    {
      "epoch": 2.9547569304236383,
      "grad_norm": 0.05540753404013863,
      "learning_rate": 7.559737370498027e-05,
      "loss": 0.8427,
      "step": 168300
    },
    {
      "epoch": 2.9549324953036393,
      "grad_norm": 0.047349409590548175,
      "learning_rate": 7.558660332475899e-05,
      "loss": 0.844,
      "step": 168310
    },
    {
      "epoch": 2.955108060183641,
      "grad_norm": 0.04957502791727881,
      "learning_rate": 7.557583326653026e-05,
      "loss": 0.8459,
      "step": 168320
    },
    {
      "epoch": 2.9552836250636423,
      "grad_norm": 0.061507967461701105,
      "learning_rate": 7.556506353043064e-05,
      "loss": 0.8434,
      "step": 168330
    },
    {
      "epoch": 2.9554591899436438,
      "grad_norm": 0.06635647581897161,
      "learning_rate": 7.555429411659654e-05,
      "loss": 0.8415,
      "step": 168340
    },
    {
      "epoch": 2.9556347548236452,
      "grad_norm": 0.07663879783747689,
      "learning_rate": 7.55435250251644e-05,
      "loss": 0.8404,
      "step": 168350
    },
    {
      "epoch": 2.9558103197036463,
      "grad_norm": 0.05736600317411771,
      "learning_rate": 7.553275625627068e-05,
      "loss": 0.8479,
      "step": 168360
    },
    {
      "epoch": 2.9559858845836477,
      "grad_norm": 0.04523658092900591,
      "learning_rate": 7.552198781005182e-05,
      "loss": 0.8431,
      "step": 168370
    },
    {
      "epoch": 2.9561614494636492,
      "grad_norm": 0.04637458802874618,
      "learning_rate": 7.551121968664427e-05,
      "loss": 0.8346,
      "step": 168380
    },
    {
      "epoch": 2.9563370143436507,
      "grad_norm": 0.05955157292837442,
      "learning_rate": 7.550045188618442e-05,
      "loss": 0.8497,
      "step": 168390
    },
    {
      "epoch": 2.956512579223652,
      "grad_norm": 0.05181297075228381,
      "learning_rate": 7.548968440880876e-05,
      "loss": 0.8418,
      "step": 168400
    },
    {
      "epoch": 2.9566881441036537,
      "grad_norm": 0.04625287632300723,
      "learning_rate": 7.547891725465369e-05,
      "loss": 0.8464,
      "step": 168410
    },
    {
      "epoch": 2.956863708983655,
      "grad_norm": 0.049351324853537014,
      "learning_rate": 7.546815042385559e-05,
      "loss": 0.8407,
      "step": 168420
    },
    {
      "epoch": 2.957039273863656,
      "grad_norm": 0.06324081034453582,
      "learning_rate": 7.545738391655097e-05,
      "loss": 0.8463,
      "step": 168430
    },
    {
      "epoch": 2.9572148387436576,
      "grad_norm": 0.05456528000535669,
      "learning_rate": 7.544661773287617e-05,
      "loss": 0.8366,
      "step": 168440
    },
    {
      "epoch": 2.957390403623659,
      "grad_norm": 0.0641583877446992,
      "learning_rate": 7.54358518729676e-05,
      "loss": 0.8427,
      "step": 168450
    },
    {
      "epoch": 2.9575659685036606,
      "grad_norm": 0.04729272686481462,
      "learning_rate": 7.542508633696171e-05,
      "loss": 0.8449,
      "step": 168460
    },
    {
      "epoch": 2.957741533383662,
      "grad_norm": 0.05111797299923122,
      "learning_rate": 7.541432112499487e-05,
      "loss": 0.839,
      "step": 168470
    },
    {
      "epoch": 2.957917098263663,
      "grad_norm": 0.08608268299739921,
      "learning_rate": 7.54035562372035e-05,
      "loss": 0.8512,
      "step": 168480
    },
    {
      "epoch": 2.9580926631436646,
      "grad_norm": 0.07918159997770169,
      "learning_rate": 7.539279167372392e-05,
      "loss": 0.8449,
      "step": 168490
    },
    {
      "epoch": 2.958268228023666,
      "grad_norm": 0.054624711597465715,
      "learning_rate": 7.538202743469264e-05,
      "loss": 0.849,
      "step": 168500
    },
    {
      "epoch": 2.9584437929036675,
      "grad_norm": 0.055918372567179214,
      "learning_rate": 7.537126352024597e-05,
      "loss": 0.8383,
      "step": 168510
    },
    {
      "epoch": 2.958619357783669,
      "grad_norm": 0.04658735199541122,
      "learning_rate": 7.536049993052026e-05,
      "loss": 0.8464,
      "step": 168520
    },
    {
      "epoch": 2.9587949226636705,
      "grad_norm": 0.05966164727810955,
      "learning_rate": 7.534973666565198e-05,
      "loss": 0.8423,
      "step": 168530
    },
    {
      "epoch": 2.958970487543672,
      "grad_norm": 0.06328942466932053,
      "learning_rate": 7.533897372577743e-05,
      "loss": 0.8487,
      "step": 168540
    },
    {
      "epoch": 2.959146052423673,
      "grad_norm": 0.05823969853584251,
      "learning_rate": 7.532821111103303e-05,
      "loss": 0.8453,
      "step": 168550
    },
    {
      "epoch": 2.9593216173036745,
      "grad_norm": 0.05292372641394586,
      "learning_rate": 7.531744882155506e-05,
      "loss": 0.8413,
      "step": 168560
    },
    {
      "epoch": 2.959497182183676,
      "grad_norm": 0.05135196983765343,
      "learning_rate": 7.530668685747998e-05,
      "loss": 0.8402,
      "step": 168570
    },
    {
      "epoch": 2.9596727470636774,
      "grad_norm": 0.042856978673269555,
      "learning_rate": 7.529592521894409e-05,
      "loss": 0.8503,
      "step": 168580
    },
    {
      "epoch": 2.959848311943679,
      "grad_norm": 0.04856271775855568,
      "learning_rate": 7.528516390608374e-05,
      "loss": 0.8293,
      "step": 168590
    },
    {
      "epoch": 2.96002387682368,
      "grad_norm": 0.06266236638722485,
      "learning_rate": 7.527440291903532e-05,
      "loss": 0.8509,
      "step": 168600
    },
    {
      "epoch": 2.9601994417036814,
      "grad_norm": 0.05991195757900381,
      "learning_rate": 7.526364225793515e-05,
      "loss": 0.8375,
      "step": 168610
    },
    {
      "epoch": 2.960375006583683,
      "grad_norm": 0.04634599055432888,
      "learning_rate": 7.525288192291953e-05,
      "loss": 0.8382,
      "step": 168620
    },
    {
      "epoch": 2.9605505714636844,
      "grad_norm": 0.04851637815715352,
      "learning_rate": 7.524212191412485e-05,
      "loss": 0.8424,
      "step": 168630
    },
    {
      "epoch": 2.960726136343686,
      "grad_norm": 0.047219451053629964,
      "learning_rate": 7.523136223168742e-05,
      "loss": 0.849,
      "step": 168640
    },
    {
      "epoch": 2.9609017012236873,
      "grad_norm": 0.050356839475001136,
      "learning_rate": 7.522060287574358e-05,
      "loss": 0.8437,
      "step": 168650
    },
    {
      "epoch": 2.961077266103689,
      "grad_norm": 0.07565658247875105,
      "learning_rate": 7.520984384642959e-05,
      "loss": 0.8442,
      "step": 168660
    },
    {
      "epoch": 2.96125283098369,
      "grad_norm": 0.04756622138999191,
      "learning_rate": 7.519908514388186e-05,
      "loss": 0.837,
      "step": 168670
    },
    {
      "epoch": 2.9614283958636913,
      "grad_norm": 0.05217294627547827,
      "learning_rate": 7.518832676823668e-05,
      "loss": 0.842,
      "step": 168680
    },
    {
      "epoch": 2.961603960743693,
      "grad_norm": 0.045228606942661724,
      "learning_rate": 7.517756871963032e-05,
      "loss": 0.8459,
      "step": 168690
    },
    {
      "epoch": 2.9617795256236943,
      "grad_norm": 0.05103652630472418,
      "learning_rate": 7.516681099819913e-05,
      "loss": 0.8465,
      "step": 168700
    },
    {
      "epoch": 2.9619550905036958,
      "grad_norm": 0.07996916682746567,
      "learning_rate": 7.515605360407941e-05,
      "loss": 0.8456,
      "step": 168710
    },
    {
      "epoch": 2.962130655383697,
      "grad_norm": 0.05569438579159169,
      "learning_rate": 7.514529653740742e-05,
      "loss": 0.8418,
      "step": 168720
    },
    {
      "epoch": 2.9623062202636987,
      "grad_norm": 0.05478965968494344,
      "learning_rate": 7.513453979831946e-05,
      "loss": 0.8372,
      "step": 168730
    },
    {
      "epoch": 2.9624817851436998,
      "grad_norm": 0.060611035871923256,
      "learning_rate": 7.512378338695186e-05,
      "loss": 0.837,
      "step": 168740
    },
    {
      "epoch": 2.9626573500237012,
      "grad_norm": 0.04846726381402895,
      "learning_rate": 7.511302730344089e-05,
      "loss": 0.8389,
      "step": 168750
    },
    {
      "epoch": 2.9628329149037027,
      "grad_norm": 0.05717131412173837,
      "learning_rate": 7.51022715479228e-05,
      "loss": 0.8447,
      "step": 168760
    },
    {
      "epoch": 2.963008479783704,
      "grad_norm": 0.04935139912725604,
      "learning_rate": 7.509151612053394e-05,
      "loss": 0.8356,
      "step": 168770
    },
    {
      "epoch": 2.9631840446637057,
      "grad_norm": 0.05130019650026619,
      "learning_rate": 7.508076102141051e-05,
      "loss": 0.8406,
      "step": 168780
    },
    {
      "epoch": 2.9633596095437067,
      "grad_norm": 0.05950380622806272,
      "learning_rate": 7.50700062506888e-05,
      "loss": 0.8357,
      "step": 168790
    },
    {
      "epoch": 2.963535174423708,
      "grad_norm": 0.06489190557976145,
      "learning_rate": 7.505925180850508e-05,
      "loss": 0.8422,
      "step": 168800
    },
    {
      "epoch": 2.9637107393037097,
      "grad_norm": 0.04530482153467937,
      "learning_rate": 7.504849769499563e-05,
      "loss": 0.8442,
      "step": 168810
    },
    {
      "epoch": 2.963886304183711,
      "grad_norm": 0.06311054040310143,
      "learning_rate": 7.503774391029668e-05,
      "loss": 0.8523,
      "step": 168820
    },
    {
      "epoch": 2.9640618690637126,
      "grad_norm": 0.05262967847609385,
      "learning_rate": 7.502699045454445e-05,
      "loss": 0.8428,
      "step": 168830
    },
    {
      "epoch": 2.9642374339437136,
      "grad_norm": 0.06038320431844586,
      "learning_rate": 7.501623732787526e-05,
      "loss": 0.8476,
      "step": 168840
    },
    {
      "epoch": 2.9644129988237156,
      "grad_norm": 0.05214931612395326,
      "learning_rate": 7.500548453042534e-05,
      "loss": 0.8502,
      "step": 168850
    },
    {
      "epoch": 2.9645885637037166,
      "grad_norm": 0.054596038313717364,
      "learning_rate": 7.499473206233087e-05,
      "loss": 0.8412,
      "step": 168860
    },
    {
      "epoch": 2.964764128583718,
      "grad_norm": 0.06420486436050703,
      "learning_rate": 7.498397992372818e-05,
      "loss": 0.8473,
      "step": 168870
    },
    {
      "epoch": 2.9649396934637195,
      "grad_norm": 0.05773873653324499,
      "learning_rate": 7.497322811475342e-05,
      "loss": 0.8466,
      "step": 168880
    },
    {
      "epoch": 2.965115258343721,
      "grad_norm": 0.05452381864483177,
      "learning_rate": 7.496247663554287e-05,
      "loss": 0.8399,
      "step": 168890
    },
    {
      "epoch": 2.9652908232237225,
      "grad_norm": 0.04992187153787078,
      "learning_rate": 7.495172548623269e-05,
      "loss": 0.8481,
      "step": 168900
    },
    {
      "epoch": 2.9654663881037235,
      "grad_norm": 0.05621121064332261,
      "learning_rate": 7.494097466695917e-05,
      "loss": 0.8398,
      "step": 168910
    },
    {
      "epoch": 2.965641952983725,
      "grad_norm": 0.05870437997399555,
      "learning_rate": 7.493022417785852e-05,
      "loss": 0.8419,
      "step": 168920
    },
    {
      "epoch": 2.9658175178637265,
      "grad_norm": 0.06904724764877426,
      "learning_rate": 7.491947401906687e-05,
      "loss": 0.8434,
      "step": 168930
    },
    {
      "epoch": 2.965993082743728,
      "grad_norm": 0.05497003361286376,
      "learning_rate": 7.49087241907205e-05,
      "loss": 0.8437,
      "step": 168940
    },
    {
      "epoch": 2.9661686476237294,
      "grad_norm": 0.06275294156145579,
      "learning_rate": 7.489797469295562e-05,
      "loss": 0.85,
      "step": 168950
    },
    {
      "epoch": 2.9663442125037305,
      "grad_norm": 0.04512596590802216,
      "learning_rate": 7.488722552590838e-05,
      "loss": 0.8341,
      "step": 168960
    },
    {
      "epoch": 2.9665197773837324,
      "grad_norm": 0.050568142447064014,
      "learning_rate": 7.487647668971501e-05,
      "loss": 0.8384,
      "step": 168970
    },
    {
      "epoch": 2.9666953422637334,
      "grad_norm": 0.05109962284296371,
      "learning_rate": 7.48657281845117e-05,
      "loss": 0.8411,
      "step": 168980
    },
    {
      "epoch": 2.966870907143735,
      "grad_norm": 0.055220741780403565,
      "learning_rate": 7.48549800104346e-05,
      "loss": 0.8529,
      "step": 168990
    },
    {
      "epoch": 2.9670464720237364,
      "grad_norm": 0.04985091822027374,
      "learning_rate": 7.484423216761989e-05,
      "loss": 0.8475,
      "step": 169000
    },
    {
      "epoch": 2.967222036903738,
      "grad_norm": 0.05272193794025351,
      "learning_rate": 7.48334846562038e-05,
      "loss": 0.8401,
      "step": 169010
    },
    {
      "epoch": 2.9673976017837393,
      "grad_norm": 0.043682503153636976,
      "learning_rate": 7.482273747632249e-05,
      "loss": 0.8405,
      "step": 169020
    },
    {
      "epoch": 2.9675731666637404,
      "grad_norm": 0.05523527768549374,
      "learning_rate": 7.481199062811208e-05,
      "loss": 0.8434,
      "step": 169030
    },
    {
      "epoch": 2.967748731543742,
      "grad_norm": 0.07783328990742736,
      "learning_rate": 7.480124411170878e-05,
      "loss": 0.8445,
      "step": 169040
    },
    {
      "epoch": 2.9679242964237433,
      "grad_norm": 0.059717492526369105,
      "learning_rate": 7.479049792724877e-05,
      "loss": 0.8485,
      "step": 169050
    },
    {
      "epoch": 2.968099861303745,
      "grad_norm": 0.05392892633772013,
      "learning_rate": 7.477975207486815e-05,
      "loss": 0.8387,
      "step": 169060
    },
    {
      "epoch": 2.9682754261837463,
      "grad_norm": 0.05391961839958929,
      "learning_rate": 7.476900655470304e-05,
      "loss": 0.8421,
      "step": 169070
    },
    {
      "epoch": 2.9684509910637478,
      "grad_norm": 0.05324704260745852,
      "learning_rate": 7.475826136688971e-05,
      "loss": 0.8455,
      "step": 169080
    },
    {
      "epoch": 2.9686265559437492,
      "grad_norm": 0.05609446963325138,
      "learning_rate": 7.474751651156424e-05,
      "loss": 0.8452,
      "step": 169090
    },
    {
      "epoch": 2.9688021208237503,
      "grad_norm": 0.05486450178186058,
      "learning_rate": 7.473677198886273e-05,
      "loss": 0.8409,
      "step": 169100
    },
    {
      "epoch": 2.9689776857037518,
      "grad_norm": 0.04831977893805946,
      "learning_rate": 7.472602779892138e-05,
      "loss": 0.8425,
      "step": 169110
    },
    {
      "epoch": 2.9691532505837532,
      "grad_norm": 0.047889772859908225,
      "learning_rate": 7.47152839418763e-05,
      "loss": 0.8401,
      "step": 169120
    },
    {
      "epoch": 2.9693288154637547,
      "grad_norm": 0.04965646926620969,
      "learning_rate": 7.47045404178636e-05,
      "loss": 0.8436,
      "step": 169130
    },
    {
      "epoch": 2.969504380343756,
      "grad_norm": 0.089480715414352,
      "learning_rate": 7.46937972270194e-05,
      "loss": 0.8427,
      "step": 169140
    },
    {
      "epoch": 2.9696799452237572,
      "grad_norm": 0.04326977789929106,
      "learning_rate": 7.468305436947986e-05,
      "loss": 0.8429,
      "step": 169150
    },
    {
      "epoch": 2.9698555101037587,
      "grad_norm": 0.05907912177242429,
      "learning_rate": 7.467231184538104e-05,
      "loss": 0.8478,
      "step": 169160
    },
    {
      "epoch": 2.97003107498376,
      "grad_norm": 0.06365626603564883,
      "learning_rate": 7.466156965485906e-05,
      "loss": 0.8503,
      "step": 169170
    },
    {
      "epoch": 2.9702066398637617,
      "grad_norm": 0.052724842427474876,
      "learning_rate": 7.465082779805005e-05,
      "loss": 0.8393,
      "step": 169180
    },
    {
      "epoch": 2.970382204743763,
      "grad_norm": 0.05808607598216973,
      "learning_rate": 7.464008627509012e-05,
      "loss": 0.8431,
      "step": 169190
    },
    {
      "epoch": 2.9705577696237646,
      "grad_norm": 0.05295360506180756,
      "learning_rate": 7.462934508611533e-05,
      "loss": 0.8516,
      "step": 169200
    },
    {
      "epoch": 2.970733334503766,
      "grad_norm": 0.055603412511300083,
      "learning_rate": 7.46186042312618e-05,
      "loss": 0.8418,
      "step": 169210
    },
    {
      "epoch": 2.970908899383767,
      "grad_norm": 0.049772046053090185,
      "learning_rate": 7.460786371066562e-05,
      "loss": 0.8443,
      "step": 169220
    },
    {
      "epoch": 2.9710844642637686,
      "grad_norm": 0.05818245298839031,
      "learning_rate": 7.459712352446288e-05,
      "loss": 0.8369,
      "step": 169230
    },
    {
      "epoch": 2.97126002914377,
      "grad_norm": 0.07157322371041915,
      "learning_rate": 7.458638367278958e-05,
      "loss": 0.8455,
      "step": 169240
    },
    {
      "epoch": 2.9714355940237716,
      "grad_norm": 0.05330694296559712,
      "learning_rate": 7.457564415578191e-05,
      "loss": 0.8448,
      "step": 169250
    },
    {
      "epoch": 2.971611158903773,
      "grad_norm": 0.05881120848448991,
      "learning_rate": 7.45649049735759e-05,
      "loss": 0.8459,
      "step": 169260
    },
    {
      "epoch": 2.971786723783774,
      "grad_norm": 0.05096062241057413,
      "learning_rate": 7.455416612630758e-05,
      "loss": 0.8429,
      "step": 169270
    },
    {
      "epoch": 2.9719622886637755,
      "grad_norm": 0.04278002872093576,
      "learning_rate": 7.454342761411307e-05,
      "loss": 0.8432,
      "step": 169280
    },
    {
      "epoch": 2.972137853543777,
      "grad_norm": 0.05835168944900167,
      "learning_rate": 7.45326894371284e-05,
      "loss": 0.8395,
      "step": 169290
    },
    {
      "epoch": 2.9723134184237785,
      "grad_norm": 0.056750955453901505,
      "learning_rate": 7.452195159548961e-05,
      "loss": 0.8345,
      "step": 169300
    },
    {
      "epoch": 2.97248898330378,
      "grad_norm": 0.050582731865255665,
      "learning_rate": 7.451121408933276e-05,
      "loss": 0.8463,
      "step": 169310
    },
    {
      "epoch": 2.9726645481837815,
      "grad_norm": 0.04832329732001909,
      "learning_rate": 7.450047691879395e-05,
      "loss": 0.8476,
      "step": 169320
    },
    {
      "epoch": 2.972840113063783,
      "grad_norm": 0.055811087676733415,
      "learning_rate": 7.448974008400914e-05,
      "loss": 0.842,
      "step": 169330
    },
    {
      "epoch": 2.973015677943784,
      "grad_norm": 0.05164443900316336,
      "learning_rate": 7.447900358511439e-05,
      "loss": 0.8453,
      "step": 169340
    },
    {
      "epoch": 2.9731912428237854,
      "grad_norm": 0.04102340773307804,
      "learning_rate": 7.446826742224576e-05,
      "loss": 0.845,
      "step": 169350
    },
    {
      "epoch": 2.973366807703787,
      "grad_norm": 0.05117725897231588,
      "learning_rate": 7.445753159553928e-05,
      "loss": 0.8488,
      "step": 169360
    },
    {
      "epoch": 2.9735423725837884,
      "grad_norm": 0.06097823058859093,
      "learning_rate": 7.444679610513095e-05,
      "loss": 0.8576,
      "step": 169370
    },
    {
      "epoch": 2.97371793746379,
      "grad_norm": 0.05113923283994533,
      "learning_rate": 7.443606095115682e-05,
      "loss": 0.8395,
      "step": 169380
    },
    {
      "epoch": 2.973893502343791,
      "grad_norm": 0.06161564574439606,
      "learning_rate": 7.442532613375288e-05,
      "loss": 0.8429,
      "step": 169390
    },
    {
      "epoch": 2.974069067223793,
      "grad_norm": 0.0614683602751196,
      "learning_rate": 7.441459165305517e-05,
      "loss": 0.8387,
      "step": 169400
    },
    {
      "epoch": 2.974244632103794,
      "grad_norm": 0.052447958837884986,
      "learning_rate": 7.440385750919962e-05,
      "loss": 0.8452,
      "step": 169410
    },
    {
      "epoch": 2.9744201969837953,
      "grad_norm": 0.05780621098119954,
      "learning_rate": 7.439312370232233e-05,
      "loss": 0.8399,
      "step": 169420
    },
    {
      "epoch": 2.974595761863797,
      "grad_norm": 0.046879501814102,
      "learning_rate": 7.438239023255926e-05,
      "loss": 0.8479,
      "step": 169430
    },
    {
      "epoch": 2.9747713267437983,
      "grad_norm": 0.055364523210159895,
      "learning_rate": 7.43716571000464e-05,
      "loss": 0.8395,
      "step": 169440
    },
    {
      "epoch": 2.9749468916237998,
      "grad_norm": 0.09025764903432552,
      "learning_rate": 7.436092430491977e-05,
      "loss": 0.8448,
      "step": 169450
    },
    {
      "epoch": 2.975122456503801,
      "grad_norm": 0.056067628483124,
      "learning_rate": 7.435019184731531e-05,
      "loss": 0.8365,
      "step": 169460
    },
    {
      "epoch": 2.9752980213838023,
      "grad_norm": 0.06836727256194225,
      "learning_rate": 7.433945972736906e-05,
      "loss": 0.8484,
      "step": 169470
    },
    {
      "epoch": 2.9754735862638038,
      "grad_norm": 0.06681496704714468,
      "learning_rate": 7.432872794521692e-05,
      "loss": 0.8484,
      "step": 169480
    },
    {
      "epoch": 2.9756491511438052,
      "grad_norm": 0.058990406033963945,
      "learning_rate": 7.431799650099495e-05,
      "loss": 0.8358,
      "step": 169490
    },
    {
      "epoch": 2.9758247160238067,
      "grad_norm": 0.06840373248176704,
      "learning_rate": 7.430726539483906e-05,
      "loss": 0.8418,
      "step": 169500
    },
    {
      "epoch": 2.9760002809038077,
      "grad_norm": 0.05814389156153244,
      "learning_rate": 7.42965346268852e-05,
      "loss": 0.8365,
      "step": 169510
    },
    {
      "epoch": 2.9761758457838097,
      "grad_norm": 0.057937669123300735,
      "learning_rate": 7.42858041972694e-05,
      "loss": 0.8315,
      "step": 169520
    },
    {
      "epoch": 2.9763514106638107,
      "grad_norm": 0.04813126130427398,
      "learning_rate": 7.427507410612758e-05,
      "loss": 0.8473,
      "step": 169530
    },
    {
      "epoch": 2.976526975543812,
      "grad_norm": 0.09863817040373764,
      "learning_rate": 7.426434435359568e-05,
      "loss": 0.8372,
      "step": 169540
    },
    {
      "epoch": 2.9767025404238137,
      "grad_norm": 0.048802869850876646,
      "learning_rate": 7.425361493980968e-05,
      "loss": 0.8396,
      "step": 169550
    },
    {
      "epoch": 2.976878105303815,
      "grad_norm": 0.045639126445160376,
      "learning_rate": 7.42428858649055e-05,
      "loss": 0.8431,
      "step": 169560
    },
    {
      "epoch": 2.9770536701838166,
      "grad_norm": 0.06370854262053151,
      "learning_rate": 7.423215712901908e-05,
      "loss": 0.8387,
      "step": 169570
    },
    {
      "epoch": 2.9772292350638176,
      "grad_norm": 0.07563635939820122,
      "learning_rate": 7.422142873228632e-05,
      "loss": 0.8356,
      "step": 169580
    },
    {
      "epoch": 2.977404799943819,
      "grad_norm": 0.05401844260063269,
      "learning_rate": 7.421070067484324e-05,
      "loss": 0.8482,
      "step": 169590
    },
    {
      "epoch": 2.9775803648238206,
      "grad_norm": 0.06383308410711917,
      "learning_rate": 7.41999729568257e-05,
      "loss": 0.8452,
      "step": 169600
    },
    {
      "epoch": 2.977755929703822,
      "grad_norm": 0.05044443075931382,
      "learning_rate": 7.418924557836964e-05,
      "loss": 0.8486,
      "step": 169610
    },
    {
      "epoch": 2.9779314945838236,
      "grad_norm": 0.060889216345067484,
      "learning_rate": 7.417851853961098e-05,
      "loss": 0.8381,
      "step": 169620
    },
    {
      "epoch": 2.9781070594638246,
      "grad_norm": 0.053852193135811924,
      "learning_rate": 7.416779184068564e-05,
      "loss": 0.8405,
      "step": 169630
    },
    {
      "epoch": 2.9782826243438265,
      "grad_norm": 0.04659075819997242,
      "learning_rate": 7.415706548172952e-05,
      "loss": 0.8377,
      "step": 169640
    },
    {
      "epoch": 2.9784581892238275,
      "grad_norm": 0.06248010595068482,
      "learning_rate": 7.41463394628785e-05,
      "loss": 0.8468,
      "step": 169650
    },
    {
      "epoch": 2.978633754103829,
      "grad_norm": 0.08084670600664634,
      "learning_rate": 7.413561378426854e-05,
      "loss": 0.8377,
      "step": 169660
    },
    {
      "epoch": 2.9788093189838305,
      "grad_norm": 0.06581696598200197,
      "learning_rate": 7.412488844603549e-05,
      "loss": 0.8436,
      "step": 169670
    },
    {
      "epoch": 2.978984883863832,
      "grad_norm": 0.061200699610134926,
      "learning_rate": 7.41141634483152e-05,
      "loss": 0.8436,
      "step": 169680
    },
    {
      "epoch": 2.9791604487438335,
      "grad_norm": 0.052678852624720905,
      "learning_rate": 7.410343879124366e-05,
      "loss": 0.8465,
      "step": 169690
    },
    {
      "epoch": 2.9793360136238345,
      "grad_norm": 0.0613733434999498,
      "learning_rate": 7.409271447495672e-05,
      "loss": 0.8432,
      "step": 169700
    },
    {
      "epoch": 2.979511578503836,
      "grad_norm": 0.05587562946276938,
      "learning_rate": 7.408199049959021e-05,
      "loss": 0.8396,
      "step": 169710
    },
    {
      "epoch": 2.9796871433838374,
      "grad_norm": 0.042005635558149956,
      "learning_rate": 7.407126686528007e-05,
      "loss": 0.8443,
      "step": 169720
    },
    {
      "epoch": 2.979862708263839,
      "grad_norm": 0.05966108414395797,
      "learning_rate": 7.406054357216213e-05,
      "loss": 0.8493,
      "step": 169730
    },
    {
      "epoch": 2.9800382731438404,
      "grad_norm": 0.0484983172245412,
      "learning_rate": 7.404982062037228e-05,
      "loss": 0.8436,
      "step": 169740
    },
    {
      "epoch": 2.980213838023842,
      "grad_norm": 0.04108956855799947,
      "learning_rate": 7.40390980100463e-05,
      "loss": 0.8441,
      "step": 169750
    },
    {
      "epoch": 2.9803894029038434,
      "grad_norm": 0.052817894521178195,
      "learning_rate": 7.402837574132018e-05,
      "loss": 0.8406,
      "step": 169760
    },
    {
      "epoch": 2.9805649677838444,
      "grad_norm": 0.060582582450352813,
      "learning_rate": 7.40176538143297e-05,
      "loss": 0.8474,
      "step": 169770
    },
    {
      "epoch": 2.980740532663846,
      "grad_norm": 0.07098972014511638,
      "learning_rate": 7.40069322292107e-05,
      "loss": 0.8381,
      "step": 169780
    },
    {
      "epoch": 2.9809160975438473,
      "grad_norm": 0.0685583609482427,
      "learning_rate": 7.399621098609906e-05,
      "loss": 0.8463,
      "step": 169790
    },
    {
      "epoch": 2.981091662423849,
      "grad_norm": 0.053552629116923496,
      "learning_rate": 7.39854900851306e-05,
      "loss": 0.8407,
      "step": 169800
    },
    {
      "epoch": 2.9812672273038503,
      "grad_norm": 0.052610871033402944,
      "learning_rate": 7.397476952644117e-05,
      "loss": 0.8433,
      "step": 169810
    },
    {
      "epoch": 2.9814427921838513,
      "grad_norm": 0.04756424901291243,
      "learning_rate": 7.396404931016654e-05,
      "loss": 0.8418,
      "step": 169820
    },
    {
      "epoch": 2.981618357063853,
      "grad_norm": 0.0664439025924535,
      "learning_rate": 7.395332943644263e-05,
      "loss": 0.8461,
      "step": 169830
    },
    {
      "epoch": 2.9817939219438543,
      "grad_norm": 0.07239490765627432,
      "learning_rate": 7.394260990540522e-05,
      "loss": 0.8496,
      "step": 169840
    },
    {
      "epoch": 2.9819694868238558,
      "grad_norm": 0.04631434563909119,
      "learning_rate": 7.393189071719009e-05,
      "loss": 0.8429,
      "step": 169850
    },
    {
      "epoch": 2.9821450517038572,
      "grad_norm": 0.05759506034646767,
      "learning_rate": 7.392117187193312e-05,
      "loss": 0.842,
      "step": 169860
    },
    {
      "epoch": 2.9823206165838587,
      "grad_norm": 0.05859004674687411,
      "learning_rate": 7.39104533697701e-05,
      "loss": 0.8482,
      "step": 169870
    },
    {
      "epoch": 2.98249618146386,
      "grad_norm": 0.049015526539613725,
      "learning_rate": 7.389973521083681e-05,
      "loss": 0.8431,
      "step": 169880
    },
    {
      "epoch": 2.9826717463438612,
      "grad_norm": 0.04372805760047643,
      "learning_rate": 7.388901739526907e-05,
      "loss": 0.8449,
      "step": 169890
    },
    {
      "epoch": 2.9828473112238627,
      "grad_norm": 0.04904656002226597,
      "learning_rate": 7.38782999232027e-05,
      "loss": 0.8469,
      "step": 169900
    },
    {
      "epoch": 2.983022876103864,
      "grad_norm": 0.05335345411860456,
      "learning_rate": 7.386758279477345e-05,
      "loss": 0.843,
      "step": 169910
    },
    {
      "epoch": 2.9831984409838657,
      "grad_norm": 0.05368055592607822,
      "learning_rate": 7.385686601011708e-05,
      "loss": 0.8332,
      "step": 169920
    },
    {
      "epoch": 2.983374005863867,
      "grad_norm": 0.05503623204134334,
      "learning_rate": 7.384614956936949e-05,
      "loss": 0.8384,
      "step": 169930
    },
    {
      "epoch": 2.983549570743868,
      "grad_norm": 0.04250525112415654,
      "learning_rate": 7.383543347266637e-05,
      "loss": 0.8413,
      "step": 169940
    },
    {
      "epoch": 2.9837251356238697,
      "grad_norm": 0.05163159906736531,
      "learning_rate": 7.382471772014349e-05,
      "loss": 0.85,
      "step": 169950
    },
    {
      "epoch": 2.983900700503871,
      "grad_norm": 0.056972638197625124,
      "learning_rate": 7.381400231193667e-05,
      "loss": 0.8341,
      "step": 169960
    },
    {
      "epoch": 2.9840762653838726,
      "grad_norm": 0.058968572616418406,
      "learning_rate": 7.380328724818166e-05,
      "loss": 0.8344,
      "step": 169970
    },
    {
      "epoch": 2.984251830263874,
      "grad_norm": 0.05367295658355187,
      "learning_rate": 7.37925725290142e-05,
      "loss": 0.8447,
      "step": 169980
    },
    {
      "epoch": 2.9844273951438756,
      "grad_norm": 0.04575284195613215,
      "learning_rate": 7.378185815457003e-05,
      "loss": 0.8376,
      "step": 169990
    },
    {
      "epoch": 2.984602960023877,
      "grad_norm": 0.0461842669296962,
      "learning_rate": 7.377114412498499e-05,
      "loss": 0.8412,
      "step": 170000
    },
    {
      "epoch": 2.984778524903878,
      "grad_norm": 0.08426101401013081,
      "learning_rate": 7.376043044039475e-05,
      "loss": 0.8396,
      "step": 170010
    },
    {
      "epoch": 2.9849540897838795,
      "grad_norm": 0.05268664330701726,
      "learning_rate": 7.374971710093502e-05,
      "loss": 0.84,
      "step": 170020
    },
    {
      "epoch": 2.985129654663881,
      "grad_norm": 0.04377089839084537,
      "learning_rate": 7.373900410674166e-05,
      "loss": 0.8445,
      "step": 170030
    },
    {
      "epoch": 2.9853052195438825,
      "grad_norm": 0.049274770251911425,
      "learning_rate": 7.372829145795034e-05,
      "loss": 0.8381,
      "step": 170040
    },
    {
      "epoch": 2.985480784423884,
      "grad_norm": 0.05208854096900774,
      "learning_rate": 7.371757915469676e-05,
      "loss": 0.8396,
      "step": 170050
    },
    {
      "epoch": 2.985656349303885,
      "grad_norm": 0.06806988922789081,
      "learning_rate": 7.370686719711673e-05,
      "loss": 0.8385,
      "step": 170060
    },
    {
      "epoch": 2.9858319141838865,
      "grad_norm": 0.05169816717309718,
      "learning_rate": 7.369615558534591e-05,
      "loss": 0.8412,
      "step": 170070
    },
    {
      "epoch": 2.986007479063888,
      "grad_norm": 0.056122059173729255,
      "learning_rate": 7.368544431952002e-05,
      "loss": 0.8387,
      "step": 170080
    },
    {
      "epoch": 2.9861830439438894,
      "grad_norm": 0.05446188474568226,
      "learning_rate": 7.367473339977475e-05,
      "loss": 0.8313,
      "step": 170090
    },
    {
      "epoch": 2.986358608823891,
      "grad_norm": 0.057327158293982966,
      "learning_rate": 7.366402282624588e-05,
      "loss": 0.8387,
      "step": 170100
    },
    {
      "epoch": 2.9865341737038924,
      "grad_norm": 0.06713353679049303,
      "learning_rate": 7.365331259906907e-05,
      "loss": 0.8362,
      "step": 170110
    },
    {
      "epoch": 2.986709738583894,
      "grad_norm": 0.06257696161893787,
      "learning_rate": 7.364260271838004e-05,
      "loss": 0.8467,
      "step": 170120
    },
    {
      "epoch": 2.986885303463895,
      "grad_norm": 0.05975831895768334,
      "learning_rate": 7.363189318431449e-05,
      "loss": 0.8455,
      "step": 170130
    },
    {
      "epoch": 2.9870608683438964,
      "grad_norm": 0.052479366756728554,
      "learning_rate": 7.362118399700806e-05,
      "loss": 0.8341,
      "step": 170140
    },
    {
      "epoch": 2.987236433223898,
      "grad_norm": 0.05184763427723244,
      "learning_rate": 7.361047515659653e-05,
      "loss": 0.8403,
      "step": 170150
    },
    {
      "epoch": 2.9874119981038993,
      "grad_norm": 0.05507909512169412,
      "learning_rate": 7.359976666321548e-05,
      "loss": 0.8408,
      "step": 170160
    },
    {
      "epoch": 2.987587562983901,
      "grad_norm": 0.05227901719345902,
      "learning_rate": 7.358905851700066e-05,
      "loss": 0.8438,
      "step": 170170
    },
    {
      "epoch": 2.987763127863902,
      "grad_norm": 0.04719913347909704,
      "learning_rate": 7.357835071808774e-05,
      "loss": 0.8345,
      "step": 170180
    },
    {
      "epoch": 2.987938692743904,
      "grad_norm": 0.06729112066219092,
      "learning_rate": 7.356764326661231e-05,
      "loss": 0.8419,
      "step": 170190
    },
    {
      "epoch": 2.988114257623905,
      "grad_norm": 0.061404153393933465,
      "learning_rate": 7.355693616271015e-05,
      "loss": 0.851,
      "step": 170200
    },
    {
      "epoch": 2.9882898225039063,
      "grad_norm": 0.07972491589198283,
      "learning_rate": 7.354622940651686e-05,
      "loss": 0.8375,
      "step": 170210
    },
    {
      "epoch": 2.9884653873839078,
      "grad_norm": 0.053420208257593374,
      "learning_rate": 7.353552299816808e-05,
      "loss": 0.8488,
      "step": 170220
    },
    {
      "epoch": 2.9886409522639092,
      "grad_norm": 0.06324240837295032,
      "learning_rate": 7.35248169377995e-05,
      "loss": 0.8416,
      "step": 170230
    },
    {
      "epoch": 2.9888165171439107,
      "grad_norm": 0.06530730467911545,
      "learning_rate": 7.351411122554677e-05,
      "loss": 0.8496,
      "step": 170240
    },
    {
      "epoch": 2.9889920820239118,
      "grad_norm": 0.06575735723547085,
      "learning_rate": 7.350340586154553e-05,
      "loss": 0.8371,
      "step": 170250
    },
    {
      "epoch": 2.9891676469039132,
      "grad_norm": 0.05289449273918068,
      "learning_rate": 7.349270084593135e-05,
      "loss": 0.8377,
      "step": 170260
    },
    {
      "epoch": 2.9893432117839147,
      "grad_norm": 0.05259459612735132,
      "learning_rate": 7.348199617883996e-05,
      "loss": 0.8434,
      "step": 170270
    },
    {
      "epoch": 2.989518776663916,
      "grad_norm": 0.057690048012012045,
      "learning_rate": 7.347129186040697e-05,
      "loss": 0.8408,
      "step": 170280
    },
    {
      "epoch": 2.9896943415439177,
      "grad_norm": 0.055747281892168384,
      "learning_rate": 7.346058789076795e-05,
      "loss": 0.8352,
      "step": 170290
    },
    {
      "epoch": 2.9898699064239187,
      "grad_norm": 0.058813061095760316,
      "learning_rate": 7.344988427005859e-05,
      "loss": 0.8456,
      "step": 170300
    },
    {
      "epoch": 2.9900454713039206,
      "grad_norm": 0.05755152956809683,
      "learning_rate": 7.343918099841447e-05,
      "loss": 0.8433,
      "step": 170310
    },
    {
      "epoch": 2.9902210361839217,
      "grad_norm": 0.06166154915908936,
      "learning_rate": 7.342847807597122e-05,
      "loss": 0.8473,
      "step": 170320
    },
    {
      "epoch": 2.990396601063923,
      "grad_norm": 0.05445228341230468,
      "learning_rate": 7.34177755028644e-05,
      "loss": 0.8515,
      "step": 170330
    },
    {
      "epoch": 2.9905721659439246,
      "grad_norm": 0.055271865997678665,
      "learning_rate": 7.340707327922969e-05,
      "loss": 0.8383,
      "step": 170340
    },
    {
      "epoch": 2.990747730823926,
      "grad_norm": 0.05864934852110265,
      "learning_rate": 7.339637140520264e-05,
      "loss": 0.8487,
      "step": 170350
    },
    {
      "epoch": 2.9909232957039276,
      "grad_norm": 0.04779014075877621,
      "learning_rate": 7.33856698809188e-05,
      "loss": 0.84,
      "step": 170360
    },
    {
      "epoch": 2.9910988605839286,
      "grad_norm": 0.05021014854916971,
      "learning_rate": 7.337496870651388e-05,
      "loss": 0.8407,
      "step": 170370
    },
    {
      "epoch": 2.99127442546393,
      "grad_norm": 0.048684150471849236,
      "learning_rate": 7.336426788212337e-05,
      "loss": 0.8389,
      "step": 170380
    },
    {
      "epoch": 2.9914499903439316,
      "grad_norm": 0.04710905721558106,
      "learning_rate": 7.33535674078829e-05,
      "loss": 0.8375,
      "step": 170390
    },
    {
      "epoch": 2.991625555223933,
      "grad_norm": 0.04790921675403385,
      "learning_rate": 7.334286728392803e-05,
      "loss": 0.8383,
      "step": 170400
    },
    {
      "epoch": 2.9918011201039345,
      "grad_norm": 0.04504366733460787,
      "learning_rate": 7.333216751039432e-05,
      "loss": 0.8303,
      "step": 170410
    },
    {
      "epoch": 2.9919766849839355,
      "grad_norm": 0.048711659597662514,
      "learning_rate": 7.332146808741736e-05,
      "loss": 0.8357,
      "step": 170420
    },
    {
      "epoch": 2.9921522498639375,
      "grad_norm": 0.057032803664389085,
      "learning_rate": 7.331076901513268e-05,
      "loss": 0.8474,
      "step": 170430
    },
    {
      "epoch": 2.9923278147439385,
      "grad_norm": 0.05441315911106953,
      "learning_rate": 7.330007029367589e-05,
      "loss": 0.8379,
      "step": 170440
    },
    {
      "epoch": 2.99250337962394,
      "grad_norm": 0.05442541798045794,
      "learning_rate": 7.32893719231825e-05,
      "loss": 0.831,
      "step": 170450
    },
    {
      "epoch": 2.9926789445039415,
      "grad_norm": 0.05598315271819905,
      "learning_rate": 7.327867390378808e-05,
      "loss": 0.8423,
      "step": 170460
    },
    {
      "epoch": 2.992854509383943,
      "grad_norm": 0.052033621090318054,
      "learning_rate": 7.32679762356282e-05,
      "loss": 0.842,
      "step": 170470
    },
    {
      "epoch": 2.9930300742639444,
      "grad_norm": 0.05314491709744866,
      "learning_rate": 7.325727891883836e-05,
      "loss": 0.8384,
      "step": 170480
    },
    {
      "epoch": 2.9932056391439454,
      "grad_norm": 0.04741689102548389,
      "learning_rate": 7.324658195355412e-05,
      "loss": 0.842,
      "step": 170490
    },
    {
      "epoch": 2.993381204023947,
      "grad_norm": 0.058568567405839704,
      "learning_rate": 7.323588533991099e-05,
      "loss": 0.8423,
      "step": 170500
    },
    {
      "epoch": 2.9935567689039484,
      "grad_norm": 0.07303103940950065,
      "learning_rate": 7.322518907804453e-05,
      "loss": 0.846,
      "step": 170510
    },
    {
      "epoch": 2.99373233378395,
      "grad_norm": 0.06710055779379079,
      "learning_rate": 7.321449316809024e-05,
      "loss": 0.8477,
      "step": 170520
    },
    {
      "epoch": 2.9939078986639514,
      "grad_norm": 0.05474660746947266,
      "learning_rate": 7.320379761018361e-05,
      "loss": 0.8434,
      "step": 170530
    },
    {
      "epoch": 2.994083463543953,
      "grad_norm": 0.04858331239753577,
      "learning_rate": 7.319310240446024e-05,
      "loss": 0.8415,
      "step": 170540
    },
    {
      "epoch": 2.9942590284239543,
      "grad_norm": 0.05246367654257503,
      "learning_rate": 7.318240755105558e-05,
      "loss": 0.8444,
      "step": 170550
    },
    {
      "epoch": 2.9944345933039553,
      "grad_norm": 0.05712871727182397,
      "learning_rate": 7.317171305010513e-05,
      "loss": 0.8508,
      "step": 170560
    },
    {
      "epoch": 2.994610158183957,
      "grad_norm": 0.05709899146791603,
      "learning_rate": 7.316101890174445e-05,
      "loss": 0.8441,
      "step": 170570
    },
    {
      "epoch": 2.9947857230639583,
      "grad_norm": 0.0458033758665017,
      "learning_rate": 7.315032510610896e-05,
      "loss": 0.8384,
      "step": 170580
    },
    {
      "epoch": 2.9949612879439598,
      "grad_norm": 0.047526396344823585,
      "learning_rate": 7.313963166333422e-05,
      "loss": 0.8356,
      "step": 170590
    },
    {
      "epoch": 2.9951368528239612,
      "grad_norm": 0.04565686111293034,
      "learning_rate": 7.312893857355562e-05,
      "loss": 0.8482,
      "step": 170600
    },
    {
      "epoch": 2.9953124177039623,
      "grad_norm": 0.07362268294179275,
      "learning_rate": 7.311824583690877e-05,
      "loss": 0.8365,
      "step": 170610
    },
    {
      "epoch": 2.9954879825839638,
      "grad_norm": 0.06156580335941738,
      "learning_rate": 7.310755345352909e-05,
      "loss": 0.8485,
      "step": 170620
    },
    {
      "epoch": 2.9956635474639652,
      "grad_norm": 0.06548819389914416,
      "learning_rate": 7.309686142355202e-05,
      "loss": 0.8369,
      "step": 170630
    },
    {
      "epoch": 2.9958391123439667,
      "grad_norm": 0.05998777108796838,
      "learning_rate": 7.308616974711309e-05,
      "loss": 0.8407,
      "step": 170640
    },
    {
      "epoch": 2.996014677223968,
      "grad_norm": 0.058372120653331916,
      "learning_rate": 7.307547842434773e-05,
      "loss": 0.8513,
      "step": 170650
    },
    {
      "epoch": 2.9961902421039697,
      "grad_norm": 0.0538071059471943,
      "learning_rate": 7.306478745539141e-05,
      "loss": 0.8503,
      "step": 170660
    },
    {
      "epoch": 2.996365806983971,
      "grad_norm": 0.04647545117155671,
      "learning_rate": 7.305409684037958e-05,
      "loss": 0.8415,
      "step": 170670
    },
    {
      "epoch": 2.996541371863972,
      "grad_norm": 0.04704973853240036,
      "learning_rate": 7.304340657944772e-05,
      "loss": 0.8409,
      "step": 170680
    },
    {
      "epoch": 2.9967169367439737,
      "grad_norm": 0.09295305566176376,
      "learning_rate": 7.303271667273124e-05,
      "loss": 0.8446,
      "step": 170690
    },
    {
      "epoch": 2.996892501623975,
      "grad_norm": 0.05095672552659034,
      "learning_rate": 7.302202712036557e-05,
      "loss": 0.841,
      "step": 170700
    },
    {
      "epoch": 2.9970680665039766,
      "grad_norm": 0.07143422218423198,
      "learning_rate": 7.301133792248621e-05,
      "loss": 0.8381,
      "step": 170710
    },
    {
      "epoch": 2.997243631383978,
      "grad_norm": 0.05408948701619399,
      "learning_rate": 7.300064907922858e-05,
      "loss": 0.8404,
      "step": 170720
    },
    {
      "epoch": 2.997419196263979,
      "grad_norm": 0.048974084355588296,
      "learning_rate": 7.298996059072804e-05,
      "loss": 0.8445,
      "step": 170730
    },
    {
      "epoch": 2.9975947611439806,
      "grad_norm": 0.05714008339213229,
      "learning_rate": 7.297927245712013e-05,
      "loss": 0.8388,
      "step": 170740
    },
    {
      "epoch": 2.997770326023982,
      "grad_norm": 0.062230533687453446,
      "learning_rate": 7.296858467854019e-05,
      "loss": 0.848,
      "step": 170750
    },
    {
      "epoch": 2.9979458909039836,
      "grad_norm": 0.06247602540043573,
      "learning_rate": 7.295789725512365e-05,
      "loss": 0.8423,
      "step": 170760
    },
    {
      "epoch": 2.998121455783985,
      "grad_norm": 0.045216712972055574,
      "learning_rate": 7.29472101870059e-05,
      "loss": 0.8361,
      "step": 170770
    },
    {
      "epoch": 2.9982970206639865,
      "grad_norm": 0.045834626304271445,
      "learning_rate": 7.293652347432238e-05,
      "loss": 0.8454,
      "step": 170780
    },
    {
      "epoch": 2.998472585543988,
      "grad_norm": 0.05962768305935381,
      "learning_rate": 7.292583711720852e-05,
      "loss": 0.8396,
      "step": 170790
    },
    {
      "epoch": 2.998648150423989,
      "grad_norm": 0.04478891749787044,
      "learning_rate": 7.291515111579964e-05,
      "loss": 0.8469,
      "step": 170800
    },
    {
      "epoch": 2.9988237153039905,
      "grad_norm": 0.06671405432050598,
      "learning_rate": 7.290446547023121e-05,
      "loss": 0.8368,
      "step": 170810
    },
    {
      "epoch": 2.998999280183992,
      "grad_norm": 0.041001479979801886,
      "learning_rate": 7.289378018063858e-05,
      "loss": 0.8348,
      "step": 170820
    },
    {
      "epoch": 2.9991748450639935,
      "grad_norm": 0.05045777104600123,
      "learning_rate": 7.288309524715717e-05,
      "loss": 0.8408,
      "step": 170830
    },
    {
      "epoch": 2.999350409943995,
      "grad_norm": 0.04890461973751653,
      "learning_rate": 7.287241066992228e-05,
      "loss": 0.8396,
      "step": 170840
    },
    {
      "epoch": 2.999525974823996,
      "grad_norm": 0.06330839726894504,
      "learning_rate": 7.286172644906936e-05,
      "loss": 0.8397,
      "step": 170850
    },
    {
      "epoch": 2.999701539703998,
      "grad_norm": 0.051508673730117326,
      "learning_rate": 7.285104258473376e-05,
      "loss": 0.8364,
      "step": 170860
    },
    {
      "epoch": 2.999877104583999,
      "grad_norm": 0.05417489216685118,
      "learning_rate": 7.284035907705081e-05,
      "loss": 0.846,
      "step": 170870
    },
    {
      "epoch": 3.0000526694640004,
      "grad_norm": 0.05742736119916887,
      "learning_rate": 7.282967592615597e-05,
      "loss": 0.8438,
      "step": 170880
    },
    {
      "epoch": 3.000228234344002,
      "grad_norm": 0.054440744752247555,
      "learning_rate": 7.281899313218449e-05,
      "loss": 0.834,
      "step": 170890
    },
    {
      "epoch": 3.0004037992240034,
      "grad_norm": 0.07013086190726067,
      "learning_rate": 7.280831069527178e-05,
      "loss": 0.8455,
      "step": 170900
    },
    {
      "epoch": 3.000579364104005,
      "grad_norm": 0.05690449144935582,
      "learning_rate": 7.279762861555318e-05,
      "loss": 0.8427,
      "step": 170910
    },
    {
      "epoch": 3.000754928984006,
      "grad_norm": 0.046283970535679356,
      "learning_rate": 7.278694689316403e-05,
      "loss": 0.8493,
      "step": 170920
    },
    {
      "epoch": 3.0009304938640073,
      "grad_norm": 0.049712548707396054,
      "learning_rate": 7.277626552823969e-05,
      "loss": 0.8398,
      "step": 170930
    },
    {
      "epoch": 3.001106058744009,
      "grad_norm": 0.04937766840043774,
      "learning_rate": 7.276558452091542e-05,
      "loss": 0.8404,
      "step": 170940
    },
    {
      "epoch": 3.0012816236240103,
      "grad_norm": 0.06380492785993028,
      "learning_rate": 7.275490387132664e-05,
      "loss": 0.8391,
      "step": 170950
    },
    {
      "epoch": 3.0014571885040118,
      "grad_norm": 0.0545731133318981,
      "learning_rate": 7.274422357960865e-05,
      "loss": 0.8388,
      "step": 170960
    },
    {
      "epoch": 3.0016327533840133,
      "grad_norm": 0.04658169765322587,
      "learning_rate": 7.273354364589676e-05,
      "loss": 0.8385,
      "step": 170970
    },
    {
      "epoch": 3.0018083182640143,
      "grad_norm": 0.06300817729775568,
      "learning_rate": 7.272286407032628e-05,
      "loss": 0.8403,
      "step": 170980
    },
    {
      "epoch": 3.0019838831440158,
      "grad_norm": 0.04840269482310942,
      "learning_rate": 7.271218485303253e-05,
      "loss": 0.8454,
      "step": 170990
    },
    {
      "epoch": 3.0021594480240172,
      "grad_norm": 0.07420735955546094,
      "learning_rate": 7.270150599415083e-05,
      "loss": 0.8446,
      "step": 171000
    },
    {
      "epoch": 3.0023350129040187,
      "grad_norm": 0.05735373342971433,
      "learning_rate": 7.269082749381647e-05,
      "loss": 0.8421,
      "step": 171010
    },
    {
      "epoch": 3.00251057778402,
      "grad_norm": 0.05452477975203114,
      "learning_rate": 7.268014935216474e-05,
      "loss": 0.8415,
      "step": 171020
    },
    {
      "epoch": 3.0026861426640217,
      "grad_norm": 0.07701152673555976,
      "learning_rate": 7.266947156933097e-05,
      "loss": 0.8381,
      "step": 171030
    },
    {
      "epoch": 3.0028617075440227,
      "grad_norm": 0.06694678444100004,
      "learning_rate": 7.265879414545036e-05,
      "loss": 0.839,
      "step": 171040
    },
    {
      "epoch": 3.003037272424024,
      "grad_norm": 0.05722504674095029,
      "learning_rate": 7.264811708065832e-05,
      "loss": 0.8459,
      "step": 171050
    },
    {
      "epoch": 3.0032128373040257,
      "grad_norm": 0.04723396941863149,
      "learning_rate": 7.263744037509005e-05,
      "loss": 0.8409,
      "step": 171060
    },
    {
      "epoch": 3.003388402184027,
      "grad_norm": 0.053006148709238726,
      "learning_rate": 7.262676402888084e-05,
      "loss": 0.8397,
      "step": 171070
    },
    {
      "epoch": 3.0035639670640286,
      "grad_norm": 0.06602440888939964,
      "learning_rate": 7.2616088042166e-05,
      "loss": 0.8419,
      "step": 171080
    },
    {
      "epoch": 3.00373953194403,
      "grad_norm": 0.05210727737864016,
      "learning_rate": 7.260541241508074e-05,
      "loss": 0.8373,
      "step": 171090
    },
    {
      "epoch": 3.003915096824031,
      "grad_norm": 0.04787416627643182,
      "learning_rate": 7.259473714776036e-05,
      "loss": 0.8498,
      "step": 171100
    },
    {
      "epoch": 3.0040906617040326,
      "grad_norm": 0.04359113376153456,
      "learning_rate": 7.258406224034006e-05,
      "loss": 0.8506,
      "step": 171110
    },
    {
      "epoch": 3.004266226584034,
      "grad_norm": 0.07167907862825396,
      "learning_rate": 7.257338769295518e-05,
      "loss": 0.8469,
      "step": 171120
    },
    {
      "epoch": 3.0044417914640356,
      "grad_norm": 0.06038987906429896,
      "learning_rate": 7.256271350574093e-05,
      "loss": 0.8379,
      "step": 171130
    },
    {
      "epoch": 3.004617356344037,
      "grad_norm": 0.05548249564124799,
      "learning_rate": 7.255203967883252e-05,
      "loss": 0.8484,
      "step": 171140
    },
    {
      "epoch": 3.0047929212240385,
      "grad_norm": 0.05515496578860297,
      "learning_rate": 7.254136621236526e-05,
      "loss": 0.845,
      "step": 171150
    },
    {
      "epoch": 3.0049684861040395,
      "grad_norm": 0.0692798052040501,
      "learning_rate": 7.253069310647433e-05,
      "loss": 0.8434,
      "step": 171160
    },
    {
      "epoch": 3.005144050984041,
      "grad_norm": 0.05781575085244356,
      "learning_rate": 7.252002036129496e-05,
      "loss": 0.8374,
      "step": 171170
    },
    {
      "epoch": 3.0053196158640425,
      "grad_norm": 0.05616345538192738,
      "learning_rate": 7.250934797696238e-05,
      "loss": 0.849,
      "step": 171180
    },
    {
      "epoch": 3.005495180744044,
      "grad_norm": 0.05742974712783364,
      "learning_rate": 7.249867595361185e-05,
      "loss": 0.8356,
      "step": 171190
    },
    {
      "epoch": 3.0056707456240455,
      "grad_norm": 0.05375911241593356,
      "learning_rate": 7.248800429137855e-05,
      "loss": 0.8365,
      "step": 171200
    },
    {
      "epoch": 3.005846310504047,
      "grad_norm": 0.06169345618960183,
      "learning_rate": 7.247733299039766e-05,
      "loss": 0.8385,
      "step": 171210
    },
    {
      "epoch": 3.006021875384048,
      "grad_norm": 0.04763401624830053,
      "learning_rate": 7.246666205080449e-05,
      "loss": 0.8436,
      "step": 171220
    },
    {
      "epoch": 3.0061974402640494,
      "grad_norm": 0.06761917752551917,
      "learning_rate": 7.245599147273414e-05,
      "loss": 0.8456,
      "step": 171230
    },
    {
      "epoch": 3.006373005144051,
      "grad_norm": 0.058375944520206044,
      "learning_rate": 7.244532125632185e-05,
      "loss": 0.842,
      "step": 171240
    },
    {
      "epoch": 3.0065485700240524,
      "grad_norm": 0.05909187457216946,
      "learning_rate": 7.243465140170283e-05,
      "loss": 0.8401,
      "step": 171250
    },
    {
      "epoch": 3.006724134904054,
      "grad_norm": 0.06321761087808035,
      "learning_rate": 7.242398190901226e-05,
      "loss": 0.8436,
      "step": 171260
    },
    {
      "epoch": 3.0068996997840554,
      "grad_norm": 0.048867133083927344,
      "learning_rate": 7.24133127783853e-05,
      "loss": 0.8441,
      "step": 171270
    },
    {
      "epoch": 3.0070752646640564,
      "grad_norm": 0.05307242951729981,
      "learning_rate": 7.240264400995713e-05,
      "loss": 0.8442,
      "step": 171280
    },
    {
      "epoch": 3.007250829544058,
      "grad_norm": 0.053621635056094005,
      "learning_rate": 7.239197560386295e-05,
      "loss": 0.8403,
      "step": 171290
    },
    {
      "epoch": 3.0074263944240593,
      "grad_norm": 0.06937257148081516,
      "learning_rate": 7.238130756023794e-05,
      "loss": 0.8442,
      "step": 171300
    },
    {
      "epoch": 3.007601959304061,
      "grad_norm": 0.048170336595598925,
      "learning_rate": 7.237063987921721e-05,
      "loss": 0.845,
      "step": 171310
    },
    {
      "epoch": 3.0077775241840623,
      "grad_norm": 0.053032698649244474,
      "learning_rate": 7.2359972560936e-05,
      "loss": 0.8475,
      "step": 171320
    },
    {
      "epoch": 3.007953089064064,
      "grad_norm": 0.04762066563536456,
      "learning_rate": 7.234930560552943e-05,
      "loss": 0.8327,
      "step": 171330
    },
    {
      "epoch": 3.008128653944065,
      "grad_norm": 0.05695093665871964,
      "learning_rate": 7.233863901313263e-05,
      "loss": 0.8438,
      "step": 171340
    },
    {
      "epoch": 3.0083042188240663,
      "grad_norm": 0.048249550987154825,
      "learning_rate": 7.232797278388075e-05,
      "loss": 0.8415,
      "step": 171350
    },
    {
      "epoch": 3.0084797837040678,
      "grad_norm": 0.0719894822417738,
      "learning_rate": 7.2317306917909e-05,
      "loss": 0.8403,
      "step": 171360
    },
    {
      "epoch": 3.0086553485840692,
      "grad_norm": 0.04776238080810231,
      "learning_rate": 7.230664141535243e-05,
      "loss": 0.8367,
      "step": 171370
    },
    {
      "epoch": 3.0088309134640707,
      "grad_norm": 0.0524983991489296,
      "learning_rate": 7.229597627634618e-05,
      "loss": 0.8452,
      "step": 171380
    },
    {
      "epoch": 3.009006478344072,
      "grad_norm": 0.04526284060080202,
      "learning_rate": 7.228531150102547e-05,
      "loss": 0.838,
      "step": 171390
    },
    {
      "epoch": 3.0091820432240737,
      "grad_norm": 0.06607048784706504,
      "learning_rate": 7.227464708952535e-05,
      "loss": 0.8451,
      "step": 171400
    },
    {
      "epoch": 3.0093576081040747,
      "grad_norm": 0.04235195032095485,
      "learning_rate": 7.226398304198094e-05,
      "loss": 0.8374,
      "step": 171410
    },
    {
      "epoch": 3.009533172984076,
      "grad_norm": 0.07098243159545824,
      "learning_rate": 7.225331935852739e-05,
      "loss": 0.843,
      "step": 171420
    },
    {
      "epoch": 3.0097087378640777,
      "grad_norm": 0.10227379195254023,
      "learning_rate": 7.224265603929979e-05,
      "loss": 0.838,
      "step": 171430
    },
    {
      "epoch": 3.009884302744079,
      "grad_norm": 0.0544888541272487,
      "learning_rate": 7.223199308443327e-05,
      "loss": 0.8472,
      "step": 171440
    },
    {
      "epoch": 3.0100598676240806,
      "grad_norm": 0.06423822575947612,
      "learning_rate": 7.222133049406286e-05,
      "loss": 0.8443,
      "step": 171450
    },
    {
      "epoch": 3.010235432504082,
      "grad_norm": 0.05222744404746236,
      "learning_rate": 7.221066826832374e-05,
      "loss": 0.848,
      "step": 171460
    },
    {
      "epoch": 3.010410997384083,
      "grad_norm": 0.053549825193916406,
      "learning_rate": 7.220000640735099e-05,
      "loss": 0.8423,
      "step": 171470
    },
    {
      "epoch": 3.0105865622640846,
      "grad_norm": 0.052359154093377554,
      "learning_rate": 7.218934491127963e-05,
      "loss": 0.8475,
      "step": 171480
    },
    {
      "epoch": 3.010762127144086,
      "grad_norm": 0.043709399334888294,
      "learning_rate": 7.217868378024481e-05,
      "loss": 0.8432,
      "step": 171490
    },
    {
      "epoch": 3.0109376920240876,
      "grad_norm": 0.04786815940725118,
      "learning_rate": 7.216802301438161e-05,
      "loss": 0.8421,
      "step": 171500
    },
    {
      "epoch": 3.011113256904089,
      "grad_norm": 0.06263380414479991,
      "learning_rate": 7.215736261382508e-05,
      "loss": 0.8404,
      "step": 171510
    },
    {
      "epoch": 3.0112888217840905,
      "grad_norm": 0.05176284729224452,
      "learning_rate": 7.214670257871028e-05,
      "loss": 0.851,
      "step": 171520
    },
    {
      "epoch": 3.0114643866640916,
      "grad_norm": 0.05039396174548485,
      "learning_rate": 7.21360429091723e-05,
      "loss": 0.845,
      "step": 171530
    },
    {
      "epoch": 3.011639951544093,
      "grad_norm": 0.046737869298228756,
      "learning_rate": 7.212538360534617e-05,
      "loss": 0.8503,
      "step": 171540
    },
    {
      "epoch": 3.0118155164240945,
      "grad_norm": 0.0575551809825628,
      "learning_rate": 7.211472466736695e-05,
      "loss": 0.8429,
      "step": 171550
    },
    {
      "epoch": 3.011991081304096,
      "grad_norm": 0.057848700675114946,
      "learning_rate": 7.210406609536974e-05,
      "loss": 0.8468,
      "step": 171560
    },
    {
      "epoch": 3.0121666461840975,
      "grad_norm": 0.05866858581038805,
      "learning_rate": 7.209340788948955e-05,
      "loss": 0.8408,
      "step": 171570
    },
    {
      "epoch": 3.012342211064099,
      "grad_norm": 0.04591232141011969,
      "learning_rate": 7.208275004986139e-05,
      "loss": 0.8393,
      "step": 171580
    },
    {
      "epoch": 3.0125177759441,
      "grad_norm": 0.06020332312490891,
      "learning_rate": 7.207209257662037e-05,
      "loss": 0.8406,
      "step": 171590
    },
    {
      "epoch": 3.0126933408241015,
      "grad_norm": 0.058296126698897235,
      "learning_rate": 7.206143546990146e-05,
      "loss": 0.8404,
      "step": 171600
    },
    {
      "epoch": 3.012868905704103,
      "grad_norm": 0.08278322265125347,
      "learning_rate": 7.205077872983972e-05,
      "loss": 0.8438,
      "step": 171610
    },
    {
      "epoch": 3.0130444705841044,
      "grad_norm": 0.044582571045944555,
      "learning_rate": 7.204012235657012e-05,
      "loss": 0.8486,
      "step": 171620
    },
    {
      "epoch": 3.013220035464106,
      "grad_norm": 0.053815371859569264,
      "learning_rate": 7.202946635022776e-05,
      "loss": 0.8478,
      "step": 171630
    },
    {
      "epoch": 3.0133956003441074,
      "grad_norm": 0.052054315648345444,
      "learning_rate": 7.201881071094761e-05,
      "loss": 0.8463,
      "step": 171640
    },
    {
      "epoch": 3.0135711652241084,
      "grad_norm": 0.0576961316232121,
      "learning_rate": 7.200815543886467e-05,
      "loss": 0.8357,
      "step": 171650
    },
    {
      "epoch": 3.01374673010411,
      "grad_norm": 0.07717729838922582,
      "learning_rate": 7.199750053411398e-05,
      "loss": 0.8407,
      "step": 171660
    },
    {
      "epoch": 3.0139222949841113,
      "grad_norm": 0.06177731913046722,
      "learning_rate": 7.198684599683049e-05,
      "loss": 0.8496,
      "step": 171670
    },
    {
      "epoch": 3.014097859864113,
      "grad_norm": 0.051107954666948306,
      "learning_rate": 7.197619182714925e-05,
      "loss": 0.8444,
      "step": 171680
    },
    {
      "epoch": 3.0142734247441143,
      "grad_norm": 0.05122203871449132,
      "learning_rate": 7.196553802520519e-05,
      "loss": 0.843,
      "step": 171690
    },
    {
      "epoch": 3.014448989624116,
      "grad_norm": 0.04864713119767333,
      "learning_rate": 7.195488459113334e-05,
      "loss": 0.8486,
      "step": 171700
    },
    {
      "epoch": 3.014624554504117,
      "grad_norm": 0.05385484110137351,
      "learning_rate": 7.194423152506869e-05,
      "loss": 0.8452,
      "step": 171710
    },
    {
      "epoch": 3.0148001193841183,
      "grad_norm": 0.05351432321201021,
      "learning_rate": 7.193357882714612e-05,
      "loss": 0.8396,
      "step": 171720
    },
    {
      "epoch": 3.0149756842641198,
      "grad_norm": 0.08588695032615334,
      "learning_rate": 7.192292649750077e-05,
      "loss": 0.8454,
      "step": 171730
    },
    {
      "epoch": 3.0151512491441212,
      "grad_norm": 0.06366792725907204,
      "learning_rate": 7.191227453626747e-05,
      "loss": 0.8432,
      "step": 171740
    },
    {
      "epoch": 3.0153268140241227,
      "grad_norm": 0.05683266223105861,
      "learning_rate": 7.190162294358121e-05,
      "loss": 0.8399,
      "step": 171750
    },
    {
      "epoch": 3.015502378904124,
      "grad_norm": 0.04895540633145298,
      "learning_rate": 7.1890971719577e-05,
      "loss": 0.8405,
      "step": 171760
    },
    {
      "epoch": 3.0156779437841252,
      "grad_norm": 0.057470048599106316,
      "learning_rate": 7.188032086438976e-05,
      "loss": 0.8475,
      "step": 171770
    },
    {
      "epoch": 3.0158535086641267,
      "grad_norm": 0.08027046011449912,
      "learning_rate": 7.186967037815442e-05,
      "loss": 0.8387,
      "step": 171780
    },
    {
      "epoch": 3.016029073544128,
      "grad_norm": 0.05613393880626938,
      "learning_rate": 7.185902026100589e-05,
      "loss": 0.8439,
      "step": 171790
    },
    {
      "epoch": 3.0162046384241297,
      "grad_norm": 0.05034488390617439,
      "learning_rate": 7.184837051307922e-05,
      "loss": 0.8459,
      "step": 171800
    },
    {
      "epoch": 3.016380203304131,
      "grad_norm": 0.051801687225974356,
      "learning_rate": 7.183772113450928e-05,
      "loss": 0.8424,
      "step": 171810
    },
    {
      "epoch": 3.0165557681841326,
      "grad_norm": 0.06366488357883712,
      "learning_rate": 7.182707212543098e-05,
      "loss": 0.8457,
      "step": 171820
    },
    {
      "epoch": 3.0167313330641337,
      "grad_norm": 0.051173270081191845,
      "learning_rate": 7.18164234859793e-05,
      "loss": 0.8411,
      "step": 171830
    },
    {
      "epoch": 3.016906897944135,
      "grad_norm": 0.06017877441095334,
      "learning_rate": 7.180577521628909e-05,
      "loss": 0.8458,
      "step": 171840
    },
    {
      "epoch": 3.0170824628241366,
      "grad_norm": 0.057958741430119816,
      "learning_rate": 7.179512731649533e-05,
      "loss": 0.8382,
      "step": 171850
    },
    {
      "epoch": 3.017258027704138,
      "grad_norm": 0.051528943650910104,
      "learning_rate": 7.178447978673289e-05,
      "loss": 0.8481,
      "step": 171860
    },
    {
      "epoch": 3.0174335925841396,
      "grad_norm": 0.06901610870083762,
      "learning_rate": 7.177383262713669e-05,
      "loss": 0.8444,
      "step": 171870
    },
    {
      "epoch": 3.017609157464141,
      "grad_norm": 0.06128238154365561,
      "learning_rate": 7.176318583784165e-05,
      "loss": 0.8428,
      "step": 171880
    },
    {
      "epoch": 3.017784722344142,
      "grad_norm": 0.062326316536093135,
      "learning_rate": 7.17525394189826e-05,
      "loss": 0.8467,
      "step": 171890
    },
    {
      "epoch": 3.0179602872241436,
      "grad_norm": 0.06115650897793941,
      "learning_rate": 7.174189337069451e-05,
      "loss": 0.8431,
      "step": 171900
    },
    {
      "epoch": 3.018135852104145,
      "grad_norm": 0.05858376550323635,
      "learning_rate": 7.173124769311226e-05,
      "loss": 0.838,
      "step": 171910
    },
    {
      "epoch": 3.0183114169841465,
      "grad_norm": 0.04373776023326719,
      "learning_rate": 7.172060238637067e-05,
      "loss": 0.8493,
      "step": 171920
    },
    {
      "epoch": 3.018486981864148,
      "grad_norm": 0.04355841401362683,
      "learning_rate": 7.17099574506047e-05,
      "loss": 0.8356,
      "step": 171930
    },
    {
      "epoch": 3.0186625467441495,
      "grad_norm": 0.06144269779960712,
      "learning_rate": 7.169931288594917e-05,
      "loss": 0.8376,
      "step": 171940
    },
    {
      "epoch": 3.0188381116241505,
      "grad_norm": 0.06149665312306409,
      "learning_rate": 7.168866869253896e-05,
      "loss": 0.8416,
      "step": 171950
    },
    {
      "epoch": 3.019013676504152,
      "grad_norm": 0.044188521065432314,
      "learning_rate": 7.167802487050889e-05,
      "loss": 0.8394,
      "step": 171960
    },
    {
      "epoch": 3.0191892413841535,
      "grad_norm": 0.04673305666200695,
      "learning_rate": 7.166738141999391e-05,
      "loss": 0.8395,
      "step": 171970
    },
    {
      "epoch": 3.019364806264155,
      "grad_norm": 0.05184130773687192,
      "learning_rate": 7.165673834112884e-05,
      "loss": 0.8432,
      "step": 171980
    },
    {
      "epoch": 3.0195403711441564,
      "grad_norm": 0.0614901274198665,
      "learning_rate": 7.164609563404849e-05,
      "loss": 0.8392,
      "step": 171990
    },
    {
      "epoch": 3.019715936024158,
      "grad_norm": 0.05252103989414253,
      "learning_rate": 7.163545329888777e-05,
      "loss": 0.8376,
      "step": 172000
    },
    {
      "epoch": 3.019891500904159,
      "grad_norm": 0.05198473228854114,
      "learning_rate": 7.162481133578147e-05,
      "loss": 0.841,
      "step": 172010
    },
    {
      "epoch": 3.0200670657841604,
      "grad_norm": 0.0867712291364977,
      "learning_rate": 7.161416974486443e-05,
      "loss": 0.8418,
      "step": 172020
    },
    {
      "epoch": 3.020242630664162,
      "grad_norm": 0.07126694789426687,
      "learning_rate": 7.160352852627148e-05,
      "loss": 0.8499,
      "step": 172030
    },
    {
      "epoch": 3.0204181955441634,
      "grad_norm": 0.07217177983302062,
      "learning_rate": 7.159288768013747e-05,
      "loss": 0.8431,
      "step": 172040
    },
    {
      "epoch": 3.020593760424165,
      "grad_norm": 0.053495210461679495,
      "learning_rate": 7.158224720659723e-05,
      "loss": 0.839,
      "step": 172050
    },
    {
      "epoch": 3.0207693253041663,
      "grad_norm": 0.05748795478301729,
      "learning_rate": 7.157160710578551e-05,
      "loss": 0.8376,
      "step": 172060
    },
    {
      "epoch": 3.0209448901841673,
      "grad_norm": 0.06940956317927163,
      "learning_rate": 7.156096737783719e-05,
      "loss": 0.8442,
      "step": 172070
    },
    {
      "epoch": 3.021120455064169,
      "grad_norm": 0.04933989099311346,
      "learning_rate": 7.155032802288706e-05,
      "loss": 0.8428,
      "step": 172080
    },
    {
      "epoch": 3.0212960199441703,
      "grad_norm": 0.07548667199647713,
      "learning_rate": 7.153968904106991e-05,
      "loss": 0.8427,
      "step": 172090
    },
    {
      "epoch": 3.0214715848241718,
      "grad_norm": 0.0652787513373655,
      "learning_rate": 7.152905043252054e-05,
      "loss": 0.8386,
      "step": 172100
    },
    {
      "epoch": 3.0216471497041733,
      "grad_norm": 0.04554663715464172,
      "learning_rate": 7.151841219737378e-05,
      "loss": 0.8378,
      "step": 172110
    },
    {
      "epoch": 3.0218227145841747,
      "grad_norm": 0.052491160939757996,
      "learning_rate": 7.150777433576438e-05,
      "loss": 0.8413,
      "step": 172120
    },
    {
      "epoch": 3.021998279464176,
      "grad_norm": 0.048555715138503035,
      "learning_rate": 7.149713684782708e-05,
      "loss": 0.8451,
      "step": 172130
    },
    {
      "epoch": 3.0221738443441772,
      "grad_norm": 0.04318917019668998,
      "learning_rate": 7.148649973369677e-05,
      "loss": 0.8352,
      "step": 172140
    },
    {
      "epoch": 3.0223494092241787,
      "grad_norm": 0.05362128619887078,
      "learning_rate": 7.147586299350815e-05,
      "loss": 0.8479,
      "step": 172150
    },
    {
      "epoch": 3.02252497410418,
      "grad_norm": 0.06973730459390685,
      "learning_rate": 7.146522662739599e-05,
      "loss": 0.8439,
      "step": 172160
    },
    {
      "epoch": 3.0227005389841817,
      "grad_norm": 0.05810731040199876,
      "learning_rate": 7.145459063549507e-05,
      "loss": 0.8403,
      "step": 172170
    },
    {
      "epoch": 3.022876103864183,
      "grad_norm": 0.056031370114583215,
      "learning_rate": 7.144395501794018e-05,
      "loss": 0.8448,
      "step": 172180
    },
    {
      "epoch": 3.0230516687441846,
      "grad_norm": 0.04792246292254009,
      "learning_rate": 7.143331977486603e-05,
      "loss": 0.8457,
      "step": 172190
    },
    {
      "epoch": 3.0232272336241857,
      "grad_norm": 0.05664421900019744,
      "learning_rate": 7.142268490640735e-05,
      "loss": 0.8293,
      "step": 172200
    },
    {
      "epoch": 3.023402798504187,
      "grad_norm": 0.07839252510459502,
      "learning_rate": 7.141205041269896e-05,
      "loss": 0.8426,
      "step": 172210
    },
    {
      "epoch": 3.0235783633841886,
      "grad_norm": 0.05924131902736301,
      "learning_rate": 7.140141629387557e-05,
      "loss": 0.8482,
      "step": 172220
    },
    {
      "epoch": 3.02375392826419,
      "grad_norm": 0.053840825121746975,
      "learning_rate": 7.139078255007184e-05,
      "loss": 0.8353,
      "step": 172230
    },
    {
      "epoch": 3.0239294931441916,
      "grad_norm": 0.05157375267306251,
      "learning_rate": 7.138014918142261e-05,
      "loss": 0.8436,
      "step": 172240
    },
    {
      "epoch": 3.024105058024193,
      "grad_norm": 0.05603717100106106,
      "learning_rate": 7.13695161880626e-05,
      "loss": 0.8432,
      "step": 172250
    },
    {
      "epoch": 3.024280622904194,
      "grad_norm": 0.06653589936129839,
      "learning_rate": 7.135888357012645e-05,
      "loss": 0.8397,
      "step": 172260
    },
    {
      "epoch": 3.0244561877841956,
      "grad_norm": 0.05386793250359579,
      "learning_rate": 7.134825132774896e-05,
      "loss": 0.851,
      "step": 172270
    },
    {
      "epoch": 3.024631752664197,
      "grad_norm": 0.0533080091843935,
      "learning_rate": 7.133761946106481e-05,
      "loss": 0.8448,
      "step": 172280
    },
    {
      "epoch": 3.0248073175441985,
      "grad_norm": 0.05711837875665944,
      "learning_rate": 7.13269879702087e-05,
      "loss": 0.8405,
      "step": 172290
    },
    {
      "epoch": 3.0249828824242,
      "grad_norm": 0.057343958901438105,
      "learning_rate": 7.13163568553153e-05,
      "loss": 0.8487,
      "step": 172300
    },
    {
      "epoch": 3.0251584473042015,
      "grad_norm": 0.05986600183176731,
      "learning_rate": 7.130572611651938e-05,
      "loss": 0.8416,
      "step": 172310
    },
    {
      "epoch": 3.0253340121842025,
      "grad_norm": 0.05578259179574344,
      "learning_rate": 7.12950957539556e-05,
      "loss": 0.8383,
      "step": 172320
    },
    {
      "epoch": 3.025509577064204,
      "grad_norm": 0.06356693380948866,
      "learning_rate": 7.128446576775863e-05,
      "loss": 0.84,
      "step": 172330
    },
    {
      "epoch": 3.0256851419442055,
      "grad_norm": 0.06323896327916934,
      "learning_rate": 7.127383615806322e-05,
      "loss": 0.8536,
      "step": 172340
    },
    {
      "epoch": 3.025860706824207,
      "grad_norm": 0.053228481140856626,
      "learning_rate": 7.126320692500399e-05,
      "loss": 0.8392,
      "step": 172350
    },
    {
      "epoch": 3.0260362717042084,
      "grad_norm": 0.07768825709070243,
      "learning_rate": 7.125257806871562e-05,
      "loss": 0.8398,
      "step": 172360
    },
    {
      "epoch": 3.02621183658421,
      "grad_norm": 0.06127705314656558,
      "learning_rate": 7.124194958933276e-05,
      "loss": 0.837,
      "step": 172370
    },
    {
      "epoch": 3.026387401464211,
      "grad_norm": 0.05243373552527549,
      "learning_rate": 7.123132148699013e-05,
      "loss": 0.8435,
      "step": 172380
    },
    {
      "epoch": 3.0265629663442124,
      "grad_norm": 0.05251648133341359,
      "learning_rate": 7.122069376182235e-05,
      "loss": 0.8428,
      "step": 172390
    },
    {
      "epoch": 3.026738531224214,
      "grad_norm": 0.04693973753096576,
      "learning_rate": 7.121006641396407e-05,
      "loss": 0.842,
      "step": 172400
    },
    {
      "epoch": 3.0269140961042154,
      "grad_norm": 0.056254665679083646,
      "learning_rate": 7.119943944354997e-05,
      "loss": 0.8515,
      "step": 172410
    },
    {
      "epoch": 3.027089660984217,
      "grad_norm": 0.08051219997879495,
      "learning_rate": 7.118881285071471e-05,
      "loss": 0.853,
      "step": 172420
    },
    {
      "epoch": 3.0272652258642183,
      "grad_norm": 0.04713868061170432,
      "learning_rate": 7.117818663559287e-05,
      "loss": 0.8319,
      "step": 172430
    },
    {
      "epoch": 3.0274407907442193,
      "grad_norm": 0.06097676174743066,
      "learning_rate": 7.116756079831915e-05,
      "loss": 0.8414,
      "step": 172440
    },
    {
      "epoch": 3.027616355624221,
      "grad_norm": 0.055711766345339,
      "learning_rate": 7.115693533902815e-05,
      "loss": 0.8383,
      "step": 172450
    },
    {
      "epoch": 3.0277919205042223,
      "grad_norm": 0.06581109058639242,
      "learning_rate": 7.11463102578545e-05,
      "loss": 0.8374,
      "step": 172460
    },
    {
      "epoch": 3.027967485384224,
      "grad_norm": 0.05500091433784306,
      "learning_rate": 7.113568555493276e-05,
      "loss": 0.8413,
      "step": 172470
    },
    {
      "epoch": 3.0281430502642253,
      "grad_norm": 0.08660285193116445,
      "learning_rate": 7.112506123039767e-05,
      "loss": 0.8473,
      "step": 172480
    },
    {
      "epoch": 3.0283186151442267,
      "grad_norm": 0.06272971920377854,
      "learning_rate": 7.111443728438378e-05,
      "loss": 0.839,
      "step": 172490
    },
    {
      "epoch": 3.0284941800242278,
      "grad_norm": 0.04755037197203741,
      "learning_rate": 7.110381371702567e-05,
      "loss": 0.8438,
      "step": 172500
    },
    {
      "epoch": 3.0286697449042292,
      "grad_norm": 0.054499343450983444,
      "learning_rate": 7.109319052845799e-05,
      "loss": 0.8435,
      "step": 172510
    },
    {
      "epoch": 3.0288453097842307,
      "grad_norm": 0.05157908899243053,
      "learning_rate": 7.108256771881531e-05,
      "loss": 0.836,
      "step": 172520
    },
    {
      "epoch": 3.029020874664232,
      "grad_norm": 0.0638731669407866,
      "learning_rate": 7.107194528823225e-05,
      "loss": 0.8415,
      "step": 172530
    },
    {
      "epoch": 3.0291964395442337,
      "grad_norm": 0.0538921121706394,
      "learning_rate": 7.106132323684334e-05,
      "loss": 0.8422,
      "step": 172540
    },
    {
      "epoch": 3.029372004424235,
      "grad_norm": 0.07324206721748568,
      "learning_rate": 7.105070156478323e-05,
      "loss": 0.8428,
      "step": 172550
    },
    {
      "epoch": 3.029547569304236,
      "grad_norm": 0.04630736792881751,
      "learning_rate": 7.104008027218648e-05,
      "loss": 0.8415,
      "step": 172560
    },
    {
      "epoch": 3.0297231341842377,
      "grad_norm": 0.04668643355531951,
      "learning_rate": 7.10294593591876e-05,
      "loss": 0.8419,
      "step": 172570
    },
    {
      "epoch": 3.029898699064239,
      "grad_norm": 0.04894442682149659,
      "learning_rate": 7.101883882592127e-05,
      "loss": 0.8447,
      "step": 172580
    },
    {
      "epoch": 3.0300742639442406,
      "grad_norm": 0.04206262350029431,
      "learning_rate": 7.100821867252198e-05,
      "loss": 0.845,
      "step": 172590
    },
    {
      "epoch": 3.030249828824242,
      "grad_norm": 0.046986050088413514,
      "learning_rate": 7.099759889912429e-05,
      "loss": 0.8427,
      "step": 172600
    },
    {
      "epoch": 3.0304253937042436,
      "grad_norm": 0.05089779702812183,
      "learning_rate": 7.09869795058628e-05,
      "loss": 0.8465,
      "step": 172610
    },
    {
      "epoch": 3.0306009585842446,
      "grad_norm": 0.0583258138170358,
      "learning_rate": 7.097636049287202e-05,
      "loss": 0.8435,
      "step": 172620
    },
    {
      "epoch": 3.030776523464246,
      "grad_norm": 0.05916986742018186,
      "learning_rate": 7.096574186028652e-05,
      "loss": 0.8353,
      "step": 172630
    },
    {
      "epoch": 3.0309520883442476,
      "grad_norm": 0.05694817444877599,
      "learning_rate": 7.095512360824078e-05,
      "loss": 0.8406,
      "step": 172640
    },
    {
      "epoch": 3.031127653224249,
      "grad_norm": 0.05997310663476509,
      "learning_rate": 7.094450573686942e-05,
      "loss": 0.8406,
      "step": 172650
    },
    {
      "epoch": 3.0313032181042505,
      "grad_norm": 0.05615986337736947,
      "learning_rate": 7.093388824630693e-05,
      "loss": 0.8381,
      "step": 172660
    },
    {
      "epoch": 3.031478782984252,
      "grad_norm": 0.05079433867206173,
      "learning_rate": 7.092327113668782e-05,
      "loss": 0.8496,
      "step": 172670
    },
    {
      "epoch": 3.031654347864253,
      "grad_norm": 0.04809195516520837,
      "learning_rate": 7.091265440814666e-05,
      "loss": 0.8406,
      "step": 172680
    },
    {
      "epoch": 3.0318299127442545,
      "grad_norm": 0.05564907077020405,
      "learning_rate": 7.090203806081792e-05,
      "loss": 0.8453,
      "step": 172690
    },
    {
      "epoch": 3.032005477624256,
      "grad_norm": 0.04921245636231936,
      "learning_rate": 7.089142209483613e-05,
      "loss": 0.8363,
      "step": 172700
    },
    {
      "epoch": 3.0321810425042575,
      "grad_norm": 0.059636878060774334,
      "learning_rate": 7.088080651033578e-05,
      "loss": 0.843,
      "step": 172710
    },
    {
      "epoch": 3.032356607384259,
      "grad_norm": 0.04965137567824993,
      "learning_rate": 7.087019130745141e-05,
      "loss": 0.8491,
      "step": 172720
    },
    {
      "epoch": 3.0325321722642604,
      "grad_norm": 0.06702268071942273,
      "learning_rate": 7.085957648631748e-05,
      "loss": 0.8354,
      "step": 172730
    },
    {
      "epoch": 3.0327077371442615,
      "grad_norm": 0.05231505185436787,
      "learning_rate": 7.084896204706846e-05,
      "loss": 0.8427,
      "step": 172740
    },
    {
      "epoch": 3.032883302024263,
      "grad_norm": 0.05090390795837563,
      "learning_rate": 7.083834798983889e-05,
      "loss": 0.8384,
      "step": 172750
    },
    {
      "epoch": 3.0330588669042644,
      "grad_norm": 0.058114870849432974,
      "learning_rate": 7.082773431476326e-05,
      "loss": 0.8439,
      "step": 172760
    },
    {
      "epoch": 3.033234431784266,
      "grad_norm": 0.057083751559024315,
      "learning_rate": 7.081712102197598e-05,
      "loss": 0.8479,
      "step": 172770
    },
    {
      "epoch": 3.0334099966642674,
      "grad_norm": 0.05363797974967333,
      "learning_rate": 7.080650811161157e-05,
      "loss": 0.8409,
      "step": 172780
    },
    {
      "epoch": 3.033585561544269,
      "grad_norm": 0.05320532278770529,
      "learning_rate": 7.07958955838045e-05,
      "loss": 0.84,
      "step": 172790
    },
    {
      "epoch": 3.03376112642427,
      "grad_norm": 0.047268831620782444,
      "learning_rate": 7.078528343868922e-05,
      "loss": 0.8439,
      "step": 172800
    },
    {
      "epoch": 3.0339366913042713,
      "grad_norm": 0.0601084984745601,
      "learning_rate": 7.077467167640015e-05,
      "loss": 0.8376,
      "step": 172810
    },
    {
      "epoch": 3.034112256184273,
      "grad_norm": 0.0461607022726688,
      "learning_rate": 7.076406029707184e-05,
      "loss": 0.8456,
      "step": 172820
    },
    {
      "epoch": 3.0342878210642743,
      "grad_norm": 0.06932266221572474,
      "learning_rate": 7.075344930083866e-05,
      "loss": 0.8432,
      "step": 172830
    },
    {
      "epoch": 3.034463385944276,
      "grad_norm": 0.04797413783053031,
      "learning_rate": 7.074283868783506e-05,
      "loss": 0.8452,
      "step": 172840
    },
    {
      "epoch": 3.0346389508242773,
      "grad_norm": 0.05329475115306119,
      "learning_rate": 7.073222845819552e-05,
      "loss": 0.8439,
      "step": 172850
    },
    {
      "epoch": 3.0348145157042783,
      "grad_norm": 0.04780938442012379,
      "learning_rate": 7.072161861205443e-05,
      "loss": 0.8442,
      "step": 172860
    },
    {
      "epoch": 3.0349900805842798,
      "grad_norm": 0.07055280482767992,
      "learning_rate": 7.071100914954625e-05,
      "loss": 0.8517,
      "step": 172870
    },
    {
      "epoch": 3.0351656454642812,
      "grad_norm": 0.04276490135469776,
      "learning_rate": 7.070040007080537e-05,
      "loss": 0.846,
      "step": 172880
    },
    {
      "epoch": 3.0353412103442827,
      "grad_norm": 0.05657317176390512,
      "learning_rate": 7.068979137596622e-05,
      "loss": 0.8408,
      "step": 172890
    },
    {
      "epoch": 3.035516775224284,
      "grad_norm": 0.06521364471169394,
      "learning_rate": 7.067918306516325e-05,
      "loss": 0.8466,
      "step": 172900
    },
    {
      "epoch": 3.0356923401042857,
      "grad_norm": 0.0543013284594152,
      "learning_rate": 7.066857513853079e-05,
      "loss": 0.8402,
      "step": 172910
    },
    {
      "epoch": 3.035867904984287,
      "grad_norm": 0.07640849931290362,
      "learning_rate": 7.065796759620334e-05,
      "loss": 0.8454,
      "step": 172920
    },
    {
      "epoch": 3.036043469864288,
      "grad_norm": 0.05345044512660462,
      "learning_rate": 7.064736043831525e-05,
      "loss": 0.8484,
      "step": 172930
    },
    {
      "epoch": 3.0362190347442897,
      "grad_norm": 0.04830476638669778,
      "learning_rate": 7.063675366500092e-05,
      "loss": 0.8411,
      "step": 172940
    },
    {
      "epoch": 3.036394599624291,
      "grad_norm": 0.04921905324757683,
      "learning_rate": 7.062614727639474e-05,
      "loss": 0.8456,
      "step": 172950
    },
    {
      "epoch": 3.0365701645042926,
      "grad_norm": 0.043879005400671224,
      "learning_rate": 7.06155412726311e-05,
      "loss": 0.8519,
      "step": 172960
    },
    {
      "epoch": 3.036745729384294,
      "grad_norm": 0.07406466257141797,
      "learning_rate": 7.060493565384438e-05,
      "loss": 0.8451,
      "step": 172970
    },
    {
      "epoch": 3.0369212942642956,
      "grad_norm": 0.06463953759465707,
      "learning_rate": 7.059433042016891e-05,
      "loss": 0.8401,
      "step": 172980
    },
    {
      "epoch": 3.0370968591442966,
      "grad_norm": 0.061753542796278475,
      "learning_rate": 7.058372557173914e-05,
      "loss": 0.8312,
      "step": 172990
    },
    {
      "epoch": 3.037272424024298,
      "grad_norm": 0.0564265341997786,
      "learning_rate": 7.05731211086894e-05,
      "loss": 0.8398,
      "step": 173000
    },
    {
      "epoch": 3.0374479889042996,
      "grad_norm": 0.060578271739294,
      "learning_rate": 7.056251703115402e-05,
      "loss": 0.8447,
      "step": 173010
    },
    {
      "epoch": 3.037623553784301,
      "grad_norm": 0.06502500628468491,
      "learning_rate": 7.05519133392674e-05,
      "loss": 0.8423,
      "step": 173020
    },
    {
      "epoch": 3.0377991186643025,
      "grad_norm": 0.05420499615889656,
      "learning_rate": 7.054131003316387e-05,
      "loss": 0.8342,
      "step": 173030
    },
    {
      "epoch": 3.037974683544304,
      "grad_norm": 0.05161584000214233,
      "learning_rate": 7.05307071129778e-05,
      "loss": 0.8398,
      "step": 173040
    },
    {
      "epoch": 3.038150248424305,
      "grad_norm": 0.05058747075460645,
      "learning_rate": 7.052010457884346e-05,
      "loss": 0.8419,
      "step": 173050
    },
    {
      "epoch": 3.0383258133043065,
      "grad_norm": 0.05348940316953709,
      "learning_rate": 7.050950243089528e-05,
      "loss": 0.8412,
      "step": 173060
    },
    {
      "epoch": 3.038501378184308,
      "grad_norm": 0.06461307732056834,
      "learning_rate": 7.049890066926755e-05,
      "loss": 0.8359,
      "step": 173070
    },
    {
      "epoch": 3.0386769430643095,
      "grad_norm": 0.06632527936374245,
      "learning_rate": 7.048829929409453e-05,
      "loss": 0.8373,
      "step": 173080
    },
    {
      "epoch": 3.038852507944311,
      "grad_norm": 0.05455534089937088,
      "learning_rate": 7.047769830551067e-05,
      "loss": 0.8408,
      "step": 173090
    },
    {
      "epoch": 3.0390280728243124,
      "grad_norm": 0.048241723224340546,
      "learning_rate": 7.046709770365021e-05,
      "loss": 0.8457,
      "step": 173100
    },
    {
      "epoch": 3.0392036377043135,
      "grad_norm": 0.0655389397385138,
      "learning_rate": 7.045649748864745e-05,
      "loss": 0.8448,
      "step": 173110
    },
    {
      "epoch": 3.039379202584315,
      "grad_norm": 0.07176156771966735,
      "learning_rate": 7.044589766063675e-05,
      "loss": 0.8495,
      "step": 173120
    },
    {
      "epoch": 3.0395547674643164,
      "grad_norm": 0.06246732258535992,
      "learning_rate": 7.043529821975237e-05,
      "loss": 0.8365,
      "step": 173130
    },
    {
      "epoch": 3.039730332344318,
      "grad_norm": 0.046905147865942194,
      "learning_rate": 7.042469916612864e-05,
      "loss": 0.8501,
      "step": 173140
    },
    {
      "epoch": 3.0399058972243194,
      "grad_norm": 0.05967482271792556,
      "learning_rate": 7.041410049989978e-05,
      "loss": 0.8357,
      "step": 173150
    },
    {
      "epoch": 3.040081462104321,
      "grad_norm": 0.04903925129634139,
      "learning_rate": 7.040350222120017e-05,
      "loss": 0.8478,
      "step": 173160
    },
    {
      "epoch": 3.040257026984322,
      "grad_norm": 0.04946229900998376,
      "learning_rate": 7.039290433016404e-05,
      "loss": 0.8388,
      "step": 173170
    },
    {
      "epoch": 3.0404325918643234,
      "grad_norm": 0.05485817701394587,
      "learning_rate": 7.038230682692567e-05,
      "loss": 0.844,
      "step": 173180
    },
    {
      "epoch": 3.040608156744325,
      "grad_norm": 0.05003868847570746,
      "learning_rate": 7.037170971161936e-05,
      "loss": 0.8431,
      "step": 173190
    },
    {
      "epoch": 3.0407837216243263,
      "grad_norm": 0.06397365191101874,
      "learning_rate": 7.036111298437935e-05,
      "loss": 0.8437,
      "step": 173200
    },
    {
      "epoch": 3.040959286504328,
      "grad_norm": 0.06188433592504548,
      "learning_rate": 7.035051664533993e-05,
      "loss": 0.8458,
      "step": 173210
    },
    {
      "epoch": 3.0411348513843293,
      "grad_norm": 0.05308355829870945,
      "learning_rate": 7.033992069463528e-05,
      "loss": 0.8419,
      "step": 173220
    },
    {
      "epoch": 3.0413104162643303,
      "grad_norm": 0.06625264467141226,
      "learning_rate": 7.032932513239977e-05,
      "loss": 0.8421,
      "step": 173230
    },
    {
      "epoch": 3.0414859811443318,
      "grad_norm": 0.05299396195324524,
      "learning_rate": 7.031872995876757e-05,
      "loss": 0.8475,
      "step": 173240
    },
    {
      "epoch": 3.0416615460243333,
      "grad_norm": 0.07670345444668518,
      "learning_rate": 7.03081351738729e-05,
      "loss": 0.837,
      "step": 173250
    },
    {
      "epoch": 3.0418371109043347,
      "grad_norm": 0.06299600949857569,
      "learning_rate": 7.029754077785008e-05,
      "loss": 0.8429,
      "step": 173260
    },
    {
      "epoch": 3.042012675784336,
      "grad_norm": 0.07997691448158156,
      "learning_rate": 7.02869467708333e-05,
      "loss": 0.8404,
      "step": 173270
    },
    {
      "epoch": 3.0421882406643377,
      "grad_norm": 0.04708935134376923,
      "learning_rate": 7.027635315295679e-05,
      "loss": 0.8348,
      "step": 173280
    },
    {
      "epoch": 3.0423638055443387,
      "grad_norm": 0.055265539765142514,
      "learning_rate": 7.026575992435479e-05,
      "loss": 0.8427,
      "step": 173290
    },
    {
      "epoch": 3.04253937042434,
      "grad_norm": 0.057410297703122365,
      "learning_rate": 7.025516708516149e-05,
      "loss": 0.8468,
      "step": 173300
    },
    {
      "epoch": 3.0427149353043417,
      "grad_norm": 0.05845430531114512,
      "learning_rate": 7.02445746355111e-05,
      "loss": 0.853,
      "step": 173310
    },
    {
      "epoch": 3.042890500184343,
      "grad_norm": 0.04977321702769646,
      "learning_rate": 7.023398257553782e-05,
      "loss": 0.8406,
      "step": 173320
    },
    {
      "epoch": 3.0430660650643446,
      "grad_norm": 0.07370513848652244,
      "learning_rate": 7.022339090537592e-05,
      "loss": 0.8435,
      "step": 173330
    },
    {
      "epoch": 3.043241629944346,
      "grad_norm": 0.051193298554062905,
      "learning_rate": 7.021279962515956e-05,
      "loss": 0.8421,
      "step": 173340
    },
    {
      "epoch": 3.043417194824347,
      "grad_norm": 0.04826401085479484,
      "learning_rate": 7.02022087350229e-05,
      "loss": 0.8357,
      "step": 173350
    },
    {
      "epoch": 3.0435927597043486,
      "grad_norm": 0.060832224072190764,
      "learning_rate": 7.019161823510018e-05,
      "loss": 0.8359,
      "step": 173360
    },
    {
      "epoch": 3.04376832458435,
      "grad_norm": 0.05087403071368211,
      "learning_rate": 7.018102812552556e-05,
      "loss": 0.839,
      "step": 173370
    },
    {
      "epoch": 3.0439438894643516,
      "grad_norm": 0.07831284944984315,
      "learning_rate": 7.017043840643323e-05,
      "loss": 0.8342,
      "step": 173380
    },
    {
      "epoch": 3.044119454344353,
      "grad_norm": 0.04225407351434069,
      "learning_rate": 7.015984907795731e-05,
      "loss": 0.8431,
      "step": 173390
    },
    {
      "epoch": 3.0442950192243545,
      "grad_norm": 0.05024200982455116,
      "learning_rate": 7.014926014023205e-05,
      "loss": 0.8436,
      "step": 173400
    },
    {
      "epoch": 3.0444705841043556,
      "grad_norm": 0.05069432546986197,
      "learning_rate": 7.013867159339157e-05,
      "loss": 0.8339,
      "step": 173410
    },
    {
      "epoch": 3.044646148984357,
      "grad_norm": 0.048480662861661,
      "learning_rate": 7.012808343757001e-05,
      "loss": 0.8257,
      "step": 173420
    },
    {
      "epoch": 3.0448217138643585,
      "grad_norm": 0.05210376152918246,
      "learning_rate": 7.011749567290157e-05,
      "loss": 0.8458,
      "step": 173430
    },
    {
      "epoch": 3.04499727874436,
      "grad_norm": 0.05657730404229541,
      "learning_rate": 7.010690829952039e-05,
      "loss": 0.8388,
      "step": 173440
    },
    {
      "epoch": 3.0451728436243615,
      "grad_norm": 0.04813330064381283,
      "learning_rate": 7.009632131756059e-05,
      "loss": 0.8385,
      "step": 173450
    },
    {
      "epoch": 3.045348408504363,
      "grad_norm": 0.05255122886194093,
      "learning_rate": 7.008573472715633e-05,
      "loss": 0.8389,
      "step": 173460
    },
    {
      "epoch": 3.045523973384364,
      "grad_norm": 0.039849894979013235,
      "learning_rate": 7.007514852844174e-05,
      "loss": 0.8411,
      "step": 173470
    },
    {
      "epoch": 3.0456995382643655,
      "grad_norm": 0.05555755551858423,
      "learning_rate": 7.006456272155094e-05,
      "loss": 0.8373,
      "step": 173480
    },
    {
      "epoch": 3.045875103144367,
      "grad_norm": 0.06562046001620803,
      "learning_rate": 7.005397730661801e-05,
      "loss": 0.8477,
      "step": 173490
    },
    {
      "epoch": 3.0460506680243684,
      "grad_norm": 0.06519058971802054,
      "learning_rate": 7.004339228377719e-05,
      "loss": 0.8416,
      "step": 173500
    },
    {
      "epoch": 3.04622623290437,
      "grad_norm": 0.05326189921297259,
      "learning_rate": 7.003280765316248e-05,
      "loss": 0.8454,
      "step": 173510
    },
    {
      "epoch": 3.0464017977843714,
      "grad_norm": 0.08044783269062097,
      "learning_rate": 7.002222341490805e-05,
      "loss": 0.844,
      "step": 173520
    },
    {
      "epoch": 3.0465773626643724,
      "grad_norm": 0.06259262853439787,
      "learning_rate": 7.001163956914799e-05,
      "loss": 0.8499,
      "step": 173530
    },
    {
      "epoch": 3.046752927544374,
      "grad_norm": 0.06547296510287215,
      "learning_rate": 7.00010561160164e-05,
      "loss": 0.8471,
      "step": 173540
    },
    {
      "epoch": 3.0469284924243754,
      "grad_norm": 0.06467025080221318,
      "learning_rate": 6.999047305564736e-05,
      "loss": 0.8466,
      "step": 173550
    },
    {
      "epoch": 3.047104057304377,
      "grad_norm": 0.08015807254613085,
      "learning_rate": 6.997989038817495e-05,
      "loss": 0.8426,
      "step": 173560
    },
    {
      "epoch": 3.0472796221843783,
      "grad_norm": 0.0744323800105358,
      "learning_rate": 6.99693081137333e-05,
      "loss": 0.8403,
      "step": 173570
    },
    {
      "epoch": 3.04745518706438,
      "grad_norm": 0.08763269363198892,
      "learning_rate": 6.995872623245648e-05,
      "loss": 0.8348,
      "step": 173580
    },
    {
      "epoch": 3.0476307519443813,
      "grad_norm": 0.05426240818597345,
      "learning_rate": 6.994814474447848e-05,
      "loss": 0.8446,
      "step": 173590
    },
    {
      "epoch": 3.0478063168243823,
      "grad_norm": 0.052981154520635856,
      "learning_rate": 6.993756364993348e-05,
      "loss": 0.8414,
      "step": 173600
    },
    {
      "epoch": 3.0479818817043838,
      "grad_norm": 0.05994469881455825,
      "learning_rate": 6.992698294895551e-05,
      "loss": 0.8417,
      "step": 173610
    },
    {
      "epoch": 3.0481574465843853,
      "grad_norm": 0.059044249704652094,
      "learning_rate": 6.99164026416786e-05,
      "loss": 0.8447,
      "step": 173620
    },
    {
      "epoch": 3.0483330114643867,
      "grad_norm": 0.05003088962790335,
      "learning_rate": 6.990582272823685e-05,
      "loss": 0.8518,
      "step": 173630
    },
    {
      "epoch": 3.048508576344388,
      "grad_norm": 0.04738719253185804,
      "learning_rate": 6.989524320876428e-05,
      "loss": 0.8378,
      "step": 173640
    },
    {
      "epoch": 3.0486841412243897,
      "grad_norm": 0.050650855342271865,
      "learning_rate": 6.988466408339493e-05,
      "loss": 0.8438,
      "step": 173650
    },
    {
      "epoch": 3.0488597061043907,
      "grad_norm": 0.06603929581385094,
      "learning_rate": 6.987408535226283e-05,
      "loss": 0.8444,
      "step": 173660
    },
    {
      "epoch": 3.049035270984392,
      "grad_norm": 0.055592158154153345,
      "learning_rate": 6.986350701550206e-05,
      "loss": 0.8427,
      "step": 173670
    },
    {
      "epoch": 3.0492108358643937,
      "grad_norm": 0.04816850471345703,
      "learning_rate": 6.985292907324663e-05,
      "loss": 0.8431,
      "step": 173680
    },
    {
      "epoch": 3.049386400744395,
      "grad_norm": 0.05013971255705688,
      "learning_rate": 6.984235152563053e-05,
      "loss": 0.8496,
      "step": 173690
    },
    {
      "epoch": 3.0495619656243966,
      "grad_norm": 0.05964723107243261,
      "learning_rate": 6.983177437278783e-05,
      "loss": 0.8391,
      "step": 173700
    },
    {
      "epoch": 3.049737530504398,
      "grad_norm": 0.0578709966732916,
      "learning_rate": 6.982119761485254e-05,
      "loss": 0.8475,
      "step": 173710
    },
    {
      "epoch": 3.049913095384399,
      "grad_norm": 0.04942109564949152,
      "learning_rate": 6.981062125195863e-05,
      "loss": 0.8387,
      "step": 173720
    },
    {
      "epoch": 3.0500886602644006,
      "grad_norm": 0.06842406417579507,
      "learning_rate": 6.980004528424012e-05,
      "loss": 0.8489,
      "step": 173730
    },
    {
      "epoch": 3.050264225144402,
      "grad_norm": 0.07206835716857488,
      "learning_rate": 6.978946971183101e-05,
      "loss": 0.8478,
      "step": 173740
    },
    {
      "epoch": 3.0504397900244036,
      "grad_norm": 0.05962457851345982,
      "learning_rate": 6.977889453486534e-05,
      "loss": 0.8406,
      "step": 173750
    },
    {
      "epoch": 3.050615354904405,
      "grad_norm": 0.06967525239292123,
      "learning_rate": 6.976831975347698e-05,
      "loss": 0.8376,
      "step": 173760
    },
    {
      "epoch": 3.0507909197844065,
      "grad_norm": 0.043820194536863664,
      "learning_rate": 6.975774536780005e-05,
      "loss": 0.8429,
      "step": 173770
    },
    {
      "epoch": 3.0509664846644076,
      "grad_norm": 0.05341571035611082,
      "learning_rate": 6.974717137796849e-05,
      "loss": 0.841,
      "step": 173780
    },
    {
      "epoch": 3.051142049544409,
      "grad_norm": 0.04986556389563427,
      "learning_rate": 6.973659778411622e-05,
      "loss": 0.8492,
      "step": 173790
    },
    {
      "epoch": 3.0513176144244105,
      "grad_norm": 0.05773644246394325,
      "learning_rate": 6.972602458637728e-05,
      "loss": 0.8453,
      "step": 173800
    },
    {
      "epoch": 3.051493179304412,
      "grad_norm": 0.04607355626680451,
      "learning_rate": 6.971545178488559e-05,
      "loss": 0.8409,
      "step": 173810
    },
    {
      "epoch": 3.0516687441844135,
      "grad_norm": 0.051457176268794036,
      "learning_rate": 6.970487937977513e-05,
      "loss": 0.8462,
      "step": 173820
    },
    {
      "epoch": 3.051844309064415,
      "grad_norm": 0.06536731029414364,
      "learning_rate": 6.96943073711798e-05,
      "loss": 0.8377,
      "step": 173830
    },
    {
      "epoch": 3.052019873944416,
      "grad_norm": 0.05826615137146788,
      "learning_rate": 6.968373575923363e-05,
      "loss": 0.8364,
      "step": 173840
    },
    {
      "epoch": 3.0521954388244175,
      "grad_norm": 0.07065246489473959,
      "learning_rate": 6.967316454407053e-05,
      "loss": 0.8443,
      "step": 173850
    },
    {
      "epoch": 3.052371003704419,
      "grad_norm": 0.04763790920865287,
      "learning_rate": 6.966259372582443e-05,
      "loss": 0.8308,
      "step": 173860
    },
    {
      "epoch": 3.0525465685844204,
      "grad_norm": 0.04769109975511524,
      "learning_rate": 6.965202330462928e-05,
      "loss": 0.8426,
      "step": 173870
    },
    {
      "epoch": 3.052722133464422,
      "grad_norm": 0.048640788517954024,
      "learning_rate": 6.964145328061902e-05,
      "loss": 0.8399,
      "step": 173880
    },
    {
      "epoch": 3.0528976983444234,
      "grad_norm": 0.0498953073140878,
      "learning_rate": 6.963088365392755e-05,
      "loss": 0.849,
      "step": 173890
    },
    {
      "epoch": 3.0530732632244244,
      "grad_norm": 0.04868965831044402,
      "learning_rate": 6.962031442468877e-05,
      "loss": 0.8413,
      "step": 173900
    },
    {
      "epoch": 3.053248828104426,
      "grad_norm": 0.05447753779115411,
      "learning_rate": 6.960974559303663e-05,
      "loss": 0.839,
      "step": 173910
    },
    {
      "epoch": 3.0534243929844274,
      "grad_norm": 0.05274451112376426,
      "learning_rate": 6.959917715910505e-05,
      "loss": 0.8394,
      "step": 173920
    },
    {
      "epoch": 3.053599957864429,
      "grad_norm": 0.07299483566835412,
      "learning_rate": 6.958860912302788e-05,
      "loss": 0.8464,
      "step": 173930
    },
    {
      "epoch": 3.0537755227444303,
      "grad_norm": 0.04719657125196367,
      "learning_rate": 6.957804148493908e-05,
      "loss": 0.843,
      "step": 173940
    },
    {
      "epoch": 3.053951087624432,
      "grad_norm": 0.05102452316362669,
      "learning_rate": 6.956747424497251e-05,
      "loss": 0.8512,
      "step": 173950
    },
    {
      "epoch": 3.054126652504433,
      "grad_norm": 0.051632547116908,
      "learning_rate": 6.955690740326205e-05,
      "loss": 0.8398,
      "step": 173960
    },
    {
      "epoch": 3.0543022173844343,
      "grad_norm": 0.05136740942668013,
      "learning_rate": 6.954634095994162e-05,
      "loss": 0.8486,
      "step": 173970
    },
    {
      "epoch": 3.054477782264436,
      "grad_norm": 0.06283400321867234,
      "learning_rate": 6.953577491514508e-05,
      "loss": 0.8408,
      "step": 173980
    },
    {
      "epoch": 3.0546533471444373,
      "grad_norm": 0.05829613338066152,
      "learning_rate": 6.952520926900631e-05,
      "loss": 0.8463,
      "step": 173990
    },
    {
      "epoch": 3.0548289120244387,
      "grad_norm": 0.04128316142372947,
      "learning_rate": 6.951464402165911e-05,
      "loss": 0.8356,
      "step": 174000
    },
    {
      "epoch": 3.05500447690444,
      "grad_norm": 0.046411070129892984,
      "learning_rate": 6.950407917323746e-05,
      "loss": 0.8424,
      "step": 174010
    },
    {
      "epoch": 3.0551800417844412,
      "grad_norm": 0.06371364872535651,
      "learning_rate": 6.949351472387516e-05,
      "loss": 0.8427,
      "step": 174020
    },
    {
      "epoch": 3.0553556066644427,
      "grad_norm": 0.05325020547797375,
      "learning_rate": 6.948295067370604e-05,
      "loss": 0.8481,
      "step": 174030
    },
    {
      "epoch": 3.055531171544444,
      "grad_norm": 0.06677676058702173,
      "learning_rate": 6.947238702286401e-05,
      "loss": 0.8423,
      "step": 174040
    },
    {
      "epoch": 3.0557067364244457,
      "grad_norm": 0.04529182246034823,
      "learning_rate": 6.946182377148285e-05,
      "loss": 0.8424,
      "step": 174050
    },
    {
      "epoch": 3.055882301304447,
      "grad_norm": 0.06733486078696131,
      "learning_rate": 6.945126091969645e-05,
      "loss": 0.8432,
      "step": 174060
    },
    {
      "epoch": 3.0560578661844486,
      "grad_norm": 0.05619954467889164,
      "learning_rate": 6.94406984676386e-05,
      "loss": 0.845,
      "step": 174070
    },
    {
      "epoch": 3.0562334310644497,
      "grad_norm": 0.05944784101518529,
      "learning_rate": 6.943013641544317e-05,
      "loss": 0.8438,
      "step": 174080
    },
    {
      "epoch": 3.056408995944451,
      "grad_norm": 0.06389497997601021,
      "learning_rate": 6.941957476324395e-05,
      "loss": 0.8371,
      "step": 174090
    },
    {
      "epoch": 3.0565845608244526,
      "grad_norm": 0.07560146476550507,
      "learning_rate": 6.940901351117475e-05,
      "loss": 0.8346,
      "step": 174100
    },
    {
      "epoch": 3.056760125704454,
      "grad_norm": 0.04792228435811361,
      "learning_rate": 6.939845265936944e-05,
      "loss": 0.8285,
      "step": 174110
    },
    {
      "epoch": 3.0569356905844556,
      "grad_norm": 0.056535474201700416,
      "learning_rate": 6.938789220796177e-05,
      "loss": 0.8362,
      "step": 174120
    },
    {
      "epoch": 3.057111255464457,
      "grad_norm": 0.046200262614949875,
      "learning_rate": 6.937733215708555e-05,
      "loss": 0.8455,
      "step": 174130
    },
    {
      "epoch": 3.057286820344458,
      "grad_norm": 0.0511625977841541,
      "learning_rate": 6.936677250687462e-05,
      "loss": 0.8383,
      "step": 174140
    },
    {
      "epoch": 3.0574623852244596,
      "grad_norm": 0.05199580670713242,
      "learning_rate": 6.935621325746275e-05,
      "loss": 0.851,
      "step": 174150
    },
    {
      "epoch": 3.057637950104461,
      "grad_norm": 0.06118858024287419,
      "learning_rate": 6.93456544089837e-05,
      "loss": 0.8403,
      "step": 174160
    },
    {
      "epoch": 3.0578135149844625,
      "grad_norm": 0.10993807774464165,
      "learning_rate": 6.933509596157125e-05,
      "loss": 0.8367,
      "step": 174170
    },
    {
      "epoch": 3.057989079864464,
      "grad_norm": 0.06633375408116819,
      "learning_rate": 6.932453791535923e-05,
      "loss": 0.8373,
      "step": 174180
    },
    {
      "epoch": 3.0581646447444655,
      "grad_norm": 0.04595142975735656,
      "learning_rate": 6.931398027048141e-05,
      "loss": 0.8426,
      "step": 174190
    },
    {
      "epoch": 3.0583402096244665,
      "grad_norm": 0.05399029337277553,
      "learning_rate": 6.93034230270715e-05,
      "loss": 0.8406,
      "step": 174200
    },
    {
      "epoch": 3.058515774504468,
      "grad_norm": 0.05043894794795184,
      "learning_rate": 6.929286618526331e-05,
      "loss": 0.8438,
      "step": 174210
    },
    {
      "epoch": 3.0586913393844695,
      "grad_norm": 0.05178264990659348,
      "learning_rate": 6.92823097451906e-05,
      "loss": 0.8431,
      "step": 174220
    },
    {
      "epoch": 3.058866904264471,
      "grad_norm": 0.0794970105199614,
      "learning_rate": 6.927175370698709e-05,
      "loss": 0.8484,
      "step": 174230
    },
    {
      "epoch": 3.0590424691444724,
      "grad_norm": 0.0531910872428408,
      "learning_rate": 6.926119807078651e-05,
      "loss": 0.851,
      "step": 174240
    },
    {
      "epoch": 3.059218034024474,
      "grad_norm": 0.05309814249247087,
      "learning_rate": 6.925064283672268e-05,
      "loss": 0.8401,
      "step": 174250
    },
    {
      "epoch": 3.059393598904475,
      "grad_norm": 0.057134115953952826,
      "learning_rate": 6.924008800492927e-05,
      "loss": 0.8343,
      "step": 174260
    },
    {
      "epoch": 3.0595691637844764,
      "grad_norm": 0.05321522419768154,
      "learning_rate": 6.922953357553999e-05,
      "loss": 0.8331,
      "step": 174270
    },
    {
      "epoch": 3.059744728664478,
      "grad_norm": 0.048069267737691004,
      "learning_rate": 6.921897954868866e-05,
      "loss": 0.8378,
      "step": 174280
    },
    {
      "epoch": 3.0599202935444794,
      "grad_norm": 0.04061403360167974,
      "learning_rate": 6.920842592450895e-05,
      "loss": 0.837,
      "step": 174290
    },
    {
      "epoch": 3.060095858424481,
      "grad_norm": 0.05095493079812211,
      "learning_rate": 6.919787270313455e-05,
      "loss": 0.8496,
      "step": 174300
    },
    {
      "epoch": 3.0602714233044823,
      "grad_norm": 0.0439606391573628,
      "learning_rate": 6.918731988469923e-05,
      "loss": 0.8464,
      "step": 174310
    },
    {
      "epoch": 3.0604469881844834,
      "grad_norm": 0.051748742201727246,
      "learning_rate": 6.917676746933666e-05,
      "loss": 0.8479,
      "step": 174320
    },
    {
      "epoch": 3.060622553064485,
      "grad_norm": 0.055651721237335514,
      "learning_rate": 6.916621545718053e-05,
      "loss": 0.8468,
      "step": 174330
    },
    {
      "epoch": 3.0607981179444863,
      "grad_norm": 0.048377611233925644,
      "learning_rate": 6.915566384836451e-05,
      "loss": 0.84,
      "step": 174340
    },
    {
      "epoch": 3.060973682824488,
      "grad_norm": 0.053292743521563685,
      "learning_rate": 6.91451126430224e-05,
      "loss": 0.8481,
      "step": 174350
    },
    {
      "epoch": 3.0611492477044893,
      "grad_norm": 0.058338226179061564,
      "learning_rate": 6.913456184128779e-05,
      "loss": 0.8391,
      "step": 174360
    },
    {
      "epoch": 3.0613248125844907,
      "grad_norm": 0.04803313050527189,
      "learning_rate": 6.912401144329438e-05,
      "loss": 0.8412,
      "step": 174370
    },
    {
      "epoch": 3.061500377464492,
      "grad_norm": 0.04499671751068121,
      "learning_rate": 6.911346144917587e-05,
      "loss": 0.8522,
      "step": 174380
    },
    {
      "epoch": 3.0616759423444933,
      "grad_norm": 0.048902781298086596,
      "learning_rate": 6.910291185906592e-05,
      "loss": 0.8452,
      "step": 174390
    },
    {
      "epoch": 3.0618515072244947,
      "grad_norm": 0.04797037196910991,
      "learning_rate": 6.909236267309819e-05,
      "loss": 0.8425,
      "step": 174400
    },
    {
      "epoch": 3.062027072104496,
      "grad_norm": 0.05688451346478783,
      "learning_rate": 6.908181389140631e-05,
      "loss": 0.8465,
      "step": 174410
    },
    {
      "epoch": 3.0622026369844977,
      "grad_norm": 0.0481924336261629,
      "learning_rate": 6.9071265514124e-05,
      "loss": 0.8479,
      "step": 174420
    },
    {
      "epoch": 3.062378201864499,
      "grad_norm": 0.05666547797694573,
      "learning_rate": 6.906071754138485e-05,
      "loss": 0.8338,
      "step": 174430
    },
    {
      "epoch": 3.0625537667445006,
      "grad_norm": 0.06354938452845219,
      "learning_rate": 6.905016997332252e-05,
      "loss": 0.843,
      "step": 174440
    },
    {
      "epoch": 3.0627293316245017,
      "grad_norm": 0.05491407511581481,
      "learning_rate": 6.903962281007069e-05,
      "loss": 0.8468,
      "step": 174450
    },
    {
      "epoch": 3.062904896504503,
      "grad_norm": 0.04309427314161035,
      "learning_rate": 6.902907605176296e-05,
      "loss": 0.8432,
      "step": 174460
    },
    {
      "epoch": 3.0630804613845046,
      "grad_norm": 0.04657696617423997,
      "learning_rate": 6.901852969853294e-05,
      "loss": 0.84,
      "step": 174470
    },
    {
      "epoch": 3.063256026264506,
      "grad_norm": 0.04825358296349173,
      "learning_rate": 6.90079837505143e-05,
      "loss": 0.8478,
      "step": 174480
    },
    {
      "epoch": 3.0634315911445076,
      "grad_norm": 0.06518439805039182,
      "learning_rate": 6.899743820784065e-05,
      "loss": 0.8393,
      "step": 174490
    },
    {
      "epoch": 3.063607156024509,
      "grad_norm": 0.045205914944833815,
      "learning_rate": 6.898689307064557e-05,
      "loss": 0.8409,
      "step": 174500
    },
    {
      "epoch": 3.06378272090451,
      "grad_norm": 0.05953694022558826,
      "learning_rate": 6.897634833906266e-05,
      "loss": 0.8336,
      "step": 174510
    },
    {
      "epoch": 3.0639582857845116,
      "grad_norm": 0.049127981059010895,
      "learning_rate": 6.896580401322559e-05,
      "loss": 0.845,
      "step": 174520
    },
    {
      "epoch": 3.064133850664513,
      "grad_norm": 0.06665356936078108,
      "learning_rate": 6.895526009326793e-05,
      "loss": 0.8357,
      "step": 174530
    },
    {
      "epoch": 3.0643094155445145,
      "grad_norm": 0.050861410333385995,
      "learning_rate": 6.894471657932325e-05,
      "loss": 0.8454,
      "step": 174540
    },
    {
      "epoch": 3.064484980424516,
      "grad_norm": 0.06374390076441983,
      "learning_rate": 6.893417347152516e-05,
      "loss": 0.8381,
      "step": 174550
    },
    {
      "epoch": 3.0646605453045175,
      "grad_norm": 0.04712434521444203,
      "learning_rate": 6.892363077000725e-05,
      "loss": 0.8421,
      "step": 174560
    },
    {
      "epoch": 3.0648361101845185,
      "grad_norm": 0.0537881767744789,
      "learning_rate": 6.891308847490307e-05,
      "loss": 0.8372,
      "step": 174570
    },
    {
      "epoch": 3.06501167506452,
      "grad_norm": 0.05324810737999309,
      "learning_rate": 6.890254658634621e-05,
      "loss": 0.8483,
      "step": 174580
    },
    {
      "epoch": 3.0651872399445215,
      "grad_norm": 0.061145040453798684,
      "learning_rate": 6.889200510447025e-05,
      "loss": 0.849,
      "step": 174590
    },
    {
      "epoch": 3.065362804824523,
      "grad_norm": 0.05605425793790212,
      "learning_rate": 6.888146402940872e-05,
      "loss": 0.8401,
      "step": 174600
    },
    {
      "epoch": 3.0655383697045244,
      "grad_norm": 0.05553808506131313,
      "learning_rate": 6.887092336129518e-05,
      "loss": 0.8379,
      "step": 174610
    },
    {
      "epoch": 3.065713934584526,
      "grad_norm": 0.05856780217924377,
      "learning_rate": 6.886038310026322e-05,
      "loss": 0.8371,
      "step": 174620
    },
    {
      "epoch": 3.065889499464527,
      "grad_norm": 0.07153262635977008,
      "learning_rate": 6.884984324644637e-05,
      "loss": 0.8367,
      "step": 174630
    },
    {
      "epoch": 3.0660650643445284,
      "grad_norm": 0.05428332975906769,
      "learning_rate": 6.883930379997818e-05,
      "loss": 0.8409,
      "step": 174640
    },
    {
      "epoch": 3.06624062922453,
      "grad_norm": 0.06646132101483684,
      "learning_rate": 6.882876476099215e-05,
      "loss": 0.8282,
      "step": 174650
    },
    {
      "epoch": 3.0664161941045314,
      "grad_norm": 0.05980256004690444,
      "learning_rate": 6.881822612962185e-05,
      "loss": 0.845,
      "step": 174660
    },
    {
      "epoch": 3.066591758984533,
      "grad_norm": 0.07475236818628811,
      "learning_rate": 6.880768790600079e-05,
      "loss": 0.8404,
      "step": 174670
    },
    {
      "epoch": 3.0667673238645343,
      "grad_norm": 0.058345500722714494,
      "learning_rate": 6.879715009026246e-05,
      "loss": 0.8471,
      "step": 174680
    },
    {
      "epoch": 3.0669428887445354,
      "grad_norm": 0.05122266154897929,
      "learning_rate": 6.878661268254045e-05,
      "loss": 0.8395,
      "step": 174690
    },
    {
      "epoch": 3.067118453624537,
      "grad_norm": 0.051580443367627876,
      "learning_rate": 6.877607568296823e-05,
      "loss": 0.8423,
      "step": 174700
    },
    {
      "epoch": 3.0672940185045383,
      "grad_norm": 0.06784917626818288,
      "learning_rate": 6.876553909167929e-05,
      "loss": 0.8457,
      "step": 174710
    },
    {
      "epoch": 3.06746958338454,
      "grad_norm": 0.05312945720232359,
      "learning_rate": 6.875500290880716e-05,
      "loss": 0.844,
      "step": 174720
    },
    {
      "epoch": 3.0676451482645413,
      "grad_norm": 0.056908426585540135,
      "learning_rate": 6.874446713448533e-05,
      "loss": 0.8415,
      "step": 174730
    },
    {
      "epoch": 3.0678207131445427,
      "grad_norm": 0.0571028198515609,
      "learning_rate": 6.873393176884728e-05,
      "loss": 0.8459,
      "step": 174740
    },
    {
      "epoch": 3.0679962780245438,
      "grad_norm": 0.05604837147419114,
      "learning_rate": 6.872339681202647e-05,
      "loss": 0.8366,
      "step": 174750
    },
    {
      "epoch": 3.0681718429045453,
      "grad_norm": 0.06581767315761806,
      "learning_rate": 6.871286226415644e-05,
      "loss": 0.8422,
      "step": 174760
    },
    {
      "epoch": 3.0683474077845467,
      "grad_norm": 0.07112452090976763,
      "learning_rate": 6.870232812537062e-05,
      "loss": 0.84,
      "step": 174770
    },
    {
      "epoch": 3.068522972664548,
      "grad_norm": 0.05459984864656078,
      "learning_rate": 6.869179439580247e-05,
      "loss": 0.8373,
      "step": 174780
    },
    {
      "epoch": 3.0686985375445497,
      "grad_norm": 0.05458543856797954,
      "learning_rate": 6.868126107558551e-05,
      "loss": 0.84,
      "step": 174790
    },
    {
      "epoch": 3.068874102424551,
      "grad_norm": 0.0507677971273891,
      "learning_rate": 6.867072816485315e-05,
      "loss": 0.8364,
      "step": 174800
    },
    {
      "epoch": 3.069049667304552,
      "grad_norm": 0.07528736848452743,
      "learning_rate": 6.866019566373888e-05,
      "loss": 0.8418,
      "step": 174810
    },
    {
      "epoch": 3.0692252321845537,
      "grad_norm": 0.05480526500026667,
      "learning_rate": 6.86496635723761e-05,
      "loss": 0.8482,
      "step": 174820
    },
    {
      "epoch": 3.069400797064555,
      "grad_norm": 0.07761961636197628,
      "learning_rate": 6.863913189089831e-05,
      "loss": 0.8401,
      "step": 174830
    },
    {
      "epoch": 3.0695763619445566,
      "grad_norm": 0.06851076582836027,
      "learning_rate": 6.862860061943892e-05,
      "loss": 0.8434,
      "step": 174840
    },
    {
      "epoch": 3.069751926824558,
      "grad_norm": 0.04836704169031059,
      "learning_rate": 6.861806975813132e-05,
      "loss": 0.8453,
      "step": 174850
    },
    {
      "epoch": 3.0699274917045596,
      "grad_norm": 0.04928533420809973,
      "learning_rate": 6.860753930710902e-05,
      "loss": 0.8396,
      "step": 174860
    },
    {
      "epoch": 3.0701030565845606,
      "grad_norm": 0.04876020981083669,
      "learning_rate": 6.859700926650542e-05,
      "loss": 0.8458,
      "step": 174870
    },
    {
      "epoch": 3.070278621464562,
      "grad_norm": 0.05198083813925165,
      "learning_rate": 6.858647963645388e-05,
      "loss": 0.8397,
      "step": 174880
    },
    {
      "epoch": 3.0704541863445636,
      "grad_norm": 0.05120276262284467,
      "learning_rate": 6.857595041708789e-05,
      "loss": 0.844,
      "step": 174890
    },
    {
      "epoch": 3.070629751224565,
      "grad_norm": 0.06470388183285733,
      "learning_rate": 6.856542160854081e-05,
      "loss": 0.8397,
      "step": 174900
    },
    {
      "epoch": 3.0708053161045665,
      "grad_norm": 0.049359980020908516,
      "learning_rate": 6.855489321094607e-05,
      "loss": 0.8451,
      "step": 174910
    },
    {
      "epoch": 3.070980880984568,
      "grad_norm": 0.0635577167935494,
      "learning_rate": 6.854436522443701e-05,
      "loss": 0.8358,
      "step": 174920
    },
    {
      "epoch": 3.071156445864569,
      "grad_norm": 0.07775946223093315,
      "learning_rate": 6.853383764914711e-05,
      "loss": 0.8447,
      "step": 174930
    },
    {
      "epoch": 3.0713320107445705,
      "grad_norm": 0.047803275137090706,
      "learning_rate": 6.852331048520971e-05,
      "loss": 0.8501,
      "step": 174940
    },
    {
      "epoch": 3.071507575624572,
      "grad_norm": 0.060496403907323276,
      "learning_rate": 6.851278373275812e-05,
      "loss": 0.8551,
      "step": 174950
    },
    {
      "epoch": 3.0716831405045735,
      "grad_norm": 0.05776512595723172,
      "learning_rate": 6.850225739192584e-05,
      "loss": 0.8429,
      "step": 174960
    },
    {
      "epoch": 3.071858705384575,
      "grad_norm": 0.04623281312838503,
      "learning_rate": 6.849173146284619e-05,
      "loss": 0.8448,
      "step": 174970
    },
    {
      "epoch": 3.0720342702645764,
      "grad_norm": 0.0515425328795,
      "learning_rate": 6.848120594565252e-05,
      "loss": 0.8382,
      "step": 174980
    },
    {
      "epoch": 3.0722098351445775,
      "grad_norm": 0.057203149662560474,
      "learning_rate": 6.847068084047819e-05,
      "loss": 0.8429,
      "step": 174990
    },
    {
      "epoch": 3.072385400024579,
      "grad_norm": 0.04053235461373034,
      "learning_rate": 6.84601561474566e-05,
      "loss": 0.8411,
      "step": 175000
    },
    {
      "epoch": 3.0725609649045804,
      "grad_norm": 0.04686865700704318,
      "learning_rate": 6.844963186672105e-05,
      "loss": 0.8485,
      "step": 175010
    },
    {
      "epoch": 3.072736529784582,
      "grad_norm": 0.06295647949797388,
      "learning_rate": 6.843910799840488e-05,
      "loss": 0.8396,
      "step": 175020
    },
    {
      "epoch": 3.0729120946645834,
      "grad_norm": 0.05486772996822058,
      "learning_rate": 6.842858454264148e-05,
      "loss": 0.8437,
      "step": 175030
    },
    {
      "epoch": 3.073087659544585,
      "grad_norm": 0.056094204974172605,
      "learning_rate": 6.841806149956417e-05,
      "loss": 0.8404,
      "step": 175040
    },
    {
      "epoch": 3.0732632244245863,
      "grad_norm": 0.06139133163686299,
      "learning_rate": 6.840753886930624e-05,
      "loss": 0.8446,
      "step": 175050
    },
    {
      "epoch": 3.0734387893045874,
      "grad_norm": 0.050802315207763675,
      "learning_rate": 6.839701665200104e-05,
      "loss": 0.843,
      "step": 175060
    },
    {
      "epoch": 3.073614354184589,
      "grad_norm": 0.04625580713468218,
      "learning_rate": 6.838649484778191e-05,
      "loss": 0.8395,
      "step": 175070
    },
    {
      "epoch": 3.0737899190645903,
      "grad_norm": 0.059882726888172894,
      "learning_rate": 6.837597345678214e-05,
      "loss": 0.8414,
      "step": 175080
    },
    {
      "epoch": 3.073965483944592,
      "grad_norm": 0.06155571828271598,
      "learning_rate": 6.836545247913501e-05,
      "loss": 0.8513,
      "step": 175090
    },
    {
      "epoch": 3.0741410488245933,
      "grad_norm": 0.08269285448885104,
      "learning_rate": 6.835493191497386e-05,
      "loss": 0.8458,
      "step": 175100
    },
    {
      "epoch": 3.0743166137045947,
      "grad_norm": 0.04657123693172788,
      "learning_rate": 6.834441176443201e-05,
      "loss": 0.8378,
      "step": 175110
    },
    {
      "epoch": 3.074492178584596,
      "grad_norm": 0.0480059044932817,
      "learning_rate": 6.833389202764265e-05,
      "loss": 0.8378,
      "step": 175120
    },
    {
      "epoch": 3.0746677434645973,
      "grad_norm": 0.0643675751427009,
      "learning_rate": 6.83233727047392e-05,
      "loss": 0.8447,
      "step": 175130
    },
    {
      "epoch": 3.0748433083445987,
      "grad_norm": 0.0694262062027441,
      "learning_rate": 6.831285379585486e-05,
      "loss": 0.8355,
      "step": 175140
    },
    {
      "epoch": 3.0750188732246,
      "grad_norm": 0.07284552273729417,
      "learning_rate": 6.830233530112294e-05,
      "loss": 0.838,
      "step": 175150
    },
    {
      "epoch": 3.0751944381046017,
      "grad_norm": 0.05345974779365334,
      "learning_rate": 6.829181722067668e-05,
      "loss": 0.8409,
      "step": 175160
    },
    {
      "epoch": 3.075370002984603,
      "grad_norm": 0.0511138065082372,
      "learning_rate": 6.828129955464937e-05,
      "loss": 0.8457,
      "step": 175170
    },
    {
      "epoch": 3.075545567864604,
      "grad_norm": 0.04477027606524673,
      "learning_rate": 6.827078230317429e-05,
      "loss": 0.8459,
      "step": 175180
    },
    {
      "epoch": 3.0757211327446057,
      "grad_norm": 0.05386443513596692,
      "learning_rate": 6.82602654663846e-05,
      "loss": 0.8432,
      "step": 175190
    },
    {
      "epoch": 3.075896697624607,
      "grad_norm": 0.06114659397655289,
      "learning_rate": 6.82497490444137e-05,
      "loss": 0.8397,
      "step": 175200
    },
    {
      "epoch": 3.0760722625046086,
      "grad_norm": 0.050591329980027265,
      "learning_rate": 6.823923303739471e-05,
      "loss": 0.8481,
      "step": 175210
    },
    {
      "epoch": 3.07624782738461,
      "grad_norm": 0.03604864912326447,
      "learning_rate": 6.822871744546093e-05,
      "loss": 0.8475,
      "step": 175220
    },
    {
      "epoch": 3.0764233922646116,
      "grad_norm": 0.043813599783656364,
      "learning_rate": 6.821820226874559e-05,
      "loss": 0.8438,
      "step": 175230
    },
    {
      "epoch": 3.0765989571446126,
      "grad_norm": 0.046396225454713835,
      "learning_rate": 6.82076875073819e-05,
      "loss": 0.8415,
      "step": 175240
    },
    {
      "epoch": 3.076774522024614,
      "grad_norm": 0.056352733081177483,
      "learning_rate": 6.819717316150312e-05,
      "loss": 0.8489,
      "step": 175250
    },
    {
      "epoch": 3.0769500869046156,
      "grad_norm": 0.07139400274647001,
      "learning_rate": 6.81866592312424e-05,
      "loss": 0.8439,
      "step": 175260
    },
    {
      "epoch": 3.077125651784617,
      "grad_norm": 0.060044101621152965,
      "learning_rate": 6.817614571673302e-05,
      "loss": 0.84,
      "step": 175270
    },
    {
      "epoch": 3.0773012166646185,
      "grad_norm": 0.04775424328424617,
      "learning_rate": 6.816563261810816e-05,
      "loss": 0.8346,
      "step": 175280
    },
    {
      "epoch": 3.07747678154462,
      "grad_norm": 0.05733613363852002,
      "learning_rate": 6.8155119935501e-05,
      "loss": 0.8458,
      "step": 175290
    },
    {
      "epoch": 3.077652346424621,
      "grad_norm": 0.0515582963293572,
      "learning_rate": 6.81446076690448e-05,
      "loss": 0.8299,
      "step": 175300
    },
    {
      "epoch": 3.0778279113046225,
      "grad_norm": 0.05405448904137167,
      "learning_rate": 6.813409581887273e-05,
      "loss": 0.8391,
      "step": 175310
    },
    {
      "epoch": 3.078003476184624,
      "grad_norm": 0.06706674988943122,
      "learning_rate": 6.812358438511794e-05,
      "loss": 0.8336,
      "step": 175320
    },
    {
      "epoch": 3.0781790410646255,
      "grad_norm": 0.054050909432187884,
      "learning_rate": 6.811307336791364e-05,
      "loss": 0.8462,
      "step": 175330
    },
    {
      "epoch": 3.078354605944627,
      "grad_norm": 0.05787285056921521,
      "learning_rate": 6.8102562767393e-05,
      "loss": 0.8367,
      "step": 175340
    },
    {
      "epoch": 3.0785301708246284,
      "grad_norm": 0.054090252481081125,
      "learning_rate": 6.809205258368922e-05,
      "loss": 0.8438,
      "step": 175350
    },
    {
      "epoch": 3.0787057357046295,
      "grad_norm": 0.06158706204185987,
      "learning_rate": 6.808154281693539e-05,
      "loss": 0.8429,
      "step": 175360
    },
    {
      "epoch": 3.078881300584631,
      "grad_norm": 0.06846877903509613,
      "learning_rate": 6.807103346726476e-05,
      "loss": 0.8419,
      "step": 175370
    },
    {
      "epoch": 3.0790568654646324,
      "grad_norm": 0.05421220977384527,
      "learning_rate": 6.806052453481044e-05,
      "loss": 0.8482,
      "step": 175380
    },
    {
      "epoch": 3.079232430344634,
      "grad_norm": 0.05748259945135824,
      "learning_rate": 6.805001601970557e-05,
      "loss": 0.8425,
      "step": 175390
    },
    {
      "epoch": 3.0794079952246354,
      "grad_norm": 0.05012459230974726,
      "learning_rate": 6.803950792208333e-05,
      "loss": 0.8482,
      "step": 175400
    },
    {
      "epoch": 3.079583560104637,
      "grad_norm": 0.05279473617199634,
      "learning_rate": 6.802900024207686e-05,
      "loss": 0.8405,
      "step": 175410
    },
    {
      "epoch": 3.079759124984638,
      "grad_norm": 0.05834227635235926,
      "learning_rate": 6.801849297981925e-05,
      "loss": 0.8338,
      "step": 175420
    },
    {
      "epoch": 3.0799346898646394,
      "grad_norm": 0.06710985352832502,
      "learning_rate": 6.800798613544365e-05,
      "loss": 0.8397,
      "step": 175430
    },
    {
      "epoch": 3.080110254744641,
      "grad_norm": 0.07236476195349621,
      "learning_rate": 6.79974797090832e-05,
      "loss": 0.8382,
      "step": 175440
    },
    {
      "epoch": 3.0802858196246423,
      "grad_norm": 0.04201045618977495,
      "learning_rate": 6.798697370087101e-05,
      "loss": 0.8394,
      "step": 175450
    },
    {
      "epoch": 3.080461384504644,
      "grad_norm": 0.060240009565047326,
      "learning_rate": 6.797646811094015e-05,
      "loss": 0.8422,
      "step": 175460
    },
    {
      "epoch": 3.0806369493846453,
      "grad_norm": 0.07941543292607534,
      "learning_rate": 6.796596293942381e-05,
      "loss": 0.8387,
      "step": 175470
    },
    {
      "epoch": 3.0808125142646463,
      "grad_norm": 0.05312550477204751,
      "learning_rate": 6.795545818645505e-05,
      "loss": 0.846,
      "step": 175480
    },
    {
      "epoch": 3.080988079144648,
      "grad_norm": 0.04254467151011606,
      "learning_rate": 6.794495385216697e-05,
      "loss": 0.8384,
      "step": 175490
    },
    {
      "epoch": 3.0811636440246493,
      "grad_norm": 0.060770883838173036,
      "learning_rate": 6.793444993669263e-05,
      "loss": 0.8476,
      "step": 175500
    },
    {
      "epoch": 3.0813392089046507,
      "grad_norm": 0.044073989113208034,
      "learning_rate": 6.792394644016518e-05,
      "loss": 0.8439,
      "step": 175510
    },
    {
      "epoch": 3.081514773784652,
      "grad_norm": 0.04559024828243668,
      "learning_rate": 6.791344336271767e-05,
      "loss": 0.8431,
      "step": 175520
    },
    {
      "epoch": 3.0816903386646537,
      "grad_norm": 0.041661859889022344,
      "learning_rate": 6.790294070448312e-05,
      "loss": 0.8383,
      "step": 175530
    },
    {
      "epoch": 3.0818659035446547,
      "grad_norm": 0.04782584135692878,
      "learning_rate": 6.78924384655947e-05,
      "loss": 0.8319,
      "step": 175540
    },
    {
      "epoch": 3.082041468424656,
      "grad_norm": 0.05157099116040579,
      "learning_rate": 6.788193664618543e-05,
      "loss": 0.8402,
      "step": 175550
    },
    {
      "epoch": 3.0822170333046577,
      "grad_norm": 0.05156657037128815,
      "learning_rate": 6.787143524638834e-05,
      "loss": 0.8466,
      "step": 175560
    },
    {
      "epoch": 3.082392598184659,
      "grad_norm": 0.09123495315940242,
      "learning_rate": 6.786093426633655e-05,
      "loss": 0.8413,
      "step": 175570
    },
    {
      "epoch": 3.0825681630646606,
      "grad_norm": 0.06611216396004564,
      "learning_rate": 6.785043370616306e-05,
      "loss": 0.8393,
      "step": 175580
    },
    {
      "epoch": 3.082743727944662,
      "grad_norm": 0.045253924977061835,
      "learning_rate": 6.783993356600095e-05,
      "loss": 0.8354,
      "step": 175590
    },
    {
      "epoch": 3.082919292824663,
      "grad_norm": 0.04904783246215093,
      "learning_rate": 6.78294338459832e-05,
      "loss": 0.8432,
      "step": 175600
    },
    {
      "epoch": 3.0830948577046646,
      "grad_norm": 0.04682271394293054,
      "learning_rate": 6.781893454624289e-05,
      "loss": 0.8522,
      "step": 175610
    },
    {
      "epoch": 3.083270422584666,
      "grad_norm": 0.05571550586570587,
      "learning_rate": 6.780843566691305e-05,
      "loss": 0.8385,
      "step": 175620
    },
    {
      "epoch": 3.0834459874646676,
      "grad_norm": 0.06299299756136877,
      "learning_rate": 6.779793720812666e-05,
      "loss": 0.838,
      "step": 175630
    },
    {
      "epoch": 3.083621552344669,
      "grad_norm": 0.0539389625226837,
      "learning_rate": 6.778743917001679e-05,
      "loss": 0.8381,
      "step": 175640
    },
    {
      "epoch": 3.0837971172246705,
      "grad_norm": 0.04583732730772104,
      "learning_rate": 6.777694155271645e-05,
      "loss": 0.8364,
      "step": 175650
    },
    {
      "epoch": 3.0839726821046716,
      "grad_norm": 0.050943267480609795,
      "learning_rate": 6.776644435635862e-05,
      "loss": 0.8441,
      "step": 175660
    },
    {
      "epoch": 3.084148246984673,
      "grad_norm": 0.05646451104510366,
      "learning_rate": 6.77559475810763e-05,
      "loss": 0.8456,
      "step": 175670
    },
    {
      "epoch": 3.0843238118646745,
      "grad_norm": 0.07566261431552623,
      "learning_rate": 6.774545122700248e-05,
      "loss": 0.8493,
      "step": 175680
    },
    {
      "epoch": 3.084499376744676,
      "grad_norm": 0.08124948062868495,
      "learning_rate": 6.77349552942702e-05,
      "loss": 0.8429,
      "step": 175690
    },
    {
      "epoch": 3.0846749416246775,
      "grad_norm": 0.043292922569825705,
      "learning_rate": 6.772445978301237e-05,
      "loss": 0.8422,
      "step": 175700
    },
    {
      "epoch": 3.084850506504679,
      "grad_norm": 0.050549397081318684,
      "learning_rate": 6.771396469336204e-05,
      "loss": 0.851,
      "step": 175710
    },
    {
      "epoch": 3.0850260713846804,
      "grad_norm": 0.06559995355768417,
      "learning_rate": 6.770347002545215e-05,
      "loss": 0.8366,
      "step": 175720
    },
    {
      "epoch": 3.0852016362646815,
      "grad_norm": 0.04670164810150193,
      "learning_rate": 6.769297577941567e-05,
      "loss": 0.8434,
      "step": 175730
    },
    {
      "epoch": 3.085377201144683,
      "grad_norm": 0.049239239332879776,
      "learning_rate": 6.768248195538559e-05,
      "loss": 0.847,
      "step": 175740
    },
    {
      "epoch": 3.0855527660246844,
      "grad_norm": 0.04919922779698342,
      "learning_rate": 6.767198855349484e-05,
      "loss": 0.8427,
      "step": 175750
    },
    {
      "epoch": 3.085728330904686,
      "grad_norm": 0.04833090629942577,
      "learning_rate": 6.766149557387638e-05,
      "loss": 0.8364,
      "step": 175760
    },
    {
      "epoch": 3.0859038957846874,
      "grad_norm": 0.08210172953866542,
      "learning_rate": 6.765100301666317e-05,
      "loss": 0.8348,
      "step": 175770
    },
    {
      "epoch": 3.0860794606646884,
      "grad_norm": 0.04054686309990024,
      "learning_rate": 6.764051088198813e-05,
      "loss": 0.841,
      "step": 175780
    },
    {
      "epoch": 3.08625502554469,
      "grad_norm": 0.05298319722869329,
      "learning_rate": 6.763001916998422e-05,
      "loss": 0.8444,
      "step": 175790
    },
    {
      "epoch": 3.0864305904246914,
      "grad_norm": 0.05618117653237754,
      "learning_rate": 6.761952788078432e-05,
      "loss": 0.8374,
      "step": 175800
    },
    {
      "epoch": 3.086606155304693,
      "grad_norm": 0.04603667412880749,
      "learning_rate": 6.760903701452144e-05,
      "loss": 0.8432,
      "step": 175810
    },
    {
      "epoch": 3.0867817201846943,
      "grad_norm": 0.06707158077092362,
      "learning_rate": 6.759854657132847e-05,
      "loss": 0.8457,
      "step": 175820
    },
    {
      "epoch": 3.086957285064696,
      "grad_norm": 0.056767607665002016,
      "learning_rate": 6.758805655133831e-05,
      "loss": 0.8384,
      "step": 175830
    },
    {
      "epoch": 3.0871328499446973,
      "grad_norm": 0.048808244670754085,
      "learning_rate": 6.757756695468386e-05,
      "loss": 0.8335,
      "step": 175840
    },
    {
      "epoch": 3.0873084148246983,
      "grad_norm": 0.05289745362125682,
      "learning_rate": 6.756707778149805e-05,
      "loss": 0.8407,
      "step": 175850
    },
    {
      "epoch": 3.0874839797047,
      "grad_norm": 0.05078149034254458,
      "learning_rate": 6.755658903191378e-05,
      "loss": 0.8449,
      "step": 175860
    },
    {
      "epoch": 3.0876595445847013,
      "grad_norm": 0.06720827218637557,
      "learning_rate": 6.754610070606391e-05,
      "loss": 0.8474,
      "step": 175870
    },
    {
      "epoch": 3.0878351094647027,
      "grad_norm": 0.047477115099590815,
      "learning_rate": 6.753561280408137e-05,
      "loss": 0.8475,
      "step": 175880
    },
    {
      "epoch": 3.088010674344704,
      "grad_norm": 0.08208660857161702,
      "learning_rate": 6.752512532609904e-05,
      "loss": 0.8408,
      "step": 175890
    },
    {
      "epoch": 3.0881862392247057,
      "grad_norm": 0.051133971096909626,
      "learning_rate": 6.751463827224977e-05,
      "loss": 0.8439,
      "step": 175900
    },
    {
      "epoch": 3.0883618041047067,
      "grad_norm": 0.046556948912207075,
      "learning_rate": 6.750415164266646e-05,
      "loss": 0.8461,
      "step": 175910
    },
    {
      "epoch": 3.088537368984708,
      "grad_norm": 0.045587662016547306,
      "learning_rate": 6.749366543748199e-05,
      "loss": 0.837,
      "step": 175920
    },
    {
      "epoch": 3.0887129338647097,
      "grad_norm": 0.08083303810312431,
      "learning_rate": 6.748317965682919e-05,
      "loss": 0.8473,
      "step": 175930
    },
    {
      "epoch": 3.088888498744711,
      "grad_norm": 0.06606754874293368,
      "learning_rate": 6.747269430084091e-05,
      "loss": 0.8404,
      "step": 175940
    },
    {
      "epoch": 3.0890640636247126,
      "grad_norm": 0.06173030206203973,
      "learning_rate": 6.746220936965003e-05,
      "loss": 0.8406,
      "step": 175950
    },
    {
      "epoch": 3.089239628504714,
      "grad_norm": 0.05005264784060539,
      "learning_rate": 6.745172486338939e-05,
      "loss": 0.8388,
      "step": 175960
    },
    {
      "epoch": 3.089415193384715,
      "grad_norm": 0.0703764318352421,
      "learning_rate": 6.744124078219177e-05,
      "loss": 0.8373,
      "step": 175970
    },
    {
      "epoch": 3.0895907582647166,
      "grad_norm": 0.05311478626961473,
      "learning_rate": 6.743075712619011e-05,
      "loss": 0.8466,
      "step": 175980
    },
    {
      "epoch": 3.089766323144718,
      "grad_norm": 0.05151508757425209,
      "learning_rate": 6.74202738955172e-05,
      "loss": 0.835,
      "step": 175990
    },
    {
      "epoch": 3.0899418880247196,
      "grad_norm": 0.06506151750285391,
      "learning_rate": 6.740979109030586e-05,
      "loss": 0.8465,
      "step": 176000
    },
    {
      "epoch": 3.090117452904721,
      "grad_norm": 0.05516643759727267,
      "learning_rate": 6.739930871068889e-05,
      "loss": 0.8513,
      "step": 176010
    },
    {
      "epoch": 3.0902930177847225,
      "grad_norm": 0.06764897001196098,
      "learning_rate": 6.73888267567991e-05,
      "loss": 0.8541,
      "step": 176020
    },
    {
      "epoch": 3.0904685826647236,
      "grad_norm": 0.03943588138619527,
      "learning_rate": 6.737834522876936e-05,
      "loss": 0.8425,
      "step": 176030
    },
    {
      "epoch": 3.090644147544725,
      "grad_norm": 0.05904913230800903,
      "learning_rate": 6.736786412673235e-05,
      "loss": 0.841,
      "step": 176040
    },
    {
      "epoch": 3.0908197124247265,
      "grad_norm": 0.04623364912743804,
      "learning_rate": 6.735738345082101e-05,
      "loss": 0.8322,
      "step": 176050
    },
    {
      "epoch": 3.090995277304728,
      "grad_norm": 0.06328085380103222,
      "learning_rate": 6.734690320116807e-05,
      "loss": 0.8498,
      "step": 176060
    },
    {
      "epoch": 3.0911708421847295,
      "grad_norm": 0.06063478765068309,
      "learning_rate": 6.73364233779063e-05,
      "loss": 0.8342,
      "step": 176070
    },
    {
      "epoch": 3.091346407064731,
      "grad_norm": 0.057848255670455755,
      "learning_rate": 6.73259439811685e-05,
      "loss": 0.8367,
      "step": 176080
    },
    {
      "epoch": 3.091521971944732,
      "grad_norm": 0.052824182087618576,
      "learning_rate": 6.731546501108746e-05,
      "loss": 0.8435,
      "step": 176090
    },
    {
      "epoch": 3.0916975368247335,
      "grad_norm": 0.06899720080919773,
      "learning_rate": 6.730498646779592e-05,
      "loss": 0.8365,
      "step": 176100
    },
    {
      "epoch": 3.091873101704735,
      "grad_norm": 0.07054744576774609,
      "learning_rate": 6.729450835142665e-05,
      "loss": 0.8462,
      "step": 176110
    },
    {
      "epoch": 3.0920486665847364,
      "grad_norm": 0.05298313185937617,
      "learning_rate": 6.728403066211243e-05,
      "loss": 0.8442,
      "step": 176120
    },
    {
      "epoch": 3.092224231464738,
      "grad_norm": 0.04631277822133381,
      "learning_rate": 6.7273553399986e-05,
      "loss": 0.8449,
      "step": 176130
    },
    {
      "epoch": 3.0923997963447394,
      "grad_norm": 0.051131948431171276,
      "learning_rate": 6.72630765651801e-05,
      "loss": 0.8552,
      "step": 176140
    },
    {
      "epoch": 3.0925753612247404,
      "grad_norm": 0.05287713288912453,
      "learning_rate": 6.72526001578275e-05,
      "loss": 0.841,
      "step": 176150
    },
    {
      "epoch": 3.092750926104742,
      "grad_norm": 0.05084416466670568,
      "learning_rate": 6.724212417806093e-05,
      "loss": 0.8452,
      "step": 176160
    },
    {
      "epoch": 3.0929264909847434,
      "grad_norm": 0.058122725910300994,
      "learning_rate": 6.723164862601311e-05,
      "loss": 0.8392,
      "step": 176170
    },
    {
      "epoch": 3.093102055864745,
      "grad_norm": 0.05989885352761236,
      "learning_rate": 6.722117350181675e-05,
      "loss": 0.8459,
      "step": 176180
    },
    {
      "epoch": 3.0932776207447463,
      "grad_norm": 0.05373275796701177,
      "learning_rate": 6.721069880560462e-05,
      "loss": 0.8519,
      "step": 176190
    },
    {
      "epoch": 3.093453185624748,
      "grad_norm": 0.05451270370316998,
      "learning_rate": 6.720022453750942e-05,
      "loss": 0.8441,
      "step": 176200
    },
    {
      "epoch": 3.093628750504749,
      "grad_norm": 0.0451687006867277,
      "learning_rate": 6.718975069766382e-05,
      "loss": 0.8435,
      "step": 176210
    },
    {
      "epoch": 3.0938043153847503,
      "grad_norm": 0.05360337902536341,
      "learning_rate": 6.717927728620057e-05,
      "loss": 0.8389,
      "step": 176220
    },
    {
      "epoch": 3.093979880264752,
      "grad_norm": 0.05076147497846708,
      "learning_rate": 6.716880430325238e-05,
      "loss": 0.8386,
      "step": 176230
    },
    {
      "epoch": 3.0941554451447533,
      "grad_norm": 0.06218577502833038,
      "learning_rate": 6.71583317489519e-05,
      "loss": 0.8442,
      "step": 176240
    },
    {
      "epoch": 3.0943310100247547,
      "grad_norm": 0.05588292932014368,
      "learning_rate": 6.714785962343186e-05,
      "loss": 0.8363,
      "step": 176250
    },
    {
      "epoch": 3.0945065749047562,
      "grad_norm": 0.05465870914154287,
      "learning_rate": 6.713738792682493e-05,
      "loss": 0.8413,
      "step": 176260
    },
    {
      "epoch": 3.0946821397847573,
      "grad_norm": 0.11220465912370961,
      "learning_rate": 6.712691665926379e-05,
      "loss": 0.8324,
      "step": 176270
    },
    {
      "epoch": 3.0948577046647587,
      "grad_norm": 0.05179901890804501,
      "learning_rate": 6.711644582088108e-05,
      "loss": 0.8376,
      "step": 176280
    },
    {
      "epoch": 3.09503326954476,
      "grad_norm": 0.04638387583813963,
      "learning_rate": 6.710597541180951e-05,
      "loss": 0.8363,
      "step": 176290
    },
    {
      "epoch": 3.0952088344247617,
      "grad_norm": 0.0672091943619568,
      "learning_rate": 6.709550543218174e-05,
      "loss": 0.8333,
      "step": 176300
    },
    {
      "epoch": 3.095384399304763,
      "grad_norm": 0.050509011827878175,
      "learning_rate": 6.708503588213037e-05,
      "loss": 0.8483,
      "step": 176310
    },
    {
      "epoch": 3.0955599641847646,
      "grad_norm": 0.055071959741475975,
      "learning_rate": 6.707456676178811e-05,
      "loss": 0.8439,
      "step": 176320
    },
    {
      "epoch": 3.0957355290647657,
      "grad_norm": 0.04676578760943654,
      "learning_rate": 6.706409807128763e-05,
      "loss": 0.846,
      "step": 176330
    },
    {
      "epoch": 3.095911093944767,
      "grad_norm": 0.057354728297267024,
      "learning_rate": 6.705362981076151e-05,
      "loss": 0.8386,
      "step": 176340
    },
    {
      "epoch": 3.0960866588247686,
      "grad_norm": 0.045402510007482424,
      "learning_rate": 6.704316198034239e-05,
      "loss": 0.8342,
      "step": 176350
    },
    {
      "epoch": 3.09626222370477,
      "grad_norm": 0.05535217172348485,
      "learning_rate": 6.703269458016293e-05,
      "loss": 0.8432,
      "step": 176360
    },
    {
      "epoch": 3.0964377885847716,
      "grad_norm": 0.07145952704896984,
      "learning_rate": 6.702222761035574e-05,
      "loss": 0.8448,
      "step": 176370
    },
    {
      "epoch": 3.096613353464773,
      "grad_norm": 0.05350954592173466,
      "learning_rate": 6.70117610710534e-05,
      "loss": 0.8435,
      "step": 176380
    },
    {
      "epoch": 3.096788918344774,
      "grad_norm": 0.051361589057380065,
      "learning_rate": 6.700129496238861e-05,
      "loss": 0.8442,
      "step": 176390
    },
    {
      "epoch": 3.0969644832247756,
      "grad_norm": 0.050063715457771094,
      "learning_rate": 6.699082928449392e-05,
      "loss": 0.8459,
      "step": 176400
    },
    {
      "epoch": 3.097140048104777,
      "grad_norm": 0.04516873235987365,
      "learning_rate": 6.698036403750192e-05,
      "loss": 0.8499,
      "step": 176410
    },
    {
      "epoch": 3.0973156129847785,
      "grad_norm": 0.05906207479194119,
      "learning_rate": 6.696989922154526e-05,
      "loss": 0.8412,
      "step": 176420
    },
    {
      "epoch": 3.09749117786478,
      "grad_norm": 0.04819318209330234,
      "learning_rate": 6.69594348367565e-05,
      "loss": 0.848,
      "step": 176430
    },
    {
      "epoch": 3.0976667427447815,
      "grad_norm": 0.05126633724485611,
      "learning_rate": 6.69489708832682e-05,
      "loss": 0.8397,
      "step": 176440
    },
    {
      "epoch": 3.0978423076247825,
      "grad_norm": 0.06078854695194317,
      "learning_rate": 6.693850736121297e-05,
      "loss": 0.8392,
      "step": 176450
    },
    {
      "epoch": 3.098017872504784,
      "grad_norm": 0.04663654232338929,
      "learning_rate": 6.692804427072339e-05,
      "loss": 0.8397,
      "step": 176460
    },
    {
      "epoch": 3.0981934373847855,
      "grad_norm": 0.05707538363705613,
      "learning_rate": 6.6917581611932e-05,
      "loss": 0.839,
      "step": 176470
    },
    {
      "epoch": 3.098369002264787,
      "grad_norm": 0.05105369212394215,
      "learning_rate": 6.690711938497137e-05,
      "loss": 0.8433,
      "step": 176480
    },
    {
      "epoch": 3.0985445671447884,
      "grad_norm": 0.08019685722200995,
      "learning_rate": 6.68966575899741e-05,
      "loss": 0.8313,
      "step": 176490
    },
    {
      "epoch": 3.09872013202479,
      "grad_norm": 0.04805858596052041,
      "learning_rate": 6.688619622707271e-05,
      "loss": 0.8376,
      "step": 176500
    },
    {
      "epoch": 3.0988956969047914,
      "grad_norm": 0.05790563856657866,
      "learning_rate": 6.687573529639976e-05,
      "loss": 0.8365,
      "step": 176510
    },
    {
      "epoch": 3.0990712617847924,
      "grad_norm": 0.04436462794553314,
      "learning_rate": 6.686527479808777e-05,
      "loss": 0.8422,
      "step": 176520
    },
    {
      "epoch": 3.099246826664794,
      "grad_norm": 0.06487651378632081,
      "learning_rate": 6.68548147322693e-05,
      "loss": 0.8454,
      "step": 176530
    },
    {
      "epoch": 3.0994223915447954,
      "grad_norm": 0.065770713312634,
      "learning_rate": 6.684435509907686e-05,
      "loss": 0.8372,
      "step": 176540
    },
    {
      "epoch": 3.099597956424797,
      "grad_norm": 0.058645581726262254,
      "learning_rate": 6.683389589864297e-05,
      "loss": 0.8366,
      "step": 176550
    },
    {
      "epoch": 3.0997735213047983,
      "grad_norm": 0.06740267347432748,
      "learning_rate": 6.682343713110019e-05,
      "loss": 0.837,
      "step": 176560
    },
    {
      "epoch": 3.0999490861848,
      "grad_norm": 0.050481602196107256,
      "learning_rate": 6.681297879658101e-05,
      "loss": 0.8343,
      "step": 176570
    },
    {
      "epoch": 3.100124651064801,
      "grad_norm": 0.04735066133663418,
      "learning_rate": 6.680252089521794e-05,
      "loss": 0.8389,
      "step": 176580
    },
    {
      "epoch": 3.1003002159448023,
      "grad_norm": 0.056510662029249474,
      "learning_rate": 6.67920634271435e-05,
      "loss": 0.8442,
      "step": 176590
    },
    {
      "epoch": 3.100475780824804,
      "grad_norm": 0.06140064798466904,
      "learning_rate": 6.678160639249016e-05,
      "loss": 0.8424,
      "step": 176600
    },
    {
      "epoch": 3.1006513457048053,
      "grad_norm": 0.05302413778783998,
      "learning_rate": 6.677114979139042e-05,
      "loss": 0.8359,
      "step": 176610
    },
    {
      "epoch": 3.1008269105848068,
      "grad_norm": 0.05032403273116645,
      "learning_rate": 6.676069362397675e-05,
      "loss": 0.8482,
      "step": 176620
    },
    {
      "epoch": 3.1010024754648082,
      "grad_norm": 0.07322200615015086,
      "learning_rate": 6.675023789038169e-05,
      "loss": 0.841,
      "step": 176630
    },
    {
      "epoch": 3.1011780403448093,
      "grad_norm": 0.04620816425499337,
      "learning_rate": 6.673978259073768e-05,
      "loss": 0.8459,
      "step": 176640
    },
    {
      "epoch": 3.1013536052248107,
      "grad_norm": 0.04132884488616638,
      "learning_rate": 6.672932772517715e-05,
      "loss": 0.8461,
      "step": 176650
    },
    {
      "epoch": 3.101529170104812,
      "grad_norm": 0.0506498732779882,
      "learning_rate": 6.671887329383264e-05,
      "loss": 0.8419,
      "step": 176660
    },
    {
      "epoch": 3.1017047349848137,
      "grad_norm": 0.05770249725276673,
      "learning_rate": 6.670841929683658e-05,
      "loss": 0.8461,
      "step": 176670
    },
    {
      "epoch": 3.101880299864815,
      "grad_norm": 0.05317057757347464,
      "learning_rate": 6.669796573432143e-05,
      "loss": 0.8408,
      "step": 176680
    },
    {
      "epoch": 3.1020558647448166,
      "grad_norm": 0.05924226787374698,
      "learning_rate": 6.668751260641959e-05,
      "loss": 0.8507,
      "step": 176690
    },
    {
      "epoch": 3.1022314296248177,
      "grad_norm": 0.05570965606153921,
      "learning_rate": 6.667705991326354e-05,
      "loss": 0.836,
      "step": 176700
    },
    {
      "epoch": 3.102406994504819,
      "grad_norm": 0.058139195393872245,
      "learning_rate": 6.666660765498575e-05,
      "loss": 0.843,
      "step": 176710
    },
    {
      "epoch": 3.1025825593848206,
      "grad_norm": 0.04211043172841682,
      "learning_rate": 6.665615583171858e-05,
      "loss": 0.85,
      "step": 176720
    },
    {
      "epoch": 3.102758124264822,
      "grad_norm": 0.04963040509697402,
      "learning_rate": 6.664570444359453e-05,
      "loss": 0.8432,
      "step": 176730
    },
    {
      "epoch": 3.1029336891448236,
      "grad_norm": 0.04916544762149151,
      "learning_rate": 6.6635253490746e-05,
      "loss": 0.8417,
      "step": 176740
    },
    {
      "epoch": 3.103109254024825,
      "grad_norm": 0.044549785156672114,
      "learning_rate": 6.662480297330535e-05,
      "loss": 0.8446,
      "step": 176750
    },
    {
      "epoch": 3.103284818904826,
      "grad_norm": 0.05475749730330051,
      "learning_rate": 6.661435289140509e-05,
      "loss": 0.8407,
      "step": 176760
    },
    {
      "epoch": 3.1034603837848276,
      "grad_norm": 0.07226929245841855,
      "learning_rate": 6.660390324517754e-05,
      "loss": 0.8399,
      "step": 176770
    },
    {
      "epoch": 3.103635948664829,
      "grad_norm": 0.04361924738475813,
      "learning_rate": 6.659345403475514e-05,
      "loss": 0.8351,
      "step": 176780
    },
    {
      "epoch": 3.1038115135448305,
      "grad_norm": 0.05299461124014582,
      "learning_rate": 6.658300526027024e-05,
      "loss": 0.8432,
      "step": 176790
    },
    {
      "epoch": 3.103987078424832,
      "grad_norm": 0.06824013958938431,
      "learning_rate": 6.657255692185528e-05,
      "loss": 0.8437,
      "step": 176800
    },
    {
      "epoch": 3.1041626433048335,
      "grad_norm": 0.0775442856086218,
      "learning_rate": 6.656210901964264e-05,
      "loss": 0.8394,
      "step": 176810
    },
    {
      "epoch": 3.1043382081848345,
      "grad_norm": 0.05931238196283119,
      "learning_rate": 6.655166155376463e-05,
      "loss": 0.8406,
      "step": 176820
    },
    {
      "epoch": 3.104513773064836,
      "grad_norm": 0.06084799071151443,
      "learning_rate": 6.65412145243537e-05,
      "loss": 0.8491,
      "step": 176830
    },
    {
      "epoch": 3.1046893379448375,
      "grad_norm": 0.054876693755206984,
      "learning_rate": 6.653076793154221e-05,
      "loss": 0.8454,
      "step": 176840
    },
    {
      "epoch": 3.104864902824839,
      "grad_norm": 0.04786554464039425,
      "learning_rate": 6.652032177546248e-05,
      "loss": 0.8393,
      "step": 176850
    },
    {
      "epoch": 3.1050404677048404,
      "grad_norm": 0.05276662942586663,
      "learning_rate": 6.650987605624686e-05,
      "loss": 0.8389,
      "step": 176860
    },
    {
      "epoch": 3.105216032584842,
      "grad_norm": 0.059039383775644036,
      "learning_rate": 6.649943077402774e-05,
      "loss": 0.845,
      "step": 176870
    },
    {
      "epoch": 3.105391597464843,
      "grad_norm": 0.07418162372790557,
      "learning_rate": 6.648898592893744e-05,
      "loss": 0.8371,
      "step": 176880
    },
    {
      "epoch": 3.1055671623448444,
      "grad_norm": 0.0437778118255116,
      "learning_rate": 6.647854152110828e-05,
      "loss": 0.85,
      "step": 176890
    },
    {
      "epoch": 3.105742727224846,
      "grad_norm": 0.06086256586511177,
      "learning_rate": 6.646809755067263e-05,
      "loss": 0.8405,
      "step": 176900
    },
    {
      "epoch": 3.1059182921048474,
      "grad_norm": 0.04707679576738242,
      "learning_rate": 6.645765401776282e-05,
      "loss": 0.8406,
      "step": 176910
    },
    {
      "epoch": 3.106093856984849,
      "grad_norm": 0.05203652217599716,
      "learning_rate": 6.644721092251113e-05,
      "loss": 0.8517,
      "step": 176920
    },
    {
      "epoch": 3.1062694218648503,
      "grad_norm": 0.05126471686367877,
      "learning_rate": 6.643676826504992e-05,
      "loss": 0.8373,
      "step": 176930
    },
    {
      "epoch": 3.1064449867448514,
      "grad_norm": 0.04537980276521672,
      "learning_rate": 6.642632604551148e-05,
      "loss": 0.8469,
      "step": 176940
    },
    {
      "epoch": 3.106620551624853,
      "grad_norm": 0.0595798849982844,
      "learning_rate": 6.641588426402811e-05,
      "loss": 0.8386,
      "step": 176950
    },
    {
      "epoch": 3.1067961165048543,
      "grad_norm": 0.058961097617391996,
      "learning_rate": 6.64054429207321e-05,
      "loss": 0.8353,
      "step": 176960
    },
    {
      "epoch": 3.106971681384856,
      "grad_norm": 0.055638564737910505,
      "learning_rate": 6.639500201575577e-05,
      "loss": 0.8373,
      "step": 176970
    },
    {
      "epoch": 3.1071472462648573,
      "grad_norm": 0.06332925779999139,
      "learning_rate": 6.638456154923141e-05,
      "loss": 0.8421,
      "step": 176980
    },
    {
      "epoch": 3.1073228111448588,
      "grad_norm": 0.04362178442791679,
      "learning_rate": 6.637412152129122e-05,
      "loss": 0.8383,
      "step": 176990
    },
    {
      "epoch": 3.10749837602486,
      "grad_norm": 0.06603978402243751,
      "learning_rate": 6.636368193206762e-05,
      "loss": 0.8459,
      "step": 177000
    },
    {
      "epoch": 3.1076739409048613,
      "grad_norm": 0.06716662486743671,
      "learning_rate": 6.63532427816928e-05,
      "loss": 0.8469,
      "step": 177010
    },
    {
      "epoch": 3.1078495057848627,
      "grad_norm": 0.05361279841961661,
      "learning_rate": 6.634280407029904e-05,
      "loss": 0.8391,
      "step": 177020
    },
    {
      "epoch": 3.108025070664864,
      "grad_norm": 0.04905073582614076,
      "learning_rate": 6.633236579801856e-05,
      "loss": 0.8316,
      "step": 177030
    },
    {
      "epoch": 3.1082006355448657,
      "grad_norm": 0.06290642479730899,
      "learning_rate": 6.632192796498368e-05,
      "loss": 0.8474,
      "step": 177040
    },
    {
      "epoch": 3.108376200424867,
      "grad_norm": 0.08892634083853725,
      "learning_rate": 6.631149057132661e-05,
      "loss": 0.8401,
      "step": 177050
    },
    {
      "epoch": 3.108551765304868,
      "grad_norm": 0.06368479781067783,
      "learning_rate": 6.630105361717956e-05,
      "loss": 0.8425,
      "step": 177060
    },
    {
      "epoch": 3.1087273301848697,
      "grad_norm": 0.052653646722849785,
      "learning_rate": 6.629061710267485e-05,
      "loss": 0.8321,
      "step": 177070
    },
    {
      "epoch": 3.108902895064871,
      "grad_norm": 0.05370815297699797,
      "learning_rate": 6.628018102794468e-05,
      "loss": 0.8355,
      "step": 177080
    },
    {
      "epoch": 3.1090784599448726,
      "grad_norm": 0.04901570217878158,
      "learning_rate": 6.626974539312123e-05,
      "loss": 0.8406,
      "step": 177090
    },
    {
      "epoch": 3.109254024824874,
      "grad_norm": 0.04954390688303785,
      "learning_rate": 6.625931019833682e-05,
      "loss": 0.8449,
      "step": 177100
    },
    {
      "epoch": 3.1094295897048756,
      "grad_norm": 0.05063946737933878,
      "learning_rate": 6.62488754437236e-05,
      "loss": 0.8516,
      "step": 177110
    },
    {
      "epoch": 3.1096051545848766,
      "grad_norm": 0.0631409712315275,
      "learning_rate": 6.623844112941377e-05,
      "loss": 0.8468,
      "step": 177120
    },
    {
      "epoch": 3.109780719464878,
      "grad_norm": 0.050072784774591685,
      "learning_rate": 6.622800725553954e-05,
      "loss": 0.8378,
      "step": 177130
    },
    {
      "epoch": 3.1099562843448796,
      "grad_norm": 0.06671118380752444,
      "learning_rate": 6.621757382223314e-05,
      "loss": 0.85,
      "step": 177140
    },
    {
      "epoch": 3.110131849224881,
      "grad_norm": 0.06683768119319104,
      "learning_rate": 6.620714082962674e-05,
      "loss": 0.8429,
      "step": 177150
    },
    {
      "epoch": 3.1103074141048825,
      "grad_norm": 0.052705652239980465,
      "learning_rate": 6.619670827785252e-05,
      "loss": 0.8434,
      "step": 177160
    },
    {
      "epoch": 3.110482978984884,
      "grad_norm": 0.05944106098368148,
      "learning_rate": 6.618627616704268e-05,
      "loss": 0.8391,
      "step": 177170
    },
    {
      "epoch": 3.1106585438648855,
      "grad_norm": 0.05080230217133314,
      "learning_rate": 6.617584449732942e-05,
      "loss": 0.8384,
      "step": 177180
    },
    {
      "epoch": 3.1108341087448865,
      "grad_norm": 0.05243556930577264,
      "learning_rate": 6.616541326884487e-05,
      "loss": 0.8449,
      "step": 177190
    },
    {
      "epoch": 3.111009673624888,
      "grad_norm": 0.043715578195933726,
      "learning_rate": 6.61549824817212e-05,
      "loss": 0.8418,
      "step": 177200
    },
    {
      "epoch": 3.1111852385048895,
      "grad_norm": 0.045669521006695274,
      "learning_rate": 6.614455213609059e-05,
      "loss": 0.8469,
      "step": 177210
    },
    {
      "epoch": 3.111360803384891,
      "grad_norm": 0.06094747656221403,
      "learning_rate": 6.61341222320852e-05,
      "loss": 0.8389,
      "step": 177220
    },
    {
      "epoch": 3.1115363682648924,
      "grad_norm": 0.057818860013321254,
      "learning_rate": 6.61236927698371e-05,
      "loss": 0.8368,
      "step": 177230
    },
    {
      "epoch": 3.1117119331448935,
      "grad_norm": 0.04582359894583239,
      "learning_rate": 6.611326374947855e-05,
      "loss": 0.8323,
      "step": 177240
    },
    {
      "epoch": 3.111887498024895,
      "grad_norm": 0.0499217360505235,
      "learning_rate": 6.610283517114165e-05,
      "loss": 0.8478,
      "step": 177250
    },
    {
      "epoch": 3.1120630629048964,
      "grad_norm": 0.04413072104359998,
      "learning_rate": 6.609240703495846e-05,
      "loss": 0.8396,
      "step": 177260
    },
    {
      "epoch": 3.112238627784898,
      "grad_norm": 0.0634393646853726,
      "learning_rate": 6.60819793410612e-05,
      "loss": 0.8407,
      "step": 177270
    },
    {
      "epoch": 3.1124141926648994,
      "grad_norm": 0.05924976277905708,
      "learning_rate": 6.607155208958196e-05,
      "loss": 0.8389,
      "step": 177280
    },
    {
      "epoch": 3.112589757544901,
      "grad_norm": 0.07737968740057356,
      "learning_rate": 6.606112528065285e-05,
      "loss": 0.8384,
      "step": 177290
    },
    {
      "epoch": 3.1127653224249023,
      "grad_norm": 0.04845540781509291,
      "learning_rate": 6.605069891440596e-05,
      "loss": 0.8437,
      "step": 177300
    },
    {
      "epoch": 3.1129408873049034,
      "grad_norm": 0.059077194773759285,
      "learning_rate": 6.604027299097341e-05,
      "loss": 0.8415,
      "step": 177310
    },
    {
      "epoch": 3.113116452184905,
      "grad_norm": 0.057515580874801985,
      "learning_rate": 6.602984751048733e-05,
      "loss": 0.8523,
      "step": 177320
    },
    {
      "epoch": 3.1132920170649063,
      "grad_norm": 0.07952376954277225,
      "learning_rate": 6.601942247307973e-05,
      "loss": 0.8367,
      "step": 177330
    },
    {
      "epoch": 3.113467581944908,
      "grad_norm": 0.049294534387541104,
      "learning_rate": 6.600899787888279e-05,
      "loss": 0.8403,
      "step": 177340
    },
    {
      "epoch": 3.1136431468249093,
      "grad_norm": 0.05251196744729007,
      "learning_rate": 6.599857372802857e-05,
      "loss": 0.8372,
      "step": 177350
    },
    {
      "epoch": 3.1138187117049108,
      "grad_norm": 0.0433265171218327,
      "learning_rate": 6.598815002064911e-05,
      "loss": 0.8369,
      "step": 177360
    },
    {
      "epoch": 3.113994276584912,
      "grad_norm": 0.057443280667234066,
      "learning_rate": 6.597772675687648e-05,
      "loss": 0.8402,
      "step": 177370
    },
    {
      "epoch": 3.1141698414649133,
      "grad_norm": 0.0422988025702983,
      "learning_rate": 6.596730393684279e-05,
      "loss": 0.8481,
      "step": 177380
    },
    {
      "epoch": 3.1143454063449147,
      "grad_norm": 0.05560227758889045,
      "learning_rate": 6.595688156068008e-05,
      "loss": 0.8461,
      "step": 177390
    },
    {
      "epoch": 3.1145209712249162,
      "grad_norm": 0.05662321151439553,
      "learning_rate": 6.594645962852034e-05,
      "loss": 0.8369,
      "step": 177400
    },
    {
      "epoch": 3.1146965361049177,
      "grad_norm": 0.04859052372818408,
      "learning_rate": 6.593603814049573e-05,
      "loss": 0.8429,
      "step": 177410
    },
    {
      "epoch": 3.114872100984919,
      "grad_norm": 0.05179253727012936,
      "learning_rate": 6.592561709673823e-05,
      "loss": 0.851,
      "step": 177420
    },
    {
      "epoch": 3.11504766586492,
      "grad_norm": 0.06962921917500237,
      "learning_rate": 6.591519649737985e-05,
      "loss": 0.8409,
      "step": 177430
    },
    {
      "epoch": 3.1152232307449217,
      "grad_norm": 0.05129597705304573,
      "learning_rate": 6.59047763425527e-05,
      "loss": 0.8398,
      "step": 177440
    },
    {
      "epoch": 3.115398795624923,
      "grad_norm": 0.05974106887912427,
      "learning_rate": 6.589435663238874e-05,
      "loss": 0.8427,
      "step": 177450
    },
    {
      "epoch": 3.1155743605049246,
      "grad_norm": 0.05287089539452583,
      "learning_rate": 6.588393736702002e-05,
      "loss": 0.8426,
      "step": 177460
    },
    {
      "epoch": 3.115749925384926,
      "grad_norm": 0.05228434276704972,
      "learning_rate": 6.587351854657852e-05,
      "loss": 0.8463,
      "step": 177470
    },
    {
      "epoch": 3.1159254902649276,
      "grad_norm": 0.042220703108511205,
      "learning_rate": 6.586310017119629e-05,
      "loss": 0.8394,
      "step": 177480
    },
    {
      "epoch": 3.1161010551449286,
      "grad_norm": 0.049338039564223814,
      "learning_rate": 6.585268224100533e-05,
      "loss": 0.8411,
      "step": 177490
    },
    {
      "epoch": 3.11627662002493,
      "grad_norm": 0.07065650648378875,
      "learning_rate": 6.584226475613756e-05,
      "loss": 0.8408,
      "step": 177500
    },
    {
      "epoch": 3.1164521849049316,
      "grad_norm": 0.05625997273676701,
      "learning_rate": 6.583184771672508e-05,
      "loss": 0.8434,
      "step": 177510
    },
    {
      "epoch": 3.116627749784933,
      "grad_norm": 0.05543963462161017,
      "learning_rate": 6.582143112289982e-05,
      "loss": 0.8464,
      "step": 177520
    },
    {
      "epoch": 3.1168033146649345,
      "grad_norm": 0.04777556372070191,
      "learning_rate": 6.58110149747938e-05,
      "loss": 0.8409,
      "step": 177530
    },
    {
      "epoch": 3.116978879544936,
      "grad_norm": 0.05238845043874038,
      "learning_rate": 6.580059927253891e-05,
      "loss": 0.8471,
      "step": 177540
    },
    {
      "epoch": 3.117154444424937,
      "grad_norm": 0.051129392155283816,
      "learning_rate": 6.579018401626722e-05,
      "loss": 0.8296,
      "step": 177550
    },
    {
      "epoch": 3.1173300093049385,
      "grad_norm": 0.0598766621646147,
      "learning_rate": 6.577976920611063e-05,
      "loss": 0.8449,
      "step": 177560
    },
    {
      "epoch": 3.11750557418494,
      "grad_norm": 0.050334157963618974,
      "learning_rate": 6.576935484220107e-05,
      "loss": 0.8444,
      "step": 177570
    },
    {
      "epoch": 3.1176811390649415,
      "grad_norm": 0.045198543657779075,
      "learning_rate": 6.575894092467059e-05,
      "loss": 0.8442,
      "step": 177580
    },
    {
      "epoch": 3.117856703944943,
      "grad_norm": 0.04941929617447806,
      "learning_rate": 6.574852745365108e-05,
      "loss": 0.8393,
      "step": 177590
    },
    {
      "epoch": 3.1180322688249444,
      "grad_norm": 0.05406099161203175,
      "learning_rate": 6.573811442927447e-05,
      "loss": 0.847,
      "step": 177600
    },
    {
      "epoch": 3.1182078337049455,
      "grad_norm": 0.0549761192786372,
      "learning_rate": 6.572770185167272e-05,
      "loss": 0.8371,
      "step": 177610
    },
    {
      "epoch": 3.118383398584947,
      "grad_norm": 0.042719478552425694,
      "learning_rate": 6.571728972097774e-05,
      "loss": 0.8482,
      "step": 177620
    },
    {
      "epoch": 3.1185589634649484,
      "grad_norm": 0.04548676998799097,
      "learning_rate": 6.570687803732148e-05,
      "loss": 0.8449,
      "step": 177630
    },
    {
      "epoch": 3.11873452834495,
      "grad_norm": 0.05102509185678363,
      "learning_rate": 6.569646680083578e-05,
      "loss": 0.841,
      "step": 177640
    },
    {
      "epoch": 3.1189100932249514,
      "grad_norm": 0.050657791357171196,
      "learning_rate": 6.568605601165268e-05,
      "loss": 0.8425,
      "step": 177650
    },
    {
      "epoch": 3.119085658104953,
      "grad_norm": 0.05272877547268544,
      "learning_rate": 6.567564566990399e-05,
      "loss": 0.8492,
      "step": 177660
    },
    {
      "epoch": 3.119261222984954,
      "grad_norm": 0.0514517185757152,
      "learning_rate": 6.56652357757216e-05,
      "loss": 0.8451,
      "step": 177670
    },
    {
      "epoch": 3.1194367878649554,
      "grad_norm": 0.0549490760175467,
      "learning_rate": 6.565482632923749e-05,
      "loss": 0.836,
      "step": 177680
    },
    {
      "epoch": 3.119612352744957,
      "grad_norm": 0.058762956298762534,
      "learning_rate": 6.56444173305835e-05,
      "loss": 0.8447,
      "step": 177690
    },
    {
      "epoch": 3.1197879176249583,
      "grad_norm": 0.07537230940047412,
      "learning_rate": 6.563400877989152e-05,
      "loss": 0.8378,
      "step": 177700
    },
    {
      "epoch": 3.11996348250496,
      "grad_norm": 0.05858591671875514,
      "learning_rate": 6.56236006772934e-05,
      "loss": 0.8352,
      "step": 177710
    },
    {
      "epoch": 3.1201390473849613,
      "grad_norm": 0.04506413212303719,
      "learning_rate": 6.561319302292109e-05,
      "loss": 0.8419,
      "step": 177720
    },
    {
      "epoch": 3.1203146122649623,
      "grad_norm": 0.06999931293305242,
      "learning_rate": 6.560278581690638e-05,
      "loss": 0.8435,
      "step": 177730
    },
    {
      "epoch": 3.120490177144964,
      "grad_norm": 0.04581097880265455,
      "learning_rate": 6.559237905938111e-05,
      "loss": 0.8422,
      "step": 177740
    },
    {
      "epoch": 3.1206657420249653,
      "grad_norm": 0.057578523798732954,
      "learning_rate": 6.558197275047725e-05,
      "loss": 0.8423,
      "step": 177750
    },
    {
      "epoch": 3.1208413069049667,
      "grad_norm": 0.06674118325822215,
      "learning_rate": 6.557156689032655e-05,
      "loss": 0.8423,
      "step": 177760
    },
    {
      "epoch": 3.1210168717849682,
      "grad_norm": 0.050372967372002725,
      "learning_rate": 6.556116147906091e-05,
      "loss": 0.8398,
      "step": 177770
    },
    {
      "epoch": 3.1211924366649697,
      "grad_norm": 0.04890613674838637,
      "learning_rate": 6.555075651681213e-05,
      "loss": 0.8458,
      "step": 177780
    },
    {
      "epoch": 3.1213680015449707,
      "grad_norm": 0.044581533984415216,
      "learning_rate": 6.554035200371207e-05,
      "loss": 0.8457,
      "step": 177790
    },
    {
      "epoch": 3.121543566424972,
      "grad_norm": 0.050548191870678216,
      "learning_rate": 6.552994793989255e-05,
      "loss": 0.8423,
      "step": 177800
    },
    {
      "epoch": 3.1217191313049737,
      "grad_norm": 0.051049445985347774,
      "learning_rate": 6.551954432548535e-05,
      "loss": 0.838,
      "step": 177810
    },
    {
      "epoch": 3.121894696184975,
      "grad_norm": 0.050643221871884074,
      "learning_rate": 6.550914116062236e-05,
      "loss": 0.845,
      "step": 177820
    },
    {
      "epoch": 3.1220702610649766,
      "grad_norm": 0.07070329059918229,
      "learning_rate": 6.549873844543536e-05,
      "loss": 0.8362,
      "step": 177830
    },
    {
      "epoch": 3.122245825944978,
      "grad_norm": 0.049110603660844736,
      "learning_rate": 6.54883361800561e-05,
      "loss": 0.8364,
      "step": 177840
    },
    {
      "epoch": 3.122421390824979,
      "grad_norm": 0.045021808045287666,
      "learning_rate": 6.547793436461645e-05,
      "loss": 0.8443,
      "step": 177850
    },
    {
      "epoch": 3.1225969557049806,
      "grad_norm": 0.05827316435279269,
      "learning_rate": 6.54675329992482e-05,
      "loss": 0.836,
      "step": 177860
    },
    {
      "epoch": 3.122772520584982,
      "grad_norm": 0.05658196055184611,
      "learning_rate": 6.545713208408311e-05,
      "loss": 0.8439,
      "step": 177870
    },
    {
      "epoch": 3.1229480854649836,
      "grad_norm": 0.06053788321796417,
      "learning_rate": 6.544673161925296e-05,
      "loss": 0.8449,
      "step": 177880
    },
    {
      "epoch": 3.123123650344985,
      "grad_norm": 0.0522538448885441,
      "learning_rate": 6.543633160488953e-05,
      "loss": 0.8522,
      "step": 177890
    },
    {
      "epoch": 3.1232992152249865,
      "grad_norm": 0.0571582676805879,
      "learning_rate": 6.542593204112461e-05,
      "loss": 0.8501,
      "step": 177900
    },
    {
      "epoch": 3.1234747801049876,
      "grad_norm": 0.08569163300346559,
      "learning_rate": 6.54155329280899e-05,
      "loss": 0.843,
      "step": 177910
    },
    {
      "epoch": 3.123650344984989,
      "grad_norm": 0.06206070894049456,
      "learning_rate": 6.540513426591726e-05,
      "loss": 0.8423,
      "step": 177920
    },
    {
      "epoch": 3.1238259098649905,
      "grad_norm": 0.05516956704649832,
      "learning_rate": 6.539473605473839e-05,
      "loss": 0.8354,
      "step": 177930
    },
    {
      "epoch": 3.124001474744992,
      "grad_norm": 0.06073482347575983,
      "learning_rate": 6.538433829468501e-05,
      "loss": 0.8391,
      "step": 177940
    },
    {
      "epoch": 3.1241770396249935,
      "grad_norm": 0.04959690200497376,
      "learning_rate": 6.53739409858889e-05,
      "loss": 0.8457,
      "step": 177950
    },
    {
      "epoch": 3.124352604504995,
      "grad_norm": 0.05311844017994634,
      "learning_rate": 6.536354412848182e-05,
      "loss": 0.8353,
      "step": 177960
    },
    {
      "epoch": 3.1245281693849964,
      "grad_norm": 0.06792895464064776,
      "learning_rate": 6.535314772259544e-05,
      "loss": 0.8441,
      "step": 177970
    },
    {
      "epoch": 3.1247037342649975,
      "grad_norm": 0.04932947436940383,
      "learning_rate": 6.534275176836148e-05,
      "loss": 0.8455,
      "step": 177980
    },
    {
      "epoch": 3.124879299144999,
      "grad_norm": 0.06037963961932143,
      "learning_rate": 6.533235626591171e-05,
      "loss": 0.8421,
      "step": 177990
    },
    {
      "epoch": 3.1250548640250004,
      "grad_norm": 0.05880350181803266,
      "learning_rate": 6.532196121537785e-05,
      "loss": 0.8447,
      "step": 178000
    },
    {
      "epoch": 3.125230428905002,
      "grad_norm": 0.04513031032796769,
      "learning_rate": 6.531156661689152e-05,
      "loss": 0.843,
      "step": 178010
    },
    {
      "epoch": 3.1254059937850034,
      "grad_norm": 0.0466876626317143,
      "learning_rate": 6.530117247058453e-05,
      "loss": 0.8487,
      "step": 178020
    },
    {
      "epoch": 3.1255815586650044,
      "grad_norm": 0.08869849851004087,
      "learning_rate": 6.529077877658851e-05,
      "loss": 0.8406,
      "step": 178030
    },
    {
      "epoch": 3.125757123545006,
      "grad_norm": 0.052984688313148974,
      "learning_rate": 6.528038553503517e-05,
      "loss": 0.8443,
      "step": 178040
    },
    {
      "epoch": 3.1259326884250074,
      "grad_norm": 0.05291178664632654,
      "learning_rate": 6.526999274605618e-05,
      "loss": 0.8459,
      "step": 178050
    },
    {
      "epoch": 3.126108253305009,
      "grad_norm": 0.0575801899515457,
      "learning_rate": 6.525960040978324e-05,
      "loss": 0.8409,
      "step": 178060
    },
    {
      "epoch": 3.1262838181850103,
      "grad_norm": 0.04974705639487808,
      "learning_rate": 6.524920852634802e-05,
      "loss": 0.8424,
      "step": 178070
    },
    {
      "epoch": 3.126459383065012,
      "grad_norm": 0.05101300547258843,
      "learning_rate": 6.523881709588214e-05,
      "loss": 0.8387,
      "step": 178080
    },
    {
      "epoch": 3.1266349479450133,
      "grad_norm": 0.05515173488164023,
      "learning_rate": 6.522842611851734e-05,
      "loss": 0.8341,
      "step": 178090
    },
    {
      "epoch": 3.1268105128250143,
      "grad_norm": 0.053378702734042466,
      "learning_rate": 6.521803559438524e-05,
      "loss": 0.8462,
      "step": 178100
    },
    {
      "epoch": 3.126986077705016,
      "grad_norm": 0.05772258034386112,
      "learning_rate": 6.520764552361746e-05,
      "loss": 0.8401,
      "step": 178110
    },
    {
      "epoch": 3.1271616425850173,
      "grad_norm": 0.04355293450467232,
      "learning_rate": 6.51972559063457e-05,
      "loss": 0.8387,
      "step": 178120
    },
    {
      "epoch": 3.1273372074650188,
      "grad_norm": 0.05455392987846172,
      "learning_rate": 6.518686674270156e-05,
      "loss": 0.85,
      "step": 178130
    },
    {
      "epoch": 3.1275127723450202,
      "grad_norm": 0.0483368953028399,
      "learning_rate": 6.517647803281667e-05,
      "loss": 0.8459,
      "step": 178140
    },
    {
      "epoch": 3.1276883372250217,
      "grad_norm": 0.048856275946127944,
      "learning_rate": 6.516608977682267e-05,
      "loss": 0.8344,
      "step": 178150
    },
    {
      "epoch": 3.1278639021050227,
      "grad_norm": 0.057198972135459145,
      "learning_rate": 6.51557019748512e-05,
      "loss": 0.8458,
      "step": 178160
    },
    {
      "epoch": 3.128039466985024,
      "grad_norm": 0.047307026355425094,
      "learning_rate": 6.514531462703383e-05,
      "loss": 0.8408,
      "step": 178170
    },
    {
      "epoch": 3.1282150318650257,
      "grad_norm": 0.05284158573568351,
      "learning_rate": 6.513492773350218e-05,
      "loss": 0.8373,
      "step": 178180
    },
    {
      "epoch": 3.128390596745027,
      "grad_norm": 0.053462496863110165,
      "learning_rate": 6.512454129438792e-05,
      "loss": 0.8529,
      "step": 178190
    },
    {
      "epoch": 3.1285661616250287,
      "grad_norm": 0.06871051625540021,
      "learning_rate": 6.511415530982258e-05,
      "loss": 0.8448,
      "step": 178200
    },
    {
      "epoch": 3.12874172650503,
      "grad_norm": 0.05140171140299039,
      "learning_rate": 6.510376977993777e-05,
      "loss": 0.8431,
      "step": 178210
    },
    {
      "epoch": 3.128917291385031,
      "grad_norm": 0.06772899811762499,
      "learning_rate": 6.509338470486506e-05,
      "loss": 0.8362,
      "step": 178220
    },
    {
      "epoch": 3.1290928562650326,
      "grad_norm": 0.07046765189879099,
      "learning_rate": 6.508300008473605e-05,
      "loss": 0.841,
      "step": 178230
    },
    {
      "epoch": 3.129268421145034,
      "grad_norm": 0.056475976777573156,
      "learning_rate": 6.507261591968232e-05,
      "loss": 0.8404,
      "step": 178240
    },
    {
      "epoch": 3.1294439860250356,
      "grad_norm": 0.06777508874136066,
      "learning_rate": 6.506223220983541e-05,
      "loss": 0.8494,
      "step": 178250
    },
    {
      "epoch": 3.129619550905037,
      "grad_norm": 0.045103898504592135,
      "learning_rate": 6.505184895532692e-05,
      "loss": 0.8446,
      "step": 178260
    },
    {
      "epoch": 3.1297951157850386,
      "grad_norm": 0.06302107889963031,
      "learning_rate": 6.504146615628843e-05,
      "loss": 0.8358,
      "step": 178270
    },
    {
      "epoch": 3.1299706806650396,
      "grad_norm": 0.044531480238633056,
      "learning_rate": 6.503108381285139e-05,
      "loss": 0.8396,
      "step": 178280
    },
    {
      "epoch": 3.130146245545041,
      "grad_norm": 0.06056271662238985,
      "learning_rate": 6.502070192514745e-05,
      "loss": 0.8424,
      "step": 178290
    },
    {
      "epoch": 3.1303218104250425,
      "grad_norm": 0.052644724285929444,
      "learning_rate": 6.501032049330812e-05,
      "loss": 0.8519,
      "step": 178300
    },
    {
      "epoch": 3.130497375305044,
      "grad_norm": 0.04584034447418287,
      "learning_rate": 6.499993951746492e-05,
      "loss": 0.8478,
      "step": 178310
    },
    {
      "epoch": 3.1306729401850455,
      "grad_norm": 0.06682958527392269,
      "learning_rate": 6.498955899774937e-05,
      "loss": 0.839,
      "step": 178320
    },
    {
      "epoch": 3.130848505065047,
      "grad_norm": 0.0524801982783204,
      "learning_rate": 6.497917893429303e-05,
      "loss": 0.8477,
      "step": 178330
    },
    {
      "epoch": 3.131024069945048,
      "grad_norm": 0.05587078694312685,
      "learning_rate": 6.49687993272274e-05,
      "loss": 0.8404,
      "step": 178340
    },
    {
      "epoch": 3.1311996348250495,
      "grad_norm": 0.06338676452876384,
      "learning_rate": 6.495842017668393e-05,
      "loss": 0.8339,
      "step": 178350
    },
    {
      "epoch": 3.131375199705051,
      "grad_norm": 0.055832188668211,
      "learning_rate": 6.494804148279424e-05,
      "loss": 0.8395,
      "step": 178360
    },
    {
      "epoch": 3.1315507645850524,
      "grad_norm": 0.05683933296879574,
      "learning_rate": 6.493766324568977e-05,
      "loss": 0.8456,
      "step": 178370
    },
    {
      "epoch": 3.131726329465054,
      "grad_norm": 0.057608682793735315,
      "learning_rate": 6.492728546550201e-05,
      "loss": 0.8437,
      "step": 178380
    },
    {
      "epoch": 3.1319018943450554,
      "grad_norm": 0.05250791383340115,
      "learning_rate": 6.491690814236244e-05,
      "loss": 0.8473,
      "step": 178390
    },
    {
      "epoch": 3.1320774592250564,
      "grad_norm": 0.05302478241963316,
      "learning_rate": 6.490653127640258e-05,
      "loss": 0.8363,
      "step": 178400
    },
    {
      "epoch": 3.132253024105058,
      "grad_norm": 0.054861884203445614,
      "learning_rate": 6.489615486775389e-05,
      "loss": 0.8402,
      "step": 178410
    },
    {
      "epoch": 3.1324285889850594,
      "grad_norm": 0.05675549303012766,
      "learning_rate": 6.48857789165478e-05,
      "loss": 0.8524,
      "step": 178420
    },
    {
      "epoch": 3.132604153865061,
      "grad_norm": 0.046365543689716654,
      "learning_rate": 6.487540342291585e-05,
      "loss": 0.8389,
      "step": 178430
    },
    {
      "epoch": 3.1327797187450623,
      "grad_norm": 0.05257065909712258,
      "learning_rate": 6.486502838698947e-05,
      "loss": 0.8362,
      "step": 178440
    },
    {
      "epoch": 3.132955283625064,
      "grad_norm": 0.054202935245921804,
      "learning_rate": 6.485465380890006e-05,
      "loss": 0.8396,
      "step": 178450
    },
    {
      "epoch": 3.133130848505065,
      "grad_norm": 0.04899213709845828,
      "learning_rate": 6.484427968877914e-05,
      "loss": 0.8492,
      "step": 178460
    },
    {
      "epoch": 3.1333064133850663,
      "grad_norm": 0.053653682023365123,
      "learning_rate": 6.483390602675814e-05,
      "loss": 0.8447,
      "step": 178470
    },
    {
      "epoch": 3.133481978265068,
      "grad_norm": 0.04508661937337517,
      "learning_rate": 6.482353282296848e-05,
      "loss": 0.8473,
      "step": 178480
    },
    {
      "epoch": 3.1336575431450693,
      "grad_norm": 0.046944902968170575,
      "learning_rate": 6.481316007754156e-05,
      "loss": 0.8453,
      "step": 178490
    },
    {
      "epoch": 3.1338331080250708,
      "grad_norm": 0.059019728002192344,
      "learning_rate": 6.480278779060888e-05,
      "loss": 0.8462,
      "step": 178500
    },
    {
      "epoch": 3.1340086729050722,
      "grad_norm": 0.07229071392889384,
      "learning_rate": 6.479241596230179e-05,
      "loss": 0.8453,
      "step": 178510
    },
    {
      "epoch": 3.1341842377850733,
      "grad_norm": 0.05474900694994087,
      "learning_rate": 6.47820445927517e-05,
      "loss": 0.8398,
      "step": 178520
    },
    {
      "epoch": 3.1343598026650747,
      "grad_norm": 0.04375012058767881,
      "learning_rate": 6.477167368209009e-05,
      "loss": 0.8458,
      "step": 178530
    },
    {
      "epoch": 3.1345353675450762,
      "grad_norm": 0.08876199821895149,
      "learning_rate": 6.47613032304483e-05,
      "loss": 0.8432,
      "step": 178540
    },
    {
      "epoch": 3.1347109324250777,
      "grad_norm": 0.04834687477612056,
      "learning_rate": 6.475093323795777e-05,
      "loss": 0.8395,
      "step": 178550
    },
    {
      "epoch": 3.134886497305079,
      "grad_norm": 0.056333186252766565,
      "learning_rate": 6.474056370474981e-05,
      "loss": 0.8432,
      "step": 178560
    },
    {
      "epoch": 3.1350620621850807,
      "grad_norm": 0.058316199963526175,
      "learning_rate": 6.473019463095591e-05,
      "loss": 0.8367,
      "step": 178570
    },
    {
      "epoch": 3.1352376270650817,
      "grad_norm": 0.07108979025378816,
      "learning_rate": 6.471982601670737e-05,
      "loss": 0.8483,
      "step": 178580
    },
    {
      "epoch": 3.135413191945083,
      "grad_norm": 0.05725074094522768,
      "learning_rate": 6.470945786213557e-05,
      "loss": 0.8467,
      "step": 178590
    },
    {
      "epoch": 3.1355887568250846,
      "grad_norm": 0.0473667105283192,
      "learning_rate": 6.469909016737193e-05,
      "loss": 0.8432,
      "step": 178600
    },
    {
      "epoch": 3.135764321705086,
      "grad_norm": 0.05539484821659648,
      "learning_rate": 6.468872293254777e-05,
      "loss": 0.8467,
      "step": 178610
    },
    {
      "epoch": 3.1359398865850876,
      "grad_norm": 0.047388617723978704,
      "learning_rate": 6.467835615779441e-05,
      "loss": 0.8433,
      "step": 178620
    },
    {
      "epoch": 3.136115451465089,
      "grad_norm": 0.054421660295673496,
      "learning_rate": 6.466798984324327e-05,
      "loss": 0.8415,
      "step": 178630
    },
    {
      "epoch": 3.1362910163450906,
      "grad_norm": 0.07174169831441726,
      "learning_rate": 6.465762398902567e-05,
      "loss": 0.8405,
      "step": 178640
    },
    {
      "epoch": 3.1364665812250916,
      "grad_norm": 0.05874783902764889,
      "learning_rate": 6.464725859527294e-05,
      "loss": 0.847,
      "step": 178650
    },
    {
      "epoch": 3.136642146105093,
      "grad_norm": 0.0411629571461877,
      "learning_rate": 6.463689366211639e-05,
      "loss": 0.8461,
      "step": 178660
    },
    {
      "epoch": 3.1368177109850945,
      "grad_norm": 0.04513011414921116,
      "learning_rate": 6.462652918968738e-05,
      "loss": 0.8438,
      "step": 178670
    },
    {
      "epoch": 3.136993275865096,
      "grad_norm": 0.0570281454505495,
      "learning_rate": 6.461616517811722e-05,
      "loss": 0.842,
      "step": 178680
    },
    {
      "epoch": 3.1371688407450975,
      "grad_norm": 0.04561104062250132,
      "learning_rate": 6.460580162753717e-05,
      "loss": 0.8472,
      "step": 178690
    },
    {
      "epoch": 3.1373444056250985,
      "grad_norm": 0.05495913863318287,
      "learning_rate": 6.459543853807864e-05,
      "loss": 0.8453,
      "step": 178700
    },
    {
      "epoch": 3.1375199705051,
      "grad_norm": 0.057362467106024,
      "learning_rate": 6.458507590987289e-05,
      "loss": 0.8383,
      "step": 178710
    },
    {
      "epoch": 3.1376955353851015,
      "grad_norm": 0.05580102352375802,
      "learning_rate": 6.457471374305119e-05,
      "loss": 0.8385,
      "step": 178720
    },
    {
      "epoch": 3.137871100265103,
      "grad_norm": 0.05275123280012451,
      "learning_rate": 6.456435203774484e-05,
      "loss": 0.8393,
      "step": 178730
    },
    {
      "epoch": 3.1380466651451044,
      "grad_norm": 0.047782537132588156,
      "learning_rate": 6.455399079408513e-05,
      "loss": 0.8386,
      "step": 178740
    },
    {
      "epoch": 3.138222230025106,
      "grad_norm": 0.05249016959511997,
      "learning_rate": 6.454363001220336e-05,
      "loss": 0.851,
      "step": 178750
    },
    {
      "epoch": 3.1383977949051074,
      "grad_norm": 0.04659148616115913,
      "learning_rate": 6.453326969223073e-05,
      "loss": 0.8357,
      "step": 178760
    },
    {
      "epoch": 3.1385733597851084,
      "grad_norm": 0.044996624513579955,
      "learning_rate": 6.452290983429863e-05,
      "loss": 0.8423,
      "step": 178770
    },
    {
      "epoch": 3.13874892466511,
      "grad_norm": 0.08138170183437587,
      "learning_rate": 6.451255043853825e-05,
      "loss": 0.841,
      "step": 178780
    },
    {
      "epoch": 3.1389244895451114,
      "grad_norm": 0.05460043834890342,
      "learning_rate": 6.45021915050808e-05,
      "loss": 0.8443,
      "step": 178790
    },
    {
      "epoch": 3.139100054425113,
      "grad_norm": 0.046086601702064474,
      "learning_rate": 6.44918330340576e-05,
      "loss": 0.8427,
      "step": 178800
    },
    {
      "epoch": 3.1392756193051143,
      "grad_norm": 0.05081017724090852,
      "learning_rate": 6.448147502559989e-05,
      "loss": 0.845,
      "step": 178810
    },
    {
      "epoch": 3.139451184185116,
      "grad_norm": 0.06520915132643486,
      "learning_rate": 6.44711174798389e-05,
      "loss": 0.8415,
      "step": 178820
    },
    {
      "epoch": 3.139626749065117,
      "grad_norm": 0.06073759458287094,
      "learning_rate": 6.44607603969058e-05,
      "loss": 0.8322,
      "step": 178830
    },
    {
      "epoch": 3.1398023139451183,
      "grad_norm": 0.056793542917856155,
      "learning_rate": 6.445040377693193e-05,
      "loss": 0.8451,
      "step": 178840
    },
    {
      "epoch": 3.13997787882512,
      "grad_norm": 0.04692638731246804,
      "learning_rate": 6.444004762004842e-05,
      "loss": 0.8382,
      "step": 178850
    },
    {
      "epoch": 3.1401534437051213,
      "grad_norm": 0.055049361543355775,
      "learning_rate": 6.442969192638647e-05,
      "loss": 0.8408,
      "step": 178860
    },
    {
      "epoch": 3.1403290085851228,
      "grad_norm": 0.04727303907667797,
      "learning_rate": 6.441933669607739e-05,
      "loss": 0.8477,
      "step": 178870
    },
    {
      "epoch": 3.1405045734651242,
      "grad_norm": 0.06422244986003404,
      "learning_rate": 6.440898192925232e-05,
      "loss": 0.848,
      "step": 178880
    },
    {
      "epoch": 3.1406801383451253,
      "grad_norm": 0.07494289049452428,
      "learning_rate": 6.439862762604244e-05,
      "loss": 0.8459,
      "step": 178890
    },
    {
      "epoch": 3.1408557032251267,
      "grad_norm": 0.047421725440594815,
      "learning_rate": 6.438827378657897e-05,
      "loss": 0.846,
      "step": 178900
    },
    {
      "epoch": 3.1410312681051282,
      "grad_norm": 0.05784004952017789,
      "learning_rate": 6.437792041099309e-05,
      "loss": 0.8452,
      "step": 178910
    },
    {
      "epoch": 3.1412068329851297,
      "grad_norm": 0.058854750341155115,
      "learning_rate": 6.436756749941598e-05,
      "loss": 0.8451,
      "step": 178920
    },
    {
      "epoch": 3.141382397865131,
      "grad_norm": 0.05827025082622834,
      "learning_rate": 6.43572150519788e-05,
      "loss": 0.8422,
      "step": 178930
    },
    {
      "epoch": 3.1415579627451327,
      "grad_norm": 0.052495763461346254,
      "learning_rate": 6.434686306881273e-05,
      "loss": 0.8352,
      "step": 178940
    },
    {
      "epoch": 3.1417335276251337,
      "grad_norm": 0.05647728291398512,
      "learning_rate": 6.433651155004895e-05,
      "loss": 0.8403,
      "step": 178950
    },
    {
      "epoch": 3.141909092505135,
      "grad_norm": 0.052186362278243116,
      "learning_rate": 6.432616049581856e-05,
      "loss": 0.8404,
      "step": 178960
    },
    {
      "epoch": 3.1420846573851366,
      "grad_norm": 0.0609229412661522,
      "learning_rate": 6.431580990625276e-05,
      "loss": 0.843,
      "step": 178970
    },
    {
      "epoch": 3.142260222265138,
      "grad_norm": 0.06077849000626897,
      "learning_rate": 6.43054597814827e-05,
      "loss": 0.837,
      "step": 178980
    },
    {
      "epoch": 3.1424357871451396,
      "grad_norm": 0.06995258551475193,
      "learning_rate": 6.429511012163949e-05,
      "loss": 0.8476,
      "step": 178990
    },
    {
      "epoch": 3.142611352025141,
      "grad_norm": 0.06472802178396476,
      "learning_rate": 6.428476092685425e-05,
      "loss": 0.8312,
      "step": 179000
    },
    {
      "epoch": 3.142786916905142,
      "grad_norm": 0.05504365584092887,
      "learning_rate": 6.427441219725815e-05,
      "loss": 0.8371,
      "step": 179010
    },
    {
      "epoch": 3.1429624817851436,
      "grad_norm": 0.05556149318751837,
      "learning_rate": 6.426406393298226e-05,
      "loss": 0.8341,
      "step": 179020
    },
    {
      "epoch": 3.143138046665145,
      "grad_norm": 0.040136408992915644,
      "learning_rate": 6.425371613415771e-05,
      "loss": 0.8338,
      "step": 179030
    },
    {
      "epoch": 3.1433136115451465,
      "grad_norm": 0.07215797935606835,
      "learning_rate": 6.424336880091564e-05,
      "loss": 0.845,
      "step": 179040
    },
    {
      "epoch": 3.143489176425148,
      "grad_norm": 0.05428939529380394,
      "learning_rate": 6.423302193338715e-05,
      "loss": 0.8499,
      "step": 179050
    },
    {
      "epoch": 3.1436647413051495,
      "grad_norm": 0.05263506993872728,
      "learning_rate": 6.422267553170329e-05,
      "loss": 0.8396,
      "step": 179060
    },
    {
      "epoch": 3.1438403061851505,
      "grad_norm": 0.05414221175979501,
      "learning_rate": 6.42123295959952e-05,
      "loss": 0.8428,
      "step": 179070
    },
    {
      "epoch": 3.144015871065152,
      "grad_norm": 0.058732870138231384,
      "learning_rate": 6.420198412639395e-05,
      "loss": 0.8376,
      "step": 179080
    },
    {
      "epoch": 3.1441914359451535,
      "grad_norm": 0.054402683194074083,
      "learning_rate": 6.419163912303059e-05,
      "loss": 0.8354,
      "step": 179090
    },
    {
      "epoch": 3.144367000825155,
      "grad_norm": 0.051207716253714494,
      "learning_rate": 6.418129458603619e-05,
      "loss": 0.8424,
      "step": 179100
    },
    {
      "epoch": 3.1445425657051564,
      "grad_norm": 0.07356963151404143,
      "learning_rate": 6.417095051554189e-05,
      "loss": 0.8481,
      "step": 179110
    },
    {
      "epoch": 3.144718130585158,
      "grad_norm": 0.05333974481641849,
      "learning_rate": 6.416060691167872e-05,
      "loss": 0.8416,
      "step": 179120
    },
    {
      "epoch": 3.144893695465159,
      "grad_norm": 0.052742192577020675,
      "learning_rate": 6.415026377457767e-05,
      "loss": 0.8371,
      "step": 179130
    },
    {
      "epoch": 3.1450692603451604,
      "grad_norm": 0.06651033098947766,
      "learning_rate": 6.413992110436986e-05,
      "loss": 0.8368,
      "step": 179140
    },
    {
      "epoch": 3.145244825225162,
      "grad_norm": 0.05642497347534963,
      "learning_rate": 6.412957890118633e-05,
      "loss": 0.8384,
      "step": 179150
    },
    {
      "epoch": 3.1454203901051634,
      "grad_norm": 0.07430361949388749,
      "learning_rate": 6.411923716515809e-05,
      "loss": 0.8368,
      "step": 179160
    },
    {
      "epoch": 3.145595954985165,
      "grad_norm": 0.04400441129508039,
      "learning_rate": 6.410889589641618e-05,
      "loss": 0.8477,
      "step": 179170
    },
    {
      "epoch": 3.1457715198651663,
      "grad_norm": 0.042755051733279315,
      "learning_rate": 6.409855509509163e-05,
      "loss": 0.8435,
      "step": 179180
    },
    {
      "epoch": 3.1459470847451674,
      "grad_norm": 0.06702723955601185,
      "learning_rate": 6.408821476131546e-05,
      "loss": 0.8353,
      "step": 179190
    },
    {
      "epoch": 3.146122649625169,
      "grad_norm": 0.05329531883500681,
      "learning_rate": 6.407787489521867e-05,
      "loss": 0.8432,
      "step": 179200
    },
    {
      "epoch": 3.1462982145051703,
      "grad_norm": 0.047773650294997236,
      "learning_rate": 6.40675354969323e-05,
      "loss": 0.845,
      "step": 179210
    },
    {
      "epoch": 3.146473779385172,
      "grad_norm": 0.05822065857073286,
      "learning_rate": 6.405719656658733e-05,
      "loss": 0.8461,
      "step": 179220
    },
    {
      "epoch": 3.1466493442651733,
      "grad_norm": 0.04761168140264076,
      "learning_rate": 6.404685810431478e-05,
      "loss": 0.8344,
      "step": 179230
    },
    {
      "epoch": 3.1468249091451748,
      "grad_norm": 0.0428686081982245,
      "learning_rate": 6.403652011024558e-05,
      "loss": 0.8411,
      "step": 179240
    },
    {
      "epoch": 3.147000474025176,
      "grad_norm": 0.04795499080781019,
      "learning_rate": 6.402618258451077e-05,
      "loss": 0.8369,
      "step": 179250
    },
    {
      "epoch": 3.1471760389051773,
      "grad_norm": 0.048448829765550906,
      "learning_rate": 6.401584552724132e-05,
      "loss": 0.8437,
      "step": 179260
    },
    {
      "epoch": 3.1473516037851788,
      "grad_norm": 0.05509264792456271,
      "learning_rate": 6.400550893856816e-05,
      "loss": 0.8485,
      "step": 179270
    },
    {
      "epoch": 3.1475271686651802,
      "grad_norm": 0.05611725235717654,
      "learning_rate": 6.399517281862233e-05,
      "loss": 0.8455,
      "step": 179280
    },
    {
      "epoch": 3.1477027335451817,
      "grad_norm": 0.046597549023473406,
      "learning_rate": 6.398483716753479e-05,
      "loss": 0.8438,
      "step": 179290
    },
    {
      "epoch": 3.147878298425183,
      "grad_norm": 0.08295332716009504,
      "learning_rate": 6.397450198543638e-05,
      "loss": 0.8344,
      "step": 179300
    },
    {
      "epoch": 3.1480538633051847,
      "grad_norm": 0.05034785425743946,
      "learning_rate": 6.396416727245816e-05,
      "loss": 0.8517,
      "step": 179310
    },
    {
      "epoch": 3.1482294281851857,
      "grad_norm": 0.044171440571531,
      "learning_rate": 6.395383302873104e-05,
      "loss": 0.8405,
      "step": 179320
    },
    {
      "epoch": 3.148404993065187,
      "grad_norm": 0.05020234400837109,
      "learning_rate": 6.394349925438597e-05,
      "loss": 0.8436,
      "step": 179330
    },
    {
      "epoch": 3.1485805579451887,
      "grad_norm": 0.05161970981505903,
      "learning_rate": 6.393316594955382e-05,
      "loss": 0.8412,
      "step": 179340
    },
    {
      "epoch": 3.14875612282519,
      "grad_norm": 0.0547235188590037,
      "learning_rate": 6.39228331143656e-05,
      "loss": 0.8465,
      "step": 179350
    },
    {
      "epoch": 3.1489316877051916,
      "grad_norm": 0.04889940475593968,
      "learning_rate": 6.391250074895217e-05,
      "loss": 0.8379,
      "step": 179360
    },
    {
      "epoch": 3.1491072525851926,
      "grad_norm": 0.05451759078277569,
      "learning_rate": 6.390216885344445e-05,
      "loss": 0.8424,
      "step": 179370
    },
    {
      "epoch": 3.149282817465194,
      "grad_norm": 0.05202645454684516,
      "learning_rate": 6.389183742797339e-05,
      "loss": 0.8357,
      "step": 179380
    },
    {
      "epoch": 3.1494583823451956,
      "grad_norm": 0.05241556934800756,
      "learning_rate": 6.388150647266985e-05,
      "loss": 0.8481,
      "step": 179390
    },
    {
      "epoch": 3.149633947225197,
      "grad_norm": 0.06045961485317918,
      "learning_rate": 6.387117598766476e-05,
      "loss": 0.8378,
      "step": 179400
    },
    {
      "epoch": 3.1498095121051986,
      "grad_norm": 0.06602853791045067,
      "learning_rate": 6.386084597308896e-05,
      "loss": 0.8468,
      "step": 179410
    },
    {
      "epoch": 3.1499850769852,
      "grad_norm": 0.048580587127177804,
      "learning_rate": 6.385051642907337e-05,
      "loss": 0.8441,
      "step": 179420
    },
    {
      "epoch": 3.1501606418652015,
      "grad_norm": 0.048875359271861464,
      "learning_rate": 6.384018735574887e-05,
      "loss": 0.8414,
      "step": 179430
    },
    {
      "epoch": 3.1503362067452025,
      "grad_norm": 0.07266881317918883,
      "learning_rate": 6.382985875324626e-05,
      "loss": 0.8392,
      "step": 179440
    },
    {
      "epoch": 3.150511771625204,
      "grad_norm": 0.0561331223402171,
      "learning_rate": 6.381953062169652e-05,
      "loss": 0.841,
      "step": 179450
    },
    {
      "epoch": 3.1506873365052055,
      "grad_norm": 0.05526824443792002,
      "learning_rate": 6.380920296123045e-05,
      "loss": 0.8399,
      "step": 179460
    },
    {
      "epoch": 3.150862901385207,
      "grad_norm": 0.05029794998574817,
      "learning_rate": 6.379887577197886e-05,
      "loss": 0.8377,
      "step": 179470
    },
    {
      "epoch": 3.1510384662652084,
      "grad_norm": 0.055607824227515965,
      "learning_rate": 6.37885490540727e-05,
      "loss": 0.8365,
      "step": 179480
    },
    {
      "epoch": 3.1512140311452095,
      "grad_norm": 0.09091493454644883,
      "learning_rate": 6.377822280764272e-05,
      "loss": 0.8345,
      "step": 179490
    },
    {
      "epoch": 3.151389596025211,
      "grad_norm": 0.06431042537821893,
      "learning_rate": 6.376789703281981e-05,
      "loss": 0.8329,
      "step": 179500
    },
    {
      "epoch": 3.1515651609052124,
      "grad_norm": 0.047913979775700395,
      "learning_rate": 6.375757172973477e-05,
      "loss": 0.8429,
      "step": 179510
    },
    {
      "epoch": 3.151740725785214,
      "grad_norm": 0.06256844542368858,
      "learning_rate": 6.374724689851845e-05,
      "loss": 0.8384,
      "step": 179520
    },
    {
      "epoch": 3.1519162906652154,
      "grad_norm": 0.04535143417856363,
      "learning_rate": 6.373692253930163e-05,
      "loss": 0.8515,
      "step": 179530
    },
    {
      "epoch": 3.152091855545217,
      "grad_norm": 0.055138511051176946,
      "learning_rate": 6.372659865221511e-05,
      "loss": 0.8393,
      "step": 179540
    },
    {
      "epoch": 3.1522674204252183,
      "grad_norm": 0.05432848103272801,
      "learning_rate": 6.371627523738979e-05,
      "loss": 0.8469,
      "step": 179550
    },
    {
      "epoch": 3.1524429853052194,
      "grad_norm": 0.057076011055602914,
      "learning_rate": 6.370595229495638e-05,
      "loss": 0.8388,
      "step": 179560
    },
    {
      "epoch": 3.152618550185221,
      "grad_norm": 0.05352763153660033,
      "learning_rate": 6.369562982504572e-05,
      "loss": 0.8449,
      "step": 179570
    },
    {
      "epoch": 3.1527941150652223,
      "grad_norm": 0.05382227035672381,
      "learning_rate": 6.368530782778857e-05,
      "loss": 0.847,
      "step": 179580
    },
    {
      "epoch": 3.152969679945224,
      "grad_norm": 0.06792562622809295,
      "learning_rate": 6.367498630331573e-05,
      "loss": 0.8455,
      "step": 179590
    },
    {
      "epoch": 3.1531452448252253,
      "grad_norm": 0.05964740227212663,
      "learning_rate": 6.366466525175794e-05,
      "loss": 0.8371,
      "step": 179600
    },
    {
      "epoch": 3.1533208097052268,
      "grad_norm": 0.04635053777027475,
      "learning_rate": 6.365434467324598e-05,
      "loss": 0.8401,
      "step": 179610
    },
    {
      "epoch": 3.153496374585228,
      "grad_norm": 0.06475959620330492,
      "learning_rate": 6.364402456791067e-05,
      "loss": 0.8376,
      "step": 179620
    },
    {
      "epoch": 3.1536719394652293,
      "grad_norm": 0.05649088765350369,
      "learning_rate": 6.363370493588276e-05,
      "loss": 0.8411,
      "step": 179630
    },
    {
      "epoch": 3.1538475043452308,
      "grad_norm": 0.06404285603878118,
      "learning_rate": 6.362338577729289e-05,
      "loss": 0.8473,
      "step": 179640
    },
    {
      "epoch": 3.1540230692252322,
      "grad_norm": 0.04986700338782178,
      "learning_rate": 6.361306709227193e-05,
      "loss": 0.8378,
      "step": 179650
    },
    {
      "epoch": 3.1541986341052337,
      "grad_norm": 0.06121834183212604,
      "learning_rate": 6.36027488809506e-05,
      "loss": 0.8332,
      "step": 179660
    },
    {
      "epoch": 3.154374198985235,
      "grad_norm": 0.05113835253916613,
      "learning_rate": 6.359243114345957e-05,
      "loss": 0.8361,
      "step": 179670
    },
    {
      "epoch": 3.1545497638652362,
      "grad_norm": 0.06353809392727275,
      "learning_rate": 6.358211387992959e-05,
      "loss": 0.8424,
      "step": 179680
    },
    {
      "epoch": 3.1547253287452377,
      "grad_norm": 0.059267217563659165,
      "learning_rate": 6.357179709049145e-05,
      "loss": 0.8345,
      "step": 179690
    },
    {
      "epoch": 3.154900893625239,
      "grad_norm": 0.05138101033357843,
      "learning_rate": 6.356148077527579e-05,
      "loss": 0.8499,
      "step": 179700
    },
    {
      "epoch": 3.1550764585052407,
      "grad_norm": 0.04755272296793122,
      "learning_rate": 6.355116493441331e-05,
      "loss": 0.8425,
      "step": 179710
    },
    {
      "epoch": 3.155252023385242,
      "grad_norm": 0.049912330167360686,
      "learning_rate": 6.354084956803478e-05,
      "loss": 0.8439,
      "step": 179720
    },
    {
      "epoch": 3.1554275882652436,
      "grad_norm": 0.049771245013577894,
      "learning_rate": 6.353053467627087e-05,
      "loss": 0.837,
      "step": 179730
    },
    {
      "epoch": 3.1556031531452446,
      "grad_norm": 0.05311701426058319,
      "learning_rate": 6.352022025925226e-05,
      "loss": 0.8402,
      "step": 179740
    },
    {
      "epoch": 3.155778718025246,
      "grad_norm": 0.04912449477067914,
      "learning_rate": 6.350990631710965e-05,
      "loss": 0.8317,
      "step": 179750
    },
    {
      "epoch": 3.1559542829052476,
      "grad_norm": 0.05983226396313665,
      "learning_rate": 6.34995928499737e-05,
      "loss": 0.837,
      "step": 179760
    },
    {
      "epoch": 3.156129847785249,
      "grad_norm": 0.061328086678976004,
      "learning_rate": 6.34892798579751e-05,
      "loss": 0.8403,
      "step": 179770
    },
    {
      "epoch": 3.1563054126652506,
      "grad_norm": 0.04998998909930003,
      "learning_rate": 6.347896734124449e-05,
      "loss": 0.8406,
      "step": 179780
    },
    {
      "epoch": 3.156480977545252,
      "grad_norm": 0.04955459625168574,
      "learning_rate": 6.346865529991258e-05,
      "loss": 0.8448,
      "step": 179790
    },
    {
      "epoch": 3.156656542425253,
      "grad_norm": 0.06202081837031431,
      "learning_rate": 6.345834373411004e-05,
      "loss": 0.8465,
      "step": 179800
    },
    {
      "epoch": 3.1568321073052545,
      "grad_norm": 0.07046672853026174,
      "learning_rate": 6.34480326439674e-05,
      "loss": 0.839,
      "step": 179810
    },
    {
      "epoch": 3.157007672185256,
      "grad_norm": 0.04876546825870116,
      "learning_rate": 6.343772202961543e-05,
      "loss": 0.8463,
      "step": 179820
    },
    {
      "epoch": 3.1571832370652575,
      "grad_norm": 0.05353984501547024,
      "learning_rate": 6.342741189118472e-05,
      "loss": 0.8399,
      "step": 179830
    },
    {
      "epoch": 3.157358801945259,
      "grad_norm": 0.06360065985506613,
      "learning_rate": 6.341710222880588e-05,
      "loss": 0.8438,
      "step": 179840
    },
    {
      "epoch": 3.1575343668252605,
      "grad_norm": 0.04984387626493659,
      "learning_rate": 6.340679304260956e-05,
      "loss": 0.8377,
      "step": 179850
    },
    {
      "epoch": 3.1577099317052615,
      "grad_norm": 0.06014650594962761,
      "learning_rate": 6.339648433272642e-05,
      "loss": 0.8463,
      "step": 179860
    },
    {
      "epoch": 3.157885496585263,
      "grad_norm": 0.06570244686012702,
      "learning_rate": 6.3386176099287e-05,
      "loss": 0.8461,
      "step": 179870
    },
    {
      "epoch": 3.1580610614652644,
      "grad_norm": 0.05302085122908752,
      "learning_rate": 6.337586834242189e-05,
      "loss": 0.8352,
      "step": 179880
    },
    {
      "epoch": 3.158236626345266,
      "grad_norm": 0.044332997665070505,
      "learning_rate": 6.33655610622618e-05,
      "loss": 0.8319,
      "step": 179890
    },
    {
      "epoch": 3.1584121912252674,
      "grad_norm": 0.051084365830957615,
      "learning_rate": 6.335525425893725e-05,
      "loss": 0.8443,
      "step": 179900
    },
    {
      "epoch": 3.158587756105269,
      "grad_norm": 0.06715008835809087,
      "learning_rate": 6.334494793257885e-05,
      "loss": 0.8344,
      "step": 179910
    },
    {
      "epoch": 3.15876332098527,
      "grad_norm": 0.05504110657395569,
      "learning_rate": 6.333464208331717e-05,
      "loss": 0.8325,
      "step": 179920
    },
    {
      "epoch": 3.1589388858652714,
      "grad_norm": 0.055172892807333396,
      "learning_rate": 6.332433671128279e-05,
      "loss": 0.8402,
      "step": 179930
    },
    {
      "epoch": 3.159114450745273,
      "grad_norm": 0.04910433914329564,
      "learning_rate": 6.331403181660629e-05,
      "loss": 0.8347,
      "step": 179940
    },
    {
      "epoch": 3.1592900156252743,
      "grad_norm": 0.047886136068208546,
      "learning_rate": 6.330372739941819e-05,
      "loss": 0.8403,
      "step": 179950
    },
    {
      "epoch": 3.159465580505276,
      "grad_norm": 0.045145246765841134,
      "learning_rate": 6.329342345984913e-05,
      "loss": 0.8413,
      "step": 179960
    },
    {
      "epoch": 3.1596411453852773,
      "grad_norm": 0.061582889259715216,
      "learning_rate": 6.328311999802964e-05,
      "loss": 0.8456,
      "step": 179970
    },
    {
      "epoch": 3.1598167102652783,
      "grad_norm": 0.046387992353655975,
      "learning_rate": 6.32728170140902e-05,
      "loss": 0.845,
      "step": 179980
    },
    {
      "epoch": 3.15999227514528,
      "grad_norm": 0.05440440243678406,
      "learning_rate": 6.326251450816142e-05,
      "loss": 0.8352,
      "step": 179990
    },
    {
      "epoch": 3.1601678400252813,
      "grad_norm": 0.047461512671409196,
      "learning_rate": 6.325221248037381e-05,
      "loss": 0.8485,
      "step": 180000
    },
    {
      "epoch": 3.1603434049052828,
      "grad_norm": 0.0518894029551176,
      "learning_rate": 6.324191093085791e-05,
      "loss": 0.8348,
      "step": 180010
    },
    {
      "epoch": 3.1605189697852842,
      "grad_norm": 0.04339280603388512,
      "learning_rate": 6.323160985974424e-05,
      "loss": 0.8421,
      "step": 180020
    },
    {
      "epoch": 3.1606945346652857,
      "grad_norm": 0.06725375397048222,
      "learning_rate": 6.322130926716331e-05,
      "loss": 0.8411,
      "step": 180030
    },
    {
      "epoch": 3.1608700995452867,
      "grad_norm": 0.045162511370106594,
      "learning_rate": 6.321100915324563e-05,
      "loss": 0.8348,
      "step": 180040
    },
    {
      "epoch": 3.1610456644252882,
      "grad_norm": 0.05104739015070901,
      "learning_rate": 6.320070951812167e-05,
      "loss": 0.8403,
      "step": 180050
    },
    {
      "epoch": 3.1612212293052897,
      "grad_norm": 0.05708030093043129,
      "learning_rate": 6.3190410361922e-05,
      "loss": 0.8378,
      "step": 180060
    },
    {
      "epoch": 3.161396794185291,
      "grad_norm": 0.07278597025012958,
      "learning_rate": 6.318011168477708e-05,
      "loss": 0.846,
      "step": 180070
    },
    {
      "epoch": 3.1615723590652927,
      "grad_norm": 0.060195178224976255,
      "learning_rate": 6.31698134868174e-05,
      "loss": 0.8516,
      "step": 180080
    },
    {
      "epoch": 3.161747923945294,
      "grad_norm": 0.052905326410571815,
      "learning_rate": 6.31595157681734e-05,
      "loss": 0.8385,
      "step": 180090
    },
    {
      "epoch": 3.1619234888252956,
      "grad_norm": 0.051875134896985475,
      "learning_rate": 6.314921852897562e-05,
      "loss": 0.8452,
      "step": 180100
    },
    {
      "epoch": 3.1620990537052966,
      "grad_norm": 0.04464887404773847,
      "learning_rate": 6.31389217693545e-05,
      "loss": 0.8347,
      "step": 180110
    },
    {
      "epoch": 3.162274618585298,
      "grad_norm": 0.04557875224433096,
      "learning_rate": 6.312862548944043e-05,
      "loss": 0.8394,
      "step": 180120
    },
    {
      "epoch": 3.1624501834652996,
      "grad_norm": 0.055086357200098426,
      "learning_rate": 6.311832968936399e-05,
      "loss": 0.8398,
      "step": 180130
    },
    {
      "epoch": 3.162625748345301,
      "grad_norm": 0.08712715535457383,
      "learning_rate": 6.31080343692556e-05,
      "loss": 0.8418,
      "step": 180140
    },
    {
      "epoch": 3.1628013132253026,
      "grad_norm": 0.046510288974922424,
      "learning_rate": 6.309773952924563e-05,
      "loss": 0.8345,
      "step": 180150
    },
    {
      "epoch": 3.1629768781053036,
      "grad_norm": 0.045026370182620436,
      "learning_rate": 6.308744516946455e-05,
      "loss": 0.8389,
      "step": 180160
    },
    {
      "epoch": 3.163152442985305,
      "grad_norm": 0.05958439175854949,
      "learning_rate": 6.307715129004285e-05,
      "loss": 0.8383,
      "step": 180170
    },
    {
      "epoch": 3.1633280078653065,
      "grad_norm": 0.058576582909534713,
      "learning_rate": 6.30668578911109e-05,
      "loss": 0.8412,
      "step": 180180
    },
    {
      "epoch": 3.163503572745308,
      "grad_norm": 0.05955069593397579,
      "learning_rate": 6.305656497279911e-05,
      "loss": 0.8506,
      "step": 180190
    },
    {
      "epoch": 3.1636791376253095,
      "grad_norm": 0.04490041517342094,
      "learning_rate": 6.304627253523793e-05,
      "loss": 0.8386,
      "step": 180200
    },
    {
      "epoch": 3.163854702505311,
      "grad_norm": 0.08534248557877444,
      "learning_rate": 6.303598057855775e-05,
      "loss": 0.8441,
      "step": 180210
    },
    {
      "epoch": 3.1640302673853125,
      "grad_norm": 0.05748936368876836,
      "learning_rate": 6.302568910288894e-05,
      "loss": 0.8401,
      "step": 180220
    },
    {
      "epoch": 3.1642058322653135,
      "grad_norm": 0.055979303012549636,
      "learning_rate": 6.301539810836196e-05,
      "loss": 0.8427,
      "step": 180230
    },
    {
      "epoch": 3.164381397145315,
      "grad_norm": 0.04943045738578314,
      "learning_rate": 6.300510759510716e-05,
      "loss": 0.8405,
      "step": 180240
    },
    {
      "epoch": 3.1645569620253164,
      "grad_norm": 0.06674567482082877,
      "learning_rate": 6.299481756325494e-05,
      "loss": 0.8469,
      "step": 180250
    },
    {
      "epoch": 3.164732526905318,
      "grad_norm": 0.048084340339558634,
      "learning_rate": 6.298452801293564e-05,
      "loss": 0.8371,
      "step": 180260
    },
    {
      "epoch": 3.1649080917853194,
      "grad_norm": 0.06796961202722816,
      "learning_rate": 6.297423894427968e-05,
      "loss": 0.845,
      "step": 180270
    },
    {
      "epoch": 3.165083656665321,
      "grad_norm": 0.07201925640858722,
      "learning_rate": 6.296395035741741e-05,
      "loss": 0.8359,
      "step": 180280
    },
    {
      "epoch": 3.165259221545322,
      "grad_norm": 0.0663537025585738,
      "learning_rate": 6.295366225247911e-05,
      "loss": 0.8408,
      "step": 180290
    },
    {
      "epoch": 3.1654347864253234,
      "grad_norm": 0.04644661502187092,
      "learning_rate": 6.294337462959526e-05,
      "loss": 0.8324,
      "step": 180300
    },
    {
      "epoch": 3.165610351305325,
      "grad_norm": 0.0479318049620793,
      "learning_rate": 6.293308748889618e-05,
      "loss": 0.8423,
      "step": 180310
    },
    {
      "epoch": 3.1657859161853263,
      "grad_norm": 0.07049427839563434,
      "learning_rate": 6.292280083051213e-05,
      "loss": 0.8507,
      "step": 180320
    },
    {
      "epoch": 3.165961481065328,
      "grad_norm": 0.06740242628621466,
      "learning_rate": 6.291251465457347e-05,
      "loss": 0.8411,
      "step": 180330
    },
    {
      "epoch": 3.1661370459453293,
      "grad_norm": 0.07570005204121985,
      "learning_rate": 6.290222896121058e-05,
      "loss": 0.8443,
      "step": 180340
    },
    {
      "epoch": 3.1663126108253303,
      "grad_norm": 0.07691770931878841,
      "learning_rate": 6.289194375055374e-05,
      "loss": 0.8544,
      "step": 180350
    },
    {
      "epoch": 3.166488175705332,
      "grad_norm": 0.04996462582109361,
      "learning_rate": 6.288165902273329e-05,
      "loss": 0.8438,
      "step": 180360
    },
    {
      "epoch": 3.1666637405853333,
      "grad_norm": 0.055989922568162265,
      "learning_rate": 6.287137477787951e-05,
      "loss": 0.8448,
      "step": 180370
    },
    {
      "epoch": 3.1668393054653348,
      "grad_norm": 0.05638742606009302,
      "learning_rate": 6.286109101612273e-05,
      "loss": 0.8411,
      "step": 180380
    },
    {
      "epoch": 3.1670148703453362,
      "grad_norm": 0.08108993076959144,
      "learning_rate": 6.28508077375932e-05,
      "loss": 0.8451,
      "step": 180390
    },
    {
      "epoch": 3.1671904352253377,
      "grad_norm": 0.054956135059987865,
      "learning_rate": 6.284052494242128e-05,
      "loss": 0.8353,
      "step": 180400
    },
    {
      "epoch": 3.1673660001053388,
      "grad_norm": 0.04881250876449361,
      "learning_rate": 6.283024263073723e-05,
      "loss": 0.8416,
      "step": 180410
    },
    {
      "epoch": 3.1675415649853402,
      "grad_norm": 0.07126810369491665,
      "learning_rate": 6.281996080267132e-05,
      "loss": 0.8399,
      "step": 180420
    },
    {
      "epoch": 3.1677171298653417,
      "grad_norm": 0.04225393113233018,
      "learning_rate": 6.28096794583538e-05,
      "loss": 0.8324,
      "step": 180430
    },
    {
      "epoch": 3.167892694745343,
      "grad_norm": 0.06385307221471702,
      "learning_rate": 6.279939859791499e-05,
      "loss": 0.8367,
      "step": 180440
    },
    {
      "epoch": 3.1680682596253447,
      "grad_norm": 0.05229878924282728,
      "learning_rate": 6.278911822148512e-05,
      "loss": 0.8378,
      "step": 180450
    },
    {
      "epoch": 3.168243824505346,
      "grad_norm": 0.057668196536324816,
      "learning_rate": 6.277883832919438e-05,
      "loss": 0.845,
      "step": 180460
    },
    {
      "epoch": 3.168419389385347,
      "grad_norm": 0.0630826884096026,
      "learning_rate": 6.276855892117316e-05,
      "loss": 0.8366,
      "step": 180470
    },
    {
      "epoch": 3.1685949542653487,
      "grad_norm": 0.061361750328243723,
      "learning_rate": 6.275827999755162e-05,
      "loss": 0.8399,
      "step": 180480
    },
    {
      "epoch": 3.16877051914535,
      "grad_norm": 0.05308782781445419,
      "learning_rate": 6.274800155845999e-05,
      "loss": 0.8431,
      "step": 180490
    },
    {
      "epoch": 3.1689460840253516,
      "grad_norm": 0.04900390663349413,
      "learning_rate": 6.273772360402848e-05,
      "loss": 0.8338,
      "step": 180500
    },
    {
      "epoch": 3.169121648905353,
      "grad_norm": 0.0551037370330427,
      "learning_rate": 6.272744613438737e-05,
      "loss": 0.8502,
      "step": 180510
    },
    {
      "epoch": 3.1692972137853546,
      "grad_norm": 0.04388734707879429,
      "learning_rate": 6.271716914966685e-05,
      "loss": 0.8461,
      "step": 180520
    },
    {
      "epoch": 3.1694727786653556,
      "grad_norm": 0.04382186446396339,
      "learning_rate": 6.270689264999711e-05,
      "loss": 0.8473,
      "step": 180530
    },
    {
      "epoch": 3.169648343545357,
      "grad_norm": 0.05664815920304455,
      "learning_rate": 6.269661663550843e-05,
      "loss": 0.8491,
      "step": 180540
    },
    {
      "epoch": 3.1698239084253585,
      "grad_norm": 0.04288222525525371,
      "learning_rate": 6.268634110633094e-05,
      "loss": 0.8424,
      "step": 180550
    },
    {
      "epoch": 3.16999947330536,
      "grad_norm": 0.046780190313187,
      "learning_rate": 6.26760660625948e-05,
      "loss": 0.842,
      "step": 180560
    },
    {
      "epoch": 3.1701750381853615,
      "grad_norm": 0.07281545254013368,
      "learning_rate": 6.26657915044303e-05,
      "loss": 0.8398,
      "step": 180570
    },
    {
      "epoch": 3.170350603065363,
      "grad_norm": 0.07344323919408062,
      "learning_rate": 6.265551743196756e-05,
      "loss": 0.8369,
      "step": 180580
    },
    {
      "epoch": 3.170526167945364,
      "grad_norm": 0.05869499282737324,
      "learning_rate": 6.264524384533678e-05,
      "loss": 0.8351,
      "step": 180590
    },
    {
      "epoch": 3.1707017328253655,
      "grad_norm": 0.06416323156103834,
      "learning_rate": 6.263497074466806e-05,
      "loss": 0.8417,
      "step": 180600
    },
    {
      "epoch": 3.170877297705367,
      "grad_norm": 0.06510864350223267,
      "learning_rate": 6.262469813009166e-05,
      "loss": 0.8398,
      "step": 180610
    },
    {
      "epoch": 3.1710528625853684,
      "grad_norm": 0.07189241475950726,
      "learning_rate": 6.261442600173769e-05,
      "loss": 0.8489,
      "step": 180620
    },
    {
      "epoch": 3.17122842746537,
      "grad_norm": 0.04973017486028625,
      "learning_rate": 6.260415435973624e-05,
      "loss": 0.849,
      "step": 180630
    },
    {
      "epoch": 3.1714039923453714,
      "grad_norm": 0.06581163487127424,
      "learning_rate": 6.259388320421757e-05,
      "loss": 0.8445,
      "step": 180640
    },
    {
      "epoch": 3.1715795572253724,
      "grad_norm": 0.04687444186734894,
      "learning_rate": 6.258361253531177e-05,
      "loss": 0.8456,
      "step": 180650
    },
    {
      "epoch": 3.171755122105374,
      "grad_norm": 0.044858954594748736,
      "learning_rate": 6.257334235314895e-05,
      "loss": 0.8408,
      "step": 180660
    },
    {
      "epoch": 3.1719306869853754,
      "grad_norm": 0.04133317007295835,
      "learning_rate": 6.25630726578592e-05,
      "loss": 0.8467,
      "step": 180670
    },
    {
      "epoch": 3.172106251865377,
      "grad_norm": 0.054479477197170736,
      "learning_rate": 6.255280344957274e-05,
      "loss": 0.8385,
      "step": 180680
    },
    {
      "epoch": 3.1722818167453783,
      "grad_norm": 0.047565039899928184,
      "learning_rate": 6.254253472841961e-05,
      "loss": 0.8467,
      "step": 180690
    },
    {
      "epoch": 3.17245738162538,
      "grad_norm": 0.05435781419475874,
      "learning_rate": 6.253226649452992e-05,
      "loss": 0.8383,
      "step": 180700
    },
    {
      "epoch": 3.172632946505381,
      "grad_norm": 0.06244136455851963,
      "learning_rate": 6.25219987480338e-05,
      "loss": 0.8414,
      "step": 180710
    },
    {
      "epoch": 3.1728085113853823,
      "grad_norm": 0.04418778103169791,
      "learning_rate": 6.251173148906135e-05,
      "loss": 0.8432,
      "step": 180720
    },
    {
      "epoch": 3.172984076265384,
      "grad_norm": 0.06006616949020596,
      "learning_rate": 6.250146471774258e-05,
      "loss": 0.8401,
      "step": 180730
    },
    {
      "epoch": 3.1731596411453853,
      "grad_norm": 0.07924975495024111,
      "learning_rate": 6.249119843420767e-05,
      "loss": 0.8335,
      "step": 180740
    },
    {
      "epoch": 3.1733352060253868,
      "grad_norm": 0.07942122586671568,
      "learning_rate": 6.248093263858668e-05,
      "loss": 0.8433,
      "step": 180750
    },
    {
      "epoch": 3.1735107709053882,
      "grad_norm": 0.053563114130436414,
      "learning_rate": 6.247066733100963e-05,
      "loss": 0.8394,
      "step": 180760
    },
    {
      "epoch": 3.1736863357853897,
      "grad_norm": 0.054344173755844036,
      "learning_rate": 6.24604025116066e-05,
      "loss": 0.8444,
      "step": 180770
    },
    {
      "epoch": 3.1738619006653908,
      "grad_norm": 0.043487309515126554,
      "learning_rate": 6.245013818050767e-05,
      "loss": 0.8348,
      "step": 180780
    },
    {
      "epoch": 3.1740374655453922,
      "grad_norm": 0.05650546979678455,
      "learning_rate": 6.243987433784289e-05,
      "loss": 0.8438,
      "step": 180790
    },
    {
      "epoch": 3.1742130304253937,
      "grad_norm": 0.055122870902469964,
      "learning_rate": 6.242961098374223e-05,
      "loss": 0.8379,
      "step": 180800
    },
    {
      "epoch": 3.174388595305395,
      "grad_norm": 0.053771277268420216,
      "learning_rate": 6.241934811833585e-05,
      "loss": 0.8435,
      "step": 180810
    },
    {
      "epoch": 3.1745641601853967,
      "grad_norm": 0.05363921449183982,
      "learning_rate": 6.240908574175375e-05,
      "loss": 0.8409,
      "step": 180820
    },
    {
      "epoch": 3.1747397250653977,
      "grad_norm": 0.04313226840719622,
      "learning_rate": 6.239882385412589e-05,
      "loss": 0.8418,
      "step": 180830
    },
    {
      "epoch": 3.174915289945399,
      "grad_norm": 0.06616985742963215,
      "learning_rate": 6.238856245558229e-05,
      "loss": 0.8451,
      "step": 180840
    },
    {
      "epoch": 3.1750908548254007,
      "grad_norm": 0.048913073128427784,
      "learning_rate": 6.237830154625305e-05,
      "loss": 0.8416,
      "step": 180850
    },
    {
      "epoch": 3.175266419705402,
      "grad_norm": 0.056263092251537146,
      "learning_rate": 6.236804112626813e-05,
      "loss": 0.8447,
      "step": 180860
    },
    {
      "epoch": 3.1754419845854036,
      "grad_norm": 0.053500813372265095,
      "learning_rate": 6.235778119575752e-05,
      "loss": 0.8439,
      "step": 180870
    },
    {
      "epoch": 3.175617549465405,
      "grad_norm": 0.052779180041148975,
      "learning_rate": 6.234752175485123e-05,
      "loss": 0.8414,
      "step": 180880
    },
    {
      "epoch": 3.1757931143454066,
      "grad_norm": 0.050666317095893025,
      "learning_rate": 6.233726280367927e-05,
      "loss": 0.8447,
      "step": 180890
    },
    {
      "epoch": 3.1759686792254076,
      "grad_norm": 0.06156017355581826,
      "learning_rate": 6.232700434237156e-05,
      "loss": 0.8498,
      "step": 180900
    },
    {
      "epoch": 3.176144244105409,
      "grad_norm": 0.05133234558526922,
      "learning_rate": 6.231674637105814e-05,
      "loss": 0.8377,
      "step": 180910
    },
    {
      "epoch": 3.1763198089854106,
      "grad_norm": 0.06374287130573988,
      "learning_rate": 6.230648888986896e-05,
      "loss": 0.8386,
      "step": 180920
    },
    {
      "epoch": 3.176495373865412,
      "grad_norm": 0.06778422070349258,
      "learning_rate": 6.229623189893399e-05,
      "loss": 0.8408,
      "step": 180930
    },
    {
      "epoch": 3.1766709387454135,
      "grad_norm": 0.05135380738206118,
      "learning_rate": 6.228597539838315e-05,
      "loss": 0.8416,
      "step": 180940
    },
    {
      "epoch": 3.1768465036254145,
      "grad_norm": 0.07703661209523938,
      "learning_rate": 6.227571938834646e-05,
      "loss": 0.8443,
      "step": 180950
    },
    {
      "epoch": 3.177022068505416,
      "grad_norm": 0.07470155358029452,
      "learning_rate": 6.22654638689538e-05,
      "loss": 0.8431,
      "step": 180960
    },
    {
      "epoch": 3.1771976333854175,
      "grad_norm": 0.06262835823580958,
      "learning_rate": 6.225520884033509e-05,
      "loss": 0.834,
      "step": 180970
    },
    {
      "epoch": 3.177373198265419,
      "grad_norm": 0.05073348026579183,
      "learning_rate": 6.224495430262039e-05,
      "loss": 0.8414,
      "step": 180980
    },
    {
      "epoch": 3.1775487631454205,
      "grad_norm": 0.05446580382597247,
      "learning_rate": 6.223470025593953e-05,
      "loss": 0.8457,
      "step": 180990
    },
    {
      "epoch": 3.177724328025422,
      "grad_norm": 0.058848903961272146,
      "learning_rate": 6.222444670042244e-05,
      "loss": 0.8454,
      "step": 181000
    },
    {
      "epoch": 3.1778998929054234,
      "grad_norm": 0.054175683436700726,
      "learning_rate": 6.221419363619902e-05,
      "loss": 0.8434,
      "step": 181010
    },
    {
      "epoch": 3.1780754577854244,
      "grad_norm": 0.05936829474541057,
      "learning_rate": 6.22039410633992e-05,
      "loss": 0.839,
      "step": 181020
    },
    {
      "epoch": 3.178251022665426,
      "grad_norm": 0.055813825163259966,
      "learning_rate": 6.219368898215291e-05,
      "loss": 0.8453,
      "step": 181030
    },
    {
      "epoch": 3.1784265875454274,
      "grad_norm": 0.08023453835282786,
      "learning_rate": 6.218343739259e-05,
      "loss": 0.842,
      "step": 181040
    },
    {
      "epoch": 3.178602152425429,
      "grad_norm": 0.05468409611556574,
      "learning_rate": 6.21731862948404e-05,
      "loss": 0.8421,
      "step": 181050
    },
    {
      "epoch": 3.1787777173054304,
      "grad_norm": 0.06196587042531229,
      "learning_rate": 6.216293568903395e-05,
      "loss": 0.8411,
      "step": 181060
    },
    {
      "epoch": 3.178953282185432,
      "grad_norm": 0.05076389099872235,
      "learning_rate": 6.215268557530052e-05,
      "loss": 0.8376,
      "step": 181070
    },
    {
      "epoch": 3.179128847065433,
      "grad_norm": 0.059748366706869935,
      "learning_rate": 6.214243595377003e-05,
      "loss": 0.8509,
      "step": 181080
    },
    {
      "epoch": 3.1793044119454343,
      "grad_norm": 0.07886641774978191,
      "learning_rate": 6.213218682457235e-05,
      "loss": 0.8377,
      "step": 181090
    },
    {
      "epoch": 3.179479976825436,
      "grad_norm": 0.05199676916923831,
      "learning_rate": 6.21219381878373e-05,
      "loss": 0.845,
      "step": 181100
    },
    {
      "epoch": 3.1796555417054373,
      "grad_norm": 0.04875496093009314,
      "learning_rate": 6.211169004369474e-05,
      "loss": 0.8428,
      "step": 181110
    },
    {
      "epoch": 3.1798311065854388,
      "grad_norm": 0.05649105908492239,
      "learning_rate": 6.210144239227451e-05,
      "loss": 0.8413,
      "step": 181120
    },
    {
      "epoch": 3.1800066714654402,
      "grad_norm": 0.050538494709163204,
      "learning_rate": 6.209119523370648e-05,
      "loss": 0.8395,
      "step": 181130
    },
    {
      "epoch": 3.1801822363454413,
      "grad_norm": 0.05918713127219318,
      "learning_rate": 6.208094856812043e-05,
      "loss": 0.8414,
      "step": 181140
    },
    {
      "epoch": 3.1803578012254428,
      "grad_norm": 0.06255356176161767,
      "learning_rate": 6.207070239564625e-05,
      "loss": 0.8329,
      "step": 181150
    },
    {
      "epoch": 3.1805333661054442,
      "grad_norm": 0.043275783748467925,
      "learning_rate": 6.206045671641374e-05,
      "loss": 0.8514,
      "step": 181160
    },
    {
      "epoch": 3.1807089309854457,
      "grad_norm": 0.06918787025699287,
      "learning_rate": 6.20502115305527e-05,
      "loss": 0.8384,
      "step": 181170
    },
    {
      "epoch": 3.180884495865447,
      "grad_norm": 0.05777847693186216,
      "learning_rate": 6.20399668381929e-05,
      "loss": 0.8488,
      "step": 181180
    },
    {
      "epoch": 3.1810600607454487,
      "grad_norm": 0.05727158584837803,
      "learning_rate": 6.202972263946423e-05,
      "loss": 0.8436,
      "step": 181190
    },
    {
      "epoch": 3.1812356256254497,
      "grad_norm": 0.058678683683133115,
      "learning_rate": 6.201947893449644e-05,
      "loss": 0.8524,
      "step": 181200
    },
    {
      "epoch": 3.181411190505451,
      "grad_norm": 0.042658535986953315,
      "learning_rate": 6.20092357234193e-05,
      "loss": 0.8379,
      "step": 181210
    },
    {
      "epoch": 3.1815867553854527,
      "grad_norm": 0.0634597026273684,
      "learning_rate": 6.199899300636266e-05,
      "loss": 0.8485,
      "step": 181220
    },
    {
      "epoch": 3.181762320265454,
      "grad_norm": 0.04315030626162649,
      "learning_rate": 6.198875078345621e-05,
      "loss": 0.844,
      "step": 181230
    },
    {
      "epoch": 3.1819378851454556,
      "grad_norm": 0.05728204171799419,
      "learning_rate": 6.197850905482976e-05,
      "loss": 0.8391,
      "step": 181240
    },
    {
      "epoch": 3.182113450025457,
      "grad_norm": 0.05797562567173369,
      "learning_rate": 6.196826782061309e-05,
      "loss": 0.8397,
      "step": 181250
    },
    {
      "epoch": 3.182289014905458,
      "grad_norm": 0.04276410301245393,
      "learning_rate": 6.195802708093598e-05,
      "loss": 0.84,
      "step": 181260
    },
    {
      "epoch": 3.1824645797854596,
      "grad_norm": 0.0788013561071987,
      "learning_rate": 6.194778683592814e-05,
      "loss": 0.8392,
      "step": 181270
    },
    {
      "epoch": 3.182640144665461,
      "grad_norm": 0.04969994692136077,
      "learning_rate": 6.193754708571928e-05,
      "loss": 0.8401,
      "step": 181280
    },
    {
      "epoch": 3.1828157095454626,
      "grad_norm": 0.05729152470400016,
      "learning_rate": 6.192730783043923e-05,
      "loss": 0.8465,
      "step": 181290
    },
    {
      "epoch": 3.182991274425464,
      "grad_norm": 0.05033735311692828,
      "learning_rate": 6.191706907021765e-05,
      "loss": 0.8374,
      "step": 181300
    },
    {
      "epoch": 3.1831668393054655,
      "grad_norm": 0.04839769019867525,
      "learning_rate": 6.190683080518428e-05,
      "loss": 0.8459,
      "step": 181310
    },
    {
      "epoch": 3.1833424041854665,
      "grad_norm": 0.054295448713569824,
      "learning_rate": 6.18965930354689e-05,
      "loss": 0.8326,
      "step": 181320
    },
    {
      "epoch": 3.183517969065468,
      "grad_norm": 0.05677980282480508,
      "learning_rate": 6.18863557612012e-05,
      "loss": 0.8384,
      "step": 181330
    },
    {
      "epoch": 3.1836935339454695,
      "grad_norm": 0.041700215182891444,
      "learning_rate": 6.187611898251082e-05,
      "loss": 0.838,
      "step": 181340
    },
    {
      "epoch": 3.183869098825471,
      "grad_norm": 0.04233545720250186,
      "learning_rate": 6.18658826995275e-05,
      "loss": 0.8398,
      "step": 181350
    },
    {
      "epoch": 3.1840446637054725,
      "grad_norm": 0.06818250642331745,
      "learning_rate": 6.185564691238099e-05,
      "loss": 0.8424,
      "step": 181360
    },
    {
      "epoch": 3.184220228585474,
      "grad_norm": 0.043878073553094706,
      "learning_rate": 6.184541162120091e-05,
      "loss": 0.8532,
      "step": 181370
    },
    {
      "epoch": 3.184395793465475,
      "grad_norm": 0.052177287958350034,
      "learning_rate": 6.183517682611696e-05,
      "loss": 0.8512,
      "step": 181380
    },
    {
      "epoch": 3.1845713583454764,
      "grad_norm": 0.04728269916327154,
      "learning_rate": 6.182494252725885e-05,
      "loss": 0.8427,
      "step": 181390
    },
    {
      "epoch": 3.184746923225478,
      "grad_norm": 0.051861131145414414,
      "learning_rate": 6.181470872475622e-05,
      "loss": 0.8382,
      "step": 181400
    },
    {
      "epoch": 3.1849224881054794,
      "grad_norm": 0.05786377192256197,
      "learning_rate": 6.18044754187387e-05,
      "loss": 0.8425,
      "step": 181410
    },
    {
      "epoch": 3.185098052985481,
      "grad_norm": 0.044645338628613966,
      "learning_rate": 6.179424260933603e-05,
      "loss": 0.8362,
      "step": 181420
    },
    {
      "epoch": 3.1852736178654824,
      "grad_norm": 0.04759075492485572,
      "learning_rate": 6.178401029667783e-05,
      "loss": 0.8363,
      "step": 181430
    },
    {
      "epoch": 3.1854491827454834,
      "grad_norm": 0.0463291058293646,
      "learning_rate": 6.177377848089371e-05,
      "loss": 0.8387,
      "step": 181440
    },
    {
      "epoch": 3.185624747625485,
      "grad_norm": 0.04447494132091603,
      "learning_rate": 6.176354716211335e-05,
      "loss": 0.841,
      "step": 181450
    },
    {
      "epoch": 3.1858003125054863,
      "grad_norm": 0.041093321443565076,
      "learning_rate": 6.175331634046636e-05,
      "loss": 0.8432,
      "step": 181460
    },
    {
      "epoch": 3.185975877385488,
      "grad_norm": 0.05668744216104256,
      "learning_rate": 6.174308601608238e-05,
      "loss": 0.8395,
      "step": 181470
    },
    {
      "epoch": 3.1861514422654893,
      "grad_norm": 0.05554044271270587,
      "learning_rate": 6.173285618909098e-05,
      "loss": 0.8467,
      "step": 181480
    },
    {
      "epoch": 3.1863270071454908,
      "grad_norm": 0.04530003860336921,
      "learning_rate": 6.172262685962187e-05,
      "loss": 0.8486,
      "step": 181490
    },
    {
      "epoch": 3.186502572025492,
      "grad_norm": 0.05836989267378719,
      "learning_rate": 6.171239802780461e-05,
      "loss": 0.8354,
      "step": 181500
    },
    {
      "epoch": 3.1866781369054933,
      "grad_norm": 0.05616014028791638,
      "learning_rate": 6.17021696937688e-05,
      "loss": 0.8403,
      "step": 181510
    },
    {
      "epoch": 3.1868537017854948,
      "grad_norm": 0.060532107929449214,
      "learning_rate": 6.169194185764398e-05,
      "loss": 0.8414,
      "step": 181520
    },
    {
      "epoch": 3.1870292666654962,
      "grad_norm": 0.037465652310939956,
      "learning_rate": 6.16817145195598e-05,
      "loss": 0.8509,
      "step": 181530
    },
    {
      "epoch": 3.1872048315454977,
      "grad_norm": 0.05799962915630295,
      "learning_rate": 6.167148767964588e-05,
      "loss": 0.8458,
      "step": 181540
    },
    {
      "epoch": 3.187380396425499,
      "grad_norm": 0.05883798270195268,
      "learning_rate": 6.166126133803169e-05,
      "loss": 0.8455,
      "step": 181550
    },
    {
      "epoch": 3.1875559613055007,
      "grad_norm": 0.04752194771728433,
      "learning_rate": 6.16510354948469e-05,
      "loss": 0.8483,
      "step": 181560
    },
    {
      "epoch": 3.1877315261855017,
      "grad_norm": 0.06528562022606547,
      "learning_rate": 6.164081015022102e-05,
      "loss": 0.8459,
      "step": 181570
    },
    {
      "epoch": 3.187907091065503,
      "grad_norm": 0.051404185889496634,
      "learning_rate": 6.16305853042836e-05,
      "loss": 0.8324,
      "step": 181580
    },
    {
      "epoch": 3.1880826559455047,
      "grad_norm": 0.048897593165146296,
      "learning_rate": 6.162036095716423e-05,
      "loss": 0.8305,
      "step": 181590
    },
    {
      "epoch": 3.188258220825506,
      "grad_norm": 0.05865969460277963,
      "learning_rate": 6.161013710899244e-05,
      "loss": 0.8414,
      "step": 181600
    },
    {
      "epoch": 3.1884337857055076,
      "grad_norm": 0.05105169867169069,
      "learning_rate": 6.159991375989776e-05,
      "loss": 0.8453,
      "step": 181610
    },
    {
      "epoch": 3.1886093505855087,
      "grad_norm": 0.04476004825346111,
      "learning_rate": 6.158969091000969e-05,
      "loss": 0.8432,
      "step": 181620
    },
    {
      "epoch": 3.18878491546551,
      "grad_norm": 0.07446555843963132,
      "learning_rate": 6.157946855945784e-05,
      "loss": 0.8353,
      "step": 181630
    },
    {
      "epoch": 3.1889604803455116,
      "grad_norm": 0.05236043385505729,
      "learning_rate": 6.156924670837165e-05,
      "loss": 0.8483,
      "step": 181640
    },
    {
      "epoch": 3.189136045225513,
      "grad_norm": 0.05425391370297193,
      "learning_rate": 6.155902535688064e-05,
      "loss": 0.8458,
      "step": 181650
    },
    {
      "epoch": 3.1893116101055146,
      "grad_norm": 0.04968178553945077,
      "learning_rate": 6.154880450511436e-05,
      "loss": 0.8393,
      "step": 181660
    },
    {
      "epoch": 3.189487174985516,
      "grad_norm": 0.045683852443946155,
      "learning_rate": 6.153858415320231e-05,
      "loss": 0.844,
      "step": 181670
    },
    {
      "epoch": 3.1896627398655175,
      "grad_norm": 0.05606401016187325,
      "learning_rate": 6.152836430127396e-05,
      "loss": 0.8391,
      "step": 181680
    },
    {
      "epoch": 3.1898383047455185,
      "grad_norm": 0.06248938447817715,
      "learning_rate": 6.151814494945873e-05,
      "loss": 0.8343,
      "step": 181690
    },
    {
      "epoch": 3.19001386962552,
      "grad_norm": 0.061730276489407224,
      "learning_rate": 6.150792609788624e-05,
      "loss": 0.8373,
      "step": 181700
    },
    {
      "epoch": 3.1901894345055215,
      "grad_norm": 0.04862548794016038,
      "learning_rate": 6.149770774668588e-05,
      "loss": 0.8503,
      "step": 181710
    },
    {
      "epoch": 3.190364999385523,
      "grad_norm": 0.049580303604039526,
      "learning_rate": 6.148748989598712e-05,
      "loss": 0.8367,
      "step": 181720
    },
    {
      "epoch": 3.1905405642655245,
      "grad_norm": 0.050588146382512,
      "learning_rate": 6.147727254591945e-05,
      "loss": 0.8405,
      "step": 181730
    },
    {
      "epoch": 3.190716129145526,
      "grad_norm": 0.05617221704408805,
      "learning_rate": 6.146705569661232e-05,
      "loss": 0.8527,
      "step": 181740
    },
    {
      "epoch": 3.190891694025527,
      "grad_norm": 0.05029204643751995,
      "learning_rate": 6.145683934819512e-05,
      "loss": 0.8413,
      "step": 181750
    },
    {
      "epoch": 3.1910672589055284,
      "grad_norm": 0.05423673687936631,
      "learning_rate": 6.144662350079737e-05,
      "loss": 0.8413,
      "step": 181760
    },
    {
      "epoch": 3.19124282378553,
      "grad_norm": 0.049820229493160506,
      "learning_rate": 6.143640815454852e-05,
      "loss": 0.8421,
      "step": 181770
    },
    {
      "epoch": 3.1914183886655314,
      "grad_norm": 0.052359303424463,
      "learning_rate": 6.142619330957793e-05,
      "loss": 0.8387,
      "step": 181780
    },
    {
      "epoch": 3.191593953545533,
      "grad_norm": 0.05775085976374844,
      "learning_rate": 6.141597896601506e-05,
      "loss": 0.8401,
      "step": 181790
    },
    {
      "epoch": 3.1917695184255344,
      "grad_norm": 0.04730359820484511,
      "learning_rate": 6.140576512398931e-05,
      "loss": 0.8343,
      "step": 181800
    },
    {
      "epoch": 3.1919450833055354,
      "grad_norm": 0.04154236867709806,
      "learning_rate": 6.139555178363012e-05,
      "loss": 0.8425,
      "step": 181810
    },
    {
      "epoch": 3.192120648185537,
      "grad_norm": 0.05140909661338718,
      "learning_rate": 6.138533894506683e-05,
      "loss": 0.8409,
      "step": 181820
    },
    {
      "epoch": 3.1922962130655383,
      "grad_norm": 0.05919962069762185,
      "learning_rate": 6.137512660842894e-05,
      "loss": 0.8393,
      "step": 181830
    },
    {
      "epoch": 3.19247177794554,
      "grad_norm": 0.05825087938017819,
      "learning_rate": 6.136491477384582e-05,
      "loss": 0.8432,
      "step": 181840
    },
    {
      "epoch": 3.1926473428255413,
      "grad_norm": 0.06781563772196901,
      "learning_rate": 6.13547034414468e-05,
      "loss": 0.8372,
      "step": 181850
    },
    {
      "epoch": 3.192822907705543,
      "grad_norm": 0.04596141614422917,
      "learning_rate": 6.134449261136122e-05,
      "loss": 0.8446,
      "step": 181860
    },
    {
      "epoch": 3.192998472585544,
      "grad_norm": 0.05642820560148716,
      "learning_rate": 6.13342822837186e-05,
      "loss": 0.8388,
      "step": 181870
    },
    {
      "epoch": 3.1931740374655453,
      "grad_norm": 0.05614329572333585,
      "learning_rate": 6.132407245864818e-05,
      "loss": 0.8417,
      "step": 181880
    },
    {
      "epoch": 3.1933496023455468,
      "grad_norm": 0.05405060808639001,
      "learning_rate": 6.131386313627938e-05,
      "loss": 0.8404,
      "step": 181890
    },
    {
      "epoch": 3.1935251672255482,
      "grad_norm": 0.05970005965031111,
      "learning_rate": 6.130365431674154e-05,
      "loss": 0.8456,
      "step": 181900
    },
    {
      "epoch": 3.1937007321055497,
      "grad_norm": 0.05468746206602166,
      "learning_rate": 6.129344600016402e-05,
      "loss": 0.8384,
      "step": 181910
    },
    {
      "epoch": 3.193876296985551,
      "grad_norm": 0.053214234079784664,
      "learning_rate": 6.12832381866761e-05,
      "loss": 0.8353,
      "step": 181920
    },
    {
      "epoch": 3.1940518618655522,
      "grad_norm": 0.04791409653456163,
      "learning_rate": 6.12730308764072e-05,
      "loss": 0.8357,
      "step": 181930
    },
    {
      "epoch": 3.1942274267455537,
      "grad_norm": 0.0488873991121105,
      "learning_rate": 6.126282406948664e-05,
      "loss": 0.8425,
      "step": 181940
    },
    {
      "epoch": 3.194402991625555,
      "grad_norm": 0.0666228415934281,
      "learning_rate": 6.12526177660437e-05,
      "loss": 0.8385,
      "step": 181950
    },
    {
      "epoch": 3.1945785565055567,
      "grad_norm": 0.05396864517891219,
      "learning_rate": 6.124241196620768e-05,
      "loss": 0.8304,
      "step": 181960
    },
    {
      "epoch": 3.194754121385558,
      "grad_norm": 0.06881622362483647,
      "learning_rate": 6.123220667010793e-05,
      "loss": 0.8452,
      "step": 181970
    },
    {
      "epoch": 3.1949296862655596,
      "grad_norm": 0.05014823767366758,
      "learning_rate": 6.122200187787377e-05,
      "loss": 0.8393,
      "step": 181980
    },
    {
      "epoch": 3.1951052511455607,
      "grad_norm": 0.054208582113349175,
      "learning_rate": 6.121179758963443e-05,
      "loss": 0.8389,
      "step": 181990
    },
    {
      "epoch": 3.195280816025562,
      "grad_norm": 0.05154122155542282,
      "learning_rate": 6.120159380551927e-05,
      "loss": 0.8475,
      "step": 182000
    },
    {
      "epoch": 3.1954563809055636,
      "grad_norm": 0.05733507048766492,
      "learning_rate": 6.119139052565754e-05,
      "loss": 0.8359,
      "step": 182010
    },
    {
      "epoch": 3.195631945785565,
      "grad_norm": 0.05079278996951382,
      "learning_rate": 6.118118775017853e-05,
      "loss": 0.8424,
      "step": 182020
    },
    {
      "epoch": 3.1958075106655666,
      "grad_norm": 0.05151582199218898,
      "learning_rate": 6.117098547921147e-05,
      "loss": 0.8391,
      "step": 182030
    },
    {
      "epoch": 3.195983075545568,
      "grad_norm": 0.04541303447347507,
      "learning_rate": 6.116078371288568e-05,
      "loss": 0.842,
      "step": 182040
    },
    {
      "epoch": 3.196158640425569,
      "grad_norm": 0.04908470278098667,
      "learning_rate": 6.11505824513304e-05,
      "loss": 0.8345,
      "step": 182050
    },
    {
      "epoch": 3.1963342053055706,
      "grad_norm": 0.047887485280586034,
      "learning_rate": 6.114038169467485e-05,
      "loss": 0.8389,
      "step": 182060
    },
    {
      "epoch": 3.196509770185572,
      "grad_norm": 0.042305080428833276,
      "learning_rate": 6.113018144304832e-05,
      "loss": 0.8412,
      "step": 182070
    },
    {
      "epoch": 3.1966853350655735,
      "grad_norm": 0.07301964348346865,
      "learning_rate": 6.111998169658005e-05,
      "loss": 0.841,
      "step": 182080
    },
    {
      "epoch": 3.196860899945575,
      "grad_norm": 0.044341564099918664,
      "learning_rate": 6.110978245539921e-05,
      "loss": 0.8439,
      "step": 182090
    },
    {
      "epoch": 3.1970364648255765,
      "grad_norm": 0.05649070817801077,
      "learning_rate": 6.10995837196351e-05,
      "loss": 0.8367,
      "step": 182100
    },
    {
      "epoch": 3.1972120297055775,
      "grad_norm": 0.066969187609454,
      "learning_rate": 6.108938548941691e-05,
      "loss": 0.8404,
      "step": 182110
    },
    {
      "epoch": 3.197387594585579,
      "grad_norm": 0.051306924391157895,
      "learning_rate": 6.107918776487388e-05,
      "loss": 0.8441,
      "step": 182120
    },
    {
      "epoch": 3.1975631594655805,
      "grad_norm": 0.053459785027181704,
      "learning_rate": 6.106899054613515e-05,
      "loss": 0.8354,
      "step": 182130
    },
    {
      "epoch": 3.197738724345582,
      "grad_norm": 0.05463038000282751,
      "learning_rate": 6.105879383332999e-05,
      "loss": 0.8444,
      "step": 182140
    },
    {
      "epoch": 3.1979142892255834,
      "grad_norm": 0.059790797376772016,
      "learning_rate": 6.104859762658757e-05,
      "loss": 0.8449,
      "step": 182150
    },
    {
      "epoch": 3.198089854105585,
      "grad_norm": 0.057597385023622896,
      "learning_rate": 6.103840192603702e-05,
      "loss": 0.8516,
      "step": 182160
    },
    {
      "epoch": 3.198265418985586,
      "grad_norm": 0.06062357707518469,
      "learning_rate": 6.1028206731807636e-05,
      "loss": 0.8366,
      "step": 182170
    },
    {
      "epoch": 3.1984409838655874,
      "grad_norm": 0.04586346898616864,
      "learning_rate": 6.101801204402854e-05,
      "loss": 0.8403,
      "step": 182180
    },
    {
      "epoch": 3.198616548745589,
      "grad_norm": 0.0652882973306885,
      "learning_rate": 6.1007817862828874e-05,
      "loss": 0.8356,
      "step": 182190
    },
    {
      "epoch": 3.1987921136255903,
      "grad_norm": 0.062079736957871695,
      "learning_rate": 6.099762418833779e-05,
      "loss": 0.8319,
      "step": 182200
    },
    {
      "epoch": 3.198967678505592,
      "grad_norm": 0.05322169419431621,
      "learning_rate": 6.098743102068451e-05,
      "loss": 0.8438,
      "step": 182210
    },
    {
      "epoch": 3.1991432433855933,
      "grad_norm": 0.0499560578471252,
      "learning_rate": 6.097723835999814e-05,
      "loss": 0.8446,
      "step": 182220
    },
    {
      "epoch": 3.199318808265595,
      "grad_norm": 0.053074478267723726,
      "learning_rate": 6.096704620640781e-05,
      "loss": 0.8488,
      "step": 182230
    },
    {
      "epoch": 3.199494373145596,
      "grad_norm": 0.047905661519592434,
      "learning_rate": 6.09568545600427e-05,
      "loss": 0.85,
      "step": 182240
    },
    {
      "epoch": 3.1996699380255973,
      "grad_norm": 0.05866739886760045,
      "learning_rate": 6.0946663421031925e-05,
      "loss": 0.8384,
      "step": 182250
    },
    {
      "epoch": 3.1998455029055988,
      "grad_norm": 0.04776624895116065,
      "learning_rate": 6.093647278950455e-05,
      "loss": 0.8468,
      "step": 182260
    },
    {
      "epoch": 3.2000210677856002,
      "grad_norm": 0.0611431658065939,
      "learning_rate": 6.0926282665589786e-05,
      "loss": 0.8325,
      "step": 182270
    },
    {
      "epoch": 3.2001966326656017,
      "grad_norm": 0.06563225729700933,
      "learning_rate": 6.09160930494167e-05,
      "loss": 0.8393,
      "step": 182280
    },
    {
      "epoch": 3.2003721975456028,
      "grad_norm": 0.053072022903428044,
      "learning_rate": 6.09059039411144e-05,
      "loss": 0.8401,
      "step": 182290
    },
    {
      "epoch": 3.2005477624256042,
      "grad_norm": 0.060454762034126905,
      "learning_rate": 6.0895715340811955e-05,
      "loss": 0.8404,
      "step": 182300
    },
    {
      "epoch": 3.2007233273056057,
      "grad_norm": 0.05975433565106528,
      "learning_rate": 6.088552724863852e-05,
      "loss": 0.8335,
      "step": 182310
    },
    {
      "epoch": 3.200898892185607,
      "grad_norm": 0.06521592761429809,
      "learning_rate": 6.087533966472312e-05,
      "loss": 0.8357,
      "step": 182320
    },
    {
      "epoch": 3.2010744570656087,
      "grad_norm": 0.05406160841041899,
      "learning_rate": 6.086515258919483e-05,
      "loss": 0.8369,
      "step": 182330
    },
    {
      "epoch": 3.20125002194561,
      "grad_norm": 0.05153329193579625,
      "learning_rate": 6.085496602218278e-05,
      "loss": 0.8466,
      "step": 182340
    },
    {
      "epoch": 3.2014255868256116,
      "grad_norm": 0.04938554590637465,
      "learning_rate": 6.0844779963816036e-05,
      "loss": 0.8424,
      "step": 182350
    },
    {
      "epoch": 3.2016011517056127,
      "grad_norm": 0.09072078701705143,
      "learning_rate": 6.0834594414223586e-05,
      "loss": 0.8491,
      "step": 182360
    },
    {
      "epoch": 3.201776716585614,
      "grad_norm": 0.04972151028487942,
      "learning_rate": 6.0824409373534494e-05,
      "loss": 0.8461,
      "step": 182370
    },
    {
      "epoch": 3.2019522814656156,
      "grad_norm": 0.06257295192642903,
      "learning_rate": 6.081422484187789e-05,
      "loss": 0.8404,
      "step": 182380
    },
    {
      "epoch": 3.202127846345617,
      "grad_norm": 0.060216047596058614,
      "learning_rate": 6.0804040819382734e-05,
      "loss": 0.8373,
      "step": 182390
    },
    {
      "epoch": 3.2023034112256186,
      "grad_norm": 0.05595882019254314,
      "learning_rate": 6.079385730617808e-05,
      "loss": 0.852,
      "step": 182400
    },
    {
      "epoch": 3.2024789761056196,
      "grad_norm": 0.0538954341546527,
      "learning_rate": 6.0783674302392965e-05,
      "loss": 0.8438,
      "step": 182410
    },
    {
      "epoch": 3.202654540985621,
      "grad_norm": 0.06123958631801769,
      "learning_rate": 6.0773491808156424e-05,
      "loss": 0.8398,
      "step": 182420
    },
    {
      "epoch": 3.2028301058656226,
      "grad_norm": 0.047406358809918454,
      "learning_rate": 6.07633098235974e-05,
      "loss": 0.8404,
      "step": 182430
    },
    {
      "epoch": 3.203005670745624,
      "grad_norm": 0.05820301410760285,
      "learning_rate": 6.075312834884499e-05,
      "loss": 0.8436,
      "step": 182440
    },
    {
      "epoch": 3.2031812356256255,
      "grad_norm": 0.06579987289206528,
      "learning_rate": 6.074294738402815e-05,
      "loss": 0.84,
      "step": 182450
    },
    {
      "epoch": 3.203356800505627,
      "grad_norm": 0.05881335184028285,
      "learning_rate": 6.0732766929275905e-05,
      "loss": 0.8403,
      "step": 182460
    },
    {
      "epoch": 3.2035323653856285,
      "grad_norm": 0.05475786496535445,
      "learning_rate": 6.072258698471718e-05,
      "loss": 0.8399,
      "step": 182470
    },
    {
      "epoch": 3.2037079302656295,
      "grad_norm": 0.06964063776722025,
      "learning_rate": 6.071240755048102e-05,
      "loss": 0.8377,
      "step": 182480
    },
    {
      "epoch": 3.203883495145631,
      "grad_norm": 0.043912630640128826,
      "learning_rate": 6.070222862669639e-05,
      "loss": 0.8394,
      "step": 182490
    },
    {
      "epoch": 3.2040590600256325,
      "grad_norm": 0.043630814570068434,
      "learning_rate": 6.06920502134922e-05,
      "loss": 0.8443,
      "step": 182500
    },
    {
      "epoch": 3.204234624905634,
      "grad_norm": 0.05676365890298951,
      "learning_rate": 6.068187231099749e-05,
      "loss": 0.841,
      "step": 182510
    },
    {
      "epoch": 3.2044101897856354,
      "grad_norm": 0.0524725399273576,
      "learning_rate": 6.067169491934122e-05,
      "loss": 0.8323,
      "step": 182520
    },
    {
      "epoch": 3.204585754665637,
      "grad_norm": 0.05903638234577931,
      "learning_rate": 6.0661518038652266e-05,
      "loss": 0.8402,
      "step": 182530
    },
    {
      "epoch": 3.204761319545638,
      "grad_norm": 0.049588624321469314,
      "learning_rate": 6.065134166905958e-05,
      "loss": 0.8413,
      "step": 182540
    },
    {
      "epoch": 3.2049368844256394,
      "grad_norm": 0.05972539313008733,
      "learning_rate": 6.064116581069216e-05,
      "loss": 0.8393,
      "step": 182550
    },
    {
      "epoch": 3.205112449305641,
      "grad_norm": 0.05256699013401026,
      "learning_rate": 6.06309904636789e-05,
      "loss": 0.8408,
      "step": 182560
    },
    {
      "epoch": 3.2052880141856424,
      "grad_norm": 0.052990688414644445,
      "learning_rate": 6.062081562814873e-05,
      "loss": 0.8377,
      "step": 182570
    },
    {
      "epoch": 3.205463579065644,
      "grad_norm": 0.05338790982220299,
      "learning_rate": 6.061064130423056e-05,
      "loss": 0.8476,
      "step": 182580
    },
    {
      "epoch": 3.2056391439456453,
      "grad_norm": 0.05381299698364561,
      "learning_rate": 6.060046749205334e-05,
      "loss": 0.8421,
      "step": 182590
    },
    {
      "epoch": 3.2058147088256463,
      "grad_norm": 0.054385987407072296,
      "learning_rate": 6.059029419174587e-05,
      "loss": 0.8372,
      "step": 182600
    },
    {
      "epoch": 3.205990273705648,
      "grad_norm": 0.06776046142586827,
      "learning_rate": 6.058012140343716e-05,
      "loss": 0.8432,
      "step": 182610
    },
    {
      "epoch": 3.2061658385856493,
      "grad_norm": 0.05906823183334407,
      "learning_rate": 6.056994912725605e-05,
      "loss": 0.8364,
      "step": 182620
    },
    {
      "epoch": 3.2063414034656508,
      "grad_norm": 0.04943058099345212,
      "learning_rate": 6.055977736333146e-05,
      "loss": 0.8384,
      "step": 182630
    },
    {
      "epoch": 3.2065169683456523,
      "grad_norm": 0.04654087457060079,
      "learning_rate": 6.054960611179221e-05,
      "loss": 0.8386,
      "step": 182640
    },
    {
      "epoch": 3.2066925332256537,
      "grad_norm": 0.056219770291054516,
      "learning_rate": 6.0539435372767214e-05,
      "loss": 0.8433,
      "step": 182650
    },
    {
      "epoch": 3.2068680981056548,
      "grad_norm": 0.05576134664981089,
      "learning_rate": 6.052926514638534e-05,
      "loss": 0.8489,
      "step": 182660
    },
    {
      "epoch": 3.2070436629856562,
      "grad_norm": 0.08577693411709156,
      "learning_rate": 6.051909543277537e-05,
      "loss": 0.8409,
      "step": 182670
    },
    {
      "epoch": 3.2072192278656577,
      "grad_norm": 0.04872839184403289,
      "learning_rate": 6.0508926232066276e-05,
      "loss": 0.8401,
      "step": 182680
    },
    {
      "epoch": 3.207394792745659,
      "grad_norm": 0.05544485125890247,
      "learning_rate": 6.049875754438686e-05,
      "loss": 0.8466,
      "step": 182690
    },
    {
      "epoch": 3.2075703576256607,
      "grad_norm": 0.061754028340142646,
      "learning_rate": 6.048858936986595e-05,
      "loss": 0.8435,
      "step": 182700
    },
    {
      "epoch": 3.207745922505662,
      "grad_norm": 0.057559244784733676,
      "learning_rate": 6.0478421708632315e-05,
      "loss": 0.8404,
      "step": 182710
    },
    {
      "epoch": 3.207921487385663,
      "grad_norm": 0.05004118082966457,
      "learning_rate": 6.046825456081488e-05,
      "loss": 0.8488,
      "step": 182720
    },
    {
      "epoch": 3.2080970522656647,
      "grad_norm": 0.06973176060086556,
      "learning_rate": 6.045808792654244e-05,
      "loss": 0.8391,
      "step": 182730
    },
    {
      "epoch": 3.208272617145666,
      "grad_norm": 0.055168568862335704,
      "learning_rate": 6.0447921805943764e-05,
      "loss": 0.8422,
      "step": 182740
    },
    {
      "epoch": 3.2084481820256676,
      "grad_norm": 0.04871311466032282,
      "learning_rate": 6.0437756199147724e-05,
      "loss": 0.8386,
      "step": 182750
    },
    {
      "epoch": 3.208623746905669,
      "grad_norm": 0.0719391803165989,
      "learning_rate": 6.0427591106283064e-05,
      "loss": 0.8387,
      "step": 182760
    },
    {
      "epoch": 3.2087993117856706,
      "grad_norm": 0.05509128335042559,
      "learning_rate": 6.041742652747858e-05,
      "loss": 0.8384,
      "step": 182770
    },
    {
      "epoch": 3.2089748766656716,
      "grad_norm": 0.05394860983339904,
      "learning_rate": 6.040726246286312e-05,
      "loss": 0.8461,
      "step": 182780
    },
    {
      "epoch": 3.209150441545673,
      "grad_norm": 0.048007156265003924,
      "learning_rate": 6.039709891256542e-05,
      "loss": 0.842,
      "step": 182790
    },
    {
      "epoch": 3.2093260064256746,
      "grad_norm": 0.04644788340640933,
      "learning_rate": 6.0386935876714254e-05,
      "loss": 0.842,
      "step": 182800
    },
    {
      "epoch": 3.209501571305676,
      "grad_norm": 0.04545644412955526,
      "learning_rate": 6.0376773355438376e-05,
      "loss": 0.852,
      "step": 182810
    },
    {
      "epoch": 3.2096771361856775,
      "grad_norm": 0.1168902964683629,
      "learning_rate": 6.036661134886659e-05,
      "loss": 0.8403,
      "step": 182820
    },
    {
      "epoch": 3.209852701065679,
      "grad_norm": 0.053658159274901106,
      "learning_rate": 6.035644985712762e-05,
      "loss": 0.839,
      "step": 182830
    },
    {
      "epoch": 3.21002826594568,
      "grad_norm": 0.06305368560410879,
      "learning_rate": 6.034628888035018e-05,
      "loss": 0.8391,
      "step": 182840
    },
    {
      "epoch": 3.2102038308256815,
      "grad_norm": 0.05813728050827306,
      "learning_rate": 6.0336128418663086e-05,
      "loss": 0.8423,
      "step": 182850
    },
    {
      "epoch": 3.210379395705683,
      "grad_norm": 0.06015076780971664,
      "learning_rate": 6.032596847219506e-05,
      "loss": 0.8316,
      "step": 182860
    },
    {
      "epoch": 3.2105549605856845,
      "grad_norm": 0.0591441269570068,
      "learning_rate": 6.03158090410748e-05,
      "loss": 0.8418,
      "step": 182870
    },
    {
      "epoch": 3.210730525465686,
      "grad_norm": 0.06019756228224484,
      "learning_rate": 6.030565012543099e-05,
      "loss": 0.8421,
      "step": 182880
    },
    {
      "epoch": 3.2109060903456874,
      "grad_norm": 0.05665530678958933,
      "learning_rate": 6.029549172539243e-05,
      "loss": 0.8417,
      "step": 182890
    },
    {
      "epoch": 3.2110816552256884,
      "grad_norm": 0.04993936186851527,
      "learning_rate": 6.02853338410878e-05,
      "loss": 0.835,
      "step": 182900
    },
    {
      "epoch": 3.21125722010569,
      "grad_norm": 0.04339131722134552,
      "learning_rate": 6.027517647264577e-05,
      "loss": 0.8386,
      "step": 182910
    },
    {
      "epoch": 3.2114327849856914,
      "grad_norm": 0.05487788614460194,
      "learning_rate": 6.026501962019506e-05,
      "loss": 0.8413,
      "step": 182920
    },
    {
      "epoch": 3.211608349865693,
      "grad_norm": 0.07762923751623055,
      "learning_rate": 6.025486328386437e-05,
      "loss": 0.8412,
      "step": 182930
    },
    {
      "epoch": 3.2117839147456944,
      "grad_norm": 0.07884404565607263,
      "learning_rate": 6.024470746378235e-05,
      "loss": 0.8438,
      "step": 182940
    },
    {
      "epoch": 3.211959479625696,
      "grad_norm": 0.06119657598579873,
      "learning_rate": 6.023455216007772e-05,
      "loss": 0.8441,
      "step": 182950
    },
    {
      "epoch": 3.212135044505697,
      "grad_norm": 0.05136704222626834,
      "learning_rate": 6.022439737287913e-05,
      "loss": 0.8381,
      "step": 182960
    },
    {
      "epoch": 3.2123106093856983,
      "grad_norm": 0.06293635023686944,
      "learning_rate": 6.0214243102315225e-05,
      "loss": 0.8405,
      "step": 182970
    },
    {
      "epoch": 3.2124861742657,
      "grad_norm": 0.050905877559539466,
      "learning_rate": 6.020408934851468e-05,
      "loss": 0.839,
      "step": 182980
    },
    {
      "epoch": 3.2126617391457013,
      "grad_norm": 0.07580665417523146,
      "learning_rate": 6.0193936111606144e-05,
      "loss": 0.8417,
      "step": 182990
    },
    {
      "epoch": 3.212837304025703,
      "grad_norm": 0.055753103211993865,
      "learning_rate": 6.0183783391718265e-05,
      "loss": 0.847,
      "step": 183000
    },
    {
      "epoch": 3.2130128689057043,
      "grad_norm": 0.05013370899580317,
      "learning_rate": 6.017363118897963e-05,
      "loss": 0.8477,
      "step": 183010
    },
    {
      "epoch": 3.2131884337857057,
      "grad_norm": 0.05541218074631912,
      "learning_rate": 6.016347950351895e-05,
      "loss": 0.8348,
      "step": 183020
    },
    {
      "epoch": 3.2133639986657068,
      "grad_norm": 0.04276668118115214,
      "learning_rate": 6.015332833546484e-05,
      "loss": 0.8476,
      "step": 183030
    },
    {
      "epoch": 3.2135395635457082,
      "grad_norm": 0.04649889202360455,
      "learning_rate": 6.014317768494586e-05,
      "loss": 0.8396,
      "step": 183040
    },
    {
      "epoch": 3.2137151284257097,
      "grad_norm": 0.04616187710762655,
      "learning_rate": 6.0133027552090616e-05,
      "loss": 0.8383,
      "step": 183050
    },
    {
      "epoch": 3.213890693305711,
      "grad_norm": 0.05224692051569096,
      "learning_rate": 6.0122877937027776e-05,
      "loss": 0.8407,
      "step": 183060
    },
    {
      "epoch": 3.2140662581857127,
      "grad_norm": 0.05203576035195779,
      "learning_rate": 6.011272883988592e-05,
      "loss": 0.8402,
      "step": 183070
    },
    {
      "epoch": 3.2142418230657137,
      "grad_norm": 0.04434661258853681,
      "learning_rate": 6.010258026079359e-05,
      "loss": 0.849,
      "step": 183080
    },
    {
      "epoch": 3.214417387945715,
      "grad_norm": 0.04715916164946819,
      "learning_rate": 6.0092432199879435e-05,
      "loss": 0.8412,
      "step": 183090
    },
    {
      "epoch": 3.2145929528257167,
      "grad_norm": 0.04933349538880327,
      "learning_rate": 6.0082284657272e-05,
      "loss": 0.8342,
      "step": 183100
    },
    {
      "epoch": 3.214768517705718,
      "grad_norm": 0.0474979508663132,
      "learning_rate": 6.0072137633099826e-05,
      "loss": 0.8473,
      "step": 183110
    },
    {
      "epoch": 3.2149440825857196,
      "grad_norm": 0.05649803995971455,
      "learning_rate": 6.006199112749155e-05,
      "loss": 0.8491,
      "step": 183120
    },
    {
      "epoch": 3.215119647465721,
      "grad_norm": 0.050262384849381996,
      "learning_rate": 6.005184514057568e-05,
      "loss": 0.8414,
      "step": 183130
    },
    {
      "epoch": 3.2152952123457226,
      "grad_norm": 0.05834383015721292,
      "learning_rate": 6.00416996724808e-05,
      "loss": 0.8358,
      "step": 183140
    },
    {
      "epoch": 3.2154707772257236,
      "grad_norm": 0.044583025618811827,
      "learning_rate": 6.003155472333542e-05,
      "loss": 0.8387,
      "step": 183150
    },
    {
      "epoch": 3.215646342105725,
      "grad_norm": 0.05082844261878605,
      "learning_rate": 6.002141029326809e-05,
      "loss": 0.8375,
      "step": 183160
    },
    {
      "epoch": 3.2158219069857266,
      "grad_norm": 0.05481453197525674,
      "learning_rate": 6.001126638240735e-05,
      "loss": 0.8431,
      "step": 183170
    },
    {
      "epoch": 3.215997471865728,
      "grad_norm": 0.05106928673592324,
      "learning_rate": 6.0001122990881684e-05,
      "loss": 0.8423,
      "step": 183180
    },
    {
      "epoch": 3.2161730367457295,
      "grad_norm": 0.04485056603545711,
      "learning_rate": 5.9990980118819695e-05,
      "loss": 0.8404,
      "step": 183190
    },
    {
      "epoch": 3.216348601625731,
      "grad_norm": 0.0668341130987412,
      "learning_rate": 5.998083776634985e-05,
      "loss": 0.8332,
      "step": 183200
    },
    {
      "epoch": 3.216524166505732,
      "grad_norm": 0.058304871266889456,
      "learning_rate": 5.997069593360064e-05,
      "loss": 0.8407,
      "step": 183210
    },
    {
      "epoch": 3.2166997313857335,
      "grad_norm": 0.061452219472675984,
      "learning_rate": 5.996055462070054e-05,
      "loss": 0.8389,
      "step": 183220
    },
    {
      "epoch": 3.216875296265735,
      "grad_norm": 0.06089344124980045,
      "learning_rate": 5.9950413827778116e-05,
      "loss": 0.8452,
      "step": 183230
    },
    {
      "epoch": 3.2170508611457365,
      "grad_norm": 0.061227601815809975,
      "learning_rate": 5.994027355496181e-05,
      "loss": 0.846,
      "step": 183240
    },
    {
      "epoch": 3.217226426025738,
      "grad_norm": 0.051034790972402815,
      "learning_rate": 5.9930133802380095e-05,
      "loss": 0.838,
      "step": 183250
    },
    {
      "epoch": 3.2174019909057394,
      "grad_norm": 0.04960386533592311,
      "learning_rate": 5.991999457016148e-05,
      "loss": 0.8376,
      "step": 183260
    },
    {
      "epoch": 3.2175775557857405,
      "grad_norm": 0.05516100860356233,
      "learning_rate": 5.9909855858434394e-05,
      "loss": 0.8427,
      "step": 183270
    },
    {
      "epoch": 3.217753120665742,
      "grad_norm": 0.05353342041424053,
      "learning_rate": 5.989971766732728e-05,
      "loss": 0.8436,
      "step": 183280
    },
    {
      "epoch": 3.2179286855457434,
      "grad_norm": 0.04606434128896412,
      "learning_rate": 5.988957999696865e-05,
      "loss": 0.8449,
      "step": 183290
    },
    {
      "epoch": 3.218104250425745,
      "grad_norm": 0.051861321732415,
      "learning_rate": 5.987944284748693e-05,
      "loss": 0.8422,
      "step": 183300
    },
    {
      "epoch": 3.2182798153057464,
      "grad_norm": 0.04829771587866355,
      "learning_rate": 5.986930621901055e-05,
      "loss": 0.8406,
      "step": 183310
    },
    {
      "epoch": 3.218455380185748,
      "grad_norm": 0.0480527520069731,
      "learning_rate": 5.9859170111667907e-05,
      "loss": 0.8402,
      "step": 183320
    },
    {
      "epoch": 3.218630945065749,
      "grad_norm": 0.050250119059590884,
      "learning_rate": 5.98490345255875e-05,
      "loss": 0.8387,
      "step": 183330
    },
    {
      "epoch": 3.2188065099457503,
      "grad_norm": 0.04987054380135817,
      "learning_rate": 5.98388994608977e-05,
      "loss": 0.8478,
      "step": 183340
    },
    {
      "epoch": 3.218982074825752,
      "grad_norm": 0.04277976154956029,
      "learning_rate": 5.98287649177269e-05,
      "loss": 0.8395,
      "step": 183350
    },
    {
      "epoch": 3.2191576397057533,
      "grad_norm": 0.05821579581173413,
      "learning_rate": 5.9818630896203575e-05,
      "loss": 0.8439,
      "step": 183360
    },
    {
      "epoch": 3.219333204585755,
      "grad_norm": 0.05080756272594297,
      "learning_rate": 5.980849739645612e-05,
      "loss": 0.8371,
      "step": 183370
    },
    {
      "epoch": 3.2195087694657563,
      "grad_norm": 0.0522340925631416,
      "learning_rate": 5.979836441861284e-05,
      "loss": 0.8452,
      "step": 183380
    },
    {
      "epoch": 3.2196843343457573,
      "grad_norm": 0.08372130266599613,
      "learning_rate": 5.978823196280219e-05,
      "loss": 0.832,
      "step": 183390
    },
    {
      "epoch": 3.2198598992257588,
      "grad_norm": 0.045802601799839464,
      "learning_rate": 5.977810002915257e-05,
      "loss": 0.8338,
      "step": 183400
    },
    {
      "epoch": 3.2200354641057602,
      "grad_norm": 0.05547046421063146,
      "learning_rate": 5.976796861779229e-05,
      "loss": 0.8341,
      "step": 183410
    },
    {
      "epoch": 3.2202110289857617,
      "grad_norm": 0.04975963732296029,
      "learning_rate": 5.975783772884976e-05,
      "loss": 0.8401,
      "step": 183420
    },
    {
      "epoch": 3.220386593865763,
      "grad_norm": 0.05661889257578196,
      "learning_rate": 5.974770736245332e-05,
      "loss": 0.8376,
      "step": 183430
    },
    {
      "epoch": 3.2205621587457647,
      "grad_norm": 0.06830168258314066,
      "learning_rate": 5.973757751873138e-05,
      "loss": 0.8439,
      "step": 183440
    },
    {
      "epoch": 3.2207377236257657,
      "grad_norm": 0.0773152281467056,
      "learning_rate": 5.972744819781218e-05,
      "loss": 0.8453,
      "step": 183450
    },
    {
      "epoch": 3.220913288505767,
      "grad_norm": 0.06231498025060511,
      "learning_rate": 5.971731939982413e-05,
      "loss": 0.8359,
      "step": 183460
    },
    {
      "epoch": 3.2210888533857687,
      "grad_norm": 0.05863489675395209,
      "learning_rate": 5.970719112489561e-05,
      "loss": 0.8392,
      "step": 183470
    },
    {
      "epoch": 3.22126441826577,
      "grad_norm": 0.060568281960057514,
      "learning_rate": 5.9697063373154836e-05,
      "loss": 0.842,
      "step": 183480
    },
    {
      "epoch": 3.2214399831457716,
      "grad_norm": 0.055324290930755624,
      "learning_rate": 5.96869361447302e-05,
      "loss": 0.8409,
      "step": 183490
    },
    {
      "epoch": 3.221615548025773,
      "grad_norm": 0.05322967844418525,
      "learning_rate": 5.9676809439749984e-05,
      "loss": 0.8386,
      "step": 183500
    },
    {
      "epoch": 3.221791112905774,
      "grad_norm": 0.06601640520249595,
      "learning_rate": 5.966668325834257e-05,
      "loss": 0.8502,
      "step": 183510
    },
    {
      "epoch": 3.2219666777857756,
      "grad_norm": 0.06521150238311982,
      "learning_rate": 5.965655760063614e-05,
      "loss": 0.8431,
      "step": 183520
    },
    {
      "epoch": 3.222142242665777,
      "grad_norm": 0.05670593922023839,
      "learning_rate": 5.964643246675905e-05,
      "loss": 0.8437,
      "step": 183530
    },
    {
      "epoch": 3.2223178075457786,
      "grad_norm": 0.043522051019804074,
      "learning_rate": 5.963630785683966e-05,
      "loss": 0.8408,
      "step": 183540
    },
    {
      "epoch": 3.22249337242578,
      "grad_norm": 0.05918710649010191,
      "learning_rate": 5.96261837710061e-05,
      "loss": 0.8476,
      "step": 183550
    },
    {
      "epoch": 3.2226689373057815,
      "grad_norm": 0.06887999560193957,
      "learning_rate": 5.9616060209386725e-05,
      "loss": 0.8367,
      "step": 183560
    },
    {
      "epoch": 3.2228445021857826,
      "grad_norm": 0.05233673931832359,
      "learning_rate": 5.960593717210985e-05,
      "loss": 0.8379,
      "step": 183570
    },
    {
      "epoch": 3.223020067065784,
      "grad_norm": 0.051320575675150686,
      "learning_rate": 5.959581465930365e-05,
      "loss": 0.8353,
      "step": 183580
    },
    {
      "epoch": 3.2231956319457855,
      "grad_norm": 0.04265433727174777,
      "learning_rate": 5.958569267109639e-05,
      "loss": 0.853,
      "step": 183590
    },
    {
      "epoch": 3.223371196825787,
      "grad_norm": 0.04164991691908554,
      "learning_rate": 5.957557120761637e-05,
      "loss": 0.8373,
      "step": 183600
    },
    {
      "epoch": 3.2235467617057885,
      "grad_norm": 0.06736083235494056,
      "learning_rate": 5.9565450268991807e-05,
      "loss": 0.8452,
      "step": 183610
    },
    {
      "epoch": 3.22372232658579,
      "grad_norm": 0.04754517686004884,
      "learning_rate": 5.9555329855350914e-05,
      "loss": 0.842,
      "step": 183620
    },
    {
      "epoch": 3.223897891465791,
      "grad_norm": 0.04910841714197757,
      "learning_rate": 5.954520996682194e-05,
      "loss": 0.8436,
      "step": 183630
    },
    {
      "epoch": 3.2240734563457925,
      "grad_norm": 0.045219140899869475,
      "learning_rate": 5.953509060353312e-05,
      "loss": 0.8482,
      "step": 183640
    },
    {
      "epoch": 3.224249021225794,
      "grad_norm": 0.05286420791951403,
      "learning_rate": 5.9524971765612614e-05,
      "loss": 0.8423,
      "step": 183650
    },
    {
      "epoch": 3.2244245861057954,
      "grad_norm": 0.06390545481050647,
      "learning_rate": 5.951485345318867e-05,
      "loss": 0.8321,
      "step": 183660
    },
    {
      "epoch": 3.224600150985797,
      "grad_norm": 0.04712143559383382,
      "learning_rate": 5.9504735666389485e-05,
      "loss": 0.841,
      "step": 183670
    },
    {
      "epoch": 3.2247757158657984,
      "grad_norm": 0.05421173545578668,
      "learning_rate": 5.9494618405343294e-05,
      "loss": 0.8401,
      "step": 183680
    },
    {
      "epoch": 3.2249512807458,
      "grad_norm": 0.04399590813400492,
      "learning_rate": 5.948450167017819e-05,
      "loss": 0.8388,
      "step": 183690
    },
    {
      "epoch": 3.225126845625801,
      "grad_norm": 0.06445653695704215,
      "learning_rate": 5.947438546102242e-05,
      "loss": 0.8385,
      "step": 183700
    },
    {
      "epoch": 3.2253024105058024,
      "grad_norm": 0.05765108368512892,
      "learning_rate": 5.946426977800419e-05,
      "loss": 0.8435,
      "step": 183710
    },
    {
      "epoch": 3.225477975385804,
      "grad_norm": 0.054151411227574595,
      "learning_rate": 5.945415462125157e-05,
      "loss": 0.8416,
      "step": 183720
    },
    {
      "epoch": 3.2256535402658053,
      "grad_norm": 0.04926472278299313,
      "learning_rate": 5.944403999089276e-05,
      "loss": 0.8359,
      "step": 183730
    },
    {
      "epoch": 3.225829105145807,
      "grad_norm": 0.04660675254526335,
      "learning_rate": 5.9433925887055974e-05,
      "loss": 0.8435,
      "step": 183740
    },
    {
      "epoch": 3.226004670025808,
      "grad_norm": 0.052790955907457665,
      "learning_rate": 5.942381230986929e-05,
      "loss": 0.8413,
      "step": 183750
    },
    {
      "epoch": 3.2261802349058093,
      "grad_norm": 0.07922403742304256,
      "learning_rate": 5.941369925946086e-05,
      "loss": 0.8426,
      "step": 183760
    },
    {
      "epoch": 3.2263557997858108,
      "grad_norm": 0.06250791351888607,
      "learning_rate": 5.940358673595883e-05,
      "loss": 0.8317,
      "step": 183770
    },
    {
      "epoch": 3.2265313646658123,
      "grad_norm": 0.062449535828149556,
      "learning_rate": 5.939347473949137e-05,
      "loss": 0.8436,
      "step": 183780
    },
    {
      "epoch": 3.2267069295458137,
      "grad_norm": 0.046780172083255384,
      "learning_rate": 5.938336327018653e-05,
      "loss": 0.8377,
      "step": 183790
    },
    {
      "epoch": 3.226882494425815,
      "grad_norm": 0.049973605335239,
      "learning_rate": 5.9373252328172424e-05,
      "loss": 0.8388,
      "step": 183800
    },
    {
      "epoch": 3.2270580593058167,
      "grad_norm": 0.04759958854193068,
      "learning_rate": 5.936314191357727e-05,
      "loss": 0.8383,
      "step": 183810
    },
    {
      "epoch": 3.2272336241858177,
      "grad_norm": 0.05181866474913734,
      "learning_rate": 5.935303202652901e-05,
      "loss": 0.843,
      "step": 183820
    },
    {
      "epoch": 3.227409189065819,
      "grad_norm": 0.06207191502647553,
      "learning_rate": 5.934292266715584e-05,
      "loss": 0.8334,
      "step": 183830
    },
    {
      "epoch": 3.2275847539458207,
      "grad_norm": 0.05897047753666602,
      "learning_rate": 5.933281383558581e-05,
      "loss": 0.8408,
      "step": 183840
    },
    {
      "epoch": 3.227760318825822,
      "grad_norm": 0.04161147693416526,
      "learning_rate": 5.9322705531947066e-05,
      "loss": 0.8513,
      "step": 183850
    },
    {
      "epoch": 3.2279358837058236,
      "grad_norm": 0.05788686137262993,
      "learning_rate": 5.9312597756367574e-05,
      "loss": 0.8417,
      "step": 183860
    },
    {
      "epoch": 3.2281114485858247,
      "grad_norm": 0.060436752803791634,
      "learning_rate": 5.9302490508975474e-05,
      "loss": 0.8438,
      "step": 183870
    },
    {
      "epoch": 3.228287013465826,
      "grad_norm": 0.05148034086205373,
      "learning_rate": 5.929238378989883e-05,
      "loss": 0.8408,
      "step": 183880
    },
    {
      "epoch": 3.2284625783458276,
      "grad_norm": 0.04967402586005994,
      "learning_rate": 5.9282277599265656e-05,
      "loss": 0.8504,
      "step": 183890
    },
    {
      "epoch": 3.228638143225829,
      "grad_norm": 0.05365496498681879,
      "learning_rate": 5.9272171937204015e-05,
      "loss": 0.8414,
      "step": 183900
    },
    {
      "epoch": 3.2288137081058306,
      "grad_norm": 0.05840809502354821,
      "learning_rate": 5.926206680384199e-05,
      "loss": 0.8364,
      "step": 183910
    },
    {
      "epoch": 3.228989272985832,
      "grad_norm": 0.04830722889911971,
      "learning_rate": 5.925196219930754e-05,
      "loss": 0.8484,
      "step": 183920
    },
    {
      "epoch": 3.2291648378658335,
      "grad_norm": 0.04576862168179371,
      "learning_rate": 5.924185812372873e-05,
      "loss": 0.8446,
      "step": 183930
    },
    {
      "epoch": 3.2293404027458346,
      "grad_norm": 0.0672394758396955,
      "learning_rate": 5.923175457723358e-05,
      "loss": 0.8451,
      "step": 183940
    },
    {
      "epoch": 3.229515967625836,
      "grad_norm": 0.05491449487091906,
      "learning_rate": 5.922165155995013e-05,
      "loss": 0.8419,
      "step": 183950
    },
    {
      "epoch": 3.2296915325058375,
      "grad_norm": 0.06693796941025162,
      "learning_rate": 5.921154907200635e-05,
      "loss": 0.845,
      "step": 183960
    },
    {
      "epoch": 3.229867097385839,
      "grad_norm": 0.05918360744421931,
      "learning_rate": 5.9201447113530225e-05,
      "loss": 0.8344,
      "step": 183970
    },
    {
      "epoch": 3.2300426622658405,
      "grad_norm": 0.04974892688485771,
      "learning_rate": 5.919134568464983e-05,
      "loss": 0.8398,
      "step": 183980
    },
    {
      "epoch": 3.230218227145842,
      "grad_norm": 0.10964720098456787,
      "learning_rate": 5.9181244785493065e-05,
      "loss": 0.8348,
      "step": 183990
    },
    {
      "epoch": 3.230393792025843,
      "grad_norm": 0.052649441345362126,
      "learning_rate": 5.917114441618793e-05,
      "loss": 0.8362,
      "step": 184000
    },
    {
      "epoch": 3.2305693569058445,
      "grad_norm": 0.05241615329197365,
      "learning_rate": 5.916104457686241e-05,
      "loss": 0.84,
      "step": 184010
    },
    {
      "epoch": 3.230744921785846,
      "grad_norm": 0.05917770832986534,
      "learning_rate": 5.915094526764452e-05,
      "loss": 0.8421,
      "step": 184020
    },
    {
      "epoch": 3.2309204866658474,
      "grad_norm": 0.050416079272688534,
      "learning_rate": 5.914084648866213e-05,
      "loss": 0.8526,
      "step": 184030
    },
    {
      "epoch": 3.231096051545849,
      "grad_norm": 0.05041624212090278,
      "learning_rate": 5.9130748240043246e-05,
      "loss": 0.8394,
      "step": 184040
    },
    {
      "epoch": 3.2312716164258504,
      "grad_norm": 0.044873587279064264,
      "learning_rate": 5.9120650521915845e-05,
      "loss": 0.8393,
      "step": 184050
    },
    {
      "epoch": 3.2314471813058514,
      "grad_norm": 0.05427813140784777,
      "learning_rate": 5.911055333440778e-05,
      "loss": 0.8384,
      "step": 184060
    },
    {
      "epoch": 3.231622746185853,
      "grad_norm": 0.050351633404048114,
      "learning_rate": 5.910045667764704e-05,
      "loss": 0.834,
      "step": 184070
    },
    {
      "epoch": 3.2317983110658544,
      "grad_norm": 0.06297068283304759,
      "learning_rate": 5.909036055176157e-05,
      "loss": 0.8344,
      "step": 184080
    },
    {
      "epoch": 3.231973875945856,
      "grad_norm": 0.052257075494015794,
      "learning_rate": 5.908026495687923e-05,
      "loss": 0.8326,
      "step": 184090
    },
    {
      "epoch": 3.2321494408258573,
      "grad_norm": 0.057869963405787,
      "learning_rate": 5.907016989312798e-05,
      "loss": 0.8326,
      "step": 184100
    },
    {
      "epoch": 3.232325005705859,
      "grad_norm": 0.05497753159713663,
      "learning_rate": 5.9060075360635714e-05,
      "loss": 0.8277,
      "step": 184110
    },
    {
      "epoch": 3.23250057058586,
      "grad_norm": 0.05616445444858763,
      "learning_rate": 5.904998135953035e-05,
      "loss": 0.8384,
      "step": 184120
    },
    {
      "epoch": 3.2326761354658613,
      "grad_norm": 0.07819876404286732,
      "learning_rate": 5.903988788993973e-05,
      "loss": 0.8426,
      "step": 184130
    },
    {
      "epoch": 3.2328517003458628,
      "grad_norm": 0.04951856588017169,
      "learning_rate": 5.902979495199178e-05,
      "loss": 0.8362,
      "step": 184140
    },
    {
      "epoch": 3.2330272652258643,
      "grad_norm": 0.06437870821698817,
      "learning_rate": 5.901970254581439e-05,
      "loss": 0.8363,
      "step": 184150
    },
    {
      "epoch": 3.2332028301058657,
      "grad_norm": 0.0498382752237453,
      "learning_rate": 5.90096106715354e-05,
      "loss": 0.8432,
      "step": 184160
    },
    {
      "epoch": 3.233378394985867,
      "grad_norm": 0.049996434389085284,
      "learning_rate": 5.8999519329282685e-05,
      "loss": 0.8396,
      "step": 184170
    },
    {
      "epoch": 3.2335539598658682,
      "grad_norm": 0.05859343670168666,
      "learning_rate": 5.89894285191841e-05,
      "loss": 0.8462,
      "step": 184180
    },
    {
      "epoch": 3.2337295247458697,
      "grad_norm": 0.05382630502868404,
      "learning_rate": 5.897933824136755e-05,
      "loss": 0.8369,
      "step": 184190
    },
    {
      "epoch": 3.233905089625871,
      "grad_norm": 0.04714161237565076,
      "learning_rate": 5.8969248495960805e-05,
      "loss": 0.8438,
      "step": 184200
    },
    {
      "epoch": 3.2340806545058727,
      "grad_norm": 0.07707918768868491,
      "learning_rate": 5.8959159283091724e-05,
      "loss": 0.8378,
      "step": 184210
    },
    {
      "epoch": 3.234256219385874,
      "grad_norm": 0.055841691031139284,
      "learning_rate": 5.894907060288819e-05,
      "loss": 0.8366,
      "step": 184220
    },
    {
      "epoch": 3.2344317842658756,
      "grad_norm": 0.04772009713756977,
      "learning_rate": 5.893898245547798e-05,
      "loss": 0.8458,
      "step": 184230
    },
    {
      "epoch": 3.2346073491458767,
      "grad_norm": 0.051881256056288294,
      "learning_rate": 5.892889484098888e-05,
      "loss": 0.8346,
      "step": 184240
    },
    {
      "epoch": 3.234782914025878,
      "grad_norm": 0.05475909407833451,
      "learning_rate": 5.8918807759548825e-05,
      "loss": 0.8452,
      "step": 184250
    },
    {
      "epoch": 3.2349584789058796,
      "grad_norm": 0.05465574307208287,
      "learning_rate": 5.8908721211285475e-05,
      "loss": 0.8529,
      "step": 184260
    },
    {
      "epoch": 3.235134043785881,
      "grad_norm": 0.05144556831330523,
      "learning_rate": 5.8898635196326714e-05,
      "loss": 0.8423,
      "step": 184270
    },
    {
      "epoch": 3.2353096086658826,
      "grad_norm": 0.053965417728487015,
      "learning_rate": 5.888854971480029e-05,
      "loss": 0.8318,
      "step": 184280
    },
    {
      "epoch": 3.235485173545884,
      "grad_norm": 0.0461011939652606,
      "learning_rate": 5.8878464766834076e-05,
      "loss": 0.8396,
      "step": 184290
    },
    {
      "epoch": 3.235660738425885,
      "grad_norm": 0.05123982242953701,
      "learning_rate": 5.886838035255573e-05,
      "loss": 0.8436,
      "step": 184300
    },
    {
      "epoch": 3.2358363033058866,
      "grad_norm": 0.0779615933800103,
      "learning_rate": 5.8858296472093076e-05,
      "loss": 0.8447,
      "step": 184310
    },
    {
      "epoch": 3.236011868185888,
      "grad_norm": 0.050231680794836475,
      "learning_rate": 5.8848213125573934e-05,
      "loss": 0.8404,
      "step": 184320
    },
    {
      "epoch": 3.2361874330658895,
      "grad_norm": 0.05149228798870723,
      "learning_rate": 5.883813031312596e-05,
      "loss": 0.8399,
      "step": 184330
    },
    {
      "epoch": 3.236362997945891,
      "grad_norm": 0.05519220327246756,
      "learning_rate": 5.882804803487697e-05,
      "loss": 0.8324,
      "step": 184340
    },
    {
      "epoch": 3.2365385628258925,
      "grad_norm": 0.04384476424110565,
      "learning_rate": 5.881796629095467e-05,
      "loss": 0.8424,
      "step": 184350
    },
    {
      "epoch": 3.2367141277058935,
      "grad_norm": 0.06203020701494398,
      "learning_rate": 5.880788508148688e-05,
      "loss": 0.8332,
      "step": 184360
    },
    {
      "epoch": 3.236889692585895,
      "grad_norm": 0.06469797748721112,
      "learning_rate": 5.879780440660122e-05,
      "loss": 0.8457,
      "step": 184370
    },
    {
      "epoch": 3.2370652574658965,
      "grad_norm": 0.051961817527365436,
      "learning_rate": 5.878772426642549e-05,
      "loss": 0.8421,
      "step": 184380
    },
    {
      "epoch": 3.237240822345898,
      "grad_norm": 0.0594154931507186,
      "learning_rate": 5.877764466108741e-05,
      "loss": 0.8471,
      "step": 184390
    },
    {
      "epoch": 3.2374163872258994,
      "grad_norm": 0.06481084755526148,
      "learning_rate": 5.8767565590714605e-05,
      "loss": 0.8389,
      "step": 184400
    },
    {
      "epoch": 3.237591952105901,
      "grad_norm": 0.06432595440013829,
      "learning_rate": 5.8757487055434856e-05,
      "loss": 0.8475,
      "step": 184410
    },
    {
      "epoch": 3.237767516985902,
      "grad_norm": 0.04659090534825186,
      "learning_rate": 5.874740905537589e-05,
      "loss": 0.8412,
      "step": 184420
    },
    {
      "epoch": 3.2379430818659034,
      "grad_norm": 0.04836031161464134,
      "learning_rate": 5.873733159066529e-05,
      "loss": 0.8393,
      "step": 184430
    },
    {
      "epoch": 3.238118646745905,
      "grad_norm": 0.05528361472318778,
      "learning_rate": 5.872725466143081e-05,
      "loss": 0.841,
      "step": 184440
    },
    {
      "epoch": 3.2382942116259064,
      "grad_norm": 0.04645003760108794,
      "learning_rate": 5.871717826780011e-05,
      "loss": 0.837,
      "step": 184450
    },
    {
      "epoch": 3.238469776505908,
      "grad_norm": 0.049358917780806456,
      "learning_rate": 5.870710240990092e-05,
      "loss": 0.8349,
      "step": 184460
    },
    {
      "epoch": 3.2386453413859093,
      "grad_norm": 0.0560844531583116,
      "learning_rate": 5.869702708786078e-05,
      "loss": 0.8335,
      "step": 184470
    },
    {
      "epoch": 3.238820906265911,
      "grad_norm": 0.05068903589402944,
      "learning_rate": 5.868695230180744e-05,
      "loss": 0.8411,
      "step": 184480
    },
    {
      "epoch": 3.238996471145912,
      "grad_norm": 0.05642652564972516,
      "learning_rate": 5.8676878051868544e-05,
      "loss": 0.8427,
      "step": 184490
    },
    {
      "epoch": 3.2391720360259133,
      "grad_norm": 0.05797892053839419,
      "learning_rate": 5.866680433817169e-05,
      "loss": 0.8428,
      "step": 184500
    },
    {
      "epoch": 3.239347600905915,
      "grad_norm": 0.056246288668489706,
      "learning_rate": 5.8656731160844525e-05,
      "loss": 0.8539,
      "step": 184510
    },
    {
      "epoch": 3.2395231657859163,
      "grad_norm": 0.05622180333995811,
      "learning_rate": 5.864665852001471e-05,
      "loss": 0.8507,
      "step": 184520
    },
    {
      "epoch": 3.2396987306659177,
      "grad_norm": 0.06898833326503896,
      "learning_rate": 5.863658641580986e-05,
      "loss": 0.8418,
      "step": 184530
    },
    {
      "epoch": 3.2398742955459188,
      "grad_norm": 0.05215319828829471,
      "learning_rate": 5.862651484835756e-05,
      "loss": 0.8417,
      "step": 184540
    },
    {
      "epoch": 3.2400498604259202,
      "grad_norm": 0.05421786380705445,
      "learning_rate": 5.861644381778543e-05,
      "loss": 0.8395,
      "step": 184550
    },
    {
      "epoch": 3.2402254253059217,
      "grad_norm": 0.045759195715106026,
      "learning_rate": 5.860637332422112e-05,
      "loss": 0.8485,
      "step": 184560
    },
    {
      "epoch": 3.240400990185923,
      "grad_norm": 0.04435382897843194,
      "learning_rate": 5.859630336779215e-05,
      "loss": 0.8444,
      "step": 184570
    },
    {
      "epoch": 3.2405765550659247,
      "grad_norm": 0.05660823812640221,
      "learning_rate": 5.858623394862615e-05,
      "loss": 0.8406,
      "step": 184580
    },
    {
      "epoch": 3.240752119945926,
      "grad_norm": 0.05785421848903796,
      "learning_rate": 5.857616506685071e-05,
      "loss": 0.8423,
      "step": 184590
    },
    {
      "epoch": 3.2409276848259276,
      "grad_norm": 0.05798908314494106,
      "learning_rate": 5.856609672259337e-05,
      "loss": 0.8345,
      "step": 184600
    },
    {
      "epoch": 3.2411032497059287,
      "grad_norm": 0.05422642233516043,
      "learning_rate": 5.855602891598171e-05,
      "loss": 0.8319,
      "step": 184610
    },
    {
      "epoch": 3.24127881458593,
      "grad_norm": 0.05811184755464429,
      "learning_rate": 5.854596164714329e-05,
      "loss": 0.8414,
      "step": 184620
    },
    {
      "epoch": 3.2414543794659316,
      "grad_norm": 0.05118443933370086,
      "learning_rate": 5.8535894916205723e-05,
      "loss": 0.8456,
      "step": 184630
    },
    {
      "epoch": 3.241629944345933,
      "grad_norm": 0.055364233549595514,
      "learning_rate": 5.8525828723296475e-05,
      "loss": 0.8412,
      "step": 184640
    },
    {
      "epoch": 3.2418055092259346,
      "grad_norm": 0.07850177611529766,
      "learning_rate": 5.8515763068543084e-05,
      "loss": 0.8371,
      "step": 184650
    },
    {
      "epoch": 3.241981074105936,
      "grad_norm": 0.05681733528276603,
      "learning_rate": 5.850569795207319e-05,
      "loss": 0.8472,
      "step": 184660
    },
    {
      "epoch": 3.242156638985937,
      "grad_norm": 0.050619696418904604,
      "learning_rate": 5.8495633374014194e-05,
      "loss": 0.8394,
      "step": 184670
    },
    {
      "epoch": 3.2423322038659386,
      "grad_norm": 0.05734114310136564,
      "learning_rate": 5.848556933449367e-05,
      "loss": 0.8313,
      "step": 184680
    },
    {
      "epoch": 3.24250776874594,
      "grad_norm": 0.04820754608430847,
      "learning_rate": 5.847550583363912e-05,
      "loss": 0.8346,
      "step": 184690
    },
    {
      "epoch": 3.2426833336259415,
      "grad_norm": 0.051014242854346734,
      "learning_rate": 5.8465442871578125e-05,
      "loss": 0.8379,
      "step": 184700
    },
    {
      "epoch": 3.242858898505943,
      "grad_norm": 0.049970429999527013,
      "learning_rate": 5.845538044843807e-05,
      "loss": 0.8436,
      "step": 184710
    },
    {
      "epoch": 3.2430344633859445,
      "grad_norm": 0.05771402651518539,
      "learning_rate": 5.84453185643465e-05,
      "loss": 0.8437,
      "step": 184720
    },
    {
      "epoch": 3.2432100282659455,
      "grad_norm": 0.053035948859372135,
      "learning_rate": 5.843525721943093e-05,
      "loss": 0.8446,
      "step": 184730
    },
    {
      "epoch": 3.243385593145947,
      "grad_norm": 0.06857716020508244,
      "learning_rate": 5.842519641381876e-05,
      "loss": 0.8399,
      "step": 184740
    },
    {
      "epoch": 3.2435611580259485,
      "grad_norm": 0.0552432223529788,
      "learning_rate": 5.8415136147637525e-05,
      "loss": 0.8354,
      "step": 184750
    },
    {
      "epoch": 3.24373672290595,
      "grad_norm": 0.06411136386925215,
      "learning_rate": 5.8405076421014726e-05,
      "loss": 0.8368,
      "step": 184760
    },
    {
      "epoch": 3.2439122877859514,
      "grad_norm": 0.07209185800077794,
      "learning_rate": 5.839501723407771e-05,
      "loss": 0.8454,
      "step": 184770
    },
    {
      "epoch": 3.244087852665953,
      "grad_norm": 0.05476757332128607,
      "learning_rate": 5.838495858695401e-05,
      "loss": 0.845,
      "step": 184780
    },
    {
      "epoch": 3.244263417545954,
      "grad_norm": 0.0473062777620129,
      "learning_rate": 5.837490047977103e-05,
      "loss": 0.8381,
      "step": 184790
    },
    {
      "epoch": 3.2444389824259554,
      "grad_norm": 0.03990737273010354,
      "learning_rate": 5.8364842912656305e-05,
      "loss": 0.8498,
      "step": 184800
    },
    {
      "epoch": 3.244614547305957,
      "grad_norm": 0.049441061129540505,
      "learning_rate": 5.8354785885737126e-05,
      "loss": 0.8459,
      "step": 184810
    },
    {
      "epoch": 3.2447901121859584,
      "grad_norm": 0.05559455795785447,
      "learning_rate": 5.834472939914098e-05,
      "loss": 0.8364,
      "step": 184820
    },
    {
      "epoch": 3.24496567706596,
      "grad_norm": 0.066305322195853,
      "learning_rate": 5.8334673452995336e-05,
      "loss": 0.8419,
      "step": 184830
    },
    {
      "epoch": 3.2451412419459613,
      "grad_norm": 0.06140602264248671,
      "learning_rate": 5.832461804742751e-05,
      "loss": 0.839,
      "step": 184840
    },
    {
      "epoch": 3.2453168068259624,
      "grad_norm": 0.056323392874587194,
      "learning_rate": 5.8314563182564966e-05,
      "loss": 0.8476,
      "step": 184850
    },
    {
      "epoch": 3.245492371705964,
      "grad_norm": 0.04781422909238236,
      "learning_rate": 5.830450885853508e-05,
      "loss": 0.8362,
      "step": 184860
    },
    {
      "epoch": 3.2456679365859653,
      "grad_norm": 0.055807842087495736,
      "learning_rate": 5.8294455075465295e-05,
      "loss": 0.8531,
      "step": 184870
    },
    {
      "epoch": 3.245843501465967,
      "grad_norm": 0.06007871093167832,
      "learning_rate": 5.82844018334829e-05,
      "loss": 0.8465,
      "step": 184880
    },
    {
      "epoch": 3.2460190663459683,
      "grad_norm": 0.048713451734531524,
      "learning_rate": 5.82743491327153e-05,
      "loss": 0.8469,
      "step": 184890
    },
    {
      "epoch": 3.2461946312259697,
      "grad_norm": 0.05817465659163844,
      "learning_rate": 5.826429697328996e-05,
      "loss": 0.8382,
      "step": 184900
    },
    {
      "epoch": 3.2463701961059708,
      "grad_norm": 0.04884703023014615,
      "learning_rate": 5.82542453553341e-05,
      "loss": 0.8356,
      "step": 184910
    },
    {
      "epoch": 3.2465457609859723,
      "grad_norm": 0.05259910580802678,
      "learning_rate": 5.8244194278975164e-05,
      "loss": 0.8442,
      "step": 184920
    },
    {
      "epoch": 3.2467213258659737,
      "grad_norm": 0.052200217986386264,
      "learning_rate": 5.8234143744340505e-05,
      "loss": 0.8498,
      "step": 184930
    },
    {
      "epoch": 3.246896890745975,
      "grad_norm": 0.06867905382344527,
      "learning_rate": 5.822409375155742e-05,
      "loss": 0.8435,
      "step": 184940
    },
    {
      "epoch": 3.2470724556259767,
      "grad_norm": 0.05620392396916582,
      "learning_rate": 5.8214044300753247e-05,
      "loss": 0.8436,
      "step": 184950
    },
    {
      "epoch": 3.247248020505978,
      "grad_norm": 0.046873439328896066,
      "learning_rate": 5.8203995392055346e-05,
      "loss": 0.8343,
      "step": 184960
    },
    {
      "epoch": 3.247423585385979,
      "grad_norm": 0.06749060035345454,
      "learning_rate": 5.8193947025591055e-05,
      "loss": 0.8359,
      "step": 184970
    },
    {
      "epoch": 3.2475991502659807,
      "grad_norm": 0.04191353379262066,
      "learning_rate": 5.8183899201487625e-05,
      "loss": 0.8435,
      "step": 184980
    },
    {
      "epoch": 3.247774715145982,
      "grad_norm": 0.047332760115030806,
      "learning_rate": 5.81738519198724e-05,
      "loss": 0.8394,
      "step": 184990
    },
    {
      "epoch": 3.2479502800259836,
      "grad_norm": 0.046378021121788855,
      "learning_rate": 5.816380518087271e-05,
      "loss": 0.8469,
      "step": 185000
    },
    {
      "epoch": 3.248125844905985,
      "grad_norm": 0.04513987011928773,
      "learning_rate": 5.8153758984615795e-05,
      "loss": 0.8349,
      "step": 185010
    },
    {
      "epoch": 3.2483014097859866,
      "grad_norm": 0.04428683458874726,
      "learning_rate": 5.814371333122895e-05,
      "loss": 0.8384,
      "step": 185020
    },
    {
      "epoch": 3.2484769746659876,
      "grad_norm": 0.06528813048374721,
      "learning_rate": 5.8133668220839475e-05,
      "loss": 0.8401,
      "step": 185030
    },
    {
      "epoch": 3.248652539545989,
      "grad_norm": 0.052261618035757976,
      "learning_rate": 5.812362365357466e-05,
      "loss": 0.8424,
      "step": 185040
    },
    {
      "epoch": 3.2488281044259906,
      "grad_norm": 0.04737975295458155,
      "learning_rate": 5.811357962956173e-05,
      "loss": 0.8476,
      "step": 185050
    },
    {
      "epoch": 3.249003669305992,
      "grad_norm": 0.05121717719654849,
      "learning_rate": 5.8103536148927964e-05,
      "loss": 0.8417,
      "step": 185060
    },
    {
      "epoch": 3.2491792341859935,
      "grad_norm": 0.05767321643913805,
      "learning_rate": 5.8093493211800666e-05,
      "loss": 0.8414,
      "step": 185070
    },
    {
      "epoch": 3.249354799065995,
      "grad_norm": 0.04392129265701165,
      "learning_rate": 5.8083450818306974e-05,
      "loss": 0.8419,
      "step": 185080
    },
    {
      "epoch": 3.249530363945996,
      "grad_norm": 0.047658790815133203,
      "learning_rate": 5.807340896857419e-05,
      "loss": 0.8418,
      "step": 185090
    },
    {
      "epoch": 3.2497059288259975,
      "grad_norm": 0.06510381453283089,
      "learning_rate": 5.8063367662729596e-05,
      "loss": 0.8368,
      "step": 185100
    },
    {
      "epoch": 3.249881493705999,
      "grad_norm": 0.05695956768042066,
      "learning_rate": 5.8053326900900305e-05,
      "loss": 0.8277,
      "step": 185110
    },
    {
      "epoch": 3.2500570585860005,
      "grad_norm": 0.04535302140321086,
      "learning_rate": 5.8043286683213606e-05,
      "loss": 0.8409,
      "step": 185120
    },
    {
      "epoch": 3.250232623466002,
      "grad_norm": 0.053516131493588706,
      "learning_rate": 5.8033247009796675e-05,
      "loss": 0.8444,
      "step": 185130
    },
    {
      "epoch": 3.2504081883460034,
      "grad_norm": 0.05643519680250108,
      "learning_rate": 5.8023207880776784e-05,
      "loss": 0.8478,
      "step": 185140
    },
    {
      "epoch": 3.250583753226005,
      "grad_norm": 0.05261608453696099,
      "learning_rate": 5.8013169296281057e-05,
      "loss": 0.8372,
      "step": 185150
    },
    {
      "epoch": 3.250759318106006,
      "grad_norm": 0.04663071297285905,
      "learning_rate": 5.8003131256436716e-05,
      "loss": 0.8321,
      "step": 185160
    },
    {
      "epoch": 3.2509348829860074,
      "grad_norm": 0.04825829634078794,
      "learning_rate": 5.7993093761370973e-05,
      "loss": 0.8431,
      "step": 185170
    },
    {
      "epoch": 3.251110447866009,
      "grad_norm": 0.054697710732549476,
      "learning_rate": 5.798305681121093e-05,
      "loss": 0.8493,
      "step": 185180
    },
    {
      "epoch": 3.2512860127460104,
      "grad_norm": 0.0675316174041659,
      "learning_rate": 5.7973020406083814e-05,
      "loss": 0.8421,
      "step": 185190
    },
    {
      "epoch": 3.251461577626012,
      "grad_norm": 0.052279274571173526,
      "learning_rate": 5.7962984546116785e-05,
      "loss": 0.8423,
      "step": 185200
    },
    {
      "epoch": 3.251637142506013,
      "grad_norm": 0.04207109579511031,
      "learning_rate": 5.7952949231437e-05,
      "loss": 0.8417,
      "step": 185210
    },
    {
      "epoch": 3.2518127073860144,
      "grad_norm": 0.06253793711616629,
      "learning_rate": 5.794291446217158e-05,
      "loss": 0.8393,
      "step": 185220
    },
    {
      "epoch": 3.251988272266016,
      "grad_norm": 0.052118138810322304,
      "learning_rate": 5.7932880238447683e-05,
      "loss": 0.8497,
      "step": 185230
    },
    {
      "epoch": 3.2521638371460173,
      "grad_norm": 0.05573204382564439,
      "learning_rate": 5.7922846560392485e-05,
      "loss": 0.8467,
      "step": 185240
    },
    {
      "epoch": 3.252339402026019,
      "grad_norm": 0.05237773534782745,
      "learning_rate": 5.791281342813304e-05,
      "loss": 0.8472,
      "step": 185250
    },
    {
      "epoch": 3.2525149669060203,
      "grad_norm": 0.06667961707976088,
      "learning_rate": 5.79027808417965e-05,
      "loss": 0.8339,
      "step": 185260
    },
    {
      "epoch": 3.2526905317860217,
      "grad_norm": 0.052009513106569594,
      "learning_rate": 5.7892748801510036e-05,
      "loss": 0.8462,
      "step": 185270
    },
    {
      "epoch": 3.2528660966660228,
      "grad_norm": 0.045965292683245604,
      "learning_rate": 5.788271730740066e-05,
      "loss": 0.8401,
      "step": 185280
    },
    {
      "epoch": 3.2530416615460243,
      "grad_norm": 0.08632097978582033,
      "learning_rate": 5.7872686359595506e-05,
      "loss": 0.8452,
      "step": 185290
    },
    {
      "epoch": 3.2532172264260257,
      "grad_norm": 0.057351329970485566,
      "learning_rate": 5.78626559582217e-05,
      "loss": 0.848,
      "step": 185300
    },
    {
      "epoch": 3.253392791306027,
      "grad_norm": 0.06519562496277555,
      "learning_rate": 5.785262610340634e-05,
      "loss": 0.8456,
      "step": 185310
    },
    {
      "epoch": 3.2535683561860287,
      "grad_norm": 0.04351018168128444,
      "learning_rate": 5.784259679527644e-05,
      "loss": 0.8457,
      "step": 185320
    },
    {
      "epoch": 3.2537439210660297,
      "grad_norm": 0.056745545160710814,
      "learning_rate": 5.7832568033959114e-05,
      "loss": 0.8483,
      "step": 185330
    },
    {
      "epoch": 3.253919485946031,
      "grad_norm": 0.06935682952276113,
      "learning_rate": 5.7822539819581454e-05,
      "loss": 0.8379,
      "step": 185340
    },
    {
      "epoch": 3.2540950508260327,
      "grad_norm": 0.06388685439551489,
      "learning_rate": 5.781251215227045e-05,
      "loss": 0.8432,
      "step": 185350
    },
    {
      "epoch": 3.254270615706034,
      "grad_norm": 0.05585297394299238,
      "learning_rate": 5.7802485032153194e-05,
      "loss": 0.8429,
      "step": 185360
    },
    {
      "epoch": 3.2544461805860356,
      "grad_norm": 0.051116606292073465,
      "learning_rate": 5.7792458459356746e-05,
      "loss": 0.8374,
      "step": 185370
    },
    {
      "epoch": 3.254621745466037,
      "grad_norm": 0.06244052216603926,
      "learning_rate": 5.778243243400815e-05,
      "loss": 0.8503,
      "step": 185380
    },
    {
      "epoch": 3.2547973103460386,
      "grad_norm": 0.04330450940020767,
      "learning_rate": 5.777240695623439e-05,
      "loss": 0.8469,
      "step": 185390
    },
    {
      "epoch": 3.2549728752260396,
      "grad_norm": 0.048760279773084524,
      "learning_rate": 5.776238202616252e-05,
      "loss": 0.8411,
      "step": 185400
    },
    {
      "epoch": 3.255148440106041,
      "grad_norm": 0.048312944767618965,
      "learning_rate": 5.77523576439196e-05,
      "loss": 0.8451,
      "step": 185410
    },
    {
      "epoch": 3.2553240049860426,
      "grad_norm": 0.05015244285744013,
      "learning_rate": 5.774233380963254e-05,
      "loss": 0.8456,
      "step": 185420
    },
    {
      "epoch": 3.255499569866044,
      "grad_norm": 0.05157513039847175,
      "learning_rate": 5.773231052342841e-05,
      "loss": 0.8412,
      "step": 185430
    },
    {
      "epoch": 3.2556751347460455,
      "grad_norm": 0.05142603478057689,
      "learning_rate": 5.772228778543424e-05,
      "loss": 0.833,
      "step": 185440
    },
    {
      "epoch": 3.2558506996260466,
      "grad_norm": 0.05960352022202547,
      "learning_rate": 5.771226559577694e-05,
      "loss": 0.8428,
      "step": 185450
    },
    {
      "epoch": 3.256026264506048,
      "grad_norm": 0.048667485073505976,
      "learning_rate": 5.770224395458352e-05,
      "loss": 0.841,
      "step": 185460
    },
    {
      "epoch": 3.2562018293860495,
      "grad_norm": 0.05055861884782479,
      "learning_rate": 5.769222286198098e-05,
      "loss": 0.8416,
      "step": 185470
    },
    {
      "epoch": 3.256377394266051,
      "grad_norm": 0.04669156146129456,
      "learning_rate": 5.768220231809631e-05,
      "loss": 0.8468,
      "step": 185480
    },
    {
      "epoch": 3.2565529591460525,
      "grad_norm": 0.052159651633536114,
      "learning_rate": 5.767218232305639e-05,
      "loss": 0.8438,
      "step": 185490
    },
    {
      "epoch": 3.256728524026054,
      "grad_norm": 0.11910068227725813,
      "learning_rate": 5.766216287698825e-05,
      "loss": 0.8471,
      "step": 185500
    },
    {
      "epoch": 3.2569040889060554,
      "grad_norm": 0.04878110165146401,
      "learning_rate": 5.765214398001882e-05,
      "loss": 0.8511,
      "step": 185510
    },
    {
      "epoch": 3.2570796537860565,
      "grad_norm": 0.05670014040826032,
      "learning_rate": 5.764212563227501e-05,
      "loss": 0.8475,
      "step": 185520
    },
    {
      "epoch": 3.257255218666058,
      "grad_norm": 0.046545767775613006,
      "learning_rate": 5.763210783388378e-05,
      "loss": 0.8473,
      "step": 185530
    },
    {
      "epoch": 3.2574307835460594,
      "grad_norm": 0.05392337551539763,
      "learning_rate": 5.7622090584972044e-05,
      "loss": 0.8382,
      "step": 185540
    },
    {
      "epoch": 3.257606348426061,
      "grad_norm": 0.047683560907619035,
      "learning_rate": 5.761207388566679e-05,
      "loss": 0.8413,
      "step": 185550
    },
    {
      "epoch": 3.2577819133060624,
      "grad_norm": 0.05674820959105953,
      "learning_rate": 5.760205773609483e-05,
      "loss": 0.8468,
      "step": 185560
    },
    {
      "epoch": 3.257957478186064,
      "grad_norm": 0.04900432846854775,
      "learning_rate": 5.7592042136383094e-05,
      "loss": 0.8432,
      "step": 185570
    },
    {
      "epoch": 3.258133043066065,
      "grad_norm": 0.05772015990832718,
      "learning_rate": 5.7582027086658564e-05,
      "loss": 0.8376,
      "step": 185580
    },
    {
      "epoch": 3.2583086079460664,
      "grad_norm": 0.052119788049831875,
      "learning_rate": 5.757201258704803e-05,
      "loss": 0.8467,
      "step": 185590
    },
    {
      "epoch": 3.258484172826068,
      "grad_norm": 0.04946754302859257,
      "learning_rate": 5.756199863767842e-05,
      "loss": 0.8445,
      "step": 185600
    },
    {
      "epoch": 3.2586597377060693,
      "grad_norm": 0.05046214096774248,
      "learning_rate": 5.7551985238676645e-05,
      "loss": 0.8462,
      "step": 185610
    },
    {
      "epoch": 3.258835302586071,
      "grad_norm": 0.05920479887418911,
      "learning_rate": 5.754197239016951e-05,
      "loss": 0.8449,
      "step": 185620
    },
    {
      "epoch": 3.2590108674660723,
      "grad_norm": 0.055714991777823644,
      "learning_rate": 5.753196009228391e-05,
      "loss": 0.8396,
      "step": 185630
    },
    {
      "epoch": 3.2591864323460733,
      "grad_norm": 0.04555082002594911,
      "learning_rate": 5.7521948345146716e-05,
      "loss": 0.8312,
      "step": 185640
    },
    {
      "epoch": 3.259361997226075,
      "grad_norm": 0.05689903205755594,
      "learning_rate": 5.7511937148884786e-05,
      "loss": 0.8409,
      "step": 185650
    },
    {
      "epoch": 3.2595375621060763,
      "grad_norm": 0.053809294023697114,
      "learning_rate": 5.750192650362494e-05,
      "loss": 0.8445,
      "step": 185660
    },
    {
      "epoch": 3.2597131269860777,
      "grad_norm": 0.05131055307855098,
      "learning_rate": 5.7491916409493994e-05,
      "loss": 0.8357,
      "step": 185670
    },
    {
      "epoch": 3.259888691866079,
      "grad_norm": 0.04786798475118406,
      "learning_rate": 5.748190686661885e-05,
      "loss": 0.8391,
      "step": 185680
    },
    {
      "epoch": 3.2600642567460807,
      "grad_norm": 0.048317815563771266,
      "learning_rate": 5.747189787512626e-05,
      "loss": 0.8452,
      "step": 185690
    },
    {
      "epoch": 3.2602398216260817,
      "grad_norm": 0.04485997567254521,
      "learning_rate": 5.746188943514305e-05,
      "loss": 0.8465,
      "step": 185700
    },
    {
      "epoch": 3.260415386506083,
      "grad_norm": 0.0473655461107759,
      "learning_rate": 5.745188154679606e-05,
      "loss": 0.8317,
      "step": 185710
    },
    {
      "epoch": 3.2605909513860847,
      "grad_norm": 0.10016425753317294,
      "learning_rate": 5.74418742102121e-05,
      "loss": 0.8388,
      "step": 185720
    },
    {
      "epoch": 3.260766516266086,
      "grad_norm": 0.05383686721102848,
      "learning_rate": 5.7431867425517915e-05,
      "loss": 0.8369,
      "step": 185730
    },
    {
      "epoch": 3.2609420811460876,
      "grad_norm": 0.05142359213776441,
      "learning_rate": 5.742186119284031e-05,
      "loss": 0.8338,
      "step": 185740
    },
    {
      "epoch": 3.261117646026089,
      "grad_norm": 0.04441037532475336,
      "learning_rate": 5.7411855512306135e-05,
      "loss": 0.8439,
      "step": 185750
    },
    {
      "epoch": 3.26129321090609,
      "grad_norm": 0.06207295270410876,
      "learning_rate": 5.740185038404204e-05,
      "loss": 0.8424,
      "step": 185760
    },
    {
      "epoch": 3.2614687757860916,
      "grad_norm": 0.05599202172871003,
      "learning_rate": 5.739184580817487e-05,
      "loss": 0.8468,
      "step": 185770
    },
    {
      "epoch": 3.261644340666093,
      "grad_norm": 0.06266605871879978,
      "learning_rate": 5.73818417848314e-05,
      "loss": 0.8349,
      "step": 185780
    },
    {
      "epoch": 3.2618199055460946,
      "grad_norm": 0.06042389031565253,
      "learning_rate": 5.7371838314138326e-05,
      "loss": 0.8475,
      "step": 185790
    },
    {
      "epoch": 3.261995470426096,
      "grad_norm": 0.04828676600639126,
      "learning_rate": 5.736183539622242e-05,
      "loss": 0.831,
      "step": 185800
    },
    {
      "epoch": 3.2621710353060975,
      "grad_norm": 0.05221482672235218,
      "learning_rate": 5.735183303121041e-05,
      "loss": 0.8442,
      "step": 185810
    },
    {
      "epoch": 3.262346600186099,
      "grad_norm": 0.06753708709123074,
      "learning_rate": 5.73418312192291e-05,
      "loss": 0.8391,
      "step": 185820
    },
    {
      "epoch": 3.2625221650661,
      "grad_norm": 0.06017301553391267,
      "learning_rate": 5.73318299604051e-05,
      "loss": 0.8367,
      "step": 185830
    },
    {
      "epoch": 3.2626977299461015,
      "grad_norm": 0.05683615477212134,
      "learning_rate": 5.732182925486519e-05,
      "loss": 0.839,
      "step": 185840
    },
    {
      "epoch": 3.262873294826103,
      "grad_norm": 0.0457756227508442,
      "learning_rate": 5.731182910273611e-05,
      "loss": 0.8412,
      "step": 185850
    },
    {
      "epoch": 3.2630488597061045,
      "grad_norm": 0.046279280952931615,
      "learning_rate": 5.7301829504144496e-05,
      "loss": 0.8431,
      "step": 185860
    },
    {
      "epoch": 3.263224424586106,
      "grad_norm": 0.0551320051615266,
      "learning_rate": 5.7291830459217064e-05,
      "loss": 0.8395,
      "step": 185870
    },
    {
      "epoch": 3.263399989466107,
      "grad_norm": 0.06165074073784295,
      "learning_rate": 5.7281831968080524e-05,
      "loss": 0.8441,
      "step": 185880
    },
    {
      "epoch": 3.2635755543461085,
      "grad_norm": 0.055201295621117,
      "learning_rate": 5.727183403086161e-05,
      "loss": 0.8371,
      "step": 185890
    },
    {
      "epoch": 3.26375111922611,
      "grad_norm": 0.04748344545537263,
      "learning_rate": 5.726183664768689e-05,
      "loss": 0.8452,
      "step": 185900
    },
    {
      "epoch": 3.2639266841061114,
      "grad_norm": 0.052060245509255136,
      "learning_rate": 5.725183981868305e-05,
      "loss": 0.8381,
      "step": 185910
    },
    {
      "epoch": 3.264102248986113,
      "grad_norm": 0.0541208274462344,
      "learning_rate": 5.724184354397686e-05,
      "loss": 0.8286,
      "step": 185920
    },
    {
      "epoch": 3.2642778138661144,
      "grad_norm": 0.08080320001898621,
      "learning_rate": 5.7231847823694837e-05,
      "loss": 0.8383,
      "step": 185930
    },
    {
      "epoch": 3.264453378746116,
      "grad_norm": 0.05326749015172998,
      "learning_rate": 5.722185265796369e-05,
      "loss": 0.8454,
      "step": 185940
    },
    {
      "epoch": 3.264628943626117,
      "grad_norm": 0.045733688249148426,
      "learning_rate": 5.7211858046910127e-05,
      "loss": 0.8427,
      "step": 185950
    },
    {
      "epoch": 3.2648045085061184,
      "grad_norm": 0.07292807197525593,
      "learning_rate": 5.7201863990660656e-05,
      "loss": 0.8395,
      "step": 185960
    },
    {
      "epoch": 3.26498007338612,
      "grad_norm": 0.04517546821302732,
      "learning_rate": 5.719187048934196e-05,
      "loss": 0.8399,
      "step": 185970
    },
    {
      "epoch": 3.2651556382661213,
      "grad_norm": 0.04640257042157944,
      "learning_rate": 5.718187754308066e-05,
      "loss": 0.8374,
      "step": 185980
    },
    {
      "epoch": 3.265331203146123,
      "grad_norm": 0.05060230167559333,
      "learning_rate": 5.717188515200342e-05,
      "loss": 0.8408,
      "step": 185990
    },
    {
      "epoch": 3.265506768026124,
      "grad_norm": 0.10714126835793661,
      "learning_rate": 5.7161893316236735e-05,
      "loss": 0.8387,
      "step": 186000
    },
    {
      "epoch": 3.2656823329061253,
      "grad_norm": 0.05912232502399635,
      "learning_rate": 5.715190203590728e-05,
      "loss": 0.8441,
      "step": 186010
    },
    {
      "epoch": 3.265857897786127,
      "grad_norm": 0.05488336250416575,
      "learning_rate": 5.714191131114166e-05,
      "loss": 0.8534,
      "step": 186020
    },
    {
      "epoch": 3.2660334626661283,
      "grad_norm": 0.04264547981548531,
      "learning_rate": 5.7131921142066404e-05,
      "loss": 0.8434,
      "step": 186030
    },
    {
      "epoch": 3.2662090275461297,
      "grad_norm": 0.05368040174960821,
      "learning_rate": 5.71219315288081e-05,
      "loss": 0.8459,
      "step": 186040
    },
    {
      "epoch": 3.266384592426131,
      "grad_norm": 0.04745138027504958,
      "learning_rate": 5.711194247149334e-05,
      "loss": 0.8372,
      "step": 186050
    },
    {
      "epoch": 3.2665601573061327,
      "grad_norm": 0.05747660365542539,
      "learning_rate": 5.710195397024871e-05,
      "loss": 0.8499,
      "step": 186060
    },
    {
      "epoch": 3.2667357221861337,
      "grad_norm": 0.05273117752836502,
      "learning_rate": 5.7091966025200716e-05,
      "loss": 0.8407,
      "step": 186070
    },
    {
      "epoch": 3.266911287066135,
      "grad_norm": 0.057392042064361874,
      "learning_rate": 5.708197863647592e-05,
      "loss": 0.8397,
      "step": 186080
    },
    {
      "epoch": 3.2670868519461367,
      "grad_norm": 0.04270922918903289,
      "learning_rate": 5.70719918042009e-05,
      "loss": 0.8443,
      "step": 186090
    },
    {
      "epoch": 3.267262416826138,
      "grad_norm": 0.05988212214675811,
      "learning_rate": 5.7062005528502125e-05,
      "loss": 0.8447,
      "step": 186100
    },
    {
      "epoch": 3.2674379817061396,
      "grad_norm": 0.06919750214356499,
      "learning_rate": 5.705201980950616e-05,
      "loss": 0.8393,
      "step": 186110
    },
    {
      "epoch": 3.2676135465861407,
      "grad_norm": 0.06437598239554464,
      "learning_rate": 5.704203464733957e-05,
      "loss": 0.8426,
      "step": 186120
    },
    {
      "epoch": 3.267789111466142,
      "grad_norm": 0.052042929224019506,
      "learning_rate": 5.703205004212879e-05,
      "loss": 0.8377,
      "step": 186130
    },
    {
      "epoch": 3.2679646763461436,
      "grad_norm": 0.05800562966916374,
      "learning_rate": 5.702206599400036e-05,
      "loss": 0.8379,
      "step": 186140
    },
    {
      "epoch": 3.268140241226145,
      "grad_norm": 0.040337497412233844,
      "learning_rate": 5.70120825030808e-05,
      "loss": 0.8381,
      "step": 186150
    },
    {
      "epoch": 3.2683158061061466,
      "grad_norm": 0.04332642319684308,
      "learning_rate": 5.700209956949659e-05,
      "loss": 0.8402,
      "step": 186160
    },
    {
      "epoch": 3.268491370986148,
      "grad_norm": 0.07550159776425847,
      "learning_rate": 5.6992117193374185e-05,
      "loss": 0.835,
      "step": 186170
    },
    {
      "epoch": 3.2686669358661495,
      "grad_norm": 0.06588981891988491,
      "learning_rate": 5.698213537484009e-05,
      "loss": 0.8513,
      "step": 186180
    },
    {
      "epoch": 3.2688425007461506,
      "grad_norm": 0.05683352257201385,
      "learning_rate": 5.697215411402082e-05,
      "loss": 0.8412,
      "step": 186190
    },
    {
      "epoch": 3.269018065626152,
      "grad_norm": 0.05516928136513448,
      "learning_rate": 5.696217341104275e-05,
      "loss": 0.8447,
      "step": 186200
    },
    {
      "epoch": 3.2691936305061535,
      "grad_norm": 0.059858644678125156,
      "learning_rate": 5.695219326603238e-05,
      "loss": 0.8464,
      "step": 186210
    },
    {
      "epoch": 3.269369195386155,
      "grad_norm": 0.04493435228136795,
      "learning_rate": 5.6942213679116174e-05,
      "loss": 0.8481,
      "step": 186220
    },
    {
      "epoch": 3.2695447602661565,
      "grad_norm": 0.06737847812819675,
      "learning_rate": 5.6932234650420605e-05,
      "loss": 0.8389,
      "step": 186230
    },
    {
      "epoch": 3.2697203251461575,
      "grad_norm": 0.04573184302577004,
      "learning_rate": 5.692225618007202e-05,
      "loss": 0.8404,
      "step": 186240
    },
    {
      "epoch": 3.269895890026159,
      "grad_norm": 0.07596613053732061,
      "learning_rate": 5.6912278268196896e-05,
      "loss": 0.8383,
      "step": 186250
    },
    {
      "epoch": 3.2700714549061605,
      "grad_norm": 0.04443716154285498,
      "learning_rate": 5.69023009149217e-05,
      "loss": 0.8397,
      "step": 186260
    },
    {
      "epoch": 3.270247019786162,
      "grad_norm": 0.04190740081803386,
      "learning_rate": 5.689232412037275e-05,
      "loss": 0.8414,
      "step": 186270
    },
    {
      "epoch": 3.2704225846661634,
      "grad_norm": 0.054579444911523986,
      "learning_rate": 5.688234788467653e-05,
      "loss": 0.8374,
      "step": 186280
    },
    {
      "epoch": 3.270598149546165,
      "grad_norm": 0.06632244594265825,
      "learning_rate": 5.687237220795945e-05,
      "loss": 0.8474,
      "step": 186290
    },
    {
      "epoch": 3.2707737144261664,
      "grad_norm": 0.07288535196055378,
      "learning_rate": 5.686239709034782e-05,
      "loss": 0.8387,
      "step": 186300
    },
    {
      "epoch": 3.2709492793061674,
      "grad_norm": 0.05174284063290461,
      "learning_rate": 5.685242253196808e-05,
      "loss": 0.839,
      "step": 186310
    },
    {
      "epoch": 3.271124844186169,
      "grad_norm": 0.04985149082358937,
      "learning_rate": 5.684244853294661e-05,
      "loss": 0.8474,
      "step": 186320
    },
    {
      "epoch": 3.2713004090661704,
      "grad_norm": 0.056871477721104624,
      "learning_rate": 5.683247509340983e-05,
      "loss": 0.8454,
      "step": 186330
    },
    {
      "epoch": 3.271475973946172,
      "grad_norm": 0.05118480899487593,
      "learning_rate": 5.682250221348401e-05,
      "loss": 0.8419,
      "step": 186340
    },
    {
      "epoch": 3.2716515388261733,
      "grad_norm": 0.047785060797934396,
      "learning_rate": 5.681252989329554e-05,
      "loss": 0.8396,
      "step": 186350
    },
    {
      "epoch": 3.271827103706175,
      "grad_norm": 0.06505976987608493,
      "learning_rate": 5.680255813297084e-05,
      "loss": 0.8419,
      "step": 186360
    },
    {
      "epoch": 3.272002668586176,
      "grad_norm": 0.04981136808005781,
      "learning_rate": 5.679258693263615e-05,
      "loss": 0.8452,
      "step": 186370
    },
    {
      "epoch": 3.2721782334661773,
      "grad_norm": 0.04719631670591931,
      "learning_rate": 5.678261629241787e-05,
      "loss": 0.8431,
      "step": 186380
    },
    {
      "epoch": 3.272353798346179,
      "grad_norm": 0.05097143409899824,
      "learning_rate": 5.677264621244229e-05,
      "loss": 0.837,
      "step": 186390
    },
    {
      "epoch": 3.2725293632261803,
      "grad_norm": 0.06152522500817044,
      "learning_rate": 5.676267669283582e-05,
      "loss": 0.8369,
      "step": 186400
    },
    {
      "epoch": 3.2727049281061817,
      "grad_norm": 0.07401715811492679,
      "learning_rate": 5.675270773372466e-05,
      "loss": 0.8381,
      "step": 186410
    },
    {
      "epoch": 3.272880492986183,
      "grad_norm": 0.04902910509474474,
      "learning_rate": 5.6742739335235184e-05,
      "loss": 0.8416,
      "step": 186420
    },
    {
      "epoch": 3.2730560578661843,
      "grad_norm": 0.05008877873897101,
      "learning_rate": 5.6732771497493716e-05,
      "loss": 0.8376,
      "step": 186430
    },
    {
      "epoch": 3.2732316227461857,
      "grad_norm": 0.060160001375544614,
      "learning_rate": 5.672280422062649e-05,
      "loss": 0.8409,
      "step": 186440
    },
    {
      "epoch": 3.273407187626187,
      "grad_norm": 0.05394545861967373,
      "learning_rate": 5.6712837504759806e-05,
      "loss": 0.8385,
      "step": 186450
    },
    {
      "epoch": 3.2735827525061887,
      "grad_norm": 0.06255260132406686,
      "learning_rate": 5.6702871350020004e-05,
      "loss": 0.8396,
      "step": 186460
    },
    {
      "epoch": 3.27375831738619,
      "grad_norm": 0.08135779037612512,
      "learning_rate": 5.669290575653328e-05,
      "loss": 0.8394,
      "step": 186470
    },
    {
      "epoch": 3.2739338822661916,
      "grad_norm": 0.04792758217787204,
      "learning_rate": 5.668294072442591e-05,
      "loss": 0.8372,
      "step": 186480
    },
    {
      "epoch": 3.274109447146193,
      "grad_norm": 0.07117059403742455,
      "learning_rate": 5.6672976253824196e-05,
      "loss": 0.8433,
      "step": 186490
    },
    {
      "epoch": 3.274285012026194,
      "grad_norm": 0.05529422476062434,
      "learning_rate": 5.66630123448544e-05,
      "loss": 0.8456,
      "step": 186500
    },
    {
      "epoch": 3.2744605769061956,
      "grad_norm": 0.06301182521276676,
      "learning_rate": 5.6653048997642675e-05,
      "loss": 0.8469,
      "step": 186510
    },
    {
      "epoch": 3.274636141786197,
      "grad_norm": 0.05937348206901403,
      "learning_rate": 5.6643086212315356e-05,
      "loss": 0.8438,
      "step": 186520
    },
    {
      "epoch": 3.2748117066661986,
      "grad_norm": 0.04500833459840994,
      "learning_rate": 5.663312398899865e-05,
      "loss": 0.8452,
      "step": 186530
    },
    {
      "epoch": 3.2749872715462,
      "grad_norm": 0.0658713344003714,
      "learning_rate": 5.6623162327818725e-05,
      "loss": 0.8374,
      "step": 186540
    },
    {
      "epoch": 3.275162836426201,
      "grad_norm": 0.04816793249617194,
      "learning_rate": 5.6613201228901856e-05,
      "loss": 0.8408,
      "step": 186550
    },
    {
      "epoch": 3.2753384013062026,
      "grad_norm": 0.055147064490275816,
      "learning_rate": 5.660324069237423e-05,
      "loss": 0.8425,
      "step": 186560
    },
    {
      "epoch": 3.275513966186204,
      "grad_norm": 0.07943823095181865,
      "learning_rate": 5.659328071836209e-05,
      "loss": 0.8412,
      "step": 186570
    },
    {
      "epoch": 3.2756895310662055,
      "grad_norm": 0.05414803389455297,
      "learning_rate": 5.6583321306991535e-05,
      "loss": 0.8402,
      "step": 186580
    },
    {
      "epoch": 3.275865095946207,
      "grad_norm": 0.05359461834193573,
      "learning_rate": 5.657336245838883e-05,
      "loss": 0.8358,
      "step": 186590
    },
    {
      "epoch": 3.2760406608262085,
      "grad_norm": 0.06739409261872199,
      "learning_rate": 5.656340417268017e-05,
      "loss": 0.8475,
      "step": 186600
    },
    {
      "epoch": 3.27621622570621,
      "grad_norm": 0.05327922752061686,
      "learning_rate": 5.6553446449991666e-05,
      "loss": 0.844,
      "step": 186610
    },
    {
      "epoch": 3.276391790586211,
      "grad_norm": 0.05071849676249465,
      "learning_rate": 5.65434892904495e-05,
      "loss": 0.836,
      "step": 186620
    },
    {
      "epoch": 3.2765673554662125,
      "grad_norm": 0.0518113008222446,
      "learning_rate": 5.653353269417989e-05,
      "loss": 0.8473,
      "step": 186630
    },
    {
      "epoch": 3.276742920346214,
      "grad_norm": 0.048041868363595655,
      "learning_rate": 5.652357666130891e-05,
      "loss": 0.8423,
      "step": 186640
    },
    {
      "epoch": 3.2769184852262154,
      "grad_norm": 0.05802544414688372,
      "learning_rate": 5.651362119196273e-05,
      "loss": 0.8368,
      "step": 186650
    },
    {
      "epoch": 3.277094050106217,
      "grad_norm": 0.05219532630452618,
      "learning_rate": 5.650366628626751e-05,
      "loss": 0.8424,
      "step": 186660
    },
    {
      "epoch": 3.277269614986218,
      "grad_norm": 0.0478576208685752,
      "learning_rate": 5.649371194434937e-05,
      "loss": 0.8392,
      "step": 186670
    },
    {
      "epoch": 3.2774451798662194,
      "grad_norm": 0.05020620525257764,
      "learning_rate": 5.648375816633442e-05,
      "loss": 0.8394,
      "step": 186680
    },
    {
      "epoch": 3.277620744746221,
      "grad_norm": 0.04934485811918082,
      "learning_rate": 5.6473804952348774e-05,
      "loss": 0.8436,
      "step": 186690
    },
    {
      "epoch": 3.2777963096262224,
      "grad_norm": 0.052262518726053246,
      "learning_rate": 5.6463852302518605e-05,
      "loss": 0.8397,
      "step": 186700
    },
    {
      "epoch": 3.277971874506224,
      "grad_norm": 0.05392748669805188,
      "learning_rate": 5.6453900216969914e-05,
      "loss": 0.8414,
      "step": 186710
    },
    {
      "epoch": 3.2781474393862253,
      "grad_norm": 0.07390758702133021,
      "learning_rate": 5.644394869582885e-05,
      "loss": 0.8427,
      "step": 186720
    },
    {
      "epoch": 3.278323004266227,
      "grad_norm": 0.06998957417625541,
      "learning_rate": 5.6433997739221476e-05,
      "loss": 0.8508,
      "step": 186730
    },
    {
      "epoch": 3.278498569146228,
      "grad_norm": 0.050861824213740706,
      "learning_rate": 5.642404734727394e-05,
      "loss": 0.8388,
      "step": 186740
    },
    {
      "epoch": 3.2786741340262293,
      "grad_norm": 0.045019896911244334,
      "learning_rate": 5.6414097520112244e-05,
      "loss": 0.8446,
      "step": 186750
    },
    {
      "epoch": 3.278849698906231,
      "grad_norm": 0.07099640666796621,
      "learning_rate": 5.6404148257862466e-05,
      "loss": 0.8446,
      "step": 186760
    },
    {
      "epoch": 3.2790252637862323,
      "grad_norm": 0.04596960900844939,
      "learning_rate": 5.6394199560650715e-05,
      "loss": 0.8315,
      "step": 186770
    },
    {
      "epoch": 3.2792008286662337,
      "grad_norm": 0.04858793331571325,
      "learning_rate": 5.638425142860296e-05,
      "loss": 0.837,
      "step": 186780
    },
    {
      "epoch": 3.279376393546235,
      "grad_norm": 0.059003085191019176,
      "learning_rate": 5.63743038618453e-05,
      "loss": 0.836,
      "step": 186790
    },
    {
      "epoch": 3.2795519584262363,
      "grad_norm": 0.06592418383348052,
      "learning_rate": 5.63643568605038e-05,
      "loss": 0.8451,
      "step": 186800
    },
    {
      "epoch": 3.2797275233062377,
      "grad_norm": 0.052258781284756375,
      "learning_rate": 5.6354410424704405e-05,
      "loss": 0.8446,
      "step": 186810
    },
    {
      "epoch": 3.279903088186239,
      "grad_norm": 0.05133502109151789,
      "learning_rate": 5.63444645545732e-05,
      "loss": 0.8416,
      "step": 186820
    },
    {
      "epoch": 3.2800786530662407,
      "grad_norm": 0.04607504127947738,
      "learning_rate": 5.6334519250236165e-05,
      "loss": 0.8418,
      "step": 186830
    },
    {
      "epoch": 3.280254217946242,
      "grad_norm": 0.03988047650996541,
      "learning_rate": 5.6324574511819396e-05,
      "loss": 0.8445,
      "step": 186840
    },
    {
      "epoch": 3.2804297828262436,
      "grad_norm": 0.053973459554948054,
      "learning_rate": 5.631463033944877e-05,
      "loss": 0.8408,
      "step": 186850
    },
    {
      "epoch": 3.2806053477062447,
      "grad_norm": 0.0567703758472805,
      "learning_rate": 5.630468673325036e-05,
      "loss": 0.8481,
      "step": 186860
    },
    {
      "epoch": 3.280780912586246,
      "grad_norm": 0.06362483545459716,
      "learning_rate": 5.629474369335016e-05,
      "loss": 0.8396,
      "step": 186870
    },
    {
      "epoch": 3.2809564774662476,
      "grad_norm": 0.05822387893932798,
      "learning_rate": 5.628480121987408e-05,
      "loss": 0.8456,
      "step": 186880
    },
    {
      "epoch": 3.281132042346249,
      "grad_norm": 0.06803077550751671,
      "learning_rate": 5.627485931294816e-05,
      "loss": 0.8471,
      "step": 186890
    },
    {
      "epoch": 3.2813076072262506,
      "grad_norm": 0.0543322223501457,
      "learning_rate": 5.626491797269834e-05,
      "loss": 0.8514,
      "step": 186900
    },
    {
      "epoch": 3.2814831721062516,
      "grad_norm": 0.06566732779730333,
      "learning_rate": 5.6254977199250616e-05,
      "loss": 0.8437,
      "step": 186910
    },
    {
      "epoch": 3.281658736986253,
      "grad_norm": 0.05879802412360698,
      "learning_rate": 5.624503699273086e-05,
      "loss": 0.838,
      "step": 186920
    },
    {
      "epoch": 3.2818343018662546,
      "grad_norm": 0.05530500498224623,
      "learning_rate": 5.623509735326508e-05,
      "loss": 0.8394,
      "step": 186930
    },
    {
      "epoch": 3.282009866746256,
      "grad_norm": 0.0448121006421922,
      "learning_rate": 5.6225158280979216e-05,
      "loss": 0.8339,
      "step": 186940
    },
    {
      "epoch": 3.2821854316262575,
      "grad_norm": 0.05526595294773155,
      "learning_rate": 5.6215219775999154e-05,
      "loss": 0.8374,
      "step": 186950
    },
    {
      "epoch": 3.282360996506259,
      "grad_norm": 0.05875178378567468,
      "learning_rate": 5.6205281838450826e-05,
      "loss": 0.842,
      "step": 186960
    },
    {
      "epoch": 3.2825365613862605,
      "grad_norm": 0.05165983769745394,
      "learning_rate": 5.61953444684602e-05,
      "loss": 0.8364,
      "step": 186970
    },
    {
      "epoch": 3.2827121262662615,
      "grad_norm": 0.03917768271540712,
      "learning_rate": 5.6185407666153125e-05,
      "loss": 0.8294,
      "step": 186980
    },
    {
      "epoch": 3.282887691146263,
      "grad_norm": 0.06766608603252015,
      "learning_rate": 5.61754714316555e-05,
      "loss": 0.8416,
      "step": 186990
    },
    {
      "epoch": 3.2830632560262645,
      "grad_norm": 0.05180737567778106,
      "learning_rate": 5.616553576509327e-05,
      "loss": 0.8388,
      "step": 187000
    },
    {
      "epoch": 3.283238820906266,
      "grad_norm": 0.07207062357215023,
      "learning_rate": 5.615560066659231e-05,
      "loss": 0.8489,
      "step": 187010
    },
    {
      "epoch": 3.2834143857862674,
      "grad_norm": 0.04980844985412598,
      "learning_rate": 5.6145666136278454e-05,
      "loss": 0.844,
      "step": 187020
    },
    {
      "epoch": 3.283589950666269,
      "grad_norm": 0.05157749889057648,
      "learning_rate": 5.613573217427761e-05,
      "loss": 0.8405,
      "step": 187030
    },
    {
      "epoch": 3.28376551554627,
      "grad_norm": 0.04300456234405837,
      "learning_rate": 5.6125798780715667e-05,
      "loss": 0.8423,
      "step": 187040
    },
    {
      "epoch": 3.2839410804262714,
      "grad_norm": 0.05339047935064294,
      "learning_rate": 5.6115865955718415e-05,
      "loss": 0.841,
      "step": 187050
    },
    {
      "epoch": 3.284116645306273,
      "grad_norm": 0.06435475430387042,
      "learning_rate": 5.610593369941175e-05,
      "loss": 0.8464,
      "step": 187060
    },
    {
      "epoch": 3.2842922101862744,
      "grad_norm": 0.06517536003941378,
      "learning_rate": 5.609600201192151e-05,
      "loss": 0.8403,
      "step": 187070
    },
    {
      "epoch": 3.284467775066276,
      "grad_norm": 0.06384751095228076,
      "learning_rate": 5.608607089337357e-05,
      "loss": 0.8439,
      "step": 187080
    },
    {
      "epoch": 3.2846433399462773,
      "grad_norm": 0.06694685670947832,
      "learning_rate": 5.607614034389367e-05,
      "loss": 0.8373,
      "step": 187090
    },
    {
      "epoch": 3.2848189048262784,
      "grad_norm": 0.06250917105546237,
      "learning_rate": 5.6066210363607686e-05,
      "loss": 0.8429,
      "step": 187100
    },
    {
      "epoch": 3.28499446970628,
      "grad_norm": 0.05809934073776498,
      "learning_rate": 5.605628095264147e-05,
      "loss": 0.8393,
      "step": 187110
    },
    {
      "epoch": 3.2851700345862813,
      "grad_norm": 0.051564480287071335,
      "learning_rate": 5.6046352111120736e-05,
      "loss": 0.8369,
      "step": 187120
    },
    {
      "epoch": 3.285345599466283,
      "grad_norm": 0.05257378410118228,
      "learning_rate": 5.603642383917135e-05,
      "loss": 0.8351,
      "step": 187130
    },
    {
      "epoch": 3.2855211643462843,
      "grad_norm": 0.04573697934711422,
      "learning_rate": 5.602649613691914e-05,
      "loss": 0.8422,
      "step": 187140
    },
    {
      "epoch": 3.2856967292262858,
      "grad_norm": 0.04745146668131821,
      "learning_rate": 5.601656900448978e-05,
      "loss": 0.8458,
      "step": 187150
    },
    {
      "epoch": 3.2858722941062872,
      "grad_norm": 0.05172564139015285,
      "learning_rate": 5.600664244200913e-05,
      "loss": 0.8487,
      "step": 187160
    },
    {
      "epoch": 3.2860478589862883,
      "grad_norm": 0.04224056221335726,
      "learning_rate": 5.5996716449602936e-05,
      "loss": 0.838,
      "step": 187170
    },
    {
      "epoch": 3.2862234238662897,
      "grad_norm": 0.05339941364855989,
      "learning_rate": 5.598679102739701e-05,
      "loss": 0.8442,
      "step": 187180
    },
    {
      "epoch": 3.286398988746291,
      "grad_norm": 0.06465513143509148,
      "learning_rate": 5.597686617551702e-05,
      "loss": 0.8368,
      "step": 187190
    },
    {
      "epoch": 3.2865745536262927,
      "grad_norm": 0.07112539106117968,
      "learning_rate": 5.596694189408878e-05,
      "loss": 0.841,
      "step": 187200
    },
    {
      "epoch": 3.286750118506294,
      "grad_norm": 0.05811772430820987,
      "learning_rate": 5.5957018183238044e-05,
      "loss": 0.8381,
      "step": 187210
    },
    {
      "epoch": 3.286925683386295,
      "grad_norm": 0.059911351775666026,
      "learning_rate": 5.594709504309049e-05,
      "loss": 0.836,
      "step": 187220
    },
    {
      "epoch": 3.2871012482662967,
      "grad_norm": 0.04019579506089568,
      "learning_rate": 5.593717247377187e-05,
      "loss": 0.8407,
      "step": 187230
    },
    {
      "epoch": 3.287276813146298,
      "grad_norm": 0.06562877616535763,
      "learning_rate": 5.592725047540791e-05,
      "loss": 0.832,
      "step": 187240
    },
    {
      "epoch": 3.2874523780262996,
      "grad_norm": 0.054038886728761115,
      "learning_rate": 5.591732904812438e-05,
      "loss": 0.842,
      "step": 187250
    },
    {
      "epoch": 3.287627942906301,
      "grad_norm": 0.04747532471925827,
      "learning_rate": 5.590740819204686e-05,
      "loss": 0.8376,
      "step": 187260
    },
    {
      "epoch": 3.2878035077863026,
      "grad_norm": 0.05691852660089238,
      "learning_rate": 5.589748790730114e-05,
      "loss": 0.8369,
      "step": 187270
    },
    {
      "epoch": 3.287979072666304,
      "grad_norm": 0.053985330615695375,
      "learning_rate": 5.5887568194012926e-05,
      "loss": 0.8535,
      "step": 187280
    },
    {
      "epoch": 3.288154637546305,
      "grad_norm": 0.08561497074201281,
      "learning_rate": 5.587764905230784e-05,
      "loss": 0.8502,
      "step": 187290
    },
    {
      "epoch": 3.2883302024263066,
      "grad_norm": 0.048729366454349166,
      "learning_rate": 5.5867730482311574e-05,
      "loss": 0.8374,
      "step": 187300
    },
    {
      "epoch": 3.288505767306308,
      "grad_norm": 0.043693253782149,
      "learning_rate": 5.5857812484149876e-05,
      "loss": 0.8528,
      "step": 187310
    },
    {
      "epoch": 3.2886813321863095,
      "grad_norm": 0.043598563399509876,
      "learning_rate": 5.584789505794827e-05,
      "loss": 0.8476,
      "step": 187320
    },
    {
      "epoch": 3.288856897066311,
      "grad_norm": 0.04743562423430732,
      "learning_rate": 5.583797820383252e-05,
      "loss": 0.8486,
      "step": 187330
    },
    {
      "epoch": 3.289032461946312,
      "grad_norm": 0.07480102633707061,
      "learning_rate": 5.582806192192823e-05,
      "loss": 0.8351,
      "step": 187340
    },
    {
      "epoch": 3.2892080268263135,
      "grad_norm": 0.0589496036136965,
      "learning_rate": 5.581814621236108e-05,
      "loss": 0.8392,
      "step": 187350
    },
    {
      "epoch": 3.289383591706315,
      "grad_norm": 0.047323060663540716,
      "learning_rate": 5.5808231075256645e-05,
      "loss": 0.8405,
      "step": 187360
    },
    {
      "epoch": 3.2895591565863165,
      "grad_norm": 0.05233577978645099,
      "learning_rate": 5.579831651074059e-05,
      "loss": 0.8365,
      "step": 187370
    },
    {
      "epoch": 3.289734721466318,
      "grad_norm": 0.052980010620280356,
      "learning_rate": 5.5788402518938565e-05,
      "loss": 0.8434,
      "step": 187380
    },
    {
      "epoch": 3.2899102863463194,
      "grad_norm": 0.057080557268860736,
      "learning_rate": 5.57784890999761e-05,
      "loss": 0.8432,
      "step": 187390
    },
    {
      "epoch": 3.290085851226321,
      "grad_norm": 0.05177662294074474,
      "learning_rate": 5.576857625397885e-05,
      "loss": 0.8452,
      "step": 187400
    },
    {
      "epoch": 3.290261416106322,
      "grad_norm": 0.06230624225761757,
      "learning_rate": 5.57586639810724e-05,
      "loss": 0.8425,
      "step": 187410
    },
    {
      "epoch": 3.2904369809863234,
      "grad_norm": 0.07163364382135151,
      "learning_rate": 5.5748752281382386e-05,
      "loss": 0.8434,
      "step": 187420
    },
    {
      "epoch": 3.290612545866325,
      "grad_norm": 0.058888944629568216,
      "learning_rate": 5.573884115503432e-05,
      "loss": 0.844,
      "step": 187430
    },
    {
      "epoch": 3.2907881107463264,
      "grad_norm": 0.04435441844953733,
      "learning_rate": 5.572893060215381e-05,
      "loss": 0.8472,
      "step": 187440
    },
    {
      "epoch": 3.290963675626328,
      "grad_norm": 0.056817234790947975,
      "learning_rate": 5.571902062286647e-05,
      "loss": 0.8343,
      "step": 187450
    },
    {
      "epoch": 3.291139240506329,
      "grad_norm": 0.058192974020247513,
      "learning_rate": 5.570911121729777e-05,
      "loss": 0.8393,
      "step": 187460
    },
    {
      "epoch": 3.2913148053863304,
      "grad_norm": 0.054780104354581784,
      "learning_rate": 5.5699202385573323e-05,
      "loss": 0.8394,
      "step": 187470
    },
    {
      "epoch": 3.291490370266332,
      "grad_norm": 0.04258755678862435,
      "learning_rate": 5.568929412781868e-05,
      "loss": 0.8397,
      "step": 187480
    },
    {
      "epoch": 3.2916659351463333,
      "grad_norm": 0.07068096033592025,
      "learning_rate": 5.567938644415934e-05,
      "loss": 0.8377,
      "step": 187490
    },
    {
      "epoch": 3.291841500026335,
      "grad_norm": 0.04762966940001732,
      "learning_rate": 5.566947933472087e-05,
      "loss": 0.8393,
      "step": 187500
    },
    {
      "epoch": 3.2920170649063363,
      "grad_norm": 0.05147168437291164,
      "learning_rate": 5.565957279962877e-05,
      "loss": 0.8334,
      "step": 187510
    },
    {
      "epoch": 3.2921926297863378,
      "grad_norm": 0.04866774534855943,
      "learning_rate": 5.5649666839008595e-05,
      "loss": 0.8391,
      "step": 187520
    },
    {
      "epoch": 3.292368194666339,
      "grad_norm": 0.049839357051387354,
      "learning_rate": 5.5639761452985816e-05,
      "loss": 0.8435,
      "step": 187530
    },
    {
      "epoch": 3.2925437595463403,
      "grad_norm": 0.04756246915714735,
      "learning_rate": 5.562985664168595e-05,
      "loss": 0.8428,
      "step": 187540
    },
    {
      "epoch": 3.2927193244263417,
      "grad_norm": 0.07230124188213527,
      "learning_rate": 5.561995240523454e-05,
      "loss": 0.8404,
      "step": 187550
    },
    {
      "epoch": 3.292894889306343,
      "grad_norm": 0.06563483938459268,
      "learning_rate": 5.561004874375698e-05,
      "loss": 0.8414,
      "step": 187560
    },
    {
      "epoch": 3.2930704541863447,
      "grad_norm": 0.06421044620034856,
      "learning_rate": 5.560014565737879e-05,
      "loss": 0.8521,
      "step": 187570
    },
    {
      "epoch": 3.2932460190663457,
      "grad_norm": 0.06412570745770863,
      "learning_rate": 5.559024314622547e-05,
      "loss": 0.8423,
      "step": 187580
    },
    {
      "epoch": 3.293421583946347,
      "grad_norm": 0.06947587687510147,
      "learning_rate": 5.558034121042251e-05,
      "loss": 0.8367,
      "step": 187590
    },
    {
      "epoch": 3.2935971488263487,
      "grad_norm": 0.0445788494465031,
      "learning_rate": 5.55704398500953e-05,
      "loss": 0.8352,
      "step": 187600
    },
    {
      "epoch": 3.29377271370635,
      "grad_norm": 0.04701610852906054,
      "learning_rate": 5.556053906536932e-05,
      "loss": 0.8383,
      "step": 187610
    },
    {
      "epoch": 3.2939482785863516,
      "grad_norm": 0.04396656528689304,
      "learning_rate": 5.555063885637005e-05,
      "loss": 0.836,
      "step": 187620
    },
    {
      "epoch": 3.294123843466353,
      "grad_norm": 0.044531779515699456,
      "learning_rate": 5.554073922322287e-05,
      "loss": 0.8386,
      "step": 187630
    },
    {
      "epoch": 3.2942994083463546,
      "grad_norm": 0.06523604040448505,
      "learning_rate": 5.553084016605322e-05,
      "loss": 0.8334,
      "step": 187640
    },
    {
      "epoch": 3.2944749732263556,
      "grad_norm": 0.054391157492333565,
      "learning_rate": 5.552094168498659e-05,
      "loss": 0.8434,
      "step": 187650
    },
    {
      "epoch": 3.294650538106357,
      "grad_norm": 0.06794470102647865,
      "learning_rate": 5.55110437801483e-05,
      "loss": 0.8388,
      "step": 187660
    },
    {
      "epoch": 3.2948261029863586,
      "grad_norm": 0.06395131913370213,
      "learning_rate": 5.550114645166381e-05,
      "loss": 0.836,
      "step": 187670
    },
    {
      "epoch": 3.29500166786636,
      "grad_norm": 0.051100393988954045,
      "learning_rate": 5.54912496996585e-05,
      "loss": 0.8373,
      "step": 187680
    },
    {
      "epoch": 3.2951772327463615,
      "grad_norm": 0.04508244158402082,
      "learning_rate": 5.548135352425783e-05,
      "loss": 0.8393,
      "step": 187690
    },
    {
      "epoch": 3.2953527976263626,
      "grad_norm": 0.04809491347075099,
      "learning_rate": 5.54714579255871e-05,
      "loss": 0.8428,
      "step": 187700
    },
    {
      "epoch": 3.295528362506364,
      "grad_norm": 0.05651575290776306,
      "learning_rate": 5.5461562903771714e-05,
      "loss": 0.8411,
      "step": 187710
    },
    {
      "epoch": 3.2957039273863655,
      "grad_norm": 0.07140438188809123,
      "learning_rate": 5.545166845893709e-05,
      "loss": 0.8434,
      "step": 187720
    },
    {
      "epoch": 3.295879492266367,
      "grad_norm": 0.0644732535545731,
      "learning_rate": 5.5441774591208516e-05,
      "loss": 0.8528,
      "step": 187730
    },
    {
      "epoch": 3.2960550571463685,
      "grad_norm": 0.048182932065583314,
      "learning_rate": 5.543188130071141e-05,
      "loss": 0.8348,
      "step": 187740
    },
    {
      "epoch": 3.29623062202637,
      "grad_norm": 0.06174170822029935,
      "learning_rate": 5.542198858757108e-05,
      "loss": 0.8387,
      "step": 187750
    },
    {
      "epoch": 3.2964061869063714,
      "grad_norm": 0.05262396585819535,
      "learning_rate": 5.541209645191294e-05,
      "loss": 0.8448,
      "step": 187760
    },
    {
      "epoch": 3.2965817517863725,
      "grad_norm": 0.06528234692301411,
      "learning_rate": 5.540220489386223e-05,
      "loss": 0.8436,
      "step": 187770
    },
    {
      "epoch": 3.296757316666374,
      "grad_norm": 0.06034157626796728,
      "learning_rate": 5.539231391354432e-05,
      "loss": 0.8465,
      "step": 187780
    },
    {
      "epoch": 3.2969328815463754,
      "grad_norm": 0.04854458807302771,
      "learning_rate": 5.538242351108458e-05,
      "loss": 0.8428,
      "step": 187790
    },
    {
      "epoch": 3.297108446426377,
      "grad_norm": 0.07987941133072252,
      "learning_rate": 5.537253368660823e-05,
      "loss": 0.8446,
      "step": 187800
    },
    {
      "epoch": 3.2972840113063784,
      "grad_norm": 0.05619849914791502,
      "learning_rate": 5.5362644440240615e-05,
      "loss": 0.8322,
      "step": 187810
    },
    {
      "epoch": 3.29745957618638,
      "grad_norm": 0.06032507748696715,
      "learning_rate": 5.535275577210709e-05,
      "loss": 0.8446,
      "step": 187820
    },
    {
      "epoch": 3.297635141066381,
      "grad_norm": 0.06506744894887932,
      "learning_rate": 5.5342867682332864e-05,
      "loss": 0.8444,
      "step": 187830
    },
    {
      "epoch": 3.2978107059463824,
      "grad_norm": 0.04639452535362073,
      "learning_rate": 5.5332980171043246e-05,
      "loss": 0.8438,
      "step": 187840
    },
    {
      "epoch": 3.297986270826384,
      "grad_norm": 0.0544268192383158,
      "learning_rate": 5.532309323836352e-05,
      "loss": 0.8449,
      "step": 187850
    },
    {
      "epoch": 3.2981618357063853,
      "grad_norm": 0.06568795883600453,
      "learning_rate": 5.531320688441901e-05,
      "loss": 0.8403,
      "step": 187860
    },
    {
      "epoch": 3.298337400586387,
      "grad_norm": 0.07151434283656274,
      "learning_rate": 5.530332110933487e-05,
      "loss": 0.8403,
      "step": 187870
    },
    {
      "epoch": 3.2985129654663883,
      "grad_norm": 0.057430523569984356,
      "learning_rate": 5.529343591323642e-05,
      "loss": 0.835,
      "step": 187880
    },
    {
      "epoch": 3.2986885303463893,
      "grad_norm": 0.07499020170784415,
      "learning_rate": 5.528355129624894e-05,
      "loss": 0.8484,
      "step": 187890
    },
    {
      "epoch": 3.298864095226391,
      "grad_norm": 0.052709835002322965,
      "learning_rate": 5.527366725849757e-05,
      "loss": 0.8492,
      "step": 187900
    },
    {
      "epoch": 3.2990396601063923,
      "grad_norm": 0.05314284873977334,
      "learning_rate": 5.5263783800107604e-05,
      "loss": 0.8401,
      "step": 187910
    },
    {
      "epoch": 3.2992152249863937,
      "grad_norm": 0.0560056525624995,
      "learning_rate": 5.525390092120427e-05,
      "loss": 0.8408,
      "step": 187920
    },
    {
      "epoch": 3.2993907898663952,
      "grad_norm": 0.05679686001759922,
      "learning_rate": 5.524401862191282e-05,
      "loss": 0.8412,
      "step": 187930
    },
    {
      "epoch": 3.2995663547463967,
      "grad_norm": 0.05376549932183391,
      "learning_rate": 5.5234136902358367e-05,
      "loss": 0.8463,
      "step": 187940
    },
    {
      "epoch": 3.299741919626398,
      "grad_norm": 0.05932417203569761,
      "learning_rate": 5.522425576266618e-05,
      "loss": 0.8386,
      "step": 187950
    },
    {
      "epoch": 3.299917484506399,
      "grad_norm": 0.05493975897743994,
      "learning_rate": 5.521437520296149e-05,
      "loss": 0.8437,
      "step": 187960
    },
    {
      "epoch": 3.3000930493864007,
      "grad_norm": 0.04395804117943977,
      "learning_rate": 5.520449522336939e-05,
      "loss": 0.8342,
      "step": 187970
    },
    {
      "epoch": 3.300268614266402,
      "grad_norm": 0.04788884595394651,
      "learning_rate": 5.519461582401514e-05,
      "loss": 0.844,
      "step": 187980
    },
    {
      "epoch": 3.3004441791464036,
      "grad_norm": 0.06029444771406243,
      "learning_rate": 5.518473700502389e-05,
      "loss": 0.8416,
      "step": 187990
    },
    {
      "epoch": 3.300619744026405,
      "grad_norm": 0.06004690076690455,
      "learning_rate": 5.517485876652079e-05,
      "loss": 0.8392,
      "step": 188000
    },
    {
      "epoch": 3.300795308906406,
      "grad_norm": 0.08131966526544993,
      "learning_rate": 5.5164981108630985e-05,
      "loss": 0.8404,
      "step": 188010
    },
    {
      "epoch": 3.3009708737864076,
      "grad_norm": 0.04102072496427912,
      "learning_rate": 5.5155104031479675e-05,
      "loss": 0.845,
      "step": 188020
    },
    {
      "epoch": 3.301146438666409,
      "grad_norm": 0.07342626265554952,
      "learning_rate": 5.5145227535192024e-05,
      "loss": 0.8402,
      "step": 188030
    },
    {
      "epoch": 3.3013220035464106,
      "grad_norm": 0.05559361838417433,
      "learning_rate": 5.513535161989308e-05,
      "loss": 0.8338,
      "step": 188040
    },
    {
      "epoch": 3.301497568426412,
      "grad_norm": 0.04871103020926843,
      "learning_rate": 5.512547628570801e-05,
      "loss": 0.8393,
      "step": 188050
    },
    {
      "epoch": 3.3016731333064135,
      "grad_norm": 0.06311042656764082,
      "learning_rate": 5.511560153276201e-05,
      "loss": 0.8505,
      "step": 188060
    },
    {
      "epoch": 3.301848698186415,
      "grad_norm": 0.05437375004412733,
      "learning_rate": 5.510572736118007e-05,
      "loss": 0.8461,
      "step": 188070
    },
    {
      "epoch": 3.302024263066416,
      "grad_norm": 0.04468874028624708,
      "learning_rate": 5.509585377108738e-05,
      "loss": 0.8469,
      "step": 188080
    },
    {
      "epoch": 3.3021998279464175,
      "grad_norm": 0.05498567128129493,
      "learning_rate": 5.508598076260901e-05,
      "loss": 0.8437,
      "step": 188090
    },
    {
      "epoch": 3.302375392826419,
      "grad_norm": 0.051416343435312484,
      "learning_rate": 5.5076108335870113e-05,
      "loss": 0.8492,
      "step": 188100
    },
    {
      "epoch": 3.3025509577064205,
      "grad_norm": 0.061728196146596144,
      "learning_rate": 5.506623649099567e-05,
      "loss": 0.8463,
      "step": 188110
    },
    {
      "epoch": 3.302726522586422,
      "grad_norm": 0.06758036418116091,
      "learning_rate": 5.505636522811082e-05,
      "loss": 0.8312,
      "step": 188120
    },
    {
      "epoch": 3.302902087466423,
      "grad_norm": 0.051310275221777175,
      "learning_rate": 5.504649454734067e-05,
      "loss": 0.8389,
      "step": 188130
    },
    {
      "epoch": 3.3030776523464245,
      "grad_norm": 0.05643359995370061,
      "learning_rate": 5.5036624448810205e-05,
      "loss": 0.847,
      "step": 188140
    },
    {
      "epoch": 3.303253217226426,
      "grad_norm": 0.05141366009232395,
      "learning_rate": 5.5026754932644506e-05,
      "loss": 0.8364,
      "step": 188150
    },
    {
      "epoch": 3.3034287821064274,
      "grad_norm": 0.06397890437337943,
      "learning_rate": 5.501688599896868e-05,
      "loss": 0.8362,
      "step": 188160
    },
    {
      "epoch": 3.303604346986429,
      "grad_norm": 0.06519832606933068,
      "learning_rate": 5.5007017647907675e-05,
      "loss": 0.8384,
      "step": 188170
    },
    {
      "epoch": 3.3037799118664304,
      "grad_norm": 0.04711253071247089,
      "learning_rate": 5.499714987958658e-05,
      "loss": 0.8475,
      "step": 188180
    },
    {
      "epoch": 3.303955476746432,
      "grad_norm": 0.04964133054691046,
      "learning_rate": 5.4987282694130394e-05,
      "loss": 0.8451,
      "step": 188190
    },
    {
      "epoch": 3.304131041626433,
      "grad_norm": 0.084973666899913,
      "learning_rate": 5.4977416091664215e-05,
      "loss": 0.8439,
      "step": 188200
    },
    {
      "epoch": 3.3043066065064344,
      "grad_norm": 0.04522139758956372,
      "learning_rate": 5.4967550072312946e-05,
      "loss": 0.8483,
      "step": 188210
    },
    {
      "epoch": 3.304482171386436,
      "grad_norm": 0.05026725464349196,
      "learning_rate": 5.495768463620164e-05,
      "loss": 0.8322,
      "step": 188220
    },
    {
      "epoch": 3.3046577362664373,
      "grad_norm": 0.055182119301050415,
      "learning_rate": 5.494781978345533e-05,
      "loss": 0.8419,
      "step": 188230
    },
    {
      "epoch": 3.304833301146439,
      "grad_norm": 0.04988263399875774,
      "learning_rate": 5.4937955514198945e-05,
      "loss": 0.8467,
      "step": 188240
    },
    {
      "epoch": 3.30500886602644,
      "grad_norm": 0.05697099996958386,
      "learning_rate": 5.4928091828557474e-05,
      "loss": 0.8471,
      "step": 188250
    },
    {
      "epoch": 3.3051844309064413,
      "grad_norm": 0.08001571268247012,
      "learning_rate": 5.491822872665592e-05,
      "loss": 0.8457,
      "step": 188260
    },
    {
      "epoch": 3.305359995786443,
      "grad_norm": 0.049720849319713996,
      "learning_rate": 5.490836620861928e-05,
      "loss": 0.8367,
      "step": 188270
    },
    {
      "epoch": 3.3055355606664443,
      "grad_norm": 0.04756646101456504,
      "learning_rate": 5.489850427457242e-05,
      "loss": 0.84,
      "step": 188280
    },
    {
      "epoch": 3.3057111255464457,
      "grad_norm": 0.05604126297435017,
      "learning_rate": 5.488864292464035e-05,
      "loss": 0.8356,
      "step": 188290
    },
    {
      "epoch": 3.3058866904264472,
      "grad_norm": 0.04921828146842735,
      "learning_rate": 5.487878215894805e-05,
      "loss": 0.8359,
      "step": 188300
    },
    {
      "epoch": 3.3060622553064487,
      "grad_norm": 0.05226327186080261,
      "learning_rate": 5.486892197762039e-05,
      "loss": 0.8348,
      "step": 188310
    },
    {
      "epoch": 3.3062378201864497,
      "grad_norm": 0.042764462068133426,
      "learning_rate": 5.4859062380782334e-05,
      "loss": 0.8417,
      "step": 188320
    },
    {
      "epoch": 3.306413385066451,
      "grad_norm": 0.05555471820821586,
      "learning_rate": 5.484920336855882e-05,
      "loss": 0.8449,
      "step": 188330
    },
    {
      "epoch": 3.3065889499464527,
      "grad_norm": 0.06159032520665692,
      "learning_rate": 5.483934494107472e-05,
      "loss": 0.8424,
      "step": 188340
    },
    {
      "epoch": 3.306764514826454,
      "grad_norm": 0.15324502131012205,
      "learning_rate": 5.4829487098454954e-05,
      "loss": 0.8476,
      "step": 188350
    },
    {
      "epoch": 3.3069400797064556,
      "grad_norm": 0.04583010517474038,
      "learning_rate": 5.481962984082445e-05,
      "loss": 0.8462,
      "step": 188360
    },
    {
      "epoch": 3.3071156445864567,
      "grad_norm": 0.08822450243125395,
      "learning_rate": 5.480977316830812e-05,
      "loss": 0.8474,
      "step": 188370
    },
    {
      "epoch": 3.307291209466458,
      "grad_norm": 0.04316264767426342,
      "learning_rate": 5.479991708103076e-05,
      "loss": 0.8522,
      "step": 188380
    },
    {
      "epoch": 3.3074667743464596,
      "grad_norm": 0.06403100813518539,
      "learning_rate": 5.479006157911732e-05,
      "loss": 0.8424,
      "step": 188390
    },
    {
      "epoch": 3.307642339226461,
      "grad_norm": 0.047957866186606246,
      "learning_rate": 5.4780206662692704e-05,
      "loss": 0.85,
      "step": 188400
    },
    {
      "epoch": 3.3078179041064626,
      "grad_norm": 0.05718167784153297,
      "learning_rate": 5.4770352331881666e-05,
      "loss": 0.8429,
      "step": 188410
    },
    {
      "epoch": 3.307993468986464,
      "grad_norm": 0.052272537413278336,
      "learning_rate": 5.476049858680916e-05,
      "loss": 0.843,
      "step": 188420
    },
    {
      "epoch": 3.3081690338664655,
      "grad_norm": 0.05953647945218844,
      "learning_rate": 5.4750645427599966e-05,
      "loss": 0.8431,
      "step": 188430
    },
    {
      "epoch": 3.3083445987464666,
      "grad_norm": 0.057026947449312436,
      "learning_rate": 5.474079285437901e-05,
      "loss": 0.837,
      "step": 188440
    },
    {
      "epoch": 3.308520163626468,
      "grad_norm": 0.052060077258774366,
      "learning_rate": 5.473094086727103e-05,
      "loss": 0.8421,
      "step": 188450
    },
    {
      "epoch": 3.3086957285064695,
      "grad_norm": 0.05036635386672244,
      "learning_rate": 5.47210894664009e-05,
      "loss": 0.8437,
      "step": 188460
    },
    {
      "epoch": 3.308871293386471,
      "grad_norm": 0.05032288033514849,
      "learning_rate": 5.471123865189348e-05,
      "loss": 0.8388,
      "step": 188470
    },
    {
      "epoch": 3.3090468582664725,
      "grad_norm": 0.051495101115482944,
      "learning_rate": 5.470138842387348e-05,
      "loss": 0.839,
      "step": 188480
    },
    {
      "epoch": 3.309222423146474,
      "grad_norm": 0.06676249421004482,
      "learning_rate": 5.469153878246579e-05,
      "loss": 0.8432,
      "step": 188490
    },
    {
      "epoch": 3.309397988026475,
      "grad_norm": 0.08454936650951901,
      "learning_rate": 5.4681689727795196e-05,
      "loss": 0.8458,
      "step": 188500
    },
    {
      "epoch": 3.3095735529064765,
      "grad_norm": 0.06627754149352143,
      "learning_rate": 5.4671841259986436e-05,
      "loss": 0.8415,
      "step": 188510
    },
    {
      "epoch": 3.309749117786478,
      "grad_norm": 0.056798327583904185,
      "learning_rate": 5.4661993379164324e-05,
      "loss": 0.8393,
      "step": 188520
    },
    {
      "epoch": 3.3099246826664794,
      "grad_norm": 0.056225789135752526,
      "learning_rate": 5.4652146085453634e-05,
      "loss": 0.8352,
      "step": 188530
    },
    {
      "epoch": 3.310100247546481,
      "grad_norm": 0.053054641897268696,
      "learning_rate": 5.464229937897918e-05,
      "loss": 0.8416,
      "step": 188540
    },
    {
      "epoch": 3.3102758124264824,
      "grad_norm": 0.04578637421801843,
      "learning_rate": 5.463245325986564e-05,
      "loss": 0.846,
      "step": 188550
    },
    {
      "epoch": 3.3104513773064834,
      "grad_norm": 0.05618874778033527,
      "learning_rate": 5.4622607728237796e-05,
      "loss": 0.8358,
      "step": 188560
    },
    {
      "epoch": 3.310626942186485,
      "grad_norm": 0.045572131678131636,
      "learning_rate": 5.461276278422044e-05,
      "loss": 0.8454,
      "step": 188570
    },
    {
      "epoch": 3.3108025070664864,
      "grad_norm": 0.062378806211410905,
      "learning_rate": 5.4602918427938244e-05,
      "loss": 0.8431,
      "step": 188580
    },
    {
      "epoch": 3.310978071946488,
      "grad_norm": 0.05486832935237594,
      "learning_rate": 5.459307465951595e-05,
      "loss": 0.8297,
      "step": 188590
    },
    {
      "epoch": 3.3111536368264893,
      "grad_norm": 0.04182835393099748,
      "learning_rate": 5.45832314790783e-05,
      "loss": 0.8493,
      "step": 188600
    },
    {
      "epoch": 3.311329201706491,
      "grad_norm": 0.050063225082640224,
      "learning_rate": 5.4573388886750047e-05,
      "loss": 0.846,
      "step": 188610
    },
    {
      "epoch": 3.3115047665864923,
      "grad_norm": 0.056759901545790664,
      "learning_rate": 5.4563546882655805e-05,
      "loss": 0.8451,
      "step": 188620
    },
    {
      "epoch": 3.3116803314664933,
      "grad_norm": 0.05808854068190132,
      "learning_rate": 5.455370546692031e-05,
      "loss": 0.8369,
      "step": 188630
    },
    {
      "epoch": 3.311855896346495,
      "grad_norm": 0.0604057319236312,
      "learning_rate": 5.454386463966833e-05,
      "loss": 0.8404,
      "step": 188640
    },
    {
      "epoch": 3.3120314612264963,
      "grad_norm": 0.047283536613036005,
      "learning_rate": 5.4534024401024444e-05,
      "loss": 0.8402,
      "step": 188650
    },
    {
      "epoch": 3.3122070261064978,
      "grad_norm": 0.055530492580010374,
      "learning_rate": 5.452418475111337e-05,
      "loss": 0.8367,
      "step": 188660
    },
    {
      "epoch": 3.3123825909864992,
      "grad_norm": 0.06039194314943583,
      "learning_rate": 5.451434569005983e-05,
      "loss": 0.8333,
      "step": 188670
    },
    {
      "epoch": 3.3125581558665003,
      "grad_norm": 0.05511140426925219,
      "learning_rate": 5.45045072179884e-05,
      "loss": 0.8367,
      "step": 188680
    },
    {
      "epoch": 3.3127337207465017,
      "grad_norm": 0.03725313901020002,
      "learning_rate": 5.449466933502377e-05,
      "loss": 0.8391,
      "step": 188690
    },
    {
      "epoch": 3.312909285626503,
      "grad_norm": 0.05636281408770627,
      "learning_rate": 5.44848320412906e-05,
      "loss": 0.8323,
      "step": 188700
    },
    {
      "epoch": 3.3130848505065047,
      "grad_norm": 0.060125247999110024,
      "learning_rate": 5.447499533691354e-05,
      "loss": 0.8388,
      "step": 188710
    },
    {
      "epoch": 3.313260415386506,
      "grad_norm": 0.05268548256624648,
      "learning_rate": 5.4465159222017184e-05,
      "loss": 0.8456,
      "step": 188720
    },
    {
      "epoch": 3.3134359802665077,
      "grad_norm": 0.04131191914627482,
      "learning_rate": 5.445532369672617e-05,
      "loss": 0.8545,
      "step": 188730
    },
    {
      "epoch": 3.313611545146509,
      "grad_norm": 0.048795539056496945,
      "learning_rate": 5.444548876116518e-05,
      "loss": 0.8394,
      "step": 188740
    },
    {
      "epoch": 3.31378711002651,
      "grad_norm": 0.058874618046789125,
      "learning_rate": 5.4435654415458716e-05,
      "loss": 0.8359,
      "step": 188750
    },
    {
      "epoch": 3.3139626749065116,
      "grad_norm": 0.04601977044907312,
      "learning_rate": 5.442582065973143e-05,
      "loss": 0.8379,
      "step": 188760
    },
    {
      "epoch": 3.314138239786513,
      "grad_norm": 0.05417863926246087,
      "learning_rate": 5.4415987494107935e-05,
      "loss": 0.8415,
      "step": 188770
    },
    {
      "epoch": 3.3143138046665146,
      "grad_norm": 0.05493349372701165,
      "learning_rate": 5.4406154918712834e-05,
      "loss": 0.8371,
      "step": 188780
    },
    {
      "epoch": 3.314489369546516,
      "grad_norm": 0.06620077437303802,
      "learning_rate": 5.439632293367065e-05,
      "loss": 0.8477,
      "step": 188790
    },
    {
      "epoch": 3.314664934426517,
      "grad_norm": 0.04750718154760229,
      "learning_rate": 5.4386491539105985e-05,
      "loss": 0.8383,
      "step": 188800
    },
    {
      "epoch": 3.3148404993065186,
      "grad_norm": 0.059073424154455455,
      "learning_rate": 5.437666073514345e-05,
      "loss": 0.8318,
      "step": 188810
    },
    {
      "epoch": 3.31501606418652,
      "grad_norm": 0.05709885731914001,
      "learning_rate": 5.436683052190751e-05,
      "loss": 0.8415,
      "step": 188820
    },
    {
      "epoch": 3.3151916290665215,
      "grad_norm": 0.059300552819550545,
      "learning_rate": 5.435700089952277e-05,
      "loss": 0.8412,
      "step": 188830
    },
    {
      "epoch": 3.315367193946523,
      "grad_norm": 0.061249559101184375,
      "learning_rate": 5.434717186811379e-05,
      "loss": 0.8473,
      "step": 188840
    },
    {
      "epoch": 3.3155427588265245,
      "grad_norm": 0.06903902508222046,
      "learning_rate": 5.4337343427805066e-05,
      "loss": 0.8364,
      "step": 188850
    },
    {
      "epoch": 3.315718323706526,
      "grad_norm": 0.04797787417009455,
      "learning_rate": 5.432751557872113e-05,
      "loss": 0.8481,
      "step": 188860
    },
    {
      "epoch": 3.315893888586527,
      "grad_norm": 0.07498501927355607,
      "learning_rate": 5.431768832098653e-05,
      "loss": 0.8358,
      "step": 188870
    },
    {
      "epoch": 3.3160694534665285,
      "grad_norm": 0.04884192028284671,
      "learning_rate": 5.43078616547258e-05,
      "loss": 0.8454,
      "step": 188880
    },
    {
      "epoch": 3.31624501834653,
      "grad_norm": 0.06717861959420066,
      "learning_rate": 5.4298035580063365e-05,
      "loss": 0.8451,
      "step": 188890
    },
    {
      "epoch": 3.3164205832265314,
      "grad_norm": 0.0527923751689476,
      "learning_rate": 5.428821009712377e-05,
      "loss": 0.8407,
      "step": 188900
    },
    {
      "epoch": 3.316596148106533,
      "grad_norm": 0.050328404856147084,
      "learning_rate": 5.4278385206031546e-05,
      "loss": 0.8346,
      "step": 188910
    },
    {
      "epoch": 3.316771712986534,
      "grad_norm": 0.08458518255591137,
      "learning_rate": 5.4268560906911104e-05,
      "loss": 0.844,
      "step": 188920
    },
    {
      "epoch": 3.3169472778665354,
      "grad_norm": 0.04848740103133456,
      "learning_rate": 5.4258737199886945e-05,
      "loss": 0.8371,
      "step": 188930
    },
    {
      "epoch": 3.317122842746537,
      "grad_norm": 0.06572282645962602,
      "learning_rate": 5.4248914085083545e-05,
      "loss": 0.8438,
      "step": 188940
    },
    {
      "epoch": 3.3172984076265384,
      "grad_norm": 0.0654673482313044,
      "learning_rate": 5.42390915626254e-05,
      "loss": 0.8368,
      "step": 188950
    },
    {
      "epoch": 3.31747397250654,
      "grad_norm": 0.0421083466636067,
      "learning_rate": 5.4229269632636895e-05,
      "loss": 0.8348,
      "step": 188960
    },
    {
      "epoch": 3.3176495373865413,
      "grad_norm": 0.053279282748816874,
      "learning_rate": 5.421944829524249e-05,
      "loss": 0.8384,
      "step": 188970
    },
    {
      "epoch": 3.317825102266543,
      "grad_norm": 0.05817711172375571,
      "learning_rate": 5.42096275505667e-05,
      "loss": 0.8372,
      "step": 188980
    },
    {
      "epoch": 3.318000667146544,
      "grad_norm": 0.05513395960740993,
      "learning_rate": 5.419980739873385e-05,
      "loss": 0.8461,
      "step": 188990
    },
    {
      "epoch": 3.3181762320265453,
      "grad_norm": 0.04824082898075808,
      "learning_rate": 5.41899878398684e-05,
      "loss": 0.8494,
      "step": 189000
    },
    {
      "epoch": 3.318351796906547,
      "grad_norm": 0.048454872104198556,
      "learning_rate": 5.418016887409483e-05,
      "loss": 0.834,
      "step": 189010
    },
    {
      "epoch": 3.3185273617865483,
      "grad_norm": 0.08192917005410971,
      "learning_rate": 5.417035050153744e-05,
      "loss": 0.8373,
      "step": 189020
    },
    {
      "epoch": 3.3187029266665498,
      "grad_norm": 0.06369356676404798,
      "learning_rate": 5.41605327223207e-05,
      "loss": 0.8491,
      "step": 189030
    },
    {
      "epoch": 3.318878491546551,
      "grad_norm": 0.06513602748356759,
      "learning_rate": 5.4150715536568976e-05,
      "loss": 0.8421,
      "step": 189040
    },
    {
      "epoch": 3.3190540564265523,
      "grad_norm": 0.054113222621474506,
      "learning_rate": 5.414089894440669e-05,
      "loss": 0.8362,
      "step": 189050
    },
    {
      "epoch": 3.3192296213065537,
      "grad_norm": 0.053822506781832924,
      "learning_rate": 5.413108294595818e-05,
      "loss": 0.8379,
      "step": 189060
    },
    {
      "epoch": 3.3194051861865552,
      "grad_norm": 0.05543233504069285,
      "learning_rate": 5.4121267541347806e-05,
      "loss": 0.8426,
      "step": 189070
    },
    {
      "epoch": 3.3195807510665567,
      "grad_norm": 0.0533414617152652,
      "learning_rate": 5.411145273070003e-05,
      "loss": 0.8373,
      "step": 189080
    },
    {
      "epoch": 3.319756315946558,
      "grad_norm": 0.07302750529827116,
      "learning_rate": 5.4101638514139086e-05,
      "loss": 0.8459,
      "step": 189090
    },
    {
      "epoch": 3.3199318808265597,
      "grad_norm": 0.06545672586754138,
      "learning_rate": 5.409182489178936e-05,
      "loss": 0.8471,
      "step": 189100
    },
    {
      "epoch": 3.3201074457065607,
      "grad_norm": 0.06358831117921779,
      "learning_rate": 5.408201186377521e-05,
      "loss": 0.8442,
      "step": 189110
    },
    {
      "epoch": 3.320283010586562,
      "grad_norm": 0.04446808951434463,
      "learning_rate": 5.407219943022098e-05,
      "loss": 0.8412,
      "step": 189120
    },
    {
      "epoch": 3.3204585754665636,
      "grad_norm": 0.04170255084503563,
      "learning_rate": 5.406238759125096e-05,
      "loss": 0.8541,
      "step": 189130
    },
    {
      "epoch": 3.320634140346565,
      "grad_norm": 0.058720881702952525,
      "learning_rate": 5.405257634698948e-05,
      "loss": 0.8473,
      "step": 189140
    },
    {
      "epoch": 3.3208097052265666,
      "grad_norm": 0.04900992965230126,
      "learning_rate": 5.404276569756088e-05,
      "loss": 0.8346,
      "step": 189150
    },
    {
      "epoch": 3.3209852701065676,
      "grad_norm": 0.051157792139762584,
      "learning_rate": 5.403295564308941e-05,
      "loss": 0.8376,
      "step": 189160
    },
    {
      "epoch": 3.321160834986569,
      "grad_norm": 0.058274119978565235,
      "learning_rate": 5.402314618369939e-05,
      "loss": 0.8378,
      "step": 189170
    },
    {
      "epoch": 3.3213363998665706,
      "grad_norm": 0.056457136316671576,
      "learning_rate": 5.4013337319515154e-05,
      "loss": 0.8406,
      "step": 189180
    },
    {
      "epoch": 3.321511964746572,
      "grad_norm": 0.05123784757207901,
      "learning_rate": 5.40035290506609e-05,
      "loss": 0.8449,
      "step": 189190
    },
    {
      "epoch": 3.3216875296265735,
      "grad_norm": 0.061720709450122285,
      "learning_rate": 5.399372137726093e-05,
      "loss": 0.8339,
      "step": 189200
    },
    {
      "epoch": 3.321863094506575,
      "grad_norm": 0.06595297885761715,
      "learning_rate": 5.3983914299439545e-05,
      "loss": 0.8342,
      "step": 189210
    },
    {
      "epoch": 3.3220386593865765,
      "grad_norm": 0.053472812420902675,
      "learning_rate": 5.3974107817320984e-05,
      "loss": 0.8426,
      "step": 189220
    },
    {
      "epoch": 3.3222142242665775,
      "grad_norm": 0.05603873009800822,
      "learning_rate": 5.396430193102947e-05,
      "loss": 0.847,
      "step": 189230
    },
    {
      "epoch": 3.322389789146579,
      "grad_norm": 0.06136008889119566,
      "learning_rate": 5.3954496640689266e-05,
      "loss": 0.8434,
      "step": 189240
    },
    {
      "epoch": 3.3225653540265805,
      "grad_norm": 0.06339201948613171,
      "learning_rate": 5.394469194642464e-05,
      "loss": 0.8331,
      "step": 189250
    },
    {
      "epoch": 3.322740918906582,
      "grad_norm": 0.055206687958636216,
      "learning_rate": 5.393488784835974e-05,
      "loss": 0.8441,
      "step": 189260
    },
    {
      "epoch": 3.3229164837865834,
      "grad_norm": 0.05963300779271633,
      "learning_rate": 5.392508434661884e-05,
      "loss": 0.8421,
      "step": 189270
    },
    {
      "epoch": 3.323092048666585,
      "grad_norm": 0.05399994955703744,
      "learning_rate": 5.391528144132614e-05,
      "loss": 0.8509,
      "step": 189280
    },
    {
      "epoch": 3.323267613546586,
      "grad_norm": 0.052015646258120316,
      "learning_rate": 5.390547913260587e-05,
      "loss": 0.8372,
      "step": 189290
    },
    {
      "epoch": 3.3234431784265874,
      "grad_norm": 0.051091587337023375,
      "learning_rate": 5.389567742058218e-05,
      "loss": 0.8364,
      "step": 189300
    },
    {
      "epoch": 3.323618743306589,
      "grad_norm": 0.04287546359278542,
      "learning_rate": 5.3885876305379304e-05,
      "loss": 0.8295,
      "step": 189310
    },
    {
      "epoch": 3.3237943081865904,
      "grad_norm": 0.07410245842997132,
      "learning_rate": 5.3876075787121415e-05,
      "loss": 0.8401,
      "step": 189320
    },
    {
      "epoch": 3.323969873066592,
      "grad_norm": 0.054410826188134094,
      "learning_rate": 5.3866275865932646e-05,
      "loss": 0.8394,
      "step": 189330
    },
    {
      "epoch": 3.3241454379465933,
      "grad_norm": 0.052568098634607235,
      "learning_rate": 5.3856476541937205e-05,
      "loss": 0.8424,
      "step": 189340
    },
    {
      "epoch": 3.3243210028265944,
      "grad_norm": 0.04291832350882859,
      "learning_rate": 5.3846677815259275e-05,
      "loss": 0.8523,
      "step": 189350
    },
    {
      "epoch": 3.324496567706596,
      "grad_norm": 0.05405553118128843,
      "learning_rate": 5.3836879686022937e-05,
      "loss": 0.8342,
      "step": 189360
    },
    {
      "epoch": 3.3246721325865973,
      "grad_norm": 0.061975073643158796,
      "learning_rate": 5.382708215435236e-05,
      "loss": 0.8404,
      "step": 189370
    },
    {
      "epoch": 3.324847697466599,
      "grad_norm": 0.05529494695958789,
      "learning_rate": 5.3817285220371704e-05,
      "loss": 0.8399,
      "step": 189380
    },
    {
      "epoch": 3.3250232623466003,
      "grad_norm": 0.05661619103709557,
      "learning_rate": 5.380748888420513e-05,
      "loss": 0.8337,
      "step": 189390
    },
    {
      "epoch": 3.3251988272266018,
      "grad_norm": 0.04571081633459759,
      "learning_rate": 5.379769314597667e-05,
      "loss": 0.8433,
      "step": 189400
    },
    {
      "epoch": 3.3253743921066032,
      "grad_norm": 0.052782579383291645,
      "learning_rate": 5.378789800581048e-05,
      "loss": 0.8361,
      "step": 189410
    },
    {
      "epoch": 3.3255499569866043,
      "grad_norm": 0.060984903364383856,
      "learning_rate": 5.377810346383071e-05,
      "loss": 0.8326,
      "step": 189420
    },
    {
      "epoch": 3.3257255218666057,
      "grad_norm": 0.0474245269928329,
      "learning_rate": 5.3768309520161384e-05,
      "loss": 0.8489,
      "step": 189430
    },
    {
      "epoch": 3.3259010867466072,
      "grad_norm": 0.049169609492749816,
      "learning_rate": 5.375851617492662e-05,
      "loss": 0.8475,
      "step": 189440
    },
    {
      "epoch": 3.3260766516266087,
      "grad_norm": 0.06616896023461705,
      "learning_rate": 5.3748723428250526e-05,
      "loss": 0.8426,
      "step": 189450
    },
    {
      "epoch": 3.32625221650661,
      "grad_norm": 0.07214393967453883,
      "learning_rate": 5.373893128025719e-05,
      "loss": 0.8418,
      "step": 189460
    },
    {
      "epoch": 3.326427781386611,
      "grad_norm": 0.04828367173674802,
      "learning_rate": 5.3729139731070613e-05,
      "loss": 0.844,
      "step": 189470
    },
    {
      "epoch": 3.3266033462666127,
      "grad_norm": 0.05187869822466638,
      "learning_rate": 5.371934878081489e-05,
      "loss": 0.8376,
      "step": 189480
    },
    {
      "epoch": 3.326778911146614,
      "grad_norm": 0.08937429387272719,
      "learning_rate": 5.370955842961414e-05,
      "loss": 0.8441,
      "step": 189490
    },
    {
      "epoch": 3.3269544760266156,
      "grad_norm": 0.04589367674415919,
      "learning_rate": 5.369976867759229e-05,
      "loss": 0.8357,
      "step": 189500
    },
    {
      "epoch": 3.327130040906617,
      "grad_norm": 0.05881932738410323,
      "learning_rate": 5.3689979524873444e-05,
      "loss": 0.8347,
      "step": 189510
    },
    {
      "epoch": 3.3273056057866186,
      "grad_norm": 0.07506330402387329,
      "learning_rate": 5.3680190971581655e-05,
      "loss": 0.8417,
      "step": 189520
    },
    {
      "epoch": 3.32748117066662,
      "grad_norm": 0.05809387842191893,
      "learning_rate": 5.367040301784089e-05,
      "loss": 0.8464,
      "step": 189530
    },
    {
      "epoch": 3.327656735546621,
      "grad_norm": 0.051913550978840706,
      "learning_rate": 5.3660615663775174e-05,
      "loss": 0.8353,
      "step": 189540
    },
    {
      "epoch": 3.3278323004266226,
      "grad_norm": 0.058541221617859435,
      "learning_rate": 5.365082890950855e-05,
      "loss": 0.8374,
      "step": 189550
    },
    {
      "epoch": 3.328007865306624,
      "grad_norm": 0.04991979491652878,
      "learning_rate": 5.364104275516502e-05,
      "loss": 0.8474,
      "step": 189560
    },
    {
      "epoch": 3.3281834301866255,
      "grad_norm": 0.050092345538252155,
      "learning_rate": 5.3631257200868515e-05,
      "loss": 0.8376,
      "step": 189570
    },
    {
      "epoch": 3.328358995066627,
      "grad_norm": 0.05093500833387768,
      "learning_rate": 5.362147224674307e-05,
      "loss": 0.8382,
      "step": 189580
    },
    {
      "epoch": 3.328534559946628,
      "grad_norm": 0.04623444043714597,
      "learning_rate": 5.3611687892912675e-05,
      "loss": 0.8368,
      "step": 189590
    },
    {
      "epoch": 3.3287101248266295,
      "grad_norm": 0.05594380336900717,
      "learning_rate": 5.360190413950125e-05,
      "loss": 0.8477,
      "step": 189600
    },
    {
      "epoch": 3.328885689706631,
      "grad_norm": 0.055031810040797866,
      "learning_rate": 5.359212098663278e-05,
      "loss": 0.8341,
      "step": 189610
    },
    {
      "epoch": 3.3290612545866325,
      "grad_norm": 0.0633188913679895,
      "learning_rate": 5.358233843443121e-05,
      "loss": 0.8402,
      "step": 189620
    },
    {
      "epoch": 3.329236819466634,
      "grad_norm": 0.042233698634467944,
      "learning_rate": 5.357255648302056e-05,
      "loss": 0.8364,
      "step": 189630
    },
    {
      "epoch": 3.3294123843466354,
      "grad_norm": 0.058998416770341344,
      "learning_rate": 5.356277513252467e-05,
      "loss": 0.8445,
      "step": 189640
    },
    {
      "epoch": 3.329587949226637,
      "grad_norm": 0.05074396786073412,
      "learning_rate": 5.355299438306749e-05,
      "loss": 0.842,
      "step": 189650
    },
    {
      "epoch": 3.329763514106638,
      "grad_norm": 0.0515964670616467,
      "learning_rate": 5.354321423477302e-05,
      "loss": 0.8369,
      "step": 189660
    },
    {
      "epoch": 3.3299390789866394,
      "grad_norm": 0.08425231340152768,
      "learning_rate": 5.353343468776508e-05,
      "loss": 0.8445,
      "step": 189670
    },
    {
      "epoch": 3.330114643866641,
      "grad_norm": 0.05592324098765087,
      "learning_rate": 5.352365574216762e-05,
      "loss": 0.8419,
      "step": 189680
    },
    {
      "epoch": 3.3302902087466424,
      "grad_norm": 0.05881557292760229,
      "learning_rate": 5.351387739810457e-05,
      "loss": 0.8405,
      "step": 189690
    },
    {
      "epoch": 3.330465773626644,
      "grad_norm": 0.15209186171572872,
      "learning_rate": 5.350409965569977e-05,
      "loss": 0.8361,
      "step": 189700
    },
    {
      "epoch": 3.330641338506645,
      "grad_norm": 0.05342823258494497,
      "learning_rate": 5.349432251507711e-05,
      "loss": 0.8389,
      "step": 189710
    },
    {
      "epoch": 3.3308169033866464,
      "grad_norm": 0.054560227889959505,
      "learning_rate": 5.34845459763605e-05,
      "loss": 0.841,
      "step": 189720
    },
    {
      "epoch": 3.330992468266648,
      "grad_norm": 0.047304482433185885,
      "learning_rate": 5.347477003967385e-05,
      "loss": 0.8407,
      "step": 189730
    },
    {
      "epoch": 3.3311680331466493,
      "grad_norm": 0.07737420694291107,
      "learning_rate": 5.3464994705140916e-05,
      "loss": 0.8435,
      "step": 189740
    },
    {
      "epoch": 3.331343598026651,
      "grad_norm": 0.0489716429813751,
      "learning_rate": 5.3455219972885605e-05,
      "loss": 0.8382,
      "step": 189750
    },
    {
      "epoch": 3.3315191629066523,
      "grad_norm": 0.060309394509053676,
      "learning_rate": 5.3445445843031826e-05,
      "loss": 0.839,
      "step": 189760
    },
    {
      "epoch": 3.3316947277866538,
      "grad_norm": 0.06146597429660073,
      "learning_rate": 5.343567231570331e-05,
      "loss": 0.8488,
      "step": 189770
    },
    {
      "epoch": 3.331870292666655,
      "grad_norm": 0.052332608740482706,
      "learning_rate": 5.3425899391023934e-05,
      "loss": 0.8452,
      "step": 189780
    },
    {
      "epoch": 3.3320458575466563,
      "grad_norm": 0.052166687694256954,
      "learning_rate": 5.341612706911754e-05,
      "loss": 0.838,
      "step": 189790
    },
    {
      "epoch": 3.3322214224266578,
      "grad_norm": 0.058188840410620425,
      "learning_rate": 5.3406355350107985e-05,
      "loss": 0.8313,
      "step": 189800
    },
    {
      "epoch": 3.3323969873066592,
      "grad_norm": 0.05235209430918068,
      "learning_rate": 5.339658423411897e-05,
      "loss": 0.8392,
      "step": 189810
    },
    {
      "epoch": 3.3325725521866607,
      "grad_norm": 0.04241793755208619,
      "learning_rate": 5.338681372127436e-05,
      "loss": 0.8434,
      "step": 189820
    },
    {
      "epoch": 3.3327481170666617,
      "grad_norm": 0.06655877848287461,
      "learning_rate": 5.337704381169799e-05,
      "loss": 0.8432,
      "step": 189830
    },
    {
      "epoch": 3.332923681946663,
      "grad_norm": 0.07108427566079556,
      "learning_rate": 5.336727450551357e-05,
      "loss": 0.8357,
      "step": 189840
    },
    {
      "epoch": 3.3330992468266647,
      "grad_norm": 0.04987518463202046,
      "learning_rate": 5.335750580284491e-05,
      "loss": 0.8448,
      "step": 189850
    },
    {
      "epoch": 3.333274811706666,
      "grad_norm": 0.04154132645543485,
      "learning_rate": 5.3347737703815816e-05,
      "loss": 0.8419,
      "step": 189860
    },
    {
      "epoch": 3.3334503765866677,
      "grad_norm": 0.08962594739910905,
      "learning_rate": 5.333797020854997e-05,
      "loss": 0.8463,
      "step": 189870
    },
    {
      "epoch": 3.333625941466669,
      "grad_norm": 0.05071444583396208,
      "learning_rate": 5.332820331717118e-05,
      "loss": 0.8462,
      "step": 189880
    },
    {
      "epoch": 3.3338015063466706,
      "grad_norm": 0.04588064239476597,
      "learning_rate": 5.3318437029803195e-05,
      "loss": 0.8382,
      "step": 189890
    },
    {
      "epoch": 3.3339770712266716,
      "grad_norm": 0.04863125218759956,
      "learning_rate": 5.330867134656979e-05,
      "loss": 0.8483,
      "step": 189900
    },
    {
      "epoch": 3.334152636106673,
      "grad_norm": 0.05696931732027708,
      "learning_rate": 5.329890626759462e-05,
      "loss": 0.8483,
      "step": 189910
    },
    {
      "epoch": 3.3343282009866746,
      "grad_norm": 0.07352158734629959,
      "learning_rate": 5.328914179300145e-05,
      "loss": 0.8408,
      "step": 189920
    },
    {
      "epoch": 3.334503765866676,
      "grad_norm": 0.05071599183702518,
      "learning_rate": 5.327937792291403e-05,
      "loss": 0.8404,
      "step": 189930
    },
    {
      "epoch": 3.3346793307466776,
      "grad_norm": 0.05133499665837229,
      "learning_rate": 5.326961465745601e-05,
      "loss": 0.842,
      "step": 189940
    },
    {
      "epoch": 3.334854895626679,
      "grad_norm": 0.051371254314752535,
      "learning_rate": 5.325985199675111e-05,
      "loss": 0.8394,
      "step": 189950
    },
    {
      "epoch": 3.33503046050668,
      "grad_norm": 0.08268500224919204,
      "learning_rate": 5.325008994092303e-05,
      "loss": 0.8463,
      "step": 189960
    },
    {
      "epoch": 3.3352060253866815,
      "grad_norm": 0.04446796982015695,
      "learning_rate": 5.3240328490095515e-05,
      "loss": 0.8443,
      "step": 189970
    },
    {
      "epoch": 3.335381590266683,
      "grad_norm": 0.061943274937434026,
      "learning_rate": 5.323056764439214e-05,
      "loss": 0.8412,
      "step": 189980
    },
    {
      "epoch": 3.3355571551466845,
      "grad_norm": 0.05319647359601944,
      "learning_rate": 5.3220807403936626e-05,
      "loss": 0.8494,
      "step": 189990
    },
    {
      "epoch": 3.335732720026686,
      "grad_norm": 0.05282210511550332,
      "learning_rate": 5.32110477688527e-05,
      "loss": 0.8363,
      "step": 190000
    },
    {
      "epoch": 3.3359082849066874,
      "grad_norm": 0.046447780115372594,
      "learning_rate": 5.32012887392639e-05,
      "loss": 0.838,
      "step": 190010
    },
    {
      "epoch": 3.3360838497866885,
      "grad_norm": 0.04729651221073054,
      "learning_rate": 5.3191530315293936e-05,
      "loss": 0.8415,
      "step": 190020
    },
    {
      "epoch": 3.33625941466669,
      "grad_norm": 0.04726186868815279,
      "learning_rate": 5.318177249706648e-05,
      "loss": 0.841,
      "step": 190030
    },
    {
      "epoch": 3.3364349795466914,
      "grad_norm": 0.0476636587605162,
      "learning_rate": 5.31720152847051e-05,
      "loss": 0.8458,
      "step": 190040
    },
    {
      "epoch": 3.336610544426693,
      "grad_norm": 0.045243972266877625,
      "learning_rate": 5.316225867833344e-05,
      "loss": 0.8417,
      "step": 190050
    },
    {
      "epoch": 3.3367861093066944,
      "grad_norm": 0.056886256271409204,
      "learning_rate": 5.315250267807513e-05,
      "loss": 0.839,
      "step": 190060
    },
    {
      "epoch": 3.336961674186696,
      "grad_norm": 0.04747029295674506,
      "learning_rate": 5.314274728405384e-05,
      "loss": 0.8388,
      "step": 190070
    },
    {
      "epoch": 3.3371372390666973,
      "grad_norm": 0.0670165965096357,
      "learning_rate": 5.3132992496393067e-05,
      "loss": 0.8449,
      "step": 190080
    },
    {
      "epoch": 3.3373128039466984,
      "grad_norm": 0.058131009093983256,
      "learning_rate": 5.312323831521644e-05,
      "loss": 0.8451,
      "step": 190090
    },
    {
      "epoch": 3.3374883688267,
      "grad_norm": 0.05304944685507681,
      "learning_rate": 5.311348474064761e-05,
      "loss": 0.8461,
      "step": 190100
    },
    {
      "epoch": 3.3376639337067013,
      "grad_norm": 0.04787622415486119,
      "learning_rate": 5.3103731772810075e-05,
      "loss": 0.8403,
      "step": 190110
    },
    {
      "epoch": 3.337839498586703,
      "grad_norm": 0.0583091607691032,
      "learning_rate": 5.309397941182741e-05,
      "loss": 0.8431,
      "step": 190120
    },
    {
      "epoch": 3.3380150634667043,
      "grad_norm": 0.038664385069156586,
      "learning_rate": 5.3084227657823225e-05,
      "loss": 0.8401,
      "step": 190130
    },
    {
      "epoch": 3.3381906283467053,
      "grad_norm": 0.04429642436867699,
      "learning_rate": 5.30744765109211e-05,
      "loss": 0.8447,
      "step": 190140
    },
    {
      "epoch": 3.338366193226707,
      "grad_norm": 0.04546357881050222,
      "learning_rate": 5.30647259712445e-05,
      "loss": 0.8407,
      "step": 190150
    },
    {
      "epoch": 3.3385417581067083,
      "grad_norm": 0.04580341044081266,
      "learning_rate": 5.3054976038917004e-05,
      "loss": 0.8403,
      "step": 190160
    },
    {
      "epoch": 3.3387173229867098,
      "grad_norm": 0.07604852658180039,
      "learning_rate": 5.30452267140622e-05,
      "loss": 0.8401,
      "step": 190170
    },
    {
      "epoch": 3.3388928878667112,
      "grad_norm": 0.05110613396996473,
      "learning_rate": 5.3035477996803505e-05,
      "loss": 0.8416,
      "step": 190180
    },
    {
      "epoch": 3.3390684527467127,
      "grad_norm": 0.05576829332538855,
      "learning_rate": 5.302572988726452e-05,
      "loss": 0.8439,
      "step": 190190
    },
    {
      "epoch": 3.339244017626714,
      "grad_norm": 0.046203485194732635,
      "learning_rate": 5.301598238556876e-05,
      "loss": 0.8362,
      "step": 190200
    },
    {
      "epoch": 3.3394195825067152,
      "grad_norm": 0.05018518535401116,
      "learning_rate": 5.300623549183966e-05,
      "loss": 0.851,
      "step": 190210
    },
    {
      "epoch": 3.3395951473867167,
      "grad_norm": 0.04505992159164537,
      "learning_rate": 5.299648920620076e-05,
      "loss": 0.846,
      "step": 190220
    },
    {
      "epoch": 3.339770712266718,
      "grad_norm": 0.038548419386683595,
      "learning_rate": 5.298674352877554e-05,
      "loss": 0.852,
      "step": 190230
    },
    {
      "epoch": 3.3399462771467197,
      "grad_norm": 0.08414073355293826,
      "learning_rate": 5.297699845968751e-05,
      "loss": 0.8421,
      "step": 190240
    },
    {
      "epoch": 3.340121842026721,
      "grad_norm": 0.05143087740026186,
      "learning_rate": 5.296725399906008e-05,
      "loss": 0.8419,
      "step": 190250
    },
    {
      "epoch": 3.340297406906722,
      "grad_norm": 0.048932662743289876,
      "learning_rate": 5.2957510147016754e-05,
      "loss": 0.846,
      "step": 190260
    },
    {
      "epoch": 3.3404729717867236,
      "grad_norm": 0.05859023747864608,
      "learning_rate": 5.294776690368101e-05,
      "loss": 0.8438,
      "step": 190270
    },
    {
      "epoch": 3.340648536666725,
      "grad_norm": 0.045353086826743216,
      "learning_rate": 5.2938024269176235e-05,
      "loss": 0.8409,
      "step": 190280
    },
    {
      "epoch": 3.3408241015467266,
      "grad_norm": 0.06091381668762484,
      "learning_rate": 5.2928282243625903e-05,
      "loss": 0.8379,
      "step": 190290
    },
    {
      "epoch": 3.340999666426728,
      "grad_norm": 0.04438870252642902,
      "learning_rate": 5.291854082715344e-05,
      "loss": 0.84,
      "step": 190300
    },
    {
      "epoch": 3.3411752313067296,
      "grad_norm": 0.04228208870922415,
      "learning_rate": 5.2908800019882334e-05,
      "loss": 0.8494,
      "step": 190310
    },
    {
      "epoch": 3.341350796186731,
      "grad_norm": 0.05961112335370485,
      "learning_rate": 5.28990598219359e-05,
      "loss": 0.8468,
      "step": 190320
    },
    {
      "epoch": 3.341526361066732,
      "grad_norm": 0.04534522105995815,
      "learning_rate": 5.28893202334376e-05,
      "loss": 0.8432,
      "step": 190330
    },
    {
      "epoch": 3.3417019259467335,
      "grad_norm": 0.05820723112193159,
      "learning_rate": 5.2879581254510865e-05,
      "loss": 0.8458,
      "step": 190340
    },
    {
      "epoch": 3.341877490826735,
      "grad_norm": 0.05112437194470751,
      "learning_rate": 5.286984288527903e-05,
      "loss": 0.8422,
      "step": 190350
    },
    {
      "epoch": 3.3420530557067365,
      "grad_norm": 0.0506952051762276,
      "learning_rate": 5.286010512586549e-05,
      "loss": 0.8418,
      "step": 190360
    },
    {
      "epoch": 3.342228620586738,
      "grad_norm": 0.04695314931791569,
      "learning_rate": 5.28503679763937e-05,
      "loss": 0.8353,
      "step": 190370
    },
    {
      "epoch": 3.342404185466739,
      "grad_norm": 0.05116800971079988,
      "learning_rate": 5.284063143698693e-05,
      "loss": 0.8442,
      "step": 190380
    },
    {
      "epoch": 3.3425797503467405,
      "grad_norm": 0.06382008732688994,
      "learning_rate": 5.2830895507768594e-05,
      "loss": 0.8379,
      "step": 190390
    },
    {
      "epoch": 3.342755315226742,
      "grad_norm": 0.05178294238931887,
      "learning_rate": 5.282116018886204e-05,
      "loss": 0.8364,
      "step": 190400
    },
    {
      "epoch": 3.3429308801067434,
      "grad_norm": 0.0470718146503585,
      "learning_rate": 5.2811425480390666e-05,
      "loss": 0.8375,
      "step": 190410
    },
    {
      "epoch": 3.343106444986745,
      "grad_norm": 0.04482021346731024,
      "learning_rate": 5.280169138247772e-05,
      "loss": 0.838,
      "step": 190420
    },
    {
      "epoch": 3.3432820098667464,
      "grad_norm": 0.05917915432872709,
      "learning_rate": 5.279195789524659e-05,
      "loss": 0.8342,
      "step": 190430
    },
    {
      "epoch": 3.343457574746748,
      "grad_norm": 0.06094321028442564,
      "learning_rate": 5.278222501882062e-05,
      "loss": 0.842,
      "step": 190440
    },
    {
      "epoch": 3.343633139626749,
      "grad_norm": 0.05343697851931588,
      "learning_rate": 5.2772492753323065e-05,
      "loss": 0.8438,
      "step": 190450
    },
    {
      "epoch": 3.3438087045067504,
      "grad_norm": 0.08207301718168192,
      "learning_rate": 5.2762761098877275e-05,
      "loss": 0.8436,
      "step": 190460
    },
    {
      "epoch": 3.343984269386752,
      "grad_norm": 0.04615646457066099,
      "learning_rate": 5.2753030055606544e-05,
      "loss": 0.838,
      "step": 190470
    },
    {
      "epoch": 3.3441598342667533,
      "grad_norm": 0.050337120050456645,
      "learning_rate": 5.274329962363421e-05,
      "loss": 0.8414,
      "step": 190480
    },
    {
      "epoch": 3.344335399146755,
      "grad_norm": 0.05075653627797945,
      "learning_rate": 5.273356980308347e-05,
      "loss": 0.8405,
      "step": 190490
    },
    {
      "epoch": 3.344510964026756,
      "grad_norm": 0.05324663515433989,
      "learning_rate": 5.272384059407766e-05,
      "loss": 0.8454,
      "step": 190500
    },
    {
      "epoch": 3.3446865289067573,
      "grad_norm": 0.04708662835511205,
      "learning_rate": 5.2714111996740065e-05,
      "loss": 0.846,
      "step": 190510
    },
    {
      "epoch": 3.344862093786759,
      "grad_norm": 0.04709992750829385,
      "learning_rate": 5.2704384011193906e-05,
      "loss": 0.8441,
      "step": 190520
    },
    {
      "epoch": 3.3450376586667603,
      "grad_norm": 0.05859445984691485,
      "learning_rate": 5.269465663756245e-05,
      "loss": 0.8367,
      "step": 190530
    },
    {
      "epoch": 3.3452132235467618,
      "grad_norm": 0.0498424374163307,
      "learning_rate": 5.268492987596899e-05,
      "loss": 0.8294,
      "step": 190540
    },
    {
      "epoch": 3.3453887884267632,
      "grad_norm": 0.04690352600700953,
      "learning_rate": 5.2675203726536685e-05,
      "loss": 0.8429,
      "step": 190550
    },
    {
      "epoch": 3.3455643533067647,
      "grad_norm": 0.04527921221620908,
      "learning_rate": 5.266547818938881e-05,
      "loss": 0.8357,
      "step": 190560
    },
    {
      "epoch": 3.3457399181867657,
      "grad_norm": 0.05587445940761521,
      "learning_rate": 5.26557532646486e-05,
      "loss": 0.8392,
      "step": 190570
    },
    {
      "epoch": 3.3459154830667672,
      "grad_norm": 0.043927965482848955,
      "learning_rate": 5.2646028952439286e-05,
      "loss": 0.8311,
      "step": 190580
    },
    {
      "epoch": 3.3460910479467687,
      "grad_norm": 0.05227686305626216,
      "learning_rate": 5.263630525288402e-05,
      "loss": 0.8421,
      "step": 190590
    },
    {
      "epoch": 3.34626661282677,
      "grad_norm": 0.06055686577473273,
      "learning_rate": 5.262658216610604e-05,
      "loss": 0.8376,
      "step": 190600
    },
    {
      "epoch": 3.3464421777067717,
      "grad_norm": 0.06115875606685929,
      "learning_rate": 5.261685969222857e-05,
      "loss": 0.8436,
      "step": 190610
    },
    {
      "epoch": 3.3466177425867727,
      "grad_norm": 0.06754793487751942,
      "learning_rate": 5.260713783137472e-05,
      "loss": 0.8348,
      "step": 190620
    },
    {
      "epoch": 3.346793307466774,
      "grad_norm": 0.04572822425898899,
      "learning_rate": 5.259741658366771e-05,
      "loss": 0.8422,
      "step": 190630
    },
    {
      "epoch": 3.3469688723467756,
      "grad_norm": 0.05122997139251615,
      "learning_rate": 5.258769594923071e-05,
      "loss": 0.8448,
      "step": 190640
    },
    {
      "epoch": 3.347144437226777,
      "grad_norm": 0.04974123550129526,
      "learning_rate": 5.257797592818692e-05,
      "loss": 0.8461,
      "step": 190650
    },
    {
      "epoch": 3.3473200021067786,
      "grad_norm": 0.07171226343235383,
      "learning_rate": 5.256825652065941e-05,
      "loss": 0.8388,
      "step": 190660
    },
    {
      "epoch": 3.34749556698678,
      "grad_norm": 0.04747523197594043,
      "learning_rate": 5.2558537726771375e-05,
      "loss": 0.8437,
      "step": 190670
    },
    {
      "epoch": 3.3476711318667816,
      "grad_norm": 0.062398389308279845,
      "learning_rate": 5.2548819546645987e-05,
      "loss": 0.8468,
      "step": 190680
    },
    {
      "epoch": 3.3478466967467826,
      "grad_norm": 0.0494899126738849,
      "learning_rate": 5.2539101980406304e-05,
      "loss": 0.8414,
      "step": 190690
    },
    {
      "epoch": 3.348022261626784,
      "grad_norm": 0.04602008765612902,
      "learning_rate": 5.2529385028175485e-05,
      "loss": 0.8503,
      "step": 190700
    },
    {
      "epoch": 3.3481978265067855,
      "grad_norm": 0.05165299434491655,
      "learning_rate": 5.2519668690076696e-05,
      "loss": 0.8362,
      "step": 190710
    },
    {
      "epoch": 3.348373391386787,
      "grad_norm": 0.06654753053359305,
      "learning_rate": 5.2509952966232945e-05,
      "loss": 0.8399,
      "step": 190720
    },
    {
      "epoch": 3.3485489562667885,
      "grad_norm": 0.07499227863837812,
      "learning_rate": 5.2500237856767386e-05,
      "loss": 0.8493,
      "step": 190730
    },
    {
      "epoch": 3.34872452114679,
      "grad_norm": 0.06075787780505727,
      "learning_rate": 5.2490523361803115e-05,
      "loss": 0.8413,
      "step": 190740
    },
    {
      "epoch": 3.348900086026791,
      "grad_norm": 0.05046717876523233,
      "learning_rate": 5.2480809481463236e-05,
      "loss": 0.8397,
      "step": 190750
    },
    {
      "epoch": 3.3490756509067925,
      "grad_norm": 0.04757993557531884,
      "learning_rate": 5.2471096215870774e-05,
      "loss": 0.8454,
      "step": 190760
    },
    {
      "epoch": 3.349251215786794,
      "grad_norm": 0.05237202352968707,
      "learning_rate": 5.246138356514881e-05,
      "loss": 0.8426,
      "step": 190770
    },
    {
      "epoch": 3.3494267806667954,
      "grad_norm": 0.04914089628240936,
      "learning_rate": 5.245167152942047e-05,
      "loss": 0.8418,
      "step": 190780
    },
    {
      "epoch": 3.349602345546797,
      "grad_norm": 0.04867320329679744,
      "learning_rate": 5.244196010880872e-05,
      "loss": 0.8428,
      "step": 190790
    },
    {
      "epoch": 3.3497779104267984,
      "grad_norm": 0.04848752585469435,
      "learning_rate": 5.2432249303436634e-05,
      "loss": 0.8353,
      "step": 190800
    },
    {
      "epoch": 3.3499534753067994,
      "grad_norm": 0.053336345822012264,
      "learning_rate": 5.242253911342725e-05,
      "loss": 0.8352,
      "step": 190810
    },
    {
      "epoch": 3.350129040186801,
      "grad_norm": 0.04530332534021975,
      "learning_rate": 5.241282953890365e-05,
      "loss": 0.8362,
      "step": 190820
    },
    {
      "epoch": 3.3503046050668024,
      "grad_norm": 0.0446044989266345,
      "learning_rate": 5.2403120579988776e-05,
      "loss": 0.8478,
      "step": 190830
    },
    {
      "epoch": 3.350480169946804,
      "grad_norm": 0.05787222833544451,
      "learning_rate": 5.239341223680567e-05,
      "loss": 0.8396,
      "step": 190840
    },
    {
      "epoch": 3.3506557348268053,
      "grad_norm": 0.05203573819910843,
      "learning_rate": 5.238370450947739e-05,
      "loss": 0.8402,
      "step": 190850
    },
    {
      "epoch": 3.350831299706807,
      "grad_norm": 0.055035385183026114,
      "learning_rate": 5.237399739812686e-05,
      "loss": 0.8443,
      "step": 190860
    },
    {
      "epoch": 3.3510068645868083,
      "grad_norm": 0.046579249913731156,
      "learning_rate": 5.2364290902877086e-05,
      "loss": 0.8354,
      "step": 190870
    },
    {
      "epoch": 3.3511824294668093,
      "grad_norm": 0.055980569446447845,
      "learning_rate": 5.235458502385112e-05,
      "loss": 0.8326,
      "step": 190880
    },
    {
      "epoch": 3.351357994346811,
      "grad_norm": 0.05612311784587836,
      "learning_rate": 5.2344879761171837e-05,
      "loss": 0.8435,
      "step": 190890
    },
    {
      "epoch": 3.3515335592268123,
      "grad_norm": 0.04390777018766173,
      "learning_rate": 5.233517511496225e-05,
      "loss": 0.8413,
      "step": 190900
    },
    {
      "epoch": 3.3517091241068138,
      "grad_norm": 0.049651480332631066,
      "learning_rate": 5.232547108534533e-05,
      "loss": 0.836,
      "step": 190910
    },
    {
      "epoch": 3.3518846889868152,
      "grad_norm": 0.0526098137351421,
      "learning_rate": 5.231576767244406e-05,
      "loss": 0.8431,
      "step": 190920
    },
    {
      "epoch": 3.3520602538668163,
      "grad_norm": 0.0645360501731601,
      "learning_rate": 5.23060648763813e-05,
      "loss": 0.8358,
      "step": 190930
    },
    {
      "epoch": 3.3522358187468178,
      "grad_norm": 0.050731070391274176,
      "learning_rate": 5.2296362697280035e-05,
      "loss": 0.8432,
      "step": 190940
    },
    {
      "epoch": 3.3524113836268192,
      "grad_norm": 0.04626447620341748,
      "learning_rate": 5.2286661135263215e-05,
      "loss": 0.8408,
      "step": 190950
    },
    {
      "epoch": 3.3525869485068207,
      "grad_norm": 0.058071643089364744,
      "learning_rate": 5.227696019045372e-05,
      "loss": 0.8398,
      "step": 190960
    },
    {
      "epoch": 3.352762513386822,
      "grad_norm": 0.045490869173398035,
      "learning_rate": 5.226725986297445e-05,
      "loss": 0.8476,
      "step": 190970
    },
    {
      "epoch": 3.3529380782668237,
      "grad_norm": 0.05018176341260143,
      "learning_rate": 5.225756015294835e-05,
      "loss": 0.8457,
      "step": 190980
    },
    {
      "epoch": 3.353113643146825,
      "grad_norm": 0.04706853521010304,
      "learning_rate": 5.224786106049834e-05,
      "loss": 0.8429,
      "step": 190990
    },
    {
      "epoch": 3.353289208026826,
      "grad_norm": 0.05073340975904758,
      "learning_rate": 5.223816258574723e-05,
      "loss": 0.8392,
      "step": 191000
    },
    {
      "epoch": 3.3534647729068277,
      "grad_norm": 0.05258735933821124,
      "learning_rate": 5.222846472881795e-05,
      "loss": 0.8495,
      "step": 191010
    },
    {
      "epoch": 3.353640337786829,
      "grad_norm": 0.07275318477813661,
      "learning_rate": 5.221876748983342e-05,
      "loss": 0.8394,
      "step": 191020
    },
    {
      "epoch": 3.3538159026668306,
      "grad_norm": 0.051964712825330564,
      "learning_rate": 5.220907086891639e-05,
      "loss": 0.8348,
      "step": 191030
    },
    {
      "epoch": 3.353991467546832,
      "grad_norm": 0.05133741667933899,
      "learning_rate": 5.219937486618981e-05,
      "loss": 0.8374,
      "step": 191040
    },
    {
      "epoch": 3.354167032426833,
      "grad_norm": 0.07298847078727372,
      "learning_rate": 5.218967948177652e-05,
      "loss": 0.84,
      "step": 191050
    },
    {
      "epoch": 3.3543425973068346,
      "grad_norm": 0.05458210620257251,
      "learning_rate": 5.2179984715799315e-05,
      "loss": 0.8427,
      "step": 191060
    },
    {
      "epoch": 3.354518162186836,
      "grad_norm": 0.0460278104520748,
      "learning_rate": 5.2170290568381066e-05,
      "loss": 0.8364,
      "step": 191070
    },
    {
      "epoch": 3.3546937270668375,
      "grad_norm": 0.048206780546615695,
      "learning_rate": 5.2160597039644586e-05,
      "loss": 0.8435,
      "step": 191080
    },
    {
      "epoch": 3.354869291946839,
      "grad_norm": 0.04840532971043061,
      "learning_rate": 5.215090412971274e-05,
      "loss": 0.8426,
      "step": 191090
    },
    {
      "epoch": 3.3550448568268405,
      "grad_norm": 0.05165681815567989,
      "learning_rate": 5.214121183870827e-05,
      "loss": 0.8401,
      "step": 191100
    },
    {
      "epoch": 3.355220421706842,
      "grad_norm": 0.07358868630576501,
      "learning_rate": 5.213152016675399e-05,
      "loss": 0.8455,
      "step": 191110
    },
    {
      "epoch": 3.355395986586843,
      "grad_norm": 0.048813698325806044,
      "learning_rate": 5.2121829113972775e-05,
      "loss": 0.8355,
      "step": 191120
    },
    {
      "epoch": 3.3555715514668445,
      "grad_norm": 0.05382755993771167,
      "learning_rate": 5.21121386804873e-05,
      "loss": 0.8495,
      "step": 191130
    },
    {
      "epoch": 3.355747116346846,
      "grad_norm": 0.05357854723340084,
      "learning_rate": 5.2102448866420416e-05,
      "loss": 0.843,
      "step": 191140
    },
    {
      "epoch": 3.3559226812268474,
      "grad_norm": 0.05592315065438146,
      "learning_rate": 5.2092759671894864e-05,
      "loss": 0.8421,
      "step": 191150
    },
    {
      "epoch": 3.356098246106849,
      "grad_norm": 0.06253019397358424,
      "learning_rate": 5.2083071097033455e-05,
      "loss": 0.8459,
      "step": 191160
    },
    {
      "epoch": 3.35627381098685,
      "grad_norm": 0.05696978542630176,
      "learning_rate": 5.207338314195891e-05,
      "loss": 0.849,
      "step": 191170
    },
    {
      "epoch": 3.3564493758668514,
      "grad_norm": 0.04718827988243918,
      "learning_rate": 5.206369580679391e-05,
      "loss": 0.8384,
      "step": 191180
    },
    {
      "epoch": 3.356624940746853,
      "grad_norm": 0.04972980005313652,
      "learning_rate": 5.2054009091661336e-05,
      "loss": 0.8408,
      "step": 191190
    },
    {
      "epoch": 3.3568005056268544,
      "grad_norm": 0.05579528610564679,
      "learning_rate": 5.20443229966838e-05,
      "loss": 0.8407,
      "step": 191200
    },
    {
      "epoch": 3.356976070506856,
      "grad_norm": 0.04945570656947227,
      "learning_rate": 5.2034637521984073e-05,
      "loss": 0.8406,
      "step": 191210
    },
    {
      "epoch": 3.3571516353868573,
      "grad_norm": 0.05295445827201562,
      "learning_rate": 5.2024952667684935e-05,
      "loss": 0.8349,
      "step": 191220
    },
    {
      "epoch": 3.357327200266859,
      "grad_norm": 0.06214827537156114,
      "learning_rate": 5.201526843390897e-05,
      "loss": 0.8446,
      "step": 191230
    },
    {
      "epoch": 3.35750276514686,
      "grad_norm": 0.05526814246049467,
      "learning_rate": 5.200558482077896e-05,
      "loss": 0.8435,
      "step": 191240
    },
    {
      "epoch": 3.3576783300268613,
      "grad_norm": 0.04435600299692391,
      "learning_rate": 5.199590182841756e-05,
      "loss": 0.8465,
      "step": 191250
    },
    {
      "epoch": 3.357853894906863,
      "grad_norm": 0.04802614076231742,
      "learning_rate": 5.198621945694753e-05,
      "loss": 0.8384,
      "step": 191260
    },
    {
      "epoch": 3.3580294597868643,
      "grad_norm": 0.05232071475621778,
      "learning_rate": 5.1976537706491446e-05,
      "loss": 0.8427,
      "step": 191270
    },
    {
      "epoch": 3.3582050246668658,
      "grad_norm": 0.055613264852500184,
      "learning_rate": 5.196685657717203e-05,
      "loss": 0.8391,
      "step": 191280
    },
    {
      "epoch": 3.358380589546867,
      "grad_norm": 0.0595309444155183,
      "learning_rate": 5.1957176069111965e-05,
      "loss": 0.8337,
      "step": 191290
    },
    {
      "epoch": 3.3585561544268683,
      "grad_norm": 0.04489667331094394,
      "learning_rate": 5.1947496182433854e-05,
      "loss": 0.8414,
      "step": 191300
    },
    {
      "epoch": 3.3587317193068698,
      "grad_norm": 0.05898817083571271,
      "learning_rate": 5.193781691726036e-05,
      "loss": 0.8385,
      "step": 191310
    },
    {
      "epoch": 3.3589072841868712,
      "grad_norm": 0.06836311669657767,
      "learning_rate": 5.1928138273714135e-05,
      "loss": 0.8366,
      "step": 191320
    },
    {
      "epoch": 3.3590828490668727,
      "grad_norm": 0.06762238956897854,
      "learning_rate": 5.1918460251917836e-05,
      "loss": 0.8335,
      "step": 191330
    },
    {
      "epoch": 3.359258413946874,
      "grad_norm": 0.06578271466290951,
      "learning_rate": 5.190878285199405e-05,
      "loss": 0.8428,
      "step": 191340
    },
    {
      "epoch": 3.3594339788268757,
      "grad_norm": 0.05593395034251765,
      "learning_rate": 5.1899106074065315e-05,
      "loss": 0.8352,
      "step": 191350
    },
    {
      "epoch": 3.3596095437068767,
      "grad_norm": 0.0443208222228264,
      "learning_rate": 5.1889429918254396e-05,
      "loss": 0.8433,
      "step": 191360
    },
    {
      "epoch": 3.359785108586878,
      "grad_norm": 0.05313140971938416,
      "learning_rate": 5.187975438468378e-05,
      "loss": 0.8357,
      "step": 191370
    },
    {
      "epoch": 3.3599606734668797,
      "grad_norm": 0.05235437181921441,
      "learning_rate": 5.187007947347608e-05,
      "loss": 0.8449,
      "step": 191380
    },
    {
      "epoch": 3.360136238346881,
      "grad_norm": 0.09068872084150574,
      "learning_rate": 5.1860405184753935e-05,
      "loss": 0.8336,
      "step": 191390
    },
    {
      "epoch": 3.3603118032268826,
      "grad_norm": 0.05573652387036431,
      "learning_rate": 5.1850731518639825e-05,
      "loss": 0.8368,
      "step": 191400
    },
    {
      "epoch": 3.360487368106884,
      "grad_norm": 0.07041833734607848,
      "learning_rate": 5.1841058475256376e-05,
      "loss": 0.8472,
      "step": 191410
    },
    {
      "epoch": 3.360662932986885,
      "grad_norm": 0.06945677529061446,
      "learning_rate": 5.183138605472613e-05,
      "loss": 0.8403,
      "step": 191420
    },
    {
      "epoch": 3.3608384978668866,
      "grad_norm": 0.05663234880852349,
      "learning_rate": 5.1821714257171695e-05,
      "loss": 0.8367,
      "step": 191430
    },
    {
      "epoch": 3.361014062746888,
      "grad_norm": 0.04279341620840209,
      "learning_rate": 5.181204308271552e-05,
      "loss": 0.847,
      "step": 191440
    },
    {
      "epoch": 3.3611896276268896,
      "grad_norm": 0.06071023712749484,
      "learning_rate": 5.1802372531480195e-05,
      "loss": 0.8351,
      "step": 191450
    },
    {
      "epoch": 3.361365192506891,
      "grad_norm": 0.05923957379804324,
      "learning_rate": 5.179270260358826e-05,
      "loss": 0.8358,
      "step": 191460
    },
    {
      "epoch": 3.3615407573868925,
      "grad_norm": 0.06364679682234625,
      "learning_rate": 5.1783033299162176e-05,
      "loss": 0.842,
      "step": 191470
    },
    {
      "epoch": 3.3617163222668935,
      "grad_norm": 0.05372783999142378,
      "learning_rate": 5.1773364618324513e-05,
      "loss": 0.8417,
      "step": 191480
    },
    {
      "epoch": 3.361891887146895,
      "grad_norm": 0.062211705138542774,
      "learning_rate": 5.176369656119774e-05,
      "loss": 0.8438,
      "step": 191490
    },
    {
      "epoch": 3.3620674520268965,
      "grad_norm": 0.045073437715726014,
      "learning_rate": 5.17540291279044e-05,
      "loss": 0.8412,
      "step": 191500
    },
    {
      "epoch": 3.362243016906898,
      "grad_norm": 0.0507509748817467,
      "learning_rate": 5.1744362318566965e-05,
      "loss": 0.8428,
      "step": 191510
    },
    {
      "epoch": 3.3624185817868995,
      "grad_norm": 0.06056414777656706,
      "learning_rate": 5.17346961333078e-05,
      "loss": 0.8403,
      "step": 191520
    },
    {
      "epoch": 3.362594146666901,
      "grad_norm": 0.06155625851065957,
      "learning_rate": 5.1725030572249576e-05,
      "loss": 0.8451,
      "step": 191530
    },
    {
      "epoch": 3.3627697115469024,
      "grad_norm": 0.07592780786338804,
      "learning_rate": 5.171536563551461e-05,
      "loss": 0.842,
      "step": 191540
    },
    {
      "epoch": 3.3629452764269034,
      "grad_norm": 0.06592555350249535,
      "learning_rate": 5.1705701323225406e-05,
      "loss": 0.8456,
      "step": 191550
    },
    {
      "epoch": 3.363120841306905,
      "grad_norm": 0.04346658818459692,
      "learning_rate": 5.1696037635504455e-05,
      "loss": 0.8405,
      "step": 191560
    },
    {
      "epoch": 3.3632964061869064,
      "grad_norm": 0.04999902101291993,
      "learning_rate": 5.168637457247412e-05,
      "loss": 0.8529,
      "step": 191570
    },
    {
      "epoch": 3.363471971066908,
      "grad_norm": 0.06162347658163204,
      "learning_rate": 5.1676712134256876e-05,
      "loss": 0.847,
      "step": 191580
    },
    {
      "epoch": 3.3636475359469094,
      "grad_norm": 0.058590562767727304,
      "learning_rate": 5.166705032097512e-05,
      "loss": 0.8365,
      "step": 191590
    },
    {
      "epoch": 3.3638231008269104,
      "grad_norm": 0.051831450495312376,
      "learning_rate": 5.165738913275134e-05,
      "loss": 0.8476,
      "step": 191600
    },
    {
      "epoch": 3.363998665706912,
      "grad_norm": 0.051505158997625373,
      "learning_rate": 5.164772856970786e-05,
      "loss": 0.8376,
      "step": 191610
    },
    {
      "epoch": 3.3641742305869133,
      "grad_norm": 0.05880654942371588,
      "learning_rate": 5.1638068631967117e-05,
      "loss": 0.8402,
      "step": 191620
    },
    {
      "epoch": 3.364349795466915,
      "grad_norm": 0.04860386560160922,
      "learning_rate": 5.162840931965155e-05,
      "loss": 0.8388,
      "step": 191630
    },
    {
      "epoch": 3.3645253603469163,
      "grad_norm": 0.056728674681503194,
      "learning_rate": 5.161875063288345e-05,
      "loss": 0.8383,
      "step": 191640
    },
    {
      "epoch": 3.3647009252269178,
      "grad_norm": 0.042502144034061336,
      "learning_rate": 5.160909257178525e-05,
      "loss": 0.8366,
      "step": 191650
    },
    {
      "epoch": 3.3648764901069192,
      "grad_norm": 0.06317656185424349,
      "learning_rate": 5.159943513647931e-05,
      "loss": 0.8432,
      "step": 191660
    },
    {
      "epoch": 3.3650520549869203,
      "grad_norm": 0.04851703091442338,
      "learning_rate": 5.158977832708803e-05,
      "loss": 0.8404,
      "step": 191670
    },
    {
      "epoch": 3.3652276198669218,
      "grad_norm": 0.04813583359949942,
      "learning_rate": 5.158012214373373e-05,
      "loss": 0.8453,
      "step": 191680
    },
    {
      "epoch": 3.3654031847469232,
      "grad_norm": 0.055897445659908375,
      "learning_rate": 5.157046658653868e-05,
      "loss": 0.8378,
      "step": 191690
    },
    {
      "epoch": 3.3655787496269247,
      "grad_norm": 0.05219186582634968,
      "learning_rate": 5.156081165562536e-05,
      "loss": 0.8336,
      "step": 191700
    },
    {
      "epoch": 3.365754314506926,
      "grad_norm": 0.0401720002365314,
      "learning_rate": 5.155115735111601e-05,
      "loss": 0.8386,
      "step": 191710
    },
    {
      "epoch": 3.3659298793869272,
      "grad_norm": 0.05946010610852469,
      "learning_rate": 5.154150367313295e-05,
      "loss": 0.8377,
      "step": 191720
    },
    {
      "epoch": 3.3661054442669287,
      "grad_norm": 0.053884823649920574,
      "learning_rate": 5.153185062179857e-05,
      "loss": 0.8445,
      "step": 191730
    },
    {
      "epoch": 3.36628100914693,
      "grad_norm": 0.08824128196160515,
      "learning_rate": 5.152219819723509e-05,
      "loss": 0.8432,
      "step": 191740
    },
    {
      "epoch": 3.3664565740269317,
      "grad_norm": 0.0495568756443934,
      "learning_rate": 5.151254639956483e-05,
      "loss": 0.8436,
      "step": 191750
    },
    {
      "epoch": 3.366632138906933,
      "grad_norm": 0.046577802705277005,
      "learning_rate": 5.1502895228910096e-05,
      "loss": 0.8458,
      "step": 191760
    },
    {
      "epoch": 3.3668077037869346,
      "grad_norm": 0.05326935994067747,
      "learning_rate": 5.149324468539321e-05,
      "loss": 0.8432,
      "step": 191770
    },
    {
      "epoch": 3.366983268666936,
      "grad_norm": 0.05916856711305766,
      "learning_rate": 5.148359476913635e-05,
      "loss": 0.8472,
      "step": 191780
    },
    {
      "epoch": 3.367158833546937,
      "grad_norm": 0.06543873456884586,
      "learning_rate": 5.1473945480261845e-05,
      "loss": 0.8487,
      "step": 191790
    },
    {
      "epoch": 3.3673343984269386,
      "grad_norm": 0.06275363568312523,
      "learning_rate": 5.1464296818891975e-05,
      "loss": 0.8413,
      "step": 191800
    },
    {
      "epoch": 3.36750996330694,
      "grad_norm": 0.04622519168504258,
      "learning_rate": 5.1454648785148915e-05,
      "loss": 0.8438,
      "step": 191810
    },
    {
      "epoch": 3.3676855281869416,
      "grad_norm": 0.05207079566811095,
      "learning_rate": 5.1445001379154955e-05,
      "loss": 0.8344,
      "step": 191820
    },
    {
      "epoch": 3.367861093066943,
      "grad_norm": 0.05538581591291978,
      "learning_rate": 5.1435354601032326e-05,
      "loss": 0.8417,
      "step": 191830
    },
    {
      "epoch": 3.368036657946944,
      "grad_norm": 0.055247490238291216,
      "learning_rate": 5.14257084509033e-05,
      "loss": 0.8325,
      "step": 191840
    },
    {
      "epoch": 3.3682122228269455,
      "grad_norm": 0.05845034804655482,
      "learning_rate": 5.141606292889003e-05,
      "loss": 0.8451,
      "step": 191850
    },
    {
      "epoch": 3.368387787706947,
      "grad_norm": 0.041857142799372964,
      "learning_rate": 5.140641803511469e-05,
      "loss": 0.8329,
      "step": 191860
    },
    {
      "epoch": 3.3685633525869485,
      "grad_norm": 0.047457777459788145,
      "learning_rate": 5.13967737696996e-05,
      "loss": 0.8379,
      "step": 191870
    },
    {
      "epoch": 3.36873891746695,
      "grad_norm": 0.05794785666784063,
      "learning_rate": 5.138713013276686e-05,
      "loss": 0.8376,
      "step": 191880
    },
    {
      "epoch": 3.3689144823469515,
      "grad_norm": 0.054210441567374916,
      "learning_rate": 5.1377487124438686e-05,
      "loss": 0.8387,
      "step": 191890
    },
    {
      "epoch": 3.369090047226953,
      "grad_norm": 0.04928852933486844,
      "learning_rate": 5.136784474483729e-05,
      "loss": 0.8437,
      "step": 191900
    },
    {
      "epoch": 3.369265612106954,
      "grad_norm": 0.05175914735161534,
      "learning_rate": 5.135820299408479e-05,
      "loss": 0.8322,
      "step": 191910
    },
    {
      "epoch": 3.3694411769869554,
      "grad_norm": 0.06813651090165655,
      "learning_rate": 5.134856187230337e-05,
      "loss": 0.8435,
      "step": 191920
    },
    {
      "epoch": 3.369616741866957,
      "grad_norm": 0.05417962259801,
      "learning_rate": 5.133892137961518e-05,
      "loss": 0.837,
      "step": 191930
    },
    {
      "epoch": 3.3697923067469584,
      "grad_norm": 0.0668328623175882,
      "learning_rate": 5.132928151614241e-05,
      "loss": 0.8435,
      "step": 191940
    },
    {
      "epoch": 3.36996787162696,
      "grad_norm": 0.06191291765279457,
      "learning_rate": 5.131964228200714e-05,
      "loss": 0.8379,
      "step": 191950
    },
    {
      "epoch": 3.370143436506961,
      "grad_norm": 0.05582153942910568,
      "learning_rate": 5.13100036773315e-05,
      "loss": 0.84,
      "step": 191960
    },
    {
      "epoch": 3.3703190013869624,
      "grad_norm": 0.046320758290574304,
      "learning_rate": 5.13003657022377e-05,
      "loss": 0.8426,
      "step": 191970
    },
    {
      "epoch": 3.370494566266964,
      "grad_norm": 0.051675984679389726,
      "learning_rate": 5.129072835684774e-05,
      "loss": 0.8338,
      "step": 191980
    },
    {
      "epoch": 3.3706701311469653,
      "grad_norm": 0.047745619098106894,
      "learning_rate": 5.1281091641283774e-05,
      "loss": 0.8461,
      "step": 191990
    },
    {
      "epoch": 3.370845696026967,
      "grad_norm": 0.05766773648025234,
      "learning_rate": 5.1271455555667914e-05,
      "loss": 0.8349,
      "step": 192000
    },
    {
      "epoch": 3.3710212609069683,
      "grad_norm": 0.048402917377727464,
      "learning_rate": 5.126182010012227e-05,
      "loss": 0.8424,
      "step": 192010
    },
    {
      "epoch": 3.3711968257869698,
      "grad_norm": 0.058862230635301466,
      "learning_rate": 5.1252185274768905e-05,
      "loss": 0.8381,
      "step": 192020
    },
    {
      "epoch": 3.371372390666971,
      "grad_norm": 0.04915711962912607,
      "learning_rate": 5.124255107972982e-05,
      "loss": 0.8442,
      "step": 192030
    },
    {
      "epoch": 3.3715479555469723,
      "grad_norm": 0.04946065812090965,
      "learning_rate": 5.123291751512721e-05,
      "loss": 0.8399,
      "step": 192040
    },
    {
      "epoch": 3.3717235204269738,
      "grad_norm": 0.05557752797608386,
      "learning_rate": 5.1223284581083044e-05,
      "loss": 0.8391,
      "step": 192050
    },
    {
      "epoch": 3.3718990853069752,
      "grad_norm": 0.05654937079336426,
      "learning_rate": 5.121365227771939e-05,
      "loss": 0.8381,
      "step": 192060
    },
    {
      "epoch": 3.3720746501869767,
      "grad_norm": 0.06430008754117306,
      "learning_rate": 5.1204020605158353e-05,
      "loss": 0.8454,
      "step": 192070
    },
    {
      "epoch": 3.3722502150669778,
      "grad_norm": 0.06481225082879648,
      "learning_rate": 5.119438956352188e-05,
      "loss": 0.8439,
      "step": 192080
    },
    {
      "epoch": 3.3724257799469792,
      "grad_norm": 0.05257766167308451,
      "learning_rate": 5.118475915293203e-05,
      "loss": 0.8365,
      "step": 192090
    },
    {
      "epoch": 3.3726013448269807,
      "grad_norm": 0.05207966497299352,
      "learning_rate": 5.117512937351083e-05,
      "loss": 0.8369,
      "step": 192100
    },
    {
      "epoch": 3.372776909706982,
      "grad_norm": 0.08291615981294996,
      "learning_rate": 5.116550022538034e-05,
      "loss": 0.8401,
      "step": 192110
    },
    {
      "epoch": 3.3729524745869837,
      "grad_norm": 0.04743404765986916,
      "learning_rate": 5.115587170866245e-05,
      "loss": 0.8439,
      "step": 192120
    },
    {
      "epoch": 3.373128039466985,
      "grad_norm": 0.051144077270566725,
      "learning_rate": 5.1146243823479244e-05,
      "loss": 0.8403,
      "step": 192130
    },
    {
      "epoch": 3.3733036043469866,
      "grad_norm": 0.0481452031033534,
      "learning_rate": 5.11366165699527e-05,
      "loss": 0.8433,
      "step": 192140
    },
    {
      "epoch": 3.3734791692269877,
      "grad_norm": 0.05962443595907689,
      "learning_rate": 5.112698994820475e-05,
      "loss": 0.8406,
      "step": 192150
    },
    {
      "epoch": 3.373654734106989,
      "grad_norm": 0.04960791706636908,
      "learning_rate": 5.111736395835741e-05,
      "loss": 0.8408,
      "step": 192160
    },
    {
      "epoch": 3.3738302989869906,
      "grad_norm": 0.049815556073349296,
      "learning_rate": 5.110773860053262e-05,
      "loss": 0.8443,
      "step": 192170
    },
    {
      "epoch": 3.374005863866992,
      "grad_norm": 0.053565484137403634,
      "learning_rate": 5.109811387485237e-05,
      "loss": 0.8393,
      "step": 192180
    },
    {
      "epoch": 3.3741814287469936,
      "grad_norm": 0.09300784162746954,
      "learning_rate": 5.108848978143859e-05,
      "loss": 0.8419,
      "step": 192190
    },
    {
      "epoch": 3.374356993626995,
      "grad_norm": 0.05335327469545468,
      "learning_rate": 5.107886632041312e-05,
      "loss": 0.8459,
      "step": 192200
    },
    {
      "epoch": 3.374532558506996,
      "grad_norm": 0.06681633703199054,
      "learning_rate": 5.106924349189808e-05,
      "loss": 0.8324,
      "step": 192210
    },
    {
      "epoch": 3.3747081233869975,
      "grad_norm": 0.0511839618763199,
      "learning_rate": 5.105962129601524e-05,
      "loss": 0.8506,
      "step": 192220
    },
    {
      "epoch": 3.374883688266999,
      "grad_norm": 0.05303850861085858,
      "learning_rate": 5.1049999732886575e-05,
      "loss": 0.8438,
      "step": 192230
    },
    {
      "epoch": 3.3750592531470005,
      "grad_norm": 0.05021874125237089,
      "learning_rate": 5.1040378802634035e-05,
      "loss": 0.8499,
      "step": 192240
    },
    {
      "epoch": 3.375234818027002,
      "grad_norm": 0.053724173014155135,
      "learning_rate": 5.1030758505379414e-05,
      "loss": 0.8432,
      "step": 192250
    },
    {
      "epoch": 3.3754103829070035,
      "grad_norm": 0.06012699836340906,
      "learning_rate": 5.1021138841244665e-05,
      "loss": 0.8449,
      "step": 192260
    },
    {
      "epoch": 3.3755859477870045,
      "grad_norm": 0.053053366828231294,
      "learning_rate": 5.1011519810351674e-05,
      "loss": 0.8444,
      "step": 192270
    },
    {
      "epoch": 3.375761512667006,
      "grad_norm": 0.052185705370771295,
      "learning_rate": 5.100190141282234e-05,
      "loss": 0.8462,
      "step": 192280
    },
    {
      "epoch": 3.3759370775470074,
      "grad_norm": 0.04927080323607024,
      "learning_rate": 5.099228364877845e-05,
      "loss": 0.8462,
      "step": 192290
    },
    {
      "epoch": 3.376112642427009,
      "grad_norm": 0.0429754736279009,
      "learning_rate": 5.0982666518341916e-05,
      "loss": 0.8414,
      "step": 192300
    },
    {
      "epoch": 3.3762882073070104,
      "grad_norm": 0.052197676697870565,
      "learning_rate": 5.097305002163463e-05,
      "loss": 0.8389,
      "step": 192310
    },
    {
      "epoch": 3.376463772187012,
      "grad_norm": 0.060674080656408606,
      "learning_rate": 5.0963434158778346e-05,
      "loss": 0.8355,
      "step": 192320
    },
    {
      "epoch": 3.3766393370670134,
      "grad_norm": 0.05302878871277978,
      "learning_rate": 5.0953818929894924e-05,
      "loss": 0.8336,
      "step": 192330
    },
    {
      "epoch": 3.3768149019470144,
      "grad_norm": 0.046573320090396664,
      "learning_rate": 5.094420433510623e-05,
      "loss": 0.8315,
      "step": 192340
    },
    {
      "epoch": 3.376990466827016,
      "grad_norm": 0.055666950866993935,
      "learning_rate": 5.093459037453409e-05,
      "loss": 0.8437,
      "step": 192350
    },
    {
      "epoch": 3.3771660317070173,
      "grad_norm": 0.05133766462278727,
      "learning_rate": 5.092497704830028e-05,
      "loss": 0.847,
      "step": 192360
    },
    {
      "epoch": 3.377341596587019,
      "grad_norm": 0.05298455408529404,
      "learning_rate": 5.091536435652653e-05,
      "loss": 0.8424,
      "step": 192370
    },
    {
      "epoch": 3.3775171614670203,
      "grad_norm": 0.05438454959913241,
      "learning_rate": 5.0905752299334794e-05,
      "loss": 0.8339,
      "step": 192380
    },
    {
      "epoch": 3.3776927263470213,
      "grad_norm": 0.06975310770957757,
      "learning_rate": 5.089614087684674e-05,
      "loss": 0.8373,
      "step": 192390
    },
    {
      "epoch": 3.377868291227023,
      "grad_norm": 0.06184992354326768,
      "learning_rate": 5.0886530089184184e-05,
      "loss": 0.835,
      "step": 192400
    },
    {
      "epoch": 3.3780438561070243,
      "grad_norm": 0.05906375575376082,
      "learning_rate": 5.0876919936468926e-05,
      "loss": 0.8409,
      "step": 192410
    },
    {
      "epoch": 3.3782194209870258,
      "grad_norm": 0.050418174274934675,
      "learning_rate": 5.086731041882268e-05,
      "loss": 0.8488,
      "step": 192420
    },
    {
      "epoch": 3.3783949858670272,
      "grad_norm": 0.04720900293329408,
      "learning_rate": 5.08577015363672e-05,
      "loss": 0.8409,
      "step": 192430
    },
    {
      "epoch": 3.3785705507470287,
      "grad_norm": 0.052797997276673496,
      "learning_rate": 5.084809328922425e-05,
      "loss": 0.8416,
      "step": 192440
    },
    {
      "epoch": 3.37874611562703,
      "grad_norm": 0.061037656332097984,
      "learning_rate": 5.083848567751561e-05,
      "loss": 0.8304,
      "step": 192450
    },
    {
      "epoch": 3.3789216805070312,
      "grad_norm": 0.05542562532067787,
      "learning_rate": 5.082887870136295e-05,
      "loss": 0.8261,
      "step": 192460
    },
    {
      "epoch": 3.3790972453870327,
      "grad_norm": 0.04199716630062572,
      "learning_rate": 5.0819272360887996e-05,
      "loss": 0.8312,
      "step": 192470
    },
    {
      "epoch": 3.379272810267034,
      "grad_norm": 0.04894132434262027,
      "learning_rate": 5.080966665621252e-05,
      "loss": 0.8483,
      "step": 192480
    },
    {
      "epoch": 3.3794483751470357,
      "grad_norm": 0.051236824239141,
      "learning_rate": 5.080006158745815e-05,
      "loss": 0.839,
      "step": 192490
    },
    {
      "epoch": 3.379623940027037,
      "grad_norm": 0.051353890651156524,
      "learning_rate": 5.079045715474661e-05,
      "loss": 0.841,
      "step": 192500
    },
    {
      "epoch": 3.379799504907038,
      "grad_norm": 0.056389490885585106,
      "learning_rate": 5.0780853358199607e-05,
      "loss": 0.8424,
      "step": 192510
    },
    {
      "epoch": 3.3799750697870397,
      "grad_norm": 0.04414363007163887,
      "learning_rate": 5.0771250197938856e-05,
      "loss": 0.8419,
      "step": 192520
    },
    {
      "epoch": 3.380150634667041,
      "grad_norm": 0.05587936206015871,
      "learning_rate": 5.076164767408599e-05,
      "loss": 0.8433,
      "step": 192530
    },
    {
      "epoch": 3.3803261995470426,
      "grad_norm": 0.052601557574251244,
      "learning_rate": 5.0752045786762604e-05,
      "loss": 0.8453,
      "step": 192540
    },
    {
      "epoch": 3.380501764427044,
      "grad_norm": 0.07488684185343178,
      "learning_rate": 5.0742444536090504e-05,
      "loss": 0.8354,
      "step": 192550
    },
    {
      "epoch": 3.3806773293070456,
      "grad_norm": 0.06051354012192152,
      "learning_rate": 5.073284392219121e-05,
      "loss": 0.8435,
      "step": 192560
    },
    {
      "epoch": 3.380852894187047,
      "grad_norm": 0.05172873410150117,
      "learning_rate": 5.072324394518643e-05,
      "loss": 0.8468,
      "step": 192570
    },
    {
      "epoch": 3.381028459067048,
      "grad_norm": 0.07131127116625621,
      "learning_rate": 5.071364460519783e-05,
      "loss": 0.8479,
      "step": 192580
    },
    {
      "epoch": 3.3812040239470496,
      "grad_norm": 0.05628512176959717,
      "learning_rate": 5.070404590234694e-05,
      "loss": 0.8434,
      "step": 192590
    },
    {
      "epoch": 3.381379588827051,
      "grad_norm": 0.060441255522993766,
      "learning_rate": 5.069444783675543e-05,
      "loss": 0.8309,
      "step": 192600
    },
    {
      "epoch": 3.3815551537070525,
      "grad_norm": 0.05193874013227099,
      "learning_rate": 5.068485040854491e-05,
      "loss": 0.8384,
      "step": 192610
    },
    {
      "epoch": 3.381730718587054,
      "grad_norm": 0.09324846530437261,
      "learning_rate": 5.0675253617837017e-05,
      "loss": 0.8388,
      "step": 192620
    },
    {
      "epoch": 3.381906283467055,
      "grad_norm": 0.05803163247360217,
      "learning_rate": 5.066565746475326e-05,
      "loss": 0.8323,
      "step": 192630
    },
    {
      "epoch": 3.3820818483470565,
      "grad_norm": 0.06954442741106749,
      "learning_rate": 5.065606194941527e-05,
      "loss": 0.8367,
      "step": 192640
    },
    {
      "epoch": 3.382257413227058,
      "grad_norm": 0.05062620043936373,
      "learning_rate": 5.064646707194467e-05,
      "loss": 0.8433,
      "step": 192650
    },
    {
      "epoch": 3.3824329781070595,
      "grad_norm": 0.046110053183824655,
      "learning_rate": 5.0636872832462944e-05,
      "loss": 0.8381,
      "step": 192660
    },
    {
      "epoch": 3.382608542987061,
      "grad_norm": 0.09950168211040983,
      "learning_rate": 5.0627279231091694e-05,
      "loss": 0.8463,
      "step": 192670
    },
    {
      "epoch": 3.3827841078670624,
      "grad_norm": 0.07997822009373241,
      "learning_rate": 5.061768626795246e-05,
      "loss": 0.8448,
      "step": 192680
    },
    {
      "epoch": 3.382959672747064,
      "grad_norm": 0.057160974489997476,
      "learning_rate": 5.060809394316683e-05,
      "loss": 0.8354,
      "step": 192690
    },
    {
      "epoch": 3.383135237627065,
      "grad_norm": 0.05517986946028042,
      "learning_rate": 5.0598502256856316e-05,
      "loss": 0.8414,
      "step": 192700
    },
    {
      "epoch": 3.3833108025070664,
      "grad_norm": 0.050684166985397586,
      "learning_rate": 5.0588911209142365e-05,
      "loss": 0.8386,
      "step": 192710
    },
    {
      "epoch": 3.383486367387068,
      "grad_norm": 0.05133565547725073,
      "learning_rate": 5.0579320800146645e-05,
      "loss": 0.841,
      "step": 192720
    },
    {
      "epoch": 3.3836619322670693,
      "grad_norm": 0.0458838601283481,
      "learning_rate": 5.056973102999055e-05,
      "loss": 0.8411,
      "step": 192730
    },
    {
      "epoch": 3.383837497147071,
      "grad_norm": 0.06057068049103058,
      "learning_rate": 5.0560141898795635e-05,
      "loss": 0.8408,
      "step": 192740
    },
    {
      "epoch": 3.384013062027072,
      "grad_norm": 0.04869097357163453,
      "learning_rate": 5.055055340668342e-05,
      "loss": 0.8393,
      "step": 192750
    },
    {
      "epoch": 3.3841886269070733,
      "grad_norm": 0.048984920320958705,
      "learning_rate": 5.0540965553775335e-05,
      "loss": 0.8393,
      "step": 192760
    },
    {
      "epoch": 3.384364191787075,
      "grad_norm": 0.05095472669972273,
      "learning_rate": 5.053137834019288e-05,
      "loss": 0.8395,
      "step": 192770
    },
    {
      "epoch": 3.3845397566670763,
      "grad_norm": 0.05039863603623326,
      "learning_rate": 5.052179176605752e-05,
      "loss": 0.8413,
      "step": 192780
    },
    {
      "epoch": 3.3847153215470778,
      "grad_norm": 0.05481204949869155,
      "learning_rate": 5.051220583149079e-05,
      "loss": 0.8406,
      "step": 192790
    },
    {
      "epoch": 3.3848908864270792,
      "grad_norm": 0.051093534914525264,
      "learning_rate": 5.050262053661404e-05,
      "loss": 0.8475,
      "step": 192800
    },
    {
      "epoch": 3.3850664513070807,
      "grad_norm": 0.050257993975720436,
      "learning_rate": 5.049303588154877e-05,
      "loss": 0.8319,
      "step": 192810
    },
    {
      "epoch": 3.3852420161870818,
      "grad_norm": 0.04887878194982596,
      "learning_rate": 5.0483451866416454e-05,
      "loss": 0.8403,
      "step": 192820
    },
    {
      "epoch": 3.3854175810670832,
      "grad_norm": 0.053492554144958945,
      "learning_rate": 5.047386849133844e-05,
      "loss": 0.8383,
      "step": 192830
    },
    {
      "epoch": 3.3855931459470847,
      "grad_norm": 0.05379453378645171,
      "learning_rate": 5.0464285756436204e-05,
      "loss": 0.8454,
      "step": 192840
    },
    {
      "epoch": 3.385768710827086,
      "grad_norm": 0.057670727773465456,
      "learning_rate": 5.0454703661831136e-05,
      "loss": 0.8457,
      "step": 192850
    },
    {
      "epoch": 3.3859442757070877,
      "grad_norm": 0.04754326516003908,
      "learning_rate": 5.0445122207644716e-05,
      "loss": 0.8379,
      "step": 192860
    },
    {
      "epoch": 3.386119840587089,
      "grad_norm": 0.04865214989489531,
      "learning_rate": 5.043554139399828e-05,
      "loss": 0.8408,
      "step": 192870
    },
    {
      "epoch": 3.38629540546709,
      "grad_norm": 0.06493854903804064,
      "learning_rate": 5.042596122101315e-05,
      "loss": 0.846,
      "step": 192880
    },
    {
      "epoch": 3.3864709703470917,
      "grad_norm": 0.04688253284227578,
      "learning_rate": 5.0416381688810854e-05,
      "loss": 0.8371,
      "step": 192890
    },
    {
      "epoch": 3.386646535227093,
      "grad_norm": 0.04722733144409066,
      "learning_rate": 5.040680279751266e-05,
      "loss": 0.8379,
      "step": 192900
    },
    {
      "epoch": 3.3868221001070946,
      "grad_norm": 0.05686962056159399,
      "learning_rate": 5.0397224547239986e-05,
      "loss": 0.8389,
      "step": 192910
    },
    {
      "epoch": 3.386997664987096,
      "grad_norm": 0.05045928263614919,
      "learning_rate": 5.0387646938114194e-05,
      "loss": 0.8529,
      "step": 192920
    },
    {
      "epoch": 3.3871732298670976,
      "grad_norm": 0.04974958182299057,
      "learning_rate": 5.037806997025659e-05,
      "loss": 0.8449,
      "step": 192930
    },
    {
      "epoch": 3.3873487947470986,
      "grad_norm": 0.05415654928207128,
      "learning_rate": 5.036849364378853e-05,
      "loss": 0.8426,
      "step": 192940
    },
    {
      "epoch": 3.3875243596271,
      "grad_norm": 0.051150427843309525,
      "learning_rate": 5.0358917958831365e-05,
      "loss": 0.841,
      "step": 192950
    },
    {
      "epoch": 3.3876999245071016,
      "grad_norm": 0.04900491601018708,
      "learning_rate": 5.034934291550645e-05,
      "loss": 0.8471,
      "step": 192960
    },
    {
      "epoch": 3.387875489387103,
      "grad_norm": 0.05413413296675288,
      "learning_rate": 5.033976851393504e-05,
      "loss": 0.8431,
      "step": 192970
    },
    {
      "epoch": 3.3880510542671045,
      "grad_norm": 0.05165896088415281,
      "learning_rate": 5.0330194754238446e-05,
      "loss": 0.8333,
      "step": 192980
    },
    {
      "epoch": 3.388226619147106,
      "grad_norm": 0.04909065607334043,
      "learning_rate": 5.0320621636538055e-05,
      "loss": 0.8398,
      "step": 192990
    },
    {
      "epoch": 3.3884021840271075,
      "grad_norm": 0.04487472017710912,
      "learning_rate": 5.031104916095506e-05,
      "loss": 0.8448,
      "step": 193000
    },
    {
      "epoch": 3.3885777489071085,
      "grad_norm": 0.06624433325119518,
      "learning_rate": 5.030147732761079e-05,
      "loss": 0.8428,
      "step": 193010
    },
    {
      "epoch": 3.38875331378711,
      "grad_norm": 0.06497497208607621,
      "learning_rate": 5.029190613662651e-05,
      "loss": 0.848,
      "step": 193020
    },
    {
      "epoch": 3.3889288786671115,
      "grad_norm": 0.046561621059193675,
      "learning_rate": 5.0282335588123544e-05,
      "loss": 0.8376,
      "step": 193030
    },
    {
      "epoch": 3.389104443547113,
      "grad_norm": 0.04897487281666881,
      "learning_rate": 5.02727656822231e-05,
      "loss": 0.8419,
      "step": 193040
    },
    {
      "epoch": 3.3892800084271144,
      "grad_norm": 0.05126618736063735,
      "learning_rate": 5.0263196419046355e-05,
      "loss": 0.8416,
      "step": 193050
    },
    {
      "epoch": 3.3894555733071154,
      "grad_norm": 0.05496008440037184,
      "learning_rate": 5.025362779871472e-05,
      "loss": 0.8462,
      "step": 193060
    },
    {
      "epoch": 3.389631138187117,
      "grad_norm": 0.06729269026943481,
      "learning_rate": 5.024405982134931e-05,
      "loss": 0.8438,
      "step": 193070
    },
    {
      "epoch": 3.3898067030671184,
      "grad_norm": 0.04676171280006008,
      "learning_rate": 5.023449248707138e-05,
      "loss": 0.8327,
      "step": 193080
    },
    {
      "epoch": 3.38998226794712,
      "grad_norm": 0.06550582417592682,
      "learning_rate": 5.0224925796002196e-05,
      "loss": 0.8411,
      "step": 193090
    },
    {
      "epoch": 3.3901578328271214,
      "grad_norm": 0.04407418414559984,
      "learning_rate": 5.021535974826291e-05,
      "loss": 0.8426,
      "step": 193100
    },
    {
      "epoch": 3.390333397707123,
      "grad_norm": 0.03941777716548067,
      "learning_rate": 5.020579434397473e-05,
      "loss": 0.8405,
      "step": 193110
    },
    {
      "epoch": 3.3905089625871243,
      "grad_norm": 0.04150227618294421,
      "learning_rate": 5.019622958325889e-05,
      "loss": 0.8493,
      "step": 193120
    },
    {
      "epoch": 3.3906845274671253,
      "grad_norm": 0.0564939902843735,
      "learning_rate": 5.018666546623658e-05,
      "loss": 0.8382,
      "step": 193130
    },
    {
      "epoch": 3.390860092347127,
      "grad_norm": 0.04713510886314179,
      "learning_rate": 5.017710199302893e-05,
      "loss": 0.8313,
      "step": 193140
    },
    {
      "epoch": 3.3910356572271283,
      "grad_norm": 0.07047094589439587,
      "learning_rate": 5.016753916375714e-05,
      "loss": 0.8457,
      "step": 193150
    },
    {
      "epoch": 3.3912112221071298,
      "grad_norm": 0.06388024781387096,
      "learning_rate": 5.01579769785424e-05,
      "loss": 0.836,
      "step": 193160
    },
    {
      "epoch": 3.3913867869871313,
      "grad_norm": 0.08382497600273339,
      "learning_rate": 5.014841543750579e-05,
      "loss": 0.8417,
      "step": 193170
    },
    {
      "epoch": 3.3915623518671323,
      "grad_norm": 0.040511059160894584,
      "learning_rate": 5.013885454076852e-05,
      "loss": 0.8459,
      "step": 193180
    },
    {
      "epoch": 3.3917379167471338,
      "grad_norm": 0.10287742179711523,
      "learning_rate": 5.01292942884517e-05,
      "loss": 0.8445,
      "step": 193190
    },
    {
      "epoch": 3.3919134816271352,
      "grad_norm": 0.05507742230680841,
      "learning_rate": 5.011973468067651e-05,
      "loss": 0.8412,
      "step": 193200
    },
    {
      "epoch": 3.3920890465071367,
      "grad_norm": 0.05785572447827023,
      "learning_rate": 5.0110175717564034e-05,
      "loss": 0.8458,
      "step": 193210
    },
    {
      "epoch": 3.392264611387138,
      "grad_norm": 0.06267531175026334,
      "learning_rate": 5.010061739923532e-05,
      "loss": 0.8442,
      "step": 193220
    },
    {
      "epoch": 3.3924401762671397,
      "grad_norm": 0.06980805878280484,
      "learning_rate": 5.0091059725811596e-05,
      "loss": 0.832,
      "step": 193230
    },
    {
      "epoch": 3.392615741147141,
      "grad_norm": 0.046756343242839664,
      "learning_rate": 5.0081502697413864e-05,
      "loss": 0.838,
      "step": 193240
    },
    {
      "epoch": 3.392791306027142,
      "grad_norm": 0.04921355428546376,
      "learning_rate": 5.0071946314163256e-05,
      "loss": 0.8353,
      "step": 193250
    },
    {
      "epoch": 3.3929668709071437,
      "grad_norm": 0.04751022705006046,
      "learning_rate": 5.0062390576180884e-05,
      "loss": 0.8436,
      "step": 193260
    },
    {
      "epoch": 3.393142435787145,
      "grad_norm": 0.057675346977721156,
      "learning_rate": 5.005283548358776e-05,
      "loss": 0.8413,
      "step": 193270
    },
    {
      "epoch": 3.3933180006671466,
      "grad_norm": 0.05299714034484361,
      "learning_rate": 5.004328103650496e-05,
      "loss": 0.8329,
      "step": 193280
    },
    {
      "epoch": 3.393493565547148,
      "grad_norm": 0.04487185110151435,
      "learning_rate": 5.0033727235053576e-05,
      "loss": 0.8448,
      "step": 193290
    },
    {
      "epoch": 3.393669130427149,
      "grad_norm": 0.05314376300515251,
      "learning_rate": 5.0024174079354644e-05,
      "loss": 0.8317,
      "step": 193300
    },
    {
      "epoch": 3.3938446953071506,
      "grad_norm": 0.06933736514059612,
      "learning_rate": 5.0014621569529177e-05,
      "loss": 0.8445,
      "step": 193310
    },
    {
      "epoch": 3.394020260187152,
      "grad_norm": 0.048409334125121585,
      "learning_rate": 5.0005069705698224e-05,
      "loss": 0.8372,
      "step": 193320
    },
    {
      "epoch": 3.3941958250671536,
      "grad_norm": 0.05103532184243316,
      "learning_rate": 4.999551848798285e-05,
      "loss": 0.8405,
      "step": 193330
    },
    {
      "epoch": 3.394371389947155,
      "grad_norm": 0.051078584343704166,
      "learning_rate": 4.998596791650399e-05,
      "loss": 0.855,
      "step": 193340
    },
    {
      "epoch": 3.3945469548271565,
      "grad_norm": 0.06372212766403614,
      "learning_rate": 4.9976417991382694e-05,
      "loss": 0.8355,
      "step": 193350
    },
    {
      "epoch": 3.394722519707158,
      "grad_norm": 0.051553052919880736,
      "learning_rate": 4.996686871273996e-05,
      "loss": 0.8463,
      "step": 193360
    },
    {
      "epoch": 3.394898084587159,
      "grad_norm": 0.04516745305503242,
      "learning_rate": 4.995732008069682e-05,
      "loss": 0.8373,
      "step": 193370
    },
    {
      "epoch": 3.3950736494671605,
      "grad_norm": 0.04659411419829517,
      "learning_rate": 4.9947772095374216e-05,
      "loss": 0.838,
      "step": 193380
    },
    {
      "epoch": 3.395249214347162,
      "grad_norm": 0.0638625881993247,
      "learning_rate": 4.993822475689305e-05,
      "loss": 0.849,
      "step": 193390
    },
    {
      "epoch": 3.3954247792271635,
      "grad_norm": 0.06394299372980679,
      "learning_rate": 4.9928678065374425e-05,
      "loss": 0.8332,
      "step": 193400
    },
    {
      "epoch": 3.395600344107165,
      "grad_norm": 0.05964343701048201,
      "learning_rate": 4.9919132020939205e-05,
      "loss": 0.8388,
      "step": 193410
    },
    {
      "epoch": 3.395775908987166,
      "grad_norm": 0.046728327701259904,
      "learning_rate": 4.9909586623708376e-05,
      "loss": 0.8405,
      "step": 193420
    },
    {
      "epoch": 3.3959514738671674,
      "grad_norm": 0.07728742144669959,
      "learning_rate": 4.9900041873802914e-05,
      "loss": 0.8273,
      "step": 193430
    },
    {
      "epoch": 3.396127038747169,
      "grad_norm": 0.0643739978007083,
      "learning_rate": 4.989049777134368e-05,
      "loss": 0.848,
      "step": 193440
    },
    {
      "epoch": 3.3963026036271704,
      "grad_norm": 0.048566313225714926,
      "learning_rate": 4.988095431645163e-05,
      "loss": 0.8443,
      "step": 193450
    },
    {
      "epoch": 3.396478168507172,
      "grad_norm": 0.061784130868471114,
      "learning_rate": 4.9871411509247695e-05,
      "loss": 0.8432,
      "step": 193460
    },
    {
      "epoch": 3.3966537333871734,
      "grad_norm": 0.04840480083937878,
      "learning_rate": 4.9861869349852796e-05,
      "loss": 0.8391,
      "step": 193470
    },
    {
      "epoch": 3.396829298267175,
      "grad_norm": 0.047603952870023065,
      "learning_rate": 4.9852327838387795e-05,
      "loss": 0.8311,
      "step": 193480
    },
    {
      "epoch": 3.397004863147176,
      "grad_norm": 0.05675104013585827,
      "learning_rate": 4.984278697497358e-05,
      "loss": 0.8341,
      "step": 193490
    },
    {
      "epoch": 3.3971804280271773,
      "grad_norm": 0.06166414260229416,
      "learning_rate": 4.983324675973111e-05,
      "loss": 0.836,
      "step": 193500
    },
    {
      "epoch": 3.397355992907179,
      "grad_norm": 0.054571701709906065,
      "learning_rate": 4.982370719278117e-05,
      "loss": 0.8416,
      "step": 193510
    },
    {
      "epoch": 3.3975315577871803,
      "grad_norm": 0.06599583646610371,
      "learning_rate": 4.9814168274244664e-05,
      "loss": 0.842,
      "step": 193520
    },
    {
      "epoch": 3.397707122667182,
      "grad_norm": 0.06383850112702254,
      "learning_rate": 4.980463000424246e-05,
      "loss": 0.8353,
      "step": 193530
    },
    {
      "epoch": 3.397882687547183,
      "grad_norm": 0.07836354551246208,
      "learning_rate": 4.979509238289545e-05,
      "loss": 0.8395,
      "step": 193540
    },
    {
      "epoch": 3.3980582524271843,
      "grad_norm": 0.06795671573605544,
      "learning_rate": 4.978555541032442e-05,
      "loss": 0.8428,
      "step": 193550
    },
    {
      "epoch": 3.3982338173071858,
      "grad_norm": 0.06537237144270112,
      "learning_rate": 4.977601908665014e-05,
      "loss": 0.8456,
      "step": 193560
    },
    {
      "epoch": 3.3984093821871872,
      "grad_norm": 0.05321635580663171,
      "learning_rate": 4.9766483411993606e-05,
      "loss": 0.842,
      "step": 193570
    },
    {
      "epoch": 3.3985849470671887,
      "grad_norm": 0.04716189841531798,
      "learning_rate": 4.9756948386475495e-05,
      "loss": 0.8388,
      "step": 193580
    },
    {
      "epoch": 3.39876051194719,
      "grad_norm": 0.05046675966325508,
      "learning_rate": 4.974741401021667e-05,
      "loss": 0.8429,
      "step": 193590
    },
    {
      "epoch": 3.3989360768271917,
      "grad_norm": 0.049520284762534304,
      "learning_rate": 4.973788028333797e-05,
      "loss": 0.8379,
      "step": 193600
    },
    {
      "epoch": 3.3991116417071927,
      "grad_norm": 0.05093249829755091,
      "learning_rate": 4.9728347205960125e-05,
      "loss": 0.8341,
      "step": 193610
    },
    {
      "epoch": 3.399287206587194,
      "grad_norm": 0.06308516580197185,
      "learning_rate": 4.971881477820393e-05,
      "loss": 0.8463,
      "step": 193620
    },
    {
      "epoch": 3.3994627714671957,
      "grad_norm": 0.05421758246723273,
      "learning_rate": 4.97092830001902e-05,
      "loss": 0.847,
      "step": 193630
    },
    {
      "epoch": 3.399638336347197,
      "grad_norm": 0.058024554111457724,
      "learning_rate": 4.96997518720397e-05,
      "loss": 0.8445,
      "step": 193640
    },
    {
      "epoch": 3.3998139012271986,
      "grad_norm": 0.05254125118496924,
      "learning_rate": 4.969022139387315e-05,
      "loss": 0.8451,
      "step": 193650
    },
    {
      "epoch": 3.3999894661072,
      "grad_norm": 0.07752010937699252,
      "learning_rate": 4.968069156581133e-05,
      "loss": 0.8455,
      "step": 193660
    },
    {
      "epoch": 3.400165030987201,
      "grad_norm": 0.06324442290406092,
      "learning_rate": 4.967116238797502e-05,
      "loss": 0.841,
      "step": 193670
    },
    {
      "epoch": 3.4003405958672026,
      "grad_norm": 0.05915850662511783,
      "learning_rate": 4.966163386048488e-05,
      "loss": 0.8397,
      "step": 193680
    },
    {
      "epoch": 3.400516160747204,
      "grad_norm": 0.05281511061946961,
      "learning_rate": 4.9652105983461677e-05,
      "loss": 0.8476,
      "step": 193690
    },
    {
      "epoch": 3.4006917256272056,
      "grad_norm": 0.06695043443062473,
      "learning_rate": 4.964257875702613e-05,
      "loss": 0.8361,
      "step": 193700
    },
    {
      "epoch": 3.400867290507207,
      "grad_norm": 0.060623540093087104,
      "learning_rate": 4.963305218129899e-05,
      "loss": 0.8392,
      "step": 193710
    },
    {
      "epoch": 3.4010428553872085,
      "grad_norm": 0.046959093603905114,
      "learning_rate": 4.962352625640093e-05,
      "loss": 0.8419,
      "step": 193720
    },
    {
      "epoch": 3.4012184202672096,
      "grad_norm": 0.05093838128782122,
      "learning_rate": 4.9614000982452566e-05,
      "loss": 0.8346,
      "step": 193730
    },
    {
      "epoch": 3.401393985147211,
      "grad_norm": 0.05162584402537782,
      "learning_rate": 4.960447635957472e-05,
      "loss": 0.8372,
      "step": 193740
    },
    {
      "epoch": 3.4015695500272125,
      "grad_norm": 0.055666742496156794,
      "learning_rate": 4.9594952387887975e-05,
      "loss": 0.8346,
      "step": 193750
    },
    {
      "epoch": 3.401745114907214,
      "grad_norm": 0.07083411742807359,
      "learning_rate": 4.958542906751303e-05,
      "loss": 0.8439,
      "step": 193760
    },
    {
      "epoch": 3.4019206797872155,
      "grad_norm": 0.05570123541333144,
      "learning_rate": 4.957590639857061e-05,
      "loss": 0.838,
      "step": 193770
    },
    {
      "epoch": 3.402096244667217,
      "grad_norm": 0.08089845271480547,
      "learning_rate": 4.9566384381181255e-05,
      "loss": 0.8372,
      "step": 193780
    },
    {
      "epoch": 3.4022718095472184,
      "grad_norm": 0.05031620555806912,
      "learning_rate": 4.955686301546567e-05,
      "loss": 0.8388,
      "step": 193790
    },
    {
      "epoch": 3.4024473744272195,
      "grad_norm": 0.04478206225198084,
      "learning_rate": 4.954734230154449e-05,
      "loss": 0.8422,
      "step": 193800
    },
    {
      "epoch": 3.402622939307221,
      "grad_norm": 0.07132328182518904,
      "learning_rate": 4.953782223953839e-05,
      "loss": 0.8417,
      "step": 193810
    },
    {
      "epoch": 3.4027985041872224,
      "grad_norm": 0.05867142893196741,
      "learning_rate": 4.95283028295679e-05,
      "loss": 0.8398,
      "step": 193820
    },
    {
      "epoch": 3.402974069067224,
      "grad_norm": 0.048920472493168735,
      "learning_rate": 4.951878407175368e-05,
      "loss": 0.8371,
      "step": 193830
    },
    {
      "epoch": 3.4031496339472254,
      "grad_norm": 0.05709892168331806,
      "learning_rate": 4.9509265966216366e-05,
      "loss": 0.8412,
      "step": 193840
    },
    {
      "epoch": 3.4033251988272264,
      "grad_norm": 0.044134644089008704,
      "learning_rate": 4.949974851307649e-05,
      "loss": 0.8408,
      "step": 193850
    },
    {
      "epoch": 3.403500763707228,
      "grad_norm": 0.05482120912349114,
      "learning_rate": 4.9490231712454666e-05,
      "loss": 0.8357,
      "step": 193860
    },
    {
      "epoch": 3.4036763285872293,
      "grad_norm": 0.0654500695575328,
      "learning_rate": 4.948071556447147e-05,
      "loss": 0.8446,
      "step": 193870
    },
    {
      "epoch": 3.403851893467231,
      "grad_norm": 0.05667092780579422,
      "learning_rate": 4.9471200069247534e-05,
      "loss": 0.8431,
      "step": 193880
    },
    {
      "epoch": 3.4040274583472323,
      "grad_norm": 0.05390243058512263,
      "learning_rate": 4.946168522690336e-05,
      "loss": 0.8453,
      "step": 193890
    },
    {
      "epoch": 3.404203023227234,
      "grad_norm": 0.06654182833834492,
      "learning_rate": 4.945217103755943e-05,
      "loss": 0.8459,
      "step": 193900
    },
    {
      "epoch": 3.4043785881072353,
      "grad_norm": 0.07514669460048515,
      "learning_rate": 4.9442657501336456e-05,
      "loss": 0.836,
      "step": 193910
    },
    {
      "epoch": 3.4045541529872363,
      "grad_norm": 0.04582838643852945,
      "learning_rate": 4.943314461835486e-05,
      "loss": 0.847,
      "step": 193920
    },
    {
      "epoch": 3.4047297178672378,
      "grad_norm": 0.06434905321097097,
      "learning_rate": 4.94236323887352e-05,
      "loss": 0.8399,
      "step": 193930
    },
    {
      "epoch": 3.4049052827472392,
      "grad_norm": 0.04177580701607192,
      "learning_rate": 4.941412081259803e-05,
      "loss": 0.8497,
      "step": 193940
    },
    {
      "epoch": 3.4050808476272407,
      "grad_norm": 0.05893963487951595,
      "learning_rate": 4.940460989006381e-05,
      "loss": 0.8461,
      "step": 193950
    },
    {
      "epoch": 3.405256412507242,
      "grad_norm": 0.051573562437512835,
      "learning_rate": 4.939509962125307e-05,
      "loss": 0.8449,
      "step": 193960
    },
    {
      "epoch": 3.4054319773872432,
      "grad_norm": 0.051165152234904275,
      "learning_rate": 4.9385590006286295e-05,
      "loss": 0.8378,
      "step": 193970
    },
    {
      "epoch": 3.4056075422672447,
      "grad_norm": 0.06714960078065206,
      "learning_rate": 4.937608104528404e-05,
      "loss": 0.8451,
      "step": 193980
    },
    {
      "epoch": 3.405783107147246,
      "grad_norm": 0.04800993719022226,
      "learning_rate": 4.936657273836669e-05,
      "loss": 0.8407,
      "step": 193990
    },
    {
      "epoch": 3.4059586720272477,
      "grad_norm": 0.05672136971746648,
      "learning_rate": 4.935706508565475e-05,
      "loss": 0.8367,
      "step": 194000
    },
    {
      "epoch": 3.406134236907249,
      "grad_norm": 0.05552057959787185,
      "learning_rate": 4.934755808726875e-05,
      "loss": 0.8486,
      "step": 194010
    },
    {
      "epoch": 3.4063098017872506,
      "grad_norm": 0.04570459502933644,
      "learning_rate": 4.933805174332903e-05,
      "loss": 0.8395,
      "step": 194020
    },
    {
      "epoch": 3.406485366667252,
      "grad_norm": 0.045660415862172024,
      "learning_rate": 4.93285460539561e-05,
      "loss": 0.8411,
      "step": 194030
    },
    {
      "epoch": 3.406660931547253,
      "grad_norm": 0.06806662084552341,
      "learning_rate": 4.93190410192704e-05,
      "loss": 0.8436,
      "step": 194040
    },
    {
      "epoch": 3.4068364964272546,
      "grad_norm": 0.0487952080070197,
      "learning_rate": 4.9309536639392386e-05,
      "loss": 0.854,
      "step": 194050
    },
    {
      "epoch": 3.407012061307256,
      "grad_norm": 0.04390802958731697,
      "learning_rate": 4.930003291444245e-05,
      "loss": 0.8413,
      "step": 194060
    },
    {
      "epoch": 3.4071876261872576,
      "grad_norm": 0.04966191170370364,
      "learning_rate": 4.929052984454094e-05,
      "loss": 0.8341,
      "step": 194070
    },
    {
      "epoch": 3.407363191067259,
      "grad_norm": 0.04569787000467765,
      "learning_rate": 4.928102742980839e-05,
      "loss": 0.8355,
      "step": 194080
    },
    {
      "epoch": 3.40753875594726,
      "grad_norm": 0.06412421520239515,
      "learning_rate": 4.927152567036511e-05,
      "loss": 0.847,
      "step": 194090
    },
    {
      "epoch": 3.4077143208272616,
      "grad_norm": 0.04521563782951533,
      "learning_rate": 4.9262024566331496e-05,
      "loss": 0.8432,
      "step": 194100
    },
    {
      "epoch": 3.407889885707263,
      "grad_norm": 0.05007326371474944,
      "learning_rate": 4.9252524117828004e-05,
      "loss": 0.8477,
      "step": 194110
    },
    {
      "epoch": 3.4080654505872645,
      "grad_norm": 0.05258204075683385,
      "learning_rate": 4.92430243249749e-05,
      "loss": 0.8345,
      "step": 194120
    },
    {
      "epoch": 3.408241015467266,
      "grad_norm": 0.04695565041007934,
      "learning_rate": 4.923352518789259e-05,
      "loss": 0.8444,
      "step": 194130
    },
    {
      "epoch": 3.4084165803472675,
      "grad_norm": 0.061998809247954485,
      "learning_rate": 4.922402670670145e-05,
      "loss": 0.8427,
      "step": 194140
    },
    {
      "epoch": 3.408592145227269,
      "grad_norm": 0.04932292675704862,
      "learning_rate": 4.921452888152184e-05,
      "loss": 0.8395,
      "step": 194150
    },
    {
      "epoch": 3.40876771010727,
      "grad_norm": 0.05564956981266433,
      "learning_rate": 4.920503171247405e-05,
      "loss": 0.8425,
      "step": 194160
    },
    {
      "epoch": 3.4089432749872715,
      "grad_norm": 0.101039056404874,
      "learning_rate": 4.9195535199678425e-05,
      "loss": 0.8356,
      "step": 194170
    },
    {
      "epoch": 3.409118839867273,
      "grad_norm": 0.07848125134755632,
      "learning_rate": 4.918603934325534e-05,
      "loss": 0.8362,
      "step": 194180
    },
    {
      "epoch": 3.4092944047472744,
      "grad_norm": 0.050206865611582274,
      "learning_rate": 4.917654414332502e-05,
      "loss": 0.8401,
      "step": 194190
    },
    {
      "epoch": 3.409469969627276,
      "grad_norm": 0.0536346011766557,
      "learning_rate": 4.916704960000782e-05,
      "loss": 0.846,
      "step": 194200
    },
    {
      "epoch": 3.409645534507277,
      "grad_norm": 0.05027977291654508,
      "learning_rate": 4.915755571342404e-05,
      "loss": 0.8376,
      "step": 194210
    },
    {
      "epoch": 3.4098210993872784,
      "grad_norm": 0.07664927121350057,
      "learning_rate": 4.9148062483694005e-05,
      "loss": 0.8458,
      "step": 194220
    },
    {
      "epoch": 3.40999666426728,
      "grad_norm": 0.055411071907644494,
      "learning_rate": 4.913856991093796e-05,
      "loss": 0.8504,
      "step": 194230
    },
    {
      "epoch": 3.4101722291472814,
      "grad_norm": 0.05635288004619002,
      "learning_rate": 4.912907799527608e-05,
      "loss": 0.8452,
      "step": 194240
    },
    {
      "epoch": 3.410347794027283,
      "grad_norm": 0.0475562236678883,
      "learning_rate": 4.9119586736828816e-05,
      "loss": 0.8442,
      "step": 194250
    },
    {
      "epoch": 3.4105233589072843,
      "grad_norm": 0.06464925628476453,
      "learning_rate": 4.911009613571628e-05,
      "loss": 0.8404,
      "step": 194260
    },
    {
      "epoch": 3.410698923787286,
      "grad_norm": 0.08472724851545443,
      "learning_rate": 4.9100606192058777e-05,
      "loss": 0.8348,
      "step": 194270
    },
    {
      "epoch": 3.410874488667287,
      "grad_norm": 0.054074708314567475,
      "learning_rate": 4.909111690597658e-05,
      "loss": 0.8433,
      "step": 194280
    },
    {
      "epoch": 3.4110500535472883,
      "grad_norm": 0.06155282216920898,
      "learning_rate": 4.908162827758984e-05,
      "loss": 0.8363,
      "step": 194290
    },
    {
      "epoch": 3.4112256184272898,
      "grad_norm": 0.04884427785058404,
      "learning_rate": 4.907214030701882e-05,
      "loss": 0.8352,
      "step": 194300
    },
    {
      "epoch": 3.4114011833072913,
      "grad_norm": 0.05491351603754522,
      "learning_rate": 4.906265299438372e-05,
      "loss": 0.8386,
      "step": 194310
    },
    {
      "epoch": 3.4115767481872927,
      "grad_norm": 0.048733291085006014,
      "learning_rate": 4.9053166339804824e-05,
      "loss": 0.8403,
      "step": 194320
    },
    {
      "epoch": 3.411752313067294,
      "grad_norm": 0.059772447649364464,
      "learning_rate": 4.904368034340222e-05,
      "loss": 0.842,
      "step": 194330
    },
    {
      "epoch": 3.4119278779472952,
      "grad_norm": 0.058232426078806666,
      "learning_rate": 4.903419500529612e-05,
      "loss": 0.8475,
      "step": 194340
    },
    {
      "epoch": 3.4121034428272967,
      "grad_norm": 0.05878455620358811,
      "learning_rate": 4.9024710325606796e-05,
      "loss": 0.8336,
      "step": 194350
    },
    {
      "epoch": 3.412279007707298,
      "grad_norm": 0.052351180578817415,
      "learning_rate": 4.9015226304454306e-05,
      "loss": 0.8342,
      "step": 194360
    },
    {
      "epoch": 3.4124545725872997,
      "grad_norm": 0.07931606680940201,
      "learning_rate": 4.9005742941958865e-05,
      "loss": 0.8405,
      "step": 194370
    },
    {
      "epoch": 3.412630137467301,
      "grad_norm": 0.05738990893547347,
      "learning_rate": 4.8996260238240615e-05,
      "loss": 0.8399,
      "step": 194380
    },
    {
      "epoch": 3.4128057023473026,
      "grad_norm": 0.04784742571584301,
      "learning_rate": 4.898677819341975e-05,
      "loss": 0.8376,
      "step": 194390
    },
    {
      "epoch": 3.4129812672273037,
      "grad_norm": 0.047129685299179574,
      "learning_rate": 4.897729680761638e-05,
      "loss": 0.8395,
      "step": 194400
    },
    {
      "epoch": 3.413156832107305,
      "grad_norm": 0.04876514712165687,
      "learning_rate": 4.8967816080950565e-05,
      "loss": 0.8327,
      "step": 194410
    },
    {
      "epoch": 3.4133323969873066,
      "grad_norm": 0.05637000800951124,
      "learning_rate": 4.895833601354255e-05,
      "loss": 0.8455,
      "step": 194420
    },
    {
      "epoch": 3.413507961867308,
      "grad_norm": 0.050729013408154944,
      "learning_rate": 4.8948856605512366e-05,
      "loss": 0.8437,
      "step": 194430
    },
    {
      "epoch": 3.4136835267473096,
      "grad_norm": 0.05634100706276529,
      "learning_rate": 4.893937785698015e-05,
      "loss": 0.8419,
      "step": 194440
    },
    {
      "epoch": 3.413859091627311,
      "grad_norm": 0.04484919277824969,
      "learning_rate": 4.8929899768066015e-05,
      "loss": 0.8382,
      "step": 194450
    },
    {
      "epoch": 3.4140346565073125,
      "grad_norm": 0.06979831523679456,
      "learning_rate": 4.892042233889e-05,
      "loss": 0.8401,
      "step": 194460
    },
    {
      "epoch": 3.4142102213873136,
      "grad_norm": 0.04718474806674701,
      "learning_rate": 4.891094556957222e-05,
      "loss": 0.8404,
      "step": 194470
    },
    {
      "epoch": 3.414385786267315,
      "grad_norm": 0.05952625607939998,
      "learning_rate": 4.890146946023273e-05,
      "loss": 0.8342,
      "step": 194480
    },
    {
      "epoch": 3.4145613511473165,
      "grad_norm": 0.06205361297454049,
      "learning_rate": 4.889199401099165e-05,
      "loss": 0.8312,
      "step": 194490
    },
    {
      "epoch": 3.414736916027318,
      "grad_norm": 0.05501928028854676,
      "learning_rate": 4.888251922196895e-05,
      "loss": 0.8455,
      "step": 194500
    },
    {
      "epoch": 3.4149124809073195,
      "grad_norm": 0.05157517778503616,
      "learning_rate": 4.887304509328472e-05,
      "loss": 0.835,
      "step": 194510
    },
    {
      "epoch": 3.4150880457873205,
      "grad_norm": 0.048002968688585966,
      "learning_rate": 4.8863571625059044e-05,
      "loss": 0.8377,
      "step": 194520
    },
    {
      "epoch": 3.415263610667322,
      "grad_norm": 0.04883995820291457,
      "learning_rate": 4.885409881741188e-05,
      "loss": 0.8424,
      "step": 194530
    },
    {
      "epoch": 3.4154391755473235,
      "grad_norm": 0.060038827219895584,
      "learning_rate": 4.884462667046325e-05,
      "loss": 0.8374,
      "step": 194540
    },
    {
      "epoch": 3.415614740427325,
      "grad_norm": 0.057890659626191984,
      "learning_rate": 4.883515518433321e-05,
      "loss": 0.839,
      "step": 194550
    },
    {
      "epoch": 3.4157903053073264,
      "grad_norm": 0.056656630810105024,
      "learning_rate": 4.882568435914179e-05,
      "loss": 0.8424,
      "step": 194560
    },
    {
      "epoch": 3.415965870187328,
      "grad_norm": 0.0430064859043324,
      "learning_rate": 4.881621419500895e-05,
      "loss": 0.8415,
      "step": 194570
    },
    {
      "epoch": 3.4161414350673294,
      "grad_norm": 0.04798302698982355,
      "learning_rate": 4.8806744692054604e-05,
      "loss": 0.8428,
      "step": 194580
    },
    {
      "epoch": 3.4163169999473304,
      "grad_norm": 0.05283042920216587,
      "learning_rate": 4.8797275850398866e-05,
      "loss": 0.8475,
      "step": 194590
    },
    {
      "epoch": 3.416492564827332,
      "grad_norm": 0.04920837852756628,
      "learning_rate": 4.8787807670161626e-05,
      "loss": 0.8499,
      "step": 194600
    },
    {
      "epoch": 3.4166681297073334,
      "grad_norm": 0.06735709885035475,
      "learning_rate": 4.877834015146286e-05,
      "loss": 0.8413,
      "step": 194610
    },
    {
      "epoch": 3.416843694587335,
      "grad_norm": 0.047733402939448624,
      "learning_rate": 4.876887329442258e-05,
      "loss": 0.8409,
      "step": 194620
    },
    {
      "epoch": 3.4170192594673363,
      "grad_norm": 0.05939315633600334,
      "learning_rate": 4.875940709916064e-05,
      "loss": 0.8351,
      "step": 194630
    },
    {
      "epoch": 3.4171948243473373,
      "grad_norm": 0.04477995753841529,
      "learning_rate": 4.8749941565797014e-05,
      "loss": 0.8404,
      "step": 194640
    },
    {
      "epoch": 3.417370389227339,
      "grad_norm": 0.04693209400527131,
      "learning_rate": 4.874047669445166e-05,
      "loss": 0.8414,
      "step": 194650
    },
    {
      "epoch": 3.4175459541073403,
      "grad_norm": 0.055605761216834936,
      "learning_rate": 4.873101248524451e-05,
      "loss": 0.8374,
      "step": 194660
    },
    {
      "epoch": 3.4177215189873418,
      "grad_norm": 0.0565069692518219,
      "learning_rate": 4.872154893829541e-05,
      "loss": 0.8438,
      "step": 194670
    },
    {
      "epoch": 3.4178970838673433,
      "grad_norm": 0.05132572348646881,
      "learning_rate": 4.871208605372431e-05,
      "loss": 0.8434,
      "step": 194680
    },
    {
      "epoch": 3.4180726487473447,
      "grad_norm": 0.05425708830610856,
      "learning_rate": 4.870262383165114e-05,
      "loss": 0.8414,
      "step": 194690
    },
    {
      "epoch": 3.418248213627346,
      "grad_norm": 0.05085700944977528,
      "learning_rate": 4.869316227219571e-05,
      "loss": 0.8444,
      "step": 194700
    },
    {
      "epoch": 3.4184237785073472,
      "grad_norm": 0.04768102503033974,
      "learning_rate": 4.8683701375477934e-05,
      "loss": 0.8433,
      "step": 194710
    },
    {
      "epoch": 3.4185993433873487,
      "grad_norm": 0.045619912367383164,
      "learning_rate": 4.867424114161769e-05,
      "loss": 0.8401,
      "step": 194720
    },
    {
      "epoch": 3.41877490826735,
      "grad_norm": 0.05250187397648932,
      "learning_rate": 4.8664781570734876e-05,
      "loss": 0.8334,
      "step": 194730
    },
    {
      "epoch": 3.4189504731473517,
      "grad_norm": 0.05177062508511226,
      "learning_rate": 4.8655322662949305e-05,
      "loss": 0.8342,
      "step": 194740
    },
    {
      "epoch": 3.419126038027353,
      "grad_norm": 0.05890094603461449,
      "learning_rate": 4.864586441838076e-05,
      "loss": 0.8467,
      "step": 194750
    },
    {
      "epoch": 3.419301602907354,
      "grad_norm": 0.06424506974306896,
      "learning_rate": 4.863640683714922e-05,
      "loss": 0.8426,
      "step": 194760
    },
    {
      "epoch": 3.4194771677873557,
      "grad_norm": 0.055414083047634004,
      "learning_rate": 4.86269499193744e-05,
      "loss": 0.8402,
      "step": 194770
    },
    {
      "epoch": 3.419652732667357,
      "grad_norm": 0.053295872946973855,
      "learning_rate": 4.861749366517617e-05,
      "loss": 0.846,
      "step": 194780
    },
    {
      "epoch": 3.4198282975473586,
      "grad_norm": 0.04674748020581778,
      "learning_rate": 4.8608038074674354e-05,
      "loss": 0.8407,
      "step": 194790
    },
    {
      "epoch": 3.42000386242736,
      "grad_norm": 0.057666588183842,
      "learning_rate": 4.859858314798871e-05,
      "loss": 0.8298,
      "step": 194800
    },
    {
      "epoch": 3.4201794273073616,
      "grad_norm": 0.05249880687976786,
      "learning_rate": 4.858912888523907e-05,
      "loss": 0.8409,
      "step": 194810
    },
    {
      "epoch": 3.420354992187363,
      "grad_norm": 0.048379581351801995,
      "learning_rate": 4.85796752865452e-05,
      "loss": 0.8359,
      "step": 194820
    },
    {
      "epoch": 3.420530557067364,
      "grad_norm": 0.0673837772401913,
      "learning_rate": 4.8570222352026925e-05,
      "loss": 0.8449,
      "step": 194830
    },
    {
      "epoch": 3.4207061219473656,
      "grad_norm": 0.05190546040459106,
      "learning_rate": 4.856077008180396e-05,
      "loss": 0.8435,
      "step": 194840
    },
    {
      "epoch": 3.420881686827367,
      "grad_norm": 0.057264249068045586,
      "learning_rate": 4.855131847599607e-05,
      "loss": 0.8409,
      "step": 194850
    },
    {
      "epoch": 3.4210572517073685,
      "grad_norm": 0.05425030822717686,
      "learning_rate": 4.854186753472307e-05,
      "loss": 0.8411,
      "step": 194860
    },
    {
      "epoch": 3.42123281658737,
      "grad_norm": 0.06180279839331746,
      "learning_rate": 4.853241725810463e-05,
      "loss": 0.837,
      "step": 194870
    },
    {
      "epoch": 3.421408381467371,
      "grad_norm": 0.061990260510265065,
      "learning_rate": 4.852296764626051e-05,
      "loss": 0.8411,
      "step": 194880
    },
    {
      "epoch": 3.4215839463473725,
      "grad_norm": 0.04812109198774781,
      "learning_rate": 4.851351869931045e-05,
      "loss": 0.8414,
      "step": 194890
    },
    {
      "epoch": 3.421759511227374,
      "grad_norm": 0.07693944249314111,
      "learning_rate": 4.850407041737421e-05,
      "loss": 0.8403,
      "step": 194900
    },
    {
      "epoch": 3.4219350761073755,
      "grad_norm": 0.04709764199890573,
      "learning_rate": 4.8494622800571444e-05,
      "loss": 0.8411,
      "step": 194910
    },
    {
      "epoch": 3.422110640987377,
      "grad_norm": 0.05353823705039114,
      "learning_rate": 4.848517584902181e-05,
      "loss": 0.8333,
      "step": 194920
    },
    {
      "epoch": 3.4222862058673784,
      "grad_norm": 0.06824954355173168,
      "learning_rate": 4.847572956284513e-05,
      "loss": 0.8386,
      "step": 194930
    },
    {
      "epoch": 3.42246177074738,
      "grad_norm": 0.042784820947494605,
      "learning_rate": 4.846628394216099e-05,
      "loss": 0.8474,
      "step": 194940
    },
    {
      "epoch": 3.422637335627381,
      "grad_norm": 0.046568957649862144,
      "learning_rate": 4.84568389870891e-05,
      "loss": 0.8366,
      "step": 194950
    },
    {
      "epoch": 3.4228129005073824,
      "grad_norm": 0.060393039520353894,
      "learning_rate": 4.844739469774917e-05,
      "loss": 0.8341,
      "step": 194960
    },
    {
      "epoch": 3.422988465387384,
      "grad_norm": 0.04581634350567502,
      "learning_rate": 4.843795107426077e-05,
      "loss": 0.8391,
      "step": 194970
    },
    {
      "epoch": 3.4231640302673854,
      "grad_norm": 0.05587560961102423,
      "learning_rate": 4.842850811674362e-05,
      "loss": 0.8343,
      "step": 194980
    },
    {
      "epoch": 3.423339595147387,
      "grad_norm": 0.057922071913518024,
      "learning_rate": 4.841906582531733e-05,
      "loss": 0.8402,
      "step": 194990
    },
    {
      "epoch": 3.4235151600273883,
      "grad_norm": 0.05190922930097808,
      "learning_rate": 4.84096242001016e-05,
      "loss": 0.8349,
      "step": 195000
    },
    {
      "epoch": 3.4236907249073893,
      "grad_norm": 0.05701682514681516,
      "learning_rate": 4.8400183241215974e-05,
      "loss": 0.8434,
      "step": 195010
    },
    {
      "epoch": 3.423866289787391,
      "grad_norm": 0.046650975848696936,
      "learning_rate": 4.83907429487801e-05,
      "loss": 0.8356,
      "step": 195020
    },
    {
      "epoch": 3.4240418546673923,
      "grad_norm": 0.04575022247750066,
      "learning_rate": 4.8381303322913644e-05,
      "loss": 0.8419,
      "step": 195030
    },
    {
      "epoch": 3.424217419547394,
      "grad_norm": 0.06077820160161449,
      "learning_rate": 4.8371864363736126e-05,
      "loss": 0.8391,
      "step": 195040
    },
    {
      "epoch": 3.4243929844273953,
      "grad_norm": 0.08785340878414707,
      "learning_rate": 4.8362426071367155e-05,
      "loss": 0.83,
      "step": 195050
    },
    {
      "epoch": 3.4245685493073967,
      "grad_norm": 0.0557000807820804,
      "learning_rate": 4.835298844592634e-05,
      "loss": 0.8451,
      "step": 195060
    },
    {
      "epoch": 3.4247441141873978,
      "grad_norm": 0.05149313680540499,
      "learning_rate": 4.83435514875333e-05,
      "loss": 0.8349,
      "step": 195070
    },
    {
      "epoch": 3.4249196790673992,
      "grad_norm": 0.07391155601493339,
      "learning_rate": 4.833411519630753e-05,
      "loss": 0.8485,
      "step": 195080
    },
    {
      "epoch": 3.4250952439474007,
      "grad_norm": 0.07162210382336336,
      "learning_rate": 4.832467957236857e-05,
      "loss": 0.8429,
      "step": 195090
    },
    {
      "epoch": 3.425270808827402,
      "grad_norm": 0.052125452458997024,
      "learning_rate": 4.831524461583607e-05,
      "loss": 0.8401,
      "step": 195100
    },
    {
      "epoch": 3.4254463737074037,
      "grad_norm": 0.06652538895620907,
      "learning_rate": 4.830581032682948e-05,
      "loss": 0.8403,
      "step": 195110
    },
    {
      "epoch": 3.425621938587405,
      "grad_norm": 0.05933363545835157,
      "learning_rate": 4.829637670546838e-05,
      "loss": 0.8433,
      "step": 195120
    },
    {
      "epoch": 3.425797503467406,
      "grad_norm": 0.06485448157837424,
      "learning_rate": 4.8286943751872324e-05,
      "loss": 0.8412,
      "step": 195130
    },
    {
      "epoch": 3.4259730683474077,
      "grad_norm": 0.06090682955754023,
      "learning_rate": 4.8277511466160744e-05,
      "loss": 0.8342,
      "step": 195140
    },
    {
      "epoch": 3.426148633227409,
      "grad_norm": 0.047449742914494705,
      "learning_rate": 4.826807984845321e-05,
      "loss": 0.8361,
      "step": 195150
    },
    {
      "epoch": 3.4263241981074106,
      "grad_norm": 0.06277850911487405,
      "learning_rate": 4.82586488988692e-05,
      "loss": 0.8434,
      "step": 195160
    },
    {
      "epoch": 3.426499762987412,
      "grad_norm": 0.06951774744109317,
      "learning_rate": 4.824921861752826e-05,
      "loss": 0.8376,
      "step": 195170
    },
    {
      "epoch": 3.4266753278674136,
      "grad_norm": 0.0629453914179549,
      "learning_rate": 4.8239789004549795e-05,
      "loss": 0.8352,
      "step": 195180
    },
    {
      "epoch": 3.4268508927474146,
      "grad_norm": 0.0541482462702964,
      "learning_rate": 4.823036006005332e-05,
      "loss": 0.8344,
      "step": 195190
    },
    {
      "epoch": 3.427026457627416,
      "grad_norm": 0.04514016745408344,
      "learning_rate": 4.822093178415832e-05,
      "loss": 0.8439,
      "step": 195200
    },
    {
      "epoch": 3.4272020225074176,
      "grad_norm": 0.05954742496296953,
      "learning_rate": 4.821150417698421e-05,
      "loss": 0.8397,
      "step": 195210
    },
    {
      "epoch": 3.427377587387419,
      "grad_norm": 0.06233899092567859,
      "learning_rate": 4.8202077238650436e-05,
      "loss": 0.8446,
      "step": 195220
    },
    {
      "epoch": 3.4275531522674205,
      "grad_norm": 0.044591043127678635,
      "learning_rate": 4.819265096927648e-05,
      "loss": 0.8502,
      "step": 195230
    },
    {
      "epoch": 3.427728717147422,
      "grad_norm": 0.05697036825598534,
      "learning_rate": 4.818322536898178e-05,
      "loss": 0.8526,
      "step": 195240
    },
    {
      "epoch": 3.4279042820274235,
      "grad_norm": 0.054868480391695076,
      "learning_rate": 4.817380043788575e-05,
      "loss": 0.8398,
      "step": 195250
    },
    {
      "epoch": 3.4280798469074245,
      "grad_norm": 0.04701127444935511,
      "learning_rate": 4.81643761761077e-05,
      "loss": 0.8373,
      "step": 195260
    },
    {
      "epoch": 3.428255411787426,
      "grad_norm": 0.05398004645174588,
      "learning_rate": 4.815495258376722e-05,
      "loss": 0.8386,
      "step": 195270
    },
    {
      "epoch": 3.4284309766674275,
      "grad_norm": 0.050123685950903496,
      "learning_rate": 4.814552966098358e-05,
      "loss": 0.8433,
      "step": 195280
    },
    {
      "epoch": 3.428606541547429,
      "grad_norm": 0.046196841788085065,
      "learning_rate": 4.8136107407876214e-05,
      "loss": 0.8343,
      "step": 195290
    },
    {
      "epoch": 3.4287821064274304,
      "grad_norm": 0.05422395952653248,
      "learning_rate": 4.812668582456452e-05,
      "loss": 0.8419,
      "step": 195300
    },
    {
      "epoch": 3.4289576713074315,
      "grad_norm": 0.050951414224088734,
      "learning_rate": 4.8117264911167825e-05,
      "loss": 0.8331,
      "step": 195310
    },
    {
      "epoch": 3.429133236187433,
      "grad_norm": 0.0658408846831527,
      "learning_rate": 4.810784466780551e-05,
      "loss": 0.8398,
      "step": 195320
    },
    {
      "epoch": 3.4293088010674344,
      "grad_norm": 0.0643751820714155,
      "learning_rate": 4.8098425094596946e-05,
      "loss": 0.846,
      "step": 195330
    },
    {
      "epoch": 3.429484365947436,
      "grad_norm": 0.051621481679089205,
      "learning_rate": 4.808900619166151e-05,
      "loss": 0.8398,
      "step": 195340
    },
    {
      "epoch": 3.4296599308274374,
      "grad_norm": 0.06700982726631366,
      "learning_rate": 4.807958795911846e-05,
      "loss": 0.8359,
      "step": 195350
    },
    {
      "epoch": 3.429835495707439,
      "grad_norm": 0.07047512680327787,
      "learning_rate": 4.8070170397087186e-05,
      "loss": 0.8416,
      "step": 195360
    },
    {
      "epoch": 3.4300110605874403,
      "grad_norm": 0.04884597009931388,
      "learning_rate": 4.8060753505687035e-05,
      "loss": 0.8351,
      "step": 195370
    },
    {
      "epoch": 3.4301866254674414,
      "grad_norm": 0.05717647041834128,
      "learning_rate": 4.805133728503723e-05,
      "loss": 0.8368,
      "step": 195380
    },
    {
      "epoch": 3.430362190347443,
      "grad_norm": 0.07469041079892912,
      "learning_rate": 4.804192173525714e-05,
      "loss": 0.838,
      "step": 195390
    },
    {
      "epoch": 3.4305377552274443,
      "grad_norm": 0.04795723233212944,
      "learning_rate": 4.803250685646605e-05,
      "loss": 0.8362,
      "step": 195400
    },
    {
      "epoch": 3.430713320107446,
      "grad_norm": 0.05576036010433958,
      "learning_rate": 4.802309264878328e-05,
      "loss": 0.8372,
      "step": 195410
    },
    {
      "epoch": 3.4308888849874473,
      "grad_norm": 0.055811944417536126,
      "learning_rate": 4.8013679112328095e-05,
      "loss": 0.839,
      "step": 195420
    },
    {
      "epoch": 3.4310644498674483,
      "grad_norm": 0.047003374260906006,
      "learning_rate": 4.8004266247219675e-05,
      "loss": 0.8381,
      "step": 195430
    },
    {
      "epoch": 3.4312400147474498,
      "grad_norm": 0.07456788103153786,
      "learning_rate": 4.799485405357742e-05,
      "loss": 0.8421,
      "step": 195440
    },
    {
      "epoch": 3.4314155796274513,
      "grad_norm": 0.051650507051066476,
      "learning_rate": 4.798544253152049e-05,
      "loss": 0.8379,
      "step": 195450
    },
    {
      "epoch": 3.4315911445074527,
      "grad_norm": 0.05744934395177363,
      "learning_rate": 4.797603168116816e-05,
      "loss": 0.8382,
      "step": 195460
    },
    {
      "epoch": 3.431766709387454,
      "grad_norm": 0.062105895781404935,
      "learning_rate": 4.796662150263971e-05,
      "loss": 0.8416,
      "step": 195470
    },
    {
      "epoch": 3.4319422742674557,
      "grad_norm": 0.05637915698492982,
      "learning_rate": 4.795721199605429e-05,
      "loss": 0.8398,
      "step": 195480
    },
    {
      "epoch": 3.432117839147457,
      "grad_norm": 0.048135043148526985,
      "learning_rate": 4.794780316153117e-05,
      "loss": 0.838,
      "step": 195490
    },
    {
      "epoch": 3.432293404027458,
      "grad_norm": 0.07528183078428438,
      "learning_rate": 4.793839499918954e-05,
      "loss": 0.8474,
      "step": 195500
    },
    {
      "epoch": 3.4324689689074597,
      "grad_norm": 0.05535158305260257,
      "learning_rate": 4.792898750914865e-05,
      "loss": 0.8391,
      "step": 195510
    },
    {
      "epoch": 3.432644533787461,
      "grad_norm": 0.05524466768095527,
      "learning_rate": 4.791958069152763e-05,
      "loss": 0.8417,
      "step": 195520
    },
    {
      "epoch": 3.4328200986674626,
      "grad_norm": 0.07073897902691206,
      "learning_rate": 4.79101745464457e-05,
      "loss": 0.838,
      "step": 195530
    },
    {
      "epoch": 3.432995663547464,
      "grad_norm": 0.058288394907415726,
      "learning_rate": 4.790076907402206e-05,
      "loss": 0.8384,
      "step": 195540
    },
    {
      "epoch": 3.433171228427465,
      "grad_norm": 0.04905045728177045,
      "learning_rate": 4.789136427437581e-05,
      "loss": 0.8483,
      "step": 195550
    },
    {
      "epoch": 3.4333467933074666,
      "grad_norm": 0.04756830470402652,
      "learning_rate": 4.788196014762617e-05,
      "loss": 0.8532,
      "step": 195560
    },
    {
      "epoch": 3.433522358187468,
      "grad_norm": 0.06862967459692124,
      "learning_rate": 4.7872556693892276e-05,
      "loss": 0.8458,
      "step": 195570
    },
    {
      "epoch": 3.4336979230674696,
      "grad_norm": 0.04916937728346072,
      "learning_rate": 4.786315391329329e-05,
      "loss": 0.835,
      "step": 195580
    },
    {
      "epoch": 3.433873487947471,
      "grad_norm": 0.051559450701315304,
      "learning_rate": 4.785375180594833e-05,
      "loss": 0.8416,
      "step": 195590
    },
    {
      "epoch": 3.4340490528274725,
      "grad_norm": 0.04864028691167062,
      "learning_rate": 4.784435037197645e-05,
      "loss": 0.8418,
      "step": 195600
    },
    {
      "epoch": 3.434224617707474,
      "grad_norm": 0.06434298584824033,
      "learning_rate": 4.783494961149692e-05,
      "loss": 0.8449,
      "step": 195610
    },
    {
      "epoch": 3.434400182587475,
      "grad_norm": 0.05087676709519241,
      "learning_rate": 4.7825549524628727e-05,
      "loss": 0.8433,
      "step": 195620
    },
    {
      "epoch": 3.4345757474674765,
      "grad_norm": 0.0523816371896874,
      "learning_rate": 4.781615011149101e-05,
      "loss": 0.834,
      "step": 195630
    },
    {
      "epoch": 3.434751312347478,
      "grad_norm": 0.053997021619054765,
      "learning_rate": 4.7806751372202896e-05,
      "loss": 0.8433,
      "step": 195640
    },
    {
      "epoch": 3.4349268772274795,
      "grad_norm": 0.06409044539296894,
      "learning_rate": 4.779735330688341e-05,
      "loss": 0.8423,
      "step": 195650
    },
    {
      "epoch": 3.435102442107481,
      "grad_norm": 0.05025471632546948,
      "learning_rate": 4.7787955915651636e-05,
      "loss": 0.8449,
      "step": 195660
    },
    {
      "epoch": 3.435278006987482,
      "grad_norm": 0.07473351660979602,
      "learning_rate": 4.7778559198626676e-05,
      "loss": 0.8405,
      "step": 195670
    },
    {
      "epoch": 3.4354535718674835,
      "grad_norm": 0.050089418911037256,
      "learning_rate": 4.7769163155927594e-05,
      "loss": 0.841,
      "step": 195680
    },
    {
      "epoch": 3.435629136747485,
      "grad_norm": 0.06221706993197521,
      "learning_rate": 4.775976778767339e-05,
      "loss": 0.8344,
      "step": 195690
    },
    {
      "epoch": 3.4358047016274864,
      "grad_norm": 0.043577046598010906,
      "learning_rate": 4.775037309398314e-05,
      "loss": 0.8432,
      "step": 195700
    },
    {
      "epoch": 3.435980266507488,
      "grad_norm": 0.057288859726025355,
      "learning_rate": 4.774097907497589e-05,
      "loss": 0.8357,
      "step": 195710
    },
    {
      "epoch": 3.4361558313874894,
      "grad_norm": 0.056726636328687685,
      "learning_rate": 4.773158573077061e-05,
      "loss": 0.8505,
      "step": 195720
    },
    {
      "epoch": 3.436331396267491,
      "grad_norm": 0.057695190027559295,
      "learning_rate": 4.772219306148635e-05,
      "loss": 0.8394,
      "step": 195730
    },
    {
      "epoch": 3.436506961147492,
      "grad_norm": 0.054912591726116795,
      "learning_rate": 4.771280106724211e-05,
      "loss": 0.8464,
      "step": 195740
    },
    {
      "epoch": 3.4366825260274934,
      "grad_norm": 0.04445308374417714,
      "learning_rate": 4.770340974815693e-05,
      "loss": 0.8463,
      "step": 195750
    },
    {
      "epoch": 3.436858090907495,
      "grad_norm": 0.05163489760046149,
      "learning_rate": 4.769401910434976e-05,
      "loss": 0.8438,
      "step": 195760
    },
    {
      "epoch": 3.4370336557874963,
      "grad_norm": 0.04276883526229845,
      "learning_rate": 4.768462913593951e-05,
      "loss": 0.836,
      "step": 195770
    },
    {
      "epoch": 3.437209220667498,
      "grad_norm": 0.04878310892940647,
      "learning_rate": 4.7675239843045306e-05,
      "loss": 0.8442,
      "step": 195780
    },
    {
      "epoch": 3.4373847855474993,
      "grad_norm": 0.045149757255204724,
      "learning_rate": 4.7665851225786e-05,
      "loss": 0.8427,
      "step": 195790
    },
    {
      "epoch": 3.4375603504275003,
      "grad_norm": 0.04963214359589337,
      "learning_rate": 4.765646328428057e-05,
      "loss": 0.8435,
      "step": 195800
    },
    {
      "epoch": 3.4377359153075018,
      "grad_norm": 0.07095920634052655,
      "learning_rate": 4.764707601864802e-05,
      "loss": 0.8481,
      "step": 195810
    },
    {
      "epoch": 3.4379114801875033,
      "grad_norm": 0.053894272493810826,
      "learning_rate": 4.7637689429007195e-05,
      "loss": 0.8399,
      "step": 195820
    },
    {
      "epoch": 3.4380870450675047,
      "grad_norm": 0.05352757809279118,
      "learning_rate": 4.7628303515477087e-05,
      "loss": 0.8424,
      "step": 195830
    },
    {
      "epoch": 3.438262609947506,
      "grad_norm": 0.05124728833010432,
      "learning_rate": 4.7618918278176586e-05,
      "loss": 0.8393,
      "step": 195840
    },
    {
      "epoch": 3.4384381748275077,
      "grad_norm": 0.043389208151813546,
      "learning_rate": 4.760953371722467e-05,
      "loss": 0.8404,
      "step": 195850
    },
    {
      "epoch": 3.4386137397075087,
      "grad_norm": 0.05567992194392075,
      "learning_rate": 4.7600149832740146e-05,
      "loss": 0.8446,
      "step": 195860
    },
    {
      "epoch": 3.43878930458751,
      "grad_norm": 0.04514571405631952,
      "learning_rate": 4.759076662484196e-05,
      "loss": 0.8348,
      "step": 195870
    },
    {
      "epoch": 3.4389648694675117,
      "grad_norm": 0.05182183789313628,
      "learning_rate": 4.758138409364904e-05,
      "loss": 0.8443,
      "step": 195880
    },
    {
      "epoch": 3.439140434347513,
      "grad_norm": 0.0445425849512007,
      "learning_rate": 4.7572002239280184e-05,
      "loss": 0.8329,
      "step": 195890
    },
    {
      "epoch": 3.4393159992275146,
      "grad_norm": 0.05118467204140835,
      "learning_rate": 4.7562621061854285e-05,
      "loss": 0.8335,
      "step": 195900
    },
    {
      "epoch": 3.439491564107516,
      "grad_norm": 0.05522424554132841,
      "learning_rate": 4.755324056149023e-05,
      "loss": 0.8362,
      "step": 195910
    },
    {
      "epoch": 3.4396671289875176,
      "grad_norm": 0.05911025675173855,
      "learning_rate": 4.75438607383069e-05,
      "loss": 0.8491,
      "step": 195920
    },
    {
      "epoch": 3.4398426938675186,
      "grad_norm": 0.06491560103306272,
      "learning_rate": 4.7534481592423095e-05,
      "loss": 0.8419,
      "step": 195930
    },
    {
      "epoch": 3.44001825874752,
      "grad_norm": 0.055971955233953,
      "learning_rate": 4.752510312395758e-05,
      "loss": 0.8391,
      "step": 195940
    },
    {
      "epoch": 3.4401938236275216,
      "grad_norm": 0.0525523249040413,
      "learning_rate": 4.751572533302933e-05,
      "loss": 0.8384,
      "step": 195950
    },
    {
      "epoch": 3.440369388507523,
      "grad_norm": 0.07371243797151461,
      "learning_rate": 4.7506348219757054e-05,
      "loss": 0.8452,
      "step": 195960
    },
    {
      "epoch": 3.4405449533875245,
      "grad_norm": 0.049010876443591006,
      "learning_rate": 4.74969717842596e-05,
      "loss": 0.8394,
      "step": 195970
    },
    {
      "epoch": 3.4407205182675256,
      "grad_norm": 0.050365371694553795,
      "learning_rate": 4.7487596026655807e-05,
      "loss": 0.8444,
      "step": 195980
    },
    {
      "epoch": 3.440896083147527,
      "grad_norm": 0.04437673014797061,
      "learning_rate": 4.747822094706439e-05,
      "loss": 0.832,
      "step": 195990
    },
    {
      "epoch": 3.4410716480275285,
      "grad_norm": 0.04632310477547041,
      "learning_rate": 4.7468846545604165e-05,
      "loss": 0.8538,
      "step": 196000
    },
    {
      "epoch": 3.44124721290753,
      "grad_norm": 0.054063253758333464,
      "learning_rate": 4.745947282239392e-05,
      "loss": 0.8366,
      "step": 196010
    },
    {
      "epoch": 3.4414227777875315,
      "grad_norm": 0.05141324482777217,
      "learning_rate": 4.7450099777552446e-05,
      "loss": 0.8383,
      "step": 196020
    },
    {
      "epoch": 3.441598342667533,
      "grad_norm": 0.06441951295374315,
      "learning_rate": 4.744072741119844e-05,
      "loss": 0.8396,
      "step": 196030
    },
    {
      "epoch": 3.4417739075475344,
      "grad_norm": 0.04980581422198285,
      "learning_rate": 4.7431355723450676e-05,
      "loss": 0.8467,
      "step": 196040
    },
    {
      "epoch": 3.4419494724275355,
      "grad_norm": 0.06909035193355492,
      "learning_rate": 4.742198471442795e-05,
      "loss": 0.8445,
      "step": 196050
    },
    {
      "epoch": 3.442125037307537,
      "grad_norm": 0.05312563478274068,
      "learning_rate": 4.741261438424889e-05,
      "loss": 0.8388,
      "step": 196060
    },
    {
      "epoch": 3.4423006021875384,
      "grad_norm": 0.06034812625925376,
      "learning_rate": 4.74032447330323e-05,
      "loss": 0.8385,
      "step": 196070
    },
    {
      "epoch": 3.44247616706754,
      "grad_norm": 0.05454329797756831,
      "learning_rate": 4.739387576089686e-05,
      "loss": 0.839,
      "step": 196080
    },
    {
      "epoch": 3.4426517319475414,
      "grad_norm": 0.047508162415083995,
      "learning_rate": 4.7384507467961315e-05,
      "loss": 0.8394,
      "step": 196090
    },
    {
      "epoch": 3.4428272968275424,
      "grad_norm": 0.04556774746781447,
      "learning_rate": 4.737513985434434e-05,
      "loss": 0.8412,
      "step": 196100
    },
    {
      "epoch": 3.443002861707544,
      "grad_norm": 0.07462893905577377,
      "learning_rate": 4.736577292016454e-05,
      "loss": 0.8419,
      "step": 196110
    },
    {
      "epoch": 3.4431784265875454,
      "grad_norm": 0.04370697333452839,
      "learning_rate": 4.735640666554077e-05,
      "loss": 0.8401,
      "step": 196120
    },
    {
      "epoch": 3.443353991467547,
      "grad_norm": 0.047887530685747554,
      "learning_rate": 4.7347041090591554e-05,
      "loss": 0.8368,
      "step": 196130
    },
    {
      "epoch": 3.4435295563475483,
      "grad_norm": 0.062191522858930245,
      "learning_rate": 4.733767619543562e-05,
      "loss": 0.8458,
      "step": 196140
    },
    {
      "epoch": 3.44370512122755,
      "grad_norm": 0.05373495960876097,
      "learning_rate": 4.7328311980191645e-05,
      "loss": 0.8394,
      "step": 196150
    },
    {
      "epoch": 3.4438806861075513,
      "grad_norm": 0.061888888770178806,
      "learning_rate": 4.7318948444978205e-05,
      "loss": 0.8366,
      "step": 196160
    },
    {
      "epoch": 3.4440562509875523,
      "grad_norm": 0.06141804252310623,
      "learning_rate": 4.730958558991398e-05,
      "loss": 0.8391,
      "step": 196170
    },
    {
      "epoch": 3.444231815867554,
      "grad_norm": 0.05291101004948447,
      "learning_rate": 4.730022341511759e-05,
      "loss": 0.8409,
      "step": 196180
    },
    {
      "epoch": 3.4444073807475553,
      "grad_norm": 0.06445769257562677,
      "learning_rate": 4.729086192070771e-05,
      "loss": 0.8439,
      "step": 196190
    },
    {
      "epoch": 3.4445829456275567,
      "grad_norm": 0.05213811459021087,
      "learning_rate": 4.728150110680285e-05,
      "loss": 0.8447,
      "step": 196200
    },
    {
      "epoch": 3.444758510507558,
      "grad_norm": 0.04499456711928696,
      "learning_rate": 4.727214097352168e-05,
      "loss": 0.8481,
      "step": 196210
    },
    {
      "epoch": 3.4449340753875592,
      "grad_norm": 0.047617729597516686,
      "learning_rate": 4.7262781520982805e-05,
      "loss": 0.8395,
      "step": 196220
    },
    {
      "epoch": 3.4451096402675607,
      "grad_norm": 0.048279069395103076,
      "learning_rate": 4.7253422749304757e-05,
      "loss": 0.8402,
      "step": 196230
    },
    {
      "epoch": 3.445285205147562,
      "grad_norm": 0.051005390339278756,
      "learning_rate": 4.724406465860615e-05,
      "loss": 0.8344,
      "step": 196240
    },
    {
      "epoch": 3.4454607700275637,
      "grad_norm": 0.05708838074618044,
      "learning_rate": 4.723470724900553e-05,
      "loss": 0.8395,
      "step": 196250
    },
    {
      "epoch": 3.445636334907565,
      "grad_norm": 0.051541659780804716,
      "learning_rate": 4.7225350520621523e-05,
      "loss": 0.8361,
      "step": 196260
    },
    {
      "epoch": 3.4458118997875666,
      "grad_norm": 0.06477743159867472,
      "learning_rate": 4.721599447357263e-05,
      "loss": 0.8406,
      "step": 196270
    },
    {
      "epoch": 3.445987464667568,
      "grad_norm": 0.0508847071031361,
      "learning_rate": 4.720663910797731e-05,
      "loss": 0.8345,
      "step": 196280
    },
    {
      "epoch": 3.446163029547569,
      "grad_norm": 0.08930115206230974,
      "learning_rate": 4.719728442395427e-05,
      "loss": 0.8422,
      "step": 196290
    },
    {
      "epoch": 3.4463385944275706,
      "grad_norm": 0.046704295684666886,
      "learning_rate": 4.71879304216219e-05,
      "loss": 0.8461,
      "step": 196300
    },
    {
      "epoch": 3.446514159307572,
      "grad_norm": 0.04710873229804889,
      "learning_rate": 4.717857710109877e-05,
      "loss": 0.8475,
      "step": 196310
    },
    {
      "epoch": 3.4466897241875736,
      "grad_norm": 0.05018702913103075,
      "learning_rate": 4.716922446250342e-05,
      "loss": 0.8475,
      "step": 196320
    },
    {
      "epoch": 3.446865289067575,
      "grad_norm": 0.0553967653487441,
      "learning_rate": 4.7159872505954266e-05,
      "loss": 0.8421,
      "step": 196330
    },
    {
      "epoch": 3.447040853947576,
      "grad_norm": 0.04864246320372023,
      "learning_rate": 4.715052123156987e-05,
      "loss": 0.8358,
      "step": 196340
    },
    {
      "epoch": 3.4472164188275776,
      "grad_norm": 0.06620373069877032,
      "learning_rate": 4.714117063946866e-05,
      "loss": 0.849,
      "step": 196350
    },
    {
      "epoch": 3.447391983707579,
      "grad_norm": 0.046525808529272225,
      "learning_rate": 4.71318207297692e-05,
      "loss": 0.8388,
      "step": 196360
    },
    {
      "epoch": 3.4475675485875805,
      "grad_norm": 0.04705927569786345,
      "learning_rate": 4.712247150258984e-05,
      "loss": 0.8422,
      "step": 196370
    },
    {
      "epoch": 3.447743113467582,
      "grad_norm": 0.05772552187229953,
      "learning_rate": 4.711312295804911e-05,
      "loss": 0.8384,
      "step": 196380
    },
    {
      "epoch": 3.4479186783475835,
      "grad_norm": 0.051643297534160734,
      "learning_rate": 4.710377509626546e-05,
      "loss": 0.8388,
      "step": 196390
    },
    {
      "epoch": 3.448094243227585,
      "grad_norm": 0.0584911690633758,
      "learning_rate": 4.709442791735728e-05,
      "loss": 0.8431,
      "step": 196400
    },
    {
      "epoch": 3.448269808107586,
      "grad_norm": 0.051898878493141454,
      "learning_rate": 4.708508142144303e-05,
      "loss": 0.8371,
      "step": 196410
    },
    {
      "epoch": 3.4484453729875875,
      "grad_norm": 0.06232864448125314,
      "learning_rate": 4.7075735608641124e-05,
      "loss": 0.838,
      "step": 196420
    },
    {
      "epoch": 3.448620937867589,
      "grad_norm": 0.049372747459990145,
      "learning_rate": 4.706639047907003e-05,
      "loss": 0.839,
      "step": 196430
    },
    {
      "epoch": 3.4487965027475904,
      "grad_norm": 0.05156009275954762,
      "learning_rate": 4.70570460328481e-05,
      "loss": 0.8352,
      "step": 196440
    },
    {
      "epoch": 3.448972067627592,
      "grad_norm": 0.06461189657105007,
      "learning_rate": 4.704770227009366e-05,
      "loss": 0.8421,
      "step": 196450
    },
    {
      "epoch": 3.4491476325075934,
      "grad_norm": 0.05592807308659329,
      "learning_rate": 4.7038359190925246e-05,
      "loss": 0.8392,
      "step": 196460
    },
    {
      "epoch": 3.4493231973875944,
      "grad_norm": 0.06099622673717255,
      "learning_rate": 4.7029016795461126e-05,
      "loss": 0.8418,
      "step": 196470
    },
    {
      "epoch": 3.449498762267596,
      "grad_norm": 0.06404593913347874,
      "learning_rate": 4.70196750838197e-05,
      "loss": 0.8428,
      "step": 196480
    },
    {
      "epoch": 3.4496743271475974,
      "grad_norm": 0.049340386641492706,
      "learning_rate": 4.7010334056119395e-05,
      "loss": 0.8407,
      "step": 196490
    },
    {
      "epoch": 3.449849892027599,
      "grad_norm": 0.04806272534527263,
      "learning_rate": 4.700099371247845e-05,
      "loss": 0.8418,
      "step": 196500
    },
    {
      "epoch": 3.4500254569076003,
      "grad_norm": 0.05577182602799346,
      "learning_rate": 4.6991654053015264e-05,
      "loss": 0.8412,
      "step": 196510
    },
    {
      "epoch": 3.450201021787602,
      "grad_norm": 0.04550103383117711,
      "learning_rate": 4.698231507784817e-05,
      "loss": 0.8461,
      "step": 196520
    },
    {
      "epoch": 3.450376586667603,
      "grad_norm": 0.05628928862894513,
      "learning_rate": 4.6972976787095534e-05,
      "loss": 0.8419,
      "step": 196530
    },
    {
      "epoch": 3.4505521515476043,
      "grad_norm": 0.059417879332688696,
      "learning_rate": 4.6963639180875604e-05,
      "loss": 0.8455,
      "step": 196540
    },
    {
      "epoch": 3.450727716427606,
      "grad_norm": 0.04071837293223006,
      "learning_rate": 4.695430225930672e-05,
      "loss": 0.8475,
      "step": 196550
    },
    {
      "epoch": 3.4509032813076073,
      "grad_norm": 0.04563148913023968,
      "learning_rate": 4.694496602250722e-05,
      "loss": 0.843,
      "step": 196560
    },
    {
      "epoch": 3.4510788461876087,
      "grad_norm": 0.047642276694462124,
      "learning_rate": 4.693563047059533e-05,
      "loss": 0.8458,
      "step": 196570
    },
    {
      "epoch": 3.45125441106761,
      "grad_norm": 0.04807230949279478,
      "learning_rate": 4.692629560368936e-05,
      "loss": 0.8479,
      "step": 196580
    },
    {
      "epoch": 3.4514299759476113,
      "grad_norm": 0.05098251638104363,
      "learning_rate": 4.6916961421907584e-05,
      "loss": 0.8406,
      "step": 196590
    },
    {
      "epoch": 3.4516055408276127,
      "grad_norm": 0.04569414677247515,
      "learning_rate": 4.690762792536831e-05,
      "loss": 0.8404,
      "step": 196600
    },
    {
      "epoch": 3.451781105707614,
      "grad_norm": 0.04509444637648617,
      "learning_rate": 4.6898295114189756e-05,
      "loss": 0.844,
      "step": 196610
    },
    {
      "epoch": 3.4519566705876157,
      "grad_norm": 0.04706107458844503,
      "learning_rate": 4.68889629884901e-05,
      "loss": 0.8489,
      "step": 196620
    },
    {
      "epoch": 3.452132235467617,
      "grad_norm": 0.04755267533130521,
      "learning_rate": 4.687963154838772e-05,
      "loss": 0.8312,
      "step": 196630
    },
    {
      "epoch": 3.4523078003476186,
      "grad_norm": 0.051193961631097094,
      "learning_rate": 4.687030079400075e-05,
      "loss": 0.8426,
      "step": 196640
    },
    {
      "epoch": 3.4524833652276197,
      "grad_norm": 0.058307866021171155,
      "learning_rate": 4.686097072544745e-05,
      "loss": 0.8471,
      "step": 196650
    },
    {
      "epoch": 3.452658930107621,
      "grad_norm": 0.052165322170707705,
      "learning_rate": 4.685164134284606e-05,
      "loss": 0.8439,
      "step": 196660
    },
    {
      "epoch": 3.4528344949876226,
      "grad_norm": 0.04414161931597973,
      "learning_rate": 4.6842312646314704e-05,
      "loss": 0.8312,
      "step": 196670
    },
    {
      "epoch": 3.453010059867624,
      "grad_norm": 0.07136209495478579,
      "learning_rate": 4.683298463597167e-05,
      "loss": 0.8425,
      "step": 196680
    },
    {
      "epoch": 3.4531856247476256,
      "grad_norm": 0.04751907216936714,
      "learning_rate": 4.682365731193502e-05,
      "loss": 0.8556,
      "step": 196690
    },
    {
      "epoch": 3.453361189627627,
      "grad_norm": 0.06164353416216412,
      "learning_rate": 4.68143306743231e-05,
      "loss": 0.842,
      "step": 196700
    },
    {
      "epoch": 3.4535367545076285,
      "grad_norm": 0.06360148100487617,
      "learning_rate": 4.680500472325396e-05,
      "loss": 0.8381,
      "step": 196710
    },
    {
      "epoch": 3.4537123193876296,
      "grad_norm": 0.0562209153493164,
      "learning_rate": 4.679567945884579e-05,
      "loss": 0.8393,
      "step": 196720
    },
    {
      "epoch": 3.453887884267631,
      "grad_norm": 0.07963791417975098,
      "learning_rate": 4.678635488121678e-05,
      "loss": 0.8433,
      "step": 196730
    },
    {
      "epoch": 3.4540634491476325,
      "grad_norm": 0.05287333998268147,
      "learning_rate": 4.677703099048502e-05,
      "loss": 0.844,
      "step": 196740
    },
    {
      "epoch": 3.454239014027634,
      "grad_norm": 0.05098885843722229,
      "learning_rate": 4.676770778676866e-05,
      "loss": 0.8387,
      "step": 196750
    },
    {
      "epoch": 3.4544145789076355,
      "grad_norm": 0.05187533973273375,
      "learning_rate": 4.675838527018584e-05,
      "loss": 0.8301,
      "step": 196760
    },
    {
      "epoch": 3.4545901437876365,
      "grad_norm": 0.06708632338762434,
      "learning_rate": 4.674906344085471e-05,
      "loss": 0.8391,
      "step": 196770
    },
    {
      "epoch": 3.454765708667638,
      "grad_norm": 0.05054505242463005,
      "learning_rate": 4.673974229889334e-05,
      "loss": 0.8462,
      "step": 196780
    },
    {
      "epoch": 3.4549412735476395,
      "grad_norm": 0.060993797276931376,
      "learning_rate": 4.673042184441977e-05,
      "loss": 0.8423,
      "step": 196790
    },
    {
      "epoch": 3.455116838427641,
      "grad_norm": 0.052402590478197494,
      "learning_rate": 4.6721102077552224e-05,
      "loss": 0.8372,
      "step": 196800
    },
    {
      "epoch": 3.4552924033076424,
      "grad_norm": 0.06343189421177822,
      "learning_rate": 4.671178299840867e-05,
      "loss": 0.8365,
      "step": 196810
    },
    {
      "epoch": 3.455467968187644,
      "grad_norm": 0.05313346190997617,
      "learning_rate": 4.6702464607107243e-05,
      "loss": 0.8406,
      "step": 196820
    },
    {
      "epoch": 3.4556435330676454,
      "grad_norm": 0.047177615564312186,
      "learning_rate": 4.669314690376603e-05,
      "loss": 0.8348,
      "step": 196830
    },
    {
      "epoch": 3.4558190979476464,
      "grad_norm": 0.059127313627491164,
      "learning_rate": 4.6683829888503024e-05,
      "loss": 0.8273,
      "step": 196840
    },
    {
      "epoch": 3.455994662827648,
      "grad_norm": 0.061776683105313944,
      "learning_rate": 4.667451356143632e-05,
      "loss": 0.8365,
      "step": 196850
    },
    {
      "epoch": 3.4561702277076494,
      "grad_norm": 0.058858403390666,
      "learning_rate": 4.666519792268388e-05,
      "loss": 0.8407,
      "step": 196860
    },
    {
      "epoch": 3.456345792587651,
      "grad_norm": 0.06190619969684676,
      "learning_rate": 4.6655882972363865e-05,
      "loss": 0.8381,
      "step": 196870
    },
    {
      "epoch": 3.4565213574676523,
      "grad_norm": 0.06320485954781564,
      "learning_rate": 4.664656871059419e-05,
      "loss": 0.8452,
      "step": 196880
    },
    {
      "epoch": 3.4566969223476534,
      "grad_norm": 0.05647091944751015,
      "learning_rate": 4.663725513749292e-05,
      "loss": 0.841,
      "step": 196890
    },
    {
      "epoch": 3.456872487227655,
      "grad_norm": 0.04646022211982402,
      "learning_rate": 4.662794225317806e-05,
      "loss": 0.8397,
      "step": 196900
    },
    {
      "epoch": 3.4570480521076563,
      "grad_norm": 0.05495049667699605,
      "learning_rate": 4.661863005776757e-05,
      "loss": 0.8373,
      "step": 196910
    },
    {
      "epoch": 3.457223616987658,
      "grad_norm": 0.05465411774507581,
      "learning_rate": 4.660931855137947e-05,
      "loss": 0.8439,
      "step": 196920
    },
    {
      "epoch": 3.4573991818676593,
      "grad_norm": 0.05523646084651684,
      "learning_rate": 4.660000773413171e-05,
      "loss": 0.8431,
      "step": 196930
    },
    {
      "epoch": 3.4575747467476607,
      "grad_norm": 0.06143275946400446,
      "learning_rate": 4.659069760614231e-05,
      "loss": 0.835,
      "step": 196940
    },
    {
      "epoch": 3.457750311627662,
      "grad_norm": 0.057708781739356375,
      "learning_rate": 4.6581388167529214e-05,
      "loss": 0.8358,
      "step": 196950
    },
    {
      "epoch": 3.4579258765076633,
      "grad_norm": 0.04949481051803071,
      "learning_rate": 4.657207941841028e-05,
      "loss": 0.8398,
      "step": 196960
    },
    {
      "epoch": 3.4581014413876647,
      "grad_norm": 0.057996845381131236,
      "learning_rate": 4.65627713589036e-05,
      "loss": 0.849,
      "step": 196970
    },
    {
      "epoch": 3.458277006267666,
      "grad_norm": 0.05227205646255294,
      "learning_rate": 4.655346398912701e-05,
      "loss": 0.8436,
      "step": 196980
    },
    {
      "epoch": 3.4584525711476677,
      "grad_norm": 0.05688085203135641,
      "learning_rate": 4.6544157309198464e-05,
      "loss": 0.8384,
      "step": 196990
    },
    {
      "epoch": 3.458628136027669,
      "grad_norm": 0.05409196746728976,
      "learning_rate": 4.6534851319235916e-05,
      "loss": 0.8448,
      "step": 197000
    },
    {
      "epoch": 3.45880370090767,
      "grad_norm": 0.05426963030771665,
      "learning_rate": 4.652554601935721e-05,
      "loss": 0.8393,
      "step": 197010
    },
    {
      "epoch": 3.4589792657876717,
      "grad_norm": 0.06610642054779073,
      "learning_rate": 4.65162414096803e-05,
      "loss": 0.8404,
      "step": 197020
    },
    {
      "epoch": 3.459154830667673,
      "grad_norm": 0.05744545196362876,
      "learning_rate": 4.650693749032298e-05,
      "loss": 0.8411,
      "step": 197030
    },
    {
      "epoch": 3.4593303955476746,
      "grad_norm": 0.0910861156489301,
      "learning_rate": 4.649763426140329e-05,
      "loss": 0.8418,
      "step": 197040
    },
    {
      "epoch": 3.459505960427676,
      "grad_norm": 0.04803828812356007,
      "learning_rate": 4.648833172303898e-05,
      "loss": 0.839,
      "step": 197050
    },
    {
      "epoch": 3.4596815253076776,
      "grad_norm": 0.05845946909320537,
      "learning_rate": 4.647902987534795e-05,
      "loss": 0.8399,
      "step": 197060
    },
    {
      "epoch": 3.459857090187679,
      "grad_norm": 0.04847139983800448,
      "learning_rate": 4.64697287184481e-05,
      "loss": 0.8442,
      "step": 197070
    },
    {
      "epoch": 3.46003265506768,
      "grad_norm": 0.05048975962148135,
      "learning_rate": 4.6460428252457204e-05,
      "loss": 0.8485,
      "step": 197080
    },
    {
      "epoch": 3.4602082199476816,
      "grad_norm": 0.06022866034701563,
      "learning_rate": 4.645112847749313e-05,
      "loss": 0.8393,
      "step": 197090
    },
    {
      "epoch": 3.460383784827683,
      "grad_norm": 0.06544654784512123,
      "learning_rate": 4.64418293936737e-05,
      "loss": 0.8354,
      "step": 197100
    },
    {
      "epoch": 3.4605593497076845,
      "grad_norm": 0.054157266231730034,
      "learning_rate": 4.64325310011168e-05,
      "loss": 0.8443,
      "step": 197110
    },
    {
      "epoch": 3.460734914587686,
      "grad_norm": 0.056394563680862125,
      "learning_rate": 4.64232332999402e-05,
      "loss": 0.8503,
      "step": 197120
    },
    {
      "epoch": 3.460910479467687,
      "grad_norm": 0.04808135158714719,
      "learning_rate": 4.64139362902616e-05,
      "loss": 0.843,
      "step": 197130
    },
    {
      "epoch": 3.4610860443476885,
      "grad_norm": 0.057282811261637014,
      "learning_rate": 4.640463997219897e-05,
      "loss": 0.8397,
      "step": 197140
    },
    {
      "epoch": 3.46126160922769,
      "grad_norm": 0.05008221084898831,
      "learning_rate": 4.639534434586998e-05,
      "loss": 0.8423,
      "step": 197150
    },
    {
      "epoch": 3.4614371741076915,
      "grad_norm": 0.05739576897050903,
      "learning_rate": 4.638604941139244e-05,
      "loss": 0.836,
      "step": 197160
    },
    {
      "epoch": 3.461612738987693,
      "grad_norm": 0.0620256739247974,
      "learning_rate": 4.6376755168884156e-05,
      "loss": 0.8414,
      "step": 197170
    },
    {
      "epoch": 3.4617883038676944,
      "grad_norm": 0.05766333694570417,
      "learning_rate": 4.636746161846282e-05,
      "loss": 0.8429,
      "step": 197180
    },
    {
      "epoch": 3.461963868747696,
      "grad_norm": 0.05099203897967336,
      "learning_rate": 4.635816876024625e-05,
      "loss": 0.8365,
      "step": 197190
    },
    {
      "epoch": 3.462139433627697,
      "grad_norm": 0.07241348176844167,
      "learning_rate": 4.6348876594352076e-05,
      "loss": 0.8492,
      "step": 197200
    },
    {
      "epoch": 3.4623149985076984,
      "grad_norm": 0.052367901004620326,
      "learning_rate": 4.633958512089818e-05,
      "loss": 0.8369,
      "step": 197210
    },
    {
      "epoch": 3.4624905633877,
      "grad_norm": 0.07042694598976819,
      "learning_rate": 4.6330294340002186e-05,
      "loss": 0.8432,
      "step": 197220
    },
    {
      "epoch": 3.4626661282677014,
      "grad_norm": 0.04644117401236317,
      "learning_rate": 4.632100425178184e-05,
      "loss": 0.8446,
      "step": 197230
    },
    {
      "epoch": 3.462841693147703,
      "grad_norm": 0.053009959765800946,
      "learning_rate": 4.6311714856354895e-05,
      "loss": 0.8351,
      "step": 197240
    },
    {
      "epoch": 3.4630172580277043,
      "grad_norm": 0.046830333267445456,
      "learning_rate": 4.6302426153838954e-05,
      "loss": 0.8366,
      "step": 197250
    },
    {
      "epoch": 3.4631928229077054,
      "grad_norm": 0.05978010935199636,
      "learning_rate": 4.629313814435176e-05,
      "loss": 0.8442,
      "step": 197260
    },
    {
      "epoch": 3.463368387787707,
      "grad_norm": 0.044582845246675946,
      "learning_rate": 4.6283850828010995e-05,
      "loss": 0.8449,
      "step": 197270
    },
    {
      "epoch": 3.4635439526677083,
      "grad_norm": 0.05589940097380329,
      "learning_rate": 4.6274564204934364e-05,
      "loss": 0.8296,
      "step": 197280
    },
    {
      "epoch": 3.46371951754771,
      "grad_norm": 0.044109919551700606,
      "learning_rate": 4.626527827523948e-05,
      "loss": 0.8359,
      "step": 197290
    },
    {
      "epoch": 3.4638950824277113,
      "grad_norm": 0.06318523543519008,
      "learning_rate": 4.625599303904394e-05,
      "loss": 0.8421,
      "step": 197300
    },
    {
      "epoch": 3.4640706473077127,
      "grad_norm": 0.04542707681151442,
      "learning_rate": 4.6246708496465534e-05,
      "loss": 0.8335,
      "step": 197310
    },
    {
      "epoch": 3.464246212187714,
      "grad_norm": 0.04768139351693255,
      "learning_rate": 4.62374246476218e-05,
      "loss": 0.844,
      "step": 197320
    },
    {
      "epoch": 3.4644217770677153,
      "grad_norm": 0.04868842882899563,
      "learning_rate": 4.6228141492630375e-05,
      "loss": 0.8378,
      "step": 197330
    },
    {
      "epoch": 3.4645973419477167,
      "grad_norm": 0.06622370637829345,
      "learning_rate": 4.621885903160894e-05,
      "loss": 0.849,
      "step": 197340
    },
    {
      "epoch": 3.464772906827718,
      "grad_norm": 0.06712889493531762,
      "learning_rate": 4.620957726467503e-05,
      "loss": 0.8425,
      "step": 197350
    },
    {
      "epoch": 3.4649484717077197,
      "grad_norm": 0.053046807417861906,
      "learning_rate": 4.620029619194629e-05,
      "loss": 0.8395,
      "step": 197360
    },
    {
      "epoch": 3.465124036587721,
      "grad_norm": 0.051734691196682926,
      "learning_rate": 4.619101581354025e-05,
      "loss": 0.8396,
      "step": 197370
    },
    {
      "epoch": 3.4652996014677226,
      "grad_norm": 0.053291255262996454,
      "learning_rate": 4.6181736129574605e-05,
      "loss": 0.8392,
      "step": 197380
    },
    {
      "epoch": 3.4654751663477237,
      "grad_norm": 0.05661649391250457,
      "learning_rate": 4.6172457140166844e-05,
      "loss": 0.848,
      "step": 197390
    },
    {
      "epoch": 3.465650731227725,
      "grad_norm": 0.04249474258497737,
      "learning_rate": 4.616317884543455e-05,
      "loss": 0.841,
      "step": 197400
    },
    {
      "epoch": 3.4658262961077266,
      "grad_norm": 0.0637778020668044,
      "learning_rate": 4.615390124549532e-05,
      "loss": 0.8375,
      "step": 197410
    },
    {
      "epoch": 3.466001860987728,
      "grad_norm": 0.06317116125085986,
      "learning_rate": 4.614462434046664e-05,
      "loss": 0.8446,
      "step": 197420
    },
    {
      "epoch": 3.4661774258677296,
      "grad_norm": 0.05465412249694534,
      "learning_rate": 4.613534813046608e-05,
      "loss": 0.8427,
      "step": 197430
    },
    {
      "epoch": 3.4663529907477306,
      "grad_norm": 0.04843960621338829,
      "learning_rate": 4.612607261561119e-05,
      "loss": 0.8444,
      "step": 197440
    },
    {
      "epoch": 3.466528555627732,
      "grad_norm": 0.06674162651487747,
      "learning_rate": 4.61167977960195e-05,
      "loss": 0.8471,
      "step": 197450
    },
    {
      "epoch": 3.4667041205077336,
      "grad_norm": 0.09256806139678862,
      "learning_rate": 4.61075236718085e-05,
      "loss": 0.8483,
      "step": 197460
    },
    {
      "epoch": 3.466879685387735,
      "grad_norm": 0.046231863196923975,
      "learning_rate": 4.609825024309562e-05,
      "loss": 0.8391,
      "step": 197470
    },
    {
      "epoch": 3.4670552502677365,
      "grad_norm": 0.047312815844425114,
      "learning_rate": 4.608897750999851e-05,
      "loss": 0.8396,
      "step": 197480
    },
    {
      "epoch": 3.467230815147738,
      "grad_norm": 0.056707586244023715,
      "learning_rate": 4.607970547263454e-05,
      "loss": 0.843,
      "step": 197490
    },
    {
      "epoch": 3.4674063800277395,
      "grad_norm": 0.0647628827746781,
      "learning_rate": 4.6070434131121234e-05,
      "loss": 0.8403,
      "step": 197500
    },
    {
      "epoch": 3.4675819449077405,
      "grad_norm": 0.06791857665917216,
      "learning_rate": 4.6061163485576075e-05,
      "loss": 0.848,
      "step": 197510
    },
    {
      "epoch": 3.467757509787742,
      "grad_norm": 0.054908550853236766,
      "learning_rate": 4.605189353611647e-05,
      "loss": 0.8436,
      "step": 197520
    },
    {
      "epoch": 3.4679330746677435,
      "grad_norm": 0.05132670737644294,
      "learning_rate": 4.6042624282859955e-05,
      "loss": 0.8364,
      "step": 197530
    },
    {
      "epoch": 3.468108639547745,
      "grad_norm": 0.05193530088008754,
      "learning_rate": 4.603335572592385e-05,
      "loss": 0.8397,
      "step": 197540
    },
    {
      "epoch": 3.4682842044277464,
      "grad_norm": 0.06643689815643512,
      "learning_rate": 4.602408786542572e-05,
      "loss": 0.8363,
      "step": 197550
    },
    {
      "epoch": 3.4684597693077475,
      "grad_norm": 0.05597872126266016,
      "learning_rate": 4.6014820701482905e-05,
      "loss": 0.8405,
      "step": 197560
    },
    {
      "epoch": 3.468635334187749,
      "grad_norm": 0.0541522457185559,
      "learning_rate": 4.600555423421285e-05,
      "loss": 0.8449,
      "step": 197570
    },
    {
      "epoch": 3.4688108990677504,
      "grad_norm": 0.05694057206831768,
      "learning_rate": 4.5996288463732994e-05,
      "loss": 0.8391,
      "step": 197580
    },
    {
      "epoch": 3.468986463947752,
      "grad_norm": 0.05093710110360561,
      "learning_rate": 4.598702339016067e-05,
      "loss": 0.8459,
      "step": 197590
    },
    {
      "epoch": 3.4691620288277534,
      "grad_norm": 0.05560472908704228,
      "learning_rate": 4.597775901361331e-05,
      "loss": 0.8406,
      "step": 197600
    },
    {
      "epoch": 3.469337593707755,
      "grad_norm": 0.053192750548827314,
      "learning_rate": 4.5968495334208286e-05,
      "loss": 0.8417,
      "step": 197610
    },
    {
      "epoch": 3.4695131585877563,
      "grad_norm": 0.05339200188265041,
      "learning_rate": 4.5959232352063e-05,
      "loss": 0.8428,
      "step": 197620
    },
    {
      "epoch": 3.4696887234677574,
      "grad_norm": 0.05676998137177405,
      "learning_rate": 4.5949970067294806e-05,
      "loss": 0.8367,
      "step": 197630
    },
    {
      "epoch": 3.469864288347759,
      "grad_norm": 0.05167349018869613,
      "learning_rate": 4.594070848002095e-05,
      "loss": 0.8506,
      "step": 197640
    },
    {
      "epoch": 3.4700398532277603,
      "grad_norm": 0.05823105148628003,
      "learning_rate": 4.593144759035895e-05,
      "loss": 0.8404,
      "step": 197650
    },
    {
      "epoch": 3.470215418107762,
      "grad_norm": 0.05010298643716784,
      "learning_rate": 4.592218739842603e-05,
      "loss": 0.8483,
      "step": 197660
    },
    {
      "epoch": 3.4703909829877633,
      "grad_norm": 0.05595713861745298,
      "learning_rate": 4.591292790433955e-05,
      "loss": 0.8403,
      "step": 197670
    },
    {
      "epoch": 3.4705665478677643,
      "grad_norm": 0.04524354322008805,
      "learning_rate": 4.5903669108216856e-05,
      "loss": 0.8431,
      "step": 197680
    },
    {
      "epoch": 3.470742112747766,
      "grad_norm": 0.06939394165191178,
      "learning_rate": 4.5894411010175206e-05,
      "loss": 0.8436,
      "step": 197690
    },
    {
      "epoch": 3.4709176776277673,
      "grad_norm": 0.05749852439403477,
      "learning_rate": 4.588515361033198e-05,
      "loss": 0.842,
      "step": 197700
    },
    {
      "epoch": 3.4710932425077687,
      "grad_norm": 0.04969936273986872,
      "learning_rate": 4.5875896908804324e-05,
      "loss": 0.844,
      "step": 197710
    },
    {
      "epoch": 3.47126880738777,
      "grad_norm": 0.06376297657652605,
      "learning_rate": 4.58666409057097e-05,
      "loss": 0.8373,
      "step": 197720
    },
    {
      "epoch": 3.4714443722677717,
      "grad_norm": 0.06360335197647755,
      "learning_rate": 4.5857385601165274e-05,
      "loss": 0.8434,
      "step": 197730
    },
    {
      "epoch": 3.471619937147773,
      "grad_norm": 0.05169260070862869,
      "learning_rate": 4.584813099528833e-05,
      "loss": 0.8346,
      "step": 197740
    },
    {
      "epoch": 3.471795502027774,
      "grad_norm": 0.04624630326809619,
      "learning_rate": 4.583887708819617e-05,
      "loss": 0.8375,
      "step": 197750
    },
    {
      "epoch": 3.4719710669077757,
      "grad_norm": 0.07501540512073937,
      "learning_rate": 4.5829623880005984e-05,
      "loss": 0.8406,
      "step": 197760
    },
    {
      "epoch": 3.472146631787777,
      "grad_norm": 0.05284102030160163,
      "learning_rate": 4.582037137083503e-05,
      "loss": 0.8352,
      "step": 197770
    },
    {
      "epoch": 3.4723221966677786,
      "grad_norm": 0.06371363796889074,
      "learning_rate": 4.5811119560800556e-05,
      "loss": 0.835,
      "step": 197780
    },
    {
      "epoch": 3.47249776154778,
      "grad_norm": 0.056658496837730964,
      "learning_rate": 4.5801868450019794e-05,
      "loss": 0.8445,
      "step": 197790
    },
    {
      "epoch": 3.472673326427781,
      "grad_norm": 0.04712676571981139,
      "learning_rate": 4.579261803860994e-05,
      "loss": 0.8421,
      "step": 197800
    },
    {
      "epoch": 3.4728488913077826,
      "grad_norm": 0.05052938804595096,
      "learning_rate": 4.578336832668813e-05,
      "loss": 0.8416,
      "step": 197810
    },
    {
      "epoch": 3.473024456187784,
      "grad_norm": 0.053770858038039875,
      "learning_rate": 4.57741193143717e-05,
      "loss": 0.8449,
      "step": 197820
    },
    {
      "epoch": 3.4732000210677856,
      "grad_norm": 0.03850132394839228,
      "learning_rate": 4.5764871001777726e-05,
      "loss": 0.8463,
      "step": 197830
    },
    {
      "epoch": 3.473375585947787,
      "grad_norm": 0.04419053080813262,
      "learning_rate": 4.575562338902341e-05,
      "loss": 0.8367,
      "step": 197840
    },
    {
      "epoch": 3.4735511508277885,
      "grad_norm": 0.05496048699616588,
      "learning_rate": 4.5746376476225986e-05,
      "loss": 0.8468,
      "step": 197850
    },
    {
      "epoch": 3.47372671570779,
      "grad_norm": 0.05358923549353526,
      "learning_rate": 4.573713026350251e-05,
      "loss": 0.8443,
      "step": 197860
    },
    {
      "epoch": 3.473902280587791,
      "grad_norm": 0.05429402525047923,
      "learning_rate": 4.572788475097023e-05,
      "loss": 0.8512,
      "step": 197870
    },
    {
      "epoch": 3.4740778454677925,
      "grad_norm": 0.05495959959688894,
      "learning_rate": 4.5718639938746164e-05,
      "loss": 0.8437,
      "step": 197880
    },
    {
      "epoch": 3.474253410347794,
      "grad_norm": 0.06537879045141148,
      "learning_rate": 4.57093958269476e-05,
      "loss": 0.8333,
      "step": 197890
    },
    {
      "epoch": 3.4744289752277955,
      "grad_norm": 0.04046299859766793,
      "learning_rate": 4.570015241569156e-05,
      "loss": 0.8441,
      "step": 197900
    },
    {
      "epoch": 3.474604540107797,
      "grad_norm": 0.05690323390153259,
      "learning_rate": 4.569090970509518e-05,
      "loss": 0.8502,
      "step": 197910
    },
    {
      "epoch": 3.4747801049877984,
      "grad_norm": 0.04646478278749478,
      "learning_rate": 4.568166769527561e-05,
      "loss": 0.8366,
      "step": 197920
    },
    {
      "epoch": 3.4749556698677995,
      "grad_norm": 0.05555858089471255,
      "learning_rate": 4.567242638634987e-05,
      "loss": 0.8327,
      "step": 197930
    },
    {
      "epoch": 3.475131234747801,
      "grad_norm": 0.06369782256578778,
      "learning_rate": 4.56631857784351e-05,
      "loss": 0.8441,
      "step": 197940
    },
    {
      "epoch": 3.4753067996278024,
      "grad_norm": 0.051428258453009565,
      "learning_rate": 4.5653945871648365e-05,
      "loss": 0.8471,
      "step": 197950
    },
    {
      "epoch": 3.475482364507804,
      "grad_norm": 0.05485406377154333,
      "learning_rate": 4.564470666610679e-05,
      "loss": 0.8419,
      "step": 197960
    },
    {
      "epoch": 3.4756579293878054,
      "grad_norm": 0.06378794240883492,
      "learning_rate": 4.5635468161927376e-05,
      "loss": 0.8466,
      "step": 197970
    },
    {
      "epoch": 3.475833494267807,
      "grad_norm": 0.05228832440621661,
      "learning_rate": 4.562623035922713e-05,
      "loss": 0.8298,
      "step": 197980
    },
    {
      "epoch": 3.476009059147808,
      "grad_norm": 0.0510199777586386,
      "learning_rate": 4.561699325812322e-05,
      "loss": 0.8384,
      "step": 197990
    },
    {
      "epoch": 3.4761846240278094,
      "grad_norm": 0.05665999697794374,
      "learning_rate": 4.560775685873259e-05,
      "loss": 0.8341,
      "step": 198000
    },
    {
      "epoch": 3.476360188907811,
      "grad_norm": 0.046489134429219,
      "learning_rate": 4.559852116117229e-05,
      "loss": 0.8437,
      "step": 198010
    },
    {
      "epoch": 3.4765357537878123,
      "grad_norm": 0.06789215803428392,
      "learning_rate": 4.558928616555939e-05,
      "loss": 0.8429,
      "step": 198020
    },
    {
      "epoch": 3.476711318667814,
      "grad_norm": 0.057693128339397395,
      "learning_rate": 4.558005187201081e-05,
      "loss": 0.841,
      "step": 198030
    },
    {
      "epoch": 3.4768868835478153,
      "grad_norm": 0.05142879750724833,
      "learning_rate": 4.557081828064364e-05,
      "loss": 0.8457,
      "step": 198040
    },
    {
      "epoch": 3.4770624484278163,
      "grad_norm": 0.06292747953786532,
      "learning_rate": 4.5561585391574746e-05,
      "loss": 0.8415,
      "step": 198050
    },
    {
      "epoch": 3.477238013307818,
      "grad_norm": 0.052603032588425004,
      "learning_rate": 4.5552353204921264e-05,
      "loss": 0.8338,
      "step": 198060
    },
    {
      "epoch": 3.4774135781878193,
      "grad_norm": 0.048083354284683154,
      "learning_rate": 4.554312172080005e-05,
      "loss": 0.8353,
      "step": 198070
    },
    {
      "epoch": 3.4775891430678207,
      "grad_norm": 0.05031273381204942,
      "learning_rate": 4.5533890939328124e-05,
      "loss": 0.8453,
      "step": 198080
    },
    {
      "epoch": 3.477764707947822,
      "grad_norm": 0.0988647347466929,
      "learning_rate": 4.552466086062246e-05,
      "loss": 0.8429,
      "step": 198090
    },
    {
      "epoch": 3.4779402728278237,
      "grad_norm": 0.039786681849430384,
      "learning_rate": 4.551543148479993e-05,
      "loss": 0.832,
      "step": 198100
    },
    {
      "epoch": 3.4781158377078247,
      "grad_norm": 0.04761208636107407,
      "learning_rate": 4.5506202811977534e-05,
      "loss": 0.8403,
      "step": 198110
    },
    {
      "epoch": 3.478291402587826,
      "grad_norm": 0.07020701011700105,
      "learning_rate": 4.5496974842272165e-05,
      "loss": 0.8362,
      "step": 198120
    },
    {
      "epoch": 3.4784669674678277,
      "grad_norm": 0.04463819613591754,
      "learning_rate": 4.548774757580082e-05,
      "loss": 0.8425,
      "step": 198130
    },
    {
      "epoch": 3.478642532347829,
      "grad_norm": 0.044956640625443155,
      "learning_rate": 4.5478521012680326e-05,
      "loss": 0.8388,
      "step": 198140
    },
    {
      "epoch": 3.4788180972278306,
      "grad_norm": 0.06015014279583066,
      "learning_rate": 4.5469295153027556e-05,
      "loss": 0.8412,
      "step": 198150
    },
    {
      "epoch": 3.478993662107832,
      "grad_norm": 0.05818242181410549,
      "learning_rate": 4.5460069996959524e-05,
      "loss": 0.8416,
      "step": 198160
    },
    {
      "epoch": 3.4791692269878336,
      "grad_norm": 0.05582248459205252,
      "learning_rate": 4.545084554459303e-05,
      "loss": 0.8469,
      "step": 198170
    },
    {
      "epoch": 3.4793447918678346,
      "grad_norm": 0.05496675289721324,
      "learning_rate": 4.5441621796044945e-05,
      "loss": 0.8404,
      "step": 198180
    },
    {
      "epoch": 3.479520356747836,
      "grad_norm": 0.052960170860099874,
      "learning_rate": 4.543239875143221e-05,
      "loss": 0.8427,
      "step": 198190
    },
    {
      "epoch": 3.4796959216278376,
      "grad_norm": 0.058810214184192106,
      "learning_rate": 4.5423176410871597e-05,
      "loss": 0.8406,
      "step": 198200
    },
    {
      "epoch": 3.479871486507839,
      "grad_norm": 0.05728741882389626,
      "learning_rate": 4.5413954774480025e-05,
      "loss": 0.8351,
      "step": 198210
    },
    {
      "epoch": 3.4800470513878405,
      "grad_norm": 0.0525635419247595,
      "learning_rate": 4.540473384237422e-05,
      "loss": 0.8418,
      "step": 198220
    },
    {
      "epoch": 3.4802226162678416,
      "grad_norm": 0.05760878410048432,
      "learning_rate": 4.539551361467117e-05,
      "loss": 0.8406,
      "step": 198230
    },
    {
      "epoch": 3.480398181147843,
      "grad_norm": 0.048023413143515555,
      "learning_rate": 4.538629409148758e-05,
      "loss": 0.8398,
      "step": 198240
    },
    {
      "epoch": 3.4805737460278445,
      "grad_norm": 0.05753044882354377,
      "learning_rate": 4.53770752729403e-05,
      "loss": 0.8355,
      "step": 198250
    },
    {
      "epoch": 3.480749310907846,
      "grad_norm": 0.06601160977307298,
      "learning_rate": 4.536785715914618e-05,
      "loss": 0.842,
      "step": 198260
    },
    {
      "epoch": 3.4809248757878475,
      "grad_norm": 0.06009458879366598,
      "learning_rate": 4.535863975022193e-05,
      "loss": 0.8403,
      "step": 198270
    },
    {
      "epoch": 3.481100440667849,
      "grad_norm": 0.06387275176650598,
      "learning_rate": 4.534942304628438e-05,
      "loss": 0.8397,
      "step": 198280
    },
    {
      "epoch": 3.4812760055478504,
      "grad_norm": 0.061317741468715145,
      "learning_rate": 4.53402070474503e-05,
      "loss": 0.8421,
      "step": 198290
    },
    {
      "epoch": 3.4814515704278515,
      "grad_norm": 0.057040033160666775,
      "learning_rate": 4.5330991753836506e-05,
      "loss": 0.8407,
      "step": 198300
    },
    {
      "epoch": 3.481627135307853,
      "grad_norm": 0.05906722123275731,
      "learning_rate": 4.532177716555971e-05,
      "loss": 0.8375,
      "step": 198310
    },
    {
      "epoch": 3.4818027001878544,
      "grad_norm": 0.050616852858088736,
      "learning_rate": 4.5312563282736605e-05,
      "loss": 0.8397,
      "step": 198320
    },
    {
      "epoch": 3.481978265067856,
      "grad_norm": 0.07150808518568874,
      "learning_rate": 4.530335010548406e-05,
      "loss": 0.8381,
      "step": 198330
    },
    {
      "epoch": 3.4821538299478574,
      "grad_norm": 0.0661394930157319,
      "learning_rate": 4.529413763391871e-05,
      "loss": 0.843,
      "step": 198340
    },
    {
      "epoch": 3.4823293948278584,
      "grad_norm": 0.045797930865483526,
      "learning_rate": 4.5284925868157304e-05,
      "loss": 0.8405,
      "step": 198350
    },
    {
      "epoch": 3.48250495970786,
      "grad_norm": 0.04653645779477515,
      "learning_rate": 4.527571480831662e-05,
      "loss": 0.8381,
      "step": 198360
    },
    {
      "epoch": 3.4826805245878614,
      "grad_norm": 0.06062186794096757,
      "learning_rate": 4.526650445451326e-05,
      "loss": 0.8423,
      "step": 198370
    },
    {
      "epoch": 3.482856089467863,
      "grad_norm": 0.04716049140760879,
      "learning_rate": 4.525729480686401e-05,
      "loss": 0.8425,
      "step": 198380
    },
    {
      "epoch": 3.4830316543478643,
      "grad_norm": 0.05221453688691067,
      "learning_rate": 4.524808586548544e-05,
      "loss": 0.8397,
      "step": 198390
    },
    {
      "epoch": 3.483207219227866,
      "grad_norm": 0.05200711498337973,
      "learning_rate": 4.5238877630494384e-05,
      "loss": 0.8415,
      "step": 198400
    },
    {
      "epoch": 3.4833827841078673,
      "grad_norm": 0.0630319936231948,
      "learning_rate": 4.52296701020074e-05,
      "loss": 0.84,
      "step": 198410
    },
    {
      "epoch": 3.4835583489878683,
      "grad_norm": 0.04606213611839157,
      "learning_rate": 4.5220463280141176e-05,
      "loss": 0.8473,
      "step": 198420
    },
    {
      "epoch": 3.48373391386787,
      "grad_norm": 0.0499733138955965,
      "learning_rate": 4.5211257165012414e-05,
      "loss": 0.8384,
      "step": 198430
    },
    {
      "epoch": 3.4839094787478713,
      "grad_norm": 0.05712738979101536,
      "learning_rate": 4.520205175673768e-05,
      "loss": 0.8369,
      "step": 198440
    },
    {
      "epoch": 3.4840850436278727,
      "grad_norm": 0.039817736035259725,
      "learning_rate": 4.5192847055433645e-05,
      "loss": 0.8368,
      "step": 198450
    },
    {
      "epoch": 3.4842606085078742,
      "grad_norm": 0.044885185175087154,
      "learning_rate": 4.518364306121693e-05,
      "loss": 0.8317,
      "step": 198460
    },
    {
      "epoch": 3.4844361733878753,
      "grad_norm": 0.05395307438713094,
      "learning_rate": 4.517443977420419e-05,
      "loss": 0.8393,
      "step": 198470
    },
    {
      "epoch": 3.4846117382678767,
      "grad_norm": 0.048931637996607644,
      "learning_rate": 4.5165237194512e-05,
      "loss": 0.8381,
      "step": 198480
    },
    {
      "epoch": 3.484787303147878,
      "grad_norm": 0.0668690523735504,
      "learning_rate": 4.515603532225689e-05,
      "loss": 0.8373,
      "step": 198490
    },
    {
      "epoch": 3.4849628680278797,
      "grad_norm": 0.05012058684816146,
      "learning_rate": 4.5146834157555574e-05,
      "loss": 0.8431,
      "step": 198500
    },
    {
      "epoch": 3.485138432907881,
      "grad_norm": 0.06762029265482566,
      "learning_rate": 4.513763370052455e-05,
      "loss": 0.838,
      "step": 198510
    },
    {
      "epoch": 3.4853139977878826,
      "grad_norm": 0.11769561642153123,
      "learning_rate": 4.5128433951280406e-05,
      "loss": 0.8353,
      "step": 198520
    },
    {
      "epoch": 3.485489562667884,
      "grad_norm": 0.04938548310529093,
      "learning_rate": 4.5119234909939746e-05,
      "loss": 0.838,
      "step": 198530
    },
    {
      "epoch": 3.485665127547885,
      "grad_norm": 0.05677441683640447,
      "learning_rate": 4.511003657661905e-05,
      "loss": 0.8504,
      "step": 198540
    },
    {
      "epoch": 3.4858406924278866,
      "grad_norm": 0.056615632182696315,
      "learning_rate": 4.5100838951434946e-05,
      "loss": 0.849,
      "step": 198550
    },
    {
      "epoch": 3.486016257307888,
      "grad_norm": 0.06928029400187821,
      "learning_rate": 4.509164203450384e-05,
      "loss": 0.8466,
      "step": 198560
    },
    {
      "epoch": 3.4861918221878896,
      "grad_norm": 0.042975568830826615,
      "learning_rate": 4.508244582594244e-05,
      "loss": 0.8353,
      "step": 198570
    },
    {
      "epoch": 3.486367387067891,
      "grad_norm": 0.05231891673525908,
      "learning_rate": 4.507325032586711e-05,
      "loss": 0.8476,
      "step": 198580
    },
    {
      "epoch": 3.486542951947892,
      "grad_norm": 0.05045689137549638,
      "learning_rate": 4.5064055534394426e-05,
      "loss": 0.8463,
      "step": 198590
    },
    {
      "epoch": 3.4867185168278936,
      "grad_norm": 0.06254107631820234,
      "learning_rate": 4.5054861451640916e-05,
      "loss": 0.841,
      "step": 198600
    },
    {
      "epoch": 3.486894081707895,
      "grad_norm": 0.043401486353036506,
      "learning_rate": 4.5045668077723e-05,
      "loss": 0.8444,
      "step": 198610
    },
    {
      "epoch": 3.4870696465878965,
      "grad_norm": 0.06363841627090265,
      "learning_rate": 4.5036475412757176e-05,
      "loss": 0.8357,
      "step": 198620
    },
    {
      "epoch": 3.487245211467898,
      "grad_norm": 0.047828848636035426,
      "learning_rate": 4.5027283456859955e-05,
      "loss": 0.8309,
      "step": 198630
    },
    {
      "epoch": 3.4874207763478995,
      "grad_norm": 0.04844071419866838,
      "learning_rate": 4.50180922101478e-05,
      "loss": 0.8455,
      "step": 198640
    },
    {
      "epoch": 3.487596341227901,
      "grad_norm": 0.05245212812302599,
      "learning_rate": 4.500890167273716e-05,
      "loss": 0.8381,
      "step": 198650
    },
    {
      "epoch": 3.487771906107902,
      "grad_norm": 0.0486482181539715,
      "learning_rate": 4.4999711844744385e-05,
      "loss": 0.8386,
      "step": 198660
    },
    {
      "epoch": 3.4879474709879035,
      "grad_norm": 0.042024935537885215,
      "learning_rate": 4.499052272628606e-05,
      "loss": 0.8417,
      "step": 198670
    },
    {
      "epoch": 3.488123035867905,
      "grad_norm": 0.05353142946127681,
      "learning_rate": 4.4981334317478515e-05,
      "loss": 0.8398,
      "step": 198680
    },
    {
      "epoch": 3.4882986007479064,
      "grad_norm": 0.05555968817836667,
      "learning_rate": 4.4972146618438204e-05,
      "loss": 0.8304,
      "step": 198690
    },
    {
      "epoch": 3.488474165627908,
      "grad_norm": 0.05433376761198443,
      "learning_rate": 4.496295962928157e-05,
      "loss": 0.8356,
      "step": 198700
    },
    {
      "epoch": 3.4886497305079094,
      "grad_norm": 0.052924656436065906,
      "learning_rate": 4.4953773350124934e-05,
      "loss": 0.8338,
      "step": 198710
    },
    {
      "epoch": 3.4888252953879104,
      "grad_norm": 0.05129501570556686,
      "learning_rate": 4.494458778108477e-05,
      "loss": 0.8433,
      "step": 198720
    },
    {
      "epoch": 3.489000860267912,
      "grad_norm": 0.047449262875638075,
      "learning_rate": 4.493540292227735e-05,
      "loss": 0.841,
      "step": 198730
    },
    {
      "epoch": 3.4891764251479134,
      "grad_norm": 0.053705376606623306,
      "learning_rate": 4.49262187738192e-05,
      "loss": 0.8378,
      "step": 198740
    },
    {
      "epoch": 3.489351990027915,
      "grad_norm": 0.06322913009185321,
      "learning_rate": 4.491703533582656e-05,
      "loss": 0.8373,
      "step": 198750
    },
    {
      "epoch": 3.4895275549079163,
      "grad_norm": 0.0511181817204554,
      "learning_rate": 4.490785260841583e-05,
      "loss": 0.8376,
      "step": 198760
    },
    {
      "epoch": 3.489703119787918,
      "grad_norm": 0.0573971115410176,
      "learning_rate": 4.48986705917034e-05,
      "loss": 0.8363,
      "step": 198770
    },
    {
      "epoch": 3.489878684667919,
      "grad_norm": 0.059441618900109836,
      "learning_rate": 4.4889489285805526e-05,
      "loss": 0.8426,
      "step": 198780
    },
    {
      "epoch": 3.4900542495479203,
      "grad_norm": 0.07427056436079586,
      "learning_rate": 4.488030869083857e-05,
      "loss": 0.8508,
      "step": 198790
    },
    {
      "epoch": 3.490229814427922,
      "grad_norm": 0.05366486366148835,
      "learning_rate": 4.4871128806918864e-05,
      "loss": 0.8325,
      "step": 198800
    },
    {
      "epoch": 3.4904053793079233,
      "grad_norm": 0.04663137483177132,
      "learning_rate": 4.4861949634162745e-05,
      "loss": 0.8286,
      "step": 198810
    },
    {
      "epoch": 3.4905809441879248,
      "grad_norm": 0.05068866285044372,
      "learning_rate": 4.485277117268648e-05,
      "loss": 0.8404,
      "step": 198820
    },
    {
      "epoch": 3.4907565090679262,
      "grad_norm": 0.05837770933162542,
      "learning_rate": 4.4843593422606306e-05,
      "loss": 0.8396,
      "step": 198830
    },
    {
      "epoch": 3.4909320739479277,
      "grad_norm": 0.06919242000008091,
      "learning_rate": 4.483441638403863e-05,
      "loss": 0.8424,
      "step": 198840
    },
    {
      "epoch": 3.4911076388279287,
      "grad_norm": 0.064968262878111,
      "learning_rate": 4.482524005709963e-05,
      "loss": 0.8429,
      "step": 198850
    },
    {
      "epoch": 3.49128320370793,
      "grad_norm": 0.06932946875053662,
      "learning_rate": 4.48160644419056e-05,
      "loss": 0.8325,
      "step": 198860
    },
    {
      "epoch": 3.4914587685879317,
      "grad_norm": 0.05005724165055143,
      "learning_rate": 4.480688953857283e-05,
      "loss": 0.8466,
      "step": 198870
    },
    {
      "epoch": 3.491634333467933,
      "grad_norm": 0.05324313451893133,
      "learning_rate": 4.479771534721752e-05,
      "loss": 0.8393,
      "step": 198880
    },
    {
      "epoch": 3.4918098983479346,
      "grad_norm": 0.05233403102634368,
      "learning_rate": 4.4788541867955954e-05,
      "loss": 0.832,
      "step": 198890
    },
    {
      "epoch": 3.4919854632279357,
      "grad_norm": 0.05426423450731101,
      "learning_rate": 4.477936910090426e-05,
      "loss": 0.8401,
      "step": 198900
    },
    {
      "epoch": 3.492161028107937,
      "grad_norm": 0.04962497089454261,
      "learning_rate": 4.477019704617881e-05,
      "loss": 0.8387,
      "step": 198910
    },
    {
      "epoch": 3.4923365929879386,
      "grad_norm": 0.045494505507683,
      "learning_rate": 4.476102570389571e-05,
      "loss": 0.8389,
      "step": 198920
    },
    {
      "epoch": 3.49251215786794,
      "grad_norm": 0.04482383696518938,
      "learning_rate": 4.475185507417118e-05,
      "loss": 0.8407,
      "step": 198930
    },
    {
      "epoch": 3.4926877227479416,
      "grad_norm": 0.04224066422025547,
      "learning_rate": 4.474268515712147e-05,
      "loss": 0.8569,
      "step": 198940
    },
    {
      "epoch": 3.492863287627943,
      "grad_norm": 0.05278145255172799,
      "learning_rate": 4.473351595286268e-05,
      "loss": 0.8275,
      "step": 198950
    },
    {
      "epoch": 3.4930388525079445,
      "grad_norm": 0.0380850092571548,
      "learning_rate": 4.4724347461511034e-05,
      "loss": 0.8441,
      "step": 198960
    },
    {
      "epoch": 3.4932144173879456,
      "grad_norm": 0.040320370135536646,
      "learning_rate": 4.471517968318269e-05,
      "loss": 0.8412,
      "step": 198970
    },
    {
      "epoch": 3.493389982267947,
      "grad_norm": 0.04551084404808535,
      "learning_rate": 4.470601261799383e-05,
      "loss": 0.8513,
      "step": 198980
    },
    {
      "epoch": 3.4935655471479485,
      "grad_norm": 0.06932628054608699,
      "learning_rate": 4.46968462660606e-05,
      "loss": 0.836,
      "step": 198990
    },
    {
      "epoch": 3.49374111202795,
      "grad_norm": 0.05629127797198223,
      "learning_rate": 4.468768062749903e-05,
      "loss": 0.8505,
      "step": 199000
    },
    {
      "epoch": 3.4939166769079515,
      "grad_norm": 0.05570253121692377,
      "learning_rate": 4.4678515702425406e-05,
      "loss": 0.8492,
      "step": 199010
    },
    {
      "epoch": 3.4940922417879525,
      "grad_norm": 0.059823004756691854,
      "learning_rate": 4.466935149095576e-05,
      "loss": 0.8307,
      "step": 199020
    },
    {
      "epoch": 3.494267806667954,
      "grad_norm": 0.05281971655574328,
      "learning_rate": 4.466018799320622e-05,
      "loss": 0.8377,
      "step": 199030
    },
    {
      "epoch": 3.4944433715479555,
      "grad_norm": 0.05229852758822509,
      "learning_rate": 4.465102520929294e-05,
      "loss": 0.8443,
      "step": 199040
    },
    {
      "epoch": 3.494618936427957,
      "grad_norm": 0.0599117718489344,
      "learning_rate": 4.464186313933192e-05,
      "loss": 0.8399,
      "step": 199050
    },
    {
      "epoch": 3.4947945013079584,
      "grad_norm": 0.052522749018502686,
      "learning_rate": 4.463270178343934e-05,
      "loss": 0.8385,
      "step": 199060
    },
    {
      "epoch": 3.49497006618796,
      "grad_norm": 0.06588677348215416,
      "learning_rate": 4.462354114173115e-05,
      "loss": 0.8431,
      "step": 199070
    },
    {
      "epoch": 3.4951456310679614,
      "grad_norm": 0.05578952581452616,
      "learning_rate": 4.461438121432359e-05,
      "loss": 0.8424,
      "step": 199080
    },
    {
      "epoch": 3.4953211959479624,
      "grad_norm": 0.051576931404738804,
      "learning_rate": 4.460522200133257e-05,
      "loss": 0.8342,
      "step": 199090
    },
    {
      "epoch": 3.495496760827964,
      "grad_norm": 0.08281251040730915,
      "learning_rate": 4.459606350287421e-05,
      "loss": 0.841,
      "step": 199100
    },
    {
      "epoch": 3.4956723257079654,
      "grad_norm": 0.052716558351412784,
      "learning_rate": 4.458690571906455e-05,
      "loss": 0.8434,
      "step": 199110
    },
    {
      "epoch": 3.495847890587967,
      "grad_norm": 0.04795098140784664,
      "learning_rate": 4.4577748650019585e-05,
      "loss": 0.8362,
      "step": 199120
    },
    {
      "epoch": 3.4960234554679683,
      "grad_norm": 0.055200436208294026,
      "learning_rate": 4.456859229585536e-05,
      "loss": 0.8478,
      "step": 199130
    },
    {
      "epoch": 3.4961990203479694,
      "grad_norm": 0.04888549722548365,
      "learning_rate": 4.4559436656687884e-05,
      "loss": 0.8356,
      "step": 199140
    },
    {
      "epoch": 3.496374585227971,
      "grad_norm": 0.04001841355385094,
      "learning_rate": 4.45502817326332e-05,
      "loss": 0.8417,
      "step": 199150
    },
    {
      "epoch": 3.4965501501079723,
      "grad_norm": 0.050247867683029475,
      "learning_rate": 4.4541127523807255e-05,
      "loss": 0.8486,
      "step": 199160
    },
    {
      "epoch": 3.496725714987974,
      "grad_norm": 0.05928902508423957,
      "learning_rate": 4.4531974030325984e-05,
      "loss": 0.84,
      "step": 199170
    },
    {
      "epoch": 3.4969012798679753,
      "grad_norm": 0.05961060969699715,
      "learning_rate": 4.452282125230549e-05,
      "loss": 0.8385,
      "step": 199180
    },
    {
      "epoch": 3.4970768447479768,
      "grad_norm": 0.046448012021312705,
      "learning_rate": 4.451366918986163e-05,
      "loss": 0.8417,
      "step": 199190
    },
    {
      "epoch": 3.4972524096279782,
      "grad_norm": 0.06011098325486774,
      "learning_rate": 4.4504517843110426e-05,
      "loss": 0.8421,
      "step": 199200
    },
    {
      "epoch": 3.4974279745079793,
      "grad_norm": 0.04977333114132336,
      "learning_rate": 4.449536721216784e-05,
      "loss": 0.8338,
      "step": 199210
    },
    {
      "epoch": 3.4976035393879807,
      "grad_norm": 0.07547598047547639,
      "learning_rate": 4.448621729714973e-05,
      "loss": 0.8454,
      "step": 199220
    },
    {
      "epoch": 3.497779104267982,
      "grad_norm": 0.06577355386345812,
      "learning_rate": 4.447706809817213e-05,
      "loss": 0.8375,
      "step": 199230
    },
    {
      "epoch": 3.4979546691479837,
      "grad_norm": 0.048491444161510154,
      "learning_rate": 4.446791961535083e-05,
      "loss": 0.8382,
      "step": 199240
    },
    {
      "epoch": 3.498130234027985,
      "grad_norm": 0.04559805669142092,
      "learning_rate": 4.445877184880189e-05,
      "loss": 0.8412,
      "step": 199250
    },
    {
      "epoch": 3.498305798907986,
      "grad_norm": 0.050416727148144075,
      "learning_rate": 4.4449624798641116e-05,
      "loss": 0.8399,
      "step": 199260
    },
    {
      "epoch": 3.4984813637879877,
      "grad_norm": 0.04808740143272317,
      "learning_rate": 4.444047846498443e-05,
      "loss": 0.842,
      "step": 199270
    },
    {
      "epoch": 3.498656928667989,
      "grad_norm": 0.04408757638972667,
      "learning_rate": 4.4431332847947773e-05,
      "loss": 0.8408,
      "step": 199280
    },
    {
      "epoch": 3.4988324935479906,
      "grad_norm": 0.05999375818582311,
      "learning_rate": 4.4422187947646916e-05,
      "loss": 0.8365,
      "step": 199290
    },
    {
      "epoch": 3.499008058427992,
      "grad_norm": 0.06343815447536395,
      "learning_rate": 4.4413043764197795e-05,
      "loss": 0.8383,
      "step": 199300
    },
    {
      "epoch": 3.4991836233079936,
      "grad_norm": 0.055205954131196354,
      "learning_rate": 4.440390029771624e-05,
      "loss": 0.8389,
      "step": 199310
    },
    {
      "epoch": 3.499359188187995,
      "grad_norm": 0.05557390942436213,
      "learning_rate": 4.439475754831816e-05,
      "loss": 0.8419,
      "step": 199320
    },
    {
      "epoch": 3.499534753067996,
      "grad_norm": 0.047573862050795654,
      "learning_rate": 4.4385615516119345e-05,
      "loss": 0.8474,
      "step": 199330
    },
    {
      "epoch": 3.4997103179479976,
      "grad_norm": 0.04846385738418312,
      "learning_rate": 4.437647420123556e-05,
      "loss": 0.844,
      "step": 199340
    },
    {
      "epoch": 3.499885882827999,
      "grad_norm": 0.051312462552984606,
      "learning_rate": 4.4367333603782766e-05,
      "loss": 0.836,
      "step": 199350
    },
    {
      "epoch": 3.5000614477080005,
      "grad_norm": 0.05003334793380585,
      "learning_rate": 4.435819372387669e-05,
      "loss": 0.8399,
      "step": 199360
    },
    {
      "epoch": 3.500237012588002,
      "grad_norm": 0.06104063300597305,
      "learning_rate": 4.4349054561633144e-05,
      "loss": 0.8362,
      "step": 199370
    },
    {
      "epoch": 3.500412577468003,
      "grad_norm": 0.05343257284699606,
      "learning_rate": 4.433991611716797e-05,
      "loss": 0.8402,
      "step": 199380
    },
    {
      "epoch": 3.500588142348005,
      "grad_norm": 0.05939546126669199,
      "learning_rate": 4.433077839059688e-05,
      "loss": 0.836,
      "step": 199390
    },
    {
      "epoch": 3.500763707228006,
      "grad_norm": 0.05717246790442558,
      "learning_rate": 4.432164138203572e-05,
      "loss": 0.8455,
      "step": 199400
    },
    {
      "epoch": 3.5009392721080075,
      "grad_norm": 0.049220676072557874,
      "learning_rate": 4.431250509160017e-05,
      "loss": 0.8398,
      "step": 199410
    },
    {
      "epoch": 3.501114836988009,
      "grad_norm": 0.05272792076784081,
      "learning_rate": 4.430336951940611e-05,
      "loss": 0.8471,
      "step": 199420
    },
    {
      "epoch": 3.5012904018680104,
      "grad_norm": 0.047039806365734856,
      "learning_rate": 4.4294234665569176e-05,
      "loss": 0.8323,
      "step": 199430
    },
    {
      "epoch": 3.501465966748012,
      "grad_norm": 0.05404512714821062,
      "learning_rate": 4.428510053020517e-05,
      "loss": 0.8371,
      "step": 199440
    },
    {
      "epoch": 3.501641531628013,
      "grad_norm": 0.051069020747838575,
      "learning_rate": 4.4275967113429845e-05,
      "loss": 0.8397,
      "step": 199450
    },
    {
      "epoch": 3.5018170965080144,
      "grad_norm": 0.06963944349306432,
      "learning_rate": 4.426683441535885e-05,
      "loss": 0.8444,
      "step": 199460
    },
    {
      "epoch": 3.501992661388016,
      "grad_norm": 0.05036422373051378,
      "learning_rate": 4.425770243610793e-05,
      "loss": 0.8394,
      "step": 199470
    },
    {
      "epoch": 3.5021682262680174,
      "grad_norm": 0.04727120206654478,
      "learning_rate": 4.4248571175792794e-05,
      "loss": 0.841,
      "step": 199480
    },
    {
      "epoch": 3.502343791148019,
      "grad_norm": 0.052080717459183246,
      "learning_rate": 4.423944063452919e-05,
      "loss": 0.8365,
      "step": 199490
    },
    {
      "epoch": 3.50251935602802,
      "grad_norm": 0.046852378172602405,
      "learning_rate": 4.423031081243273e-05,
      "loss": 0.8361,
      "step": 199500
    },
    {
      "epoch": 3.502694920908022,
      "grad_norm": 0.07338125713976561,
      "learning_rate": 4.422118170961905e-05,
      "loss": 0.8346,
      "step": 199510
    },
    {
      "epoch": 3.502870485788023,
      "grad_norm": 0.05838099555089345,
      "learning_rate": 4.421205332620395e-05,
      "loss": 0.8425,
      "step": 199520
    },
    {
      "epoch": 3.5030460506680243,
      "grad_norm": 0.054424063564864546,
      "learning_rate": 4.420292566230298e-05,
      "loss": 0.8305,
      "step": 199530
    },
    {
      "epoch": 3.503221615548026,
      "grad_norm": 0.04571187389809651,
      "learning_rate": 4.4193798718031826e-05,
      "loss": 0.8385,
      "step": 199540
    },
    {
      "epoch": 3.5033971804280273,
      "grad_norm": 0.053701317244260104,
      "learning_rate": 4.4184672493506177e-05,
      "loss": 0.8464,
      "step": 199550
    },
    {
      "epoch": 3.5035727453080288,
      "grad_norm": 0.04975336375676813,
      "learning_rate": 4.4175546988841574e-05,
      "loss": 0.8412,
      "step": 199560
    },
    {
      "epoch": 3.50374831018803,
      "grad_norm": 0.05349716609259876,
      "learning_rate": 4.4166422204153715e-05,
      "loss": 0.8411,
      "step": 199570
    },
    {
      "epoch": 3.5039238750680313,
      "grad_norm": 0.05764846408355311,
      "learning_rate": 4.41572981395581e-05,
      "loss": 0.843,
      "step": 199580
    },
    {
      "epoch": 3.5040994399480327,
      "grad_norm": 0.043083211069038715,
      "learning_rate": 4.414817479517049e-05,
      "loss": 0.8465,
      "step": 199590
    },
    {
      "epoch": 3.5042750048280342,
      "grad_norm": 0.06683720724620766,
      "learning_rate": 4.413905217110637e-05,
      "loss": 0.8412,
      "step": 199600
    },
    {
      "epoch": 3.5044505697080357,
      "grad_norm": 0.059198696087760246,
      "learning_rate": 4.4129930267481344e-05,
      "loss": 0.8439,
      "step": 199610
    },
    {
      "epoch": 3.5046261345880367,
      "grad_norm": 0.051453111657391574,
      "learning_rate": 4.4120809084411045e-05,
      "loss": 0.843,
      "step": 199620
    },
    {
      "epoch": 3.5048016994680387,
      "grad_norm": 0.0637875768502769,
      "learning_rate": 4.411168862201096e-05,
      "loss": 0.8449,
      "step": 199630
    },
    {
      "epoch": 3.5049772643480397,
      "grad_norm": 0.059553692862372484,
      "learning_rate": 4.410256888039669e-05,
      "loss": 0.8493,
      "step": 199640
    },
    {
      "epoch": 3.505152829228041,
      "grad_norm": 0.05969456867233194,
      "learning_rate": 4.409344985968376e-05,
      "loss": 0.8396,
      "step": 199650
    },
    {
      "epoch": 3.5053283941080426,
      "grad_norm": 0.06261529841503233,
      "learning_rate": 4.4084331559987765e-05,
      "loss": 0.8374,
      "step": 199660
    },
    {
      "epoch": 3.505503958988044,
      "grad_norm": 0.04985268679278403,
      "learning_rate": 4.407521398142421e-05,
      "loss": 0.8452,
      "step": 199670
    },
    {
      "epoch": 3.5056795238680456,
      "grad_norm": 0.06216403230619467,
      "learning_rate": 4.406609712410852e-05,
      "loss": 0.842,
      "step": 199680
    },
    {
      "epoch": 3.5058550887480466,
      "grad_norm": 0.05057590446131984,
      "learning_rate": 4.405698098815636e-05,
      "loss": 0.841,
      "step": 199690
    },
    {
      "epoch": 3.506030653628048,
      "grad_norm": 0.07504400190400216,
      "learning_rate": 4.4047865573683136e-05,
      "loss": 0.8432,
      "step": 199700
    },
    {
      "epoch": 3.5062062185080496,
      "grad_norm": 0.06737206326313772,
      "learning_rate": 4.403875088080437e-05,
      "loss": 0.8416,
      "step": 199710
    },
    {
      "epoch": 3.506381783388051,
      "grad_norm": 0.0579214300502351,
      "learning_rate": 4.402963690963559e-05,
      "loss": 0.8391,
      "step": 199720
    },
    {
      "epoch": 3.5065573482680525,
      "grad_norm": 0.05544136228138098,
      "learning_rate": 4.402052366029218e-05,
      "loss": 0.8434,
      "step": 199730
    },
    {
      "epoch": 3.506732913148054,
      "grad_norm": 0.05603347221882146,
      "learning_rate": 4.401141113288969e-05,
      "loss": 0.8389,
      "step": 199740
    },
    {
      "epoch": 3.5069084780280555,
      "grad_norm": 0.05223532816489186,
      "learning_rate": 4.400229932754348e-05,
      "loss": 0.8404,
      "step": 199750
    },
    {
      "epoch": 3.5070840429080565,
      "grad_norm": 0.0511681736375196,
      "learning_rate": 4.3993188244369125e-05,
      "loss": 0.835,
      "step": 199760
    },
    {
      "epoch": 3.507259607788058,
      "grad_norm": 0.08044437566910967,
      "learning_rate": 4.398407788348197e-05,
      "loss": 0.8434,
      "step": 199770
    },
    {
      "epoch": 3.5074351726680595,
      "grad_norm": 0.05100619581084862,
      "learning_rate": 4.3974968244997474e-05,
      "loss": 0.8391,
      "step": 199780
    },
    {
      "epoch": 3.507610737548061,
      "grad_norm": 0.04286863585419777,
      "learning_rate": 4.3965859329031087e-05,
      "loss": 0.834,
      "step": 199790
    },
    {
      "epoch": 3.5077863024280624,
      "grad_norm": 0.05573230248057698,
      "learning_rate": 4.395675113569818e-05,
      "loss": 0.8417,
      "step": 199800
    },
    {
      "epoch": 3.5079618673080635,
      "grad_norm": 0.05003810963086917,
      "learning_rate": 4.394764366511414e-05,
      "loss": 0.8478,
      "step": 199810
    },
    {
      "epoch": 3.508137432188065,
      "grad_norm": 0.05516782959138597,
      "learning_rate": 4.393853691739441e-05,
      "loss": 0.8521,
      "step": 199820
    },
    {
      "epoch": 3.5083129970680664,
      "grad_norm": 0.04949345414195553,
      "learning_rate": 4.3929430892654374e-05,
      "loss": 0.8374,
      "step": 199830
    },
    {
      "epoch": 3.508488561948068,
      "grad_norm": 0.06215372162539031,
      "learning_rate": 4.39203255910094e-05,
      "loss": 0.8352,
      "step": 199840
    },
    {
      "epoch": 3.5086641268280694,
      "grad_norm": 0.05024716155493693,
      "learning_rate": 4.391122101257476e-05,
      "loss": 0.8406,
      "step": 199850
    },
    {
      "epoch": 3.508839691708071,
      "grad_norm": 0.06830393332795988,
      "learning_rate": 4.390211715746596e-05,
      "loss": 0.8439,
      "step": 199860
    },
    {
      "epoch": 3.5090152565880723,
      "grad_norm": 0.0746347745019505,
      "learning_rate": 4.3893014025798245e-05,
      "loss": 0.844,
      "step": 199870
    },
    {
      "epoch": 3.5091908214680734,
      "grad_norm": 0.05255087471140886,
      "learning_rate": 4.3883911617687e-05,
      "loss": 0.846,
      "step": 199880
    },
    {
      "epoch": 3.509366386348075,
      "grad_norm": 0.04454381245598604,
      "learning_rate": 4.387480993324756e-05,
      "loss": 0.8404,
      "step": 199890
    },
    {
      "epoch": 3.5095419512280763,
      "grad_norm": 0.06482225269417662,
      "learning_rate": 4.3865708972595185e-05,
      "loss": 0.8421,
      "step": 199900
    },
    {
      "epoch": 3.509717516108078,
      "grad_norm": 0.04377953591739905,
      "learning_rate": 4.385660873584528e-05,
      "loss": 0.8463,
      "step": 199910
    },
    {
      "epoch": 3.5098930809880793,
      "grad_norm": 0.04745999480944986,
      "learning_rate": 4.384750922311302e-05,
      "loss": 0.8377,
      "step": 199920
    },
    {
      "epoch": 3.5100686458680803,
      "grad_norm": 0.06493829804113885,
      "learning_rate": 4.383841043451384e-05,
      "loss": 0.8346,
      "step": 199930
    },
    {
      "epoch": 3.510244210748082,
      "grad_norm": 0.0589265140652444,
      "learning_rate": 4.382931237016291e-05,
      "loss": 0.8368,
      "step": 199940
    },
    {
      "epoch": 3.5104197756280833,
      "grad_norm": 0.060380696158208166,
      "learning_rate": 4.382021503017555e-05,
      "loss": 0.8424,
      "step": 199950
    },
    {
      "epoch": 3.5105953405080847,
      "grad_norm": 0.050706940901943734,
      "learning_rate": 4.3811118414667074e-05,
      "loss": 0.8478,
      "step": 199960
    },
    {
      "epoch": 3.5107709053880862,
      "grad_norm": 0.04938966974334838,
      "learning_rate": 4.380202252375264e-05,
      "loss": 0.8377,
      "step": 199970
    },
    {
      "epoch": 3.5109464702680877,
      "grad_norm": 0.04991631325885726,
      "learning_rate": 4.379292735754755e-05,
      "loss": 0.8445,
      "step": 199980
    },
    {
      "epoch": 3.511122035148089,
      "grad_norm": 0.05063680576262859,
      "learning_rate": 4.378383291616703e-05,
      "loss": 0.8433,
      "step": 199990
    },
    {
      "epoch": 3.51129760002809,
      "grad_norm": 0.045921490217110464,
      "learning_rate": 4.377473919972634e-05,
      "loss": 0.8377,
      "step": 200000
    },
    {
      "epoch": 3.5114731649080917,
      "grad_norm": 0.060426711658194215,
      "learning_rate": 4.376564620834068e-05,
      "loss": 0.8436,
      "step": 200010
    },
    {
      "epoch": 3.511648729788093,
      "grad_norm": 0.07501624178548578,
      "learning_rate": 4.375655394212518e-05,
      "loss": 0.84,
      "step": 200020
    },
    {
      "epoch": 3.5118242946680946,
      "grad_norm": 0.051672644605706024,
      "learning_rate": 4.3747462401195194e-05,
      "loss": 0.835,
      "step": 200030
    },
    {
      "epoch": 3.511999859548096,
      "grad_norm": 0.04874787225489718,
      "learning_rate": 4.3738371585665775e-05,
      "loss": 0.8339,
      "step": 200040
    },
    {
      "epoch": 3.512175424428097,
      "grad_norm": 0.06404115252799217,
      "learning_rate": 4.3729281495652174e-05,
      "loss": 0.8392,
      "step": 200050
    },
    {
      "epoch": 3.512350989308099,
      "grad_norm": 0.05442809462692098,
      "learning_rate": 4.3720192131269595e-05,
      "loss": 0.8429,
      "step": 200060
    },
    {
      "epoch": 3.5125265541881,
      "grad_norm": 0.04703821381212055,
      "learning_rate": 4.371110349263311e-05,
      "loss": 0.8403,
      "step": 200070
    },
    {
      "epoch": 3.5127021190681016,
      "grad_norm": 0.07621777699175948,
      "learning_rate": 4.370201557985797e-05,
      "loss": 0.8358,
      "step": 200080
    },
    {
      "epoch": 3.512877683948103,
      "grad_norm": 0.04284788835023576,
      "learning_rate": 4.3692928393059195e-05,
      "loss": 0.8408,
      "step": 200090
    },
    {
      "epoch": 3.5130532488281045,
      "grad_norm": 0.061793403053717104,
      "learning_rate": 4.368384193235208e-05,
      "loss": 0.837,
      "step": 200100
    },
    {
      "epoch": 3.513228813708106,
      "grad_norm": 0.060153546628811484,
      "learning_rate": 4.3674756197851626e-05,
      "loss": 0.8325,
      "step": 200110
    },
    {
      "epoch": 3.513404378588107,
      "grad_norm": 0.06594220254423455,
      "learning_rate": 4.3665671189673e-05,
      "loss": 0.8427,
      "step": 200120
    },
    {
      "epoch": 3.5135799434681085,
      "grad_norm": 0.040698407574556894,
      "learning_rate": 4.365658690793134e-05,
      "loss": 0.8437,
      "step": 200130
    },
    {
      "epoch": 3.51375550834811,
      "grad_norm": 0.0631111114245094,
      "learning_rate": 4.364750335274169e-05,
      "loss": 0.841,
      "step": 200140
    },
    {
      "epoch": 3.5139310732281115,
      "grad_norm": 0.060637114329597314,
      "learning_rate": 4.363842052421915e-05,
      "loss": 0.8474,
      "step": 200150
    },
    {
      "epoch": 3.514106638108113,
      "grad_norm": 0.05096224853498794,
      "learning_rate": 4.3629338422478834e-05,
      "loss": 0.8384,
      "step": 200160
    },
    {
      "epoch": 3.514282202988114,
      "grad_norm": 0.06012695491238213,
      "learning_rate": 4.3620257047635814e-05,
      "loss": 0.8422,
      "step": 200170
    },
    {
      "epoch": 3.514457767868116,
      "grad_norm": 0.05557316116861647,
      "learning_rate": 4.3611176399805136e-05,
      "loss": 0.841,
      "step": 200180
    },
    {
      "epoch": 3.514633332748117,
      "grad_norm": 0.06299761939125324,
      "learning_rate": 4.360209647910179e-05,
      "loss": 0.8417,
      "step": 200190
    },
    {
      "epoch": 3.5148088976281184,
      "grad_norm": 0.048451978478948936,
      "learning_rate": 4.3593017285640944e-05,
      "loss": 0.8405,
      "step": 200200
    },
    {
      "epoch": 3.51498446250812,
      "grad_norm": 0.05208389474457796,
      "learning_rate": 4.358393881953753e-05,
      "loss": 0.8426,
      "step": 200210
    },
    {
      "epoch": 3.5151600273881214,
      "grad_norm": 0.05544420520321481,
      "learning_rate": 4.3574861080906624e-05,
      "loss": 0.8467,
      "step": 200220
    },
    {
      "epoch": 3.515335592268123,
      "grad_norm": 0.05076551425067777,
      "learning_rate": 4.3565784069863276e-05,
      "loss": 0.8412,
      "step": 200230
    },
    {
      "epoch": 3.515511157148124,
      "grad_norm": 0.051521259570557434,
      "learning_rate": 4.35567077865224e-05,
      "loss": 0.8475,
      "step": 200240
    },
    {
      "epoch": 3.5156867220281254,
      "grad_norm": 0.040401266873476824,
      "learning_rate": 4.354763223099909e-05,
      "loss": 0.8452,
      "step": 200250
    },
    {
      "epoch": 3.515862286908127,
      "grad_norm": 0.04688066282121088,
      "learning_rate": 4.353855740340822e-05,
      "loss": 0.8435,
      "step": 200260
    },
    {
      "epoch": 3.5160378517881283,
      "grad_norm": 0.049432269706259614,
      "learning_rate": 4.35294833038649e-05,
      "loss": 0.8386,
      "step": 200270
    },
    {
      "epoch": 3.51621341666813,
      "grad_norm": 0.0501292903133393,
      "learning_rate": 4.3520409932484025e-05,
      "loss": 0.8421,
      "step": 200280
    },
    {
      "epoch": 3.516388981548131,
      "grad_norm": 0.04284844601911539,
      "learning_rate": 4.351133728938055e-05,
      "loss": 0.8488,
      "step": 200290
    },
    {
      "epoch": 3.5165645464281328,
      "grad_norm": 0.051977273537785765,
      "learning_rate": 4.350226537466951e-05,
      "loss": 0.8345,
      "step": 200300
    },
    {
      "epoch": 3.516740111308134,
      "grad_norm": 0.047303334100046196,
      "learning_rate": 4.349319418846573e-05,
      "loss": 0.8357,
      "step": 200310
    },
    {
      "epoch": 3.5169156761881353,
      "grad_norm": 0.046097528262605,
      "learning_rate": 4.348412373088422e-05,
      "loss": 0.8384,
      "step": 200320
    },
    {
      "epoch": 3.5170912410681368,
      "grad_norm": 0.04962340780071639,
      "learning_rate": 4.347505400203987e-05,
      "loss": 0.8381,
      "step": 200330
    },
    {
      "epoch": 3.5172668059481382,
      "grad_norm": 0.05749064573442487,
      "learning_rate": 4.346598500204765e-05,
      "loss": 0.8416,
      "step": 200340
    },
    {
      "epoch": 3.5174423708281397,
      "grad_norm": 0.054715299091778406,
      "learning_rate": 4.345691673102242e-05,
      "loss": 0.8472,
      "step": 200350
    },
    {
      "epoch": 3.5176179357081407,
      "grad_norm": 0.050159142306578544,
      "learning_rate": 4.344784918907902e-05,
      "loss": 0.8377,
      "step": 200360
    },
    {
      "epoch": 3.517793500588142,
      "grad_norm": 0.07058592307034914,
      "learning_rate": 4.343878237633247e-05,
      "loss": 0.8415,
      "step": 200370
    },
    {
      "epoch": 3.5179690654681437,
      "grad_norm": 0.04998301831233742,
      "learning_rate": 4.342971629289754e-05,
      "loss": 0.8375,
      "step": 200380
    },
    {
      "epoch": 3.518144630348145,
      "grad_norm": 0.05460613765618193,
      "learning_rate": 4.3420650938889136e-05,
      "loss": 0.8356,
      "step": 200390
    },
    {
      "epoch": 3.5183201952281467,
      "grad_norm": 0.05154408626693823,
      "learning_rate": 4.3411586314422165e-05,
      "loss": 0.8337,
      "step": 200400
    },
    {
      "epoch": 3.518495760108148,
      "grad_norm": 0.048505066758184265,
      "learning_rate": 4.340252241961139e-05,
      "loss": 0.8447,
      "step": 200410
    },
    {
      "epoch": 3.5186713249881496,
      "grad_norm": 0.07605049429855563,
      "learning_rate": 4.339345925457173e-05,
      "loss": 0.8381,
      "step": 200420
    },
    {
      "epoch": 3.5188468898681506,
      "grad_norm": 0.046504410820231094,
      "learning_rate": 4.338439681941792e-05,
      "loss": 0.8388,
      "step": 200430
    },
    {
      "epoch": 3.519022454748152,
      "grad_norm": 0.06161400844388744,
      "learning_rate": 4.33753351142649e-05,
      "loss": 0.8383,
      "step": 200440
    },
    {
      "epoch": 3.5191980196281536,
      "grad_norm": 0.060346243684677246,
      "learning_rate": 4.3366274139227396e-05,
      "loss": 0.8379,
      "step": 200450
    },
    {
      "epoch": 3.519373584508155,
      "grad_norm": 0.05928492553013754,
      "learning_rate": 4.3357213894420254e-05,
      "loss": 0.8466,
      "step": 200460
    },
    {
      "epoch": 3.5195491493881566,
      "grad_norm": 0.05710179586020797,
      "learning_rate": 4.3348154379958286e-05,
      "loss": 0.8463,
      "step": 200470
    },
    {
      "epoch": 3.5197247142681576,
      "grad_norm": 0.06250548178000355,
      "learning_rate": 4.333909559595621e-05,
      "loss": 0.8386,
      "step": 200480
    },
    {
      "epoch": 3.519900279148159,
      "grad_norm": 0.051309685216415345,
      "learning_rate": 4.333003754252886e-05,
      "loss": 0.8386,
      "step": 200490
    },
    {
      "epoch": 3.5200758440281605,
      "grad_norm": 0.054221509151039994,
      "learning_rate": 4.3320980219790975e-05,
      "loss": 0.8415,
      "step": 200500
    },
    {
      "epoch": 3.520251408908162,
      "grad_norm": 0.06538182838555578,
      "learning_rate": 4.331192362785736e-05,
      "loss": 0.839,
      "step": 200510
    },
    {
      "epoch": 3.5204269737881635,
      "grad_norm": 0.04594577672926932,
      "learning_rate": 4.330286776684274e-05,
      "loss": 0.837,
      "step": 200520
    },
    {
      "epoch": 3.520602538668165,
      "grad_norm": 0.046746072361353176,
      "learning_rate": 4.329381263686176e-05,
      "loss": 0.8445,
      "step": 200530
    },
    {
      "epoch": 3.5207781035481664,
      "grad_norm": 0.043033327925548266,
      "learning_rate": 4.3284758238029304e-05,
      "loss": 0.8328,
      "step": 200540
    },
    {
      "epoch": 3.5209536684281675,
      "grad_norm": 0.04319302898493033,
      "learning_rate": 4.3275704570459994e-05,
      "loss": 0.8465,
      "step": 200550
    },
    {
      "epoch": 3.521129233308169,
      "grad_norm": 0.05351123324724292,
      "learning_rate": 4.326665163426856e-05,
      "loss": 0.8397,
      "step": 200560
    },
    {
      "epoch": 3.5213047981881704,
      "grad_norm": 0.0549013181792766,
      "learning_rate": 4.325759942956976e-05,
      "loss": 0.8469,
      "step": 200570
    },
    {
      "epoch": 3.521480363068172,
      "grad_norm": 0.0508505728736163,
      "learning_rate": 4.324854795647821e-05,
      "loss": 0.8432,
      "step": 200580
    },
    {
      "epoch": 3.5216559279481734,
      "grad_norm": 0.052093035184010185,
      "learning_rate": 4.323949721510865e-05,
      "loss": 0.8334,
      "step": 200590
    },
    {
      "epoch": 3.5218314928281744,
      "grad_norm": 0.06919383874189088,
      "learning_rate": 4.3230447205575656e-05,
      "loss": 0.8418,
      "step": 200600
    },
    {
      "epoch": 3.522007057708176,
      "grad_norm": 0.04945167602892831,
      "learning_rate": 4.322139792799405e-05,
      "loss": 0.8363,
      "step": 200610
    },
    {
      "epoch": 3.5221826225881774,
      "grad_norm": 0.05561150877150249,
      "learning_rate": 4.3212349382478365e-05,
      "loss": 0.8409,
      "step": 200620
    },
    {
      "epoch": 3.522358187468179,
      "grad_norm": 0.05708291841874271,
      "learning_rate": 4.320330156914329e-05,
      "loss": 0.8399,
      "step": 200630
    },
    {
      "epoch": 3.5225337523481803,
      "grad_norm": 0.04783558247722127,
      "learning_rate": 4.319425448810348e-05,
      "loss": 0.8412,
      "step": 200640
    },
    {
      "epoch": 3.522709317228182,
      "grad_norm": 0.06457255324371095,
      "learning_rate": 4.318520813947353e-05,
      "loss": 0.8388,
      "step": 200650
    },
    {
      "epoch": 3.5228848821081833,
      "grad_norm": 0.05053482786127865,
      "learning_rate": 4.317616252336807e-05,
      "loss": 0.8453,
      "step": 200660
    },
    {
      "epoch": 3.5230604469881843,
      "grad_norm": 0.058359960472974086,
      "learning_rate": 4.316711763990171e-05,
      "loss": 0.8463,
      "step": 200670
    },
    {
      "epoch": 3.523236011868186,
      "grad_norm": 0.04737950395158914,
      "learning_rate": 4.315807348918908e-05,
      "loss": 0.8431,
      "step": 200680
    },
    {
      "epoch": 3.5234115767481873,
      "grad_norm": 0.06340915772919703,
      "learning_rate": 4.314903007134476e-05,
      "loss": 0.843,
      "step": 200690
    },
    {
      "epoch": 3.5235871416281888,
      "grad_norm": 0.04536663239342469,
      "learning_rate": 4.3139987386483226e-05,
      "loss": 0.832,
      "step": 200700
    },
    {
      "epoch": 3.5237627065081902,
      "grad_norm": 0.056985546756060046,
      "learning_rate": 4.3130945434719223e-05,
      "loss": 0.8401,
      "step": 200710
    },
    {
      "epoch": 3.5239382713881913,
      "grad_norm": 0.04522381996816818,
      "learning_rate": 4.31219042161672e-05,
      "loss": 0.8372,
      "step": 200720
    },
    {
      "epoch": 3.5241138362681927,
      "grad_norm": 0.0525512118165931,
      "learning_rate": 4.311286373094174e-05,
      "loss": 0.8353,
      "step": 200730
    },
    {
      "epoch": 3.5242894011481942,
      "grad_norm": 0.043526638043746665,
      "learning_rate": 4.3103823979157424e-05,
      "loss": 0.8487,
      "step": 200740
    },
    {
      "epoch": 3.5244649660281957,
      "grad_norm": 0.06008058536811828,
      "learning_rate": 4.309478496092871e-05,
      "loss": 0.8426,
      "step": 200750
    },
    {
      "epoch": 3.524640530908197,
      "grad_norm": 0.05359132617709003,
      "learning_rate": 4.3085746676370214e-05,
      "loss": 0.8446,
      "step": 200760
    },
    {
      "epoch": 3.5248160957881987,
      "grad_norm": 0.04905551921850902,
      "learning_rate": 4.307670912559634e-05,
      "loss": 0.8447,
      "step": 200770
    },
    {
      "epoch": 3.5249916606682,
      "grad_norm": 0.06869321591238502,
      "learning_rate": 4.306767230872172e-05,
      "loss": 0.8353,
      "step": 200780
    },
    {
      "epoch": 3.525167225548201,
      "grad_norm": 0.042750199100669566,
      "learning_rate": 4.305863622586078e-05,
      "loss": 0.843,
      "step": 200790
    },
    {
      "epoch": 3.5253427904282026,
      "grad_norm": 0.05346122387441395,
      "learning_rate": 4.3049600877127996e-05,
      "loss": 0.8521,
      "step": 200800
    },
    {
      "epoch": 3.525518355308204,
      "grad_norm": 0.0528441221934692,
      "learning_rate": 4.304056626263793e-05,
      "loss": 0.8453,
      "step": 200810
    },
    {
      "epoch": 3.5256939201882056,
      "grad_norm": 0.04527935388183837,
      "learning_rate": 4.3031532382504965e-05,
      "loss": 0.8409,
      "step": 200820
    },
    {
      "epoch": 3.525869485068207,
      "grad_norm": 0.10367704298820406,
      "learning_rate": 4.302249923684357e-05,
      "loss": 0.8367,
      "step": 200830
    },
    {
      "epoch": 3.526045049948208,
      "grad_norm": 0.05866343978003809,
      "learning_rate": 4.3013466825768244e-05,
      "loss": 0.833,
      "step": 200840
    },
    {
      "epoch": 3.52622061482821,
      "grad_norm": 0.048623479220839295,
      "learning_rate": 4.300443514939344e-05,
      "loss": 0.8359,
      "step": 200850
    },
    {
      "epoch": 3.526396179708211,
      "grad_norm": 0.04864697716908669,
      "learning_rate": 4.299540420783354e-05,
      "loss": 0.846,
      "step": 200860
    },
    {
      "epoch": 3.5265717445882125,
      "grad_norm": 0.0527306001702434,
      "learning_rate": 4.298637400120294e-05,
      "loss": 0.8412,
      "step": 200870
    },
    {
      "epoch": 3.526747309468214,
      "grad_norm": 0.05436052382282177,
      "learning_rate": 4.2977344529616146e-05,
      "loss": 0.8373,
      "step": 200880
    },
    {
      "epoch": 3.5269228743482155,
      "grad_norm": 0.04838642063469154,
      "learning_rate": 4.29683157931875e-05,
      "loss": 0.8414,
      "step": 200890
    },
    {
      "epoch": 3.527098439228217,
      "grad_norm": 0.06933131659805908,
      "learning_rate": 4.2959287792031396e-05,
      "loss": 0.8342,
      "step": 200900
    },
    {
      "epoch": 3.527274004108218,
      "grad_norm": 0.0567725618347821,
      "learning_rate": 4.295026052626229e-05,
      "loss": 0.8463,
      "step": 200910
    },
    {
      "epoch": 3.5274495689882195,
      "grad_norm": 0.059354647316631017,
      "learning_rate": 4.294123399599447e-05,
      "loss": 0.8412,
      "step": 200920
    },
    {
      "epoch": 3.527625133868221,
      "grad_norm": 0.07152719936550288,
      "learning_rate": 4.293220820134238e-05,
      "loss": 0.8343,
      "step": 200930
    },
    {
      "epoch": 3.5278006987482224,
      "grad_norm": 0.06451670529074172,
      "learning_rate": 4.292318314242025e-05,
      "loss": 0.8469,
      "step": 200940
    },
    {
      "epoch": 3.527976263628224,
      "grad_norm": 0.054274443477392506,
      "learning_rate": 4.2914158819342615e-05,
      "loss": 0.8419,
      "step": 200950
    },
    {
      "epoch": 3.528151828508225,
      "grad_norm": 0.05370735960782895,
      "learning_rate": 4.290513523222368e-05,
      "loss": 0.846,
      "step": 200960
    },
    {
      "epoch": 3.528327393388227,
      "grad_norm": 0.046032293894312196,
      "learning_rate": 4.2896112381177807e-05,
      "loss": 0.8405,
      "step": 200970
    },
    {
      "epoch": 3.528502958268228,
      "grad_norm": 0.05418648322146241,
      "learning_rate": 4.2887090266319374e-05,
      "loss": 0.8367,
      "step": 200980
    },
    {
      "epoch": 3.5286785231482294,
      "grad_norm": 0.04953875351859977,
      "learning_rate": 4.28780688877626e-05,
      "loss": 0.8446,
      "step": 200990
    },
    {
      "epoch": 3.528854088028231,
      "grad_norm": 0.0562676623498435,
      "learning_rate": 4.2869048245621825e-05,
      "loss": 0.8397,
      "step": 201000
    },
    {
      "epoch": 3.5290296529082323,
      "grad_norm": 0.05450705292008742,
      "learning_rate": 4.286002834001136e-05,
      "loss": 0.8387,
      "step": 201010
    },
    {
      "epoch": 3.529205217788234,
      "grad_norm": 0.04344498216675427,
      "learning_rate": 4.285100917104551e-05,
      "loss": 0.8378,
      "step": 201020
    },
    {
      "epoch": 3.529380782668235,
      "grad_norm": 0.0518387944986638,
      "learning_rate": 4.2841990738838515e-05,
      "loss": 0.8398,
      "step": 201030
    },
    {
      "epoch": 3.5295563475482363,
      "grad_norm": 0.05343799360328918,
      "learning_rate": 4.283297304350458e-05,
      "loss": 0.8424,
      "step": 201040
    },
    {
      "epoch": 3.529731912428238,
      "grad_norm": 0.06737733149927795,
      "learning_rate": 4.282395608515809e-05,
      "loss": 0.8336,
      "step": 201050
    },
    {
      "epoch": 3.5299074773082393,
      "grad_norm": 0.0523217549803375,
      "learning_rate": 4.28149398639132e-05,
      "loss": 0.8419,
      "step": 201060
    },
    {
      "epoch": 3.5300830421882408,
      "grad_norm": 0.06014229952626238,
      "learning_rate": 4.2805924379884155e-05,
      "loss": 0.8501,
      "step": 201070
    },
    {
      "epoch": 3.530258607068242,
      "grad_norm": 0.04811005731577163,
      "learning_rate": 4.279690963318525e-05,
      "loss": 0.8507,
      "step": 201080
    },
    {
      "epoch": 3.5304341719482437,
      "grad_norm": 0.06211751577596833,
      "learning_rate": 4.278789562393062e-05,
      "loss": 0.8324,
      "step": 201090
    },
    {
      "epoch": 3.5306097368282447,
      "grad_norm": 0.04899702202597878,
      "learning_rate": 4.277888235223454e-05,
      "loss": 0.8423,
      "step": 201100
    },
    {
      "epoch": 3.5307853017082462,
      "grad_norm": 0.05288289677461504,
      "learning_rate": 4.27698698182111e-05,
      "loss": 0.8333,
      "step": 201110
    },
    {
      "epoch": 3.5309608665882477,
      "grad_norm": 0.04825875734468679,
      "learning_rate": 4.276085802197465e-05,
      "loss": 0.8355,
      "step": 201120
    },
    {
      "epoch": 3.531136431468249,
      "grad_norm": 0.054957521256606995,
      "learning_rate": 4.2751846963639255e-05,
      "loss": 0.84,
      "step": 201130
    },
    {
      "epoch": 3.5313119963482507,
      "grad_norm": 0.04586619716459698,
      "learning_rate": 4.274283664331912e-05,
      "loss": 0.8415,
      "step": 201140
    },
    {
      "epoch": 3.5314875612282517,
      "grad_norm": 0.06653702320531148,
      "learning_rate": 4.273382706112844e-05,
      "loss": 0.8384,
      "step": 201150
    },
    {
      "epoch": 3.531663126108253,
      "grad_norm": 0.061280756754452076,
      "learning_rate": 4.2724818217181305e-05,
      "loss": 0.8358,
      "step": 201160
    },
    {
      "epoch": 3.5318386909882546,
      "grad_norm": 0.06029525898069388,
      "learning_rate": 4.2715810111591904e-05,
      "loss": 0.8446,
      "step": 201170
    },
    {
      "epoch": 3.532014255868256,
      "grad_norm": 0.06006420745862922,
      "learning_rate": 4.270680274447434e-05,
      "loss": 0.8398,
      "step": 201180
    },
    {
      "epoch": 3.5321898207482576,
      "grad_norm": 0.056744105621414645,
      "learning_rate": 4.2697796115942804e-05,
      "loss": 0.8452,
      "step": 201190
    },
    {
      "epoch": 3.532365385628259,
      "grad_norm": 0.05840971198728881,
      "learning_rate": 4.2688790226111355e-05,
      "loss": 0.8429,
      "step": 201200
    },
    {
      "epoch": 3.5325409505082606,
      "grad_norm": 0.068777989934829,
      "learning_rate": 4.2679785075094055e-05,
      "loss": 0.84,
      "step": 201210
    },
    {
      "epoch": 3.5327165153882616,
      "grad_norm": 0.06460305610724093,
      "learning_rate": 4.267078066300511e-05,
      "loss": 0.8472,
      "step": 201220
    },
    {
      "epoch": 3.532892080268263,
      "grad_norm": 0.05761812687298573,
      "learning_rate": 4.266177698995852e-05,
      "loss": 0.8431,
      "step": 201230
    },
    {
      "epoch": 3.5330676451482645,
      "grad_norm": 0.0513825693913799,
      "learning_rate": 4.265277405606841e-05,
      "loss": 0.8399,
      "step": 201240
    },
    {
      "epoch": 3.533243210028266,
      "grad_norm": 0.07529745142997883,
      "learning_rate": 4.264377186144886e-05,
      "loss": 0.8384,
      "step": 201250
    },
    {
      "epoch": 3.5334187749082675,
      "grad_norm": 0.05740060929159923,
      "learning_rate": 4.263477040621387e-05,
      "loss": 0.8398,
      "step": 201260
    },
    {
      "epoch": 3.5335943397882685,
      "grad_norm": 0.06174356692341561,
      "learning_rate": 4.262576969047756e-05,
      "loss": 0.8546,
      "step": 201270
    },
    {
      "epoch": 3.53376990466827,
      "grad_norm": 0.05383807159638765,
      "learning_rate": 4.261676971435385e-05,
      "loss": 0.8423,
      "step": 201280
    },
    {
      "epoch": 3.5339454695482715,
      "grad_norm": 0.05526515147242702,
      "learning_rate": 4.2607770477956955e-05,
      "loss": 0.8429,
      "step": 201290
    },
    {
      "epoch": 3.534121034428273,
      "grad_norm": 0.06794025869132729,
      "learning_rate": 4.259877198140075e-05,
      "loss": 0.8392,
      "step": 201300
    },
    {
      "epoch": 3.5342965993082744,
      "grad_norm": 0.05390338530655235,
      "learning_rate": 4.258977422479929e-05,
      "loss": 0.8375,
      "step": 201310
    },
    {
      "epoch": 3.534472164188276,
      "grad_norm": 0.06232352863114447,
      "learning_rate": 4.258077720826663e-05,
      "loss": 0.842,
      "step": 201320
    },
    {
      "epoch": 3.5346477290682774,
      "grad_norm": 0.055822282410021996,
      "learning_rate": 4.257178093191668e-05,
      "loss": 0.8343,
      "step": 201330
    },
    {
      "epoch": 3.5348232939482784,
      "grad_norm": 0.058092520511197066,
      "learning_rate": 4.2562785395863454e-05,
      "loss": 0.8363,
      "step": 201340
    },
    {
      "epoch": 3.53499885882828,
      "grad_norm": 0.04435025589131343,
      "learning_rate": 4.2553790600220955e-05,
      "loss": 0.8486,
      "step": 201350
    },
    {
      "epoch": 3.5351744237082814,
      "grad_norm": 0.052657310697442485,
      "learning_rate": 4.254479654510315e-05,
      "loss": 0.8398,
      "step": 201360
    },
    {
      "epoch": 3.535349988588283,
      "grad_norm": 0.04899493858366757,
      "learning_rate": 4.253580323062398e-05,
      "loss": 0.8406,
      "step": 201370
    },
    {
      "epoch": 3.5355255534682843,
      "grad_norm": 0.05271448966053136,
      "learning_rate": 4.252681065689732e-05,
      "loss": 0.8335,
      "step": 201380
    },
    {
      "epoch": 3.5357011183482854,
      "grad_norm": 0.06390582867674015,
      "learning_rate": 4.251781882403723e-05,
      "loss": 0.8403,
      "step": 201390
    },
    {
      "epoch": 3.535876683228287,
      "grad_norm": 0.09197194866727099,
      "learning_rate": 4.2508827732157576e-05,
      "loss": 0.8336,
      "step": 201400
    },
    {
      "epoch": 3.5360522481082883,
      "grad_norm": 0.0482145070071272,
      "learning_rate": 4.249983738137227e-05,
      "loss": 0.8408,
      "step": 201410
    },
    {
      "epoch": 3.53622781298829,
      "grad_norm": 0.04504749963476472,
      "learning_rate": 4.249084777179528e-05,
      "loss": 0.8396,
      "step": 201420
    },
    {
      "epoch": 3.5364033778682913,
      "grad_norm": 0.04679672764023624,
      "learning_rate": 4.2481858903540415e-05,
      "loss": 0.8429,
      "step": 201430
    },
    {
      "epoch": 3.5365789427482928,
      "grad_norm": 0.06127377894028651,
      "learning_rate": 4.2472870776721646e-05,
      "loss": 0.8413,
      "step": 201440
    },
    {
      "epoch": 3.5367545076282942,
      "grad_norm": 0.06834177909302482,
      "learning_rate": 4.246388339145276e-05,
      "loss": 0.84,
      "step": 201450
    },
    {
      "epoch": 3.5369300725082953,
      "grad_norm": 0.06396679257824499,
      "learning_rate": 4.2454896747847756e-05,
      "loss": 0.8386,
      "step": 201460
    },
    {
      "epoch": 3.5371056373882968,
      "grad_norm": 0.05556740075290939,
      "learning_rate": 4.244591084602039e-05,
      "loss": 0.8397,
      "step": 201470
    },
    {
      "epoch": 3.5372812022682982,
      "grad_norm": 0.056128287259415946,
      "learning_rate": 4.2436925686084567e-05,
      "loss": 0.85,
      "step": 201480
    },
    {
      "epoch": 3.5374567671482997,
      "grad_norm": 0.05558697365864548,
      "learning_rate": 4.242794126815415e-05,
      "loss": 0.8384,
      "step": 201490
    },
    {
      "epoch": 3.537632332028301,
      "grad_norm": 0.051895609397885886,
      "learning_rate": 4.2418957592342904e-05,
      "loss": 0.8409,
      "step": 201500
    },
    {
      "epoch": 3.537807896908302,
      "grad_norm": 0.059973566189913514,
      "learning_rate": 4.240997465876469e-05,
      "loss": 0.8399,
      "step": 201510
    },
    {
      "epoch": 3.537983461788304,
      "grad_norm": 0.043842334543189004,
      "learning_rate": 4.2400992467533334e-05,
      "loss": 0.8403,
      "step": 201520
    },
    {
      "epoch": 3.538159026668305,
      "grad_norm": 0.0557468648781786,
      "learning_rate": 4.239201101876267e-05,
      "loss": 0.8443,
      "step": 201530
    },
    {
      "epoch": 3.5383345915483067,
      "grad_norm": 0.060151814973549905,
      "learning_rate": 4.238303031256645e-05,
      "loss": 0.8378,
      "step": 201540
    },
    {
      "epoch": 3.538510156428308,
      "grad_norm": 0.05657778645928298,
      "learning_rate": 4.237405034905841e-05,
      "loss": 0.837,
      "step": 201550
    },
    {
      "epoch": 3.5386857213083096,
      "grad_norm": 0.05572474609075534,
      "learning_rate": 4.236507112835245e-05,
      "loss": 0.8365,
      "step": 201560
    },
    {
      "epoch": 3.538861286188311,
      "grad_norm": 0.05921028226383846,
      "learning_rate": 4.235609265056224e-05,
      "loss": 0.8385,
      "step": 201570
    },
    {
      "epoch": 3.539036851068312,
      "grad_norm": 0.05317515894513794,
      "learning_rate": 4.234711491580158e-05,
      "loss": 0.8337,
      "step": 201580
    },
    {
      "epoch": 3.5392124159483136,
      "grad_norm": 0.05182159976608913,
      "learning_rate": 4.2338137924184233e-05,
      "loss": 0.8435,
      "step": 201590
    },
    {
      "epoch": 3.539387980828315,
      "grad_norm": 0.039810414390963444,
      "learning_rate": 4.232916167582391e-05,
      "loss": 0.8414,
      "step": 201600
    },
    {
      "epoch": 3.5395635457083165,
      "grad_norm": 0.05534585151638206,
      "learning_rate": 4.232018617083438e-05,
      "loss": 0.8396,
      "step": 201610
    },
    {
      "epoch": 3.539739110588318,
      "grad_norm": 0.052290184934389534,
      "learning_rate": 4.231121140932927e-05,
      "loss": 0.8413,
      "step": 201620
    },
    {
      "epoch": 3.539914675468319,
      "grad_norm": 0.054413821981248466,
      "learning_rate": 4.230223739142242e-05,
      "loss": 0.843,
      "step": 201630
    },
    {
      "epoch": 3.540090240348321,
      "grad_norm": 0.050725868030932215,
      "learning_rate": 4.2293264117227434e-05,
      "loss": 0.8348,
      "step": 201640
    },
    {
      "epoch": 3.540265805228322,
      "grad_norm": 0.05769677718067033,
      "learning_rate": 4.228429158685805e-05,
      "loss": 0.8379,
      "step": 201650
    },
    {
      "epoch": 3.5404413701083235,
      "grad_norm": 0.06771822867749121,
      "learning_rate": 4.227531980042798e-05,
      "loss": 0.8434,
      "step": 201660
    },
    {
      "epoch": 3.540616934988325,
      "grad_norm": 0.05753052573653131,
      "learning_rate": 4.2266348758050825e-05,
      "loss": 0.844,
      "step": 201670
    },
    {
      "epoch": 3.5407924998683264,
      "grad_norm": 0.055005529824736416,
      "learning_rate": 4.2257378459840295e-05,
      "loss": 0.8369,
      "step": 201680
    },
    {
      "epoch": 3.540968064748328,
      "grad_norm": 0.04997246293044959,
      "learning_rate": 4.224840890591003e-05,
      "loss": 0.8444,
      "step": 201690
    },
    {
      "epoch": 3.541143629628329,
      "grad_norm": 0.04140668048217736,
      "learning_rate": 4.223944009637372e-05,
      "loss": 0.8439,
      "step": 201700
    },
    {
      "epoch": 3.5413191945083304,
      "grad_norm": 0.05507477734404706,
      "learning_rate": 4.223047203134497e-05,
      "loss": 0.8424,
      "step": 201710
    },
    {
      "epoch": 3.541494759388332,
      "grad_norm": 0.061305070212260185,
      "learning_rate": 4.2221504710937324e-05,
      "loss": 0.8412,
      "step": 201720
    },
    {
      "epoch": 3.5416703242683334,
      "grad_norm": 0.0686124377756459,
      "learning_rate": 4.221253813526455e-05,
      "loss": 0.8353,
      "step": 201730
    },
    {
      "epoch": 3.541845889148335,
      "grad_norm": 0.07199832125226793,
      "learning_rate": 4.220357230444015e-05,
      "loss": 0.8403,
      "step": 201740
    },
    {
      "epoch": 3.542021454028336,
      "grad_norm": 0.0640870363474303,
      "learning_rate": 4.2194607218577763e-05,
      "loss": 0.8371,
      "step": 201750
    },
    {
      "epoch": 3.542197018908338,
      "grad_norm": 0.05442392487745125,
      "learning_rate": 4.218564287779101e-05,
      "loss": 0.8506,
      "step": 201760
    },
    {
      "epoch": 3.542372583788339,
      "grad_norm": 0.05143432806152093,
      "learning_rate": 4.217667928219339e-05,
      "loss": 0.8379,
      "step": 201770
    },
    {
      "epoch": 3.5425481486683403,
      "grad_norm": 0.0532560769852996,
      "learning_rate": 4.2167716431898546e-05,
      "loss": 0.8452,
      "step": 201780
    },
    {
      "epoch": 3.542723713548342,
      "grad_norm": 0.04958556734858083,
      "learning_rate": 4.2158754327019954e-05,
      "loss": 0.8459,
      "step": 201790
    },
    {
      "epoch": 3.5428992784283433,
      "grad_norm": 0.04723383673247859,
      "learning_rate": 4.214979296767128e-05,
      "loss": 0.8275,
      "step": 201800
    },
    {
      "epoch": 3.5430748433083448,
      "grad_norm": 0.05408189052213284,
      "learning_rate": 4.2140832353965984e-05,
      "loss": 0.8405,
      "step": 201810
    },
    {
      "epoch": 3.543250408188346,
      "grad_norm": 0.04572748951677873,
      "learning_rate": 4.213187248601762e-05,
      "loss": 0.8402,
      "step": 201820
    },
    {
      "epoch": 3.5434259730683473,
      "grad_norm": 0.05164368026723292,
      "learning_rate": 4.212291336393974e-05,
      "loss": 0.8406,
      "step": 201830
    },
    {
      "epoch": 3.5436015379483488,
      "grad_norm": 0.04233450146974177,
      "learning_rate": 4.21139549878458e-05,
      "loss": 0.8412,
      "step": 201840
    },
    {
      "epoch": 3.5437771028283502,
      "grad_norm": 0.04213621604336203,
      "learning_rate": 4.210499735784934e-05,
      "loss": 0.8342,
      "step": 201850
    },
    {
      "epoch": 3.5439526677083517,
      "grad_norm": 0.0528052708691953,
      "learning_rate": 4.209604047406385e-05,
      "loss": 0.8407,
      "step": 201860
    },
    {
      "epoch": 3.544128232588353,
      "grad_norm": 0.06032320249162949,
      "learning_rate": 4.2087084336602845e-05,
      "loss": 0.8411,
      "step": 201870
    },
    {
      "epoch": 3.5443037974683547,
      "grad_norm": 0.047129713590149065,
      "learning_rate": 4.2078128945579785e-05,
      "loss": 0.8412,
      "step": 201880
    },
    {
      "epoch": 3.5444793623483557,
      "grad_norm": 0.04553866619285527,
      "learning_rate": 4.206917430110804e-05,
      "loss": 0.8349,
      "step": 201890
    },
    {
      "epoch": 3.544654927228357,
      "grad_norm": 0.06492408687639692,
      "learning_rate": 4.206022040330123e-05,
      "loss": 0.8344,
      "step": 201900
    },
    {
      "epoch": 3.5448304921083587,
      "grad_norm": 0.05534338829534055,
      "learning_rate": 4.2051267252272675e-05,
      "loss": 0.84,
      "step": 201910
    },
    {
      "epoch": 3.54500605698836,
      "grad_norm": 0.04890774708605247,
      "learning_rate": 4.204231484813587e-05,
      "loss": 0.8549,
      "step": 201920
    },
    {
      "epoch": 3.5451816218683616,
      "grad_norm": 0.05011052805679417,
      "learning_rate": 4.203336319100428e-05,
      "loss": 0.8343,
      "step": 201930
    },
    {
      "epoch": 3.5453571867483626,
      "grad_norm": 0.07776652514246181,
      "learning_rate": 4.202441228099122e-05,
      "loss": 0.8404,
      "step": 201940
    },
    {
      "epoch": 3.545532751628364,
      "grad_norm": 0.03941805929930275,
      "learning_rate": 4.201546211821021e-05,
      "loss": 0.8407,
      "step": 201950
    },
    {
      "epoch": 3.5457083165083656,
      "grad_norm": 0.0582557028494856,
      "learning_rate": 4.200651270277454e-05,
      "loss": 0.8411,
      "step": 201960
    },
    {
      "epoch": 3.545883881388367,
      "grad_norm": 0.05569094157416411,
      "learning_rate": 4.199756403479771e-05,
      "loss": 0.8324,
      "step": 201970
    },
    {
      "epoch": 3.5460594462683686,
      "grad_norm": 0.050083648282018806,
      "learning_rate": 4.198861611439303e-05,
      "loss": 0.8374,
      "step": 201980
    },
    {
      "epoch": 3.54623501114837,
      "grad_norm": 0.05361188233807025,
      "learning_rate": 4.197966894167388e-05,
      "loss": 0.8413,
      "step": 201990
    },
    {
      "epoch": 3.5464105760283715,
      "grad_norm": 0.05144418915026631,
      "learning_rate": 4.197072251675369e-05,
      "loss": 0.8354,
      "step": 202000
    },
    {
      "epoch": 3.5465861409083725,
      "grad_norm": 0.06439262003378204,
      "learning_rate": 4.196177683974571e-05,
      "loss": 0.8429,
      "step": 202010
    },
    {
      "epoch": 3.546761705788374,
      "grad_norm": 0.05658721792952366,
      "learning_rate": 4.195283191076336e-05,
      "loss": 0.849,
      "step": 202020
    },
    {
      "epoch": 3.5469372706683755,
      "grad_norm": 0.058842674475488695,
      "learning_rate": 4.1943887729919896e-05,
      "loss": 0.842,
      "step": 202030
    },
    {
      "epoch": 3.547112835548377,
      "grad_norm": 0.04566019387411297,
      "learning_rate": 4.193494429732875e-05,
      "loss": 0.8408,
      "step": 202040
    },
    {
      "epoch": 3.5472884004283785,
      "grad_norm": 0.05562917604068131,
      "learning_rate": 4.192600161310318e-05,
      "loss": 0.8283,
      "step": 202050
    },
    {
      "epoch": 3.5474639653083795,
      "grad_norm": 0.049642852304689324,
      "learning_rate": 4.1917059677356425e-05,
      "loss": 0.8425,
      "step": 202060
    },
    {
      "epoch": 3.547639530188381,
      "grad_norm": 0.04900916811387846,
      "learning_rate": 4.1908118490201916e-05,
      "loss": 0.831,
      "step": 202070
    },
    {
      "epoch": 3.5478150950683824,
      "grad_norm": 0.051460572172552184,
      "learning_rate": 4.189917805175283e-05,
      "loss": 0.8453,
      "step": 202080
    },
    {
      "epoch": 3.547990659948384,
      "grad_norm": 0.05186559451715673,
      "learning_rate": 4.189023836212249e-05,
      "loss": 0.8336,
      "step": 202090
    },
    {
      "epoch": 3.5481662248283854,
      "grad_norm": 0.0835786660714101,
      "learning_rate": 4.188129942142419e-05,
      "loss": 0.8425,
      "step": 202100
    },
    {
      "epoch": 3.548341789708387,
      "grad_norm": 0.05708805642588019,
      "learning_rate": 4.187236122977113e-05,
      "loss": 0.8395,
      "step": 202110
    },
    {
      "epoch": 3.5485173545883884,
      "grad_norm": 0.05223368829452231,
      "learning_rate": 4.186342378727661e-05,
      "loss": 0.8457,
      "step": 202120
    },
    {
      "epoch": 3.5486929194683894,
      "grad_norm": 0.0712990768310104,
      "learning_rate": 4.185448709405378e-05,
      "loss": 0.8406,
      "step": 202130
    },
    {
      "epoch": 3.548868484348391,
      "grad_norm": 0.04459239130009099,
      "learning_rate": 4.1845551150216e-05,
      "loss": 0.8343,
      "step": 202140
    },
    {
      "epoch": 3.5490440492283923,
      "grad_norm": 0.06812340645276509,
      "learning_rate": 4.183661595587638e-05,
      "loss": 0.8363,
      "step": 202150
    },
    {
      "epoch": 3.549219614108394,
      "grad_norm": 0.050801589733177784,
      "learning_rate": 4.1827681511148186e-05,
      "loss": 0.8404,
      "step": 202160
    },
    {
      "epoch": 3.5493951789883953,
      "grad_norm": 0.0873724583652728,
      "learning_rate": 4.1818747816144646e-05,
      "loss": 0.8382,
      "step": 202170
    },
    {
      "epoch": 3.5495707438683963,
      "grad_norm": 0.0427164449413722,
      "learning_rate": 4.180981487097887e-05,
      "loss": 0.847,
      "step": 202180
    },
    {
      "epoch": 3.549746308748398,
      "grad_norm": 0.045833859169462775,
      "learning_rate": 4.1800882675764124e-05,
      "loss": 0.8401,
      "step": 202190
    },
    {
      "epoch": 3.5499218736283993,
      "grad_norm": 0.0596688177222984,
      "learning_rate": 4.1791951230613476e-05,
      "loss": 0.8314,
      "step": 202200
    },
    {
      "epoch": 3.5500974385084008,
      "grad_norm": 0.05856868314200318,
      "learning_rate": 4.178302053564021e-05,
      "loss": 0.8424,
      "step": 202210
    },
    {
      "epoch": 3.5502730033884022,
      "grad_norm": 0.07062133216357847,
      "learning_rate": 4.177409059095743e-05,
      "loss": 0.8365,
      "step": 202220
    },
    {
      "epoch": 3.5504485682684037,
      "grad_norm": 0.04275938854001201,
      "learning_rate": 4.17651613966782e-05,
      "loss": 0.8383,
      "step": 202230
    },
    {
      "epoch": 3.550624133148405,
      "grad_norm": 0.05672602918209526,
      "learning_rate": 4.1756232952915785e-05,
      "loss": 0.838,
      "step": 202240
    },
    {
      "epoch": 3.5507996980284062,
      "grad_norm": 0.045234960715627624,
      "learning_rate": 4.1747305259783236e-05,
      "loss": 0.8382,
      "step": 202250
    },
    {
      "epoch": 3.5509752629084077,
      "grad_norm": 0.04782335218777148,
      "learning_rate": 4.173837831739368e-05,
      "loss": 0.8385,
      "step": 202260
    },
    {
      "epoch": 3.551150827788409,
      "grad_norm": 0.05253011824523701,
      "learning_rate": 4.172945212586026e-05,
      "loss": 0.8469,
      "step": 202270
    },
    {
      "epoch": 3.5513263926684107,
      "grad_norm": 0.05578869191666809,
      "learning_rate": 4.172052668529601e-05,
      "loss": 0.8433,
      "step": 202280
    },
    {
      "epoch": 3.551501957548412,
      "grad_norm": 0.049992607179486725,
      "learning_rate": 4.171160199581408e-05,
      "loss": 0.8398,
      "step": 202290
    },
    {
      "epoch": 3.551677522428413,
      "grad_norm": 0.06656533147827975,
      "learning_rate": 4.1702678057527445e-05,
      "loss": 0.8443,
      "step": 202300
    },
    {
      "epoch": 3.551853087308415,
      "grad_norm": 0.045782272780197676,
      "learning_rate": 4.169375487054932e-05,
      "loss": 0.8432,
      "step": 202310
    },
    {
      "epoch": 3.552028652188416,
      "grad_norm": 0.05271271831102063,
      "learning_rate": 4.168483243499265e-05,
      "loss": 0.8405,
      "step": 202320
    },
    {
      "epoch": 3.5522042170684176,
      "grad_norm": 0.06386151959883696,
      "learning_rate": 4.1675910750970506e-05,
      "loss": 0.8472,
      "step": 202330
    },
    {
      "epoch": 3.552379781948419,
      "grad_norm": 0.05753717583141441,
      "learning_rate": 4.166698981859599e-05,
      "loss": 0.8428,
      "step": 202340
    },
    {
      "epoch": 3.5525553468284206,
      "grad_norm": 0.05829183750975896,
      "learning_rate": 4.1658069637982053e-05,
      "loss": 0.8354,
      "step": 202350
    },
    {
      "epoch": 3.552730911708422,
      "grad_norm": 0.04525614193209865,
      "learning_rate": 4.164915020924177e-05,
      "loss": 0.8356,
      "step": 202360
    },
    {
      "epoch": 3.552906476588423,
      "grad_norm": 0.0554474696914655,
      "learning_rate": 4.164023153248807e-05,
      "loss": 0.8418,
      "step": 202370
    },
    {
      "epoch": 3.5530820414684245,
      "grad_norm": 0.056123977958867304,
      "learning_rate": 4.163131360783407e-05,
      "loss": 0.8387,
      "step": 202380
    },
    {
      "epoch": 3.553257606348426,
      "grad_norm": 0.04682982716980217,
      "learning_rate": 4.162239643539272e-05,
      "loss": 0.8428,
      "step": 202390
    },
    {
      "epoch": 3.5534331712284275,
      "grad_norm": 0.056302710663250925,
      "learning_rate": 4.16134800152769e-05,
      "loss": 0.8443,
      "step": 202400
    },
    {
      "epoch": 3.553608736108429,
      "grad_norm": 0.05353204510118847,
      "learning_rate": 4.160456434759975e-05,
      "loss": 0.8396,
      "step": 202410
    },
    {
      "epoch": 3.55378430098843,
      "grad_norm": 0.043131618332352836,
      "learning_rate": 4.159564943247412e-05,
      "loss": 0.839,
      "step": 202420
    },
    {
      "epoch": 3.553959865868432,
      "grad_norm": 0.05099243352552083,
      "learning_rate": 4.1586735270012984e-05,
      "loss": 0.8477,
      "step": 202430
    },
    {
      "epoch": 3.554135430748433,
      "grad_norm": 0.05756520207757817,
      "learning_rate": 4.157782186032934e-05,
      "loss": 0.8395,
      "step": 202440
    },
    {
      "epoch": 3.5543109956284344,
      "grad_norm": 0.04849570403860387,
      "learning_rate": 4.1568909203536046e-05,
      "loss": 0.8358,
      "step": 202450
    },
    {
      "epoch": 3.554486560508436,
      "grad_norm": 0.07681580272349718,
      "learning_rate": 4.155999729974611e-05,
      "loss": 0.8446,
      "step": 202460
    },
    {
      "epoch": 3.5546621253884374,
      "grad_norm": 0.05484062510705913,
      "learning_rate": 4.155108614907231e-05,
      "loss": 0.8366,
      "step": 202470
    },
    {
      "epoch": 3.554837690268439,
      "grad_norm": 0.05692925950926333,
      "learning_rate": 4.154217575162772e-05,
      "loss": 0.8452,
      "step": 202480
    },
    {
      "epoch": 3.55501325514844,
      "grad_norm": 0.06384814514583219,
      "learning_rate": 4.153326610752513e-05,
      "loss": 0.8385,
      "step": 202490
    },
    {
      "epoch": 3.5551888200284414,
      "grad_norm": 0.059933017677313835,
      "learning_rate": 4.1524357216877454e-05,
      "loss": 0.8377,
      "step": 202500
    },
    {
      "epoch": 3.555364384908443,
      "grad_norm": 0.05190822824900191,
      "learning_rate": 4.151544907979759e-05,
      "loss": 0.8399,
      "step": 202510
    },
    {
      "epoch": 3.5555399497884443,
      "grad_norm": 0.05897556912746814,
      "learning_rate": 4.150654169639836e-05,
      "loss": 0.8371,
      "step": 202520
    },
    {
      "epoch": 3.555715514668446,
      "grad_norm": 0.055389475768358225,
      "learning_rate": 4.149763506679269e-05,
      "loss": 0.8414,
      "step": 202530
    },
    {
      "epoch": 3.555891079548447,
      "grad_norm": 0.06049300130611548,
      "learning_rate": 4.148872919109332e-05,
      "loss": 0.8409,
      "step": 202540
    },
    {
      "epoch": 3.5560666444284488,
      "grad_norm": 0.06258305308499724,
      "learning_rate": 4.1479824069413206e-05,
      "loss": 0.8342,
      "step": 202550
    },
    {
      "epoch": 3.55624220930845,
      "grad_norm": 0.04925570555700529,
      "learning_rate": 4.147091970186515e-05,
      "loss": 0.8469,
      "step": 202560
    },
    {
      "epoch": 3.5564177741884513,
      "grad_norm": 0.05494974008369435,
      "learning_rate": 4.146201608856187e-05,
      "loss": 0.8426,
      "step": 202570
    },
    {
      "epoch": 3.5565933390684528,
      "grad_norm": 0.04577489898155138,
      "learning_rate": 4.145311322961633e-05,
      "loss": 0.844,
      "step": 202580
    },
    {
      "epoch": 3.5567689039484542,
      "grad_norm": 0.055549322653693736,
      "learning_rate": 4.144421112514122e-05,
      "loss": 0.8392,
      "step": 202590
    },
    {
      "epoch": 3.5569444688284557,
      "grad_norm": 0.05304639264826229,
      "learning_rate": 4.143530977524936e-05,
      "loss": 0.8421,
      "step": 202600
    },
    {
      "epoch": 3.5571200337084568,
      "grad_norm": 0.062293863860261825,
      "learning_rate": 4.14264091800536e-05,
      "loss": 0.8407,
      "step": 202610
    },
    {
      "epoch": 3.5572955985884582,
      "grad_norm": 0.05422314390967318,
      "learning_rate": 4.14175093396666e-05,
      "loss": 0.8388,
      "step": 202620
    },
    {
      "epoch": 3.5574711634684597,
      "grad_norm": 0.054306372736674574,
      "learning_rate": 4.140861025420121e-05,
      "loss": 0.8431,
      "step": 202630
    },
    {
      "epoch": 3.557646728348461,
      "grad_norm": 0.06211548648470621,
      "learning_rate": 4.139971192377009e-05,
      "loss": 0.8412,
      "step": 202640
    },
    {
      "epoch": 3.5578222932284627,
      "grad_norm": 0.04950475590343682,
      "learning_rate": 4.13908143484861e-05,
      "loss": 0.8292,
      "step": 202650
    },
    {
      "epoch": 3.557997858108464,
      "grad_norm": 0.058567028664314374,
      "learning_rate": 4.138191752846188e-05,
      "loss": 0.8333,
      "step": 202660
    },
    {
      "epoch": 3.5581734229884656,
      "grad_norm": 0.07073776040709467,
      "learning_rate": 4.1373021463810206e-05,
      "loss": 0.8404,
      "step": 202670
    },
    {
      "epoch": 3.5583489878684667,
      "grad_norm": 0.04699363637424547,
      "learning_rate": 4.13641261546438e-05,
      "loss": 0.8385,
      "step": 202680
    },
    {
      "epoch": 3.558524552748468,
      "grad_norm": 0.04920509816073665,
      "learning_rate": 4.135523160107531e-05,
      "loss": 0.8406,
      "step": 202690
    },
    {
      "epoch": 3.5587001176284696,
      "grad_norm": 0.054469556173625554,
      "learning_rate": 4.1346337803217514e-05,
      "loss": 0.8457,
      "step": 202700
    },
    {
      "epoch": 3.558875682508471,
      "grad_norm": 0.047264725350148286,
      "learning_rate": 4.133744476118296e-05,
      "loss": 0.8453,
      "step": 202710
    },
    {
      "epoch": 3.5590512473884726,
      "grad_norm": 0.05935997797528456,
      "learning_rate": 4.1328552475084495e-05,
      "loss": 0.8525,
      "step": 202720
    },
    {
      "epoch": 3.5592268122684736,
      "grad_norm": 0.04822267649608381,
      "learning_rate": 4.131966094503471e-05,
      "loss": 0.834,
      "step": 202730
    },
    {
      "epoch": 3.559402377148475,
      "grad_norm": 0.05441704519615671,
      "learning_rate": 4.131077017114619e-05,
      "loss": 0.8444,
      "step": 202740
    },
    {
      "epoch": 3.5595779420284765,
      "grad_norm": 0.05563811759070384,
      "learning_rate": 4.1301880153531705e-05,
      "loss": 0.8441,
      "step": 202750
    },
    {
      "epoch": 3.559753506908478,
      "grad_norm": 0.09059775388365314,
      "learning_rate": 4.129299089230382e-05,
      "loss": 0.8431,
      "step": 202760
    },
    {
      "epoch": 3.5599290717884795,
      "grad_norm": 0.04860350001193106,
      "learning_rate": 4.128410238757518e-05,
      "loss": 0.8432,
      "step": 202770
    },
    {
      "epoch": 3.560104636668481,
      "grad_norm": 0.05666766351577844,
      "learning_rate": 4.127521463945844e-05,
      "loss": 0.8343,
      "step": 202780
    },
    {
      "epoch": 3.5602802015484825,
      "grad_norm": 0.04917878503774935,
      "learning_rate": 4.126632764806615e-05,
      "loss": 0.8378,
      "step": 202790
    },
    {
      "epoch": 3.5604557664284835,
      "grad_norm": 0.049211717103859555,
      "learning_rate": 4.125744141351097e-05,
      "loss": 0.8386,
      "step": 202800
    },
    {
      "epoch": 3.560631331308485,
      "grad_norm": 0.06303250754395291,
      "learning_rate": 4.124855593590537e-05,
      "loss": 0.839,
      "step": 202810
    },
    {
      "epoch": 3.5608068961884864,
      "grad_norm": 0.07174008101092028,
      "learning_rate": 4.123967121536211e-05,
      "loss": 0.8424,
      "step": 202820
    },
    {
      "epoch": 3.560982461068488,
      "grad_norm": 0.054384886721889916,
      "learning_rate": 4.1230787251993626e-05,
      "loss": 0.8364,
      "step": 202830
    },
    {
      "epoch": 3.5611580259484894,
      "grad_norm": 0.05212891562461495,
      "learning_rate": 4.1221904045912516e-05,
      "loss": 0.8456,
      "step": 202840
    },
    {
      "epoch": 3.5613335908284904,
      "grad_norm": 0.04443412864739048,
      "learning_rate": 4.1213021597231375e-05,
      "loss": 0.8458,
      "step": 202850
    },
    {
      "epoch": 3.561509155708492,
      "grad_norm": 0.05714945114234988,
      "learning_rate": 4.1204139906062674e-05,
      "loss": 0.8338,
      "step": 202860
    },
    {
      "epoch": 3.5616847205884934,
      "grad_norm": 0.049674807923311255,
      "learning_rate": 4.119525897251902e-05,
      "loss": 0.843,
      "step": 202870
    },
    {
      "epoch": 3.561860285468495,
      "grad_norm": 0.05106627280403374,
      "learning_rate": 4.118637879671283e-05,
      "loss": 0.8439,
      "step": 202880
    },
    {
      "epoch": 3.5620358503484963,
      "grad_norm": 0.07058066402984899,
      "learning_rate": 4.117749937875674e-05,
      "loss": 0.8432,
      "step": 202890
    },
    {
      "epoch": 3.562211415228498,
      "grad_norm": 0.04403428552019013,
      "learning_rate": 4.1168620718763196e-05,
      "loss": 0.8365,
      "step": 202900
    },
    {
      "epoch": 3.5623869801084993,
      "grad_norm": 0.08499784350790578,
      "learning_rate": 4.115974281684464e-05,
      "loss": 0.8413,
      "step": 202910
    },
    {
      "epoch": 3.5625625449885003,
      "grad_norm": 0.05276800048834633,
      "learning_rate": 4.1150865673113675e-05,
      "loss": 0.8335,
      "step": 202920
    },
    {
      "epoch": 3.562738109868502,
      "grad_norm": 0.06022356543608555,
      "learning_rate": 4.1141989287682675e-05,
      "loss": 0.8345,
      "step": 202930
    },
    {
      "epoch": 3.5629136747485033,
      "grad_norm": 0.043723620453777624,
      "learning_rate": 4.113311366066414e-05,
      "loss": 0.8446,
      "step": 202940
    },
    {
      "epoch": 3.5630892396285048,
      "grad_norm": 0.05023270590943099,
      "learning_rate": 4.1124238792170575e-05,
      "loss": 0.8377,
      "step": 202950
    },
    {
      "epoch": 3.5632648045085062,
      "grad_norm": 0.06960507605648897,
      "learning_rate": 4.111536468231433e-05,
      "loss": 0.8437,
      "step": 202960
    },
    {
      "epoch": 3.5634403693885073,
      "grad_norm": 0.05835610732478577,
      "learning_rate": 4.110649133120793e-05,
      "loss": 0.8376,
      "step": 202970
    },
    {
      "epoch": 3.563615934268509,
      "grad_norm": 0.0562142610590623,
      "learning_rate": 4.109761873896369e-05,
      "loss": 0.8425,
      "step": 202980
    },
    {
      "epoch": 3.5637914991485102,
      "grad_norm": 0.058810379577349386,
      "learning_rate": 4.1088746905694185e-05,
      "loss": 0.8454,
      "step": 202990
    },
    {
      "epoch": 3.5639670640285117,
      "grad_norm": 0.0552412548479127,
      "learning_rate": 4.1079875831511696e-05,
      "loss": 0.8401,
      "step": 203000
    },
    {
      "epoch": 3.564142628908513,
      "grad_norm": 0.05018610659422982,
      "learning_rate": 4.1071005516528654e-05,
      "loss": 0.8392,
      "step": 203010
    },
    {
      "epoch": 3.5643181937885147,
      "grad_norm": 0.048676245628602635,
      "learning_rate": 4.106213596085751e-05,
      "loss": 0.8408,
      "step": 203020
    },
    {
      "epoch": 3.564493758668516,
      "grad_norm": 0.043905972137325756,
      "learning_rate": 4.1053267164610545e-05,
      "loss": 0.8393,
      "step": 203030
    },
    {
      "epoch": 3.564669323548517,
      "grad_norm": 0.04926263508264267,
      "learning_rate": 4.104439912790021e-05,
      "loss": 0.8419,
      "step": 203040
    },
    {
      "epoch": 3.5648448884285187,
      "grad_norm": 0.0647272489065477,
      "learning_rate": 4.1035531850838756e-05,
      "loss": 0.844,
      "step": 203050
    },
    {
      "epoch": 3.56502045330852,
      "grad_norm": 0.052857568552382245,
      "learning_rate": 4.102666533353869e-05,
      "loss": 0.8431,
      "step": 203060
    },
    {
      "epoch": 3.5651960181885216,
      "grad_norm": 0.05514856923540912,
      "learning_rate": 4.101779957611225e-05,
      "loss": 0.8379,
      "step": 203070
    },
    {
      "epoch": 3.565371583068523,
      "grad_norm": 0.04719526684672021,
      "learning_rate": 4.100893457867172e-05,
      "loss": 0.8413,
      "step": 203080
    },
    {
      "epoch": 3.565547147948524,
      "grad_norm": 0.060341597954230004,
      "learning_rate": 4.100007034132956e-05,
      "loss": 0.8388,
      "step": 203090
    },
    {
      "epoch": 3.565722712828526,
      "grad_norm": 0.062367611970927446,
      "learning_rate": 4.0991206864197966e-05,
      "loss": 0.848,
      "step": 203100
    },
    {
      "epoch": 3.565898277708527,
      "grad_norm": 0.056058023677040374,
      "learning_rate": 4.098234414738929e-05,
      "loss": 0.8359,
      "step": 203110
    },
    {
      "epoch": 3.5660738425885286,
      "grad_norm": 0.04424588217058093,
      "learning_rate": 4.0973482191015844e-05,
      "loss": 0.8325,
      "step": 203120
    },
    {
      "epoch": 3.56624940746853,
      "grad_norm": 0.053134280099364946,
      "learning_rate": 4.096462099518986e-05,
      "loss": 0.8352,
      "step": 203130
    },
    {
      "epoch": 3.5664249723485315,
      "grad_norm": 0.05051734508317647,
      "learning_rate": 4.0955760560023655e-05,
      "loss": 0.8366,
      "step": 203140
    },
    {
      "epoch": 3.566600537228533,
      "grad_norm": 0.0657756387204699,
      "learning_rate": 4.0946900885629405e-05,
      "loss": 0.8392,
      "step": 203150
    },
    {
      "epoch": 3.566776102108534,
      "grad_norm": 0.06267971219693,
      "learning_rate": 4.09380419721195e-05,
      "loss": 0.8425,
      "step": 203160
    },
    {
      "epoch": 3.5669516669885355,
      "grad_norm": 0.05531373696262727,
      "learning_rate": 4.0929183819606076e-05,
      "loss": 0.8352,
      "step": 203170
    },
    {
      "epoch": 3.567127231868537,
      "grad_norm": 0.059137362197684694,
      "learning_rate": 4.092032642820141e-05,
      "loss": 0.8329,
      "step": 203180
    },
    {
      "epoch": 3.5673027967485385,
      "grad_norm": 0.04997041305769301,
      "learning_rate": 4.091146979801776e-05,
      "loss": 0.8373,
      "step": 203190
    },
    {
      "epoch": 3.56747836162854,
      "grad_norm": 0.05211164184956024,
      "learning_rate": 4.0902613929167265e-05,
      "loss": 0.844,
      "step": 203200
    },
    {
      "epoch": 3.567653926508541,
      "grad_norm": 0.05215929272867722,
      "learning_rate": 4.0893758821762204e-05,
      "loss": 0.839,
      "step": 203210
    },
    {
      "epoch": 3.567829491388543,
      "grad_norm": 0.05592047709582128,
      "learning_rate": 4.088490447591467e-05,
      "loss": 0.8407,
      "step": 203220
    },
    {
      "epoch": 3.568005056268544,
      "grad_norm": 0.04441345397438826,
      "learning_rate": 4.0876050891736984e-05,
      "loss": 0.8412,
      "step": 203230
    },
    {
      "epoch": 3.5681806211485454,
      "grad_norm": 0.0628030923881106,
      "learning_rate": 4.0867198069341256e-05,
      "loss": 0.8393,
      "step": 203240
    },
    {
      "epoch": 3.568356186028547,
      "grad_norm": 0.061767690831085534,
      "learning_rate": 4.085834600883959e-05,
      "loss": 0.8412,
      "step": 203250
    },
    {
      "epoch": 3.5685317509085484,
      "grad_norm": 0.05543397171612809,
      "learning_rate": 4.0849494710344265e-05,
      "loss": 0.8417,
      "step": 203260
    },
    {
      "epoch": 3.56870731578855,
      "grad_norm": 0.0473728266300272,
      "learning_rate": 4.0840644173967344e-05,
      "loss": 0.8415,
      "step": 203270
    },
    {
      "epoch": 3.568882880668551,
      "grad_norm": 0.05212631742208889,
      "learning_rate": 4.0831794399820986e-05,
      "loss": 0.8441,
      "step": 203280
    },
    {
      "epoch": 3.5690584455485523,
      "grad_norm": 0.05894944561356138,
      "learning_rate": 4.0822945388017365e-05,
      "loss": 0.8397,
      "step": 203290
    },
    {
      "epoch": 3.569234010428554,
      "grad_norm": 0.05168372537498453,
      "learning_rate": 4.0814097138668514e-05,
      "loss": 0.8453,
      "step": 203300
    },
    {
      "epoch": 3.5694095753085553,
      "grad_norm": 0.06392651958885176,
      "learning_rate": 4.080524965188663e-05,
      "loss": 0.8395,
      "step": 203310
    },
    {
      "epoch": 3.5695851401885568,
      "grad_norm": 0.05519640541424056,
      "learning_rate": 4.079640292778369e-05,
      "loss": 0.8366,
      "step": 203320
    },
    {
      "epoch": 3.5697607050685582,
      "grad_norm": 0.06269251959870974,
      "learning_rate": 4.0787556966471933e-05,
      "loss": 0.8494,
      "step": 203330
    },
    {
      "epoch": 3.5699362699485597,
      "grad_norm": 0.04676732525232947,
      "learning_rate": 4.077871176806335e-05,
      "loss": 0.8453,
      "step": 203340
    },
    {
      "epoch": 3.5701118348285608,
      "grad_norm": 0.05692040134585112,
      "learning_rate": 4.0769867332670015e-05,
      "loss": 0.8368,
      "step": 203350
    },
    {
      "epoch": 3.5702873997085622,
      "grad_norm": 0.05874838279287591,
      "learning_rate": 4.0761023660404045e-05,
      "loss": 0.848,
      "step": 203360
    },
    {
      "epoch": 3.5704629645885637,
      "grad_norm": 0.05318390010441011,
      "learning_rate": 4.075218075137742e-05,
      "loss": 0.8397,
      "step": 203370
    },
    {
      "epoch": 3.570638529468565,
      "grad_norm": 0.062118498454934126,
      "learning_rate": 4.074333860570223e-05,
      "loss": 0.841,
      "step": 203380
    },
    {
      "epoch": 3.5708140943485667,
      "grad_norm": 0.06786988413455702,
      "learning_rate": 4.0734497223490435e-05,
      "loss": 0.8432,
      "step": 203390
    },
    {
      "epoch": 3.5709896592285677,
      "grad_norm": 0.0472913827684817,
      "learning_rate": 4.072565660485418e-05,
      "loss": 0.8451,
      "step": 203400
    },
    {
      "epoch": 3.571165224108569,
      "grad_norm": 0.051010779853757245,
      "learning_rate": 4.071681674990539e-05,
      "loss": 0.8442,
      "step": 203410
    },
    {
      "epoch": 3.5713407889885707,
      "grad_norm": 0.05559682996026318,
      "learning_rate": 4.070797765875603e-05,
      "loss": 0.8442,
      "step": 203420
    },
    {
      "epoch": 3.571516353868572,
      "grad_norm": 0.05451190723504474,
      "learning_rate": 4.069913933151821e-05,
      "loss": 0.8464,
      "step": 203430
    },
    {
      "epoch": 3.5716919187485736,
      "grad_norm": 0.05223389657143302,
      "learning_rate": 4.069030176830382e-05,
      "loss": 0.8475,
      "step": 203440
    },
    {
      "epoch": 3.571867483628575,
      "grad_norm": 0.06565356224027362,
      "learning_rate": 4.068146496922487e-05,
      "loss": 0.8389,
      "step": 203450
    },
    {
      "epoch": 3.5720430485085766,
      "grad_norm": 0.056951755640559855,
      "learning_rate": 4.0672628934393355e-05,
      "loss": 0.8267,
      "step": 203460
    },
    {
      "epoch": 3.5722186133885776,
      "grad_norm": 0.042842283214547396,
      "learning_rate": 4.066379366392114e-05,
      "loss": 0.834,
      "step": 203470
    },
    {
      "epoch": 3.572394178268579,
      "grad_norm": 0.06392268482816076,
      "learning_rate": 4.065495915792028e-05,
      "loss": 0.8439,
      "step": 203480
    },
    {
      "epoch": 3.5725697431485806,
      "grad_norm": 0.054664258166242305,
      "learning_rate": 4.0646125416502565e-05,
      "loss": 0.8406,
      "step": 203490
    },
    {
      "epoch": 3.572745308028582,
      "grad_norm": 0.06577153496965708,
      "learning_rate": 4.063729243978008e-05,
      "loss": 0.8441,
      "step": 203500
    },
    {
      "epoch": 3.5729208729085835,
      "grad_norm": 0.048750923422640764,
      "learning_rate": 4.062846022786464e-05,
      "loss": 0.8404,
      "step": 203510
    },
    {
      "epoch": 3.5730964377885845,
      "grad_norm": 0.06080642207217739,
      "learning_rate": 4.0619628780868185e-05,
      "loss": 0.845,
      "step": 203520
    },
    {
      "epoch": 3.573272002668586,
      "grad_norm": 0.055595359858964584,
      "learning_rate": 4.061079809890262e-05,
      "loss": 0.8386,
      "step": 203530
    },
    {
      "epoch": 3.5734475675485875,
      "grad_norm": 0.05639471008767674,
      "learning_rate": 4.06019681820798e-05,
      "loss": 0.8422,
      "step": 203540
    },
    {
      "epoch": 3.573623132428589,
      "grad_norm": 0.047287642238419955,
      "learning_rate": 4.059313903051165e-05,
      "loss": 0.8388,
      "step": 203550
    },
    {
      "epoch": 3.5737986973085905,
      "grad_norm": 0.07907564200901977,
      "learning_rate": 4.0584310644309934e-05,
      "loss": 0.8353,
      "step": 203560
    },
    {
      "epoch": 3.573974262188592,
      "grad_norm": 0.06610837939658015,
      "learning_rate": 4.057548302358665e-05,
      "loss": 0.8347,
      "step": 203570
    },
    {
      "epoch": 3.5741498270685934,
      "grad_norm": 0.053973978678530585,
      "learning_rate": 4.0566656168453574e-05,
      "loss": 0.847,
      "step": 203580
    },
    {
      "epoch": 3.5743253919485944,
      "grad_norm": 0.051973333485142006,
      "learning_rate": 4.055783007902248e-05,
      "loss": 0.837,
      "step": 203590
    },
    {
      "epoch": 3.574500956828596,
      "grad_norm": 0.0702230287905264,
      "learning_rate": 4.0549004755405326e-05,
      "loss": 0.8336,
      "step": 203600
    },
    {
      "epoch": 3.5746765217085974,
      "grad_norm": 0.044784239291725994,
      "learning_rate": 4.054018019771384e-05,
      "loss": 0.8375,
      "step": 203610
    },
    {
      "epoch": 3.574852086588599,
      "grad_norm": 0.05503499102236917,
      "learning_rate": 4.053135640605985e-05,
      "loss": 0.8394,
      "step": 203620
    },
    {
      "epoch": 3.5750276514686004,
      "grad_norm": 0.055131560551788736,
      "learning_rate": 4.052253338055521e-05,
      "loss": 0.8363,
      "step": 203630
    },
    {
      "epoch": 3.5752032163486014,
      "grad_norm": 0.05005782045798044,
      "learning_rate": 4.051371112131161e-05,
      "loss": 0.8473,
      "step": 203640
    },
    {
      "epoch": 3.575378781228603,
      "grad_norm": 0.05750058981820543,
      "learning_rate": 4.0504889628440926e-05,
      "loss": 0.8405,
      "step": 203650
    },
    {
      "epoch": 3.5755543461086043,
      "grad_norm": 0.0719880241557586,
      "learning_rate": 4.0496068902054824e-05,
      "loss": 0.8374,
      "step": 203660
    },
    {
      "epoch": 3.575729910988606,
      "grad_norm": 0.059919521174012134,
      "learning_rate": 4.048724894226518e-05,
      "loss": 0.8389,
      "step": 203670
    },
    {
      "epoch": 3.5759054758686073,
      "grad_norm": 0.08135570151373045,
      "learning_rate": 4.047842974918366e-05,
      "loss": 0.8403,
      "step": 203680
    },
    {
      "epoch": 3.5760810407486088,
      "grad_norm": 0.05609337692813264,
      "learning_rate": 4.046961132292204e-05,
      "loss": 0.8454,
      "step": 203690
    },
    {
      "epoch": 3.5762566056286103,
      "grad_norm": 0.04714103623834301,
      "learning_rate": 4.046079366359207e-05,
      "loss": 0.841,
      "step": 203700
    },
    {
      "epoch": 3.5764321705086113,
      "grad_norm": 0.04653102130308885,
      "learning_rate": 4.0451976771305416e-05,
      "loss": 0.8438,
      "step": 203710
    },
    {
      "epoch": 3.5766077353886128,
      "grad_norm": 0.056402895055871656,
      "learning_rate": 4.044316064617385e-05,
      "loss": 0.8434,
      "step": 203720
    },
    {
      "epoch": 3.5767833002686142,
      "grad_norm": 0.06184846607343485,
      "learning_rate": 4.043434528830899e-05,
      "loss": 0.8458,
      "step": 203730
    },
    {
      "epoch": 3.5769588651486157,
      "grad_norm": 0.056935364931794324,
      "learning_rate": 4.042553069782264e-05,
      "loss": 0.8356,
      "step": 203740
    },
    {
      "epoch": 3.577134430028617,
      "grad_norm": 0.06603122215129295,
      "learning_rate": 4.041671687482643e-05,
      "loss": 0.8448,
      "step": 203750
    },
    {
      "epoch": 3.5773099949086182,
      "grad_norm": 0.055494467210946447,
      "learning_rate": 4.040790381943195e-05,
      "loss": 0.8318,
      "step": 203760
    },
    {
      "epoch": 3.57748555978862,
      "grad_norm": 0.04986798167969904,
      "learning_rate": 4.0399091531751014e-05,
      "loss": 0.8438,
      "step": 203770
    },
    {
      "epoch": 3.577661124668621,
      "grad_norm": 0.06696173068089792,
      "learning_rate": 4.039028001189517e-05,
      "loss": 0.8423,
      "step": 203780
    },
    {
      "epoch": 3.5778366895486227,
      "grad_norm": 0.05688598677536515,
      "learning_rate": 4.03814692599761e-05,
      "loss": 0.8451,
      "step": 203790
    },
    {
      "epoch": 3.578012254428624,
      "grad_norm": 0.04818910663596762,
      "learning_rate": 4.0372659276105455e-05,
      "loss": 0.8429,
      "step": 203800
    },
    {
      "epoch": 3.5781878193086256,
      "grad_norm": 0.044793666118685765,
      "learning_rate": 4.036385006039481e-05,
      "loss": 0.8395,
      "step": 203810
    },
    {
      "epoch": 3.578363384188627,
      "grad_norm": 0.06292232257896128,
      "learning_rate": 4.0355041612955834e-05,
      "loss": 0.8428,
      "step": 203820
    },
    {
      "epoch": 3.578538949068628,
      "grad_norm": 0.06377873030263963,
      "learning_rate": 4.034623393390004e-05,
      "loss": 0.8427,
      "step": 203830
    },
    {
      "epoch": 3.5787145139486296,
      "grad_norm": 0.05041851045295028,
      "learning_rate": 4.033742702333915e-05,
      "loss": 0.8331,
      "step": 203840
    },
    {
      "epoch": 3.578890078828631,
      "grad_norm": 0.050458019741886115,
      "learning_rate": 4.032862088138465e-05,
      "loss": 0.8444,
      "step": 203850
    },
    {
      "epoch": 3.5790656437086326,
      "grad_norm": 0.05887360930801073,
      "learning_rate": 4.0319815508148164e-05,
      "loss": 0.8371,
      "step": 203860
    },
    {
      "epoch": 3.579241208588634,
      "grad_norm": 0.050189886463340355,
      "learning_rate": 4.031101090374126e-05,
      "loss": 0.833,
      "step": 203870
    },
    {
      "epoch": 3.579416773468635,
      "grad_norm": 0.0525093874906584,
      "learning_rate": 4.0302207068275464e-05,
      "loss": 0.8388,
      "step": 203880
    },
    {
      "epoch": 3.579592338348637,
      "grad_norm": 0.05798603792512317,
      "learning_rate": 4.029340400186236e-05,
      "loss": 0.8412,
      "step": 203890
    },
    {
      "epoch": 3.579767903228638,
      "grad_norm": 0.06048998231733206,
      "learning_rate": 4.0284601704613403e-05,
      "loss": 0.8389,
      "step": 203900
    },
    {
      "epoch": 3.5799434681086395,
      "grad_norm": 0.06215780172567295,
      "learning_rate": 4.0275800176640235e-05,
      "loss": 0.8358,
      "step": 203910
    },
    {
      "epoch": 3.580119032988641,
      "grad_norm": 0.06516999417498093,
      "learning_rate": 4.026699941805432e-05,
      "loss": 0.8448,
      "step": 203920
    },
    {
      "epoch": 3.5802945978686425,
      "grad_norm": 0.04535283128208282,
      "learning_rate": 4.0258199428967094e-05,
      "loss": 0.845,
      "step": 203930
    },
    {
      "epoch": 3.580470162748644,
      "grad_norm": 0.04622808292374348,
      "learning_rate": 4.024940020949019e-05,
      "loss": 0.8433,
      "step": 203940
    },
    {
      "epoch": 3.580645727628645,
      "grad_norm": 0.04712096460423334,
      "learning_rate": 4.0240601759735e-05,
      "loss": 0.8521,
      "step": 203950
    },
    {
      "epoch": 3.5808212925086464,
      "grad_norm": 0.043223905466618695,
      "learning_rate": 4.023180407981302e-05,
      "loss": 0.8384,
      "step": 203960
    },
    {
      "epoch": 3.580996857388648,
      "grad_norm": 0.050162278922400286,
      "learning_rate": 4.022300716983576e-05,
      "loss": 0.8356,
      "step": 203970
    },
    {
      "epoch": 3.5811724222686494,
      "grad_norm": 0.058472085314676944,
      "learning_rate": 4.021421102991463e-05,
      "loss": 0.8412,
      "step": 203980
    },
    {
      "epoch": 3.581347987148651,
      "grad_norm": 0.053769121019810025,
      "learning_rate": 4.0205415660161115e-05,
      "loss": 0.8451,
      "step": 203990
    },
    {
      "epoch": 3.581523552028652,
      "grad_norm": 0.043511421998256056,
      "learning_rate": 4.0196621060686575e-05,
      "loss": 0.8448,
      "step": 204000
    },
    {
      "epoch": 3.581699116908654,
      "grad_norm": 0.05447542808550285,
      "learning_rate": 4.018782723160256e-05,
      "loss": 0.8409,
      "step": 204010
    },
    {
      "epoch": 3.581874681788655,
      "grad_norm": 0.0475065970265568,
      "learning_rate": 4.01790341730204e-05,
      "loss": 0.8373,
      "step": 204020
    },
    {
      "epoch": 3.5820502466686563,
      "grad_norm": 0.06647503068357817,
      "learning_rate": 4.017024188505154e-05,
      "loss": 0.8356,
      "step": 204030
    },
    {
      "epoch": 3.582225811548658,
      "grad_norm": 0.06291575588216783,
      "learning_rate": 4.016145036780739e-05,
      "loss": 0.8366,
      "step": 204040
    },
    {
      "epoch": 3.5824013764286593,
      "grad_norm": 0.05716841535008424,
      "learning_rate": 4.0152659621399295e-05,
      "loss": 0.8436,
      "step": 204050
    },
    {
      "epoch": 3.582576941308661,
      "grad_norm": 0.06972153520253196,
      "learning_rate": 4.014386964593871e-05,
      "loss": 0.8428,
      "step": 204060
    },
    {
      "epoch": 3.582752506188662,
      "grad_norm": 0.04489099078782428,
      "learning_rate": 4.013508044153688e-05,
      "loss": 0.8367,
      "step": 204070
    },
    {
      "epoch": 3.5829280710686633,
      "grad_norm": 0.0455980628800626,
      "learning_rate": 4.012629200830532e-05,
      "loss": 0.8515,
      "step": 204080
    },
    {
      "epoch": 3.5831036359486648,
      "grad_norm": 0.05969502702524924,
      "learning_rate": 4.0117504346355304e-05,
      "loss": 0.8417,
      "step": 204090
    },
    {
      "epoch": 3.5832792008286662,
      "grad_norm": 0.04379808265284023,
      "learning_rate": 4.01087174557981e-05,
      "loss": 0.8415,
      "step": 204100
    },
    {
      "epoch": 3.5834547657086677,
      "grad_norm": 0.056597865405464055,
      "learning_rate": 4.0099931336745196e-05,
      "loss": 0.8434,
      "step": 204110
    },
    {
      "epoch": 3.583630330588669,
      "grad_norm": 0.06699591871944781,
      "learning_rate": 4.0091145989307794e-05,
      "loss": 0.8335,
      "step": 204120
    },
    {
      "epoch": 3.5838058954686707,
      "grad_norm": 0.04555225835889251,
      "learning_rate": 4.0082361413597244e-05,
      "loss": 0.8371,
      "step": 204130
    },
    {
      "epoch": 3.5839814603486717,
      "grad_norm": 0.060080464240844025,
      "learning_rate": 4.0073577609724896e-05,
      "loss": 0.8377,
      "step": 204140
    },
    {
      "epoch": 3.584157025228673,
      "grad_norm": 0.04812101164596821,
      "learning_rate": 4.006479457780195e-05,
      "loss": 0.8457,
      "step": 204150
    },
    {
      "epoch": 3.5843325901086747,
      "grad_norm": 0.0489332553459495,
      "learning_rate": 4.005601231793977e-05,
      "loss": 0.8361,
      "step": 204160
    },
    {
      "epoch": 3.584508154988676,
      "grad_norm": 0.04835316200155569,
      "learning_rate": 4.004723083024952e-05,
      "loss": 0.8318,
      "step": 204170
    },
    {
      "epoch": 3.5846837198686776,
      "grad_norm": 0.16322349974429587,
      "learning_rate": 4.003845011484262e-05,
      "loss": 0.8477,
      "step": 204180
    },
    {
      "epoch": 3.5848592847486787,
      "grad_norm": 0.045349477365258024,
      "learning_rate": 4.002967017183019e-05,
      "loss": 0.85,
      "step": 204190
    },
    {
      "epoch": 3.58503484962868,
      "grad_norm": 0.051186666214025685,
      "learning_rate": 4.0020891001323544e-05,
      "loss": 0.842,
      "step": 204200
    },
    {
      "epoch": 3.5852104145086816,
      "grad_norm": 0.03750551568835322,
      "learning_rate": 4.001211260343391e-05,
      "loss": 0.8466,
      "step": 204210
    },
    {
      "epoch": 3.585385979388683,
      "grad_norm": 0.04556284580972863,
      "learning_rate": 4.0003334978272466e-05,
      "loss": 0.8399,
      "step": 204220
    },
    {
      "epoch": 3.5855615442686846,
      "grad_norm": 0.046865219660420246,
      "learning_rate": 3.9994558125950496e-05,
      "loss": 0.8429,
      "step": 204230
    },
    {
      "epoch": 3.585737109148686,
      "grad_norm": 0.04456512067272074,
      "learning_rate": 3.99857820465791e-05,
      "loss": 0.8413,
      "step": 204240
    },
    {
      "epoch": 3.5859126740286875,
      "grad_norm": 0.04926872852726842,
      "learning_rate": 3.997700674026961e-05,
      "loss": 0.8298,
      "step": 204250
    },
    {
      "epoch": 3.5860882389086886,
      "grad_norm": 0.0496480466769681,
      "learning_rate": 3.9968232207133134e-05,
      "loss": 0.8397,
      "step": 204260
    },
    {
      "epoch": 3.58626380378869,
      "grad_norm": 0.04821883830851315,
      "learning_rate": 3.995945844728078e-05,
      "loss": 0.8435,
      "step": 204270
    },
    {
      "epoch": 3.5864393686686915,
      "grad_norm": 0.05491757557960276,
      "learning_rate": 3.995068546082386e-05,
      "loss": 0.8407,
      "step": 204280
    },
    {
      "epoch": 3.586614933548693,
      "grad_norm": 0.05089817973005301,
      "learning_rate": 3.994191324787342e-05,
      "loss": 0.8334,
      "step": 204290
    },
    {
      "epoch": 3.5867904984286945,
      "grad_norm": 0.06961722959406363,
      "learning_rate": 3.9933141808540644e-05,
      "loss": 0.8442,
      "step": 204300
    },
    {
      "epoch": 3.5869660633086955,
      "grad_norm": 0.04321426585939357,
      "learning_rate": 3.9924371142936695e-05,
      "loss": 0.8417,
      "step": 204310
    },
    {
      "epoch": 3.587141628188697,
      "grad_norm": 0.0636543276861387,
      "learning_rate": 3.9915601251172635e-05,
      "loss": 0.8428,
      "step": 204320
    },
    {
      "epoch": 3.5873171930686985,
      "grad_norm": 0.05056435988226329,
      "learning_rate": 3.990683213335965e-05,
      "loss": 0.837,
      "step": 204330
    },
    {
      "epoch": 3.5874927579487,
      "grad_norm": 0.04109556600883279,
      "learning_rate": 3.9898063789608744e-05,
      "loss": 0.8433,
      "step": 204340
    },
    {
      "epoch": 3.5876683228287014,
      "grad_norm": 0.06119306234595807,
      "learning_rate": 3.988929622003115e-05,
      "loss": 0.8424,
      "step": 204350
    },
    {
      "epoch": 3.587843887708703,
      "grad_norm": 0.054764875019544634,
      "learning_rate": 3.988052942473784e-05,
      "loss": 0.846,
      "step": 204360
    },
    {
      "epoch": 3.5880194525887044,
      "grad_norm": 0.04655053785345462,
      "learning_rate": 3.987176340383994e-05,
      "loss": 0.8346,
      "step": 204370
    },
    {
      "epoch": 3.5881950174687054,
      "grad_norm": 0.055528761084133384,
      "learning_rate": 3.986299815744855e-05,
      "loss": 0.8422,
      "step": 204380
    },
    {
      "epoch": 3.588370582348707,
      "grad_norm": 0.046749292441046586,
      "learning_rate": 3.985423368567466e-05,
      "loss": 0.8396,
      "step": 204390
    },
    {
      "epoch": 3.5885461472287083,
      "grad_norm": 0.045128756945547965,
      "learning_rate": 3.984546998862937e-05,
      "loss": 0.8388,
      "step": 204400
    },
    {
      "epoch": 3.58872171210871,
      "grad_norm": 0.053612154672146285,
      "learning_rate": 3.983670706642364e-05,
      "loss": 0.843,
      "step": 204410
    },
    {
      "epoch": 3.5888972769887113,
      "grad_norm": 0.053299931509505016,
      "learning_rate": 3.982794491916861e-05,
      "loss": 0.8372,
      "step": 204420
    },
    {
      "epoch": 3.5890728418687123,
      "grad_norm": 0.04520492809169053,
      "learning_rate": 3.981918354697525e-05,
      "loss": 0.8427,
      "step": 204430
    },
    {
      "epoch": 3.5892484067487143,
      "grad_norm": 0.054514115105250824,
      "learning_rate": 3.9810422949954485e-05,
      "loss": 0.8357,
      "step": 204440
    },
    {
      "epoch": 3.5894239716287153,
      "grad_norm": 0.04930783690307168,
      "learning_rate": 3.9801663128217465e-05,
      "loss": 0.8398,
      "step": 204450
    },
    {
      "epoch": 3.5895995365087168,
      "grad_norm": 0.055539474788534327,
      "learning_rate": 3.979290408187505e-05,
      "loss": 0.8384,
      "step": 204460
    },
    {
      "epoch": 3.5897751013887182,
      "grad_norm": 0.06281783055519903,
      "learning_rate": 3.978414581103829e-05,
      "loss": 0.837,
      "step": 204470
    },
    {
      "epoch": 3.5899506662687197,
      "grad_norm": 0.06506606588206723,
      "learning_rate": 3.9775388315818165e-05,
      "loss": 0.8386,
      "step": 204480
    },
    {
      "epoch": 3.590126231148721,
      "grad_norm": 0.05005927545891067,
      "learning_rate": 3.976663159632556e-05,
      "loss": 0.8373,
      "step": 204490
    },
    {
      "epoch": 3.5903017960287222,
      "grad_norm": 0.05557592715492525,
      "learning_rate": 3.975787565267151e-05,
      "loss": 0.8321,
      "step": 204500
    },
    {
      "epoch": 3.5904773609087237,
      "grad_norm": 0.048668258672571164,
      "learning_rate": 3.974912048496684e-05,
      "loss": 0.8361,
      "step": 204510
    },
    {
      "epoch": 3.590652925788725,
      "grad_norm": 0.052751446044337666,
      "learning_rate": 3.9740366093322624e-05,
      "loss": 0.8364,
      "step": 204520
    },
    {
      "epoch": 3.5908284906687267,
      "grad_norm": 0.05998474394491238,
      "learning_rate": 3.973161247784968e-05,
      "loss": 0.842,
      "step": 204530
    },
    {
      "epoch": 3.591004055548728,
      "grad_norm": 0.04068300147267656,
      "learning_rate": 3.972285963865894e-05,
      "loss": 0.8447,
      "step": 204540
    },
    {
      "epoch": 3.591179620428729,
      "grad_norm": 0.04148912892458969,
      "learning_rate": 3.9714107575861345e-05,
      "loss": 0.8456,
      "step": 204550
    },
    {
      "epoch": 3.591355185308731,
      "grad_norm": 0.058109584524336205,
      "learning_rate": 3.970535628956773e-05,
      "loss": 0.8462,
      "step": 204560
    },
    {
      "epoch": 3.591530750188732,
      "grad_norm": 0.06208247715264872,
      "learning_rate": 3.969660577988902e-05,
      "loss": 0.8499,
      "step": 204570
    },
    {
      "epoch": 3.5917063150687336,
      "grad_norm": 0.060244207185258684,
      "learning_rate": 3.9687856046935996e-05,
      "loss": 0.8496,
      "step": 204580
    },
    {
      "epoch": 3.591881879948735,
      "grad_norm": 0.057014059319014716,
      "learning_rate": 3.967910709081966e-05,
      "loss": 0.8437,
      "step": 204590
    },
    {
      "epoch": 3.5920574448287366,
      "grad_norm": 0.05354175560781497,
      "learning_rate": 3.967035891165079e-05,
      "loss": 0.8365,
      "step": 204600
    },
    {
      "epoch": 3.592233009708738,
      "grad_norm": 0.04952497051423705,
      "learning_rate": 3.9661611509540154e-05,
      "loss": 0.8434,
      "step": 204610
    },
    {
      "epoch": 3.592408574588739,
      "grad_norm": 0.05199113265511496,
      "learning_rate": 3.9652864884598726e-05,
      "loss": 0.8423,
      "step": 204620
    },
    {
      "epoch": 3.5925841394687406,
      "grad_norm": 0.042752107604686465,
      "learning_rate": 3.964411903693722e-05,
      "loss": 0.8411,
      "step": 204630
    },
    {
      "epoch": 3.592759704348742,
      "grad_norm": 0.058396589090821016,
      "learning_rate": 3.963537396666648e-05,
      "loss": 0.8435,
      "step": 204640
    },
    {
      "epoch": 3.5929352692287435,
      "grad_norm": 0.04426060478742266,
      "learning_rate": 3.962662967389736e-05,
      "loss": 0.8428,
      "step": 204650
    },
    {
      "epoch": 3.593110834108745,
      "grad_norm": 0.06726764283460383,
      "learning_rate": 3.961788615874056e-05,
      "loss": 0.8375,
      "step": 204660
    },
    {
      "epoch": 3.593286398988746,
      "grad_norm": 0.06492381408433007,
      "learning_rate": 3.9609143421306935e-05,
      "loss": 0.8376,
      "step": 204670
    },
    {
      "epoch": 3.593461963868748,
      "grad_norm": 0.0635092876736587,
      "learning_rate": 3.9600401461707174e-05,
      "loss": 0.8374,
      "step": 204680
    },
    {
      "epoch": 3.593637528748749,
      "grad_norm": 0.05313179254238296,
      "learning_rate": 3.959166028005216e-05,
      "loss": 0.8445,
      "step": 204690
    },
    {
      "epoch": 3.5938130936287505,
      "grad_norm": 0.050107896142732644,
      "learning_rate": 3.9582919876452545e-05,
      "loss": 0.8432,
      "step": 204700
    },
    {
      "epoch": 3.593988658508752,
      "grad_norm": 0.0758501328126826,
      "learning_rate": 3.95741802510191e-05,
      "loss": 0.8445,
      "step": 204710
    },
    {
      "epoch": 3.5941642233887534,
      "grad_norm": 0.06156026046656243,
      "learning_rate": 3.9565441403862597e-05,
      "loss": 0.8393,
      "step": 204720
    },
    {
      "epoch": 3.594339788268755,
      "grad_norm": 0.05137165778247093,
      "learning_rate": 3.9556703335093696e-05,
      "loss": 0.8476,
      "step": 204730
    },
    {
      "epoch": 3.594515353148756,
      "grad_norm": 0.07064476975960243,
      "learning_rate": 3.954796604482316e-05,
      "loss": 0.8485,
      "step": 204740
    },
    {
      "epoch": 3.5946909180287574,
      "grad_norm": 0.061450855796457704,
      "learning_rate": 3.953922953316163e-05,
      "loss": 0.8434,
      "step": 204750
    },
    {
      "epoch": 3.594866482908759,
      "grad_norm": 0.05544320313533874,
      "learning_rate": 3.953049380021989e-05,
      "loss": 0.8428,
      "step": 204760
    },
    {
      "epoch": 3.5950420477887604,
      "grad_norm": 0.05421355594516983,
      "learning_rate": 3.952175884610858e-05,
      "loss": 0.8463,
      "step": 204770
    },
    {
      "epoch": 3.595217612668762,
      "grad_norm": 0.05017102184770904,
      "learning_rate": 3.95130246709383e-05,
      "loss": 0.8412,
      "step": 204780
    },
    {
      "epoch": 3.5953931775487633,
      "grad_norm": 0.04677073399474747,
      "learning_rate": 3.950429127481984e-05,
      "loss": 0.8404,
      "step": 204790
    },
    {
      "epoch": 3.595568742428765,
      "grad_norm": 0.050960541088465414,
      "learning_rate": 3.949555865786377e-05,
      "loss": 0.8382,
      "step": 204800
    },
    {
      "epoch": 3.595744307308766,
      "grad_norm": 0.061062939452717006,
      "learning_rate": 3.9486826820180757e-05,
      "loss": 0.8462,
      "step": 204810
    },
    {
      "epoch": 3.5959198721887673,
      "grad_norm": 0.051380066182515026,
      "learning_rate": 3.9478095761881474e-05,
      "loss": 0.8408,
      "step": 204820
    },
    {
      "epoch": 3.5960954370687688,
      "grad_norm": 0.06995943111900167,
      "learning_rate": 3.946936548307647e-05,
      "loss": 0.8499,
      "step": 204830
    },
    {
      "epoch": 3.5962710019487703,
      "grad_norm": 0.04575708256248788,
      "learning_rate": 3.9460635983876435e-05,
      "loss": 0.8409,
      "step": 204840
    },
    {
      "epoch": 3.5964465668287717,
      "grad_norm": 0.05365746305898582,
      "learning_rate": 3.9451907264391854e-05,
      "loss": 0.8343,
      "step": 204850
    },
    {
      "epoch": 3.5966221317087728,
      "grad_norm": 0.04537718928167727,
      "learning_rate": 3.9443179324733483e-05,
      "loss": 0.8434,
      "step": 204860
    },
    {
      "epoch": 3.5967976965887742,
      "grad_norm": 0.057417647591066885,
      "learning_rate": 3.9434452165011795e-05,
      "loss": 0.8434,
      "step": 204870
    },
    {
      "epoch": 3.5969732614687757,
      "grad_norm": 0.058430749169114116,
      "learning_rate": 3.94257257853374e-05,
      "loss": 0.8434,
      "step": 204880
    },
    {
      "epoch": 3.597148826348777,
      "grad_norm": 0.07077381702810077,
      "learning_rate": 3.941700018582089e-05,
      "loss": 0.8369,
      "step": 204890
    },
    {
      "epoch": 3.5973243912287787,
      "grad_norm": 0.05728535322418454,
      "learning_rate": 3.940827536657275e-05,
      "loss": 0.8447,
      "step": 204900
    },
    {
      "epoch": 3.59749995610878,
      "grad_norm": 0.03733040770720047,
      "learning_rate": 3.939955132770361e-05,
      "loss": 0.8439,
      "step": 204910
    },
    {
      "epoch": 3.5976755209887816,
      "grad_norm": 0.05676995733419339,
      "learning_rate": 3.939082806932389e-05,
      "loss": 0.8398,
      "step": 204920
    },
    {
      "epoch": 3.5978510858687827,
      "grad_norm": 0.05439654773710176,
      "learning_rate": 3.9382105591544247e-05,
      "loss": 0.8348,
      "step": 204930
    },
    {
      "epoch": 3.598026650748784,
      "grad_norm": 0.04746052300164268,
      "learning_rate": 3.937338389447513e-05,
      "loss": 0.8411,
      "step": 204940
    },
    {
      "epoch": 3.5982022156287856,
      "grad_norm": 0.048560135141762974,
      "learning_rate": 3.9364662978226996e-05,
      "loss": 0.8354,
      "step": 204950
    },
    {
      "epoch": 3.598377780508787,
      "grad_norm": 0.05303638706241667,
      "learning_rate": 3.935594284291046e-05,
      "loss": 0.8414,
      "step": 204960
    },
    {
      "epoch": 3.5985533453887886,
      "grad_norm": 0.04985022539138696,
      "learning_rate": 3.934722348863591e-05,
      "loss": 0.8496,
      "step": 204970
    },
    {
      "epoch": 3.5987289102687896,
      "grad_norm": 0.058668657152235244,
      "learning_rate": 3.933850491551385e-05,
      "loss": 0.8377,
      "step": 204980
    },
    {
      "epoch": 3.598904475148791,
      "grad_norm": 0.05125664206009921,
      "learning_rate": 3.9329787123654784e-05,
      "loss": 0.8378,
      "step": 204990
    },
    {
      "epoch": 3.5990800400287926,
      "grad_norm": 0.051398140372506916,
      "learning_rate": 3.93210701131691e-05,
      "loss": 0.8463,
      "step": 205000
    },
    {
      "epoch": 3.599255604908794,
      "grad_norm": 0.06055750173302328,
      "learning_rate": 3.931235388416731e-05,
      "loss": 0.8186,
      "step": 205010
    },
    {
      "epoch": 3.5994311697887955,
      "grad_norm": 0.05198755123260614,
      "learning_rate": 3.930363843675976e-05,
      "loss": 0.8421,
      "step": 205020
    },
    {
      "epoch": 3.599606734668797,
      "grad_norm": 0.043392092560392934,
      "learning_rate": 3.929492377105699e-05,
      "loss": 0.8356,
      "step": 205030
    },
    {
      "epoch": 3.5997822995487985,
      "grad_norm": 0.04287648588073483,
      "learning_rate": 3.928620988716933e-05,
      "loss": 0.8462,
      "step": 205040
    },
    {
      "epoch": 3.5999578644287995,
      "grad_norm": 0.06927978297844112,
      "learning_rate": 3.927749678520722e-05,
      "loss": 0.8398,
      "step": 205050
    },
    {
      "epoch": 3.600133429308801,
      "grad_norm": 0.06803504008897289,
      "learning_rate": 3.926878446528109e-05,
      "loss": 0.8464,
      "step": 205060
    },
    {
      "epoch": 3.6003089941888025,
      "grad_norm": 0.04358008714444811,
      "learning_rate": 3.9260072927501264e-05,
      "loss": 0.8326,
      "step": 205070
    },
    {
      "epoch": 3.600484559068804,
      "grad_norm": 0.052450348517060943,
      "learning_rate": 3.9251362171978166e-05,
      "loss": 0.8444,
      "step": 205080
    },
    {
      "epoch": 3.6006601239488054,
      "grad_norm": 0.05635574224695734,
      "learning_rate": 3.924265219882209e-05,
      "loss": 0.8353,
      "step": 205090
    },
    {
      "epoch": 3.6008356888288064,
      "grad_norm": 0.04643866577186989,
      "learning_rate": 3.923394300814351e-05,
      "loss": 0.8374,
      "step": 205100
    },
    {
      "epoch": 3.601011253708808,
      "grad_norm": 0.04687123181143815,
      "learning_rate": 3.92252346000527e-05,
      "loss": 0.8415,
      "step": 205110
    },
    {
      "epoch": 3.6011868185888094,
      "grad_norm": 0.07381795863738276,
      "learning_rate": 3.921652697465995e-05,
      "loss": 0.8401,
      "step": 205120
    },
    {
      "epoch": 3.601362383468811,
      "grad_norm": 0.044798046763569944,
      "learning_rate": 3.9207820132075715e-05,
      "loss": 0.8382,
      "step": 205130
    },
    {
      "epoch": 3.6015379483488124,
      "grad_norm": 0.04561657906502529,
      "learning_rate": 3.919911407241019e-05,
      "loss": 0.8496,
      "step": 205140
    },
    {
      "epoch": 3.601713513228814,
      "grad_norm": 0.07555048640121428,
      "learning_rate": 3.919040879577375e-05,
      "loss": 0.837,
      "step": 205150
    },
    {
      "epoch": 3.6018890781088153,
      "grad_norm": 0.05339499803919997,
      "learning_rate": 3.91817043022767e-05,
      "loss": 0.8272,
      "step": 205160
    },
    {
      "epoch": 3.6020646429888163,
      "grad_norm": 0.05081296803064304,
      "learning_rate": 3.917300059202928e-05,
      "loss": 0.8349,
      "step": 205170
    },
    {
      "epoch": 3.602240207868818,
      "grad_norm": 0.05411707996630344,
      "learning_rate": 3.9164297665141816e-05,
      "loss": 0.8406,
      "step": 205180
    },
    {
      "epoch": 3.6024157727488193,
      "grad_norm": 0.04953020559180365,
      "learning_rate": 3.915559552172449e-05,
      "loss": 0.8426,
      "step": 205190
    },
    {
      "epoch": 3.6025913376288208,
      "grad_norm": 0.04799327735216028,
      "learning_rate": 3.914689416188769e-05,
      "loss": 0.8362,
      "step": 205200
    },
    {
      "epoch": 3.6027669025088223,
      "grad_norm": 0.05077452688504373,
      "learning_rate": 3.913819358574155e-05,
      "loss": 0.8492,
      "step": 205210
    },
    {
      "epoch": 3.6029424673888233,
      "grad_norm": 0.049182278042932875,
      "learning_rate": 3.912949379339637e-05,
      "loss": 0.8356,
      "step": 205220
    },
    {
      "epoch": 3.603118032268825,
      "grad_norm": 0.04472791617996965,
      "learning_rate": 3.912079478496238e-05,
      "loss": 0.8284,
      "step": 205230
    },
    {
      "epoch": 3.6032935971488262,
      "grad_norm": 0.052217825150400084,
      "learning_rate": 3.9112096560549745e-05,
      "loss": 0.8348,
      "step": 205240
    },
    {
      "epoch": 3.6034691620288277,
      "grad_norm": 0.05421949937660823,
      "learning_rate": 3.910339912026876e-05,
      "loss": 0.836,
      "step": 205250
    },
    {
      "epoch": 3.603644726908829,
      "grad_norm": 0.05926507954741897,
      "learning_rate": 3.909470246422948e-05,
      "loss": 0.8364,
      "step": 205260
    },
    {
      "epoch": 3.6038202917888307,
      "grad_norm": 0.07183488476387236,
      "learning_rate": 3.9086006592542265e-05,
      "loss": 0.8471,
      "step": 205270
    },
    {
      "epoch": 3.603995856668832,
      "grad_norm": 0.04545244485283583,
      "learning_rate": 3.907731150531721e-05,
      "loss": 0.8403,
      "step": 205280
    },
    {
      "epoch": 3.604171421548833,
      "grad_norm": 0.04845101775753622,
      "learning_rate": 3.906861720266441e-05,
      "loss": 0.845,
      "step": 205290
    },
    {
      "epoch": 3.6043469864288347,
      "grad_norm": 0.056701221906107026,
      "learning_rate": 3.905992368469418e-05,
      "loss": 0.8407,
      "step": 205300
    },
    {
      "epoch": 3.604522551308836,
      "grad_norm": 0.057408701242166,
      "learning_rate": 3.905123095151654e-05,
      "loss": 0.8394,
      "step": 205310
    },
    {
      "epoch": 3.6046981161888376,
      "grad_norm": 0.04621694729443932,
      "learning_rate": 3.904253900324169e-05,
      "loss": 0.8489,
      "step": 205320
    },
    {
      "epoch": 3.604873681068839,
      "grad_norm": 0.046230276568850384,
      "learning_rate": 3.903384783997977e-05,
      "loss": 0.8401,
      "step": 205330
    },
    {
      "epoch": 3.60504924594884,
      "grad_norm": 0.06264349765374169,
      "learning_rate": 3.902515746184085e-05,
      "loss": 0.841,
      "step": 205340
    },
    {
      "epoch": 3.605224810828842,
      "grad_norm": 0.05582212017999714,
      "learning_rate": 3.901646786893508e-05,
      "loss": 0.847,
      "step": 205350
    },
    {
      "epoch": 3.605400375708843,
      "grad_norm": 0.05520935873241114,
      "learning_rate": 3.900777906137248e-05,
      "loss": 0.8401,
      "step": 205360
    },
    {
      "epoch": 3.6055759405888446,
      "grad_norm": 0.054035228331527074,
      "learning_rate": 3.899909103926327e-05,
      "loss": 0.8435,
      "step": 205370
    },
    {
      "epoch": 3.605751505468846,
      "grad_norm": 0.04712151243625597,
      "learning_rate": 3.899040380271743e-05,
      "loss": 0.8454,
      "step": 205380
    },
    {
      "epoch": 3.6059270703488475,
      "grad_norm": 0.05634423836593053,
      "learning_rate": 3.898171735184505e-05,
      "loss": 0.8423,
      "step": 205390
    },
    {
      "epoch": 3.606102635228849,
      "grad_norm": 0.05033326573343576,
      "learning_rate": 3.897303168675624e-05,
      "loss": 0.8441,
      "step": 205400
    },
    {
      "epoch": 3.60627820010885,
      "grad_norm": 0.05956989644420946,
      "learning_rate": 3.896434680756096e-05,
      "loss": 0.8402,
      "step": 205410
    },
    {
      "epoch": 3.6064537649888515,
      "grad_norm": 0.08873274419657348,
      "learning_rate": 3.895566271436934e-05,
      "loss": 0.8415,
      "step": 205420
    },
    {
      "epoch": 3.606629329868853,
      "grad_norm": 0.05680151411637027,
      "learning_rate": 3.894697940729129e-05,
      "loss": 0.8372,
      "step": 205430
    },
    {
      "epoch": 3.6068048947488545,
      "grad_norm": 0.05747579759280873,
      "learning_rate": 3.893829688643696e-05,
      "loss": 0.844,
      "step": 205440
    },
    {
      "epoch": 3.606980459628856,
      "grad_norm": 0.06372835148996667,
      "learning_rate": 3.892961515191633e-05,
      "loss": 0.8415,
      "step": 205450
    },
    {
      "epoch": 3.607156024508857,
      "grad_norm": 0.05222856610704827,
      "learning_rate": 3.892093420383928e-05,
      "loss": 0.8477,
      "step": 205460
    },
    {
      "epoch": 3.607331589388859,
      "grad_norm": 0.07088707315735987,
      "learning_rate": 3.891225404231597e-05,
      "loss": 0.8516,
      "step": 205470
    },
    {
      "epoch": 3.60750715426886,
      "grad_norm": 0.04989059129260786,
      "learning_rate": 3.8903574667456264e-05,
      "loss": 0.8418,
      "step": 205480
    },
    {
      "epoch": 3.6076827191488614,
      "grad_norm": 0.05327714673891411,
      "learning_rate": 3.889489607937017e-05,
      "loss": 0.8427,
      "step": 205490
    },
    {
      "epoch": 3.607858284028863,
      "grad_norm": 0.06067340965083805,
      "learning_rate": 3.8886218278167676e-05,
      "loss": 0.8457,
      "step": 205500
    },
    {
      "epoch": 3.6080338489088644,
      "grad_norm": 0.07339359310357513,
      "learning_rate": 3.887754126395866e-05,
      "loss": 0.8392,
      "step": 205510
    },
    {
      "epoch": 3.608209413788866,
      "grad_norm": 0.05268136270198482,
      "learning_rate": 3.8868865036853144e-05,
      "loss": 0.8408,
      "step": 205520
    },
    {
      "epoch": 3.608384978668867,
      "grad_norm": 0.04622489179103295,
      "learning_rate": 3.886018959696095e-05,
      "loss": 0.8418,
      "step": 205530
    },
    {
      "epoch": 3.6085605435488683,
      "grad_norm": 0.05559462171428013,
      "learning_rate": 3.8851514944392124e-05,
      "loss": 0.8378,
      "step": 205540
    },
    {
      "epoch": 3.60873610842887,
      "grad_norm": 0.05120516176971408,
      "learning_rate": 3.884284107925649e-05,
      "loss": 0.8522,
      "step": 205550
    },
    {
      "epoch": 3.6089116733088713,
      "grad_norm": 0.052227965172714295,
      "learning_rate": 3.883416800166396e-05,
      "loss": 0.8405,
      "step": 205560
    },
    {
      "epoch": 3.609087238188873,
      "grad_norm": 0.05336247145481215,
      "learning_rate": 3.882549571172447e-05,
      "loss": 0.845,
      "step": 205570
    },
    {
      "epoch": 3.6092628030688743,
      "grad_norm": 0.04634123985089257,
      "learning_rate": 3.881682420954784e-05,
      "loss": 0.8333,
      "step": 205580
    },
    {
      "epoch": 3.6094383679488757,
      "grad_norm": 0.053583393390799754,
      "learning_rate": 3.880815349524399e-05,
      "loss": 0.8388,
      "step": 205590
    },
    {
      "epoch": 3.6096139328288768,
      "grad_norm": 0.049951565724785804,
      "learning_rate": 3.8799483568922697e-05,
      "loss": 0.8392,
      "step": 205600
    },
    {
      "epoch": 3.6097894977088782,
      "grad_norm": 0.06183337513198959,
      "learning_rate": 3.8790814430693926e-05,
      "loss": 0.84,
      "step": 205610
    },
    {
      "epoch": 3.6099650625888797,
      "grad_norm": 0.05833007666109794,
      "learning_rate": 3.878214608066747e-05,
      "loss": 0.8491,
      "step": 205620
    },
    {
      "epoch": 3.610140627468881,
      "grad_norm": 0.054275177841680775,
      "learning_rate": 3.877347851895307e-05,
      "loss": 0.8444,
      "step": 205630
    },
    {
      "epoch": 3.6103161923488827,
      "grad_norm": 0.055208003051766395,
      "learning_rate": 3.876481174566069e-05,
      "loss": 0.836,
      "step": 205640
    },
    {
      "epoch": 3.6104917572288837,
      "grad_norm": 0.0435465960383356,
      "learning_rate": 3.875614576090004e-05,
      "loss": 0.842,
      "step": 205650
    },
    {
      "epoch": 3.610667322108885,
      "grad_norm": 0.053880028810798855,
      "learning_rate": 3.874748056478095e-05,
      "loss": 0.8382,
      "step": 205660
    },
    {
      "epoch": 3.6108428869888867,
      "grad_norm": 0.052297215929060024,
      "learning_rate": 3.8738816157413246e-05,
      "loss": 0.8393,
      "step": 205670
    },
    {
      "epoch": 3.611018451868888,
      "grad_norm": 0.06195818724401174,
      "learning_rate": 3.873015253890665e-05,
      "loss": 0.8333,
      "step": 205680
    },
    {
      "epoch": 3.6111940167488896,
      "grad_norm": 0.047339024016607396,
      "learning_rate": 3.872148970937098e-05,
      "loss": 0.8373,
      "step": 205690
    },
    {
      "epoch": 3.611369581628891,
      "grad_norm": 0.05458551659256949,
      "learning_rate": 3.87128276689159e-05,
      "loss": 0.8385,
      "step": 205700
    },
    {
      "epoch": 3.6115451465088926,
      "grad_norm": 0.05482436318924736,
      "learning_rate": 3.8704166417651314e-05,
      "loss": 0.8361,
      "step": 205710
    },
    {
      "epoch": 3.6117207113888936,
      "grad_norm": 0.05054892788381928,
      "learning_rate": 3.869550595568682e-05,
      "loss": 0.8433,
      "step": 205720
    },
    {
      "epoch": 3.611896276268895,
      "grad_norm": 0.06021391401244948,
      "learning_rate": 3.868684628313223e-05,
      "loss": 0.8441,
      "step": 205730
    },
    {
      "epoch": 3.6120718411488966,
      "grad_norm": 0.07310333702765937,
      "learning_rate": 3.867818740009726e-05,
      "loss": 0.8405,
      "step": 205740
    },
    {
      "epoch": 3.612247406028898,
      "grad_norm": 0.05681457612712172,
      "learning_rate": 3.866952930669157e-05,
      "loss": 0.8384,
      "step": 205750
    },
    {
      "epoch": 3.6124229709088995,
      "grad_norm": 0.047820006658318946,
      "learning_rate": 3.8660872003024915e-05,
      "loss": 0.8336,
      "step": 205760
    },
    {
      "epoch": 3.6125985357889006,
      "grad_norm": 0.06429024545419013,
      "learning_rate": 3.86522154892069e-05,
      "loss": 0.8376,
      "step": 205770
    },
    {
      "epoch": 3.612774100668902,
      "grad_norm": 0.0516275166386642,
      "learning_rate": 3.864355976534734e-05,
      "loss": 0.8342,
      "step": 205780
    },
    {
      "epoch": 3.6129496655489035,
      "grad_norm": 0.049801170651265415,
      "learning_rate": 3.8634904831555815e-05,
      "loss": 0.8352,
      "step": 205790
    },
    {
      "epoch": 3.613125230428905,
      "grad_norm": 0.05029178258873204,
      "learning_rate": 3.8626250687941935e-05,
      "loss": 0.8287,
      "step": 205800
    },
    {
      "epoch": 3.6133007953089065,
      "grad_norm": 0.051623569774772164,
      "learning_rate": 3.8617597334615466e-05,
      "loss": 0.8389,
      "step": 205810
    },
    {
      "epoch": 3.613476360188908,
      "grad_norm": 0.05544721158993487,
      "learning_rate": 3.860894477168598e-05,
      "loss": 0.8398,
      "step": 205820
    },
    {
      "epoch": 3.6136519250689094,
      "grad_norm": 0.047565347582713816,
      "learning_rate": 3.860029299926311e-05,
      "loss": 0.8438,
      "step": 205830
    },
    {
      "epoch": 3.6138274899489105,
      "grad_norm": 0.061203683609826025,
      "learning_rate": 3.859164201745652e-05,
      "loss": 0.8353,
      "step": 205840
    },
    {
      "epoch": 3.614003054828912,
      "grad_norm": 0.05332895454644122,
      "learning_rate": 3.858299182637577e-05,
      "loss": 0.8416,
      "step": 205850
    },
    {
      "epoch": 3.6141786197089134,
      "grad_norm": 0.04389786701703956,
      "learning_rate": 3.8574342426130494e-05,
      "loss": 0.8444,
      "step": 205860
    },
    {
      "epoch": 3.614354184588915,
      "grad_norm": 0.06383018709991686,
      "learning_rate": 3.85656938168302e-05,
      "loss": 0.8347,
      "step": 205870
    },
    {
      "epoch": 3.6145297494689164,
      "grad_norm": 0.04782890198088906,
      "learning_rate": 3.855704599858459e-05,
      "loss": 0.8315,
      "step": 205880
    },
    {
      "epoch": 3.6147053143489174,
      "grad_norm": 0.04715849153760157,
      "learning_rate": 3.854839897150315e-05,
      "loss": 0.8348,
      "step": 205890
    },
    {
      "epoch": 3.6148808792289193,
      "grad_norm": 0.04774392650030874,
      "learning_rate": 3.8539752735695456e-05,
      "loss": 0.8443,
      "step": 205900
    },
    {
      "epoch": 3.6150564441089204,
      "grad_norm": 0.044066630434360765,
      "learning_rate": 3.853110729127111e-05,
      "loss": 0.829,
      "step": 205910
    },
    {
      "epoch": 3.615232008988922,
      "grad_norm": 0.053614839788542895,
      "learning_rate": 3.8522462638339566e-05,
      "loss": 0.8418,
      "step": 205920
    },
    {
      "epoch": 3.6154075738689233,
      "grad_norm": 0.04564620893872445,
      "learning_rate": 3.851381877701043e-05,
      "loss": 0.8342,
      "step": 205930
    },
    {
      "epoch": 3.615583138748925,
      "grad_norm": 0.0532526513338837,
      "learning_rate": 3.850517570739312e-05,
      "loss": 0.8469,
      "step": 205940
    },
    {
      "epoch": 3.6157587036289263,
      "grad_norm": 0.05385917546445037,
      "learning_rate": 3.849653342959727e-05,
      "loss": 0.847,
      "step": 205950
    },
    {
      "epoch": 3.6159342685089273,
      "grad_norm": 0.07133877523512872,
      "learning_rate": 3.848789194373233e-05,
      "loss": 0.8427,
      "step": 205960
    },
    {
      "epoch": 3.6161098333889288,
      "grad_norm": 0.04304159778462057,
      "learning_rate": 3.847925124990771e-05,
      "loss": 0.8368,
      "step": 205970
    },
    {
      "epoch": 3.6162853982689303,
      "grad_norm": 0.05969400719428752,
      "learning_rate": 3.847061134823302e-05,
      "loss": 0.8377,
      "step": 205980
    },
    {
      "epoch": 3.6164609631489317,
      "grad_norm": 0.04950107243832494,
      "learning_rate": 3.846197223881764e-05,
      "loss": 0.8354,
      "step": 205990
    },
    {
      "epoch": 3.616636528028933,
      "grad_norm": 0.04812405896738595,
      "learning_rate": 3.845333392177106e-05,
      "loss": 0.8413,
      "step": 206000
    },
    {
      "epoch": 3.6168120929089342,
      "grad_norm": 0.044200551724439055,
      "learning_rate": 3.844469639720275e-05,
      "loss": 0.8418,
      "step": 206010
    },
    {
      "epoch": 3.616987657788936,
      "grad_norm": 0.06081836588899363,
      "learning_rate": 3.843605966522209e-05,
      "loss": 0.8431,
      "step": 206020
    },
    {
      "epoch": 3.617163222668937,
      "grad_norm": 0.051067786115450355,
      "learning_rate": 3.842742372593859e-05,
      "loss": 0.8436,
      "step": 206030
    },
    {
      "epoch": 3.6173387875489387,
      "grad_norm": 0.053376288577542695,
      "learning_rate": 3.8418788579461554e-05,
      "loss": 0.8381,
      "step": 206040
    },
    {
      "epoch": 3.61751435242894,
      "grad_norm": 0.05130427154109696,
      "learning_rate": 3.841015422590052e-05,
      "loss": 0.845,
      "step": 206050
    },
    {
      "epoch": 3.6176899173089416,
      "grad_norm": 0.04093502558692911,
      "learning_rate": 3.840152066536479e-05,
      "loss": 0.8491,
      "step": 206060
    },
    {
      "epoch": 3.617865482188943,
      "grad_norm": 0.07628326013346551,
      "learning_rate": 3.8392887897963806e-05,
      "loss": 0.8395,
      "step": 206070
    },
    {
      "epoch": 3.618041047068944,
      "grad_norm": 0.05518679821728958,
      "learning_rate": 3.838425592380695e-05,
      "loss": 0.8424,
      "step": 206080
    },
    {
      "epoch": 3.6182166119489456,
      "grad_norm": 0.0670995177317251,
      "learning_rate": 3.837562474300355e-05,
      "loss": 0.8314,
      "step": 206090
    },
    {
      "epoch": 3.618392176828947,
      "grad_norm": 0.05035822525168887,
      "learning_rate": 3.836699435566301e-05,
      "loss": 0.839,
      "step": 206100
    },
    {
      "epoch": 3.6185677417089486,
      "grad_norm": 0.05599632016160772,
      "learning_rate": 3.835836476189459e-05,
      "loss": 0.8363,
      "step": 206110
    },
    {
      "epoch": 3.61874330658895,
      "grad_norm": 0.06123110149596059,
      "learning_rate": 3.834973596180777e-05,
      "loss": 0.8424,
      "step": 206120
    },
    {
      "epoch": 3.618918871468951,
      "grad_norm": 0.055097016723922684,
      "learning_rate": 3.83411079555118e-05,
      "loss": 0.8412,
      "step": 206130
    },
    {
      "epoch": 3.619094436348953,
      "grad_norm": 0.053104643558551545,
      "learning_rate": 3.833248074311592e-05,
      "loss": 0.8312,
      "step": 206140
    },
    {
      "epoch": 3.619270001228954,
      "grad_norm": 0.06708783452971621,
      "learning_rate": 3.8323854324729604e-05,
      "loss": 0.841,
      "step": 206150
    },
    {
      "epoch": 3.6194455661089555,
      "grad_norm": 0.06694328705197654,
      "learning_rate": 3.8315228700462026e-05,
      "loss": 0.8424,
      "step": 206160
    },
    {
      "epoch": 3.619621130988957,
      "grad_norm": 0.04916098478172704,
      "learning_rate": 3.83066038704225e-05,
      "loss": 0.8419,
      "step": 206170
    },
    {
      "epoch": 3.6197966958689585,
      "grad_norm": 0.055964557594899496,
      "learning_rate": 3.829797983472037e-05,
      "loss": 0.8372,
      "step": 206180
    },
    {
      "epoch": 3.61997226074896,
      "grad_norm": 0.04575873824684855,
      "learning_rate": 3.828935659346482e-05,
      "loss": 0.8427,
      "step": 206190
    },
    {
      "epoch": 3.620147825628961,
      "grad_norm": 0.05713333562760868,
      "learning_rate": 3.8280734146765155e-05,
      "loss": 0.8419,
      "step": 206200
    },
    {
      "epoch": 3.6203233905089625,
      "grad_norm": 0.0725651423546486,
      "learning_rate": 3.827211249473055e-05,
      "loss": 0.8405,
      "step": 206210
    },
    {
      "epoch": 3.620498955388964,
      "grad_norm": 0.05260416696395593,
      "learning_rate": 3.8263491637470375e-05,
      "loss": 0.8428,
      "step": 206220
    },
    {
      "epoch": 3.6206745202689654,
      "grad_norm": 0.06452169178070956,
      "learning_rate": 3.825487157509374e-05,
      "loss": 0.8324,
      "step": 206230
    },
    {
      "epoch": 3.620850085148967,
      "grad_norm": 0.048812875011157573,
      "learning_rate": 3.8246252307709914e-05,
      "loss": 0.8374,
      "step": 206240
    },
    {
      "epoch": 3.6210256500289684,
      "grad_norm": 0.05039713321673325,
      "learning_rate": 3.823763383542813e-05,
      "loss": 0.8354,
      "step": 206250
    },
    {
      "epoch": 3.62120121490897,
      "grad_norm": 0.043539241481947225,
      "learning_rate": 3.8229016158357526e-05,
      "loss": 0.8363,
      "step": 206260
    },
    {
      "epoch": 3.621376779788971,
      "grad_norm": 0.05835300006825358,
      "learning_rate": 3.822039927660734e-05,
      "loss": 0.843,
      "step": 206270
    },
    {
      "epoch": 3.6215523446689724,
      "grad_norm": 0.048539226699357306,
      "learning_rate": 3.821178319028667e-05,
      "loss": 0.8397,
      "step": 206280
    },
    {
      "epoch": 3.621727909548974,
      "grad_norm": 0.0505786266005709,
      "learning_rate": 3.82031678995048e-05,
      "loss": 0.8472,
      "step": 206290
    },
    {
      "epoch": 3.6219034744289753,
      "grad_norm": 0.05376126825107002,
      "learning_rate": 3.819455340437083e-05,
      "loss": 0.8367,
      "step": 206300
    },
    {
      "epoch": 3.622079039308977,
      "grad_norm": 0.043866455637275745,
      "learning_rate": 3.818593970499383e-05,
      "loss": 0.838,
      "step": 206310
    },
    {
      "epoch": 3.622254604188978,
      "grad_norm": 0.05829822814889147,
      "learning_rate": 3.8177326801483096e-05,
      "loss": 0.8405,
      "step": 206320
    },
    {
      "epoch": 3.6224301690689793,
      "grad_norm": 0.05149512205591227,
      "learning_rate": 3.816871469394763e-05,
      "loss": 0.8371,
      "step": 206330
    },
    {
      "epoch": 3.6226057339489808,
      "grad_norm": 0.05847530308499657,
      "learning_rate": 3.816010338249659e-05,
      "loss": 0.8371,
      "step": 206340
    },
    {
      "epoch": 3.6227812988289823,
      "grad_norm": 0.06004125718103159,
      "learning_rate": 3.815149286723912e-05,
      "loss": 0.8373,
      "step": 206350
    },
    {
      "epoch": 3.6229568637089837,
      "grad_norm": 0.04616018779920488,
      "learning_rate": 3.814288314828425e-05,
      "loss": 0.8366,
      "step": 206360
    },
    {
      "epoch": 3.623132428588985,
      "grad_norm": 0.06809861558078194,
      "learning_rate": 3.8134274225741125e-05,
      "loss": 0.8352,
      "step": 206370
    },
    {
      "epoch": 3.6233079934689867,
      "grad_norm": 0.0551047293064294,
      "learning_rate": 3.8125666099718716e-05,
      "loss": 0.8544,
      "step": 206380
    },
    {
      "epoch": 3.6234835583489877,
      "grad_norm": 0.06712575936336419,
      "learning_rate": 3.811705877032624e-05,
      "loss": 0.8382,
      "step": 206390
    },
    {
      "epoch": 3.623659123228989,
      "grad_norm": 0.06508140155927838,
      "learning_rate": 3.810845223767265e-05,
      "loss": 0.8477,
      "step": 206400
    },
    {
      "epoch": 3.6238346881089907,
      "grad_norm": 0.07290756342540618,
      "learning_rate": 3.809984650186703e-05,
      "loss": 0.839,
      "step": 206410
    },
    {
      "epoch": 3.624010252988992,
      "grad_norm": 0.04985690292632335,
      "learning_rate": 3.809124156301844e-05,
      "loss": 0.8385,
      "step": 206420
    },
    {
      "epoch": 3.6241858178689936,
      "grad_norm": 0.046799534293092464,
      "learning_rate": 3.808263742123584e-05,
      "loss": 0.8404,
      "step": 206430
    },
    {
      "epoch": 3.6243613827489947,
      "grad_norm": 0.05736580525830358,
      "learning_rate": 3.807403407662832e-05,
      "loss": 0.8396,
      "step": 206440
    },
    {
      "epoch": 3.624536947628996,
      "grad_norm": 0.06913738628558357,
      "learning_rate": 3.806543152930477e-05,
      "loss": 0.8383,
      "step": 206450
    },
    {
      "epoch": 3.6247125125089976,
      "grad_norm": 0.05665875881666919,
      "learning_rate": 3.8056829779374345e-05,
      "loss": 0.8429,
      "step": 206460
    },
    {
      "epoch": 3.624888077388999,
      "grad_norm": 0.052040673477495164,
      "learning_rate": 3.8048228826945936e-05,
      "loss": 0.8389,
      "step": 206470
    },
    {
      "epoch": 3.6250636422690006,
      "grad_norm": 0.05995725153018644,
      "learning_rate": 3.8039628672128474e-05,
      "loss": 0.8404,
      "step": 206480
    },
    {
      "epoch": 3.625239207149002,
      "grad_norm": 0.05239430104042522,
      "learning_rate": 3.803102931503105e-05,
      "loss": 0.839,
      "step": 206490
    },
    {
      "epoch": 3.6254147720290035,
      "grad_norm": 0.048386591402977734,
      "learning_rate": 3.802243075576252e-05,
      "loss": 0.8432,
      "step": 206500
    },
    {
      "epoch": 3.6255903369090046,
      "grad_norm": 0.07512464717994759,
      "learning_rate": 3.8013832994431845e-05,
      "loss": 0.8407,
      "step": 206510
    },
    {
      "epoch": 3.625765901789006,
      "grad_norm": 0.04993971159162274,
      "learning_rate": 3.8005236031148036e-05,
      "loss": 0.8395,
      "step": 206520
    },
    {
      "epoch": 3.6259414666690075,
      "grad_norm": 0.048844414580567354,
      "learning_rate": 3.7996639866019904e-05,
      "loss": 0.8349,
      "step": 206530
    },
    {
      "epoch": 3.626117031549009,
      "grad_norm": 0.04713639528947886,
      "learning_rate": 3.7988044499156465e-05,
      "loss": 0.8428,
      "step": 206540
    },
    {
      "epoch": 3.6262925964290105,
      "grad_norm": 0.05605772707236479,
      "learning_rate": 3.797944993066652e-05,
      "loss": 0.8408,
      "step": 206550
    },
    {
      "epoch": 3.6264681613090115,
      "grad_norm": 0.055607924171770774,
      "learning_rate": 3.797085616065907e-05,
      "loss": 0.8387,
      "step": 206560
    },
    {
      "epoch": 3.626643726189013,
      "grad_norm": 0.049416292077697835,
      "learning_rate": 3.796226318924293e-05,
      "loss": 0.8374,
      "step": 206570
    },
    {
      "epoch": 3.6268192910690145,
      "grad_norm": 0.049424556307928594,
      "learning_rate": 3.7953671016526985e-05,
      "loss": 0.8443,
      "step": 206580
    },
    {
      "epoch": 3.626994855949016,
      "grad_norm": 0.05525255381294665,
      "learning_rate": 3.794507964262016e-05,
      "loss": 0.8369,
      "step": 206590
    },
    {
      "epoch": 3.6271704208290174,
      "grad_norm": 0.06130387710275784,
      "learning_rate": 3.793648906763121e-05,
      "loss": 0.8382,
      "step": 206600
    },
    {
      "epoch": 3.627345985709019,
      "grad_norm": 0.05286148979032406,
      "learning_rate": 3.792789929166908e-05,
      "loss": 0.843,
      "step": 206610
    },
    {
      "epoch": 3.6275215505890204,
      "grad_norm": 0.04817752804282527,
      "learning_rate": 3.791931031484249e-05,
      "loss": 0.8348,
      "step": 206620
    },
    {
      "epoch": 3.6276971154690214,
      "grad_norm": 0.05607929932615158,
      "learning_rate": 3.791072213726038e-05,
      "loss": 0.8363,
      "step": 206630
    },
    {
      "epoch": 3.627872680349023,
      "grad_norm": 0.0661608615296563,
      "learning_rate": 3.7902134759031513e-05,
      "loss": 0.8384,
      "step": 206640
    },
    {
      "epoch": 3.6280482452290244,
      "grad_norm": 0.05787240988631175,
      "learning_rate": 3.789354818026462e-05,
      "loss": 0.8389,
      "step": 206650
    },
    {
      "epoch": 3.628223810109026,
      "grad_norm": 0.0675580752365929,
      "learning_rate": 3.788496240106865e-05,
      "loss": 0.8375,
      "step": 206660
    },
    {
      "epoch": 3.6283993749890273,
      "grad_norm": 0.04656848651336428,
      "learning_rate": 3.787637742155226e-05,
      "loss": 0.8444,
      "step": 206670
    },
    {
      "epoch": 3.6285749398690283,
      "grad_norm": 0.10675013013946438,
      "learning_rate": 3.7867793241824265e-05,
      "loss": 0.8371,
      "step": 206680
    },
    {
      "epoch": 3.6287505047490303,
      "grad_norm": 0.05241834011334658,
      "learning_rate": 3.7859209861993456e-05,
      "loss": 0.8408,
      "step": 206690
    },
    {
      "epoch": 3.6289260696290313,
      "grad_norm": 0.06895221611559156,
      "learning_rate": 3.7850627282168535e-05,
      "loss": 0.8474,
      "step": 206700
    },
    {
      "epoch": 3.629101634509033,
      "grad_norm": 0.043235798537422626,
      "learning_rate": 3.78420455024583e-05,
      "loss": 0.843,
      "step": 206710
    },
    {
      "epoch": 3.6292771993890343,
      "grad_norm": 0.04945948022728828,
      "learning_rate": 3.783346452297137e-05,
      "loss": 0.8376,
      "step": 206720
    },
    {
      "epoch": 3.6294527642690357,
      "grad_norm": 0.04940295692070118,
      "learning_rate": 3.7824884343816644e-05,
      "loss": 0.8366,
      "step": 206730
    },
    {
      "epoch": 3.629628329149037,
      "grad_norm": 0.04892515547978079,
      "learning_rate": 3.781630496510269e-05,
      "loss": 0.8378,
      "step": 206740
    },
    {
      "epoch": 3.6298038940290382,
      "grad_norm": 0.06207454861361658,
      "learning_rate": 3.780772638693826e-05,
      "loss": 0.8378,
      "step": 206750
    },
    {
      "epoch": 3.6299794589090397,
      "grad_norm": 0.06047101949372692,
      "learning_rate": 3.7799148609432094e-05,
      "loss": 0.839,
      "step": 206760
    },
    {
      "epoch": 3.630155023789041,
      "grad_norm": 0.043041360592303395,
      "learning_rate": 3.779057163269279e-05,
      "loss": 0.8324,
      "step": 206770
    },
    {
      "epoch": 3.6303305886690427,
      "grad_norm": 0.0445899312963184,
      "learning_rate": 3.778199545682909e-05,
      "loss": 0.8394,
      "step": 206780
    },
    {
      "epoch": 3.630506153549044,
      "grad_norm": 0.04825648469281864,
      "learning_rate": 3.7773420081949544e-05,
      "loss": 0.8402,
      "step": 206790
    },
    {
      "epoch": 3.630681718429045,
      "grad_norm": 0.047162866501062335,
      "learning_rate": 3.776484550816297e-05,
      "loss": 0.842,
      "step": 206800
    },
    {
      "epoch": 3.630857283309047,
      "grad_norm": 0.06095831120279771,
      "learning_rate": 3.775627173557791e-05,
      "loss": 0.8298,
      "step": 206810
    },
    {
      "epoch": 3.631032848189048,
      "grad_norm": 0.049419346917703985,
      "learning_rate": 3.774769876430293e-05,
      "loss": 0.8426,
      "step": 206820
    },
    {
      "epoch": 3.6312084130690496,
      "grad_norm": 0.045562119848599196,
      "learning_rate": 3.773912659444682e-05,
      "loss": 0.8419,
      "step": 206830
    },
    {
      "epoch": 3.631383977949051,
      "grad_norm": 0.05322112391395556,
      "learning_rate": 3.773055522611805e-05,
      "loss": 0.8382,
      "step": 206840
    },
    {
      "epoch": 3.6315595428290526,
      "grad_norm": 0.06718791561213731,
      "learning_rate": 3.772198465942528e-05,
      "loss": 0.8389,
      "step": 206850
    },
    {
      "epoch": 3.631735107709054,
      "grad_norm": 0.05563110962706107,
      "learning_rate": 3.771341489447712e-05,
      "loss": 0.8339,
      "step": 206860
    },
    {
      "epoch": 3.631910672589055,
      "grad_norm": 0.06366855094173428,
      "learning_rate": 3.770484593138209e-05,
      "loss": 0.8418,
      "step": 206870
    },
    {
      "epoch": 3.6320862374690566,
      "grad_norm": 0.06364165464547106,
      "learning_rate": 3.7696277770248825e-05,
      "loss": 0.8257,
      "step": 206880
    },
    {
      "epoch": 3.632261802349058,
      "grad_norm": 0.05314683981206108,
      "learning_rate": 3.76877104111858e-05,
      "loss": 0.8414,
      "step": 206890
    },
    {
      "epoch": 3.6324373672290595,
      "grad_norm": 0.0433636091864561,
      "learning_rate": 3.7679143854301656e-05,
      "loss": 0.8429,
      "step": 206900
    },
    {
      "epoch": 3.632612932109061,
      "grad_norm": 0.06528285726070457,
      "learning_rate": 3.767057809970488e-05,
      "loss": 0.8422,
      "step": 206910
    },
    {
      "epoch": 3.632788496989062,
      "grad_norm": 0.05502964623761306,
      "learning_rate": 3.766201314750402e-05,
      "loss": 0.8413,
      "step": 206920
    },
    {
      "epoch": 3.632964061869064,
      "grad_norm": 0.05621620277537167,
      "learning_rate": 3.7653448997807616e-05,
      "loss": 0.8376,
      "step": 206930
    },
    {
      "epoch": 3.633139626749065,
      "grad_norm": 0.0482714620927815,
      "learning_rate": 3.764488565072412e-05,
      "loss": 0.8387,
      "step": 206940
    },
    {
      "epoch": 3.6333151916290665,
      "grad_norm": 0.05576458963408952,
      "learning_rate": 3.7636323106362104e-05,
      "loss": 0.8433,
      "step": 206950
    },
    {
      "epoch": 3.633490756509068,
      "grad_norm": 0.06066924050985779,
      "learning_rate": 3.7627761364829944e-05,
      "loss": 0.8334,
      "step": 206960
    },
    {
      "epoch": 3.6336663213890694,
      "grad_norm": 0.05475117773966392,
      "learning_rate": 3.761920042623627e-05,
      "loss": 0.8428,
      "step": 206970
    },
    {
      "epoch": 3.633841886269071,
      "grad_norm": 0.057105337826052795,
      "learning_rate": 3.761064029068947e-05,
      "loss": 0.834,
      "step": 206980
    },
    {
      "epoch": 3.634017451149072,
      "grad_norm": 0.05980819069561032,
      "learning_rate": 3.760208095829793e-05,
      "loss": 0.8407,
      "step": 206990
    },
    {
      "epoch": 3.6341930160290734,
      "grad_norm": 0.04283269203300795,
      "learning_rate": 3.7593522429170255e-05,
      "loss": 0.8441,
      "step": 207000
    },
    {
      "epoch": 3.634368580909075,
      "grad_norm": 0.05278019825148156,
      "learning_rate": 3.758496470341475e-05,
      "loss": 0.8467,
      "step": 207010
    },
    {
      "epoch": 3.6345441457890764,
      "grad_norm": 0.04402144815741203,
      "learning_rate": 3.757640778113992e-05,
      "loss": 0.8409,
      "step": 207020
    },
    {
      "epoch": 3.634719710669078,
      "grad_norm": 0.07039610896902142,
      "learning_rate": 3.756785166245417e-05,
      "loss": 0.8337,
      "step": 207030
    },
    {
      "epoch": 3.6348952755490793,
      "grad_norm": 0.05430228694627302,
      "learning_rate": 3.755929634746588e-05,
      "loss": 0.838,
      "step": 207040
    },
    {
      "epoch": 3.635070840429081,
      "grad_norm": 0.06423375643508548,
      "learning_rate": 3.7550741836283474e-05,
      "loss": 0.8406,
      "step": 207050
    },
    {
      "epoch": 3.635246405309082,
      "grad_norm": 0.047374250098700235,
      "learning_rate": 3.754218812901529e-05,
      "loss": 0.8402,
      "step": 207060
    },
    {
      "epoch": 3.6354219701890833,
      "grad_norm": 0.05503793898805805,
      "learning_rate": 3.753363522576979e-05,
      "loss": 0.844,
      "step": 207070
    },
    {
      "epoch": 3.635597535069085,
      "grad_norm": 0.04998968803100997,
      "learning_rate": 3.7525083126655276e-05,
      "loss": 0.841,
      "step": 207080
    },
    {
      "epoch": 3.6357730999490863,
      "grad_norm": 0.062244397700354094,
      "learning_rate": 3.751653183178011e-05,
      "loss": 0.837,
      "step": 207090
    },
    {
      "epoch": 3.6359486648290877,
      "grad_norm": 0.04888387207477947,
      "learning_rate": 3.75079813412527e-05,
      "loss": 0.8461,
      "step": 207100
    },
    {
      "epoch": 3.6361242297090888,
      "grad_norm": 0.04129128475053713,
      "learning_rate": 3.749943165518129e-05,
      "loss": 0.8433,
      "step": 207110
    },
    {
      "epoch": 3.6362997945890903,
      "grad_norm": 0.07497712841007043,
      "learning_rate": 3.749088277367429e-05,
      "loss": 0.8447,
      "step": 207120
    },
    {
      "epoch": 3.6364753594690917,
      "grad_norm": 0.05086545998171429,
      "learning_rate": 3.748233469683991e-05,
      "loss": 0.8384,
      "step": 207130
    },
    {
      "epoch": 3.636650924349093,
      "grad_norm": 0.04836266065871575,
      "learning_rate": 3.747378742478659e-05,
      "loss": 0.8464,
      "step": 207140
    },
    {
      "epoch": 3.6368264892290947,
      "grad_norm": 0.05105571221958425,
      "learning_rate": 3.746524095762255e-05,
      "loss": 0.8325,
      "step": 207150
    },
    {
      "epoch": 3.637002054109096,
      "grad_norm": 0.061926528730138326,
      "learning_rate": 3.745669529545603e-05,
      "loss": 0.8295,
      "step": 207160
    },
    {
      "epoch": 3.6371776189890976,
      "grad_norm": 0.04662248523986968,
      "learning_rate": 3.7448150438395415e-05,
      "loss": 0.8403,
      "step": 207170
    },
    {
      "epoch": 3.6373531838690987,
      "grad_norm": 0.05603760566279675,
      "learning_rate": 3.7439606386548876e-05,
      "loss": 0.8377,
      "step": 207180
    },
    {
      "epoch": 3.6375287487491,
      "grad_norm": 0.05331352910409461,
      "learning_rate": 3.743106314002472e-05,
      "loss": 0.8435,
      "step": 207190
    },
    {
      "epoch": 3.6377043136291016,
      "grad_norm": 0.04891343311418888,
      "learning_rate": 3.74225206989312e-05,
      "loss": 0.8384,
      "step": 207200
    },
    {
      "epoch": 3.637879878509103,
      "grad_norm": 0.05659100900083427,
      "learning_rate": 3.7413979063376496e-05,
      "loss": 0.8437,
      "step": 207210
    },
    {
      "epoch": 3.6380554433891046,
      "grad_norm": 0.04514120848182934,
      "learning_rate": 3.74054382334689e-05,
      "loss": 0.8434,
      "step": 207220
    },
    {
      "epoch": 3.6382310082691056,
      "grad_norm": 0.052072440720698394,
      "learning_rate": 3.7396898209316514e-05,
      "loss": 0.8426,
      "step": 207230
    },
    {
      "epoch": 3.638406573149107,
      "grad_norm": 0.045519624334827466,
      "learning_rate": 3.73883589910277e-05,
      "loss": 0.8383,
      "step": 207240
    },
    {
      "epoch": 3.6385821380291086,
      "grad_norm": 0.07305785938916413,
      "learning_rate": 3.737982057871052e-05,
      "loss": 0.8459,
      "step": 207250
    },
    {
      "epoch": 3.63875770290911,
      "grad_norm": 0.0509833639454058,
      "learning_rate": 3.737128297247321e-05,
      "loss": 0.8403,
      "step": 207260
    },
    {
      "epoch": 3.6389332677891115,
      "grad_norm": 0.058771976698932374,
      "learning_rate": 3.736274617242397e-05,
      "loss": 0.8415,
      "step": 207270
    },
    {
      "epoch": 3.639108832669113,
      "grad_norm": 0.059192640781358545,
      "learning_rate": 3.7354210178670903e-05,
      "loss": 0.8383,
      "step": 207280
    },
    {
      "epoch": 3.6392843975491145,
      "grad_norm": 0.048446883583668915,
      "learning_rate": 3.734567499132223e-05,
      "loss": 0.8445,
      "step": 207290
    },
    {
      "epoch": 3.6394599624291155,
      "grad_norm": 0.04742567564614266,
      "learning_rate": 3.7337140610485974e-05,
      "loss": 0.8341,
      "step": 207300
    },
    {
      "epoch": 3.639635527309117,
      "grad_norm": 0.05951340477158117,
      "learning_rate": 3.732860703627042e-05,
      "loss": 0.8508,
      "step": 207310
    },
    {
      "epoch": 3.6398110921891185,
      "grad_norm": 0.04967577794841141,
      "learning_rate": 3.732007426878362e-05,
      "loss": 0.8454,
      "step": 207320
    },
    {
      "epoch": 3.63998665706912,
      "grad_norm": 0.06775774676177379,
      "learning_rate": 3.731154230813361e-05,
      "loss": 0.8418,
      "step": 207330
    },
    {
      "epoch": 3.6401622219491214,
      "grad_norm": 0.05306842214266816,
      "learning_rate": 3.730301115442864e-05,
      "loss": 0.8423,
      "step": 207340
    },
    {
      "epoch": 3.6403377868291225,
      "grad_norm": 0.04960942986136163,
      "learning_rate": 3.729448080777669e-05,
      "loss": 0.8428,
      "step": 207350
    },
    {
      "epoch": 3.6405133517091244,
      "grad_norm": 0.0718637261927499,
      "learning_rate": 3.728595126828588e-05,
      "loss": 0.8298,
      "step": 207360
    },
    {
      "epoch": 3.6406889165891254,
      "grad_norm": 0.055136435431068456,
      "learning_rate": 3.727742253606431e-05,
      "loss": 0.8449,
      "step": 207370
    },
    {
      "epoch": 3.640864481469127,
      "grad_norm": 0.05205667573311479,
      "learning_rate": 3.726889461121997e-05,
      "loss": 0.8444,
      "step": 207380
    },
    {
      "epoch": 3.6410400463491284,
      "grad_norm": 0.06799965277889375,
      "learning_rate": 3.7260367493860984e-05,
      "loss": 0.8354,
      "step": 207390
    },
    {
      "epoch": 3.64121561122913,
      "grad_norm": 0.050503526102781124,
      "learning_rate": 3.72518411840953e-05,
      "loss": 0.8421,
      "step": 207400
    },
    {
      "epoch": 3.6413911761091313,
      "grad_norm": 0.041889422020677344,
      "learning_rate": 3.724331568203107e-05,
      "loss": 0.8407,
      "step": 207410
    },
    {
      "epoch": 3.6415667409891324,
      "grad_norm": 0.04784572330528736,
      "learning_rate": 3.723479098777621e-05,
      "loss": 0.8379,
      "step": 207420
    },
    {
      "epoch": 3.641742305869134,
      "grad_norm": 0.0551437536951535,
      "learning_rate": 3.7226267101438764e-05,
      "loss": 0.8447,
      "step": 207430
    },
    {
      "epoch": 3.6419178707491353,
      "grad_norm": 0.04094951266987548,
      "learning_rate": 3.721774402312678e-05,
      "loss": 0.8437,
      "step": 207440
    },
    {
      "epoch": 3.642093435629137,
      "grad_norm": 0.04909469916016027,
      "learning_rate": 3.720922175294816e-05,
      "loss": 0.8438,
      "step": 207450
    },
    {
      "epoch": 3.6422690005091383,
      "grad_norm": 0.0516606072225462,
      "learning_rate": 3.720070029101096e-05,
      "loss": 0.84,
      "step": 207460
    },
    {
      "epoch": 3.6424445653891393,
      "grad_norm": 0.04894456276118399,
      "learning_rate": 3.719217963742305e-05,
      "loss": 0.8435,
      "step": 207470
    },
    {
      "epoch": 3.642620130269141,
      "grad_norm": 0.04570873504660865,
      "learning_rate": 3.718365979229253e-05,
      "loss": 0.8301,
      "step": 207480
    },
    {
      "epoch": 3.6427956951491423,
      "grad_norm": 0.054550617134065595,
      "learning_rate": 3.717514075572725e-05,
      "loss": 0.8399,
      "step": 207490
    },
    {
      "epoch": 3.6429712600291437,
      "grad_norm": 0.043413964279711004,
      "learning_rate": 3.7166622527835125e-05,
      "loss": 0.847,
      "step": 207500
    },
    {
      "epoch": 3.643146824909145,
      "grad_norm": 0.04419167820603915,
      "learning_rate": 3.7158105108724186e-05,
      "loss": 0.8394,
      "step": 207510
    },
    {
      "epoch": 3.6433223897891467,
      "grad_norm": 0.05765268571678897,
      "learning_rate": 3.714958849850225e-05,
      "loss": 0.8459,
      "step": 207520
    },
    {
      "epoch": 3.643497954669148,
      "grad_norm": 0.05791250316344679,
      "learning_rate": 3.71410726972773e-05,
      "loss": 0.8403,
      "step": 207530
    },
    {
      "epoch": 3.643673519549149,
      "grad_norm": 0.05018405305979568,
      "learning_rate": 3.7132557705157165e-05,
      "loss": 0.8425,
      "step": 207540
    },
    {
      "epoch": 3.6438490844291507,
      "grad_norm": 0.06458812747505101,
      "learning_rate": 3.712404352224976e-05,
      "loss": 0.8388,
      "step": 207550
    },
    {
      "epoch": 3.644024649309152,
      "grad_norm": 0.0457438418267871,
      "learning_rate": 3.7115530148663e-05,
      "loss": 0.8373,
      "step": 207560
    },
    {
      "epoch": 3.6442002141891536,
      "grad_norm": 0.06066990853705161,
      "learning_rate": 3.7107017584504656e-05,
      "loss": 0.8333,
      "step": 207570
    },
    {
      "epoch": 3.644375779069155,
      "grad_norm": 0.04579613667671043,
      "learning_rate": 3.709850582988271e-05,
      "loss": 0.8301,
      "step": 207580
    },
    {
      "epoch": 3.644551343949156,
      "grad_norm": 0.07007426837512568,
      "learning_rate": 3.7089994884904904e-05,
      "loss": 0.8356,
      "step": 207590
    },
    {
      "epoch": 3.644726908829158,
      "grad_norm": 0.05385638596104467,
      "learning_rate": 3.708148474967912e-05,
      "loss": 0.8351,
      "step": 207600
    },
    {
      "epoch": 3.644902473709159,
      "grad_norm": 0.08359913164077053,
      "learning_rate": 3.7072975424313193e-05,
      "loss": 0.8402,
      "step": 207610
    },
    {
      "epoch": 3.6450780385891606,
      "grad_norm": 0.06448518563558289,
      "learning_rate": 3.706446690891489e-05,
      "loss": 0.8438,
      "step": 207620
    },
    {
      "epoch": 3.645253603469162,
      "grad_norm": 0.049552692419663986,
      "learning_rate": 3.705595920359208e-05,
      "loss": 0.8471,
      "step": 207630
    },
    {
      "epoch": 3.6454291683491635,
      "grad_norm": 0.05717464503449682,
      "learning_rate": 3.704745230845245e-05,
      "loss": 0.8395,
      "step": 207640
    },
    {
      "epoch": 3.645604733229165,
      "grad_norm": 0.05321924917936218,
      "learning_rate": 3.703894622360394e-05,
      "loss": 0.8409,
      "step": 207650
    },
    {
      "epoch": 3.645780298109166,
      "grad_norm": 0.06124325868479975,
      "learning_rate": 3.703044094915422e-05,
      "loss": 0.8307,
      "step": 207660
    },
    {
      "epoch": 3.6459558629891675,
      "grad_norm": 0.05517198255433807,
      "learning_rate": 3.702193648521102e-05,
      "loss": 0.8443,
      "step": 207670
    },
    {
      "epoch": 3.646131427869169,
      "grad_norm": 0.05714146477168648,
      "learning_rate": 3.70134328318822e-05,
      "loss": 0.8433,
      "step": 207680
    },
    {
      "epoch": 3.6463069927491705,
      "grad_norm": 0.04551001661291158,
      "learning_rate": 3.700492998927543e-05,
      "loss": 0.8448,
      "step": 207690
    },
    {
      "epoch": 3.646482557629172,
      "grad_norm": 0.06249414452714749,
      "learning_rate": 3.699642795749848e-05,
      "loss": 0.8363,
      "step": 207700
    },
    {
      "epoch": 3.6466581225091734,
      "grad_norm": 0.07639409020859114,
      "learning_rate": 3.6987926736659035e-05,
      "loss": 0.8419,
      "step": 207710
    },
    {
      "epoch": 3.646833687389175,
      "grad_norm": 0.05342337842743454,
      "learning_rate": 3.697942632686481e-05,
      "loss": 0.8325,
      "step": 207720
    },
    {
      "epoch": 3.647009252269176,
      "grad_norm": 0.04933454452230972,
      "learning_rate": 3.6970926728223566e-05,
      "loss": 0.8416,
      "step": 207730
    },
    {
      "epoch": 3.6471848171491774,
      "grad_norm": 0.04728596775590516,
      "learning_rate": 3.6962427940842894e-05,
      "loss": 0.8358,
      "step": 207740
    },
    {
      "epoch": 3.647360382029179,
      "grad_norm": 0.0471377770321644,
      "learning_rate": 3.6953929964830576e-05,
      "loss": 0.8389,
      "step": 207750
    },
    {
      "epoch": 3.6475359469091804,
      "grad_norm": 0.058753210529965426,
      "learning_rate": 3.694543280029422e-05,
      "loss": 0.8378,
      "step": 207760
    },
    {
      "epoch": 3.647711511789182,
      "grad_norm": 0.05261366440297615,
      "learning_rate": 3.6936936447341505e-05,
      "loss": 0.8414,
      "step": 207770
    },
    {
      "epoch": 3.647887076669183,
      "grad_norm": 0.04715569162594859,
      "learning_rate": 3.692844090608012e-05,
      "loss": 0.8392,
      "step": 207780
    },
    {
      "epoch": 3.6480626415491844,
      "grad_norm": 0.044951555915934256,
      "learning_rate": 3.691994617661763e-05,
      "loss": 0.8448,
      "step": 207790
    },
    {
      "epoch": 3.648238206429186,
      "grad_norm": 0.0520457009921428,
      "learning_rate": 3.691145225906174e-05,
      "loss": 0.84,
      "step": 207800
    },
    {
      "epoch": 3.6484137713091873,
      "grad_norm": 0.05704691414645937,
      "learning_rate": 3.690295915351997e-05,
      "loss": 0.8464,
      "step": 207810
    },
    {
      "epoch": 3.648589336189189,
      "grad_norm": 0.05572295058819523,
      "learning_rate": 3.689446686010005e-05,
      "loss": 0.8444,
      "step": 207820
    },
    {
      "epoch": 3.6487649010691903,
      "grad_norm": 0.04966330813133378,
      "learning_rate": 3.688597537890952e-05,
      "loss": 0.8364,
      "step": 207830
    },
    {
      "epoch": 3.6489404659491917,
      "grad_norm": 0.05232230674040612,
      "learning_rate": 3.687748471005591e-05,
      "loss": 0.846,
      "step": 207840
    },
    {
      "epoch": 3.649116030829193,
      "grad_norm": 0.06356125630718655,
      "learning_rate": 3.686899485364692e-05,
      "loss": 0.8456,
      "step": 207850
    },
    {
      "epoch": 3.6492915957091943,
      "grad_norm": 0.045060897979728405,
      "learning_rate": 3.6860505809790024e-05,
      "loss": 0.8441,
      "step": 207860
    },
    {
      "epoch": 3.6494671605891957,
      "grad_norm": 0.05674898152878474,
      "learning_rate": 3.685201757859284e-05,
      "loss": 0.8381,
      "step": 207870
    },
    {
      "epoch": 3.649642725469197,
      "grad_norm": 0.04458809466164217,
      "learning_rate": 3.684353016016287e-05,
      "loss": 0.8408,
      "step": 207880
    },
    {
      "epoch": 3.6498182903491987,
      "grad_norm": 0.05381535285125529,
      "learning_rate": 3.683504355460764e-05,
      "loss": 0.8396,
      "step": 207890
    },
    {
      "epoch": 3.6499938552291997,
      "grad_norm": 0.038697282479788986,
      "learning_rate": 3.682655776203475e-05,
      "loss": 0.8521,
      "step": 207900
    },
    {
      "epoch": 3.650169420109201,
      "grad_norm": 0.0729200379942147,
      "learning_rate": 3.6818072782551625e-05,
      "loss": 0.8382,
      "step": 207910
    },
    {
      "epoch": 3.6503449849892027,
      "grad_norm": 0.06782085971985817,
      "learning_rate": 3.6809588616265856e-05,
      "loss": 0.8365,
      "step": 207920
    },
    {
      "epoch": 3.650520549869204,
      "grad_norm": 0.07162589155073502,
      "learning_rate": 3.6801105263284894e-05,
      "loss": 0.8475,
      "step": 207930
    },
    {
      "epoch": 3.6506961147492056,
      "grad_norm": 0.05038640687444504,
      "learning_rate": 3.6792622723716217e-05,
      "loss": 0.8452,
      "step": 207940
    },
    {
      "epoch": 3.650871679629207,
      "grad_norm": 0.057077505325780035,
      "learning_rate": 3.6784140997667344e-05,
      "loss": 0.8401,
      "step": 207950
    },
    {
      "epoch": 3.6510472445092086,
      "grad_norm": 0.07501665900011711,
      "learning_rate": 3.6775660085245685e-05,
      "loss": 0.8359,
      "step": 207960
    },
    {
      "epoch": 3.6512228093892096,
      "grad_norm": 0.05398682937908255,
      "learning_rate": 3.676717998655876e-05,
      "loss": 0.842,
      "step": 207970
    },
    {
      "epoch": 3.651398374269211,
      "grad_norm": 0.048477727575171756,
      "learning_rate": 3.6758700701713906e-05,
      "loss": 0.8326,
      "step": 207980
    },
    {
      "epoch": 3.6515739391492126,
      "grad_norm": 0.050545255402797806,
      "learning_rate": 3.675022223081869e-05,
      "loss": 0.8398,
      "step": 207990
    },
    {
      "epoch": 3.651749504029214,
      "grad_norm": 0.05535414293461249,
      "learning_rate": 3.674174457398047e-05,
      "loss": 0.8451,
      "step": 208000
    },
    {
      "epoch": 3.6519250689092155,
      "grad_norm": 0.06010722368631157,
      "learning_rate": 3.673326773130659e-05,
      "loss": 0.8522,
      "step": 208010
    },
    {
      "epoch": 3.6521006337892166,
      "grad_norm": 0.04745787697738982,
      "learning_rate": 3.6724791702904606e-05,
      "loss": 0.8428,
      "step": 208020
    },
    {
      "epoch": 3.652276198669218,
      "grad_norm": 0.05797526261801778,
      "learning_rate": 3.671631648888178e-05,
      "loss": 0.8385,
      "step": 208030
    },
    {
      "epoch": 3.6524517635492195,
      "grad_norm": 0.05117262950917112,
      "learning_rate": 3.670784208934559e-05,
      "loss": 0.84,
      "step": 208040
    },
    {
      "epoch": 3.652627328429221,
      "grad_norm": 0.06966015755943702,
      "learning_rate": 3.669936850440331e-05,
      "loss": 0.8374,
      "step": 208050
    },
    {
      "epoch": 3.6528028933092225,
      "grad_norm": 0.04258384030799303,
      "learning_rate": 3.6690895734162376e-05,
      "loss": 0.8332,
      "step": 208060
    },
    {
      "epoch": 3.652978458189224,
      "grad_norm": 0.04831305024540532,
      "learning_rate": 3.668242377873014e-05,
      "loss": 0.85,
      "step": 208070
    },
    {
      "epoch": 3.6531540230692254,
      "grad_norm": 0.055906127433624615,
      "learning_rate": 3.667395263821386e-05,
      "loss": 0.8394,
      "step": 208080
    },
    {
      "epoch": 3.6533295879492265,
      "grad_norm": 0.05358729994167299,
      "learning_rate": 3.666548231272099e-05,
      "loss": 0.8438,
      "step": 208090
    },
    {
      "epoch": 3.653505152829228,
      "grad_norm": 0.04477733904239037,
      "learning_rate": 3.665701280235876e-05,
      "loss": 0.8481,
      "step": 208100
    },
    {
      "epoch": 3.6536807177092294,
      "grad_norm": 0.07407789356933361,
      "learning_rate": 3.66485441072345e-05,
      "loss": 0.841,
      "step": 208110
    },
    {
      "epoch": 3.653856282589231,
      "grad_norm": 0.046454584462398286,
      "learning_rate": 3.664007622745556e-05,
      "loss": 0.8433,
      "step": 208120
    },
    {
      "epoch": 3.6540318474692324,
      "grad_norm": 0.05199743267115952,
      "learning_rate": 3.6631609163129155e-05,
      "loss": 0.8445,
      "step": 208130
    },
    {
      "epoch": 3.6542074123492334,
      "grad_norm": 0.05292160078304449,
      "learning_rate": 3.6623142914362634e-05,
      "loss": 0.8354,
      "step": 208140
    },
    {
      "epoch": 3.6543829772292353,
      "grad_norm": 0.04879994080188687,
      "learning_rate": 3.661467748126316e-05,
      "loss": 0.8373,
      "step": 208150
    },
    {
      "epoch": 3.6545585421092364,
      "grad_norm": 0.07496208931159937,
      "learning_rate": 3.660621286393813e-05,
      "loss": 0.8399,
      "step": 208160
    },
    {
      "epoch": 3.654734106989238,
      "grad_norm": 0.04452517595804591,
      "learning_rate": 3.659774906249472e-05,
      "loss": 0.8468,
      "step": 208170
    },
    {
      "epoch": 3.6549096718692393,
      "grad_norm": 0.05967222932536949,
      "learning_rate": 3.6589286077040114e-05,
      "loss": 0.8367,
      "step": 208180
    },
    {
      "epoch": 3.655085236749241,
      "grad_norm": 0.05381750289073436,
      "learning_rate": 3.6580823907681655e-05,
      "loss": 0.8436,
      "step": 208190
    },
    {
      "epoch": 3.6552608016292423,
      "grad_norm": 0.050067816937429684,
      "learning_rate": 3.6572362554526475e-05,
      "loss": 0.8388,
      "step": 208200
    },
    {
      "epoch": 3.6554363665092433,
      "grad_norm": 0.05188376114642477,
      "learning_rate": 3.6563902017681836e-05,
      "loss": 0.8354,
      "step": 208210
    },
    {
      "epoch": 3.655611931389245,
      "grad_norm": 0.04598389192530864,
      "learning_rate": 3.6555442297254884e-05,
      "loss": 0.8433,
      "step": 208220
    },
    {
      "epoch": 3.6557874962692463,
      "grad_norm": 0.044667311270740834,
      "learning_rate": 3.6546983393352825e-05,
      "loss": 0.8389,
      "step": 208230
    },
    {
      "epoch": 3.6559630611492477,
      "grad_norm": 0.0620876979661096,
      "learning_rate": 3.653852530608287e-05,
      "loss": 0.8375,
      "step": 208240
    },
    {
      "epoch": 3.656138626029249,
      "grad_norm": 0.07200496211781039,
      "learning_rate": 3.653006803555208e-05,
      "loss": 0.8369,
      "step": 208250
    },
    {
      "epoch": 3.6563141909092503,
      "grad_norm": 0.06310951323834026,
      "learning_rate": 3.6521611581867755e-05,
      "loss": 0.8415,
      "step": 208260
    },
    {
      "epoch": 3.656489755789252,
      "grad_norm": 0.04671926052530355,
      "learning_rate": 3.651315594513693e-05,
      "loss": 0.8453,
      "step": 208270
    },
    {
      "epoch": 3.656665320669253,
      "grad_norm": 0.04679221290179048,
      "learning_rate": 3.6504701125466795e-05,
      "loss": 0.8405,
      "step": 208280
    },
    {
      "epoch": 3.6568408855492547,
      "grad_norm": 0.044634406139705346,
      "learning_rate": 3.649624712296448e-05,
      "loss": 0.8427,
      "step": 208290
    },
    {
      "epoch": 3.657016450429256,
      "grad_norm": 0.06740093183323445,
      "learning_rate": 3.6487793937737044e-05,
      "loss": 0.8486,
      "step": 208300
    },
    {
      "epoch": 3.6571920153092576,
      "grad_norm": 0.06450459700017051,
      "learning_rate": 3.647934156989165e-05,
      "loss": 0.84,
      "step": 208310
    },
    {
      "epoch": 3.657367580189259,
      "grad_norm": 0.05818036734951797,
      "learning_rate": 3.6470890019535307e-05,
      "loss": 0.8468,
      "step": 208320
    },
    {
      "epoch": 3.65754314506926,
      "grad_norm": 0.053287390725955536,
      "learning_rate": 3.646243928677522e-05,
      "loss": 0.8353,
      "step": 208330
    },
    {
      "epoch": 3.6577187099492616,
      "grad_norm": 0.0512535714779883,
      "learning_rate": 3.645398937171839e-05,
      "loss": 0.8419,
      "step": 208340
    },
    {
      "epoch": 3.657894274829263,
      "grad_norm": 0.04573258798853829,
      "learning_rate": 3.644554027447182e-05,
      "loss": 0.841,
      "step": 208350
    },
    {
      "epoch": 3.6580698397092646,
      "grad_norm": 0.04223633725473875,
      "learning_rate": 3.643709199514269e-05,
      "loss": 0.8411,
      "step": 208360
    },
    {
      "epoch": 3.658245404589266,
      "grad_norm": 0.059951669324911624,
      "learning_rate": 3.6428644533837954e-05,
      "loss": 0.8352,
      "step": 208370
    },
    {
      "epoch": 3.658420969469267,
      "grad_norm": 0.06859793157145337,
      "learning_rate": 3.642019789066469e-05,
      "loss": 0.8371,
      "step": 208380
    },
    {
      "epoch": 3.658596534349269,
      "grad_norm": 0.05907623596167393,
      "learning_rate": 3.641175206572986e-05,
      "loss": 0.8376,
      "step": 208390
    },
    {
      "epoch": 3.65877209922927,
      "grad_norm": 0.0570502019717151,
      "learning_rate": 3.640330705914052e-05,
      "loss": 0.8421,
      "step": 208400
    },
    {
      "epoch": 3.6589476641092715,
      "grad_norm": 0.055690345786462966,
      "learning_rate": 3.63948628710037e-05,
      "loss": 0.8412,
      "step": 208410
    },
    {
      "epoch": 3.659123228989273,
      "grad_norm": 0.057166883906601754,
      "learning_rate": 3.6386419501426266e-05,
      "loss": 0.8386,
      "step": 208420
    },
    {
      "epoch": 3.6592987938692745,
      "grad_norm": 0.043737855240824686,
      "learning_rate": 3.637797695051535e-05,
      "loss": 0.8367,
      "step": 208430
    },
    {
      "epoch": 3.659474358749276,
      "grad_norm": 0.05115362610634701,
      "learning_rate": 3.636953521837781e-05,
      "loss": 0.8397,
      "step": 208440
    },
    {
      "epoch": 3.659649923629277,
      "grad_norm": 0.04828477445869321,
      "learning_rate": 3.636109430512066e-05,
      "loss": 0.8375,
      "step": 208450
    },
    {
      "epoch": 3.6598254885092785,
      "grad_norm": 0.05588431551675621,
      "learning_rate": 3.6352654210850864e-05,
      "loss": 0.8419,
      "step": 208460
    },
    {
      "epoch": 3.66000105338928,
      "grad_norm": 0.05148097592621106,
      "learning_rate": 3.63442149356753e-05,
      "loss": 0.8393,
      "step": 208470
    },
    {
      "epoch": 3.6601766182692814,
      "grad_norm": 0.08544739427833326,
      "learning_rate": 3.633577647970095e-05,
      "loss": 0.8382,
      "step": 208480
    },
    {
      "epoch": 3.660352183149283,
      "grad_norm": 0.05593346554677629,
      "learning_rate": 3.6327338843034624e-05,
      "loss": 0.836,
      "step": 208490
    },
    {
      "epoch": 3.6605277480292844,
      "grad_norm": 0.04577013254973488,
      "learning_rate": 3.63189020257834e-05,
      "loss": 0.8416,
      "step": 208500
    },
    {
      "epoch": 3.660703312909286,
      "grad_norm": 0.06254390144070464,
      "learning_rate": 3.631046602805407e-05,
      "loss": 0.846,
      "step": 208510
    },
    {
      "epoch": 3.660878877789287,
      "grad_norm": 0.0478976001704165,
      "learning_rate": 3.630203084995347e-05,
      "loss": 0.8385,
      "step": 208520
    },
    {
      "epoch": 3.6610544426692884,
      "grad_norm": 0.05192258910685251,
      "learning_rate": 3.6293596491588605e-05,
      "loss": 0.8396,
      "step": 208530
    },
    {
      "epoch": 3.66123000754929,
      "grad_norm": 0.05888999763278196,
      "learning_rate": 3.628516295306623e-05,
      "loss": 0.8372,
      "step": 208540
    },
    {
      "epoch": 3.6614055724292913,
      "grad_norm": 0.04941437295629919,
      "learning_rate": 3.62767302344933e-05,
      "loss": 0.8371,
      "step": 208550
    },
    {
      "epoch": 3.661581137309293,
      "grad_norm": 0.0573821011538737,
      "learning_rate": 3.6268298335976555e-05,
      "loss": 0.8442,
      "step": 208560
    },
    {
      "epoch": 3.661756702189294,
      "grad_norm": 0.05313222120162089,
      "learning_rate": 3.625986725762287e-05,
      "loss": 0.8379,
      "step": 208570
    },
    {
      "epoch": 3.6619322670692953,
      "grad_norm": 0.054818415688642756,
      "learning_rate": 3.6251436999539103e-05,
      "loss": 0.8465,
      "step": 208580
    },
    {
      "epoch": 3.662107831949297,
      "grad_norm": 0.06349366387508558,
      "learning_rate": 3.624300756183199e-05,
      "loss": 0.8396,
      "step": 208590
    },
    {
      "epoch": 3.6622833968292983,
      "grad_norm": 0.058323068151535755,
      "learning_rate": 3.623457894460844e-05,
      "loss": 0.8406,
      "step": 208600
    },
    {
      "epoch": 3.6624589617092997,
      "grad_norm": 0.04964005673222531,
      "learning_rate": 3.622615114797516e-05,
      "loss": 0.835,
      "step": 208610
    },
    {
      "epoch": 3.662634526589301,
      "grad_norm": 0.045485576671676936,
      "learning_rate": 3.621772417203895e-05,
      "loss": 0.8346,
      "step": 208620
    },
    {
      "epoch": 3.6628100914693027,
      "grad_norm": 0.06825594397715319,
      "learning_rate": 3.6209298016906615e-05,
      "loss": 0.8396,
      "step": 208630
    },
    {
      "epoch": 3.6629856563493037,
      "grad_norm": 0.05969809300166471,
      "learning_rate": 3.620087268268488e-05,
      "loss": 0.8409,
      "step": 208640
    },
    {
      "epoch": 3.663161221229305,
      "grad_norm": 0.05818269527656864,
      "learning_rate": 3.6192448169480526e-05,
      "loss": 0.8423,
      "step": 208650
    },
    {
      "epoch": 3.6633367861093067,
      "grad_norm": 0.059272105255753764,
      "learning_rate": 3.6184024477400214e-05,
      "loss": 0.8454,
      "step": 208660
    },
    {
      "epoch": 3.663512350989308,
      "grad_norm": 0.06905215194339684,
      "learning_rate": 3.617560160655079e-05,
      "loss": 0.8418,
      "step": 208670
    },
    {
      "epoch": 3.6636879158693096,
      "grad_norm": 0.04401353734941677,
      "learning_rate": 3.616717955703893e-05,
      "loss": 0.8399,
      "step": 208680
    },
    {
      "epoch": 3.6638634807493107,
      "grad_norm": 0.049347951637373474,
      "learning_rate": 3.6158758328971254e-05,
      "loss": 0.8382,
      "step": 208690
    },
    {
      "epoch": 3.664039045629312,
      "grad_norm": 0.05106824972695485,
      "learning_rate": 3.615033792245461e-05,
      "loss": 0.8397,
      "step": 208700
    },
    {
      "epoch": 3.6642146105093136,
      "grad_norm": 0.061346775547312035,
      "learning_rate": 3.6141918337595575e-05,
      "loss": 0.8443,
      "step": 208710
    },
    {
      "epoch": 3.664390175389315,
      "grad_norm": 0.04813900473853539,
      "learning_rate": 3.613349957450091e-05,
      "loss": 0.8466,
      "step": 208720
    },
    {
      "epoch": 3.6645657402693166,
      "grad_norm": 0.057897858187791515,
      "learning_rate": 3.612508163327718e-05,
      "loss": 0.8392,
      "step": 208730
    },
    {
      "epoch": 3.664741305149318,
      "grad_norm": 0.06286048869780683,
      "learning_rate": 3.611666451403112e-05,
      "loss": 0.8402,
      "step": 208740
    },
    {
      "epoch": 3.6649168700293195,
      "grad_norm": 0.049828994675352706,
      "learning_rate": 3.6108248216869375e-05,
      "loss": 0.8438,
      "step": 208750
    },
    {
      "epoch": 3.6650924349093206,
      "grad_norm": 0.04430303371893461,
      "learning_rate": 3.60998327418985e-05,
      "loss": 0.8435,
      "step": 208760
    },
    {
      "epoch": 3.665267999789322,
      "grad_norm": 0.04963174857208082,
      "learning_rate": 3.609141808922526e-05,
      "loss": 0.8457,
      "step": 208770
    },
    {
      "epoch": 3.6654435646693235,
      "grad_norm": 0.08154664489137695,
      "learning_rate": 3.6083004258956143e-05,
      "loss": 0.8354,
      "step": 208780
    },
    {
      "epoch": 3.665619129549325,
      "grad_norm": 0.05066194203693672,
      "learning_rate": 3.607459125119781e-05,
      "loss": 0.8338,
      "step": 208790
    },
    {
      "epoch": 3.6657946944293265,
      "grad_norm": 0.05051109872065273,
      "learning_rate": 3.606617906605688e-05,
      "loss": 0.8424,
      "step": 208800
    },
    {
      "epoch": 3.6659702593093275,
      "grad_norm": 0.05710826119669858,
      "learning_rate": 3.605776770363987e-05,
      "loss": 0.8419,
      "step": 208810
    },
    {
      "epoch": 3.6661458241893294,
      "grad_norm": 0.052896895039653204,
      "learning_rate": 3.604935716405343e-05,
      "loss": 0.8359,
      "step": 208820
    },
    {
      "epoch": 3.6663213890693305,
      "grad_norm": 0.05666591105481921,
      "learning_rate": 3.604094744740401e-05,
      "loss": 0.8445,
      "step": 208830
    },
    {
      "epoch": 3.666496953949332,
      "grad_norm": 0.07069487347034938,
      "learning_rate": 3.6032538553798315e-05,
      "loss": 0.8365,
      "step": 208840
    },
    {
      "epoch": 3.6666725188293334,
      "grad_norm": 0.05031799261997027,
      "learning_rate": 3.602413048334279e-05,
      "loss": 0.8489,
      "step": 208850
    },
    {
      "epoch": 3.666848083709335,
      "grad_norm": 0.0462364968655923,
      "learning_rate": 3.601572323614392e-05,
      "loss": 0.8412,
      "step": 208860
    },
    {
      "epoch": 3.6670236485893364,
      "grad_norm": 0.05956316205802832,
      "learning_rate": 3.6007316812308364e-05,
      "loss": 0.8386,
      "step": 208870
    },
    {
      "epoch": 3.6671992134693374,
      "grad_norm": 0.048050309341962624,
      "learning_rate": 3.599891121194252e-05,
      "loss": 0.8448,
      "step": 208880
    },
    {
      "epoch": 3.667374778349339,
      "grad_norm": 0.06247629596303087,
      "learning_rate": 3.599050643515297e-05,
      "loss": 0.8365,
      "step": 208890
    },
    {
      "epoch": 3.6675503432293404,
      "grad_norm": 0.09732977738862077,
      "learning_rate": 3.5982102482046126e-05,
      "loss": 0.8412,
      "step": 208900
    },
    {
      "epoch": 3.667725908109342,
      "grad_norm": 0.04642005973541344,
      "learning_rate": 3.59736993527285e-05,
      "loss": 0.8417,
      "step": 208910
    },
    {
      "epoch": 3.6679014729893433,
      "grad_norm": 0.07454068649892699,
      "learning_rate": 3.59652970473066e-05,
      "loss": 0.8388,
      "step": 208920
    },
    {
      "epoch": 3.6680770378693444,
      "grad_norm": 0.05385261054386089,
      "learning_rate": 3.595689556588679e-05,
      "loss": 0.8479,
      "step": 208930
    },
    {
      "epoch": 3.6682526027493463,
      "grad_norm": 0.0510012981938576,
      "learning_rate": 3.594849490857564e-05,
      "loss": 0.8447,
      "step": 208940
    },
    {
      "epoch": 3.6684281676293473,
      "grad_norm": 0.05613786628140446,
      "learning_rate": 3.59400950754795e-05,
      "loss": 0.8383,
      "step": 208950
    },
    {
      "epoch": 3.668603732509349,
      "grad_norm": 0.0640982903858166,
      "learning_rate": 3.593169606670481e-05,
      "loss": 0.8447,
      "step": 208960
    },
    {
      "epoch": 3.6687792973893503,
      "grad_norm": 0.06008124631737059,
      "learning_rate": 3.5923297882358044e-05,
      "loss": 0.8372,
      "step": 208970
    },
    {
      "epoch": 3.6689548622693517,
      "grad_norm": 0.05550561540715296,
      "learning_rate": 3.591490052254553e-05,
      "loss": 0.8425,
      "step": 208980
    },
    {
      "epoch": 3.6691304271493532,
      "grad_norm": 0.06694942016244387,
      "learning_rate": 3.590650398737374e-05,
      "loss": 0.8409,
      "step": 208990
    },
    {
      "epoch": 3.6693059920293543,
      "grad_norm": 0.05095930210256873,
      "learning_rate": 3.5898108276948945e-05,
      "loss": 0.8311,
      "step": 209000
    },
    {
      "epoch": 3.6694815569093557,
      "grad_norm": 0.04966062436683267,
      "learning_rate": 3.5889713391377664e-05,
      "loss": 0.8494,
      "step": 209010
    },
    {
      "epoch": 3.669657121789357,
      "grad_norm": 0.06506234333678754,
      "learning_rate": 3.5881319330766186e-05,
      "loss": 0.8377,
      "step": 209020
    },
    {
      "epoch": 3.6698326866693587,
      "grad_norm": 0.049012715818810594,
      "learning_rate": 3.587292609522081e-05,
      "loss": 0.8421,
      "step": 209030
    },
    {
      "epoch": 3.67000825154936,
      "grad_norm": 0.051358718061736204,
      "learning_rate": 3.586453368484802e-05,
      "loss": 0.8384,
      "step": 209040
    },
    {
      "epoch": 3.670183816429361,
      "grad_norm": 0.049941857563207966,
      "learning_rate": 3.5856142099754025e-05,
      "loss": 0.8403,
      "step": 209050
    },
    {
      "epoch": 3.670359381309363,
      "grad_norm": 0.05451926941047834,
      "learning_rate": 3.5847751340045224e-05,
      "loss": 0.8391,
      "step": 209060
    },
    {
      "epoch": 3.670534946189364,
      "grad_norm": 0.05350046726385339,
      "learning_rate": 3.583936140582788e-05,
      "loss": 0.8386,
      "step": 209070
    },
    {
      "epoch": 3.6707105110693656,
      "grad_norm": 0.05782311890564411,
      "learning_rate": 3.583097229720831e-05,
      "loss": 0.8346,
      "step": 209080
    },
    {
      "epoch": 3.670886075949367,
      "grad_norm": 0.06678645120959097,
      "learning_rate": 3.5822584014292835e-05,
      "loss": 0.8373,
      "step": 209090
    },
    {
      "epoch": 3.6710616408293686,
      "grad_norm": 0.05190534905949794,
      "learning_rate": 3.5814196557187654e-05,
      "loss": 0.8362,
      "step": 209100
    },
    {
      "epoch": 3.67123720570937,
      "grad_norm": 0.04699605688914498,
      "learning_rate": 3.580580992599916e-05,
      "loss": 0.8437,
      "step": 209110
    },
    {
      "epoch": 3.671412770589371,
      "grad_norm": 0.05822669496876252,
      "learning_rate": 3.579742412083352e-05,
      "loss": 0.8395,
      "step": 209120
    },
    {
      "epoch": 3.6715883354693726,
      "grad_norm": 0.05436052138408788,
      "learning_rate": 3.5789039141797e-05,
      "loss": 0.8438,
      "step": 209130
    },
    {
      "epoch": 3.671763900349374,
      "grad_norm": 0.05217609259126647,
      "learning_rate": 3.57806549889959e-05,
      "loss": 0.8442,
      "step": 209140
    },
    {
      "epoch": 3.6719394652293755,
      "grad_norm": 0.054378250190861926,
      "learning_rate": 3.5772271662536354e-05,
      "loss": 0.8397,
      "step": 209150
    },
    {
      "epoch": 3.672115030109377,
      "grad_norm": 0.045113726797002394,
      "learning_rate": 3.576388916252467e-05,
      "loss": 0.839,
      "step": 209160
    },
    {
      "epoch": 3.6722905949893785,
      "grad_norm": 0.05008542468036958,
      "learning_rate": 3.575550748906695e-05,
      "loss": 0.8376,
      "step": 209170
    },
    {
      "epoch": 3.67246615986938,
      "grad_norm": 0.06853338596325818,
      "learning_rate": 3.5747126642269515e-05,
      "loss": 0.8378,
      "step": 209180
    },
    {
      "epoch": 3.672641724749381,
      "grad_norm": 0.06380437325818106,
      "learning_rate": 3.5738746622238483e-05,
      "loss": 0.835,
      "step": 209190
    },
    {
      "epoch": 3.6728172896293825,
      "grad_norm": 0.05683944121639501,
      "learning_rate": 3.573036742907998e-05,
      "loss": 0.8366,
      "step": 209200
    },
    {
      "epoch": 3.672992854509384,
      "grad_norm": 0.05891443277948683,
      "learning_rate": 3.572198906290029e-05,
      "loss": 0.8392,
      "step": 209210
    },
    {
      "epoch": 3.6731684193893854,
      "grad_norm": 0.05463926489464742,
      "learning_rate": 3.571361152380549e-05,
      "loss": 0.8365,
      "step": 209220
    },
    {
      "epoch": 3.673343984269387,
      "grad_norm": 0.04526191013166191,
      "learning_rate": 3.570523481190176e-05,
      "loss": 0.8344,
      "step": 209230
    },
    {
      "epoch": 3.673519549149388,
      "grad_norm": 0.053139806407865024,
      "learning_rate": 3.5696858927295206e-05,
      "loss": 0.8427,
      "step": 209240
    },
    {
      "epoch": 3.6736951140293894,
      "grad_norm": 0.052045039786561303,
      "learning_rate": 3.5688483870091944e-05,
      "loss": 0.8363,
      "step": 209250
    },
    {
      "epoch": 3.673870678909391,
      "grad_norm": 0.04639947446605266,
      "learning_rate": 3.568010964039816e-05,
      "loss": 0.8406,
      "step": 209260
    },
    {
      "epoch": 3.6740462437893924,
      "grad_norm": 0.05969403310274521,
      "learning_rate": 3.5671736238319826e-05,
      "loss": 0.8331,
      "step": 209270
    },
    {
      "epoch": 3.674221808669394,
      "grad_norm": 0.06438463920663688,
      "learning_rate": 3.566336366396318e-05,
      "loss": 0.8334,
      "step": 209280
    },
    {
      "epoch": 3.6743973735493953,
      "grad_norm": 0.06570155678892459,
      "learning_rate": 3.565499191743421e-05,
      "loss": 0.8482,
      "step": 209290
    },
    {
      "epoch": 3.674572938429397,
      "grad_norm": 0.04436215419674458,
      "learning_rate": 3.564662099883901e-05,
      "loss": 0.8427,
      "step": 209300
    },
    {
      "epoch": 3.674748503309398,
      "grad_norm": 0.0526405631930618,
      "learning_rate": 3.5638250908283686e-05,
      "loss": 0.8467,
      "step": 209310
    },
    {
      "epoch": 3.6749240681893993,
      "grad_norm": 0.0461042771129446,
      "learning_rate": 3.5629881645874214e-05,
      "loss": 0.8339,
      "step": 209320
    },
    {
      "epoch": 3.675099633069401,
      "grad_norm": 0.05455268622875595,
      "learning_rate": 3.562151321171671e-05,
      "loss": 0.8406,
      "step": 209330
    },
    {
      "epoch": 3.6752751979494023,
      "grad_norm": 0.05276252739433889,
      "learning_rate": 3.5613145605917087e-05,
      "loss": 0.8452,
      "step": 209340
    },
    {
      "epoch": 3.6754507628294038,
      "grad_norm": 0.04583324182886622,
      "learning_rate": 3.5604778828581515e-05,
      "loss": 0.8402,
      "step": 209350
    },
    {
      "epoch": 3.675626327709405,
      "grad_norm": 0.048238608692808174,
      "learning_rate": 3.5596412879815924e-05,
      "loss": 0.8423,
      "step": 209360
    },
    {
      "epoch": 3.6758018925894063,
      "grad_norm": 0.06673467932177382,
      "learning_rate": 3.558804775972625e-05,
      "loss": 0.8472,
      "step": 209370
    },
    {
      "epoch": 3.6759774574694077,
      "grad_norm": 0.042503016590045725,
      "learning_rate": 3.557968346841861e-05,
      "loss": 0.8423,
      "step": 209380
    },
    {
      "epoch": 3.676153022349409,
      "grad_norm": 0.04891614911561813,
      "learning_rate": 3.557132000599888e-05,
      "loss": 0.8367,
      "step": 209390
    },
    {
      "epoch": 3.6763285872294107,
      "grad_norm": 0.05563710852315078,
      "learning_rate": 3.5562957372573114e-05,
      "loss": 0.8321,
      "step": 209400
    },
    {
      "epoch": 3.676504152109412,
      "grad_norm": 0.06825643602309228,
      "learning_rate": 3.555459556824717e-05,
      "loss": 0.8405,
      "step": 209410
    },
    {
      "epoch": 3.6766797169894136,
      "grad_norm": 0.047447908560182585,
      "learning_rate": 3.554623459312705e-05,
      "loss": 0.8311,
      "step": 209420
    },
    {
      "epoch": 3.6768552818694147,
      "grad_norm": 0.05084135322878747,
      "learning_rate": 3.553787444731872e-05,
      "loss": 0.8327,
      "step": 209430
    },
    {
      "epoch": 3.677030846749416,
      "grad_norm": 0.04261076035161414,
      "learning_rate": 3.5529515130928e-05,
      "loss": 0.8396,
      "step": 209440
    },
    {
      "epoch": 3.6772064116294176,
      "grad_norm": 0.04892802909483145,
      "learning_rate": 3.552115664406093e-05,
      "loss": 0.8425,
      "step": 209450
    },
    {
      "epoch": 3.677381976509419,
      "grad_norm": 0.044897166050385115,
      "learning_rate": 3.551279898682332e-05,
      "loss": 0.8327,
      "step": 209460
    },
    {
      "epoch": 3.6775575413894206,
      "grad_norm": 0.07022123739631393,
      "learning_rate": 3.55044421593211e-05,
      "loss": 0.845,
      "step": 209470
    },
    {
      "epoch": 3.6777331062694216,
      "grad_norm": 0.05271928098046604,
      "learning_rate": 3.5496086161660184e-05,
      "loss": 0.8365,
      "step": 209480
    },
    {
      "epoch": 3.677908671149423,
      "grad_norm": 0.04283167719606166,
      "learning_rate": 3.5487730993946385e-05,
      "loss": 0.8441,
      "step": 209490
    },
    {
      "epoch": 3.6780842360294246,
      "grad_norm": 0.0632902305027325,
      "learning_rate": 3.5479376656285605e-05,
      "loss": 0.8346,
      "step": 209500
    },
    {
      "epoch": 3.678259800909426,
      "grad_norm": 0.058209295043646476,
      "learning_rate": 3.547102314878363e-05,
      "loss": 0.8359,
      "step": 209510
    },
    {
      "epoch": 3.6784353657894275,
      "grad_norm": 0.05802471911337925,
      "learning_rate": 3.5462670471546406e-05,
      "loss": 0.8434,
      "step": 209520
    },
    {
      "epoch": 3.678610930669429,
      "grad_norm": 0.04869962598769102,
      "learning_rate": 3.545431862467972e-05,
      "loss": 0.8432,
      "step": 209530
    },
    {
      "epoch": 3.6787864955494305,
      "grad_norm": 0.05736165577638583,
      "learning_rate": 3.54459676082893e-05,
      "loss": 0.8426,
      "step": 209540
    },
    {
      "epoch": 3.6789620604294315,
      "grad_norm": 0.04515809953465004,
      "learning_rate": 3.543761742248111e-05,
      "loss": 0.8398,
      "step": 209550
    },
    {
      "epoch": 3.679137625309433,
      "grad_norm": 0.05524635997288191,
      "learning_rate": 3.5429268067360826e-05,
      "loss": 0.846,
      "step": 209560
    },
    {
      "epoch": 3.6793131901894345,
      "grad_norm": 0.06824330425870924,
      "learning_rate": 3.542091954303432e-05,
      "loss": 0.8479,
      "step": 209570
    },
    {
      "epoch": 3.679488755069436,
      "grad_norm": 0.045433468916140515,
      "learning_rate": 3.5412571849607305e-05,
      "loss": 0.8428,
      "step": 209580
    },
    {
      "epoch": 3.6796643199494374,
      "grad_norm": 0.05241392468354748,
      "learning_rate": 3.540422498718557e-05,
      "loss": 0.8399,
      "step": 209590
    },
    {
      "epoch": 3.6798398848294385,
      "grad_norm": 0.06314386681889751,
      "learning_rate": 3.539587895587491e-05,
      "loss": 0.8414,
      "step": 209600
    },
    {
      "epoch": 3.6800154497094404,
      "grad_norm": 0.05183188333424722,
      "learning_rate": 3.5387533755780966e-05,
      "loss": 0.8353,
      "step": 209610
    },
    {
      "epoch": 3.6801910145894414,
      "grad_norm": 0.04438937766083191,
      "learning_rate": 3.537918938700961e-05,
      "loss": 0.8486,
      "step": 209620
    },
    {
      "epoch": 3.680366579469443,
      "grad_norm": 0.051266467520106274,
      "learning_rate": 3.537084584966648e-05,
      "loss": 0.8369,
      "step": 209630
    },
    {
      "epoch": 3.6805421443494444,
      "grad_norm": 0.05028991361303178,
      "learning_rate": 3.5362503143857305e-05,
      "loss": 0.8477,
      "step": 209640
    },
    {
      "epoch": 3.680717709229446,
      "grad_norm": 0.05787851043456389,
      "learning_rate": 3.535416126968782e-05,
      "loss": 0.8391,
      "step": 209650
    },
    {
      "epoch": 3.6808932741094473,
      "grad_norm": 0.04908856769294585,
      "learning_rate": 3.534582022726367e-05,
      "loss": 0.8475,
      "step": 209660
    },
    {
      "epoch": 3.6810688389894484,
      "grad_norm": 0.052496279195455416,
      "learning_rate": 3.533748001669059e-05,
      "loss": 0.8381,
      "step": 209670
    },
    {
      "epoch": 3.68124440386945,
      "grad_norm": 0.05543051244924703,
      "learning_rate": 3.5329140638074144e-05,
      "loss": 0.8395,
      "step": 209680
    },
    {
      "epoch": 3.6814199687494513,
      "grad_norm": 0.05555950331423421,
      "learning_rate": 3.5320802091520156e-05,
      "loss": 0.8393,
      "step": 209690
    },
    {
      "epoch": 3.681595533629453,
      "grad_norm": 0.04425555653795772,
      "learning_rate": 3.5312464377134175e-05,
      "loss": 0.8391,
      "step": 209700
    },
    {
      "epoch": 3.6817710985094543,
      "grad_norm": 0.06440033837124004,
      "learning_rate": 3.53041274950218e-05,
      "loss": 0.8426,
      "step": 209710
    },
    {
      "epoch": 3.6819466633894553,
      "grad_norm": 0.04031719910680272,
      "learning_rate": 3.52957914452888e-05,
      "loss": 0.842,
      "step": 209720
    },
    {
      "epoch": 3.6821222282694572,
      "grad_norm": 0.04651255561848049,
      "learning_rate": 3.528745622804067e-05,
      "loss": 0.8325,
      "step": 209730
    },
    {
      "epoch": 3.6822977931494583,
      "grad_norm": 0.05086678722693569,
      "learning_rate": 3.527912184338309e-05,
      "loss": 0.8455,
      "step": 209740
    },
    {
      "epoch": 3.6824733580294597,
      "grad_norm": 0.06145401315238536,
      "learning_rate": 3.527078829142161e-05,
      "loss": 0.8408,
      "step": 209750
    },
    {
      "epoch": 3.682648922909461,
      "grad_norm": 0.05632151939105467,
      "learning_rate": 3.526245557226183e-05,
      "loss": 0.842,
      "step": 209760
    },
    {
      "epoch": 3.6828244877894627,
      "grad_norm": 0.059086528572803235,
      "learning_rate": 3.5254123686009377e-05,
      "loss": 0.8397,
      "step": 209770
    },
    {
      "epoch": 3.683000052669464,
      "grad_norm": 0.04791571772472563,
      "learning_rate": 3.5245792632769714e-05,
      "loss": 0.8398,
      "step": 209780
    },
    {
      "epoch": 3.683175617549465,
      "grad_norm": 0.05337017898768397,
      "learning_rate": 3.523746241264852e-05,
      "loss": 0.8388,
      "step": 209790
    },
    {
      "epoch": 3.6833511824294667,
      "grad_norm": 0.04858693193663825,
      "learning_rate": 3.5229133025751246e-05,
      "loss": 0.8332,
      "step": 209800
    },
    {
      "epoch": 3.683526747309468,
      "grad_norm": 0.0540985641620731,
      "learning_rate": 3.522080447218346e-05,
      "loss": 0.8415,
      "step": 209810
    },
    {
      "epoch": 3.6837023121894696,
      "grad_norm": 0.048507781743368665,
      "learning_rate": 3.521247675205073e-05,
      "loss": 0.8386,
      "step": 209820
    },
    {
      "epoch": 3.683877877069471,
      "grad_norm": 0.05090566155797095,
      "learning_rate": 3.5204149865458494e-05,
      "loss": 0.8407,
      "step": 209830
    },
    {
      "epoch": 3.684053441949472,
      "grad_norm": 0.0480323333372483,
      "learning_rate": 3.519582381251231e-05,
      "loss": 0.843,
      "step": 209840
    },
    {
      "epoch": 3.684229006829474,
      "grad_norm": 0.04538673183861194,
      "learning_rate": 3.518749859331759e-05,
      "loss": 0.8374,
      "step": 209850
    },
    {
      "epoch": 3.684404571709475,
      "grad_norm": 0.05622929718241364,
      "learning_rate": 3.517917420797994e-05,
      "loss": 0.8316,
      "step": 209860
    },
    {
      "epoch": 3.6845801365894766,
      "grad_norm": 0.046967025773946035,
      "learning_rate": 3.517085065660478e-05,
      "loss": 0.8306,
      "step": 209870
    },
    {
      "epoch": 3.684755701469478,
      "grad_norm": 0.053799044547350565,
      "learning_rate": 3.5162527939297476e-05,
      "loss": 0.8348,
      "step": 209880
    },
    {
      "epoch": 3.6849312663494795,
      "grad_norm": 0.05875970514613972,
      "learning_rate": 3.515420605616363e-05,
      "loss": 0.8455,
      "step": 209890
    },
    {
      "epoch": 3.685106831229481,
      "grad_norm": 0.05031070524760897,
      "learning_rate": 3.5145885007308584e-05,
      "loss": 0.8408,
      "step": 209900
    },
    {
      "epoch": 3.685282396109482,
      "grad_norm": 0.04737942034684283,
      "learning_rate": 3.513756479283782e-05,
      "loss": 0.8363,
      "step": 209910
    },
    {
      "epoch": 3.6854579609894835,
      "grad_norm": 0.050476157987447985,
      "learning_rate": 3.512924541285671e-05,
      "loss": 0.8495,
      "step": 209920
    },
    {
      "epoch": 3.685633525869485,
      "grad_norm": 0.06345381311924793,
      "learning_rate": 3.5120926867470675e-05,
      "loss": 0.8394,
      "step": 209930
    },
    {
      "epoch": 3.6858090907494865,
      "grad_norm": 0.05495175195509386,
      "learning_rate": 3.511260915678517e-05,
      "loss": 0.8426,
      "step": 209940
    },
    {
      "epoch": 3.685984655629488,
      "grad_norm": 0.04184916256133415,
      "learning_rate": 3.510429228090545e-05,
      "loss": 0.8386,
      "step": 209950
    },
    {
      "epoch": 3.6861602205094894,
      "grad_norm": 0.05338183503916748,
      "learning_rate": 3.509597623993706e-05,
      "loss": 0.8445,
      "step": 209960
    },
    {
      "epoch": 3.686335785389491,
      "grad_norm": 0.046390924762384274,
      "learning_rate": 3.5087661033985245e-05,
      "loss": 0.8304,
      "step": 209970
    },
    {
      "epoch": 3.686511350269492,
      "grad_norm": 0.04778240944001487,
      "learning_rate": 3.50793466631554e-05,
      "loss": 0.8423,
      "step": 209980
    },
    {
      "epoch": 3.6866869151494934,
      "grad_norm": 0.05088295780055026,
      "learning_rate": 3.5071033127552876e-05,
      "loss": 0.8371,
      "step": 209990
    },
    {
      "epoch": 3.686862480029495,
      "grad_norm": 0.05113699384689641,
      "learning_rate": 3.506272042728297e-05,
      "loss": 0.8469,
      "step": 210000
    },
    {
      "epoch": 3.6870380449094964,
      "grad_norm": 0.05287865018254363,
      "learning_rate": 3.505440856245109e-05,
      "loss": 0.8315,
      "step": 210010
    },
    {
      "epoch": 3.687213609789498,
      "grad_norm": 0.04796640546190377,
      "learning_rate": 3.504609753316241e-05,
      "loss": 0.8471,
      "step": 210020
    },
    {
      "epoch": 3.687389174669499,
      "grad_norm": 0.0630256706357599,
      "learning_rate": 3.503778733952238e-05,
      "loss": 0.8456,
      "step": 210030
    },
    {
      "epoch": 3.6875647395495004,
      "grad_norm": 0.053769184082084656,
      "learning_rate": 3.5029477981636225e-05,
      "loss": 0.8371,
      "step": 210040
    },
    {
      "epoch": 3.687740304429502,
      "grad_norm": 0.05775731651042885,
      "learning_rate": 3.502116945960917e-05,
      "loss": 0.8372,
      "step": 210050
    },
    {
      "epoch": 3.6879158693095033,
      "grad_norm": 0.07039092712835965,
      "learning_rate": 3.50128617735466e-05,
      "loss": 0.8462,
      "step": 210060
    },
    {
      "epoch": 3.688091434189505,
      "grad_norm": 0.051025847947780895,
      "learning_rate": 3.500455492355368e-05,
      "loss": 0.8353,
      "step": 210070
    },
    {
      "epoch": 3.6882669990695063,
      "grad_norm": 0.043335999178118555,
      "learning_rate": 3.499624890973575e-05,
      "loss": 0.8412,
      "step": 210080
    },
    {
      "epoch": 3.6884425639495078,
      "grad_norm": 0.05819689852347623,
      "learning_rate": 3.4987943732197955e-05,
      "loss": 0.8374,
      "step": 210090
    },
    {
      "epoch": 3.688618128829509,
      "grad_norm": 0.07082453208490216,
      "learning_rate": 3.497963939104555e-05,
      "loss": 0.8364,
      "step": 210100
    },
    {
      "epoch": 3.6887936937095103,
      "grad_norm": 0.07024028896967274,
      "learning_rate": 3.497133588638381e-05,
      "loss": 0.8392,
      "step": 210110
    },
    {
      "epoch": 3.6889692585895117,
      "grad_norm": 0.05646461529926115,
      "learning_rate": 3.496303321831782e-05,
      "loss": 0.8403,
      "step": 210120
    },
    {
      "epoch": 3.6891448234695132,
      "grad_norm": 0.05228576079499608,
      "learning_rate": 3.4954731386952945e-05,
      "loss": 0.8459,
      "step": 210130
    },
    {
      "epoch": 3.6893203883495147,
      "grad_norm": 0.06494811645641756,
      "learning_rate": 3.494643039239423e-05,
      "loss": 0.846,
      "step": 210140
    },
    {
      "epoch": 3.6894959532295157,
      "grad_norm": 0.05921939491000753,
      "learning_rate": 3.493813023474692e-05,
      "loss": 0.8462,
      "step": 210150
    },
    {
      "epoch": 3.689671518109517,
      "grad_norm": 0.06366318737305707,
      "learning_rate": 3.492983091411617e-05,
      "loss": 0.838,
      "step": 210160
    },
    {
      "epoch": 3.6898470829895187,
      "grad_norm": 0.05121097759929135,
      "learning_rate": 3.4921532430607103e-05,
      "loss": 0.8379,
      "step": 210170
    },
    {
      "epoch": 3.69002264786952,
      "grad_norm": 0.06804341002830244,
      "learning_rate": 3.491323478432492e-05,
      "loss": 0.8382,
      "step": 210180
    },
    {
      "epoch": 3.6901982127495216,
      "grad_norm": 0.047306282675720845,
      "learning_rate": 3.490493797537464e-05,
      "loss": 0.8477,
      "step": 210190
    },
    {
      "epoch": 3.690373777629523,
      "grad_norm": 0.04951966913731846,
      "learning_rate": 3.489664200386154e-05,
      "loss": 0.8371,
      "step": 210200
    },
    {
      "epoch": 3.6905493425095246,
      "grad_norm": 0.05083679948295019,
      "learning_rate": 3.4888346869890655e-05,
      "loss": 0.8411,
      "step": 210210
    },
    {
      "epoch": 3.6907249073895256,
      "grad_norm": 0.05708556933009567,
      "learning_rate": 3.488005257356701e-05,
      "loss": 0.8488,
      "step": 210220
    },
    {
      "epoch": 3.690900472269527,
      "grad_norm": 0.0471508940421651,
      "learning_rate": 3.487175911499584e-05,
      "loss": 0.8388,
      "step": 210230
    },
    {
      "epoch": 3.6910760371495286,
      "grad_norm": 0.054092171321633555,
      "learning_rate": 3.486346649428213e-05,
      "loss": 0.8391,
      "step": 210240
    },
    {
      "epoch": 3.69125160202953,
      "grad_norm": 0.05584477477393784,
      "learning_rate": 3.4855174711531e-05,
      "loss": 0.8417,
      "step": 210250
    },
    {
      "epoch": 3.6914271669095315,
      "grad_norm": 0.06369450294772958,
      "learning_rate": 3.484688376684745e-05,
      "loss": 0.8387,
      "step": 210260
    },
    {
      "epoch": 3.6916027317895326,
      "grad_norm": 0.05002187629833076,
      "learning_rate": 3.483859366033657e-05,
      "loss": 0.8404,
      "step": 210270
    },
    {
      "epoch": 3.6917782966695345,
      "grad_norm": 0.04486636271093967,
      "learning_rate": 3.4830304392103405e-05,
      "loss": 0.8361,
      "step": 210280
    },
    {
      "epoch": 3.6919538615495355,
      "grad_norm": 0.058453583012730286,
      "learning_rate": 3.4822015962252915e-05,
      "loss": 0.8369,
      "step": 210290
    },
    {
      "epoch": 3.692129426429537,
      "grad_norm": 0.06041372303009075,
      "learning_rate": 3.481372837089023e-05,
      "loss": 0.8401,
      "step": 210300
    },
    {
      "epoch": 3.6923049913095385,
      "grad_norm": 0.04646546655286257,
      "learning_rate": 3.480544161812025e-05,
      "loss": 0.8412,
      "step": 210310
    },
    {
      "epoch": 3.69248055618954,
      "grad_norm": 0.04295704181718986,
      "learning_rate": 3.4797155704048015e-05,
      "loss": 0.8355,
      "step": 210320
    },
    {
      "epoch": 3.6926561210695414,
      "grad_norm": 0.049841448171084574,
      "learning_rate": 3.478887062877853e-05,
      "loss": 0.8454,
      "step": 210330
    },
    {
      "epoch": 3.6928316859495425,
      "grad_norm": 0.05030247452886533,
      "learning_rate": 3.4780586392416716e-05,
      "loss": 0.8341,
      "step": 210340
    },
    {
      "epoch": 3.693007250829544,
      "grad_norm": 0.06628950647410442,
      "learning_rate": 3.47723029950676e-05,
      "loss": 0.8378,
      "step": 210350
    },
    {
      "epoch": 3.6931828157095454,
      "grad_norm": 0.07876878824639971,
      "learning_rate": 3.476402043683603e-05,
      "loss": 0.8407,
      "step": 210360
    },
    {
      "epoch": 3.693358380589547,
      "grad_norm": 0.04843939911863175,
      "learning_rate": 3.475573871782708e-05,
      "loss": 0.8356,
      "step": 210370
    },
    {
      "epoch": 3.6935339454695484,
      "grad_norm": 0.04275820323819817,
      "learning_rate": 3.474745783814561e-05,
      "loss": 0.8387,
      "step": 210380
    },
    {
      "epoch": 3.6937095103495494,
      "grad_norm": 0.04903354967694946,
      "learning_rate": 3.4739177797896496e-05,
      "loss": 0.8422,
      "step": 210390
    },
    {
      "epoch": 3.6938850752295513,
      "grad_norm": 0.076001576749364,
      "learning_rate": 3.473089859718475e-05,
      "loss": 0.8348,
      "step": 210400
    },
    {
      "epoch": 3.6940606401095524,
      "grad_norm": 0.05416793407229062,
      "learning_rate": 3.4722620236115184e-05,
      "loss": 0.8384,
      "step": 210410
    },
    {
      "epoch": 3.694236204989554,
      "grad_norm": 0.07690899144676797,
      "learning_rate": 3.4714342714792756e-05,
      "loss": 0.8416,
      "step": 210420
    },
    {
      "epoch": 3.6944117698695553,
      "grad_norm": 0.050007872098113346,
      "learning_rate": 3.470606603332229e-05,
      "loss": 0.8369,
      "step": 210430
    },
    {
      "epoch": 3.694587334749557,
      "grad_norm": 0.054861575066963944,
      "learning_rate": 3.4697790191808664e-05,
      "loss": 0.8376,
      "step": 210440
    },
    {
      "epoch": 3.6947628996295583,
      "grad_norm": 0.04521762132544243,
      "learning_rate": 3.468951519035677e-05,
      "loss": 0.8369,
      "step": 210450
    },
    {
      "epoch": 3.6949384645095593,
      "grad_norm": 0.04477505684011138,
      "learning_rate": 3.468124102907137e-05,
      "loss": 0.8411,
      "step": 210460
    },
    {
      "epoch": 3.695114029389561,
      "grad_norm": 0.051997585945196885,
      "learning_rate": 3.467296770805743e-05,
      "loss": 0.8373,
      "step": 210470
    },
    {
      "epoch": 3.6952895942695623,
      "grad_norm": 0.05242067797721026,
      "learning_rate": 3.4664695227419646e-05,
      "loss": 0.8323,
      "step": 210480
    },
    {
      "epoch": 3.6954651591495637,
      "grad_norm": 0.05795709326077588,
      "learning_rate": 3.4656423587262896e-05,
      "loss": 0.8398,
      "step": 210490
    },
    {
      "epoch": 3.6956407240295652,
      "grad_norm": 0.06602683613047441,
      "learning_rate": 3.4648152787692015e-05,
      "loss": 0.8307,
      "step": 210500
    },
    {
      "epoch": 3.6958162889095663,
      "grad_norm": 0.05080485472683296,
      "learning_rate": 3.4639882828811714e-05,
      "loss": 0.8419,
      "step": 210510
    },
    {
      "epoch": 3.695991853789568,
      "grad_norm": 0.050030361585325776,
      "learning_rate": 3.4631613710726845e-05,
      "loss": 0.8414,
      "step": 210520
    },
    {
      "epoch": 3.696167418669569,
      "grad_norm": 0.044574211329613776,
      "learning_rate": 3.4623345433542085e-05,
      "loss": 0.8392,
      "step": 210530
    },
    {
      "epoch": 3.6963429835495707,
      "grad_norm": 0.05010340731938054,
      "learning_rate": 3.461507799736233e-05,
      "loss": 0.8358,
      "step": 210540
    },
    {
      "epoch": 3.696518548429572,
      "grad_norm": 0.05059269616559478,
      "learning_rate": 3.460681140229226e-05,
      "loss": 0.8441,
      "step": 210550
    },
    {
      "epoch": 3.6966941133095736,
      "grad_norm": 0.06341830589343718,
      "learning_rate": 3.459854564843654e-05,
      "loss": 0.8418,
      "step": 210560
    },
    {
      "epoch": 3.696869678189575,
      "grad_norm": 0.05779893844511105,
      "learning_rate": 3.459028073590005e-05,
      "loss": 0.8337,
      "step": 210570
    },
    {
      "epoch": 3.697045243069576,
      "grad_norm": 0.05208090540987079,
      "learning_rate": 3.45820166647874e-05,
      "loss": 0.8401,
      "step": 210580
    },
    {
      "epoch": 3.6972208079495776,
      "grad_norm": 0.058700088959292074,
      "learning_rate": 3.457375343520335e-05,
      "loss": 0.8429,
      "step": 210590
    },
    {
      "epoch": 3.697396372829579,
      "grad_norm": 0.046987212520536586,
      "learning_rate": 3.456549104725254e-05,
      "loss": 0.8386,
      "step": 210600
    },
    {
      "epoch": 3.6975719377095806,
      "grad_norm": 0.05542721051985387,
      "learning_rate": 3.4557229501039695e-05,
      "loss": 0.8349,
      "step": 210610
    },
    {
      "epoch": 3.697747502589582,
      "grad_norm": 0.04492566221969696,
      "learning_rate": 3.454896879666951e-05,
      "loss": 0.8432,
      "step": 210620
    },
    {
      "epoch": 3.6979230674695835,
      "grad_norm": 0.04865270248364846,
      "learning_rate": 3.454070893424657e-05,
      "loss": 0.8431,
      "step": 210630
    },
    {
      "epoch": 3.698098632349585,
      "grad_norm": 0.04187115414609849,
      "learning_rate": 3.4532449913875647e-05,
      "loss": 0.8328,
      "step": 210640
    },
    {
      "epoch": 3.698274197229586,
      "grad_norm": 0.053485016976013704,
      "learning_rate": 3.4524191735661294e-05,
      "loss": 0.8374,
      "step": 210650
    },
    {
      "epoch": 3.6984497621095875,
      "grad_norm": 0.05558163197089914,
      "learning_rate": 3.451593439970817e-05,
      "loss": 0.8411,
      "step": 210660
    },
    {
      "epoch": 3.698625326989589,
      "grad_norm": 0.06447959188721396,
      "learning_rate": 3.4507677906120934e-05,
      "loss": 0.8377,
      "step": 210670
    },
    {
      "epoch": 3.6988008918695905,
      "grad_norm": 0.11819902647307576,
      "learning_rate": 3.449942225500412e-05,
      "loss": 0.8425,
      "step": 210680
    },
    {
      "epoch": 3.698976456749592,
      "grad_norm": 0.04232464880583048,
      "learning_rate": 3.44911674464624e-05,
      "loss": 0.8373,
      "step": 210690
    },
    {
      "epoch": 3.699152021629593,
      "grad_norm": 0.047652104914990376,
      "learning_rate": 3.4482913480600284e-05,
      "loss": 0.8391,
      "step": 210700
    },
    {
      "epoch": 3.6993275865095945,
      "grad_norm": 0.0574735920581166,
      "learning_rate": 3.4474660357522464e-05,
      "loss": 0.8384,
      "step": 210710
    },
    {
      "epoch": 3.699503151389596,
      "grad_norm": 0.054226390927207796,
      "learning_rate": 3.4466408077333445e-05,
      "loss": 0.8417,
      "step": 210720
    },
    {
      "epoch": 3.6996787162695974,
      "grad_norm": 0.051028774715313766,
      "learning_rate": 3.445815664013771e-05,
      "loss": 0.8374,
      "step": 210730
    },
    {
      "epoch": 3.699854281149599,
      "grad_norm": 0.06745172834842866,
      "learning_rate": 3.444990604603997e-05,
      "loss": 0.838,
      "step": 210740
    },
    {
      "epoch": 3.7000298460296004,
      "grad_norm": 0.06632664223643357,
      "learning_rate": 3.4441656295144625e-05,
      "loss": 0.8326,
      "step": 210750
    },
    {
      "epoch": 3.700205410909602,
      "grad_norm": 0.04143125439049729,
      "learning_rate": 3.443340738755629e-05,
      "loss": 0.8412,
      "step": 210760
    },
    {
      "epoch": 3.700380975789603,
      "grad_norm": 0.04586338034728872,
      "learning_rate": 3.442515932337941e-05,
      "loss": 0.8304,
      "step": 210770
    },
    {
      "epoch": 3.7005565406696044,
      "grad_norm": 0.052968735159252274,
      "learning_rate": 3.441691210271852e-05,
      "loss": 0.8407,
      "step": 210780
    },
    {
      "epoch": 3.700732105549606,
      "grad_norm": 0.04880326610507707,
      "learning_rate": 3.440866572567815e-05,
      "loss": 0.8405,
      "step": 210790
    },
    {
      "epoch": 3.7009076704296073,
      "grad_norm": 0.07361666858173403,
      "learning_rate": 3.4400420192362665e-05,
      "loss": 0.8361,
      "step": 210800
    },
    {
      "epoch": 3.701083235309609,
      "grad_norm": 0.04995755587397494,
      "learning_rate": 3.4392175502876695e-05,
      "loss": 0.8366,
      "step": 210810
    },
    {
      "epoch": 3.70125880018961,
      "grad_norm": 0.060481327921479075,
      "learning_rate": 3.4383931657324604e-05,
      "loss": 0.8427,
      "step": 210820
    },
    {
      "epoch": 3.7014343650696113,
      "grad_norm": 0.04809399665385191,
      "learning_rate": 3.437568865581086e-05,
      "loss": 0.8417,
      "step": 210830
    },
    {
      "epoch": 3.701609929949613,
      "grad_norm": 0.057368890664269284,
      "learning_rate": 3.436744649843994e-05,
      "loss": 0.8527,
      "step": 210840
    },
    {
      "epoch": 3.7017854948296143,
      "grad_norm": 0.04920479967588251,
      "learning_rate": 3.435920518531621e-05,
      "loss": 0.8282,
      "step": 210850
    },
    {
      "epoch": 3.7019610597096158,
      "grad_norm": 0.05423460742587199,
      "learning_rate": 3.435096471654415e-05,
      "loss": 0.8341,
      "step": 210860
    },
    {
      "epoch": 3.7021366245896172,
      "grad_norm": 0.05096298622470347,
      "learning_rate": 3.4342725092228085e-05,
      "loss": 0.8401,
      "step": 210870
    },
    {
      "epoch": 3.7023121894696187,
      "grad_norm": 0.051575256071760264,
      "learning_rate": 3.4334486312472524e-05,
      "loss": 0.8427,
      "step": 210880
    },
    {
      "epoch": 3.7024877543496197,
      "grad_norm": 0.06132158790919768,
      "learning_rate": 3.4326248377381796e-05,
      "loss": 0.8449,
      "step": 210890
    },
    {
      "epoch": 3.702663319229621,
      "grad_norm": 0.049004046552582385,
      "learning_rate": 3.431801128706023e-05,
      "loss": 0.8572,
      "step": 210900
    },
    {
      "epoch": 3.7028388841096227,
      "grad_norm": 0.043836373906013346,
      "learning_rate": 3.4309775041612296e-05,
      "loss": 0.8395,
      "step": 210910
    },
    {
      "epoch": 3.703014448989624,
      "grad_norm": 0.08939843313707578,
      "learning_rate": 3.430153964114225e-05,
      "loss": 0.8501,
      "step": 210920
    },
    {
      "epoch": 3.7031900138696257,
      "grad_norm": 0.045758243512221346,
      "learning_rate": 3.429330508575453e-05,
      "loss": 0.8395,
      "step": 210930
    },
    {
      "epoch": 3.7033655787496267,
      "grad_norm": 0.04538770957028324,
      "learning_rate": 3.428507137555337e-05,
      "loss": 0.8372,
      "step": 210940
    },
    {
      "epoch": 3.703541143629628,
      "grad_norm": 0.048061062634587676,
      "learning_rate": 3.427683851064316e-05,
      "loss": 0.8394,
      "step": 210950
    },
    {
      "epoch": 3.7037167085096296,
      "grad_norm": 0.044043391927720485,
      "learning_rate": 3.4268606491128204e-05,
      "loss": 0.8399,
      "step": 210960
    },
    {
      "epoch": 3.703892273389631,
      "grad_norm": 0.06890499962125894,
      "learning_rate": 3.426037531711275e-05,
      "loss": 0.8359,
      "step": 210970
    },
    {
      "epoch": 3.7040678382696326,
      "grad_norm": 0.04994532190146109,
      "learning_rate": 3.425214498870118e-05,
      "loss": 0.8452,
      "step": 210980
    },
    {
      "epoch": 3.704243403149634,
      "grad_norm": 0.04684929595158864,
      "learning_rate": 3.4243915505997696e-05,
      "loss": 0.8454,
      "step": 210990
    },
    {
      "epoch": 3.7044189680296356,
      "grad_norm": 0.061148589777111385,
      "learning_rate": 3.42356868691066e-05,
      "loss": 0.8347,
      "step": 211000
    },
    {
      "epoch": 3.7045945329096366,
      "grad_norm": 0.07020875000480606,
      "learning_rate": 3.422745907813218e-05,
      "loss": 0.8455,
      "step": 211010
    },
    {
      "epoch": 3.704770097789638,
      "grad_norm": 0.04792958826868132,
      "learning_rate": 3.421923213317861e-05,
      "loss": 0.8358,
      "step": 211020
    },
    {
      "epoch": 3.7049456626696395,
      "grad_norm": 0.045625088916198514,
      "learning_rate": 3.421100603435021e-05,
      "loss": 0.8433,
      "step": 211030
    },
    {
      "epoch": 3.705121227549641,
      "grad_norm": 0.04884313080810672,
      "learning_rate": 3.420278078175113e-05,
      "loss": 0.8369,
      "step": 211040
    },
    {
      "epoch": 3.7052967924296425,
      "grad_norm": 0.04507021273343106,
      "learning_rate": 3.419455637548561e-05,
      "loss": 0.8397,
      "step": 211050
    },
    {
      "epoch": 3.7054723573096435,
      "grad_norm": 0.04451762240257929,
      "learning_rate": 3.418633281565791e-05,
      "loss": 0.8334,
      "step": 211060
    },
    {
      "epoch": 3.7056479221896454,
      "grad_norm": 0.06389012496252895,
      "learning_rate": 3.41781101023721e-05,
      "loss": 0.836,
      "step": 211070
    },
    {
      "epoch": 3.7058234870696465,
      "grad_norm": 0.05830000591891933,
      "learning_rate": 3.416988823573251e-05,
      "loss": 0.8421,
      "step": 211080
    },
    {
      "epoch": 3.705999051949648,
      "grad_norm": 0.057202211408835295,
      "learning_rate": 3.416166721584322e-05,
      "loss": 0.8387,
      "step": 211090
    },
    {
      "epoch": 3.7061746168296494,
      "grad_norm": 0.05437616102428297,
      "learning_rate": 3.415344704280844e-05,
      "loss": 0.8395,
      "step": 211100
    },
    {
      "epoch": 3.706350181709651,
      "grad_norm": 0.060318531340682484,
      "learning_rate": 3.414522771673226e-05,
      "loss": 0.8419,
      "step": 211110
    },
    {
      "epoch": 3.7065257465896524,
      "grad_norm": 0.051899113802191144,
      "learning_rate": 3.413700923771887e-05,
      "loss": 0.8321,
      "step": 211120
    },
    {
      "epoch": 3.7067013114696534,
      "grad_norm": 0.04934325190891691,
      "learning_rate": 3.4128791605872414e-05,
      "loss": 0.8423,
      "step": 211130
    },
    {
      "epoch": 3.706876876349655,
      "grad_norm": 0.060311204605374494,
      "learning_rate": 3.412057482129692e-05,
      "loss": 0.8415,
      "step": 211140
    },
    {
      "epoch": 3.7070524412296564,
      "grad_norm": 0.05346088066454043,
      "learning_rate": 3.411235888409662e-05,
      "loss": 0.8397,
      "step": 211150
    },
    {
      "epoch": 3.707228006109658,
      "grad_norm": 0.052525068915386185,
      "learning_rate": 3.4104143794375527e-05,
      "loss": 0.8352,
      "step": 211160
    },
    {
      "epoch": 3.7074035709896593,
      "grad_norm": 0.051372483834923516,
      "learning_rate": 3.4095929552237746e-05,
      "loss": 0.8311,
      "step": 211170
    },
    {
      "epoch": 3.7075791358696604,
      "grad_norm": 0.054321220518864975,
      "learning_rate": 3.40877161577874e-05,
      "loss": 0.838,
      "step": 211180
    },
    {
      "epoch": 3.7077547007496623,
      "grad_norm": 0.05503642962173508,
      "learning_rate": 3.407950361112847e-05,
      "loss": 0.8458,
      "step": 211190
    },
    {
      "epoch": 3.7079302656296633,
      "grad_norm": 0.05184289859693415,
      "learning_rate": 3.4071291912365095e-05,
      "loss": 0.8293,
      "step": 211200
    },
    {
      "epoch": 3.708105830509665,
      "grad_norm": 0.0509117364739118,
      "learning_rate": 3.406308106160125e-05,
      "loss": 0.839,
      "step": 211210
    },
    {
      "epoch": 3.7082813953896663,
      "grad_norm": 0.061126858707889684,
      "learning_rate": 3.405487105894099e-05,
      "loss": 0.8441,
      "step": 211220
    },
    {
      "epoch": 3.7084569602696678,
      "grad_norm": 0.061876622247823566,
      "learning_rate": 3.404666190448838e-05,
      "loss": 0.8421,
      "step": 211230
    },
    {
      "epoch": 3.7086325251496692,
      "grad_norm": 0.05267957443400821,
      "learning_rate": 3.403845359834732e-05,
      "loss": 0.841,
      "step": 211240
    },
    {
      "epoch": 3.7088080900296703,
      "grad_norm": 0.05032043330746887,
      "learning_rate": 3.403024614062196e-05,
      "loss": 0.8394,
      "step": 211250
    },
    {
      "epoch": 3.7089836549096717,
      "grad_norm": 0.0590903588063783,
      "learning_rate": 3.4022039531416186e-05,
      "loss": 0.84,
      "step": 211260
    },
    {
      "epoch": 3.7091592197896732,
      "grad_norm": 0.05217741931538674,
      "learning_rate": 3.401383377083403e-05,
      "loss": 0.8451,
      "step": 211270
    },
    {
      "epoch": 3.7093347846696747,
      "grad_norm": 0.08187858055814767,
      "learning_rate": 3.4005628858979414e-05,
      "loss": 0.8352,
      "step": 211280
    },
    {
      "epoch": 3.709510349549676,
      "grad_norm": 0.049278127190988355,
      "learning_rate": 3.399742479595631e-05,
      "loss": 0.8343,
      "step": 211290
    },
    {
      "epoch": 3.709685914429677,
      "grad_norm": 0.04620167727725859,
      "learning_rate": 3.398922158186872e-05,
      "loss": 0.8359,
      "step": 211300
    },
    {
      "epoch": 3.709861479309679,
      "grad_norm": 0.05426954197865697,
      "learning_rate": 3.398101921682046e-05,
      "loss": 0.8431,
      "step": 211310
    },
    {
      "epoch": 3.71003704418968,
      "grad_norm": 0.053349923282222045,
      "learning_rate": 3.39728177009156e-05,
      "loss": 0.8388,
      "step": 211320
    },
    {
      "epoch": 3.7102126090696816,
      "grad_norm": 0.06566209991368654,
      "learning_rate": 3.396461703425796e-05,
      "loss": 0.8337,
      "step": 211330
    },
    {
      "epoch": 3.710388173949683,
      "grad_norm": 0.04489939394555453,
      "learning_rate": 3.3956417216951456e-05,
      "loss": 0.8413,
      "step": 211340
    },
    {
      "epoch": 3.7105637388296846,
      "grad_norm": 0.06818272257671404,
      "learning_rate": 3.3948218249100034e-05,
      "loss": 0.843,
      "step": 211350
    },
    {
      "epoch": 3.710739303709686,
      "grad_norm": 0.05981376833956065,
      "learning_rate": 3.3940020130807506e-05,
      "loss": 0.8338,
      "step": 211360
    },
    {
      "epoch": 3.710914868589687,
      "grad_norm": 0.06200623992580753,
      "learning_rate": 3.393182286217781e-05,
      "loss": 0.8402,
      "step": 211370
    },
    {
      "epoch": 3.7110904334696886,
      "grad_norm": 0.05805804370033374,
      "learning_rate": 3.392362644331473e-05,
      "loss": 0.8425,
      "step": 211380
    },
    {
      "epoch": 3.71126599834969,
      "grad_norm": 0.04480896663990068,
      "learning_rate": 3.391543087432216e-05,
      "loss": 0.8423,
      "step": 211390
    },
    {
      "epoch": 3.7114415632296915,
      "grad_norm": 0.038222225515256315,
      "learning_rate": 3.390723615530398e-05,
      "loss": 0.8522,
      "step": 211400
    },
    {
      "epoch": 3.711617128109693,
      "grad_norm": 0.04136648659982268,
      "learning_rate": 3.38990422863639e-05,
      "loss": 0.8407,
      "step": 211410
    },
    {
      "epoch": 3.7117926929896945,
      "grad_norm": 0.05390155238474064,
      "learning_rate": 3.389084926760589e-05,
      "loss": 0.8397,
      "step": 211420
    },
    {
      "epoch": 3.711968257869696,
      "grad_norm": 0.059507628774379255,
      "learning_rate": 3.3882657099133635e-05,
      "loss": 0.8384,
      "step": 211430
    },
    {
      "epoch": 3.712143822749697,
      "grad_norm": 0.05345525621741854,
      "learning_rate": 3.387446578105101e-05,
      "loss": 0.8419,
      "step": 211440
    },
    {
      "epoch": 3.7123193876296985,
      "grad_norm": 0.06851458102237448,
      "learning_rate": 3.386627531346175e-05,
      "loss": 0.8413,
      "step": 211450
    },
    {
      "epoch": 3.7124949525097,
      "grad_norm": 0.05488877101736562,
      "learning_rate": 3.385808569646964e-05,
      "loss": 0.8463,
      "step": 211460
    },
    {
      "epoch": 3.7126705173897014,
      "grad_norm": 0.046601859496149436,
      "learning_rate": 3.384989693017849e-05,
      "loss": 0.8481,
      "step": 211470
    },
    {
      "epoch": 3.712846082269703,
      "grad_norm": 0.0840410132708092,
      "learning_rate": 3.384170901469194e-05,
      "loss": 0.8367,
      "step": 211480
    },
    {
      "epoch": 3.713021647149704,
      "grad_norm": 0.041225452991110384,
      "learning_rate": 3.383352195011388e-05,
      "loss": 0.8407,
      "step": 211490
    },
    {
      "epoch": 3.7131972120297054,
      "grad_norm": 0.051726006492100934,
      "learning_rate": 3.382533573654793e-05,
      "loss": 0.8349,
      "step": 211500
    },
    {
      "epoch": 3.713372776909707,
      "grad_norm": 0.07038255134504118,
      "learning_rate": 3.381715037409786e-05,
      "loss": 0.8338,
      "step": 211510
    },
    {
      "epoch": 3.7135483417897084,
      "grad_norm": 0.05068089597812051,
      "learning_rate": 3.38089658628674e-05,
      "loss": 0.8331,
      "step": 211520
    },
    {
      "epoch": 3.71372390666971,
      "grad_norm": 0.05258460607923388,
      "learning_rate": 3.380078220296018e-05,
      "loss": 0.8368,
      "step": 211530
    },
    {
      "epoch": 3.7138994715497113,
      "grad_norm": 0.05435118974023497,
      "learning_rate": 3.379259939447997e-05,
      "loss": 0.838,
      "step": 211540
    },
    {
      "epoch": 3.714075036429713,
      "grad_norm": 0.06047560336737044,
      "learning_rate": 3.378441743753037e-05,
      "loss": 0.8411,
      "step": 211550
    },
    {
      "epoch": 3.714250601309714,
      "grad_norm": 0.05882458176028383,
      "learning_rate": 3.377623633221508e-05,
      "loss": 0.8391,
      "step": 211560
    },
    {
      "epoch": 3.7144261661897153,
      "grad_norm": 0.06023305131532877,
      "learning_rate": 3.376805607863779e-05,
      "loss": 0.8392,
      "step": 211570
    },
    {
      "epoch": 3.714601731069717,
      "grad_norm": 0.05874294828857853,
      "learning_rate": 3.375987667690206e-05,
      "loss": 0.8352,
      "step": 211580
    },
    {
      "epoch": 3.7147772959497183,
      "grad_norm": 0.04875939714100281,
      "learning_rate": 3.3751698127111624e-05,
      "loss": 0.8366,
      "step": 211590
    },
    {
      "epoch": 3.7149528608297198,
      "grad_norm": 0.04939511399166342,
      "learning_rate": 3.3743520429370055e-05,
      "loss": 0.8402,
      "step": 211600
    },
    {
      "epoch": 3.715128425709721,
      "grad_norm": 0.047450908840441834,
      "learning_rate": 3.373534358378099e-05,
      "loss": 0.8409,
      "step": 211610
    },
    {
      "epoch": 3.7153039905897223,
      "grad_norm": 0.06928157392591927,
      "learning_rate": 3.372716759044798e-05,
      "loss": 0.8359,
      "step": 211620
    },
    {
      "epoch": 3.7154795554697237,
      "grad_norm": 0.06462175430331334,
      "learning_rate": 3.3718992449474644e-05,
      "loss": 0.8315,
      "step": 211630
    },
    {
      "epoch": 3.7156551203497252,
      "grad_norm": 0.054856528552533,
      "learning_rate": 3.371081816096461e-05,
      "loss": 0.8435,
      "step": 211640
    },
    {
      "epoch": 3.7158306852297267,
      "grad_norm": 0.05569156508195927,
      "learning_rate": 3.370264472502134e-05,
      "loss": 0.8446,
      "step": 211650
    },
    {
      "epoch": 3.716006250109728,
      "grad_norm": 0.12912569573053617,
      "learning_rate": 3.3694472141748515e-05,
      "loss": 0.8413,
      "step": 211660
    },
    {
      "epoch": 3.7161818149897297,
      "grad_norm": 0.05753633787610997,
      "learning_rate": 3.368630041124961e-05,
      "loss": 0.8417,
      "step": 211670
    },
    {
      "epoch": 3.7163573798697307,
      "grad_norm": 0.04501624909553479,
      "learning_rate": 3.367812953362816e-05,
      "loss": 0.8381,
      "step": 211680
    },
    {
      "epoch": 3.716532944749732,
      "grad_norm": 0.04304739130829031,
      "learning_rate": 3.3669959508987765e-05,
      "loss": 0.8392,
      "step": 211690
    },
    {
      "epoch": 3.7167085096297336,
      "grad_norm": 0.05331967262727377,
      "learning_rate": 3.3661790337431834e-05,
      "loss": 0.8407,
      "step": 211700
    },
    {
      "epoch": 3.716884074509735,
      "grad_norm": 0.05803085658349286,
      "learning_rate": 3.365362201906397e-05,
      "loss": 0.834,
      "step": 211710
    },
    {
      "epoch": 3.7170596393897366,
      "grad_norm": 0.052279928490086745,
      "learning_rate": 3.3645454553987583e-05,
      "loss": 0.837,
      "step": 211720
    },
    {
      "epoch": 3.7172352042697376,
      "grad_norm": 0.08247353683865608,
      "learning_rate": 3.363728794230619e-05,
      "loss": 0.8397,
      "step": 211730
    },
    {
      "epoch": 3.7174107691497396,
      "grad_norm": 0.060532099423884246,
      "learning_rate": 3.362912218412331e-05,
      "loss": 0.8379,
      "step": 211740
    },
    {
      "epoch": 3.7175863340297406,
      "grad_norm": 0.04702709830096514,
      "learning_rate": 3.3620957279542304e-05,
      "loss": 0.8451,
      "step": 211750
    },
    {
      "epoch": 3.717761898909742,
      "grad_norm": 0.05027737067295813,
      "learning_rate": 3.361279322866674e-05,
      "loss": 0.846,
      "step": 211760
    },
    {
      "epoch": 3.7179374637897435,
      "grad_norm": 0.051426700578984956,
      "learning_rate": 3.3604630031599976e-05,
      "loss": 0.8397,
      "step": 211770
    },
    {
      "epoch": 3.718113028669745,
      "grad_norm": 0.046740327697241756,
      "learning_rate": 3.359646768844549e-05,
      "loss": 0.8484,
      "step": 211780
    },
    {
      "epoch": 3.7182885935497465,
      "grad_norm": 0.049756238202780974,
      "learning_rate": 3.3588306199306644e-05,
      "loss": 0.8379,
      "step": 211790
    },
    {
      "epoch": 3.7184641584297475,
      "grad_norm": 0.04307002480812925,
      "learning_rate": 3.358014556428688e-05,
      "loss": 0.8419,
      "step": 211800
    },
    {
      "epoch": 3.718639723309749,
      "grad_norm": 0.051941765712855026,
      "learning_rate": 3.3571985783489615e-05,
      "loss": 0.8406,
      "step": 211810
    },
    {
      "epoch": 3.7188152881897505,
      "grad_norm": 0.05389997941886982,
      "learning_rate": 3.356382685701816e-05,
      "loss": 0.8364,
      "step": 211820
    },
    {
      "epoch": 3.718990853069752,
      "grad_norm": 0.04402519095147719,
      "learning_rate": 3.355566878497601e-05,
      "loss": 0.8352,
      "step": 211830
    },
    {
      "epoch": 3.7191664179497534,
      "grad_norm": 0.05157071093856418,
      "learning_rate": 3.3547511567466417e-05,
      "loss": 0.8444,
      "step": 211840
    },
    {
      "epoch": 3.7193419828297545,
      "grad_norm": 0.048695223281801814,
      "learning_rate": 3.3539355204592795e-05,
      "loss": 0.8434,
      "step": 211850
    },
    {
      "epoch": 3.7195175477097564,
      "grad_norm": 0.05392470736880503,
      "learning_rate": 3.353119969645849e-05,
      "loss": 0.8378,
      "step": 211860
    },
    {
      "epoch": 3.7196931125897574,
      "grad_norm": 0.047699334396605904,
      "learning_rate": 3.352304504316679e-05,
      "loss": 0.8325,
      "step": 211870
    },
    {
      "epoch": 3.719868677469759,
      "grad_norm": 0.05212614602894159,
      "learning_rate": 3.351489124482108e-05,
      "loss": 0.8428,
      "step": 211880
    },
    {
      "epoch": 3.7200442423497604,
      "grad_norm": 0.06159602156912912,
      "learning_rate": 3.350673830152458e-05,
      "loss": 0.844,
      "step": 211890
    },
    {
      "epoch": 3.720219807229762,
      "grad_norm": 0.05375782164352109,
      "learning_rate": 3.349858621338066e-05,
      "loss": 0.843,
      "step": 211900
    },
    {
      "epoch": 3.7203953721097633,
      "grad_norm": 0.05219486738874985,
      "learning_rate": 3.3490434980492604e-05,
      "loss": 0.8391,
      "step": 211910
    },
    {
      "epoch": 3.7205709369897644,
      "grad_norm": 0.05604292479382059,
      "learning_rate": 3.348228460296362e-05,
      "loss": 0.8395,
      "step": 211920
    },
    {
      "epoch": 3.720746501869766,
      "grad_norm": 0.0507674263825966,
      "learning_rate": 3.3474135080897096e-05,
      "loss": 0.848,
      "step": 211930
    },
    {
      "epoch": 3.7209220667497673,
      "grad_norm": 0.055901546566579516,
      "learning_rate": 3.346598641439618e-05,
      "loss": 0.8436,
      "step": 211940
    },
    {
      "epoch": 3.721097631629769,
      "grad_norm": 0.0451306061586399,
      "learning_rate": 3.345783860356421e-05,
      "loss": 0.8365,
      "step": 211950
    },
    {
      "epoch": 3.7212731965097703,
      "grad_norm": 0.05337101361749384,
      "learning_rate": 3.344969164850432e-05,
      "loss": 0.8414,
      "step": 211960
    },
    {
      "epoch": 3.7214487613897713,
      "grad_norm": 0.04662389819927643,
      "learning_rate": 3.344154554931979e-05,
      "loss": 0.8415,
      "step": 211970
    },
    {
      "epoch": 3.7216243262697732,
      "grad_norm": 0.048586630364780205,
      "learning_rate": 3.3433400306113856e-05,
      "loss": 0.8394,
      "step": 211980
    },
    {
      "epoch": 3.7217998911497743,
      "grad_norm": 0.04213945952742679,
      "learning_rate": 3.3425255918989626e-05,
      "loss": 0.829,
      "step": 211990
    },
    {
      "epoch": 3.7219754560297758,
      "grad_norm": 0.05333755891131583,
      "learning_rate": 3.341711238805041e-05,
      "loss": 0.8322,
      "step": 212000
    },
    {
      "epoch": 3.7221510209097772,
      "grad_norm": 0.06394852355850848,
      "learning_rate": 3.340896971339931e-05,
      "loss": 0.8394,
      "step": 212010
    },
    {
      "epoch": 3.7223265857897787,
      "grad_norm": 0.06813276325580744,
      "learning_rate": 3.3400827895139514e-05,
      "loss": 0.8384,
      "step": 212020
    },
    {
      "epoch": 3.72250215066978,
      "grad_norm": 0.05653496427436215,
      "learning_rate": 3.3392686933374225e-05,
      "loss": 0.8444,
      "step": 212030
    },
    {
      "epoch": 3.722677715549781,
      "grad_norm": 0.05836730445372674,
      "learning_rate": 3.3384546828206506e-05,
      "loss": 0.8367,
      "step": 212040
    },
    {
      "epoch": 3.7228532804297827,
      "grad_norm": 0.06029061575185375,
      "learning_rate": 3.337640757973958e-05,
      "loss": 0.8457,
      "step": 212050
    },
    {
      "epoch": 3.723028845309784,
      "grad_norm": 0.05196354596632216,
      "learning_rate": 3.336826918807648e-05,
      "loss": 0.8411,
      "step": 212060
    },
    {
      "epoch": 3.7232044101897857,
      "grad_norm": 0.058684666282881756,
      "learning_rate": 3.336013165332039e-05,
      "loss": 0.8418,
      "step": 212070
    },
    {
      "epoch": 3.723379975069787,
      "grad_norm": 0.060137666732620124,
      "learning_rate": 3.335199497557442e-05,
      "loss": 0.8375,
      "step": 212080
    },
    {
      "epoch": 3.7235555399497886,
      "grad_norm": 0.057179229427399676,
      "learning_rate": 3.334385915494156e-05,
      "loss": 0.8381,
      "step": 212090
    },
    {
      "epoch": 3.72373110482979,
      "grad_norm": 0.04438574810694763,
      "learning_rate": 3.333572419152505e-05,
      "loss": 0.8383,
      "step": 212100
    },
    {
      "epoch": 3.723906669709791,
      "grad_norm": 0.07130878496480113,
      "learning_rate": 3.3327590085427855e-05,
      "loss": 0.8389,
      "step": 212110
    },
    {
      "epoch": 3.7240822345897926,
      "grad_norm": 0.054191471894848055,
      "learning_rate": 3.331945683675308e-05,
      "loss": 0.844,
      "step": 212120
    },
    {
      "epoch": 3.724257799469794,
      "grad_norm": 0.07899086215703228,
      "learning_rate": 3.3311324445603745e-05,
      "loss": 0.8315,
      "step": 212130
    },
    {
      "epoch": 3.7244333643497955,
      "grad_norm": 0.04878875553823231,
      "learning_rate": 3.3303192912082883e-05,
      "loss": 0.8365,
      "step": 212140
    },
    {
      "epoch": 3.724608929229797,
      "grad_norm": 0.060380881831167764,
      "learning_rate": 3.329506223629358e-05,
      "loss": 0.8371,
      "step": 212150
    },
    {
      "epoch": 3.724784494109798,
      "grad_norm": 0.04862085145415083,
      "learning_rate": 3.328693241833875e-05,
      "loss": 0.8382,
      "step": 212160
    },
    {
      "epoch": 3.7249600589897995,
      "grad_norm": 0.11551659781587167,
      "learning_rate": 3.327880345832153e-05,
      "loss": 0.8305,
      "step": 212170
    },
    {
      "epoch": 3.725135623869801,
      "grad_norm": 0.06553234191613798,
      "learning_rate": 3.3270675356344814e-05,
      "loss": 0.8449,
      "step": 212180
    },
    {
      "epoch": 3.7253111887498025,
      "grad_norm": 0.055278431735996406,
      "learning_rate": 3.3262548112511624e-05,
      "loss": 0.849,
      "step": 212190
    },
    {
      "epoch": 3.725486753629804,
      "grad_norm": 0.06703463675007522,
      "learning_rate": 3.325442172692496e-05,
      "loss": 0.8399,
      "step": 212200
    },
    {
      "epoch": 3.7256623185098054,
      "grad_norm": 0.04959644594201424,
      "learning_rate": 3.324629619968773e-05,
      "loss": 0.8453,
      "step": 212210
    },
    {
      "epoch": 3.725837883389807,
      "grad_norm": 0.04748227768387074,
      "learning_rate": 3.323817153090294e-05,
      "loss": 0.8421,
      "step": 212220
    },
    {
      "epoch": 3.726013448269808,
      "grad_norm": 0.0506557650207054,
      "learning_rate": 3.3230047720673474e-05,
      "loss": 0.8329,
      "step": 212230
    },
    {
      "epoch": 3.7261890131498094,
      "grad_norm": 0.05808881295052908,
      "learning_rate": 3.3221924769102284e-05,
      "loss": 0.8439,
      "step": 212240
    },
    {
      "epoch": 3.726364578029811,
      "grad_norm": 0.05300734040378929,
      "learning_rate": 3.321380267629233e-05,
      "loss": 0.8385,
      "step": 212250
    },
    {
      "epoch": 3.7265401429098124,
      "grad_norm": 0.0694981988424773,
      "learning_rate": 3.3205681442346424e-05,
      "loss": 0.8304,
      "step": 212260
    },
    {
      "epoch": 3.726715707789814,
      "grad_norm": 0.04255067545850356,
      "learning_rate": 3.3197561067367595e-05,
      "loss": 0.844,
      "step": 212270
    },
    {
      "epoch": 3.726891272669815,
      "grad_norm": 0.05465040874152766,
      "learning_rate": 3.318944155145863e-05,
      "loss": 0.8445,
      "step": 212280
    },
    {
      "epoch": 3.7270668375498164,
      "grad_norm": 0.050188947139312406,
      "learning_rate": 3.318132289472247e-05,
      "loss": 0.8381,
      "step": 212290
    },
    {
      "epoch": 3.727242402429818,
      "grad_norm": 0.05792413368015231,
      "learning_rate": 3.31732050972619e-05,
      "loss": 0.832,
      "step": 212300
    },
    {
      "epoch": 3.7274179673098193,
      "grad_norm": 0.058238972373513886,
      "learning_rate": 3.316508815917983e-05,
      "loss": 0.8369,
      "step": 212310
    },
    {
      "epoch": 3.727593532189821,
      "grad_norm": 0.04409602404997595,
      "learning_rate": 3.315697208057914e-05,
      "loss": 0.8413,
      "step": 212320
    },
    {
      "epoch": 3.7277690970698223,
      "grad_norm": 0.0628565005293095,
      "learning_rate": 3.3148856861562534e-05,
      "loss": 0.8417,
      "step": 212330
    },
    {
      "epoch": 3.7279446619498238,
      "grad_norm": 0.05400111387220291,
      "learning_rate": 3.3140742502232974e-05,
      "loss": 0.8453,
      "step": 212340
    },
    {
      "epoch": 3.728120226829825,
      "grad_norm": 0.05913862315173438,
      "learning_rate": 3.313262900269319e-05,
      "loss": 0.8386,
      "step": 212350
    },
    {
      "epoch": 3.7282957917098263,
      "grad_norm": 0.06036711694021032,
      "learning_rate": 3.312451636304601e-05,
      "loss": 0.8405,
      "step": 212360
    },
    {
      "epoch": 3.7284713565898278,
      "grad_norm": 0.04807688011569783,
      "learning_rate": 3.311640458339424e-05,
      "loss": 0.8416,
      "step": 212370
    },
    {
      "epoch": 3.7286469214698292,
      "grad_norm": 0.045189318857777015,
      "learning_rate": 3.3108293663840595e-05,
      "loss": 0.8418,
      "step": 212380
    },
    {
      "epoch": 3.7288224863498307,
      "grad_norm": 0.061674477545877,
      "learning_rate": 3.310018360448792e-05,
      "loss": 0.8373,
      "step": 212390
    },
    {
      "epoch": 3.7289980512298317,
      "grad_norm": 0.04519026934853633,
      "learning_rate": 3.3092074405438903e-05,
      "loss": 0.8391,
      "step": 212400
    },
    {
      "epoch": 3.7291736161098332,
      "grad_norm": 0.05799034002010161,
      "learning_rate": 3.3083966066796304e-05,
      "loss": 0.8363,
      "step": 212410
    },
    {
      "epoch": 3.7293491809898347,
      "grad_norm": 0.05624267961572428,
      "learning_rate": 3.307585858866292e-05,
      "loss": 0.8484,
      "step": 212420
    },
    {
      "epoch": 3.729524745869836,
      "grad_norm": 0.05169199630375892,
      "learning_rate": 3.3067751971141354e-05,
      "loss": 0.834,
      "step": 212430
    },
    {
      "epoch": 3.7297003107498377,
      "grad_norm": 0.057516438588558746,
      "learning_rate": 3.3059646214334454e-05,
      "loss": 0.8338,
      "step": 212440
    },
    {
      "epoch": 3.729875875629839,
      "grad_norm": 0.07584227653420184,
      "learning_rate": 3.3051541318344817e-05,
      "loss": 0.8368,
      "step": 212450
    },
    {
      "epoch": 3.7300514405098406,
      "grad_norm": 0.062296055092585,
      "learning_rate": 3.3043437283275225e-05,
      "loss": 0.8411,
      "step": 212460
    },
    {
      "epoch": 3.7302270053898416,
      "grad_norm": 0.050438747538415415,
      "learning_rate": 3.303533410922826e-05,
      "loss": 0.8412,
      "step": 212470
    },
    {
      "epoch": 3.730402570269843,
      "grad_norm": 0.040833666917075684,
      "learning_rate": 3.302723179630664e-05,
      "loss": 0.8387,
      "step": 212480
    },
    {
      "epoch": 3.7305781351498446,
      "grad_norm": 0.05210734949061622,
      "learning_rate": 3.301913034461305e-05,
      "loss": 0.8385,
      "step": 212490
    },
    {
      "epoch": 3.730753700029846,
      "grad_norm": 0.05873540177572583,
      "learning_rate": 3.301102975425005e-05,
      "loss": 0.8431,
      "step": 212500
    },
    {
      "epoch": 3.7309292649098476,
      "grad_norm": 0.05704858192109554,
      "learning_rate": 3.3002930025320394e-05,
      "loss": 0.8428,
      "step": 212510
    },
    {
      "epoch": 3.7311048297898486,
      "grad_norm": 0.06337763740106835,
      "learning_rate": 3.2994831157926606e-05,
      "loss": 0.839,
      "step": 212520
    },
    {
      "epoch": 3.7312803946698505,
      "grad_norm": 0.045357381613857606,
      "learning_rate": 3.298673315217136e-05,
      "loss": 0.8408,
      "step": 212530
    },
    {
      "epoch": 3.7314559595498515,
      "grad_norm": 0.0621416034388573,
      "learning_rate": 3.297863600815725e-05,
      "loss": 0.8303,
      "step": 212540
    },
    {
      "epoch": 3.731631524429853,
      "grad_norm": 0.05617285597137504,
      "learning_rate": 3.2970539725986835e-05,
      "loss": 0.8438,
      "step": 212550
    },
    {
      "epoch": 3.7318070893098545,
      "grad_norm": 0.04628398008447487,
      "learning_rate": 3.2962444305762755e-05,
      "loss": 0.8402,
      "step": 212560
    },
    {
      "epoch": 3.731982654189856,
      "grad_norm": 0.04463173371420275,
      "learning_rate": 3.29543497475875e-05,
      "loss": 0.8427,
      "step": 212570
    },
    {
      "epoch": 3.7321582190698575,
      "grad_norm": 0.06288880794561229,
      "learning_rate": 3.2946256051563686e-05,
      "loss": 0.8435,
      "step": 212580
    },
    {
      "epoch": 3.7323337839498585,
      "grad_norm": 0.05146609206765846,
      "learning_rate": 3.293816321779388e-05,
      "loss": 0.8309,
      "step": 212590
    },
    {
      "epoch": 3.73250934882986,
      "grad_norm": 0.05535632597032615,
      "learning_rate": 3.293007124638053e-05,
      "loss": 0.8428,
      "step": 212600
    },
    {
      "epoch": 3.7326849137098614,
      "grad_norm": 0.04881312737481225,
      "learning_rate": 3.292198013742628e-05,
      "loss": 0.8389,
      "step": 212610
    },
    {
      "epoch": 3.732860478589863,
      "grad_norm": 0.0537407593086242,
      "learning_rate": 3.2913889891033564e-05,
      "loss": 0.8412,
      "step": 212620
    },
    {
      "epoch": 3.7330360434698644,
      "grad_norm": 0.05041473028290572,
      "learning_rate": 3.290580050730493e-05,
      "loss": 0.8384,
      "step": 212630
    },
    {
      "epoch": 3.7332116083498654,
      "grad_norm": 0.06015939528140917,
      "learning_rate": 3.2897711986342835e-05,
      "loss": 0.8417,
      "step": 212640
    },
    {
      "epoch": 3.7333871732298674,
      "grad_norm": 0.04964331812214613,
      "learning_rate": 3.288962432824979e-05,
      "loss": 0.8438,
      "step": 212650
    },
    {
      "epoch": 3.7335627381098684,
      "grad_norm": 0.05900943179909198,
      "learning_rate": 3.288153753312827e-05,
      "loss": 0.8333,
      "step": 212660
    },
    {
      "epoch": 3.73373830298987,
      "grad_norm": 0.04751677025590802,
      "learning_rate": 3.2873451601080663e-05,
      "loss": 0.836,
      "step": 212670
    },
    {
      "epoch": 3.7339138678698713,
      "grad_norm": 0.04407667864348086,
      "learning_rate": 3.286536653220954e-05,
      "loss": 0.841,
      "step": 212680
    },
    {
      "epoch": 3.734089432749873,
      "grad_norm": 0.045556471523579604,
      "learning_rate": 3.285728232661727e-05,
      "loss": 0.8498,
      "step": 212690
    },
    {
      "epoch": 3.7342649976298743,
      "grad_norm": 0.04230991139030103,
      "learning_rate": 3.2849198984406274e-05,
      "loss": 0.8447,
      "step": 212700
    },
    {
      "epoch": 3.7344405625098753,
      "grad_norm": 0.042535770334563724,
      "learning_rate": 3.2841116505679025e-05,
      "loss": 0.8356,
      "step": 212710
    },
    {
      "epoch": 3.734616127389877,
      "grad_norm": 0.058166415210747356,
      "learning_rate": 3.283303489053787e-05,
      "loss": 0.8342,
      "step": 212720
    },
    {
      "epoch": 3.7347916922698783,
      "grad_norm": 0.05469852175769484,
      "learning_rate": 3.282495413908525e-05,
      "loss": 0.8325,
      "step": 212730
    },
    {
      "epoch": 3.7349672571498798,
      "grad_norm": 0.05183615904564911,
      "learning_rate": 3.281687425142349e-05,
      "loss": 0.8391,
      "step": 212740
    },
    {
      "epoch": 3.7351428220298812,
      "grad_norm": 0.05339543169953448,
      "learning_rate": 3.280879522765501e-05,
      "loss": 0.8374,
      "step": 212750
    },
    {
      "epoch": 3.7353183869098823,
      "grad_norm": 0.0562632219609613,
      "learning_rate": 3.280071706788219e-05,
      "loss": 0.8425,
      "step": 212760
    },
    {
      "epoch": 3.735493951789884,
      "grad_norm": 0.046443974022541655,
      "learning_rate": 3.27926397722073e-05,
      "loss": 0.8324,
      "step": 212770
    },
    {
      "epoch": 3.7356695166698852,
      "grad_norm": 0.053582151914072756,
      "learning_rate": 3.278456334073278e-05,
      "loss": 0.8348,
      "step": 212780
    },
    {
      "epoch": 3.7358450815498867,
      "grad_norm": 0.07331862059097,
      "learning_rate": 3.277648777356091e-05,
      "loss": 0.8338,
      "step": 212790
    },
    {
      "epoch": 3.736020646429888,
      "grad_norm": 0.054479239002870036,
      "learning_rate": 3.276841307079402e-05,
      "loss": 0.8474,
      "step": 212800
    },
    {
      "epoch": 3.7361962113098897,
      "grad_norm": 0.05663536542361728,
      "learning_rate": 3.2760339232534395e-05,
      "loss": 0.8483,
      "step": 212810
    },
    {
      "epoch": 3.736371776189891,
      "grad_norm": 0.05167780158042225,
      "learning_rate": 3.275226625888435e-05,
      "loss": 0.8416,
      "step": 212820
    },
    {
      "epoch": 3.736547341069892,
      "grad_norm": 0.06033617562744012,
      "learning_rate": 3.274419414994621e-05,
      "loss": 0.8365,
      "step": 212830
    },
    {
      "epoch": 3.7367229059498936,
      "grad_norm": 0.04518845207965968,
      "learning_rate": 3.2736122905822136e-05,
      "loss": 0.8419,
      "step": 212840
    },
    {
      "epoch": 3.736898470829895,
      "grad_norm": 0.05095376681503745,
      "learning_rate": 3.272805252661453e-05,
      "loss": 0.8423,
      "step": 212850
    },
    {
      "epoch": 3.7370740357098966,
      "grad_norm": 0.045997131138368424,
      "learning_rate": 3.2719983012425565e-05,
      "loss": 0.8459,
      "step": 212860
    },
    {
      "epoch": 3.737249600589898,
      "grad_norm": 0.051280497066916245,
      "learning_rate": 3.271191436335749e-05,
      "loss": 0.837,
      "step": 212870
    },
    {
      "epoch": 3.7374251654698996,
      "grad_norm": 0.0431800636543663,
      "learning_rate": 3.270384657951257e-05,
      "loss": 0.8388,
      "step": 212880
    },
    {
      "epoch": 3.737600730349901,
      "grad_norm": 0.06959542722354593,
      "learning_rate": 3.2695779660992984e-05,
      "loss": 0.8377,
      "step": 212890
    },
    {
      "epoch": 3.737776295229902,
      "grad_norm": 0.05167459234588063,
      "learning_rate": 3.2687713607900995e-05,
      "loss": 0.8406,
      "step": 212900
    },
    {
      "epoch": 3.7379518601099035,
      "grad_norm": 0.05168351659881412,
      "learning_rate": 3.267964842033873e-05,
      "loss": 0.8359,
      "step": 212910
    },
    {
      "epoch": 3.738127424989905,
      "grad_norm": 0.06691859158525736,
      "learning_rate": 3.2671584098408413e-05,
      "loss": 0.8379,
      "step": 212920
    },
    {
      "epoch": 3.7383029898699065,
      "grad_norm": 0.047076622863356184,
      "learning_rate": 3.266352064221225e-05,
      "loss": 0.843,
      "step": 212930
    },
    {
      "epoch": 3.738478554749908,
      "grad_norm": 0.0663041058916244,
      "learning_rate": 3.265545805185232e-05,
      "loss": 0.831,
      "step": 212940
    },
    {
      "epoch": 3.738654119629909,
      "grad_norm": 0.05222548693285011,
      "learning_rate": 3.26473963274309e-05,
      "loss": 0.8387,
      "step": 212950
    },
    {
      "epoch": 3.7388296845099105,
      "grad_norm": 0.04962120821444543,
      "learning_rate": 3.263933546905002e-05,
      "loss": 0.8379,
      "step": 212960
    },
    {
      "epoch": 3.739005249389912,
      "grad_norm": 0.06169394914642336,
      "learning_rate": 3.26312754768119e-05,
      "loss": 0.837,
      "step": 212970
    },
    {
      "epoch": 3.7391808142699134,
      "grad_norm": 0.048581970174286976,
      "learning_rate": 3.26232163508186e-05,
      "loss": 0.8414,
      "step": 212980
    },
    {
      "epoch": 3.739356379149915,
      "grad_norm": 0.04754009649103007,
      "learning_rate": 3.2615158091172245e-05,
      "loss": 0.8377,
      "step": 212990
    },
    {
      "epoch": 3.7395319440299164,
      "grad_norm": 0.0863690383297236,
      "learning_rate": 3.260710069797498e-05,
      "loss": 0.8335,
      "step": 213000
    },
    {
      "epoch": 3.739707508909918,
      "grad_norm": 0.05645693513982186,
      "learning_rate": 3.259904417132879e-05,
      "loss": 0.8424,
      "step": 213010
    },
    {
      "epoch": 3.739883073789919,
      "grad_norm": 0.0576053572986354,
      "learning_rate": 3.2590988511335875e-05,
      "loss": 0.8317,
      "step": 213020
    },
    {
      "epoch": 3.7400586386699204,
      "grad_norm": 0.059334842213628676,
      "learning_rate": 3.258293371809822e-05,
      "loss": 0.8423,
      "step": 213030
    },
    {
      "epoch": 3.740234203549922,
      "grad_norm": 0.049204802788126306,
      "learning_rate": 3.257487979171794e-05,
      "loss": 0.8348,
      "step": 213040
    },
    {
      "epoch": 3.7404097684299233,
      "grad_norm": 0.06519948768962308,
      "learning_rate": 3.2566826732297014e-05,
      "loss": 0.8413,
      "step": 213050
    },
    {
      "epoch": 3.740585333309925,
      "grad_norm": 0.055494773664093056,
      "learning_rate": 3.255877453993751e-05,
      "loss": 0.8397,
      "step": 213060
    },
    {
      "epoch": 3.740760898189926,
      "grad_norm": 0.053377860562986205,
      "learning_rate": 3.255072321474148e-05,
      "loss": 0.8373,
      "step": 213070
    },
    {
      "epoch": 3.7409364630699273,
      "grad_norm": 0.05359409376772787,
      "learning_rate": 3.254267275681088e-05,
      "loss": 0.8492,
      "step": 213080
    },
    {
      "epoch": 3.741112027949929,
      "grad_norm": 0.04834014408127544,
      "learning_rate": 3.253462316624773e-05,
      "loss": 0.8419,
      "step": 213090
    },
    {
      "epoch": 3.7412875928299303,
      "grad_norm": 0.043912593123834415,
      "learning_rate": 3.252657444315405e-05,
      "loss": 0.8419,
      "step": 213100
    },
    {
      "epoch": 3.7414631577099318,
      "grad_norm": 0.04785295730062014,
      "learning_rate": 3.2518526587631745e-05,
      "loss": 0.8447,
      "step": 213110
    },
    {
      "epoch": 3.7416387225899332,
      "grad_norm": 0.05574305330249405,
      "learning_rate": 3.251047959978289e-05,
      "loss": 0.8457,
      "step": 213120
    },
    {
      "epoch": 3.7418142874699347,
      "grad_norm": 0.06981787808186828,
      "learning_rate": 3.250243347970935e-05,
      "loss": 0.8367,
      "step": 213130
    },
    {
      "epoch": 3.7419898523499358,
      "grad_norm": 0.05365835055250077,
      "learning_rate": 3.2494388227513143e-05,
      "loss": 0.8447,
      "step": 213140
    },
    {
      "epoch": 3.7421654172299372,
      "grad_norm": 0.0645492551735919,
      "learning_rate": 3.2486343843296135e-05,
      "loss": 0.8353,
      "step": 213150
    },
    {
      "epoch": 3.7423409821099387,
      "grad_norm": 0.04144643397756968,
      "learning_rate": 3.247830032716027e-05,
      "loss": 0.846,
      "step": 213160
    },
    {
      "epoch": 3.74251654698994,
      "grad_norm": 0.05608663679280813,
      "learning_rate": 3.247025767920751e-05,
      "loss": 0.8427,
      "step": 213170
    },
    {
      "epoch": 3.7426921118699417,
      "grad_norm": 0.05051438938773103,
      "learning_rate": 3.2462215899539656e-05,
      "loss": 0.8462,
      "step": 213180
    },
    {
      "epoch": 3.7428676767499427,
      "grad_norm": 0.04994025353065721,
      "learning_rate": 3.245417498825874e-05,
      "loss": 0.8372,
      "step": 213190
    },
    {
      "epoch": 3.7430432416299446,
      "grad_norm": 0.07026939395869763,
      "learning_rate": 3.2446134945466504e-05,
      "loss": 0.8382,
      "step": 213200
    },
    {
      "epoch": 3.7432188065099457,
      "grad_norm": 0.044959209177840215,
      "learning_rate": 3.243809577126492e-05,
      "loss": 0.8406,
      "step": 213210
    },
    {
      "epoch": 3.743394371389947,
      "grad_norm": 0.05531900276404251,
      "learning_rate": 3.243005746575578e-05,
      "loss": 0.8441,
      "step": 213220
    },
    {
      "epoch": 3.7435699362699486,
      "grad_norm": 0.05576124121854887,
      "learning_rate": 3.242202002904094e-05,
      "loss": 0.84,
      "step": 213230
    },
    {
      "epoch": 3.74374550114995,
      "grad_norm": 0.05664139272585752,
      "learning_rate": 3.2413983461222286e-05,
      "loss": 0.8428,
      "step": 213240
    },
    {
      "epoch": 3.7439210660299516,
      "grad_norm": 0.05297850456416652,
      "learning_rate": 3.240594776240158e-05,
      "loss": 0.8443,
      "step": 213250
    },
    {
      "epoch": 3.7440966309099526,
      "grad_norm": 0.05494653088553492,
      "learning_rate": 3.239791293268065e-05,
      "loss": 0.8387,
      "step": 213260
    },
    {
      "epoch": 3.744272195789954,
      "grad_norm": 0.04772555079806115,
      "learning_rate": 3.238987897216135e-05,
      "loss": 0.8419,
      "step": 213270
    },
    {
      "epoch": 3.7444477606699555,
      "grad_norm": 0.050993507986515925,
      "learning_rate": 3.238184588094537e-05,
      "loss": 0.8416,
      "step": 213280
    },
    {
      "epoch": 3.744623325549957,
      "grad_norm": 0.04668356267232632,
      "learning_rate": 3.237381365913462e-05,
      "loss": 0.8347,
      "step": 213290
    },
    {
      "epoch": 3.7447988904299585,
      "grad_norm": 0.05747486312542385,
      "learning_rate": 3.236578230683076e-05,
      "loss": 0.8365,
      "step": 213300
    },
    {
      "epoch": 3.7449744553099595,
      "grad_norm": 0.048419427879040335,
      "learning_rate": 3.235775182413563e-05,
      "loss": 0.8392,
      "step": 213310
    },
    {
      "epoch": 3.7451500201899615,
      "grad_norm": 0.05188142097898497,
      "learning_rate": 3.2349722211150914e-05,
      "loss": 0.8394,
      "step": 213320
    },
    {
      "epoch": 3.7453255850699625,
      "grad_norm": 0.060438773503954095,
      "learning_rate": 3.234169346797836e-05,
      "loss": 0.8385,
      "step": 213330
    },
    {
      "epoch": 3.745501149949964,
      "grad_norm": 0.06562432787830273,
      "learning_rate": 3.233366559471975e-05,
      "loss": 0.8307,
      "step": 213340
    },
    {
      "epoch": 3.7456767148299654,
      "grad_norm": 0.05596276046089051,
      "learning_rate": 3.232563859147669e-05,
      "loss": 0.8301,
      "step": 213350
    },
    {
      "epoch": 3.745852279709967,
      "grad_norm": 0.05111357869950322,
      "learning_rate": 3.231761245835102e-05,
      "loss": 0.8448,
      "step": 213360
    },
    {
      "epoch": 3.7460278445899684,
      "grad_norm": 0.06104920430788736,
      "learning_rate": 3.2309587195444326e-05,
      "loss": 0.8436,
      "step": 213370
    },
    {
      "epoch": 3.7462034094699694,
      "grad_norm": 0.04315939170816055,
      "learning_rate": 3.2301562802858355e-05,
      "loss": 0.8367,
      "step": 213380
    },
    {
      "epoch": 3.746378974349971,
      "grad_norm": 0.05121739738738398,
      "learning_rate": 3.229353928069472e-05,
      "loss": 0.8396,
      "step": 213390
    },
    {
      "epoch": 3.7465545392299724,
      "grad_norm": 0.04762320389251111,
      "learning_rate": 3.228551662905511e-05,
      "loss": 0.8336,
      "step": 213400
    },
    {
      "epoch": 3.746730104109974,
      "grad_norm": 0.05264540865600846,
      "learning_rate": 3.2277494848041215e-05,
      "loss": 0.8389,
      "step": 213410
    },
    {
      "epoch": 3.7469056689899753,
      "grad_norm": 0.05206159906368804,
      "learning_rate": 3.226947393775459e-05,
      "loss": 0.8378,
      "step": 213420
    },
    {
      "epoch": 3.7470812338699764,
      "grad_norm": 0.04967815024854766,
      "learning_rate": 3.226145389829689e-05,
      "loss": 0.8349,
      "step": 213430
    },
    {
      "epoch": 3.7472567987499783,
      "grad_norm": 0.05500309692702122,
      "learning_rate": 3.2253434729769786e-05,
      "loss": 0.8509,
      "step": 213440
    },
    {
      "epoch": 3.7474323636299793,
      "grad_norm": 0.06060620454170647,
      "learning_rate": 3.224541643227478e-05,
      "loss": 0.835,
      "step": 213450
    },
    {
      "epoch": 3.747607928509981,
      "grad_norm": 0.09249681476633348,
      "learning_rate": 3.2237399005913586e-05,
      "loss": 0.8343,
      "step": 213460
    },
    {
      "epoch": 3.7477834933899823,
      "grad_norm": 0.04816263909446041,
      "learning_rate": 3.222938245078769e-05,
      "loss": 0.8302,
      "step": 213470
    },
    {
      "epoch": 3.7479590582699838,
      "grad_norm": 0.052577426731901,
      "learning_rate": 3.222136676699873e-05,
      "loss": 0.8472,
      "step": 213480
    },
    {
      "epoch": 3.7481346231499852,
      "grad_norm": 0.054376931701859796,
      "learning_rate": 3.221335195464821e-05,
      "loss": 0.8413,
      "step": 213490
    },
    {
      "epoch": 3.7483101880299863,
      "grad_norm": 0.0614935105535678,
      "learning_rate": 3.2205338013837705e-05,
      "loss": 0.8402,
      "step": 213500
    },
    {
      "epoch": 3.7484857529099878,
      "grad_norm": 0.04978736210627789,
      "learning_rate": 3.2197324944668765e-05,
      "loss": 0.8436,
      "step": 213510
    },
    {
      "epoch": 3.7486613177899892,
      "grad_norm": 0.06594781793569407,
      "learning_rate": 3.2189312747242864e-05,
      "loss": 0.8381,
      "step": 213520
    },
    {
      "epoch": 3.7488368826699907,
      "grad_norm": 0.07056900841448353,
      "learning_rate": 3.21813014216616e-05,
      "loss": 0.8386,
      "step": 213530
    },
    {
      "epoch": 3.749012447549992,
      "grad_norm": 0.06395671849155876,
      "learning_rate": 3.217329096802642e-05,
      "loss": 0.8298,
      "step": 213540
    },
    {
      "epoch": 3.7491880124299937,
      "grad_norm": 0.047459862464976334,
      "learning_rate": 3.216528138643886e-05,
      "loss": 0.8372,
      "step": 213550
    },
    {
      "epoch": 3.749363577309995,
      "grad_norm": 0.04704212463139129,
      "learning_rate": 3.215727267700034e-05,
      "loss": 0.8404,
      "step": 213560
    },
    {
      "epoch": 3.749539142189996,
      "grad_norm": 0.056722427534958646,
      "learning_rate": 3.214926483981237e-05,
      "loss": 0.8344,
      "step": 213570
    },
    {
      "epoch": 3.7497147070699977,
      "grad_norm": 0.04955617832678936,
      "learning_rate": 3.2141257874976455e-05,
      "loss": 0.8459,
      "step": 213580
    },
    {
      "epoch": 3.749890271949999,
      "grad_norm": 0.06899175510832217,
      "learning_rate": 3.213325178259396e-05,
      "loss": 0.8327,
      "step": 213590
    },
    {
      "epoch": 3.7500658368300006,
      "grad_norm": 0.04705441625306072,
      "learning_rate": 3.212524656276636e-05,
      "loss": 0.8418,
      "step": 213600
    },
    {
      "epoch": 3.750241401710002,
      "grad_norm": 0.046864475198779516,
      "learning_rate": 3.211724221559511e-05,
      "loss": 0.8454,
      "step": 213610
    },
    {
      "epoch": 3.750416966590003,
      "grad_norm": 0.05860389818951293,
      "learning_rate": 3.2109238741181555e-05,
      "loss": 0.8465,
      "step": 213620
    },
    {
      "epoch": 3.7505925314700046,
      "grad_norm": 0.05357046673281773,
      "learning_rate": 3.21012361396272e-05,
      "loss": 0.8417,
      "step": 213630
    },
    {
      "epoch": 3.750768096350006,
      "grad_norm": 0.05986542209869112,
      "learning_rate": 3.209323441103336e-05,
      "loss": 0.8444,
      "step": 213640
    },
    {
      "epoch": 3.7509436612300076,
      "grad_norm": 0.054366022337615866,
      "learning_rate": 3.208523355550148e-05,
      "loss": 0.8398,
      "step": 213650
    },
    {
      "epoch": 3.751119226110009,
      "grad_norm": 0.0506811598667191,
      "learning_rate": 3.207723357313286e-05,
      "loss": 0.8425,
      "step": 213660
    },
    {
      "epoch": 3.7512947909900105,
      "grad_norm": 0.04125123623459474,
      "learning_rate": 3.206923446402891e-05,
      "loss": 0.8464,
      "step": 213670
    },
    {
      "epoch": 3.751470355870012,
      "grad_norm": 0.045597924349650196,
      "learning_rate": 3.2061236228290994e-05,
      "loss": 0.8393,
      "step": 213680
    },
    {
      "epoch": 3.751645920750013,
      "grad_norm": 0.04328383646848937,
      "learning_rate": 3.205323886602036e-05,
      "loss": 0.8335,
      "step": 213690
    },
    {
      "epoch": 3.7518214856300145,
      "grad_norm": 0.06740286412648847,
      "learning_rate": 3.2045242377318474e-05,
      "loss": 0.8352,
      "step": 213700
    },
    {
      "epoch": 3.751997050510016,
      "grad_norm": 0.04099787781398983,
      "learning_rate": 3.2037246762286553e-05,
      "loss": 0.8358,
      "step": 213710
    },
    {
      "epoch": 3.7521726153900175,
      "grad_norm": 0.04985108823289829,
      "learning_rate": 3.202925202102596e-05,
      "loss": 0.8355,
      "step": 213720
    },
    {
      "epoch": 3.752348180270019,
      "grad_norm": 0.07003135922803369,
      "learning_rate": 3.202125815363792e-05,
      "loss": 0.8348,
      "step": 213730
    },
    {
      "epoch": 3.75252374515002,
      "grad_norm": 0.06644724657068816,
      "learning_rate": 3.2013265160223775e-05,
      "loss": 0.8427,
      "step": 213740
    },
    {
      "epoch": 3.7526993100300214,
      "grad_norm": 0.05447165400608822,
      "learning_rate": 3.2005273040884806e-05,
      "loss": 0.8327,
      "step": 213750
    },
    {
      "epoch": 3.752874874910023,
      "grad_norm": 0.07998833882818103,
      "learning_rate": 3.199728179572222e-05,
      "loss": 0.8347,
      "step": 213760
    },
    {
      "epoch": 3.7530504397900244,
      "grad_norm": 0.04143686460733736,
      "learning_rate": 3.1989291424837294e-05,
      "loss": 0.8404,
      "step": 213770
    },
    {
      "epoch": 3.753226004670026,
      "grad_norm": 0.05703999206148614,
      "learning_rate": 3.1981301928331306e-05,
      "loss": 0.8356,
      "step": 213780
    },
    {
      "epoch": 3.7534015695500274,
      "grad_norm": 0.05821492262757193,
      "learning_rate": 3.1973313306305386e-05,
      "loss": 0.8354,
      "step": 213790
    },
    {
      "epoch": 3.753577134430029,
      "grad_norm": 0.0558246106230869,
      "learning_rate": 3.196532555886088e-05,
      "loss": 0.8439,
      "step": 213800
    },
    {
      "epoch": 3.75375269931003,
      "grad_norm": 0.046871973928098526,
      "learning_rate": 3.195733868609889e-05,
      "loss": 0.8479,
      "step": 213810
    },
    {
      "epoch": 3.7539282641900313,
      "grad_norm": 0.053174997930230984,
      "learning_rate": 3.19493526881207e-05,
      "loss": 0.8448,
      "step": 213820
    },
    {
      "epoch": 3.754103829070033,
      "grad_norm": 0.06634945551396064,
      "learning_rate": 3.19413675650274e-05,
      "loss": 0.8444,
      "step": 213830
    },
    {
      "epoch": 3.7542793939500343,
      "grad_norm": 0.08395378315318411,
      "learning_rate": 3.193338331692021e-05,
      "loss": 0.8417,
      "step": 213840
    },
    {
      "epoch": 3.7544549588300358,
      "grad_norm": 0.053491354587455756,
      "learning_rate": 3.192539994390033e-05,
      "loss": 0.8439,
      "step": 213850
    },
    {
      "epoch": 3.754630523710037,
      "grad_norm": 0.04519558021713755,
      "learning_rate": 3.19174174460688e-05,
      "loss": 0.84,
      "step": 213860
    },
    {
      "epoch": 3.7548060885900387,
      "grad_norm": 0.05113229809872996,
      "learning_rate": 3.190943582352689e-05,
      "loss": 0.8407,
      "step": 213870
    },
    {
      "epoch": 3.7549816534700398,
      "grad_norm": 0.06001862333303167,
      "learning_rate": 3.190145507637564e-05,
      "loss": 0.8438,
      "step": 213880
    },
    {
      "epoch": 3.7551572183500412,
      "grad_norm": 0.06615345390117666,
      "learning_rate": 3.1893475204716236e-05,
      "loss": 0.8328,
      "step": 213890
    },
    {
      "epoch": 3.7553327832300427,
      "grad_norm": 0.05083237032673574,
      "learning_rate": 3.18854962086497e-05,
      "loss": 0.8456,
      "step": 213900
    },
    {
      "epoch": 3.755508348110044,
      "grad_norm": 0.05662274195860441,
      "learning_rate": 3.187751808827718e-05,
      "loss": 0.8331,
      "step": 213910
    },
    {
      "epoch": 3.7556839129900457,
      "grad_norm": 0.059098721563691926,
      "learning_rate": 3.18695408436998e-05,
      "loss": 0.8392,
      "step": 213920
    },
    {
      "epoch": 3.7558594778700467,
      "grad_norm": 0.05391597183858666,
      "learning_rate": 3.186156447501853e-05,
      "loss": 0.8403,
      "step": 213930
    },
    {
      "epoch": 3.756035042750048,
      "grad_norm": 0.044621503805228786,
      "learning_rate": 3.1853588982334504e-05,
      "loss": 0.841,
      "step": 213940
    },
    {
      "epoch": 3.7562106076300497,
      "grad_norm": 0.05258137856841265,
      "learning_rate": 3.184561436574878e-05,
      "loss": 0.8399,
      "step": 213950
    },
    {
      "epoch": 3.756386172510051,
      "grad_norm": 0.0447774772178877,
      "learning_rate": 3.183764062536232e-05,
      "loss": 0.8484,
      "step": 213960
    },
    {
      "epoch": 3.7565617373900526,
      "grad_norm": 0.04956275412081876,
      "learning_rate": 3.182966776127626e-05,
      "loss": 0.8429,
      "step": 213970
    },
    {
      "epoch": 3.7567373022700536,
      "grad_norm": 0.03885121968021897,
      "learning_rate": 3.1821695773591545e-05,
      "loss": 0.8404,
      "step": 213980
    },
    {
      "epoch": 3.7569128671500556,
      "grad_norm": 0.04748589779999779,
      "learning_rate": 3.181372466240923e-05,
      "loss": 0.8439,
      "step": 213990
    },
    {
      "epoch": 3.7570884320300566,
      "grad_norm": 0.0519219003196956,
      "learning_rate": 3.180575442783025e-05,
      "loss": 0.841,
      "step": 214000
    },
    {
      "epoch": 3.757263996910058,
      "grad_norm": 0.051529702831242134,
      "learning_rate": 3.179778506995562e-05,
      "loss": 0.8344,
      "step": 214010
    },
    {
      "epoch": 3.7574395617900596,
      "grad_norm": 0.047060446307170246,
      "learning_rate": 3.1789816588886364e-05,
      "loss": 0.8388,
      "step": 214020
    },
    {
      "epoch": 3.757615126670061,
      "grad_norm": 0.055677310971003126,
      "learning_rate": 3.1781848984723315e-05,
      "loss": 0.8431,
      "step": 214030
    },
    {
      "epoch": 3.7577906915500625,
      "grad_norm": 0.06897643085017577,
      "learning_rate": 3.1773882257567576e-05,
      "loss": 0.844,
      "step": 214040
    },
    {
      "epoch": 3.7579662564300635,
      "grad_norm": 0.0442353697057626,
      "learning_rate": 3.176591640751998e-05,
      "loss": 0.8446,
      "step": 214050
    },
    {
      "epoch": 3.758141821310065,
      "grad_norm": 0.057845810033544005,
      "learning_rate": 3.175795143468153e-05,
      "loss": 0.8376,
      "step": 214060
    },
    {
      "epoch": 3.7583173861900665,
      "grad_norm": 0.05713517601033358,
      "learning_rate": 3.1749987339153066e-05,
      "loss": 0.8468,
      "step": 214070
    },
    {
      "epoch": 3.758492951070068,
      "grad_norm": 0.04449143447031028,
      "learning_rate": 3.1742024121035534e-05,
      "loss": 0.8499,
      "step": 214080
    },
    {
      "epoch": 3.7586685159500695,
      "grad_norm": 0.05026651155529376,
      "learning_rate": 3.173406178042987e-05,
      "loss": 0.8418,
      "step": 214090
    },
    {
      "epoch": 3.7588440808300705,
      "grad_norm": 0.06450275738445041,
      "learning_rate": 3.1726100317436874e-05,
      "loss": 0.8316,
      "step": 214100
    },
    {
      "epoch": 3.7590196457100724,
      "grad_norm": 0.06434621798063817,
      "learning_rate": 3.1718139732157475e-05,
      "loss": 0.8405,
      "step": 214110
    },
    {
      "epoch": 3.7591952105900734,
      "grad_norm": 0.054184979828047944,
      "learning_rate": 3.171018002469254e-05,
      "loss": 0.8342,
      "step": 214120
    },
    {
      "epoch": 3.759370775470075,
      "grad_norm": 0.06918711935945493,
      "learning_rate": 3.170222119514286e-05,
      "loss": 0.8342,
      "step": 214130
    },
    {
      "epoch": 3.7595463403500764,
      "grad_norm": 0.0467718396242105,
      "learning_rate": 3.169426324360937e-05,
      "loss": 0.842,
      "step": 214140
    },
    {
      "epoch": 3.759721905230078,
      "grad_norm": 0.04757587984929311,
      "learning_rate": 3.168630617019282e-05,
      "loss": 0.8384,
      "step": 214150
    },
    {
      "epoch": 3.7598974701100794,
      "grad_norm": 0.0412468609287616,
      "learning_rate": 3.1678349974994084e-05,
      "loss": 0.8422,
      "step": 214160
    },
    {
      "epoch": 3.7600730349900804,
      "grad_norm": 0.06775811921951674,
      "learning_rate": 3.167039465811391e-05,
      "loss": 0.8447,
      "step": 214170
    },
    {
      "epoch": 3.760248599870082,
      "grad_norm": 0.06932886932636603,
      "learning_rate": 3.166244021965311e-05,
      "loss": 0.8433,
      "step": 214180
    },
    {
      "epoch": 3.7604241647500833,
      "grad_norm": 0.05087246654773243,
      "learning_rate": 3.165448665971253e-05,
      "loss": 0.8517,
      "step": 214190
    },
    {
      "epoch": 3.760599729630085,
      "grad_norm": 0.05445977732139236,
      "learning_rate": 3.164653397839282e-05,
      "loss": 0.834,
      "step": 214200
    },
    {
      "epoch": 3.7607752945100863,
      "grad_norm": 0.047695214394711354,
      "learning_rate": 3.163858217579488e-05,
      "loss": 0.8473,
      "step": 214210
    },
    {
      "epoch": 3.7609508593900873,
      "grad_norm": 0.05145803433670018,
      "learning_rate": 3.163063125201937e-05,
      "loss": 0.8387,
      "step": 214220
    },
    {
      "epoch": 3.7611264242700893,
      "grad_norm": 0.05226971287621462,
      "learning_rate": 3.162268120716708e-05,
      "loss": 0.833,
      "step": 214230
    },
    {
      "epoch": 3.7613019891500903,
      "grad_norm": 0.03938489541344219,
      "learning_rate": 3.16147320413387e-05,
      "loss": 0.843,
      "step": 214240
    },
    {
      "epoch": 3.7614775540300918,
      "grad_norm": 0.05448312895268502,
      "learning_rate": 3.160678375463495e-05,
      "loss": 0.8455,
      "step": 214250
    },
    {
      "epoch": 3.7616531189100932,
      "grad_norm": 0.050366819185102926,
      "learning_rate": 3.159883634715659e-05,
      "loss": 0.8498,
      "step": 214260
    },
    {
      "epoch": 3.7618286837900947,
      "grad_norm": 0.06605223411829785,
      "learning_rate": 3.159088981900423e-05,
      "loss": 0.8441,
      "step": 214270
    },
    {
      "epoch": 3.762004248670096,
      "grad_norm": 0.04857349802126919,
      "learning_rate": 3.15829441702786e-05,
      "loss": 0.8379,
      "step": 214280
    },
    {
      "epoch": 3.7621798135500972,
      "grad_norm": 0.07048388432624449,
      "learning_rate": 3.157499940108041e-05,
      "loss": 0.8407,
      "step": 214290
    },
    {
      "epoch": 3.7623553784300987,
      "grad_norm": 0.10332142879629053,
      "learning_rate": 3.156705551151021e-05,
      "loss": 0.8378,
      "step": 214300
    },
    {
      "epoch": 3.7625309433101,
      "grad_norm": 0.05030540057669689,
      "learning_rate": 3.1559112501668795e-05,
      "loss": 0.837,
      "step": 214310
    },
    {
      "epoch": 3.7627065081901017,
      "grad_norm": 0.049760810080321516,
      "learning_rate": 3.155117037165671e-05,
      "loss": 0.8397,
      "step": 214320
    },
    {
      "epoch": 3.762882073070103,
      "grad_norm": 0.0514411586564239,
      "learning_rate": 3.154322912157462e-05,
      "loss": 0.8309,
      "step": 214330
    },
    {
      "epoch": 3.7630576379501046,
      "grad_norm": 0.048846354078957446,
      "learning_rate": 3.1535288751523094e-05,
      "loss": 0.8423,
      "step": 214340
    },
    {
      "epoch": 3.763233202830106,
      "grad_norm": 0.05629231890525461,
      "learning_rate": 3.152734926160279e-05,
      "loss": 0.8359,
      "step": 214350
    },
    {
      "epoch": 3.763408767710107,
      "grad_norm": 0.05257035444178978,
      "learning_rate": 3.151941065191431e-05,
      "loss": 0.8381,
      "step": 214360
    },
    {
      "epoch": 3.7635843325901086,
      "grad_norm": 0.05054777424359432,
      "learning_rate": 3.151147292255815e-05,
      "loss": 0.8379,
      "step": 214370
    },
    {
      "epoch": 3.76375989747011,
      "grad_norm": 0.053490764587493314,
      "learning_rate": 3.1503536073635006e-05,
      "loss": 0.8384,
      "step": 214380
    },
    {
      "epoch": 3.7639354623501116,
      "grad_norm": 0.06545251587475875,
      "learning_rate": 3.1495600105245345e-05,
      "loss": 0.8355,
      "step": 214390
    },
    {
      "epoch": 3.764111027230113,
      "grad_norm": 0.053474648589710544,
      "learning_rate": 3.1487665017489785e-05,
      "loss": 0.8347,
      "step": 214400
    },
    {
      "epoch": 3.764286592110114,
      "grad_norm": 0.05340810821084325,
      "learning_rate": 3.14797308104688e-05,
      "loss": 0.8379,
      "step": 214410
    },
    {
      "epoch": 3.7644621569901155,
      "grad_norm": 0.06483030126410083,
      "learning_rate": 3.147179748428295e-05,
      "loss": 0.8389,
      "step": 214420
    },
    {
      "epoch": 3.764637721870117,
      "grad_norm": 0.05697506856417742,
      "learning_rate": 3.146386503903278e-05,
      "loss": 0.828,
      "step": 214430
    },
    {
      "epoch": 3.7648132867501185,
      "grad_norm": 0.05680427472198276,
      "learning_rate": 3.145593347481875e-05,
      "loss": 0.8379,
      "step": 214440
    },
    {
      "epoch": 3.76498885163012,
      "grad_norm": 0.053239357163208025,
      "learning_rate": 3.144800279174135e-05,
      "loss": 0.846,
      "step": 214450
    },
    {
      "epoch": 3.7651644165101215,
      "grad_norm": 0.04103739888452774,
      "learning_rate": 3.144007298990112e-05,
      "loss": 0.8424,
      "step": 214460
    },
    {
      "epoch": 3.765339981390123,
      "grad_norm": 0.05558846496694252,
      "learning_rate": 3.143214406939845e-05,
      "loss": 0.8391,
      "step": 214470
    },
    {
      "epoch": 3.765515546270124,
      "grad_norm": 0.04613545897139286,
      "learning_rate": 3.142421603033389e-05,
      "loss": 0.8364,
      "step": 214480
    },
    {
      "epoch": 3.7656911111501254,
      "grad_norm": 0.055878749250400296,
      "learning_rate": 3.141628887280782e-05,
      "loss": 0.837,
      "step": 214490
    },
    {
      "epoch": 3.765866676030127,
      "grad_norm": 0.04819806765361641,
      "learning_rate": 3.1408362596920745e-05,
      "loss": 0.8335,
      "step": 214500
    },
    {
      "epoch": 3.7660422409101284,
      "grad_norm": 0.05313837444408743,
      "learning_rate": 3.1400437202773006e-05,
      "loss": 0.8361,
      "step": 214510
    },
    {
      "epoch": 3.76621780579013,
      "grad_norm": 0.051149374024312154,
      "learning_rate": 3.139251269046508e-05,
      "loss": 0.8376,
      "step": 214520
    },
    {
      "epoch": 3.766393370670131,
      "grad_norm": 0.049805921886823756,
      "learning_rate": 3.138458906009739e-05,
      "loss": 0.8359,
      "step": 214530
    },
    {
      "epoch": 3.7665689355501324,
      "grad_norm": 0.05183940309262988,
      "learning_rate": 3.137666631177024e-05,
      "loss": 0.8405,
      "step": 214540
    },
    {
      "epoch": 3.766744500430134,
      "grad_norm": 0.05421425552293453,
      "learning_rate": 3.1368744445584125e-05,
      "loss": 0.8372,
      "step": 214550
    },
    {
      "epoch": 3.7669200653101353,
      "grad_norm": 0.06399373108711016,
      "learning_rate": 3.1360823461639335e-05,
      "loss": 0.8331,
      "step": 214560
    },
    {
      "epoch": 3.767095630190137,
      "grad_norm": 0.050621161299657415,
      "learning_rate": 3.135290336003628e-05,
      "loss": 0.8462,
      "step": 214570
    },
    {
      "epoch": 3.7672711950701383,
      "grad_norm": 0.05075418015990819,
      "learning_rate": 3.134498414087526e-05,
      "loss": 0.8369,
      "step": 214580
    },
    {
      "epoch": 3.76744675995014,
      "grad_norm": 0.05422739600880449,
      "learning_rate": 3.1337065804256644e-05,
      "loss": 0.8422,
      "step": 214590
    },
    {
      "epoch": 3.767622324830141,
      "grad_norm": 0.05295289122255324,
      "learning_rate": 3.132914835028079e-05,
      "loss": 0.8397,
      "step": 214600
    },
    {
      "epoch": 3.7677978897101423,
      "grad_norm": 0.05526757842744108,
      "learning_rate": 3.132123177904794e-05,
      "loss": 0.8352,
      "step": 214610
    },
    {
      "epoch": 3.7679734545901438,
      "grad_norm": 0.06328790620015426,
      "learning_rate": 3.1313316090658446e-05,
      "loss": 0.8373,
      "step": 214620
    },
    {
      "epoch": 3.7681490194701452,
      "grad_norm": 0.047944903709310815,
      "learning_rate": 3.130540128521261e-05,
      "loss": 0.8328,
      "step": 214630
    },
    {
      "epoch": 3.7683245843501467,
      "grad_norm": 0.04863315004914211,
      "learning_rate": 3.129748736281064e-05,
      "loss": 0.8406,
      "step": 214640
    },
    {
      "epoch": 3.7685001492301478,
      "grad_norm": 0.05292718383359451,
      "learning_rate": 3.128957432355293e-05,
      "loss": 0.8349,
      "step": 214650
    },
    {
      "epoch": 3.7686757141101497,
      "grad_norm": 0.06542805809791198,
      "learning_rate": 3.128166216753964e-05,
      "loss": 0.8359,
      "step": 214660
    },
    {
      "epoch": 3.7688512789901507,
      "grad_norm": 0.05390036070654228,
      "learning_rate": 3.1273750894871086e-05,
      "loss": 0.8382,
      "step": 214670
    },
    {
      "epoch": 3.769026843870152,
      "grad_norm": 0.05916077538144203,
      "learning_rate": 3.1265840505647435e-05,
      "loss": 0.8412,
      "step": 214680
    },
    {
      "epoch": 3.7692024087501537,
      "grad_norm": 0.04646706497880775,
      "learning_rate": 3.1257930999968954e-05,
      "loss": 0.839,
      "step": 214690
    },
    {
      "epoch": 3.769377973630155,
      "grad_norm": 0.0676954843135536,
      "learning_rate": 3.1250022377935877e-05,
      "loss": 0.851,
      "step": 214700
    },
    {
      "epoch": 3.7695535385101566,
      "grad_norm": 0.057352281986241906,
      "learning_rate": 3.124211463964833e-05,
      "loss": 0.8355,
      "step": 214710
    },
    {
      "epoch": 3.7697291033901577,
      "grad_norm": 0.06521074955126209,
      "learning_rate": 3.1234207785206616e-05,
      "loss": 0.8409,
      "step": 214720
    },
    {
      "epoch": 3.769904668270159,
      "grad_norm": 0.07554077281564216,
      "learning_rate": 3.122630181471083e-05,
      "loss": 0.8385,
      "step": 214730
    },
    {
      "epoch": 3.7700802331501606,
      "grad_norm": 0.05582010746249529,
      "learning_rate": 3.121839672826121e-05,
      "loss": 0.8474,
      "step": 214740
    },
    {
      "epoch": 3.770255798030162,
      "grad_norm": 0.056577948435699035,
      "learning_rate": 3.121049252595784e-05,
      "loss": 0.8403,
      "step": 214750
    },
    {
      "epoch": 3.7704313629101636,
      "grad_norm": 0.05032330714640675,
      "learning_rate": 3.12025892079009e-05,
      "loss": 0.8347,
      "step": 214760
    },
    {
      "epoch": 3.7706069277901646,
      "grad_norm": 0.046431535376464786,
      "learning_rate": 3.119468677419057e-05,
      "loss": 0.8507,
      "step": 214770
    },
    {
      "epoch": 3.7707824926701665,
      "grad_norm": 0.05384863718373892,
      "learning_rate": 3.1186785224926906e-05,
      "loss": 0.8422,
      "step": 214780
    },
    {
      "epoch": 3.7709580575501676,
      "grad_norm": 0.05168688932333141,
      "learning_rate": 3.1178884560210044e-05,
      "loss": 0.8386,
      "step": 214790
    },
    {
      "epoch": 3.771133622430169,
      "grad_norm": 0.04699318020312147,
      "learning_rate": 3.117098478014014e-05,
      "loss": 0.8457,
      "step": 214800
    },
    {
      "epoch": 3.7713091873101705,
      "grad_norm": 0.06316851576965823,
      "learning_rate": 3.116308588481717e-05,
      "loss": 0.8395,
      "step": 214810
    },
    {
      "epoch": 3.771484752190172,
      "grad_norm": 0.06176929971716442,
      "learning_rate": 3.1155187874341344e-05,
      "loss": 0.8321,
      "step": 214820
    },
    {
      "epoch": 3.7716603170701735,
      "grad_norm": 0.052677107687728625,
      "learning_rate": 3.114729074881265e-05,
      "loss": 0.8487,
      "step": 214830
    },
    {
      "epoch": 3.7718358819501745,
      "grad_norm": 0.05965270599862889,
      "learning_rate": 3.11393945083312e-05,
      "loss": 0.8376,
      "step": 214840
    },
    {
      "epoch": 3.772011446830176,
      "grad_norm": 0.06299103570941512,
      "learning_rate": 3.113149915299698e-05,
      "loss": 0.8433,
      "step": 214850
    },
    {
      "epoch": 3.7721870117101775,
      "grad_norm": 0.050630840790215,
      "learning_rate": 3.112360468291006e-05,
      "loss": 0.8432,
      "step": 214860
    },
    {
      "epoch": 3.772362576590179,
      "grad_norm": 0.04546468742196688,
      "learning_rate": 3.111571109817048e-05,
      "loss": 0.838,
      "step": 214870
    },
    {
      "epoch": 3.7725381414701804,
      "grad_norm": 0.05498642226426605,
      "learning_rate": 3.110781839887817e-05,
      "loss": 0.8346,
      "step": 214880
    },
    {
      "epoch": 3.7727137063501814,
      "grad_norm": 0.05259495462475592,
      "learning_rate": 3.109992658513327e-05,
      "loss": 0.8361,
      "step": 214890
    },
    {
      "epoch": 3.7728892712301834,
      "grad_norm": 0.06251902781500743,
      "learning_rate": 3.109203565703566e-05,
      "loss": 0.8402,
      "step": 214900
    },
    {
      "epoch": 3.7730648361101844,
      "grad_norm": 0.06534369251106235,
      "learning_rate": 3.108414561468539e-05,
      "loss": 0.8451,
      "step": 214910
    },
    {
      "epoch": 3.773240400990186,
      "grad_norm": 0.04734509617916443,
      "learning_rate": 3.107625645818236e-05,
      "loss": 0.84,
      "step": 214920
    },
    {
      "epoch": 3.7734159658701873,
      "grad_norm": 0.0473443245470149,
      "learning_rate": 3.106836818762657e-05,
      "loss": 0.8462,
      "step": 214930
    },
    {
      "epoch": 3.773591530750189,
      "grad_norm": 0.07728783719469803,
      "learning_rate": 3.106048080311797e-05,
      "loss": 0.8368,
      "step": 214940
    },
    {
      "epoch": 3.7737670956301903,
      "grad_norm": 0.0622053457924769,
      "learning_rate": 3.105259430475647e-05,
      "loss": 0.8335,
      "step": 214950
    },
    {
      "epoch": 3.7739426605101913,
      "grad_norm": 0.045920607428883424,
      "learning_rate": 3.1044708692641995e-05,
      "loss": 0.843,
      "step": 214960
    },
    {
      "epoch": 3.774118225390193,
      "grad_norm": 0.06016679314445572,
      "learning_rate": 3.10368239668745e-05,
      "loss": 0.8423,
      "step": 214970
    },
    {
      "epoch": 3.7742937902701943,
      "grad_norm": 0.059098989339960696,
      "learning_rate": 3.102894012755379e-05,
      "loss": 0.836,
      "step": 214980
    },
    {
      "epoch": 3.7744693551501958,
      "grad_norm": 0.06331632995032194,
      "learning_rate": 3.1021057174779884e-05,
      "loss": 0.8355,
      "step": 214990
    },
    {
      "epoch": 3.7746449200301972,
      "grad_norm": 0.0455579665691247,
      "learning_rate": 3.101317510865257e-05,
      "loss": 0.8371,
      "step": 215000
    },
    {
      "epoch": 3.7748204849101987,
      "grad_norm": 0.056270028737222276,
      "learning_rate": 3.1005293929271756e-05,
      "loss": 0.8385,
      "step": 215010
    },
    {
      "epoch": 3.7749960497902,
      "grad_norm": 0.06619506342523751,
      "learning_rate": 3.099741363673726e-05,
      "loss": 0.8381,
      "step": 215020
    },
    {
      "epoch": 3.7751716146702012,
      "grad_norm": 0.052349983403559786,
      "learning_rate": 3.0989534231148954e-05,
      "loss": 0.8352,
      "step": 215030
    },
    {
      "epoch": 3.7753471795502027,
      "grad_norm": 0.061581640658897395,
      "learning_rate": 3.098165571260669e-05,
      "loss": 0.8386,
      "step": 215040
    },
    {
      "epoch": 3.775522744430204,
      "grad_norm": 0.05873422074326308,
      "learning_rate": 3.097377808121022e-05,
      "loss": 0.8406,
      "step": 215050
    },
    {
      "epoch": 3.7756983093102057,
      "grad_norm": 0.06118814351093374,
      "learning_rate": 3.0965901337059445e-05,
      "loss": 0.8412,
      "step": 215060
    },
    {
      "epoch": 3.775873874190207,
      "grad_norm": 0.05597853768959964,
      "learning_rate": 3.0958025480254094e-05,
      "loss": 0.8391,
      "step": 215070
    },
    {
      "epoch": 3.776049439070208,
      "grad_norm": 0.05044440052805297,
      "learning_rate": 3.095015051089402e-05,
      "loss": 0.8406,
      "step": 215080
    },
    {
      "epoch": 3.7762250039502097,
      "grad_norm": 0.09688000308909343,
      "learning_rate": 3.094227642907893e-05,
      "loss": 0.84,
      "step": 215090
    },
    {
      "epoch": 3.776400568830211,
      "grad_norm": 0.05919136246187355,
      "learning_rate": 3.0934403234908624e-05,
      "loss": 0.8397,
      "step": 215100
    },
    {
      "epoch": 3.7765761337102126,
      "grad_norm": 0.050213728958515946,
      "learning_rate": 3.0926530928482895e-05,
      "loss": 0.8362,
      "step": 215110
    },
    {
      "epoch": 3.776751698590214,
      "grad_norm": 0.05634246820129658,
      "learning_rate": 3.091865950990141e-05,
      "loss": 0.8428,
      "step": 215120
    },
    {
      "epoch": 3.7769272634702156,
      "grad_norm": 0.06825150029685785,
      "learning_rate": 3.091078897926394e-05,
      "loss": 0.8353,
      "step": 215130
    },
    {
      "epoch": 3.777102828350217,
      "grad_norm": 0.04967298662440711,
      "learning_rate": 3.0902919336670244e-05,
      "loss": 0.8373,
      "step": 215140
    },
    {
      "epoch": 3.777278393230218,
      "grad_norm": 0.045602355802380576,
      "learning_rate": 3.0895050582219926e-05,
      "loss": 0.8457,
      "step": 215150
    },
    {
      "epoch": 3.7774539581102196,
      "grad_norm": 0.04789382313825578,
      "learning_rate": 3.088718271601282e-05,
      "loss": 0.8415,
      "step": 215160
    },
    {
      "epoch": 3.777629522990221,
      "grad_norm": 0.041357313624311054,
      "learning_rate": 3.087931573814851e-05,
      "loss": 0.8435,
      "step": 215170
    },
    {
      "epoch": 3.7778050878702225,
      "grad_norm": 0.062122638352257675,
      "learning_rate": 3.087144964872674e-05,
      "loss": 0.8379,
      "step": 215180
    },
    {
      "epoch": 3.777980652750224,
      "grad_norm": 0.05875042793546874,
      "learning_rate": 3.086358444784712e-05,
      "loss": 0.8294,
      "step": 215190
    },
    {
      "epoch": 3.778156217630225,
      "grad_norm": 0.05914537848232735,
      "learning_rate": 3.0855720135609316e-05,
      "loss": 0.837,
      "step": 215200
    },
    {
      "epoch": 3.7783317825102265,
      "grad_norm": 0.06904318211613426,
      "learning_rate": 3.0847856712113006e-05,
      "loss": 0.8213,
      "step": 215210
    },
    {
      "epoch": 3.778507347390228,
      "grad_norm": 0.06829360851028685,
      "learning_rate": 3.0839994177457746e-05,
      "loss": 0.8406,
      "step": 215220
    },
    {
      "epoch": 3.7786829122702295,
      "grad_norm": 0.05679604729608953,
      "learning_rate": 3.083213253174327e-05,
      "loss": 0.8382,
      "step": 215230
    },
    {
      "epoch": 3.778858477150231,
      "grad_norm": 0.047989893449967846,
      "learning_rate": 3.082427177506909e-05,
      "loss": 0.842,
      "step": 215240
    },
    {
      "epoch": 3.7790340420302324,
      "grad_norm": 0.051064314686044304,
      "learning_rate": 3.081641190753485e-05,
      "loss": 0.836,
      "step": 215250
    },
    {
      "epoch": 3.779209606910234,
      "grad_norm": 0.05735085245991922,
      "learning_rate": 3.080855292924011e-05,
      "loss": 0.85,
      "step": 215260
    },
    {
      "epoch": 3.779385171790235,
      "grad_norm": 0.05731658130751086,
      "learning_rate": 3.0800694840284454e-05,
      "loss": 0.8302,
      "step": 215270
    },
    {
      "epoch": 3.7795607366702364,
      "grad_norm": 0.049953436565690255,
      "learning_rate": 3.079283764076748e-05,
      "loss": 0.8446,
      "step": 215280
    },
    {
      "epoch": 3.779736301550238,
      "grad_norm": 0.055522066832945555,
      "learning_rate": 3.0784981330788665e-05,
      "loss": 0.8422,
      "step": 215290
    },
    {
      "epoch": 3.7799118664302394,
      "grad_norm": 0.07063263770227932,
      "learning_rate": 3.07771259104476e-05,
      "loss": 0.8423,
      "step": 215300
    },
    {
      "epoch": 3.780087431310241,
      "grad_norm": 0.06584139501401648,
      "learning_rate": 3.076927137984384e-05,
      "loss": 0.8435,
      "step": 215310
    },
    {
      "epoch": 3.780262996190242,
      "grad_norm": 0.06019011430017405,
      "learning_rate": 3.076141773907681e-05,
      "loss": 0.8429,
      "step": 215320
    },
    {
      "epoch": 3.780438561070244,
      "grad_norm": 0.05542399200891208,
      "learning_rate": 3.075356498824614e-05,
      "loss": 0.8382,
      "step": 215330
    },
    {
      "epoch": 3.780614125950245,
      "grad_norm": 0.05859702644091471,
      "learning_rate": 3.074571312745124e-05,
      "loss": 0.8417,
      "step": 215340
    },
    {
      "epoch": 3.7807896908302463,
      "grad_norm": 0.05185774892913654,
      "learning_rate": 3.073786215679164e-05,
      "loss": 0.8412,
      "step": 215350
    },
    {
      "epoch": 3.7809652557102478,
      "grad_norm": 0.052722795215097554,
      "learning_rate": 3.073001207636677e-05,
      "loss": 0.8347,
      "step": 215360
    },
    {
      "epoch": 3.7811408205902493,
      "grad_norm": 0.05609874504596849,
      "learning_rate": 3.0722162886276104e-05,
      "loss": 0.832,
      "step": 215370
    },
    {
      "epoch": 3.7813163854702507,
      "grad_norm": 0.04909030779154749,
      "learning_rate": 3.071431458661914e-05,
      "loss": 0.8354,
      "step": 215380
    },
    {
      "epoch": 3.7814919503502518,
      "grad_norm": 0.06083218635190068,
      "learning_rate": 3.070646717749521e-05,
      "loss": 0.8407,
      "step": 215390
    },
    {
      "epoch": 3.7816675152302532,
      "grad_norm": 0.04981362465027949,
      "learning_rate": 3.0698620659003884e-05,
      "loss": 0.8371,
      "step": 215400
    },
    {
      "epoch": 3.7818430801102547,
      "grad_norm": 0.06316740045395042,
      "learning_rate": 3.069077503124447e-05,
      "loss": 0.8388,
      "step": 215410
    },
    {
      "epoch": 3.782018644990256,
      "grad_norm": 0.06442246711654936,
      "learning_rate": 3.0682930294316444e-05,
      "loss": 0.8433,
      "step": 215420
    },
    {
      "epoch": 3.7821942098702577,
      "grad_norm": 0.05881522721056101,
      "learning_rate": 3.0675086448319145e-05,
      "loss": 0.8309,
      "step": 215430
    },
    {
      "epoch": 3.7823697747502587,
      "grad_norm": 0.05754131021574009,
      "learning_rate": 3.066724349335197e-05,
      "loss": 0.8382,
      "step": 215440
    },
    {
      "epoch": 3.7825453396302606,
      "grad_norm": 0.04494329725005815,
      "learning_rate": 3.065940142951432e-05,
      "loss": 0.8323,
      "step": 215450
    },
    {
      "epoch": 3.7827209045102617,
      "grad_norm": 0.05403052969315236,
      "learning_rate": 3.065156025690552e-05,
      "loss": 0.8388,
      "step": 215460
    },
    {
      "epoch": 3.782896469390263,
      "grad_norm": 0.04719214404733549,
      "learning_rate": 3.0643719975624934e-05,
      "loss": 0.8361,
      "step": 215470
    },
    {
      "epoch": 3.7830720342702646,
      "grad_norm": 0.042984672109623534,
      "learning_rate": 3.0635880585771924e-05,
      "loss": 0.8392,
      "step": 215480
    },
    {
      "epoch": 3.783247599150266,
      "grad_norm": 0.06414685739086344,
      "learning_rate": 3.062804208744573e-05,
      "loss": 0.8381,
      "step": 215490
    },
    {
      "epoch": 3.7834231640302676,
      "grad_norm": 0.07203712267869565,
      "learning_rate": 3.062020448074579e-05,
      "loss": 0.8379,
      "step": 215500
    },
    {
      "epoch": 3.7835987289102686,
      "grad_norm": 0.045351065210622,
      "learning_rate": 3.0612367765771325e-05,
      "loss": 0.8382,
      "step": 215510
    },
    {
      "epoch": 3.78377429379027,
      "grad_norm": 0.044828275661691196,
      "learning_rate": 3.060453194262168e-05,
      "loss": 0.8452,
      "step": 215520
    },
    {
      "epoch": 3.7839498586702716,
      "grad_norm": 0.058324104224894084,
      "learning_rate": 3.059669701139607e-05,
      "loss": 0.835,
      "step": 215530
    },
    {
      "epoch": 3.784125423550273,
      "grad_norm": 0.04591848667297986,
      "learning_rate": 3.058886297219379e-05,
      "loss": 0.8415,
      "step": 215540
    },
    {
      "epoch": 3.7843009884302745,
      "grad_norm": 0.051982067761349504,
      "learning_rate": 3.058102982511416e-05,
      "loss": 0.8337,
      "step": 215550
    },
    {
      "epoch": 3.7844765533102755,
      "grad_norm": 0.07623465098646076,
      "learning_rate": 3.057319757025631e-05,
      "loss": 0.8356,
      "step": 215560
    },
    {
      "epoch": 3.7846521181902775,
      "grad_norm": 0.05385933264806462,
      "learning_rate": 3.056536620771961e-05,
      "loss": 0.8405,
      "step": 215570
    },
    {
      "epoch": 3.7848276830702785,
      "grad_norm": 0.054243158673813685,
      "learning_rate": 3.0557535737603184e-05,
      "loss": 0.8442,
      "step": 215580
    },
    {
      "epoch": 3.78500324795028,
      "grad_norm": 0.06176167312794796,
      "learning_rate": 3.0549706160006306e-05,
      "loss": 0.833,
      "step": 215590
    },
    {
      "epoch": 3.7851788128302815,
      "grad_norm": 0.05363638634939989,
      "learning_rate": 3.054187747502813e-05,
      "loss": 0.834,
      "step": 215600
    },
    {
      "epoch": 3.785354377710283,
      "grad_norm": 0.0513489153381955,
      "learning_rate": 3.0534049682767865e-05,
      "loss": 0.8378,
      "step": 215610
    },
    {
      "epoch": 3.7855299425902844,
      "grad_norm": 0.05073797200069079,
      "learning_rate": 3.0526222783324726e-05,
      "loss": 0.8392,
      "step": 215620
    },
    {
      "epoch": 3.7857055074702854,
      "grad_norm": 0.05720564721924178,
      "learning_rate": 3.0518396776797834e-05,
      "loss": 0.849,
      "step": 215630
    },
    {
      "epoch": 3.785881072350287,
      "grad_norm": 0.05785800554680885,
      "learning_rate": 3.0510571663286337e-05,
      "loss": 0.8381,
      "step": 215640
    },
    {
      "epoch": 3.7860566372302884,
      "grad_norm": 0.05768608986415673,
      "learning_rate": 3.0502747442889456e-05,
      "loss": 0.8428,
      "step": 215650
    },
    {
      "epoch": 3.78623220211029,
      "grad_norm": 0.044682022768163,
      "learning_rate": 3.049492411570621e-05,
      "loss": 0.8453,
      "step": 215660
    },
    {
      "epoch": 3.7864077669902914,
      "grad_norm": 0.049629449531778835,
      "learning_rate": 3.0487101681835838e-05,
      "loss": 0.8419,
      "step": 215670
    },
    {
      "epoch": 3.7865833318702924,
      "grad_norm": 0.04659385074331223,
      "learning_rate": 3.0479280141377385e-05,
      "loss": 0.8401,
      "step": 215680
    },
    {
      "epoch": 3.7867588967502943,
      "grad_norm": 0.047276483028826685,
      "learning_rate": 3.047145949443e-05,
      "loss": 0.8338,
      "step": 215690
    },
    {
      "epoch": 3.7869344616302953,
      "grad_norm": 0.04481572044479173,
      "learning_rate": 3.046363974109269e-05,
      "loss": 0.8452,
      "step": 215700
    },
    {
      "epoch": 3.787110026510297,
      "grad_norm": 0.0504226049332478,
      "learning_rate": 3.04558208814646e-05,
      "loss": 0.8411,
      "step": 215710
    },
    {
      "epoch": 3.7872855913902983,
      "grad_norm": 0.046138093716134906,
      "learning_rate": 3.0448002915644814e-05,
      "loss": 0.8427,
      "step": 215720
    },
    {
      "epoch": 3.7874611562703,
      "grad_norm": 0.040137714957505934,
      "learning_rate": 3.0440185843732273e-05,
      "loss": 0.847,
      "step": 215730
    },
    {
      "epoch": 3.7876367211503013,
      "grad_norm": 0.05880731775692375,
      "learning_rate": 3.0432369665826178e-05,
      "loss": 0.8332,
      "step": 215740
    },
    {
      "epoch": 3.7878122860303023,
      "grad_norm": 0.05230460677304659,
      "learning_rate": 3.0424554382025443e-05,
      "loss": 0.8405,
      "step": 215750
    },
    {
      "epoch": 3.7879878509103038,
      "grad_norm": 0.0453767812425392,
      "learning_rate": 3.0416739992429168e-05,
      "loss": 0.8428,
      "step": 215760
    },
    {
      "epoch": 3.7881634157903052,
      "grad_norm": 0.048097577578783766,
      "learning_rate": 3.0408926497136292e-05,
      "loss": 0.8384,
      "step": 215770
    },
    {
      "epoch": 3.7883389806703067,
      "grad_norm": 0.05159412773793071,
      "learning_rate": 3.0401113896245838e-05,
      "loss": 0.8383,
      "step": 215780
    },
    {
      "epoch": 3.788514545550308,
      "grad_norm": 0.06519927067735892,
      "learning_rate": 3.0393302189856843e-05,
      "loss": 0.8427,
      "step": 215790
    },
    {
      "epoch": 3.7886901104303097,
      "grad_norm": 0.04688890583436825,
      "learning_rate": 3.0385491378068197e-05,
      "loss": 0.8388,
      "step": 215800
    },
    {
      "epoch": 3.788865675310311,
      "grad_norm": 0.04475959146942576,
      "learning_rate": 3.037768146097891e-05,
      "loss": 0.8406,
      "step": 215810
    },
    {
      "epoch": 3.789041240190312,
      "grad_norm": 0.05714126259530708,
      "learning_rate": 3.0369872438687974e-05,
      "loss": 0.8336,
      "step": 215820
    },
    {
      "epoch": 3.7892168050703137,
      "grad_norm": 0.053006246549597784,
      "learning_rate": 3.0362064311294224e-05,
      "loss": 0.8323,
      "step": 215830
    },
    {
      "epoch": 3.789392369950315,
      "grad_norm": 0.060909875352333456,
      "learning_rate": 3.035425707889672e-05,
      "loss": 0.8388,
      "step": 215840
    },
    {
      "epoch": 3.7895679348303166,
      "grad_norm": 0.0686159035642891,
      "learning_rate": 3.0346450741594284e-05,
      "loss": 0.8372,
      "step": 215850
    },
    {
      "epoch": 3.789743499710318,
      "grad_norm": 0.06706599858714622,
      "learning_rate": 3.0338645299485898e-05,
      "loss": 0.8324,
      "step": 215860
    },
    {
      "epoch": 3.789919064590319,
      "grad_norm": 0.05085721336997347,
      "learning_rate": 3.033084075267038e-05,
      "loss": 0.8424,
      "step": 215870
    },
    {
      "epoch": 3.7900946294703206,
      "grad_norm": 0.056098779542174446,
      "learning_rate": 3.0323037101246653e-05,
      "loss": 0.8332,
      "step": 215880
    },
    {
      "epoch": 3.790270194350322,
      "grad_norm": 0.04908236633510771,
      "learning_rate": 3.031523434531362e-05,
      "loss": 0.8387,
      "step": 215890
    },
    {
      "epoch": 3.7904457592303236,
      "grad_norm": 0.05344264463249456,
      "learning_rate": 3.0307432484970054e-05,
      "loss": 0.841,
      "step": 215900
    },
    {
      "epoch": 3.790621324110325,
      "grad_norm": 0.04800740411056466,
      "learning_rate": 3.0299631520314926e-05,
      "loss": 0.8375,
      "step": 215910
    },
    {
      "epoch": 3.7907968889903265,
      "grad_norm": 0.0517770488714822,
      "learning_rate": 3.0291831451446994e-05,
      "loss": 0.8495,
      "step": 215920
    },
    {
      "epoch": 3.790972453870328,
      "grad_norm": 0.05435176594216623,
      "learning_rate": 3.0284032278465135e-05,
      "loss": 0.8422,
      "step": 215930
    },
    {
      "epoch": 3.791148018750329,
      "grad_norm": 0.05652932896738733,
      "learning_rate": 3.027623400146811e-05,
      "loss": 0.8419,
      "step": 215940
    },
    {
      "epoch": 3.7913235836303305,
      "grad_norm": 0.042922287660263125,
      "learning_rate": 3.0268436620554753e-05,
      "loss": 0.8431,
      "step": 215950
    },
    {
      "epoch": 3.791499148510332,
      "grad_norm": 0.08675728623317822,
      "learning_rate": 3.0260640135823887e-05,
      "loss": 0.8399,
      "step": 215960
    },
    {
      "epoch": 3.7916747133903335,
      "grad_norm": 0.04695875546909925,
      "learning_rate": 3.0252844547374242e-05,
      "loss": 0.8399,
      "step": 215970
    },
    {
      "epoch": 3.791850278270335,
      "grad_norm": 0.04419327745122121,
      "learning_rate": 3.0245049855304607e-05,
      "loss": 0.8406,
      "step": 215980
    },
    {
      "epoch": 3.792025843150336,
      "grad_norm": 0.05171283888085386,
      "learning_rate": 3.02372560597138e-05,
      "loss": 0.8329,
      "step": 215990
    },
    {
      "epoch": 3.7922014080303375,
      "grad_norm": 0.045043946781931805,
      "learning_rate": 3.0229463160700444e-05,
      "loss": 0.8412,
      "step": 216000
    },
    {
      "epoch": 3.792376972910339,
      "grad_norm": 0.04745036642823334,
      "learning_rate": 3.0221671158363414e-05,
      "loss": 0.8328,
      "step": 216010
    },
    {
      "epoch": 3.7925525377903404,
      "grad_norm": 0.04430336122347092,
      "learning_rate": 3.0213880052801348e-05,
      "loss": 0.834,
      "step": 216020
    },
    {
      "epoch": 3.792728102670342,
      "grad_norm": 0.054788536892417916,
      "learning_rate": 3.0206089844113017e-05,
      "loss": 0.8424,
      "step": 216030
    },
    {
      "epoch": 3.7929036675503434,
      "grad_norm": 0.04705830014176271,
      "learning_rate": 3.0198300532397062e-05,
      "loss": 0.8355,
      "step": 216040
    },
    {
      "epoch": 3.793079232430345,
      "grad_norm": 0.05165207884799079,
      "learning_rate": 3.019051211775221e-05,
      "loss": 0.8458,
      "step": 216050
    },
    {
      "epoch": 3.793254797310346,
      "grad_norm": 0.06695726288812064,
      "learning_rate": 3.018272460027718e-05,
      "loss": 0.8407,
      "step": 216060
    },
    {
      "epoch": 3.7934303621903473,
      "grad_norm": 0.06348701039852789,
      "learning_rate": 3.0174937980070533e-05,
      "loss": 0.8391,
      "step": 216070
    },
    {
      "epoch": 3.793605927070349,
      "grad_norm": 0.05299575240985735,
      "learning_rate": 3.016715225723105e-05,
      "loss": 0.8408,
      "step": 216080
    },
    {
      "epoch": 3.7937814919503503,
      "grad_norm": 0.06974765748102253,
      "learning_rate": 3.0159367431857293e-05,
      "loss": 0.8342,
      "step": 216090
    },
    {
      "epoch": 3.793957056830352,
      "grad_norm": 0.06505903751834409,
      "learning_rate": 3.015158350404797e-05,
      "loss": 0.8414,
      "step": 216100
    },
    {
      "epoch": 3.794132621710353,
      "grad_norm": 0.05818494687857544,
      "learning_rate": 3.0143800473901623e-05,
      "loss": 0.8346,
      "step": 216110
    },
    {
      "epoch": 3.7943081865903547,
      "grad_norm": 0.04621913954323976,
      "learning_rate": 3.0136018341516908e-05,
      "loss": 0.8348,
      "step": 216120
    },
    {
      "epoch": 3.7944837514703558,
      "grad_norm": 0.04986290948578882,
      "learning_rate": 3.012823710699245e-05,
      "loss": 0.8374,
      "step": 216130
    },
    {
      "epoch": 3.7946593163503572,
      "grad_norm": 0.05761866035611304,
      "learning_rate": 3.012045677042678e-05,
      "loss": 0.8387,
      "step": 216140
    },
    {
      "epoch": 3.7948348812303587,
      "grad_norm": 0.060544201284411125,
      "learning_rate": 3.0112677331918504e-05,
      "loss": 0.8324,
      "step": 216150
    },
    {
      "epoch": 3.79501044611036,
      "grad_norm": 0.06106005084873182,
      "learning_rate": 3.0104898791566226e-05,
      "loss": 0.8403,
      "step": 216160
    },
    {
      "epoch": 3.7951860109903617,
      "grad_norm": 0.04713909400776089,
      "learning_rate": 3.00971211494684e-05,
      "loss": 0.8436,
      "step": 216170
    },
    {
      "epoch": 3.7953615758703627,
      "grad_norm": 0.05301026233147863,
      "learning_rate": 3.0089344405723703e-05,
      "loss": 0.842,
      "step": 216180
    },
    {
      "epoch": 3.795537140750364,
      "grad_norm": 0.045722882112358136,
      "learning_rate": 3.0081568560430574e-05,
      "loss": 0.8503,
      "step": 216190
    },
    {
      "epoch": 3.7957127056303657,
      "grad_norm": 0.04422667076540221,
      "learning_rate": 3.0073793613687585e-05,
      "loss": 0.8399,
      "step": 216200
    },
    {
      "epoch": 3.795888270510367,
      "grad_norm": 0.05702470594432797,
      "learning_rate": 3.00660195655932e-05,
      "loss": 0.8467,
      "step": 216210
    },
    {
      "epoch": 3.7960638353903686,
      "grad_norm": 0.0809340764196382,
      "learning_rate": 3.0058246416245945e-05,
      "loss": 0.8403,
      "step": 216220
    },
    {
      "epoch": 3.7962394002703697,
      "grad_norm": 0.05472050769910334,
      "learning_rate": 3.005047416574433e-05,
      "loss": 0.8388,
      "step": 216230
    },
    {
      "epoch": 3.7964149651503716,
      "grad_norm": 0.06071982031961426,
      "learning_rate": 3.0042702814186753e-05,
      "loss": 0.8442,
      "step": 216240
    },
    {
      "epoch": 3.7965905300303726,
      "grad_norm": 0.05327637295423789,
      "learning_rate": 3.0034932361671776e-05,
      "loss": 0.8487,
      "step": 216250
    },
    {
      "epoch": 3.796766094910374,
      "grad_norm": 0.07394347323106831,
      "learning_rate": 3.0027162808297777e-05,
      "loss": 0.8349,
      "step": 216260
    },
    {
      "epoch": 3.7969416597903756,
      "grad_norm": 0.05488228934149554,
      "learning_rate": 3.0019394154163265e-05,
      "loss": 0.8463,
      "step": 216270
    },
    {
      "epoch": 3.797117224670377,
      "grad_norm": 0.060324492110900796,
      "learning_rate": 3.0011626399366606e-05,
      "loss": 0.8398,
      "step": 216280
    },
    {
      "epoch": 3.7972927895503785,
      "grad_norm": 0.04740797918408799,
      "learning_rate": 3.0003859544006236e-05,
      "loss": 0.8354,
      "step": 216290
    },
    {
      "epoch": 3.7974683544303796,
      "grad_norm": 0.09224114113472429,
      "learning_rate": 2.9996093588180607e-05,
      "loss": 0.84,
      "step": 216300
    },
    {
      "epoch": 3.797643919310381,
      "grad_norm": 0.04333030648682951,
      "learning_rate": 2.9988328531988047e-05,
      "loss": 0.8396,
      "step": 216310
    },
    {
      "epoch": 3.7978194841903825,
      "grad_norm": 0.049587864642088135,
      "learning_rate": 2.9980564375526972e-05,
      "loss": 0.8409,
      "step": 216320
    },
    {
      "epoch": 3.797995049070384,
      "grad_norm": 0.04512311919726466,
      "learning_rate": 2.9972801118895794e-05,
      "loss": 0.8345,
      "step": 216330
    },
    {
      "epoch": 3.7981706139503855,
      "grad_norm": 0.05698511157083809,
      "learning_rate": 2.9965038762192764e-05,
      "loss": 0.8383,
      "step": 216340
    },
    {
      "epoch": 3.7983461788303865,
      "grad_norm": 0.06816185259066948,
      "learning_rate": 2.995727730551638e-05,
      "loss": 0.8436,
      "step": 216350
    },
    {
      "epoch": 3.7985217437103884,
      "grad_norm": 0.043120561561021335,
      "learning_rate": 2.9949516748964878e-05,
      "loss": 0.8412,
      "step": 216360
    },
    {
      "epoch": 3.7986973085903895,
      "grad_norm": 0.047531372703378816,
      "learning_rate": 2.9941757092636647e-05,
      "loss": 0.8454,
      "step": 216370
    },
    {
      "epoch": 3.798872873470391,
      "grad_norm": 0.052057346933477695,
      "learning_rate": 2.993399833662993e-05,
      "loss": 0.8405,
      "step": 216380
    },
    {
      "epoch": 3.7990484383503924,
      "grad_norm": 0.06443667393672588,
      "learning_rate": 2.992624048104309e-05,
      "loss": 0.8445,
      "step": 216390
    },
    {
      "epoch": 3.799224003230394,
      "grad_norm": 0.07379486440828018,
      "learning_rate": 2.9918483525974418e-05,
      "loss": 0.8379,
      "step": 216400
    },
    {
      "epoch": 3.7993995681103954,
      "grad_norm": 0.044012235940425255,
      "learning_rate": 2.9910727471522126e-05,
      "loss": 0.8416,
      "step": 216410
    },
    {
      "epoch": 3.7995751329903964,
      "grad_norm": 0.06407130833236857,
      "learning_rate": 2.9902972317784612e-05,
      "loss": 0.8344,
      "step": 216420
    },
    {
      "epoch": 3.799750697870398,
      "grad_norm": 0.06017323556711089,
      "learning_rate": 2.9895218064860024e-05,
      "loss": 0.8348,
      "step": 216430
    },
    {
      "epoch": 3.7999262627503994,
      "grad_norm": 0.06522120674104633,
      "learning_rate": 2.9887464712846692e-05,
      "loss": 0.8403,
      "step": 216440
    },
    {
      "epoch": 3.800101827630401,
      "grad_norm": 0.07172474265932748,
      "learning_rate": 2.9879712261842767e-05,
      "loss": 0.8401,
      "step": 216450
    },
    {
      "epoch": 3.8002773925104023,
      "grad_norm": 0.06420950050621067,
      "learning_rate": 2.987196071194652e-05,
      "loss": 0.8429,
      "step": 216460
    },
    {
      "epoch": 3.800452957390404,
      "grad_norm": 0.04872366409263709,
      "learning_rate": 2.98642100632562e-05,
      "loss": 0.8461,
      "step": 216470
    },
    {
      "epoch": 3.8006285222704053,
      "grad_norm": 0.05573915750218048,
      "learning_rate": 2.985646031586993e-05,
      "loss": 0.8378,
      "step": 216480
    },
    {
      "epoch": 3.8008040871504063,
      "grad_norm": 0.046346114220855786,
      "learning_rate": 2.9848711469885955e-05,
      "loss": 0.8439,
      "step": 216490
    },
    {
      "epoch": 3.8009796520304078,
      "grad_norm": 0.06462167109912452,
      "learning_rate": 2.9840963525402465e-05,
      "loss": 0.8407,
      "step": 216500
    },
    {
      "epoch": 3.8011552169104093,
      "grad_norm": 0.05667411028739348,
      "learning_rate": 2.9833216482517544e-05,
      "loss": 0.8464,
      "step": 216510
    },
    {
      "epoch": 3.8013307817904107,
      "grad_norm": 0.04863056416384048,
      "learning_rate": 2.982547034132947e-05,
      "loss": 0.8349,
      "step": 216520
    },
    {
      "epoch": 3.801506346670412,
      "grad_norm": 0.045643384446940015,
      "learning_rate": 2.98177251019363e-05,
      "loss": 0.8351,
      "step": 216530
    },
    {
      "epoch": 3.8016819115504132,
      "grad_norm": 0.06114555025814227,
      "learning_rate": 2.980998076443623e-05,
      "loss": 0.8463,
      "step": 216540
    },
    {
      "epoch": 3.8018574764304147,
      "grad_norm": 0.05158082667860177,
      "learning_rate": 2.98022373289273e-05,
      "loss": 0.8451,
      "step": 216550
    },
    {
      "epoch": 3.802033041310416,
      "grad_norm": 0.06181672117244222,
      "learning_rate": 2.9794494795507683e-05,
      "loss": 0.829,
      "step": 216560
    },
    {
      "epoch": 3.8022086061904177,
      "grad_norm": 0.04756833104005501,
      "learning_rate": 2.978675316427549e-05,
      "loss": 0.8457,
      "step": 216570
    },
    {
      "epoch": 3.802384171070419,
      "grad_norm": 0.060691207112528435,
      "learning_rate": 2.9779012435328724e-05,
      "loss": 0.8394,
      "step": 216580
    },
    {
      "epoch": 3.8025597359504206,
      "grad_norm": 0.05342490287842647,
      "learning_rate": 2.977127260876558e-05,
      "loss": 0.837,
      "step": 216590
    },
    {
      "epoch": 3.802735300830422,
      "grad_norm": 0.0445617363323546,
      "learning_rate": 2.9763533684684024e-05,
      "loss": 0.8444,
      "step": 216600
    },
    {
      "epoch": 3.802910865710423,
      "grad_norm": 0.06231984547932772,
      "learning_rate": 2.975579566318219e-05,
      "loss": 0.8263,
      "step": 216610
    },
    {
      "epoch": 3.8030864305904246,
      "grad_norm": 0.059800258591130535,
      "learning_rate": 2.974805854435804e-05,
      "loss": 0.8341,
      "step": 216620
    },
    {
      "epoch": 3.803261995470426,
      "grad_norm": 0.046250147139957384,
      "learning_rate": 2.9740322328309646e-05,
      "loss": 0.8419,
      "step": 216630
    },
    {
      "epoch": 3.8034375603504276,
      "grad_norm": 0.06319462832953164,
      "learning_rate": 2.973258701513506e-05,
      "loss": 0.8437,
      "step": 216640
    },
    {
      "epoch": 3.803613125230429,
      "grad_norm": 0.07758933135715074,
      "learning_rate": 2.9724852604932218e-05,
      "loss": 0.8471,
      "step": 216650
    },
    {
      "epoch": 3.80378869011043,
      "grad_norm": 0.055051234989771054,
      "learning_rate": 2.9717119097799155e-05,
      "loss": 0.8449,
      "step": 216660
    },
    {
      "epoch": 3.8039642549904316,
      "grad_norm": 0.03867482231691754,
      "learning_rate": 2.9709386493833892e-05,
      "loss": 0.8386,
      "step": 216670
    },
    {
      "epoch": 3.804139819870433,
      "grad_norm": 0.060432878383078655,
      "learning_rate": 2.9701654793134294e-05,
      "loss": 0.8423,
      "step": 216680
    },
    {
      "epoch": 3.8043153847504345,
      "grad_norm": 0.10601325263165481,
      "learning_rate": 2.969392399579847e-05,
      "loss": 0.8446,
      "step": 216690
    },
    {
      "epoch": 3.804490949630436,
      "grad_norm": 0.05723182107102692,
      "learning_rate": 2.968619410192424e-05,
      "loss": 0.8459,
      "step": 216700
    },
    {
      "epoch": 3.8046665145104375,
      "grad_norm": 0.13962706150904752,
      "learning_rate": 2.9678465111609643e-05,
      "loss": 0.8417,
      "step": 216710
    },
    {
      "epoch": 3.804842079390439,
      "grad_norm": 0.05100983371127837,
      "learning_rate": 2.967073702495254e-05,
      "loss": 0.8372,
      "step": 216720
    },
    {
      "epoch": 3.80501764427044,
      "grad_norm": 0.0479867968628532,
      "learning_rate": 2.966300984205086e-05,
      "loss": 0.8464,
      "step": 216730
    },
    {
      "epoch": 3.8051932091504415,
      "grad_norm": 0.051614516248306604,
      "learning_rate": 2.9655283563002533e-05,
      "loss": 0.843,
      "step": 216740
    },
    {
      "epoch": 3.805368774030443,
      "grad_norm": 0.0527799857992353,
      "learning_rate": 2.9647558187905395e-05,
      "loss": 0.8434,
      "step": 216750
    },
    {
      "epoch": 3.8055443389104444,
      "grad_norm": 0.057872429703993186,
      "learning_rate": 2.9639833716857428e-05,
      "loss": 0.8342,
      "step": 216760
    },
    {
      "epoch": 3.805719903790446,
      "grad_norm": 0.05133491071281844,
      "learning_rate": 2.9632110149956403e-05,
      "loss": 0.8415,
      "step": 216770
    },
    {
      "epoch": 3.805895468670447,
      "grad_norm": 0.05359066689784784,
      "learning_rate": 2.9624387487300265e-05,
      "loss": 0.849,
      "step": 216780
    },
    {
      "epoch": 3.806071033550449,
      "grad_norm": 0.0712960261676301,
      "learning_rate": 2.961666572898677e-05,
      "loss": 0.8435,
      "step": 216790
    },
    {
      "epoch": 3.80624659843045,
      "grad_norm": 0.04885324988916581,
      "learning_rate": 2.9608944875113815e-05,
      "loss": 0.8468,
      "step": 216800
    },
    {
      "epoch": 3.8064221633104514,
      "grad_norm": 0.057888907847736704,
      "learning_rate": 2.9601224925779235e-05,
      "loss": 0.8365,
      "step": 216810
    },
    {
      "epoch": 3.806597728190453,
      "grad_norm": 0.047553237305033684,
      "learning_rate": 2.959350588108079e-05,
      "loss": 0.8342,
      "step": 216820
    },
    {
      "epoch": 3.8067732930704543,
      "grad_norm": 0.060380486743219386,
      "learning_rate": 2.958578774111631e-05,
      "loss": 0.8295,
      "step": 216830
    },
    {
      "epoch": 3.806948857950456,
      "grad_norm": 0.04451877502300554,
      "learning_rate": 2.9578070505983617e-05,
      "loss": 0.8408,
      "step": 216840
    },
    {
      "epoch": 3.807124422830457,
      "grad_norm": 0.05396986283160004,
      "learning_rate": 2.9570354175780392e-05,
      "loss": 0.8372,
      "step": 216850
    },
    {
      "epoch": 3.8072999877104583,
      "grad_norm": 0.05160566057346454,
      "learning_rate": 2.9562638750604532e-05,
      "loss": 0.8392,
      "step": 216860
    },
    {
      "epoch": 3.8074755525904598,
      "grad_norm": 0.06213281791003979,
      "learning_rate": 2.9554924230553698e-05,
      "loss": 0.8373,
      "step": 216870
    },
    {
      "epoch": 3.8076511174704613,
      "grad_norm": 0.05162443440737965,
      "learning_rate": 2.9547210615725706e-05,
      "loss": 0.842,
      "step": 216880
    },
    {
      "epoch": 3.8078266823504627,
      "grad_norm": 0.05239839618028367,
      "learning_rate": 2.9539497906218205e-05,
      "loss": 0.8383,
      "step": 216890
    },
    {
      "epoch": 3.8080022472304638,
      "grad_norm": 0.04499409468630796,
      "learning_rate": 2.953178610212897e-05,
      "loss": 0.8382,
      "step": 216900
    },
    {
      "epoch": 3.8081778121104657,
      "grad_norm": 0.05886410355732132,
      "learning_rate": 2.952407520355574e-05,
      "loss": 0.8399,
      "step": 216910
    },
    {
      "epoch": 3.8083533769904667,
      "grad_norm": 0.054736348826220575,
      "learning_rate": 2.9516365210596096e-05,
      "loss": 0.8411,
      "step": 216920
    },
    {
      "epoch": 3.808528941870468,
      "grad_norm": 0.054152477206964256,
      "learning_rate": 2.9508656123347883e-05,
      "loss": 0.8314,
      "step": 216930
    },
    {
      "epoch": 3.8087045067504697,
      "grad_norm": 0.04135410457107819,
      "learning_rate": 2.9500947941908656e-05,
      "loss": 0.8368,
      "step": 216940
    },
    {
      "epoch": 3.808880071630471,
      "grad_norm": 0.06515203382446555,
      "learning_rate": 2.9493240666376155e-05,
      "loss": 0.8344,
      "step": 216950
    },
    {
      "epoch": 3.8090556365104726,
      "grad_norm": 0.04954539049885234,
      "learning_rate": 2.948553429684798e-05,
      "loss": 0.8418,
      "step": 216960
    },
    {
      "epoch": 3.8092312013904737,
      "grad_norm": 0.05812799113499499,
      "learning_rate": 2.9477828833421772e-05,
      "loss": 0.8359,
      "step": 216970
    },
    {
      "epoch": 3.809406766270475,
      "grad_norm": 0.04232553363656654,
      "learning_rate": 2.9470124276195228e-05,
      "loss": 0.8348,
      "step": 216980
    },
    {
      "epoch": 3.8095823311504766,
      "grad_norm": 0.06253004756238967,
      "learning_rate": 2.946242062526587e-05,
      "loss": 0.8408,
      "step": 216990
    },
    {
      "epoch": 3.809757896030478,
      "grad_norm": 0.04760882881421293,
      "learning_rate": 2.9454717880731365e-05,
      "loss": 0.84,
      "step": 217000
    },
    {
      "epoch": 3.8099334609104796,
      "grad_norm": 0.04942275574157079,
      "learning_rate": 2.9447016042689317e-05,
      "loss": 0.8399,
      "step": 217010
    },
    {
      "epoch": 3.8101090257904806,
      "grad_norm": 0.05824313618163893,
      "learning_rate": 2.9439315111237238e-05,
      "loss": 0.8299,
      "step": 217020
    },
    {
      "epoch": 3.8102845906704825,
      "grad_norm": 0.05995559316748785,
      "learning_rate": 2.9431615086472806e-05,
      "loss": 0.8352,
      "step": 217030
    },
    {
      "epoch": 3.8104601555504836,
      "grad_norm": 0.04696841576170053,
      "learning_rate": 2.94239159684935e-05,
      "loss": 0.8346,
      "step": 217040
    },
    {
      "epoch": 3.810635720430485,
      "grad_norm": 0.056660098500586516,
      "learning_rate": 2.941621775739692e-05,
      "loss": 0.8381,
      "step": 217050
    },
    {
      "epoch": 3.8108112853104865,
      "grad_norm": 0.05908190047778835,
      "learning_rate": 2.9408520453280557e-05,
      "loss": 0.839,
      "step": 217060
    },
    {
      "epoch": 3.810986850190488,
      "grad_norm": 0.06124991396504309,
      "learning_rate": 2.940082405624196e-05,
      "loss": 0.8317,
      "step": 217070
    },
    {
      "epoch": 3.8111624150704895,
      "grad_norm": 0.05614191039614267,
      "learning_rate": 2.9393128566378675e-05,
      "loss": 0.8404,
      "step": 217080
    },
    {
      "epoch": 3.8113379799504905,
      "grad_norm": 0.06362326572803861,
      "learning_rate": 2.938543398378811e-05,
      "loss": 0.8378,
      "step": 217090
    },
    {
      "epoch": 3.811513544830492,
      "grad_norm": 0.06862774612917868,
      "learning_rate": 2.937774030856788e-05,
      "loss": 0.8461,
      "step": 217100
    },
    {
      "epoch": 3.8116891097104935,
      "grad_norm": 0.05333532586253141,
      "learning_rate": 2.9370047540815375e-05,
      "loss": 0.8416,
      "step": 217110
    },
    {
      "epoch": 3.811864674590495,
      "grad_norm": 0.044566917871036504,
      "learning_rate": 2.9362355680628133e-05,
      "loss": 0.8422,
      "step": 217120
    },
    {
      "epoch": 3.8120402394704964,
      "grad_norm": 0.05749830654168588,
      "learning_rate": 2.935466472810354e-05,
      "loss": 0.8418,
      "step": 217130
    },
    {
      "epoch": 3.8122158043504975,
      "grad_norm": 0.05331162033863563,
      "learning_rate": 2.9346974683339078e-05,
      "loss": 0.8424,
      "step": 217140
    },
    {
      "epoch": 3.8123913692304994,
      "grad_norm": 0.04901158490204302,
      "learning_rate": 2.93392855464322e-05,
      "loss": 0.829,
      "step": 217150
    },
    {
      "epoch": 3.8125669341105004,
      "grad_norm": 0.05093363608691648,
      "learning_rate": 2.9331597317480285e-05,
      "loss": 0.8419,
      "step": 217160
    },
    {
      "epoch": 3.812742498990502,
      "grad_norm": 0.06479845150369613,
      "learning_rate": 2.932390999658077e-05,
      "loss": 0.8401,
      "step": 217170
    },
    {
      "epoch": 3.8129180638705034,
      "grad_norm": 0.046874291255262664,
      "learning_rate": 2.9316223583831088e-05,
      "loss": 0.8363,
      "step": 217180
    },
    {
      "epoch": 3.813093628750505,
      "grad_norm": 0.05260339360580579,
      "learning_rate": 2.930853807932851e-05,
      "loss": 0.8405,
      "step": 217190
    },
    {
      "epoch": 3.8132691936305063,
      "grad_norm": 0.05257058887720699,
      "learning_rate": 2.9300853483170572e-05,
      "loss": 0.8382,
      "step": 217200
    },
    {
      "epoch": 3.8134447585105073,
      "grad_norm": 0.053998082252720105,
      "learning_rate": 2.9293169795454527e-05,
      "loss": 0.8406,
      "step": 217210
    },
    {
      "epoch": 3.813620323390509,
      "grad_norm": 0.05686674069052367,
      "learning_rate": 2.9285487016277792e-05,
      "loss": 0.832,
      "step": 217220
    },
    {
      "epoch": 3.8137958882705103,
      "grad_norm": 0.05392356848327632,
      "learning_rate": 2.9277805145737653e-05,
      "loss": 0.839,
      "step": 217230
    },
    {
      "epoch": 3.813971453150512,
      "grad_norm": 0.05553312844528772,
      "learning_rate": 2.9270124183931465e-05,
      "loss": 0.8457,
      "step": 217240
    },
    {
      "epoch": 3.8141470180305133,
      "grad_norm": 0.07503974622572539,
      "learning_rate": 2.9262444130956586e-05,
      "loss": 0.8372,
      "step": 217250
    },
    {
      "epoch": 3.8143225829105147,
      "grad_norm": 0.05815113545577752,
      "learning_rate": 2.9254764986910214e-05,
      "loss": 0.843,
      "step": 217260
    },
    {
      "epoch": 3.814498147790516,
      "grad_norm": 0.046662134764316666,
      "learning_rate": 2.924708675188979e-05,
      "loss": 0.8413,
      "step": 217270
    },
    {
      "epoch": 3.8146737126705172,
      "grad_norm": 0.06136568561607182,
      "learning_rate": 2.9239409425992498e-05,
      "loss": 0.8435,
      "step": 217280
    },
    {
      "epoch": 3.8148492775505187,
      "grad_norm": 0.04496130311075327,
      "learning_rate": 2.9231733009315672e-05,
      "loss": 0.8421,
      "step": 217290
    },
    {
      "epoch": 3.81502484243052,
      "grad_norm": 0.05251405047814573,
      "learning_rate": 2.9224057501956508e-05,
      "loss": 0.8327,
      "step": 217300
    },
    {
      "epoch": 3.8152004073105217,
      "grad_norm": 0.05457367396153035,
      "learning_rate": 2.9216382904012302e-05,
      "loss": 0.8383,
      "step": 217310
    },
    {
      "epoch": 3.815375972190523,
      "grad_norm": 0.05138439740947332,
      "learning_rate": 2.92087092155803e-05,
      "loss": 0.8468,
      "step": 217320
    },
    {
      "epoch": 3.815551537070524,
      "grad_norm": 0.05075604632020879,
      "learning_rate": 2.9201036436757695e-05,
      "loss": 0.8359,
      "step": 217330
    },
    {
      "epoch": 3.8157271019505257,
      "grad_norm": 0.05506286422402078,
      "learning_rate": 2.91933645676417e-05,
      "loss": 0.8415,
      "step": 217340
    },
    {
      "epoch": 3.815902666830527,
      "grad_norm": 0.053542496753468574,
      "learning_rate": 2.9185693608329584e-05,
      "loss": 0.833,
      "step": 217350
    },
    {
      "epoch": 3.8160782317105286,
      "grad_norm": 0.04709276070923575,
      "learning_rate": 2.917802355891842e-05,
      "loss": 0.8485,
      "step": 217360
    },
    {
      "epoch": 3.81625379659053,
      "grad_norm": 0.04965160060176507,
      "learning_rate": 2.9170354419505534e-05,
      "loss": 0.8371,
      "step": 217370
    },
    {
      "epoch": 3.8164293614705316,
      "grad_norm": 0.05830861227375066,
      "learning_rate": 2.9162686190187987e-05,
      "loss": 0.8388,
      "step": 217380
    },
    {
      "epoch": 3.816604926350533,
      "grad_norm": 0.05024693016577014,
      "learning_rate": 2.9155018871063e-05,
      "loss": 0.8361,
      "step": 217390
    },
    {
      "epoch": 3.816780491230534,
      "grad_norm": 0.07220789898643996,
      "learning_rate": 2.914735246222767e-05,
      "loss": 0.8418,
      "step": 217400
    },
    {
      "epoch": 3.8169560561105356,
      "grad_norm": 0.051325071909766276,
      "learning_rate": 2.9139686963779156e-05,
      "loss": 0.8415,
      "step": 217410
    },
    {
      "epoch": 3.817131620990537,
      "grad_norm": 0.05186060457237809,
      "learning_rate": 2.9132022375814605e-05,
      "loss": 0.8342,
      "step": 217420
    },
    {
      "epoch": 3.8173071858705385,
      "grad_norm": 0.043625592829566746,
      "learning_rate": 2.912435869843105e-05,
      "loss": 0.8364,
      "step": 217430
    },
    {
      "epoch": 3.81748275075054,
      "grad_norm": 0.04603764544549811,
      "learning_rate": 2.9116695931725703e-05,
      "loss": 0.8352,
      "step": 217440
    },
    {
      "epoch": 3.817658315630541,
      "grad_norm": 0.045966051396095185,
      "learning_rate": 2.910903407579556e-05,
      "loss": 0.8397,
      "step": 217450
    },
    {
      "epoch": 3.8178338805105425,
      "grad_norm": 0.06776435742153147,
      "learning_rate": 2.9101373130737765e-05,
      "loss": 0.8363,
      "step": 217460
    },
    {
      "epoch": 3.818009445390544,
      "grad_norm": 0.052196758557891486,
      "learning_rate": 2.9093713096649323e-05,
      "loss": 0.838,
      "step": 217470
    },
    {
      "epoch": 3.8181850102705455,
      "grad_norm": 0.053590190660677404,
      "learning_rate": 2.9086053973627314e-05,
      "loss": 0.8405,
      "step": 217480
    },
    {
      "epoch": 3.818360575150547,
      "grad_norm": 0.05067688265083248,
      "learning_rate": 2.907839576176881e-05,
      "loss": 0.836,
      "step": 217490
    },
    {
      "epoch": 3.8185361400305484,
      "grad_norm": 0.04909734318283264,
      "learning_rate": 2.9070738461170795e-05,
      "loss": 0.8378,
      "step": 217500
    },
    {
      "epoch": 3.81871170491055,
      "grad_norm": 0.05496334567904103,
      "learning_rate": 2.906308207193031e-05,
      "loss": 0.832,
      "step": 217510
    },
    {
      "epoch": 3.818887269790551,
      "grad_norm": 0.05631008877521826,
      "learning_rate": 2.905542659414439e-05,
      "loss": 0.8441,
      "step": 217520
    },
    {
      "epoch": 3.8190628346705524,
      "grad_norm": 0.05678317368765007,
      "learning_rate": 2.9047772027909936e-05,
      "loss": 0.8478,
      "step": 217530
    },
    {
      "epoch": 3.819238399550554,
      "grad_norm": 0.05359000575235598,
      "learning_rate": 2.9040118373324062e-05,
      "loss": 0.8348,
      "step": 217540
    },
    {
      "epoch": 3.8194139644305554,
      "grad_norm": 0.04772043831295848,
      "learning_rate": 2.9032465630483656e-05,
      "loss": 0.8407,
      "step": 217550
    },
    {
      "epoch": 3.819589529310557,
      "grad_norm": 0.06696473542791347,
      "learning_rate": 2.9024813799485724e-05,
      "loss": 0.8367,
      "step": 217560
    },
    {
      "epoch": 3.819765094190558,
      "grad_norm": 0.06514849206232294,
      "learning_rate": 2.9017162880427167e-05,
      "loss": 0.8342,
      "step": 217570
    },
    {
      "epoch": 3.81994065907056,
      "grad_norm": 0.06125027259968786,
      "learning_rate": 2.900951287340496e-05,
      "loss": 0.8377,
      "step": 217580
    },
    {
      "epoch": 3.820116223950561,
      "grad_norm": 0.05395400304987546,
      "learning_rate": 2.9001863778516035e-05,
      "loss": 0.8376,
      "step": 217590
    },
    {
      "epoch": 3.8202917888305623,
      "grad_norm": 0.04386858689623785,
      "learning_rate": 2.8994215595857255e-05,
      "loss": 0.8435,
      "step": 217600
    },
    {
      "epoch": 3.820467353710564,
      "grad_norm": 0.052919265230094485,
      "learning_rate": 2.8986568325525614e-05,
      "loss": 0.8379,
      "step": 217610
    },
    {
      "epoch": 3.8206429185905653,
      "grad_norm": 0.0469909856580591,
      "learning_rate": 2.8978921967617922e-05,
      "loss": 0.8359,
      "step": 217620
    },
    {
      "epoch": 3.8208184834705667,
      "grad_norm": 0.05796953206880837,
      "learning_rate": 2.8971276522231122e-05,
      "loss": 0.8328,
      "step": 217630
    },
    {
      "epoch": 3.8209940483505678,
      "grad_norm": 0.07386189347663745,
      "learning_rate": 2.8963631989462025e-05,
      "loss": 0.8367,
      "step": 217640
    },
    {
      "epoch": 3.8211696132305693,
      "grad_norm": 0.05077625535813406,
      "learning_rate": 2.8955988369407517e-05,
      "loss": 0.836,
      "step": 217650
    },
    {
      "epoch": 3.8213451781105707,
      "grad_norm": 0.06433096663627376,
      "learning_rate": 2.894834566216446e-05,
      "loss": 0.8257,
      "step": 217660
    },
    {
      "epoch": 3.821520742990572,
      "grad_norm": 0.08254608903882428,
      "learning_rate": 2.8940703867829656e-05,
      "loss": 0.8417,
      "step": 217670
    },
    {
      "epoch": 3.8216963078705737,
      "grad_norm": 0.054692110726047245,
      "learning_rate": 2.8933062986499943e-05,
      "loss": 0.8394,
      "step": 217680
    },
    {
      "epoch": 3.8218718727505747,
      "grad_norm": 0.0421971690743456,
      "learning_rate": 2.8925423018272163e-05,
      "loss": 0.843,
      "step": 217690
    },
    {
      "epoch": 3.8220474376305766,
      "grad_norm": 0.06618326744882382,
      "learning_rate": 2.8917783963243028e-05,
      "loss": 0.8419,
      "step": 217700
    },
    {
      "epoch": 3.8222230025105777,
      "grad_norm": 0.061053564484274646,
      "learning_rate": 2.8910145821509445e-05,
      "loss": 0.8437,
      "step": 217710
    },
    {
      "epoch": 3.822398567390579,
      "grad_norm": 0.05727020617273264,
      "learning_rate": 2.890250859316809e-05,
      "loss": 0.8406,
      "step": 217720
    },
    {
      "epoch": 3.8225741322705806,
      "grad_norm": 0.06864105600709133,
      "learning_rate": 2.889487227831581e-05,
      "loss": 0.8365,
      "step": 217730
    },
    {
      "epoch": 3.822749697150582,
      "grad_norm": 0.06559234990720195,
      "learning_rate": 2.8887236877049283e-05,
      "loss": 0.8344,
      "step": 217740
    },
    {
      "epoch": 3.8229252620305836,
      "grad_norm": 0.051088176523832865,
      "learning_rate": 2.8879602389465277e-05,
      "loss": 0.8417,
      "step": 217750
    },
    {
      "epoch": 3.8231008269105846,
      "grad_norm": 0.05469541270208154,
      "learning_rate": 2.887196881566058e-05,
      "loss": 0.8424,
      "step": 217760
    },
    {
      "epoch": 3.823276391790586,
      "grad_norm": 0.04819606655624512,
      "learning_rate": 2.886433615573178e-05,
      "loss": 0.8348,
      "step": 217770
    },
    {
      "epoch": 3.8234519566705876,
      "grad_norm": 0.06066112448470286,
      "learning_rate": 2.8856704409775748e-05,
      "loss": 0.8403,
      "step": 217780
    },
    {
      "epoch": 3.823627521550589,
      "grad_norm": 0.0578070037322324,
      "learning_rate": 2.8849073577889046e-05,
      "loss": 0.8433,
      "step": 217790
    },
    {
      "epoch": 3.8238030864305905,
      "grad_norm": 0.04021456416061309,
      "learning_rate": 2.8841443660168455e-05,
      "loss": 0.8406,
      "step": 217800
    },
    {
      "epoch": 3.8239786513105916,
      "grad_norm": 0.04689269823392733,
      "learning_rate": 2.8833814656710566e-05,
      "loss": 0.8358,
      "step": 217810
    },
    {
      "epoch": 3.8241542161905935,
      "grad_norm": 0.059586372589268,
      "learning_rate": 2.882618656761207e-05,
      "loss": 0.8341,
      "step": 217820
    },
    {
      "epoch": 3.8243297810705945,
      "grad_norm": 0.05048393301332705,
      "learning_rate": 2.8818559392969664e-05,
      "loss": 0.8454,
      "step": 217830
    },
    {
      "epoch": 3.824505345950596,
      "grad_norm": 0.05515371634432426,
      "learning_rate": 2.8810933132879902e-05,
      "loss": 0.8351,
      "step": 217840
    },
    {
      "epoch": 3.8246809108305975,
      "grad_norm": 0.037974768694121405,
      "learning_rate": 2.8803307787439455e-05,
      "loss": 0.8437,
      "step": 217850
    },
    {
      "epoch": 3.824856475710599,
      "grad_norm": 0.07384821010002073,
      "learning_rate": 2.8795683356744968e-05,
      "loss": 0.8438,
      "step": 217860
    },
    {
      "epoch": 3.8250320405906004,
      "grad_norm": 0.045116627362539664,
      "learning_rate": 2.8788059840892955e-05,
      "loss": 0.8486,
      "step": 217870
    },
    {
      "epoch": 3.8252076054706015,
      "grad_norm": 0.05137816335365439,
      "learning_rate": 2.8780437239980112e-05,
      "loss": 0.8371,
      "step": 217880
    },
    {
      "epoch": 3.825383170350603,
      "grad_norm": 0.0635493055194883,
      "learning_rate": 2.877281555410295e-05,
      "loss": 0.8353,
      "step": 217890
    },
    {
      "epoch": 3.8255587352306044,
      "grad_norm": 0.0814291578961263,
      "learning_rate": 2.876519478335809e-05,
      "loss": 0.845,
      "step": 217900
    },
    {
      "epoch": 3.825734300110606,
      "grad_norm": 0.04915275085982769,
      "learning_rate": 2.8757574927842014e-05,
      "loss": 0.8354,
      "step": 217910
    },
    {
      "epoch": 3.8259098649906074,
      "grad_norm": 0.05672688110338907,
      "learning_rate": 2.8749955987651317e-05,
      "loss": 0.8392,
      "step": 217920
    },
    {
      "epoch": 3.826085429870609,
      "grad_norm": 0.04610049327477415,
      "learning_rate": 2.8742337962882547e-05,
      "loss": 0.8346,
      "step": 217930
    },
    {
      "epoch": 3.8262609947506103,
      "grad_norm": 0.06777879873542218,
      "learning_rate": 2.8734720853632153e-05,
      "loss": 0.8339,
      "step": 217940
    },
    {
      "epoch": 3.8264365596306114,
      "grad_norm": 0.04391237411164208,
      "learning_rate": 2.8727104659996752e-05,
      "loss": 0.8426,
      "step": 217950
    },
    {
      "epoch": 3.826612124510613,
      "grad_norm": 0.058975691442167966,
      "learning_rate": 2.8719489382072746e-05,
      "loss": 0.836,
      "step": 217960
    },
    {
      "epoch": 3.8267876893906143,
      "grad_norm": 0.04931930086000837,
      "learning_rate": 2.8711875019956706e-05,
      "loss": 0.8398,
      "step": 217970
    },
    {
      "epoch": 3.826963254270616,
      "grad_norm": 0.05741677401108841,
      "learning_rate": 2.870426157374503e-05,
      "loss": 0.8413,
      "step": 217980
    },
    {
      "epoch": 3.8271388191506173,
      "grad_norm": 0.07181429712060745,
      "learning_rate": 2.8696649043534213e-05,
      "loss": 0.8315,
      "step": 217990
    },
    {
      "epoch": 3.8273143840306183,
      "grad_norm": 0.09330259707602125,
      "learning_rate": 2.8689037429420734e-05,
      "loss": 0.8388,
      "step": 218000
    },
    {
      "epoch": 3.8274899489106198,
      "grad_norm": 0.052211788444250014,
      "learning_rate": 2.8681426731500988e-05,
      "loss": 0.8395,
      "step": 218010
    },
    {
      "epoch": 3.8276655137906213,
      "grad_norm": 0.04315322754962974,
      "learning_rate": 2.8673816949871424e-05,
      "loss": 0.843,
      "step": 218020
    },
    {
      "epoch": 3.8278410786706227,
      "grad_norm": 0.05160784568054489,
      "learning_rate": 2.8666208084628477e-05,
      "loss": 0.8352,
      "step": 218030
    },
    {
      "epoch": 3.828016643550624,
      "grad_norm": 0.055811994579850824,
      "learning_rate": 2.865860013586849e-05,
      "loss": 0.8387,
      "step": 218040
    },
    {
      "epoch": 3.8281922084306257,
      "grad_norm": 0.04513846018871603,
      "learning_rate": 2.8650993103687968e-05,
      "loss": 0.843,
      "step": 218050
    },
    {
      "epoch": 3.828367773310627,
      "grad_norm": 0.0440722421388137,
      "learning_rate": 2.8643386988183184e-05,
      "loss": 0.8353,
      "step": 218060
    },
    {
      "epoch": 3.828543338190628,
      "grad_norm": 0.057086227456068915,
      "learning_rate": 2.86357817894506e-05,
      "loss": 0.8403,
      "step": 218070
    },
    {
      "epoch": 3.8287189030706297,
      "grad_norm": 0.05456573226152338,
      "learning_rate": 2.8628177507586484e-05,
      "loss": 0.8372,
      "step": 218080
    },
    {
      "epoch": 3.828894467950631,
      "grad_norm": 0.06429449434450157,
      "learning_rate": 2.862057414268724e-05,
      "loss": 0.8335,
      "step": 218090
    },
    {
      "epoch": 3.8290700328306326,
      "grad_norm": 0.049037838027752766,
      "learning_rate": 2.861297169484921e-05,
      "loss": 0.8374,
      "step": 218100
    },
    {
      "epoch": 3.829245597710634,
      "grad_norm": 0.055989380561605795,
      "learning_rate": 2.860537016416866e-05,
      "loss": 0.8449,
      "step": 218110
    },
    {
      "epoch": 3.829421162590635,
      "grad_norm": 0.047041374913952515,
      "learning_rate": 2.8597769550741992e-05,
      "loss": 0.8404,
      "step": 218120
    },
    {
      "epoch": 3.8295967274706366,
      "grad_norm": 0.055331828870274186,
      "learning_rate": 2.8590169854665434e-05,
      "loss": 0.8384,
      "step": 218130
    },
    {
      "epoch": 3.829772292350638,
      "grad_norm": 0.060853224020179335,
      "learning_rate": 2.8582571076035326e-05,
      "loss": 0.8401,
      "step": 218140
    },
    {
      "epoch": 3.8299478572306396,
      "grad_norm": 0.04805027525511307,
      "learning_rate": 2.8574973214947905e-05,
      "loss": 0.8337,
      "step": 218150
    },
    {
      "epoch": 3.830123422110641,
      "grad_norm": 0.04925848259285232,
      "learning_rate": 2.8567376271499435e-05,
      "loss": 0.8349,
      "step": 218160
    },
    {
      "epoch": 3.8302989869906425,
      "grad_norm": 0.08251862772793578,
      "learning_rate": 2.8559780245786233e-05,
      "loss": 0.8434,
      "step": 218170
    },
    {
      "epoch": 3.830474551870644,
      "grad_norm": 0.06677559085774283,
      "learning_rate": 2.8552185137904458e-05,
      "loss": 0.839,
      "step": 218180
    },
    {
      "epoch": 3.830650116750645,
      "grad_norm": 0.049540208311358996,
      "learning_rate": 2.8544590947950394e-05,
      "loss": 0.8375,
      "step": 218190
    },
    {
      "epoch": 3.8308256816306465,
      "grad_norm": 0.050911789926577536,
      "learning_rate": 2.8536997676020265e-05,
      "loss": 0.8502,
      "step": 218200
    },
    {
      "epoch": 3.831001246510648,
      "grad_norm": 0.05945922816175219,
      "learning_rate": 2.8529405322210214e-05,
      "loss": 0.8385,
      "step": 218210
    },
    {
      "epoch": 3.8311768113906495,
      "grad_norm": 0.062153917777248636,
      "learning_rate": 2.852181388661654e-05,
      "loss": 0.8379,
      "step": 218220
    },
    {
      "epoch": 3.831352376270651,
      "grad_norm": 0.0744492683464073,
      "learning_rate": 2.8514223369335345e-05,
      "loss": 0.8473,
      "step": 218230
    },
    {
      "epoch": 3.831527941150652,
      "grad_norm": 0.05471730002340675,
      "learning_rate": 2.850663377046286e-05,
      "loss": 0.8415,
      "step": 218240
    },
    {
      "epoch": 3.831703506030654,
      "grad_norm": 0.06192529510611837,
      "learning_rate": 2.84990450900952e-05,
      "loss": 0.8447,
      "step": 218250
    },
    {
      "epoch": 3.831879070910655,
      "grad_norm": 0.05653469580190741,
      "learning_rate": 2.8491457328328526e-05,
      "loss": 0.8392,
      "step": 218260
    },
    {
      "epoch": 3.8320546357906564,
      "grad_norm": 0.06047528120682877,
      "learning_rate": 2.8483870485259007e-05,
      "loss": 0.8403,
      "step": 218270
    },
    {
      "epoch": 3.832230200670658,
      "grad_norm": 0.05087370914830749,
      "learning_rate": 2.847628456098269e-05,
      "loss": 0.8424,
      "step": 218280
    },
    {
      "epoch": 3.8324057655506594,
      "grad_norm": 0.04687244494809302,
      "learning_rate": 2.8468699555595813e-05,
      "loss": 0.8399,
      "step": 218290
    },
    {
      "epoch": 3.832581330430661,
      "grad_norm": 0.05535476382325871,
      "learning_rate": 2.8461115469194384e-05,
      "loss": 0.8362,
      "step": 218300
    },
    {
      "epoch": 3.832756895310662,
      "grad_norm": 0.04647643017570771,
      "learning_rate": 2.8453532301874542e-05,
      "loss": 0.831,
      "step": 218310
    },
    {
      "epoch": 3.8329324601906634,
      "grad_norm": 0.05094426988029317,
      "learning_rate": 2.844595005373233e-05,
      "loss": 0.8467,
      "step": 218320
    },
    {
      "epoch": 3.833108025070665,
      "grad_norm": 0.07458383075135919,
      "learning_rate": 2.8438368724863835e-05,
      "loss": 0.8496,
      "step": 218330
    },
    {
      "epoch": 3.8332835899506663,
      "grad_norm": 0.04755210771512372,
      "learning_rate": 2.8430788315365148e-05,
      "loss": 0.8427,
      "step": 218340
    },
    {
      "epoch": 3.833459154830668,
      "grad_norm": 0.05598820424806286,
      "learning_rate": 2.8423208825332242e-05,
      "loss": 0.8354,
      "step": 218350
    },
    {
      "epoch": 3.833634719710669,
      "grad_norm": 0.047705029492315754,
      "learning_rate": 2.841563025486119e-05,
      "loss": 0.8376,
      "step": 218360
    },
    {
      "epoch": 3.8338102845906707,
      "grad_norm": 0.05452524272249354,
      "learning_rate": 2.8408052604048047e-05,
      "loss": 0.8353,
      "step": 218370
    },
    {
      "epoch": 3.833985849470672,
      "grad_norm": 0.06595402603784706,
      "learning_rate": 2.840047587298873e-05,
      "loss": 0.8452,
      "step": 218380
    },
    {
      "epoch": 3.8341614143506733,
      "grad_norm": 0.05029788778667228,
      "learning_rate": 2.8392900061779358e-05,
      "loss": 0.8404,
      "step": 218390
    },
    {
      "epoch": 3.8343369792306747,
      "grad_norm": 0.061210028038722185,
      "learning_rate": 2.8385325170515826e-05,
      "loss": 0.8328,
      "step": 218400
    },
    {
      "epoch": 3.834512544110676,
      "grad_norm": 0.0511658572771602,
      "learning_rate": 2.837775119929416e-05,
      "loss": 0.836,
      "step": 218410
    },
    {
      "epoch": 3.8346881089906777,
      "grad_norm": 0.05447957013563409,
      "learning_rate": 2.8370178148210293e-05,
      "loss": 0.8342,
      "step": 218420
    },
    {
      "epoch": 3.8348636738706787,
      "grad_norm": 0.051541719534348496,
      "learning_rate": 2.8362606017360176e-05,
      "loss": 0.8419,
      "step": 218430
    },
    {
      "epoch": 3.83503923875068,
      "grad_norm": 0.05284625414103271,
      "learning_rate": 2.8355034806839792e-05,
      "loss": 0.8363,
      "step": 218440
    },
    {
      "epoch": 3.8352148036306817,
      "grad_norm": 0.059232496685376355,
      "learning_rate": 2.8347464516744986e-05,
      "loss": 0.843,
      "step": 218450
    },
    {
      "epoch": 3.835390368510683,
      "grad_norm": 0.043039978245789305,
      "learning_rate": 2.8339895147171784e-05,
      "loss": 0.84,
      "step": 218460
    },
    {
      "epoch": 3.8355659333906846,
      "grad_norm": 0.06845129705140372,
      "learning_rate": 2.8332326698216e-05,
      "loss": 0.8431,
      "step": 218470
    },
    {
      "epoch": 3.8357414982706857,
      "grad_norm": 0.055852932708701876,
      "learning_rate": 2.832475916997361e-05,
      "loss": 0.8304,
      "step": 218480
    },
    {
      "epoch": 3.8359170631506876,
      "grad_norm": 0.04750815592257536,
      "learning_rate": 2.8317192562540407e-05,
      "loss": 0.8362,
      "step": 218490
    },
    {
      "epoch": 3.8360926280306886,
      "grad_norm": 0.05317315294702896,
      "learning_rate": 2.8309626876012308e-05,
      "loss": 0.8381,
      "step": 218500
    },
    {
      "epoch": 3.83626819291069,
      "grad_norm": 0.09352393991311736,
      "learning_rate": 2.8302062110485196e-05,
      "loss": 0.8388,
      "step": 218510
    },
    {
      "epoch": 3.8364437577906916,
      "grad_norm": 0.043923905637185415,
      "learning_rate": 2.8294498266054857e-05,
      "loss": 0.8372,
      "step": 218520
    },
    {
      "epoch": 3.836619322670693,
      "grad_norm": 0.05355854751146365,
      "learning_rate": 2.8286935342817155e-05,
      "loss": 0.8431,
      "step": 218530
    },
    {
      "epoch": 3.8367948875506945,
      "grad_norm": 0.051689072993707494,
      "learning_rate": 2.8279373340867954e-05,
      "loss": 0.847,
      "step": 218540
    },
    {
      "epoch": 3.8369704524306956,
      "grad_norm": 0.0485079503679546,
      "learning_rate": 2.8271812260302998e-05,
      "loss": 0.8387,
      "step": 218550
    },
    {
      "epoch": 3.837146017310697,
      "grad_norm": 0.050165946133104024,
      "learning_rate": 2.8264252101218125e-05,
      "loss": 0.8448,
      "step": 218560
    },
    {
      "epoch": 3.8373215821906985,
      "grad_norm": 0.049819104218074506,
      "learning_rate": 2.825669286370911e-05,
      "loss": 0.8416,
      "step": 218570
    },
    {
      "epoch": 3.8374971470707,
      "grad_norm": 0.05945672059659792,
      "learning_rate": 2.824913454787178e-05,
      "loss": 0.8304,
      "step": 218580
    },
    {
      "epoch": 3.8376727119507015,
      "grad_norm": 0.0633014695748565,
      "learning_rate": 2.824157715380181e-05,
      "loss": 0.8367,
      "step": 218590
    },
    {
      "epoch": 3.8378482768307025,
      "grad_norm": 0.0582198727770911,
      "learning_rate": 2.8234020681595015e-05,
      "loss": 0.8288,
      "step": 218600
    },
    {
      "epoch": 3.8380238417107044,
      "grad_norm": 0.05437772782170409,
      "learning_rate": 2.8226465131347152e-05,
      "loss": 0.8469,
      "step": 218610
    },
    {
      "epoch": 3.8381994065907055,
      "grad_norm": 0.05166921122087655,
      "learning_rate": 2.821891050315385e-05,
      "loss": 0.8384,
      "step": 218620
    },
    {
      "epoch": 3.838374971470707,
      "grad_norm": 0.052751713991498826,
      "learning_rate": 2.8211356797110982e-05,
      "loss": 0.8432,
      "step": 218630
    },
    {
      "epoch": 3.8385505363507084,
      "grad_norm": 0.05840720095844199,
      "learning_rate": 2.820380401331412e-05,
      "loss": 0.851,
      "step": 218640
    },
    {
      "epoch": 3.83872610123071,
      "grad_norm": 0.05596873595176756,
      "learning_rate": 2.8196252151859054e-05,
      "loss": 0.8344,
      "step": 218650
    },
    {
      "epoch": 3.8389016661107114,
      "grad_norm": 0.04513060176858437,
      "learning_rate": 2.8188701212841395e-05,
      "loss": 0.8401,
      "step": 218660
    },
    {
      "epoch": 3.8390772309907124,
      "grad_norm": 0.044955824974307376,
      "learning_rate": 2.8181151196356853e-05,
      "loss": 0.85,
      "step": 218670
    },
    {
      "epoch": 3.839252795870714,
      "grad_norm": 0.06490679519715688,
      "learning_rate": 2.81736021025011e-05,
      "loss": 0.8427,
      "step": 218680
    },
    {
      "epoch": 3.8394283607507154,
      "grad_norm": 0.04644190358315839,
      "learning_rate": 2.816605393136974e-05,
      "loss": 0.8383,
      "step": 218690
    },
    {
      "epoch": 3.839603925630717,
      "grad_norm": 0.07216746717606763,
      "learning_rate": 2.815850668305844e-05,
      "loss": 0.8419,
      "step": 218700
    },
    {
      "epoch": 3.8397794905107183,
      "grad_norm": 0.05102942508802779,
      "learning_rate": 2.815096035766285e-05,
      "loss": 0.8394,
      "step": 218710
    },
    {
      "epoch": 3.83995505539072,
      "grad_norm": 0.05705069223036557,
      "learning_rate": 2.8143414955278514e-05,
      "loss": 0.8358,
      "step": 218720
    },
    {
      "epoch": 3.8401306202707213,
      "grad_norm": 0.05118189626464809,
      "learning_rate": 2.813587047600107e-05,
      "loss": 0.8384,
      "step": 218730
    },
    {
      "epoch": 3.8403061851507223,
      "grad_norm": 0.043962997021085304,
      "learning_rate": 2.812832691992613e-05,
      "loss": 0.8368,
      "step": 218740
    },
    {
      "epoch": 3.840481750030724,
      "grad_norm": 0.045758617622988274,
      "learning_rate": 2.8120784287149278e-05,
      "loss": 0.8411,
      "step": 218750
    },
    {
      "epoch": 3.8406573149107253,
      "grad_norm": 0.05811093720257758,
      "learning_rate": 2.8113242577766014e-05,
      "loss": 0.8457,
      "step": 218760
    },
    {
      "epoch": 3.8408328797907267,
      "grad_norm": 0.06408202706290622,
      "learning_rate": 2.8105701791871946e-05,
      "loss": 0.8442,
      "step": 218770
    },
    {
      "epoch": 3.841008444670728,
      "grad_norm": 0.05604478048462434,
      "learning_rate": 2.8098161929562628e-05,
      "loss": 0.8385,
      "step": 218780
    },
    {
      "epoch": 3.8411840095507293,
      "grad_norm": 0.04500563980143818,
      "learning_rate": 2.8090622990933523e-05,
      "loss": 0.843,
      "step": 218790
    },
    {
      "epoch": 3.8413595744307307,
      "grad_norm": 0.06166870751693459,
      "learning_rate": 2.808308497608024e-05,
      "loss": 0.8377,
      "step": 218800
    },
    {
      "epoch": 3.841535139310732,
      "grad_norm": 0.05785331788004471,
      "learning_rate": 2.807554788509823e-05,
      "loss": 0.842,
      "step": 218810
    },
    {
      "epoch": 3.8417107041907337,
      "grad_norm": 0.06194902497771558,
      "learning_rate": 2.8068011718083036e-05,
      "loss": 0.835,
      "step": 218820
    },
    {
      "epoch": 3.841886269070735,
      "grad_norm": 0.05585043216398764,
      "learning_rate": 2.806047647513007e-05,
      "loss": 0.84,
      "step": 218830
    },
    {
      "epoch": 3.8420618339507366,
      "grad_norm": 0.046718488503334155,
      "learning_rate": 2.8052942156334854e-05,
      "loss": 0.8399,
      "step": 218840
    },
    {
      "epoch": 3.842237398830738,
      "grad_norm": 0.047197391820726284,
      "learning_rate": 2.8045408761792863e-05,
      "loss": 0.832,
      "step": 218850
    },
    {
      "epoch": 3.842412963710739,
      "grad_norm": 0.04658474858146443,
      "learning_rate": 2.80378762915995e-05,
      "loss": 0.8488,
      "step": 218860
    },
    {
      "epoch": 3.8425885285907406,
      "grad_norm": 0.041758361819475835,
      "learning_rate": 2.803034474585023e-05,
      "loss": 0.838,
      "step": 218870
    },
    {
      "epoch": 3.842764093470742,
      "grad_norm": 0.06277328586329195,
      "learning_rate": 2.8022814124640502e-05,
      "loss": 0.8394,
      "step": 218880
    },
    {
      "epoch": 3.8429396583507436,
      "grad_norm": 0.06499075806884659,
      "learning_rate": 2.801528442806567e-05,
      "loss": 0.8384,
      "step": 218890
    },
    {
      "epoch": 3.843115223230745,
      "grad_norm": 0.04786474407327228,
      "learning_rate": 2.8007755656221178e-05,
      "loss": 0.8462,
      "step": 218900
    },
    {
      "epoch": 3.843290788110746,
      "grad_norm": 0.0442158492408872,
      "learning_rate": 2.8000227809202417e-05,
      "loss": 0.8411,
      "step": 218910
    },
    {
      "epoch": 3.8434663529907476,
      "grad_norm": 0.059701992922136585,
      "learning_rate": 2.7992700887104782e-05,
      "loss": 0.8451,
      "step": 218920
    },
    {
      "epoch": 3.843641917870749,
      "grad_norm": 0.048779777010636946,
      "learning_rate": 2.798517489002359e-05,
      "loss": 0.8432,
      "step": 218930
    },
    {
      "epoch": 3.8438174827507505,
      "grad_norm": 0.04363513962422197,
      "learning_rate": 2.797764981805423e-05,
      "loss": 0.8348,
      "step": 218940
    },
    {
      "epoch": 3.843993047630752,
      "grad_norm": 0.050645610861058336,
      "learning_rate": 2.7970125671292063e-05,
      "loss": 0.8348,
      "step": 218950
    },
    {
      "epoch": 3.8441686125107535,
      "grad_norm": 0.04838799353295888,
      "learning_rate": 2.796260244983234e-05,
      "loss": 0.833,
      "step": 218960
    },
    {
      "epoch": 3.844344177390755,
      "grad_norm": 0.06237712831766086,
      "learning_rate": 2.79550801537705e-05,
      "loss": 0.8422,
      "step": 218970
    },
    {
      "epoch": 3.844519742270756,
      "grad_norm": 0.04255864791706977,
      "learning_rate": 2.7947558783201776e-05,
      "loss": 0.85,
      "step": 218980
    },
    {
      "epoch": 3.8446953071507575,
      "grad_norm": 0.06966975190174646,
      "learning_rate": 2.7940038338221503e-05,
      "loss": 0.8475,
      "step": 218990
    },
    {
      "epoch": 3.844870872030759,
      "grad_norm": 0.05713200132591878,
      "learning_rate": 2.7932518818924924e-05,
      "loss": 0.8309,
      "step": 219000
    },
    {
      "epoch": 3.8450464369107604,
      "grad_norm": 0.05060929306082336,
      "learning_rate": 2.7925000225407327e-05,
      "loss": 0.8362,
      "step": 219010
    },
    {
      "epoch": 3.845222001790762,
      "grad_norm": 0.04827013355956168,
      "learning_rate": 2.791748255776402e-05,
      "loss": 0.842,
      "step": 219020
    },
    {
      "epoch": 3.845397566670763,
      "grad_norm": 0.05710141796963532,
      "learning_rate": 2.790996581609018e-05,
      "loss": 0.8435,
      "step": 219030
    },
    {
      "epoch": 3.845573131550765,
      "grad_norm": 0.07920795934782872,
      "learning_rate": 2.790245000048108e-05,
      "loss": 0.84,
      "step": 219040
    },
    {
      "epoch": 3.845748696430766,
      "grad_norm": 0.055841374294081134,
      "learning_rate": 2.789493511103197e-05,
      "loss": 0.8349,
      "step": 219050
    },
    {
      "epoch": 3.8459242613107674,
      "grad_norm": 0.07859509973885805,
      "learning_rate": 2.7887421147838018e-05,
      "loss": 0.84,
      "step": 219060
    },
    {
      "epoch": 3.846099826190769,
      "grad_norm": 0.06759180872247454,
      "learning_rate": 2.7879908110994458e-05,
      "loss": 0.8374,
      "step": 219070
    },
    {
      "epoch": 3.8462753910707703,
      "grad_norm": 0.04704867044123688,
      "learning_rate": 2.7872396000596474e-05,
      "loss": 0.8368,
      "step": 219080
    },
    {
      "epoch": 3.846450955950772,
      "grad_norm": 0.07737852523372644,
      "learning_rate": 2.786488481673928e-05,
      "loss": 0.8337,
      "step": 219090
    },
    {
      "epoch": 3.846626520830773,
      "grad_norm": 0.058699116156820864,
      "learning_rate": 2.785737455951798e-05,
      "loss": 0.8391,
      "step": 219100
    },
    {
      "epoch": 3.8468020857107743,
      "grad_norm": 0.05087541050617973,
      "learning_rate": 2.7849865229027748e-05,
      "loss": 0.829,
      "step": 219110
    },
    {
      "epoch": 3.846977650590776,
      "grad_norm": 0.047261070885180556,
      "learning_rate": 2.7842356825363787e-05,
      "loss": 0.8369,
      "step": 219120
    },
    {
      "epoch": 3.8471532154707773,
      "grad_norm": 0.047084120701732055,
      "learning_rate": 2.7834849348621123e-05,
      "loss": 0.8472,
      "step": 219130
    },
    {
      "epoch": 3.8473287803507787,
      "grad_norm": 0.06734470497782259,
      "learning_rate": 2.7827342798895006e-05,
      "loss": 0.8403,
      "step": 219140
    },
    {
      "epoch": 3.8475043452307798,
      "grad_norm": 0.03862217525995662,
      "learning_rate": 2.7819837176280433e-05,
      "loss": 0.8402,
      "step": 219150
    },
    {
      "epoch": 3.8476799101107817,
      "grad_norm": 0.051618892189436126,
      "learning_rate": 2.7812332480872587e-05,
      "loss": 0.8424,
      "step": 219160
    },
    {
      "epoch": 3.8478554749907827,
      "grad_norm": 0.04737372027883426,
      "learning_rate": 2.7804828712766494e-05,
      "loss": 0.8364,
      "step": 219170
    },
    {
      "epoch": 3.848031039870784,
      "grad_norm": 0.04815093514194723,
      "learning_rate": 2.779732587205724e-05,
      "loss": 0.845,
      "step": 219180
    },
    {
      "epoch": 3.8482066047507857,
      "grad_norm": 0.05293595872308211,
      "learning_rate": 2.7789823958839927e-05,
      "loss": 0.8383,
      "step": 219190
    },
    {
      "epoch": 3.848382169630787,
      "grad_norm": 0.053440074841171525,
      "learning_rate": 2.778232297320954e-05,
      "loss": 0.8403,
      "step": 219200
    },
    {
      "epoch": 3.8485577345107886,
      "grad_norm": 0.04752215232546679,
      "learning_rate": 2.777482291526116e-05,
      "loss": 0.841,
      "step": 219210
    },
    {
      "epoch": 3.8487332993907897,
      "grad_norm": 0.0513170924959201,
      "learning_rate": 2.776732378508983e-05,
      "loss": 0.8323,
      "step": 219220
    },
    {
      "epoch": 3.848908864270791,
      "grad_norm": 0.05520270351482909,
      "learning_rate": 2.7759825582790523e-05,
      "loss": 0.8495,
      "step": 219230
    },
    {
      "epoch": 3.8490844291507926,
      "grad_norm": 0.04720325976061258,
      "learning_rate": 2.7752328308458254e-05,
      "loss": 0.8334,
      "step": 219240
    },
    {
      "epoch": 3.849259994030794,
      "grad_norm": 0.042642514744637,
      "learning_rate": 2.774483196218803e-05,
      "loss": 0.834,
      "step": 219250
    },
    {
      "epoch": 3.8494355589107956,
      "grad_norm": 0.043755223113268384,
      "learning_rate": 2.7737336544074854e-05,
      "loss": 0.8396,
      "step": 219260
    },
    {
      "epoch": 3.8496111237907966,
      "grad_norm": 0.05370640005363481,
      "learning_rate": 2.7729842054213627e-05,
      "loss": 0.8417,
      "step": 219270
    },
    {
      "epoch": 3.8497866886707985,
      "grad_norm": 0.05756470015149237,
      "learning_rate": 2.7722348492699345e-05,
      "loss": 0.8435,
      "step": 219280
    },
    {
      "epoch": 3.8499622535507996,
      "grad_norm": 0.046676551854337646,
      "learning_rate": 2.7714855859626987e-05,
      "loss": 0.8401,
      "step": 219290
    },
    {
      "epoch": 3.850137818430801,
      "grad_norm": 0.048272264890516066,
      "learning_rate": 2.7707364155091394e-05,
      "loss": 0.8329,
      "step": 219300
    },
    {
      "epoch": 3.8503133833108025,
      "grad_norm": 0.06607993492842663,
      "learning_rate": 2.7699873379187592e-05,
      "loss": 0.8368,
      "step": 219310
    },
    {
      "epoch": 3.850488948190804,
      "grad_norm": 0.053986448515217576,
      "learning_rate": 2.7692383532010413e-05,
      "loss": 0.8364,
      "step": 219320
    },
    {
      "epoch": 3.8506645130708055,
      "grad_norm": 0.05422751882106154,
      "learning_rate": 2.768489461365482e-05,
      "loss": 0.8395,
      "step": 219330
    },
    {
      "epoch": 3.8508400779508065,
      "grad_norm": 0.04697555290739746,
      "learning_rate": 2.7677406624215635e-05,
      "loss": 0.8353,
      "step": 219340
    },
    {
      "epoch": 3.851015642830808,
      "grad_norm": 0.055276727725604595,
      "learning_rate": 2.7669919563787755e-05,
      "loss": 0.8402,
      "step": 219350
    },
    {
      "epoch": 3.8511912077108095,
      "grad_norm": 0.04796310299523721,
      "learning_rate": 2.7662433432466073e-05,
      "loss": 0.8404,
      "step": 219360
    },
    {
      "epoch": 3.851366772590811,
      "grad_norm": 0.0648017285478401,
      "learning_rate": 2.7654948230345395e-05,
      "loss": 0.8396,
      "step": 219370
    },
    {
      "epoch": 3.8515423374708124,
      "grad_norm": 0.08281337946208898,
      "learning_rate": 2.7647463957520577e-05,
      "loss": 0.8373,
      "step": 219380
    },
    {
      "epoch": 3.851717902350814,
      "grad_norm": 0.04601857719279901,
      "learning_rate": 2.7639980614086474e-05,
      "loss": 0.8412,
      "step": 219390
    },
    {
      "epoch": 3.8518934672308154,
      "grad_norm": 0.06140057852682502,
      "learning_rate": 2.763249820013785e-05,
      "loss": 0.8376,
      "step": 219400
    },
    {
      "epoch": 3.8520690321108164,
      "grad_norm": 0.05154354951646932,
      "learning_rate": 2.762501671576953e-05,
      "loss": 0.8442,
      "step": 219410
    },
    {
      "epoch": 3.852244596990818,
      "grad_norm": 0.054919854762860286,
      "learning_rate": 2.761753616107631e-05,
      "loss": 0.8444,
      "step": 219420
    },
    {
      "epoch": 3.8524201618708194,
      "grad_norm": 0.1315625722015553,
      "learning_rate": 2.7610056536153006e-05,
      "loss": 0.8395,
      "step": 219430
    },
    {
      "epoch": 3.852595726750821,
      "grad_norm": 0.04711718851976625,
      "learning_rate": 2.7602577841094317e-05,
      "loss": 0.8452,
      "step": 219440
    },
    {
      "epoch": 3.8527712916308223,
      "grad_norm": 0.06178117331809661,
      "learning_rate": 2.7595100075995047e-05,
      "loss": 0.8328,
      "step": 219450
    },
    {
      "epoch": 3.8529468565108234,
      "grad_norm": 0.04838934072706186,
      "learning_rate": 2.7587623240949946e-05,
      "loss": 0.8398,
      "step": 219460
    },
    {
      "epoch": 3.853122421390825,
      "grad_norm": 0.06263911156102961,
      "learning_rate": 2.7580147336053656e-05,
      "loss": 0.8414,
      "step": 219470
    },
    {
      "epoch": 3.8532979862708263,
      "grad_norm": 0.04385880024959037,
      "learning_rate": 2.757267236140104e-05,
      "loss": 0.8329,
      "step": 219480
    },
    {
      "epoch": 3.853473551150828,
      "grad_norm": 0.04807344946702375,
      "learning_rate": 2.756519831708671e-05,
      "loss": 0.8384,
      "step": 219490
    },
    {
      "epoch": 3.8536491160308293,
      "grad_norm": 0.05520861849586192,
      "learning_rate": 2.755772520320543e-05,
      "loss": 0.8449,
      "step": 219500
    },
    {
      "epoch": 3.8538246809108307,
      "grad_norm": 0.04594010517425029,
      "learning_rate": 2.7550253019851807e-05,
      "loss": 0.8398,
      "step": 219510
    },
    {
      "epoch": 3.8540002457908322,
      "grad_norm": 0.04444677894261225,
      "learning_rate": 2.754278176712055e-05,
      "loss": 0.8451,
      "step": 219520
    },
    {
      "epoch": 3.8541758106708333,
      "grad_norm": 0.06401653575513012,
      "learning_rate": 2.753531144510636e-05,
      "loss": 0.8419,
      "step": 219530
    },
    {
      "epoch": 3.8543513755508347,
      "grad_norm": 0.05587957322349888,
      "learning_rate": 2.752784205390383e-05,
      "loss": 0.8381,
      "step": 219540
    },
    {
      "epoch": 3.854526940430836,
      "grad_norm": 0.04691480143224526,
      "learning_rate": 2.7520373593607613e-05,
      "loss": 0.8379,
      "step": 219550
    },
    {
      "epoch": 3.8547025053108377,
      "grad_norm": 0.057707112635505,
      "learning_rate": 2.7512906064312378e-05,
      "loss": 0.8431,
      "step": 219560
    },
    {
      "epoch": 3.854878070190839,
      "grad_norm": 0.06142877287486154,
      "learning_rate": 2.7505439466112676e-05,
      "loss": 0.841,
      "step": 219570
    },
    {
      "epoch": 3.85505363507084,
      "grad_norm": 0.04622621599027552,
      "learning_rate": 2.7497973799103144e-05,
      "loss": 0.8409,
      "step": 219580
    },
    {
      "epoch": 3.8552291999508417,
      "grad_norm": 0.044986058361318214,
      "learning_rate": 2.749050906337837e-05,
      "loss": 0.8409,
      "step": 219590
    },
    {
      "epoch": 3.855404764830843,
      "grad_norm": 0.07789103256657255,
      "learning_rate": 2.7483045259032962e-05,
      "loss": 0.839,
      "step": 219600
    },
    {
      "epoch": 3.8555803297108446,
      "grad_norm": 0.04304814923066816,
      "learning_rate": 2.747558238616142e-05,
      "loss": 0.8426,
      "step": 219610
    },
    {
      "epoch": 3.855755894590846,
      "grad_norm": 0.056027129552569076,
      "learning_rate": 2.746812044485836e-05,
      "loss": 0.8346,
      "step": 219620
    },
    {
      "epoch": 3.8559314594708476,
      "grad_norm": 0.04877868982638914,
      "learning_rate": 2.7460659435218323e-05,
      "loss": 0.8412,
      "step": 219630
    },
    {
      "epoch": 3.856107024350849,
      "grad_norm": 0.05131186744240273,
      "learning_rate": 2.7453199357335774e-05,
      "loss": 0.8384,
      "step": 219640
    },
    {
      "epoch": 3.85628258923085,
      "grad_norm": 0.05050551424641952,
      "learning_rate": 2.744574021130533e-05,
      "loss": 0.8322,
      "step": 219650
    },
    {
      "epoch": 3.8564581541108516,
      "grad_norm": 0.052580156217489564,
      "learning_rate": 2.7438281997221443e-05,
      "loss": 0.8386,
      "step": 219660
    },
    {
      "epoch": 3.856633718990853,
      "grad_norm": 0.0516205556855635,
      "learning_rate": 2.7430824715178632e-05,
      "loss": 0.8471,
      "step": 219670
    },
    {
      "epoch": 3.8568092838708545,
      "grad_norm": 0.05913543396641111,
      "learning_rate": 2.7423368365271356e-05,
      "loss": 0.8409,
      "step": 219680
    },
    {
      "epoch": 3.856984848750856,
      "grad_norm": 0.04611177466452878,
      "learning_rate": 2.7415912947594097e-05,
      "loss": 0.8436,
      "step": 219690
    },
    {
      "epoch": 3.857160413630857,
      "grad_norm": 0.05051800936031428,
      "learning_rate": 2.7408458462241348e-05,
      "loss": 0.8485,
      "step": 219700
    },
    {
      "epoch": 3.857335978510859,
      "grad_norm": 0.056106112646638076,
      "learning_rate": 2.740100490930752e-05,
      "loss": 0.8322,
      "step": 219710
    },
    {
      "epoch": 3.85751154339086,
      "grad_norm": 0.058245128026988204,
      "learning_rate": 2.7393552288887048e-05,
      "loss": 0.8411,
      "step": 219720
    },
    {
      "epoch": 3.8576871082708615,
      "grad_norm": 0.0569899594654877,
      "learning_rate": 2.7386100601074422e-05,
      "loss": 0.8364,
      "step": 219730
    },
    {
      "epoch": 3.857862673150863,
      "grad_norm": 0.05074267068500313,
      "learning_rate": 2.7378649845963968e-05,
      "loss": 0.8474,
      "step": 219740
    },
    {
      "epoch": 3.8580382380308644,
      "grad_norm": 0.06001791742419816,
      "learning_rate": 2.7371200023650135e-05,
      "loss": 0.8408,
      "step": 219750
    },
    {
      "epoch": 3.858213802910866,
      "grad_norm": 0.050315371637615076,
      "learning_rate": 2.7363751134227305e-05,
      "loss": 0.8413,
      "step": 219760
    },
    {
      "epoch": 3.858389367790867,
      "grad_norm": 0.04667407663644479,
      "learning_rate": 2.7356303177789887e-05,
      "loss": 0.8354,
      "step": 219770
    },
    {
      "epoch": 3.8585649326708684,
      "grad_norm": 0.04294186151886764,
      "learning_rate": 2.7348856154432197e-05,
      "loss": 0.8382,
      "step": 219780
    },
    {
      "epoch": 3.85874049755087,
      "grad_norm": 0.04814456719860335,
      "learning_rate": 2.7341410064248614e-05,
      "loss": 0.8434,
      "step": 219790
    },
    {
      "epoch": 3.8589160624308714,
      "grad_norm": 0.04391458677720737,
      "learning_rate": 2.733396490733351e-05,
      "loss": 0.8375,
      "step": 219800
    },
    {
      "epoch": 3.859091627310873,
      "grad_norm": 0.05996067588479444,
      "learning_rate": 2.732652068378113e-05,
      "loss": 0.837,
      "step": 219810
    },
    {
      "epoch": 3.859267192190874,
      "grad_norm": 0.06486119562804217,
      "learning_rate": 2.7319077393685916e-05,
      "loss": 0.8445,
      "step": 219820
    },
    {
      "epoch": 3.859442757070876,
      "grad_norm": 0.04909671484502833,
      "learning_rate": 2.7311635037142073e-05,
      "loss": 0.8452,
      "step": 219830
    },
    {
      "epoch": 3.859618321950877,
      "grad_norm": 0.05819495021616859,
      "learning_rate": 2.730419361424398e-05,
      "loss": 0.8392,
      "step": 219840
    },
    {
      "epoch": 3.8597938868308783,
      "grad_norm": 0.05050589900022409,
      "learning_rate": 2.7296753125085832e-05,
      "loss": 0.8341,
      "step": 219850
    },
    {
      "epoch": 3.85996945171088,
      "grad_norm": 0.059585097748732184,
      "learning_rate": 2.728931356976196e-05,
      "loss": 0.8364,
      "step": 219860
    },
    {
      "epoch": 3.8601450165908813,
      "grad_norm": 0.05032041959338376,
      "learning_rate": 2.7281874948366635e-05,
      "loss": 0.8419,
      "step": 219870
    },
    {
      "epoch": 3.8603205814708828,
      "grad_norm": 0.05632569125041839,
      "learning_rate": 2.7274437260994055e-05,
      "loss": 0.8451,
      "step": 219880
    },
    {
      "epoch": 3.860496146350884,
      "grad_norm": 0.04717146926614923,
      "learning_rate": 2.7267000507738483e-05,
      "loss": 0.8423,
      "step": 219890
    },
    {
      "epoch": 3.8606717112308853,
      "grad_norm": 0.07084589564966515,
      "learning_rate": 2.725956468869418e-05,
      "loss": 0.8376,
      "step": 219900
    },
    {
      "epoch": 3.8608472761108867,
      "grad_norm": 0.05450275988637983,
      "learning_rate": 2.7252129803955283e-05,
      "loss": 0.8458,
      "step": 219910
    },
    {
      "epoch": 3.861022840990888,
      "grad_norm": 0.04429101652341201,
      "learning_rate": 2.7244695853616047e-05,
      "loss": 0.8406,
      "step": 219920
    },
    {
      "epoch": 3.8611984058708897,
      "grad_norm": 0.05005283784487549,
      "learning_rate": 2.723726283777065e-05,
      "loss": 0.8389,
      "step": 219930
    },
    {
      "epoch": 3.8613739707508907,
      "grad_norm": 0.05366841644020899,
      "learning_rate": 2.7229830756513287e-05,
      "loss": 0.8415,
      "step": 219940
    },
    {
      "epoch": 3.8615495356308926,
      "grad_norm": 0.05389590765979715,
      "learning_rate": 2.722239960993808e-05,
      "loss": 0.8424,
      "step": 219950
    },
    {
      "epoch": 3.8617251005108937,
      "grad_norm": 0.060807823931956204,
      "learning_rate": 2.721496939813922e-05,
      "loss": 0.8371,
      "step": 219960
    },
    {
      "epoch": 3.861900665390895,
      "grad_norm": 0.06649100002467083,
      "learning_rate": 2.7207540121210855e-05,
      "loss": 0.8387,
      "step": 219970
    },
    {
      "epoch": 3.8620762302708966,
      "grad_norm": 0.06519891249278198,
      "learning_rate": 2.720011177924705e-05,
      "loss": 0.8337,
      "step": 219980
    },
    {
      "epoch": 3.862251795150898,
      "grad_norm": 0.05232684192002305,
      "learning_rate": 2.719268437234203e-05,
      "loss": 0.8416,
      "step": 219990
    },
    {
      "epoch": 3.8624273600308996,
      "grad_norm": 0.05722671260921137,
      "learning_rate": 2.7185257900589815e-05,
      "loss": 0.8401,
      "step": 220000
    },
    {
      "epoch": 3.8626029249109006,
      "grad_norm": 0.053624587032104744,
      "learning_rate": 2.7177832364084556e-05,
      "loss": 0.8472,
      "step": 220010
    },
    {
      "epoch": 3.862778489790902,
      "grad_norm": 0.04881976284620709,
      "learning_rate": 2.717040776292028e-05,
      "loss": 0.8389,
      "step": 220020
    },
    {
      "epoch": 3.8629540546709036,
      "grad_norm": 0.047886141191370496,
      "learning_rate": 2.716298409719108e-05,
      "loss": 0.8393,
      "step": 220030
    },
    {
      "epoch": 3.863129619550905,
      "grad_norm": 0.05457806072062649,
      "learning_rate": 2.7155561366991056e-05,
      "loss": 0.8391,
      "step": 220040
    },
    {
      "epoch": 3.8633051844309065,
      "grad_norm": 0.04846938046021353,
      "learning_rate": 2.71481395724142e-05,
      "loss": 0.8368,
      "step": 220050
    },
    {
      "epoch": 3.8634807493109076,
      "grad_norm": 0.06532820578125792,
      "learning_rate": 2.7140718713554563e-05,
      "loss": 0.8357,
      "step": 220060
    },
    {
      "epoch": 3.8636563141909095,
      "grad_norm": 0.04458342277601709,
      "learning_rate": 2.7133298790506208e-05,
      "loss": 0.8447,
      "step": 220070
    },
    {
      "epoch": 3.8638318790709105,
      "grad_norm": 0.05969064847551637,
      "learning_rate": 2.712587980336308e-05,
      "loss": 0.8451,
      "step": 220080
    },
    {
      "epoch": 3.864007443950912,
      "grad_norm": 0.06443918176955588,
      "learning_rate": 2.7118461752219214e-05,
      "loss": 0.838,
      "step": 220090
    },
    {
      "epoch": 3.8641830088309135,
      "grad_norm": 0.06979883279936493,
      "learning_rate": 2.7111044637168597e-05,
      "loss": 0.839,
      "step": 220100
    },
    {
      "epoch": 3.864358573710915,
      "grad_norm": 0.05685488500775629,
      "learning_rate": 2.7103628458305237e-05,
      "loss": 0.8425,
      "step": 220110
    },
    {
      "epoch": 3.8645341385909164,
      "grad_norm": 0.046975027223178256,
      "learning_rate": 2.7096213215723038e-05,
      "loss": 0.8457,
      "step": 220120
    },
    {
      "epoch": 3.8647097034709175,
      "grad_norm": 0.05302184229414454,
      "learning_rate": 2.7088798909515967e-05,
      "loss": 0.8397,
      "step": 220130
    },
    {
      "epoch": 3.864885268350919,
      "grad_norm": 0.05951799153742412,
      "learning_rate": 2.7081385539778027e-05,
      "loss": 0.8345,
      "step": 220140
    },
    {
      "epoch": 3.8650608332309204,
      "grad_norm": 0.04060953475397635,
      "learning_rate": 2.7073973106603025e-05,
      "loss": 0.8469,
      "step": 220150
    },
    {
      "epoch": 3.865236398110922,
      "grad_norm": 0.05727645565186854,
      "learning_rate": 2.7066561610085027e-05,
      "loss": 0.8469,
      "step": 220160
    },
    {
      "epoch": 3.8654119629909234,
      "grad_norm": 0.06176063818797023,
      "learning_rate": 2.7059151050317815e-05,
      "loss": 0.8374,
      "step": 220170
    },
    {
      "epoch": 3.865587527870925,
      "grad_norm": 0.0673599264729277,
      "learning_rate": 2.7051741427395372e-05,
      "loss": 0.848,
      "step": 220180
    },
    {
      "epoch": 3.8657630927509263,
      "grad_norm": 0.05557940120785957,
      "learning_rate": 2.7044332741411505e-05,
      "loss": 0.8293,
      "step": 220190
    },
    {
      "epoch": 3.8659386576309274,
      "grad_norm": 0.046946939461686585,
      "learning_rate": 2.7036924992460116e-05,
      "loss": 0.844,
      "step": 220200
    },
    {
      "epoch": 3.866114222510929,
      "grad_norm": 0.047504131632685215,
      "learning_rate": 2.7029518180635087e-05,
      "loss": 0.8325,
      "step": 220210
    },
    {
      "epoch": 3.8662897873909303,
      "grad_norm": 0.055738167020618984,
      "learning_rate": 2.702211230603021e-05,
      "loss": 0.83,
      "step": 220220
    },
    {
      "epoch": 3.866465352270932,
      "grad_norm": 0.0497215717921726,
      "learning_rate": 2.7014707368739354e-05,
      "loss": 0.8347,
      "step": 220230
    },
    {
      "epoch": 3.8666409171509333,
      "grad_norm": 0.05931664951213601,
      "learning_rate": 2.7007303368856367e-05,
      "loss": 0.8345,
      "step": 220240
    },
    {
      "epoch": 3.8668164820309343,
      "grad_norm": 0.0484587104498967,
      "learning_rate": 2.6999900306474984e-05,
      "loss": 0.8435,
      "step": 220250
    },
    {
      "epoch": 3.866992046910936,
      "grad_norm": 0.05750181634594695,
      "learning_rate": 2.6992498181689063e-05,
      "loss": 0.8395,
      "step": 220260
    },
    {
      "epoch": 3.8671676117909373,
      "grad_norm": 0.05566113139043902,
      "learning_rate": 2.698509699459235e-05,
      "loss": 0.8353,
      "step": 220270
    },
    {
      "epoch": 3.8673431766709387,
      "grad_norm": 0.06630100486406547,
      "learning_rate": 2.6977696745278696e-05,
      "loss": 0.8492,
      "step": 220280
    },
    {
      "epoch": 3.86751874155094,
      "grad_norm": 0.052657499664860205,
      "learning_rate": 2.6970297433841775e-05,
      "loss": 0.8375,
      "step": 220290
    },
    {
      "epoch": 3.8676943064309417,
      "grad_norm": 0.07131768821310311,
      "learning_rate": 2.6962899060375363e-05,
      "loss": 0.8428,
      "step": 220300
    },
    {
      "epoch": 3.867869871310943,
      "grad_norm": 0.060320645301357646,
      "learning_rate": 2.695550162497325e-05,
      "loss": 0.8453,
      "step": 220310
    },
    {
      "epoch": 3.868045436190944,
      "grad_norm": 0.05187010362532609,
      "learning_rate": 2.6948105127729052e-05,
      "loss": 0.8418,
      "step": 220320
    },
    {
      "epoch": 3.8682210010709457,
      "grad_norm": 0.05962915842061145,
      "learning_rate": 2.6940709568736626e-05,
      "loss": 0.8318,
      "step": 220330
    },
    {
      "epoch": 3.868396565950947,
      "grad_norm": 0.046812808605633194,
      "learning_rate": 2.693331494808956e-05,
      "loss": 0.8361,
      "step": 220340
    },
    {
      "epoch": 3.8685721308309486,
      "grad_norm": 0.05518716822310761,
      "learning_rate": 2.6925921265881636e-05,
      "loss": 0.8399,
      "step": 220350
    },
    {
      "epoch": 3.86874769571095,
      "grad_norm": 0.04851631283382004,
      "learning_rate": 2.6918528522206437e-05,
      "loss": 0.8409,
      "step": 220360
    },
    {
      "epoch": 3.868923260590951,
      "grad_norm": 0.05270176661893151,
      "learning_rate": 2.6911136717157693e-05,
      "loss": 0.8351,
      "step": 220370
    },
    {
      "epoch": 3.8690988254709526,
      "grad_norm": 0.051455604797407885,
      "learning_rate": 2.6903745850829076e-05,
      "loss": 0.836,
      "step": 220380
    },
    {
      "epoch": 3.869274390350954,
      "grad_norm": 0.047157502840221924,
      "learning_rate": 2.6896355923314155e-05,
      "loss": 0.8383,
      "step": 220390
    },
    {
      "epoch": 3.8694499552309556,
      "grad_norm": 0.053858488258449314,
      "learning_rate": 2.6888966934706617e-05,
      "loss": 0.8421,
      "step": 220400
    },
    {
      "epoch": 3.869625520110957,
      "grad_norm": 0.0759310734525289,
      "learning_rate": 2.6881578885100102e-05,
      "loss": 0.8347,
      "step": 220410
    },
    {
      "epoch": 3.8698010849909585,
      "grad_norm": 0.04960780661828362,
      "learning_rate": 2.6874191774588154e-05,
      "loss": 0.8366,
      "step": 220420
    },
    {
      "epoch": 3.86997664987096,
      "grad_norm": 0.05689180925936284,
      "learning_rate": 2.68668056032644e-05,
      "loss": 0.8415,
      "step": 220430
    },
    {
      "epoch": 3.870152214750961,
      "grad_norm": 0.05229229612174201,
      "learning_rate": 2.6859420371222432e-05,
      "loss": 0.8459,
      "step": 220440
    },
    {
      "epoch": 3.8703277796309625,
      "grad_norm": 0.04504153012892235,
      "learning_rate": 2.6852036078555843e-05,
      "loss": 0.8313,
      "step": 220450
    },
    {
      "epoch": 3.870503344510964,
      "grad_norm": 0.04491783278185149,
      "learning_rate": 2.6844652725358144e-05,
      "loss": 0.8407,
      "step": 220460
    },
    {
      "epoch": 3.8706789093909655,
      "grad_norm": 0.04318533483295248,
      "learning_rate": 2.68372703117229e-05,
      "loss": 0.8406,
      "step": 220470
    },
    {
      "epoch": 3.870854474270967,
      "grad_norm": 0.04925671442926208,
      "learning_rate": 2.682988883774369e-05,
      "loss": 0.8364,
      "step": 220480
    },
    {
      "epoch": 3.871030039150968,
      "grad_norm": 0.05438872879104268,
      "learning_rate": 2.682250830351394e-05,
      "loss": 0.8425,
      "step": 220490
    },
    {
      "epoch": 3.87120560403097,
      "grad_norm": 0.05195200259622004,
      "learning_rate": 2.681512870912728e-05,
      "loss": 0.8412,
      "step": 220500
    },
    {
      "epoch": 3.871381168910971,
      "grad_norm": 0.046271007792777964,
      "learning_rate": 2.6807750054677133e-05,
      "loss": 0.8337,
      "step": 220510
    },
    {
      "epoch": 3.8715567337909724,
      "grad_norm": 0.06291780767024757,
      "learning_rate": 2.680037234025704e-05,
      "loss": 0.8401,
      "step": 220520
    },
    {
      "epoch": 3.871732298670974,
      "grad_norm": 0.09308963747220991,
      "learning_rate": 2.6792995565960417e-05,
      "loss": 0.8395,
      "step": 220530
    },
    {
      "epoch": 3.8719078635509754,
      "grad_norm": 0.058544086619930574,
      "learning_rate": 2.678561973188076e-05,
      "loss": 0.8385,
      "step": 220540
    },
    {
      "epoch": 3.872083428430977,
      "grad_norm": 0.061352725322756034,
      "learning_rate": 2.6778244838111567e-05,
      "loss": 0.8403,
      "step": 220550
    },
    {
      "epoch": 3.872258993310978,
      "grad_norm": 0.047951729834129,
      "learning_rate": 2.67708708847462e-05,
      "loss": 0.8401,
      "step": 220560
    },
    {
      "epoch": 3.8724345581909794,
      "grad_norm": 0.06538290401109517,
      "learning_rate": 2.6763497871878118e-05,
      "loss": 0.8261,
      "step": 220570
    },
    {
      "epoch": 3.872610123070981,
      "grad_norm": 0.055707354191190356,
      "learning_rate": 2.6756125799600777e-05,
      "loss": 0.8363,
      "step": 220580
    },
    {
      "epoch": 3.8727856879509823,
      "grad_norm": 0.05206706863538927,
      "learning_rate": 2.6748754668007538e-05,
      "loss": 0.8383,
      "step": 220590
    },
    {
      "epoch": 3.872961252830984,
      "grad_norm": 0.04362964080349281,
      "learning_rate": 2.6741384477191783e-05,
      "loss": 0.8388,
      "step": 220600
    },
    {
      "epoch": 3.873136817710985,
      "grad_norm": 0.05368436319273393,
      "learning_rate": 2.6734015227246935e-05,
      "loss": 0.8457,
      "step": 220610
    },
    {
      "epoch": 3.8733123825909868,
      "grad_norm": 0.05469917198355411,
      "learning_rate": 2.6726646918266373e-05,
      "loss": 0.8377,
      "step": 220620
    },
    {
      "epoch": 3.873487947470988,
      "grad_norm": 0.04865378306889885,
      "learning_rate": 2.67192795503434e-05,
      "loss": 0.8369,
      "step": 220630
    },
    {
      "epoch": 3.8736635123509893,
      "grad_norm": 0.052306257010666324,
      "learning_rate": 2.671191312357138e-05,
      "loss": 0.8357,
      "step": 220640
    },
    {
      "epoch": 3.8738390772309907,
      "grad_norm": 0.05326203132754801,
      "learning_rate": 2.67045476380437e-05,
      "loss": 0.8434,
      "step": 220650
    },
    {
      "epoch": 3.8740146421109922,
      "grad_norm": 0.04734024574186834,
      "learning_rate": 2.6697183093853584e-05,
      "loss": 0.8525,
      "step": 220660
    },
    {
      "epoch": 3.8741902069909937,
      "grad_norm": 0.059027279408854806,
      "learning_rate": 2.668981949109444e-05,
      "loss": 0.8464,
      "step": 220670
    },
    {
      "epoch": 3.8743657718709947,
      "grad_norm": 0.04677270543880885,
      "learning_rate": 2.668245682985951e-05,
      "loss": 0.8445,
      "step": 220680
    },
    {
      "epoch": 3.874541336750996,
      "grad_norm": 0.04985757364837736,
      "learning_rate": 2.6675095110242125e-05,
      "loss": 0.8344,
      "step": 220690
    },
    {
      "epoch": 3.8747169016309977,
      "grad_norm": 0.05519134261390024,
      "learning_rate": 2.6667734332335493e-05,
      "loss": 0.8427,
      "step": 220700
    },
    {
      "epoch": 3.874892466510999,
      "grad_norm": 0.05499054073448064,
      "learning_rate": 2.6660374496232916e-05,
      "loss": 0.8475,
      "step": 220710
    },
    {
      "epoch": 3.8750680313910006,
      "grad_norm": 0.04592358681721682,
      "learning_rate": 2.665301560202767e-05,
      "loss": 0.8294,
      "step": 220720
    },
    {
      "epoch": 3.8752435962710017,
      "grad_norm": 0.05098805163174184,
      "learning_rate": 2.6645657649812934e-05,
      "loss": 0.8386,
      "step": 220730
    },
    {
      "epoch": 3.8754191611510036,
      "grad_norm": 0.06009740109607623,
      "learning_rate": 2.6638300639681964e-05,
      "loss": 0.8402,
      "step": 220740
    },
    {
      "epoch": 3.8755947260310046,
      "grad_norm": 0.0473954147535542,
      "learning_rate": 2.6630944571728e-05,
      "loss": 0.8407,
      "step": 220750
    },
    {
      "epoch": 3.875770290911006,
      "grad_norm": 0.04444707969126709,
      "learning_rate": 2.6623589446044206e-05,
      "loss": 0.8383,
      "step": 220760
    },
    {
      "epoch": 3.8759458557910076,
      "grad_norm": 0.04205555989837212,
      "learning_rate": 2.6616235262723776e-05,
      "loss": 0.8361,
      "step": 220770
    },
    {
      "epoch": 3.876121420671009,
      "grad_norm": 0.06029900567834572,
      "learning_rate": 2.660888202185991e-05,
      "loss": 0.8456,
      "step": 220780
    },
    {
      "epoch": 3.8762969855510105,
      "grad_norm": 0.04441306795619205,
      "learning_rate": 2.6601529723545788e-05,
      "loss": 0.8334,
      "step": 220790
    },
    {
      "epoch": 3.8764725504310116,
      "grad_norm": 0.04723494867467694,
      "learning_rate": 2.6594178367874517e-05,
      "loss": 0.8376,
      "step": 220800
    },
    {
      "epoch": 3.876648115311013,
      "grad_norm": 0.05192946058457212,
      "learning_rate": 2.658682795493927e-05,
      "loss": 0.8341,
      "step": 220810
    },
    {
      "epoch": 3.8768236801910145,
      "grad_norm": 0.06264820941800259,
      "learning_rate": 2.6579478484833194e-05,
      "loss": 0.843,
      "step": 220820
    },
    {
      "epoch": 3.876999245071016,
      "grad_norm": 0.06964618847190272,
      "learning_rate": 2.657212995764934e-05,
      "loss": 0.8437,
      "step": 220830
    },
    {
      "epoch": 3.8771748099510175,
      "grad_norm": 0.04860131945867386,
      "learning_rate": 2.6564782373480907e-05,
      "loss": 0.8356,
      "step": 220840
    },
    {
      "epoch": 3.877350374831019,
      "grad_norm": 0.0926491811231367,
      "learning_rate": 2.6557435732420936e-05,
      "loss": 0.8399,
      "step": 220850
    },
    {
      "epoch": 3.8775259397110204,
      "grad_norm": 0.04546755759971942,
      "learning_rate": 2.6550090034562537e-05,
      "loss": 0.8497,
      "step": 220860
    },
    {
      "epoch": 3.8777015045910215,
      "grad_norm": 0.06411854006673885,
      "learning_rate": 2.6542745279998744e-05,
      "loss": 0.8375,
      "step": 220870
    },
    {
      "epoch": 3.877877069471023,
      "grad_norm": 0.049630875439706304,
      "learning_rate": 2.6535401468822636e-05,
      "loss": 0.8412,
      "step": 220880
    },
    {
      "epoch": 3.8780526343510244,
      "grad_norm": 0.045109660703627374,
      "learning_rate": 2.6528058601127288e-05,
      "loss": 0.8447,
      "step": 220890
    },
    {
      "epoch": 3.878228199231026,
      "grad_norm": 0.058932022756407736,
      "learning_rate": 2.6520716677005685e-05,
      "loss": 0.8339,
      "step": 220900
    },
    {
      "epoch": 3.8784037641110274,
      "grad_norm": 0.06875773026184422,
      "learning_rate": 2.6513375696550874e-05,
      "loss": 0.8395,
      "step": 220910
    },
    {
      "epoch": 3.8785793289910284,
      "grad_norm": 0.044494615665701426,
      "learning_rate": 2.6506035659855894e-05,
      "loss": 0.8442,
      "step": 220920
    },
    {
      "epoch": 3.87875489387103,
      "grad_norm": 0.053182608161193974,
      "learning_rate": 2.649869656701369e-05,
      "loss": 0.834,
      "step": 220930
    },
    {
      "epoch": 3.8789304587510314,
      "grad_norm": 0.051196599472405284,
      "learning_rate": 2.649135841811729e-05,
      "loss": 0.8399,
      "step": 220940
    },
    {
      "epoch": 3.879106023631033,
      "grad_norm": 0.05402838229673597,
      "learning_rate": 2.6484021213259647e-05,
      "loss": 0.8391,
      "step": 220950
    },
    {
      "epoch": 3.8792815885110343,
      "grad_norm": 0.0481260627499626,
      "learning_rate": 2.647668495253377e-05,
      "loss": 0.8418,
      "step": 220960
    },
    {
      "epoch": 3.879457153391036,
      "grad_norm": 0.061047032594105176,
      "learning_rate": 2.646934963603255e-05,
      "loss": 0.8305,
      "step": 220970
    },
    {
      "epoch": 3.8796327182710373,
      "grad_norm": 0.043041020409989986,
      "learning_rate": 2.6462015263848943e-05,
      "loss": 0.8372,
      "step": 220980
    },
    {
      "epoch": 3.8798082831510383,
      "grad_norm": 0.05964476306250617,
      "learning_rate": 2.6454681836075916e-05,
      "loss": 0.8433,
      "step": 220990
    },
    {
      "epoch": 3.87998384803104,
      "grad_norm": 0.04755558741769873,
      "learning_rate": 2.6447349352806316e-05,
      "loss": 0.8411,
      "step": 221000
    },
    {
      "epoch": 3.8801594129110413,
      "grad_norm": 0.05637797372418505,
      "learning_rate": 2.6440017814133128e-05,
      "loss": 0.8454,
      "step": 221010
    },
    {
      "epoch": 3.8803349777910427,
      "grad_norm": 0.07004730159323973,
      "learning_rate": 2.643268722014919e-05,
      "loss": 0.8332,
      "step": 221020
    },
    {
      "epoch": 3.8805105426710442,
      "grad_norm": 0.06671428449294227,
      "learning_rate": 2.6425357570947412e-05,
      "loss": 0.8439,
      "step": 221030
    },
    {
      "epoch": 3.8806861075510453,
      "grad_norm": 0.04745360962211555,
      "learning_rate": 2.641802886662063e-05,
      "loss": 0.8388,
      "step": 221040
    },
    {
      "epoch": 3.8808616724310467,
      "grad_norm": 0.06372545258433138,
      "learning_rate": 2.6410701107261697e-05,
      "loss": 0.8366,
      "step": 221050
    },
    {
      "epoch": 3.881037237311048,
      "grad_norm": 0.04380193423821364,
      "learning_rate": 2.6403374292963524e-05,
      "loss": 0.854,
      "step": 221060
    },
    {
      "epoch": 3.8812128021910497,
      "grad_norm": 0.059633111697103225,
      "learning_rate": 2.6396048423818855e-05,
      "loss": 0.8312,
      "step": 221070
    },
    {
      "epoch": 3.881388367071051,
      "grad_norm": 0.05596707411411138,
      "learning_rate": 2.6388723499920553e-05,
      "loss": 0.8362,
      "step": 221080
    },
    {
      "epoch": 3.8815639319510526,
      "grad_norm": 0.05141994283552374,
      "learning_rate": 2.638139952136145e-05,
      "loss": 0.8346,
      "step": 221090
    },
    {
      "epoch": 3.881739496831054,
      "grad_norm": 0.044690151323756394,
      "learning_rate": 2.6374076488234283e-05,
      "loss": 0.8478,
      "step": 221100
    },
    {
      "epoch": 3.881915061711055,
      "grad_norm": 0.05264210470411027,
      "learning_rate": 2.6366754400631887e-05,
      "loss": 0.8343,
      "step": 221110
    },
    {
      "epoch": 3.8820906265910566,
      "grad_norm": 0.04951642373798584,
      "learning_rate": 2.6359433258647e-05,
      "loss": 0.8371,
      "step": 221120
    },
    {
      "epoch": 3.882266191471058,
      "grad_norm": 0.05714658635520264,
      "learning_rate": 2.6352113062372437e-05,
      "loss": 0.8377,
      "step": 221130
    },
    {
      "epoch": 3.8824417563510596,
      "grad_norm": 0.04611798068470271,
      "learning_rate": 2.634479381190087e-05,
      "loss": 0.8461,
      "step": 221140
    },
    {
      "epoch": 3.882617321231061,
      "grad_norm": 0.05585648113663818,
      "learning_rate": 2.6337475507325073e-05,
      "loss": 0.843,
      "step": 221150
    },
    {
      "epoch": 3.882792886111062,
      "grad_norm": 0.04890887846388081,
      "learning_rate": 2.6330158148737806e-05,
      "loss": 0.8414,
      "step": 221160
    },
    {
      "epoch": 3.882968450991064,
      "grad_norm": 0.050848528885300844,
      "learning_rate": 2.6322841736231683e-05,
      "loss": 0.8377,
      "step": 221170
    },
    {
      "epoch": 3.883144015871065,
      "grad_norm": 0.04857309997621667,
      "learning_rate": 2.631552626989952e-05,
      "loss": 0.8399,
      "step": 221180
    },
    {
      "epoch": 3.8833195807510665,
      "grad_norm": 0.05051736294807755,
      "learning_rate": 2.6308211749833935e-05,
      "loss": 0.8462,
      "step": 221190
    },
    {
      "epoch": 3.883495145631068,
      "grad_norm": 0.048538715411217896,
      "learning_rate": 2.630089817612765e-05,
      "loss": 0.8289,
      "step": 221200
    },
    {
      "epoch": 3.8836707105110695,
      "grad_norm": 0.04993145268228837,
      "learning_rate": 2.6293585548873268e-05,
      "loss": 0.838,
      "step": 221210
    },
    {
      "epoch": 3.883846275391071,
      "grad_norm": 0.05100416921850043,
      "learning_rate": 2.628627386816347e-05,
      "loss": 0.8421,
      "step": 221220
    },
    {
      "epoch": 3.884021840271072,
      "grad_norm": 0.067147655005026,
      "learning_rate": 2.627896313409094e-05,
      "loss": 0.8395,
      "step": 221230
    },
    {
      "epoch": 3.8841974051510735,
      "grad_norm": 0.05460481134926325,
      "learning_rate": 2.627165334674823e-05,
      "loss": 0.841,
      "step": 221240
    },
    {
      "epoch": 3.884372970031075,
      "grad_norm": 0.0880934572400214,
      "learning_rate": 2.6264344506227993e-05,
      "loss": 0.8454,
      "step": 221250
    },
    {
      "epoch": 3.8845485349110764,
      "grad_norm": 0.045851070784151106,
      "learning_rate": 2.6257036612622865e-05,
      "loss": 0.8455,
      "step": 221260
    },
    {
      "epoch": 3.884724099791078,
      "grad_norm": 0.05666348989771427,
      "learning_rate": 2.624972966602538e-05,
      "loss": 0.8382,
      "step": 221270
    },
    {
      "epoch": 3.884899664671079,
      "grad_norm": 0.05696527513668593,
      "learning_rate": 2.624242366652815e-05,
      "loss": 0.8364,
      "step": 221280
    },
    {
      "epoch": 3.885075229551081,
      "grad_norm": 0.06191582764334084,
      "learning_rate": 2.623511861422373e-05,
      "loss": 0.8373,
      "step": 221290
    },
    {
      "epoch": 3.885250794431082,
      "grad_norm": 0.05954732362202053,
      "learning_rate": 2.6227814509204717e-05,
      "loss": 0.8392,
      "step": 221300
    },
    {
      "epoch": 3.8854263593110834,
      "grad_norm": 0.05405977265516696,
      "learning_rate": 2.6220511351563604e-05,
      "loss": 0.8409,
      "step": 221310
    },
    {
      "epoch": 3.885601924191085,
      "grad_norm": 0.04516498340521064,
      "learning_rate": 2.6213209141392936e-05,
      "loss": 0.8391,
      "step": 221320
    },
    {
      "epoch": 3.8857774890710863,
      "grad_norm": 0.04938668422737259,
      "learning_rate": 2.6205907878785273e-05,
      "loss": 0.8428,
      "step": 221330
    },
    {
      "epoch": 3.885953053951088,
      "grad_norm": 0.05082432603714264,
      "learning_rate": 2.6198607563833034e-05,
      "loss": 0.8345,
      "step": 221340
    },
    {
      "epoch": 3.886128618831089,
      "grad_norm": 0.05095689770716723,
      "learning_rate": 2.6191308196628822e-05,
      "loss": 0.8443,
      "step": 221350
    },
    {
      "epoch": 3.8863041837110903,
      "grad_norm": 0.061100295854960476,
      "learning_rate": 2.618400977726505e-05,
      "loss": 0.8338,
      "step": 221360
    },
    {
      "epoch": 3.886479748591092,
      "grad_norm": 0.0630951503872218,
      "learning_rate": 2.6176712305834245e-05,
      "loss": 0.8395,
      "step": 221370
    },
    {
      "epoch": 3.8866553134710933,
      "grad_norm": 0.058124228747362235,
      "learning_rate": 2.61694157824288e-05,
      "loss": 0.8415,
      "step": 221380
    },
    {
      "epoch": 3.8868308783510948,
      "grad_norm": 0.06765894808491713,
      "learning_rate": 2.6162120207141205e-05,
      "loss": 0.8449,
      "step": 221390
    },
    {
      "epoch": 3.887006443231096,
      "grad_norm": 0.05702866046239002,
      "learning_rate": 2.6154825580063918e-05,
      "loss": 0.8373,
      "step": 221400
    },
    {
      "epoch": 3.8871820081110977,
      "grad_norm": 0.12672473007735027,
      "learning_rate": 2.6147531901289302e-05,
      "loss": 0.8436,
      "step": 221410
    },
    {
      "epoch": 3.8873575729910987,
      "grad_norm": 0.048531262490116134,
      "learning_rate": 2.6140239170909796e-05,
      "loss": 0.8376,
      "step": 221420
    },
    {
      "epoch": 3.8875331378711,
      "grad_norm": 0.0588763928253645,
      "learning_rate": 2.6132947389017842e-05,
      "loss": 0.8384,
      "step": 221430
    },
    {
      "epoch": 3.8877087027511017,
      "grad_norm": 0.04918391462975673,
      "learning_rate": 2.6125656555705763e-05,
      "loss": 0.842,
      "step": 221440
    },
    {
      "epoch": 3.887884267631103,
      "grad_norm": 0.06437199137155032,
      "learning_rate": 2.611836667106597e-05,
      "loss": 0.8457,
      "step": 221450
    },
    {
      "epoch": 3.8880598325111047,
      "grad_norm": 0.04837937037341058,
      "learning_rate": 2.6111077735190814e-05,
      "loss": 0.8403,
      "step": 221460
    },
    {
      "epoch": 3.8882353973911057,
      "grad_norm": 0.056199380030482145,
      "learning_rate": 2.6103789748172695e-05,
      "loss": 0.8324,
      "step": 221470
    },
    {
      "epoch": 3.888410962271107,
      "grad_norm": 0.06764731318739224,
      "learning_rate": 2.609650271010387e-05,
      "loss": 0.8376,
      "step": 221480
    },
    {
      "epoch": 3.8885865271511086,
      "grad_norm": 0.06471251520020949,
      "learning_rate": 2.608921662107672e-05,
      "loss": 0.843,
      "step": 221490
    },
    {
      "epoch": 3.88876209203111,
      "grad_norm": 0.042444448896490954,
      "learning_rate": 2.608193148118358e-05,
      "loss": 0.8382,
      "step": 221500
    },
    {
      "epoch": 3.8889376569111116,
      "grad_norm": 0.05601594057171357,
      "learning_rate": 2.6074647290516673e-05,
      "loss": 0.8416,
      "step": 221510
    },
    {
      "epoch": 3.8891132217911126,
      "grad_norm": 0.06816686113507768,
      "learning_rate": 2.6067364049168398e-05,
      "loss": 0.84,
      "step": 221520
    },
    {
      "epoch": 3.8892887866711146,
      "grad_norm": 0.04383837233685804,
      "learning_rate": 2.6060081757230954e-05,
      "loss": 0.8311,
      "step": 221530
    },
    {
      "epoch": 3.8894643515511156,
      "grad_norm": 0.0450656091452457,
      "learning_rate": 2.605280041479665e-05,
      "loss": 0.8328,
      "step": 221540
    },
    {
      "epoch": 3.889639916431117,
      "grad_norm": 0.0571600267638123,
      "learning_rate": 2.6045520021957724e-05,
      "loss": 0.8281,
      "step": 221550
    },
    {
      "epoch": 3.8898154813111185,
      "grad_norm": 0.05065484279452021,
      "learning_rate": 2.6038240578806412e-05,
      "loss": 0.8422,
      "step": 221560
    },
    {
      "epoch": 3.88999104619112,
      "grad_norm": 0.04442978672185212,
      "learning_rate": 2.6030962085434983e-05,
      "loss": 0.8306,
      "step": 221570
    },
    {
      "epoch": 3.8901666110711215,
      "grad_norm": 0.05425915897445851,
      "learning_rate": 2.6023684541935612e-05,
      "loss": 0.8331,
      "step": 221580
    },
    {
      "epoch": 3.8903421759511225,
      "grad_norm": 0.08008081813327972,
      "learning_rate": 2.6016407948400524e-05,
      "loss": 0.8296,
      "step": 221590
    },
    {
      "epoch": 3.890517740831124,
      "grad_norm": 0.05158280917298065,
      "learning_rate": 2.600913230492194e-05,
      "loss": 0.8337,
      "step": 221600
    },
    {
      "epoch": 3.8906933057111255,
      "grad_norm": 0.04323987843528949,
      "learning_rate": 2.6001857611592e-05,
      "loss": 0.8402,
      "step": 221610
    },
    {
      "epoch": 3.890868870591127,
      "grad_norm": 0.05789280742983715,
      "learning_rate": 2.599458386850289e-05,
      "loss": 0.8432,
      "step": 221620
    },
    {
      "epoch": 3.8910444354711284,
      "grad_norm": 0.05888907422715911,
      "learning_rate": 2.598731107574679e-05,
      "loss": 0.837,
      "step": 221630
    },
    {
      "epoch": 3.89122000035113,
      "grad_norm": 0.08355688057742996,
      "learning_rate": 2.5980039233415848e-05,
      "loss": 0.8407,
      "step": 221640
    },
    {
      "epoch": 3.8913955652311314,
      "grad_norm": 0.05752788501927827,
      "learning_rate": 2.5972768341602176e-05,
      "loss": 0.8408,
      "step": 221650
    },
    {
      "epoch": 3.8915711301111324,
      "grad_norm": 0.08106848103639307,
      "learning_rate": 2.5965498400397897e-05,
      "loss": 0.8398,
      "step": 221660
    },
    {
      "epoch": 3.891746694991134,
      "grad_norm": 0.0463938259909158,
      "learning_rate": 2.595822940989516e-05,
      "loss": 0.8371,
      "step": 221670
    },
    {
      "epoch": 3.8919222598711354,
      "grad_norm": 0.054462707500172354,
      "learning_rate": 2.5950961370185984e-05,
      "loss": 0.8396,
      "step": 221680
    },
    {
      "epoch": 3.892097824751137,
      "grad_norm": 0.07357944811362693,
      "learning_rate": 2.5943694281362574e-05,
      "loss": 0.8431,
      "step": 221690
    },
    {
      "epoch": 3.8922733896311383,
      "grad_norm": 0.04883245922076961,
      "learning_rate": 2.5936428143516907e-05,
      "loss": 0.8365,
      "step": 221700
    },
    {
      "epoch": 3.8924489545111394,
      "grad_norm": 0.05678096960201468,
      "learning_rate": 2.59291629567411e-05,
      "loss": 0.8431,
      "step": 221710
    },
    {
      "epoch": 3.892624519391141,
      "grad_norm": 0.05769700335055701,
      "learning_rate": 2.5921898721127176e-05,
      "loss": 0.8377,
      "step": 221720
    },
    {
      "epoch": 3.8928000842711423,
      "grad_norm": 0.05388362650624751,
      "learning_rate": 2.5914635436767176e-05,
      "loss": 0.8464,
      "step": 221730
    },
    {
      "epoch": 3.892975649151144,
      "grad_norm": 0.04721682633638237,
      "learning_rate": 2.5907373103753162e-05,
      "loss": 0.8365,
      "step": 221740
    },
    {
      "epoch": 3.8931512140311453,
      "grad_norm": 0.05714723778255929,
      "learning_rate": 2.5900111722177095e-05,
      "loss": 0.841,
      "step": 221750
    },
    {
      "epoch": 3.8933267789111468,
      "grad_norm": 0.0496341538276018,
      "learning_rate": 2.5892851292130998e-05,
      "loss": 0.838,
      "step": 221760
    },
    {
      "epoch": 3.8935023437911482,
      "grad_norm": 0.05330902098541864,
      "learning_rate": 2.588559181370691e-05,
      "loss": 0.8342,
      "step": 221770
    },
    {
      "epoch": 3.8936779086711493,
      "grad_norm": 0.061977449480415066,
      "learning_rate": 2.587833328699673e-05,
      "loss": 0.8359,
      "step": 221780
    },
    {
      "epoch": 3.8938534735511507,
      "grad_norm": 0.06436290876382533,
      "learning_rate": 2.5871075712092473e-05,
      "loss": 0.8385,
      "step": 221790
    },
    {
      "epoch": 3.8940290384311522,
      "grad_norm": 0.05687975192158918,
      "learning_rate": 2.5863819089086077e-05,
      "loss": 0.8437,
      "step": 221800
    },
    {
      "epoch": 3.8942046033111537,
      "grad_norm": 0.053255077985922133,
      "learning_rate": 2.5856563418069528e-05,
      "loss": 0.84,
      "step": 221810
    },
    {
      "epoch": 3.894380168191155,
      "grad_norm": 0.046433818341630954,
      "learning_rate": 2.5849308699134688e-05,
      "loss": 0.8388,
      "step": 221820
    },
    {
      "epoch": 3.894555733071156,
      "grad_norm": 0.05872615443950612,
      "learning_rate": 2.5842054932373518e-05,
      "loss": 0.8366,
      "step": 221830
    },
    {
      "epoch": 3.8947312979511577,
      "grad_norm": 0.06381822892035627,
      "learning_rate": 2.5834802117877937e-05,
      "loss": 0.8341,
      "step": 221840
    },
    {
      "epoch": 3.894906862831159,
      "grad_norm": 0.045807874902925406,
      "learning_rate": 2.582755025573976e-05,
      "loss": 0.8443,
      "step": 221850
    },
    {
      "epoch": 3.8950824277111606,
      "grad_norm": 0.05483373104433313,
      "learning_rate": 2.5820299346050992e-05,
      "loss": 0.8407,
      "step": 221860
    },
    {
      "epoch": 3.895257992591162,
      "grad_norm": 0.050293626507445933,
      "learning_rate": 2.5813049388903417e-05,
      "loss": 0.8336,
      "step": 221870
    },
    {
      "epoch": 3.8954335574711636,
      "grad_norm": 0.05247647076160996,
      "learning_rate": 2.5805800384388925e-05,
      "loss": 0.8333,
      "step": 221880
    },
    {
      "epoch": 3.895609122351165,
      "grad_norm": 0.051838947019950135,
      "learning_rate": 2.579855233259934e-05,
      "loss": 0.8427,
      "step": 221890
    },
    {
      "epoch": 3.895784687231166,
      "grad_norm": 0.056035980815287226,
      "learning_rate": 2.57913052336265e-05,
      "loss": 0.8417,
      "step": 221900
    },
    {
      "epoch": 3.8959602521111676,
      "grad_norm": 0.049111023638129504,
      "learning_rate": 2.578405908756227e-05,
      "loss": 0.8372,
      "step": 221910
    },
    {
      "epoch": 3.896135816991169,
      "grad_norm": 0.04513243291012058,
      "learning_rate": 2.5776813894498393e-05,
      "loss": 0.844,
      "step": 221920
    },
    {
      "epoch": 3.8963113818711705,
      "grad_norm": 0.04688634761483277,
      "learning_rate": 2.5769569654526705e-05,
      "loss": 0.8417,
      "step": 221930
    },
    {
      "epoch": 3.896486946751172,
      "grad_norm": 0.049516552876564784,
      "learning_rate": 2.576232636773901e-05,
      "loss": 0.8367,
      "step": 221940
    },
    {
      "epoch": 3.896662511631173,
      "grad_norm": 0.050403581731815285,
      "learning_rate": 2.575508403422704e-05,
      "loss": 0.8436,
      "step": 221950
    },
    {
      "epoch": 3.896838076511175,
      "grad_norm": 0.05665490476461767,
      "learning_rate": 2.574784265408257e-05,
      "loss": 0.8382,
      "step": 221960
    },
    {
      "epoch": 3.897013641391176,
      "grad_norm": 0.05407461198483254,
      "learning_rate": 2.574060222739736e-05,
      "loss": 0.839,
      "step": 221970
    },
    {
      "epoch": 3.8971892062711775,
      "grad_norm": 0.05317010732949234,
      "learning_rate": 2.5733362754263167e-05,
      "loss": 0.8364,
      "step": 221980
    },
    {
      "epoch": 3.897364771151179,
      "grad_norm": 0.048500308357381006,
      "learning_rate": 2.5726124234771664e-05,
      "loss": 0.8475,
      "step": 221990
    },
    {
      "epoch": 3.8975403360311804,
      "grad_norm": 0.05095050383443403,
      "learning_rate": 2.5718886669014598e-05,
      "loss": 0.8428,
      "step": 222000
    },
    {
      "epoch": 3.897715900911182,
      "grad_norm": 0.05506128428183534,
      "learning_rate": 2.571165005708369e-05,
      "loss": 0.8376,
      "step": 222010
    },
    {
      "epoch": 3.897891465791183,
      "grad_norm": 0.04815780362998678,
      "learning_rate": 2.5704414399070543e-05,
      "loss": 0.8293,
      "step": 222020
    },
    {
      "epoch": 3.8980670306711844,
      "grad_norm": 0.05019755258429041,
      "learning_rate": 2.5697179695066964e-05,
      "loss": 0.8395,
      "step": 222030
    },
    {
      "epoch": 3.898242595551186,
      "grad_norm": 0.06931072562208648,
      "learning_rate": 2.5689945945164512e-05,
      "loss": 0.8388,
      "step": 222040
    },
    {
      "epoch": 3.8984181604311874,
      "grad_norm": 0.06027425320331265,
      "learning_rate": 2.5682713149454918e-05,
      "loss": 0.8395,
      "step": 222050
    },
    {
      "epoch": 3.898593725311189,
      "grad_norm": 0.0433124535990823,
      "learning_rate": 2.5675481308029735e-05,
      "loss": 0.8357,
      "step": 222060
    },
    {
      "epoch": 3.89876929019119,
      "grad_norm": 0.04423496727264816,
      "learning_rate": 2.566825042098065e-05,
      "loss": 0.8353,
      "step": 222070
    },
    {
      "epoch": 3.898944855071192,
      "grad_norm": 0.06260774065816824,
      "learning_rate": 2.56610204883993e-05,
      "loss": 0.836,
      "step": 222080
    },
    {
      "epoch": 3.899120419951193,
      "grad_norm": 0.04151114340919156,
      "learning_rate": 2.565379151037724e-05,
      "loss": 0.8324,
      "step": 222090
    },
    {
      "epoch": 3.8992959848311943,
      "grad_norm": 0.0723105555101005,
      "learning_rate": 2.564656348700607e-05,
      "loss": 0.8362,
      "step": 222100
    },
    {
      "epoch": 3.899471549711196,
      "grad_norm": 0.060377660018217975,
      "learning_rate": 2.5639336418377408e-05,
      "loss": 0.8403,
      "step": 222110
    },
    {
      "epoch": 3.8996471145911973,
      "grad_norm": 0.04742296464886442,
      "learning_rate": 2.5632110304582767e-05,
      "loss": 0.8463,
      "step": 222120
    },
    {
      "epoch": 3.8998226794711988,
      "grad_norm": 0.05569296182646124,
      "learning_rate": 2.5624885145713745e-05,
      "loss": 0.8412,
      "step": 222130
    },
    {
      "epoch": 3.8999982443512,
      "grad_norm": 0.05399414827649697,
      "learning_rate": 2.561766094186187e-05,
      "loss": 0.8423,
      "step": 222140
    },
    {
      "epoch": 3.9001738092312013,
      "grad_norm": 0.04555694879220789,
      "learning_rate": 2.5610437693118695e-05,
      "loss": 0.828,
      "step": 222150
    },
    {
      "epoch": 3.9003493741112027,
      "grad_norm": 0.0508733229572191,
      "learning_rate": 2.5603215399575697e-05,
      "loss": 0.8319,
      "step": 222160
    },
    {
      "epoch": 3.9005249389912042,
      "grad_norm": 0.05312127686073041,
      "learning_rate": 2.559599406132442e-05,
      "loss": 0.8407,
      "step": 222170
    },
    {
      "epoch": 3.9007005038712057,
      "grad_norm": 0.0504648945768219,
      "learning_rate": 2.558877367845637e-05,
      "loss": 0.8284,
      "step": 222180
    },
    {
      "epoch": 3.9008760687512067,
      "grad_norm": 0.05516512839154054,
      "learning_rate": 2.558155425106296e-05,
      "loss": 0.8452,
      "step": 222190
    },
    {
      "epoch": 3.9010516336312087,
      "grad_norm": 0.052836307774292275,
      "learning_rate": 2.5574335779235765e-05,
      "loss": 0.8409,
      "step": 222200
    },
    {
      "epoch": 3.9012271985112097,
      "grad_norm": 0.06100688053162658,
      "learning_rate": 2.5567118263066154e-05,
      "loss": 0.8473,
      "step": 222210
    },
    {
      "epoch": 3.901402763391211,
      "grad_norm": 0.06505172320941134,
      "learning_rate": 2.555990170264565e-05,
      "loss": 0.8424,
      "step": 222220
    },
    {
      "epoch": 3.9015783282712126,
      "grad_norm": 0.060319490356756524,
      "learning_rate": 2.555268609806562e-05,
      "loss": 0.8381,
      "step": 222230
    },
    {
      "epoch": 3.901753893151214,
      "grad_norm": 0.053201003892375844,
      "learning_rate": 2.5545471449417513e-05,
      "loss": 0.8426,
      "step": 222240
    },
    {
      "epoch": 3.9019294580312156,
      "grad_norm": 0.04524932792105024,
      "learning_rate": 2.5538257756792772e-05,
      "loss": 0.8461,
      "step": 222250
    },
    {
      "epoch": 3.9021050229112166,
      "grad_norm": 0.05949112376619402,
      "learning_rate": 2.5531045020282746e-05,
      "loss": 0.8432,
      "step": 222260
    },
    {
      "epoch": 3.902280587791218,
      "grad_norm": 0.05861888546190387,
      "learning_rate": 2.5523833239978828e-05,
      "loss": 0.8323,
      "step": 222270
    },
    {
      "epoch": 3.9024561526712196,
      "grad_norm": 0.04570210707248754,
      "learning_rate": 2.5516622415972452e-05,
      "loss": 0.8404,
      "step": 222280
    },
    {
      "epoch": 3.902631717551221,
      "grad_norm": 0.05172482949654708,
      "learning_rate": 2.5509412548354906e-05,
      "loss": 0.836,
      "step": 222290
    },
    {
      "epoch": 3.9028072824312225,
      "grad_norm": 0.06895212495911308,
      "learning_rate": 2.550220363721756e-05,
      "loss": 0.8348,
      "step": 222300
    },
    {
      "epoch": 3.902982847311224,
      "grad_norm": 0.04647172875207578,
      "learning_rate": 2.549499568265177e-05,
      "loss": 0.8365,
      "step": 222310
    },
    {
      "epoch": 3.9031584121912255,
      "grad_norm": 0.06002378225453959,
      "learning_rate": 2.5487788684748882e-05,
      "loss": 0.8374,
      "step": 222320
    },
    {
      "epoch": 3.9033339770712265,
      "grad_norm": 0.047296776128430765,
      "learning_rate": 2.5480582643600155e-05,
      "loss": 0.8452,
      "step": 222330
    },
    {
      "epoch": 3.903509541951228,
      "grad_norm": 0.04788747362780445,
      "learning_rate": 2.5473377559296925e-05,
      "loss": 0.8467,
      "step": 222340
    },
    {
      "epoch": 3.9036851068312295,
      "grad_norm": 0.07831503650157705,
      "learning_rate": 2.5466173431930502e-05,
      "loss": 0.8412,
      "step": 222350
    },
    {
      "epoch": 3.903860671711231,
      "grad_norm": 0.06050165277749668,
      "learning_rate": 2.5458970261592074e-05,
      "loss": 0.8366,
      "step": 222360
    },
    {
      "epoch": 3.9040362365912324,
      "grad_norm": 0.039923632348503615,
      "learning_rate": 2.545176804837303e-05,
      "loss": 0.8393,
      "step": 222370
    },
    {
      "epoch": 3.9042118014712335,
      "grad_norm": 0.048646008894863674,
      "learning_rate": 2.5444566792364533e-05,
      "loss": 0.8467,
      "step": 222380
    },
    {
      "epoch": 3.904387366351235,
      "grad_norm": 0.05671811778747898,
      "learning_rate": 2.5437366493657888e-05,
      "loss": 0.8407,
      "step": 222390
    },
    {
      "epoch": 3.9045629312312364,
      "grad_norm": 0.045406767968990236,
      "learning_rate": 2.5430167152344265e-05,
      "loss": 0.8359,
      "step": 222400
    },
    {
      "epoch": 3.904738496111238,
      "grad_norm": 0.052824139107193115,
      "learning_rate": 2.54229687685149e-05,
      "loss": 0.8364,
      "step": 222410
    },
    {
      "epoch": 3.9049140609912394,
      "grad_norm": 0.052288614124564814,
      "learning_rate": 2.5415771342261042e-05,
      "loss": 0.8363,
      "step": 222420
    },
    {
      "epoch": 3.905089625871241,
      "grad_norm": 0.06532166439578808,
      "learning_rate": 2.540857487367381e-05,
      "loss": 0.8438,
      "step": 222430
    },
    {
      "epoch": 3.9052651907512423,
      "grad_norm": 0.059699475300794745,
      "learning_rate": 2.5401379362844428e-05,
      "loss": 0.842,
      "step": 222440
    },
    {
      "epoch": 3.9054407556312434,
      "grad_norm": 0.05570751724812584,
      "learning_rate": 2.539418480986408e-05,
      "loss": 0.8426,
      "step": 222450
    },
    {
      "epoch": 3.905616320511245,
      "grad_norm": 0.051137068739079465,
      "learning_rate": 2.5386991214823874e-05,
      "loss": 0.8412,
      "step": 222460
    },
    {
      "epoch": 3.9057918853912463,
      "grad_norm": 0.04394206565820816,
      "learning_rate": 2.537979857781498e-05,
      "loss": 0.8466,
      "step": 222470
    },
    {
      "epoch": 3.905967450271248,
      "grad_norm": 0.0947206603928735,
      "learning_rate": 2.537260689892853e-05,
      "loss": 0.8416,
      "step": 222480
    },
    {
      "epoch": 3.9061430151512493,
      "grad_norm": 0.03911333069736279,
      "learning_rate": 2.5365416178255675e-05,
      "loss": 0.8398,
      "step": 222490
    },
    {
      "epoch": 3.9063185800312503,
      "grad_norm": 0.048972361473956005,
      "learning_rate": 2.535822641588747e-05,
      "loss": 0.8361,
      "step": 222500
    },
    {
      "epoch": 3.906494144911252,
      "grad_norm": 0.06554427823946492,
      "learning_rate": 2.535103761191502e-05,
      "loss": 0.8405,
      "step": 222510
    },
    {
      "epoch": 3.9066697097912533,
      "grad_norm": 0.05195906888506268,
      "learning_rate": 2.5343849766429455e-05,
      "loss": 0.8381,
      "step": 222520
    },
    {
      "epoch": 3.9068452746712548,
      "grad_norm": 0.06238015094059804,
      "learning_rate": 2.5336662879521757e-05,
      "loss": 0.8383,
      "step": 222530
    },
    {
      "epoch": 3.9070208395512562,
      "grad_norm": 0.04952976044438362,
      "learning_rate": 2.53294769512831e-05,
      "loss": 0.8314,
      "step": 222540
    },
    {
      "epoch": 3.9071964044312577,
      "grad_norm": 0.04559724506494129,
      "learning_rate": 2.5322291981804443e-05,
      "loss": 0.8413,
      "step": 222550
    },
    {
      "epoch": 3.907371969311259,
      "grad_norm": 0.052456440618128715,
      "learning_rate": 2.5315107971176878e-05,
      "loss": 0.8428,
      "step": 222560
    },
    {
      "epoch": 3.90754753419126,
      "grad_norm": 0.04568018729326008,
      "learning_rate": 2.5307924919491372e-05,
      "loss": 0.8341,
      "step": 222570
    },
    {
      "epoch": 3.9077230990712617,
      "grad_norm": 0.05644382600402056,
      "learning_rate": 2.5300742826838958e-05,
      "loss": 0.8417,
      "step": 222580
    },
    {
      "epoch": 3.907898663951263,
      "grad_norm": 0.04794315833333682,
      "learning_rate": 2.5293561693310675e-05,
      "loss": 0.8426,
      "step": 222590
    },
    {
      "epoch": 3.9080742288312647,
      "grad_norm": 0.057916646751041675,
      "learning_rate": 2.5286381518997443e-05,
      "loss": 0.8392,
      "step": 222600
    },
    {
      "epoch": 3.908249793711266,
      "grad_norm": 0.05250066924167504,
      "learning_rate": 2.5279202303990274e-05,
      "loss": 0.8342,
      "step": 222610
    },
    {
      "epoch": 3.908425358591267,
      "grad_norm": 0.053054903836349705,
      "learning_rate": 2.5272024048380137e-05,
      "loss": 0.8372,
      "step": 222620
    },
    {
      "epoch": 3.908600923471269,
      "grad_norm": 0.05210758544404214,
      "learning_rate": 2.5264846752257955e-05,
      "loss": 0.8456,
      "step": 222630
    },
    {
      "epoch": 3.90877648835127,
      "grad_norm": 0.04049142517473293,
      "learning_rate": 2.525767041571467e-05,
      "loss": 0.836,
      "step": 222640
    },
    {
      "epoch": 3.9089520532312716,
      "grad_norm": 0.04610985032062517,
      "learning_rate": 2.5250495038841232e-05,
      "loss": 0.8438,
      "step": 222650
    },
    {
      "epoch": 3.909127618111273,
      "grad_norm": 0.05750822127642141,
      "learning_rate": 2.5243320621728557e-05,
      "loss": 0.8414,
      "step": 222660
    },
    {
      "epoch": 3.9093031829912746,
      "grad_norm": 0.07489920738508794,
      "learning_rate": 2.5236147164467495e-05,
      "loss": 0.8352,
      "step": 222670
    },
    {
      "epoch": 3.909478747871276,
      "grad_norm": 0.04973911310201513,
      "learning_rate": 2.5228974667148968e-05,
      "loss": 0.8442,
      "step": 222680
    },
    {
      "epoch": 3.909654312751277,
      "grad_norm": 0.05761714179913874,
      "learning_rate": 2.5221803129863884e-05,
      "loss": 0.8382,
      "step": 222690
    },
    {
      "epoch": 3.9098298776312785,
      "grad_norm": 0.05798019315246162,
      "learning_rate": 2.5214632552703027e-05,
      "loss": 0.843,
      "step": 222700
    },
    {
      "epoch": 3.91000544251128,
      "grad_norm": 0.05790663208551657,
      "learning_rate": 2.520746293575736e-05,
      "loss": 0.8355,
      "step": 222710
    },
    {
      "epoch": 3.9101810073912815,
      "grad_norm": 0.051900359291227535,
      "learning_rate": 2.520029427911762e-05,
      "loss": 0.8431,
      "step": 222720
    },
    {
      "epoch": 3.910356572271283,
      "grad_norm": 0.0532603819385904,
      "learning_rate": 2.5193126582874717e-05,
      "loss": 0.8369,
      "step": 222730
    },
    {
      "epoch": 3.910532137151284,
      "grad_norm": 0.04891053679878505,
      "learning_rate": 2.51859598471194e-05,
      "loss": 0.8432,
      "step": 222740
    },
    {
      "epoch": 3.910707702031286,
      "grad_norm": 0.05635723215794897,
      "learning_rate": 2.5178794071942497e-05,
      "loss": 0.8345,
      "step": 222750
    },
    {
      "epoch": 3.910883266911287,
      "grad_norm": 0.04745852694065883,
      "learning_rate": 2.517162925743483e-05,
      "loss": 0.8415,
      "step": 222760
    },
    {
      "epoch": 3.9110588317912884,
      "grad_norm": 0.049571360046278226,
      "learning_rate": 2.5164465403687138e-05,
      "loss": 0.8445,
      "step": 222770
    },
    {
      "epoch": 3.91123439667129,
      "grad_norm": 0.05358081082111596,
      "learning_rate": 2.5157302510790187e-05,
      "loss": 0.8354,
      "step": 222780
    },
    {
      "epoch": 3.9114099615512914,
      "grad_norm": 0.04480936441606455,
      "learning_rate": 2.5150140578834785e-05,
      "loss": 0.843,
      "step": 222790
    },
    {
      "epoch": 3.911585526431293,
      "grad_norm": 0.05135699245890941,
      "learning_rate": 2.5142979607911605e-05,
      "loss": 0.8306,
      "step": 222800
    },
    {
      "epoch": 3.911761091311294,
      "grad_norm": 0.04784014493148808,
      "learning_rate": 2.513581959811142e-05,
      "loss": 0.8379,
      "step": 222810
    },
    {
      "epoch": 3.9119366561912954,
      "grad_norm": 0.047636982311832506,
      "learning_rate": 2.5128660549524936e-05,
      "loss": 0.8376,
      "step": 222820
    },
    {
      "epoch": 3.912112221071297,
      "grad_norm": 0.07585041254467774,
      "learning_rate": 2.5121502462242897e-05,
      "loss": 0.8407,
      "step": 222830
    },
    {
      "epoch": 3.9122877859512983,
      "grad_norm": 0.056758557372683524,
      "learning_rate": 2.511434533635592e-05,
      "loss": 0.8361,
      "step": 222840
    },
    {
      "epoch": 3.9124633508313,
      "grad_norm": 0.052679176917707515,
      "learning_rate": 2.5107189171954748e-05,
      "loss": 0.8347,
      "step": 222850
    },
    {
      "epoch": 3.912638915711301,
      "grad_norm": 0.05681462775667279,
      "learning_rate": 2.5100033969130054e-05,
      "loss": 0.8326,
      "step": 222860
    },
    {
      "epoch": 3.9128144805913028,
      "grad_norm": 0.05252644959309877,
      "learning_rate": 2.509287972797241e-05,
      "loss": 0.8353,
      "step": 222870
    },
    {
      "epoch": 3.912990045471304,
      "grad_norm": 0.065855881824067,
      "learning_rate": 2.50857264485726e-05,
      "loss": 0.8398,
      "step": 222880
    },
    {
      "epoch": 3.9131656103513053,
      "grad_norm": 0.0682574561875848,
      "learning_rate": 2.507857413102115e-05,
      "loss": 0.8449,
      "step": 222890
    },
    {
      "epoch": 3.9133411752313068,
      "grad_norm": 0.049085464328467214,
      "learning_rate": 2.507142277540874e-05,
      "loss": 0.8372,
      "step": 222900
    },
    {
      "epoch": 3.9135167401113082,
      "grad_norm": 0.057471239967051814,
      "learning_rate": 2.506427238182594e-05,
      "loss": 0.8436,
      "step": 222910
    },
    {
      "epoch": 3.9136923049913097,
      "grad_norm": 0.04952618242859789,
      "learning_rate": 2.5057122950363355e-05,
      "loss": 0.8421,
      "step": 222920
    },
    {
      "epoch": 3.9138678698713107,
      "grad_norm": 0.08035761768240005,
      "learning_rate": 2.5049974481111595e-05,
      "loss": 0.8312,
      "step": 222930
    },
    {
      "epoch": 3.9140434347513122,
      "grad_norm": 0.0487278079069478,
      "learning_rate": 2.50428269741612e-05,
      "loss": 0.8399,
      "step": 222940
    },
    {
      "epoch": 3.9142189996313137,
      "grad_norm": 0.0816181795021592,
      "learning_rate": 2.5035680429602738e-05,
      "loss": 0.8428,
      "step": 222950
    },
    {
      "epoch": 3.914394564511315,
      "grad_norm": 0.06680148582254974,
      "learning_rate": 2.502853484752679e-05,
      "loss": 0.8351,
      "step": 222960
    },
    {
      "epoch": 3.9145701293913167,
      "grad_norm": 0.04608058533976283,
      "learning_rate": 2.5021390228023838e-05,
      "loss": 0.8423,
      "step": 222970
    },
    {
      "epoch": 3.9147456942713177,
      "grad_norm": 0.048243771479153416,
      "learning_rate": 2.501424657118444e-05,
      "loss": 0.8408,
      "step": 222980
    },
    {
      "epoch": 3.9149212591513196,
      "grad_norm": 0.04561159599788216,
      "learning_rate": 2.500710387709909e-05,
      "loss": 0.8356,
      "step": 222990
    },
    {
      "epoch": 3.9150968240313206,
      "grad_norm": 0.05067786512171242,
      "learning_rate": 2.4999962145858332e-05,
      "loss": 0.8415,
      "step": 223000
    },
    {
      "epoch": 3.915272388911322,
      "grad_norm": 0.04792632939687021,
      "learning_rate": 2.499282137755259e-05,
      "loss": 0.8353,
      "step": 223010
    },
    {
      "epoch": 3.9154479537913236,
      "grad_norm": 0.04708025888635383,
      "learning_rate": 2.498568157227237e-05,
      "loss": 0.835,
      "step": 223020
    },
    {
      "epoch": 3.915623518671325,
      "grad_norm": 0.05512401020216134,
      "learning_rate": 2.4978542730108153e-05,
      "loss": 0.8306,
      "step": 223030
    },
    {
      "epoch": 3.9157990835513266,
      "grad_norm": 0.053294569416631084,
      "learning_rate": 2.4971404851150327e-05,
      "loss": 0.8462,
      "step": 223040
    },
    {
      "epoch": 3.9159746484313276,
      "grad_norm": 0.04964393280292346,
      "learning_rate": 2.496426793548942e-05,
      "loss": 0.8362,
      "step": 223050
    },
    {
      "epoch": 3.916150213311329,
      "grad_norm": 0.053129865908775685,
      "learning_rate": 2.4957131983215777e-05,
      "loss": 0.8336,
      "step": 223060
    },
    {
      "epoch": 3.9163257781913305,
      "grad_norm": 0.055949936585148186,
      "learning_rate": 2.494999699441987e-05,
      "loss": 0.8378,
      "step": 223070
    },
    {
      "epoch": 3.916501343071332,
      "grad_norm": 0.05023851661608079,
      "learning_rate": 2.4942862969192063e-05,
      "loss": 0.8419,
      "step": 223080
    },
    {
      "epoch": 3.9166769079513335,
      "grad_norm": 0.059835608739477325,
      "learning_rate": 2.4935729907622754e-05,
      "loss": 0.8387,
      "step": 223090
    },
    {
      "epoch": 3.916852472831335,
      "grad_norm": 0.05763509156512283,
      "learning_rate": 2.4928597809802355e-05,
      "loss": 0.8316,
      "step": 223100
    },
    {
      "epoch": 3.9170280377113365,
      "grad_norm": 0.05022630400949895,
      "learning_rate": 2.492146667582116e-05,
      "loss": 0.8401,
      "step": 223110
    },
    {
      "epoch": 3.9172036025913375,
      "grad_norm": 0.052385232723080065,
      "learning_rate": 2.491433650576957e-05,
      "loss": 0.8394,
      "step": 223120
    },
    {
      "epoch": 3.917379167471339,
      "grad_norm": 0.05002254262703842,
      "learning_rate": 2.490720729973795e-05,
      "loss": 0.843,
      "step": 223130
    },
    {
      "epoch": 3.9175547323513404,
      "grad_norm": 0.048839952583107346,
      "learning_rate": 2.4900079057816563e-05,
      "loss": 0.8394,
      "step": 223140
    },
    {
      "epoch": 3.917730297231342,
      "grad_norm": 0.04908643989855063,
      "learning_rate": 2.4892951780095765e-05,
      "loss": 0.8503,
      "step": 223150
    },
    {
      "epoch": 3.9179058621113434,
      "grad_norm": 0.06714767961869435,
      "learning_rate": 2.4885825466665853e-05,
      "loss": 0.8296,
      "step": 223160
    },
    {
      "epoch": 3.9180814269913444,
      "grad_norm": 0.05235203221316881,
      "learning_rate": 2.4878700117617145e-05,
      "loss": 0.8416,
      "step": 223170
    },
    {
      "epoch": 3.918256991871346,
      "grad_norm": 0.05655042391345255,
      "learning_rate": 2.4871575733039862e-05,
      "loss": 0.837,
      "step": 223180
    },
    {
      "epoch": 3.9184325567513474,
      "grad_norm": 0.050462290620073715,
      "learning_rate": 2.4864452313024316e-05,
      "loss": 0.8342,
      "step": 223190
    },
    {
      "epoch": 3.918608121631349,
      "grad_norm": 0.04637210504593296,
      "learning_rate": 2.4857329857660774e-05,
      "loss": 0.8352,
      "step": 223200
    },
    {
      "epoch": 3.9187836865113503,
      "grad_norm": 0.064379678635108,
      "learning_rate": 2.4850208367039396e-05,
      "loss": 0.8404,
      "step": 223210
    },
    {
      "epoch": 3.918959251391352,
      "grad_norm": 0.0465860113420379,
      "learning_rate": 2.484308784125052e-05,
      "loss": 0.8427,
      "step": 223220
    },
    {
      "epoch": 3.9191348162713533,
      "grad_norm": 0.06271684623781769,
      "learning_rate": 2.4835968280384307e-05,
      "loss": 0.8398,
      "step": 223230
    },
    {
      "epoch": 3.9193103811513543,
      "grad_norm": 0.05160943889064425,
      "learning_rate": 2.4828849684530987e-05,
      "loss": 0.8366,
      "step": 223240
    },
    {
      "epoch": 3.919485946031356,
      "grad_norm": 0.03924001714865068,
      "learning_rate": 2.4821732053780714e-05,
      "loss": 0.8295,
      "step": 223250
    },
    {
      "epoch": 3.9196615109113573,
      "grad_norm": 0.041132229107978986,
      "learning_rate": 2.4814615388223696e-05,
      "loss": 0.8305,
      "step": 223260
    },
    {
      "epoch": 3.9198370757913588,
      "grad_norm": 0.05727879564856762,
      "learning_rate": 2.480749968795014e-05,
      "loss": 0.8462,
      "step": 223270
    },
    {
      "epoch": 3.9200126406713602,
      "grad_norm": 0.047632396495137175,
      "learning_rate": 2.4800384953050125e-05,
      "loss": 0.8394,
      "step": 223280
    },
    {
      "epoch": 3.9201882055513613,
      "grad_norm": 0.05324215493316091,
      "learning_rate": 2.4793271183613842e-05,
      "loss": 0.8339,
      "step": 223290
    },
    {
      "epoch": 3.9203637704313627,
      "grad_norm": 0.0510860358835914,
      "learning_rate": 2.4786158379731445e-05,
      "loss": 0.8366,
      "step": 223300
    },
    {
      "epoch": 3.9205393353113642,
      "grad_norm": 0.05893799589145969,
      "learning_rate": 2.4779046541492995e-05,
      "loss": 0.838,
      "step": 223310
    },
    {
      "epoch": 3.9207149001913657,
      "grad_norm": 0.04297664764834778,
      "learning_rate": 2.4771935668988636e-05,
      "loss": 0.8407,
      "step": 223320
    },
    {
      "epoch": 3.920890465071367,
      "grad_norm": 0.17901914086912823,
      "learning_rate": 2.4764825762308462e-05,
      "loss": 0.8391,
      "step": 223330
    },
    {
      "epoch": 3.9210660299513687,
      "grad_norm": 0.04751149715552077,
      "learning_rate": 2.4757716821542584e-05,
      "loss": 0.8389,
      "step": 223340
    },
    {
      "epoch": 3.92124159483137,
      "grad_norm": 0.05359505790382573,
      "learning_rate": 2.4750608846781018e-05,
      "loss": 0.8462,
      "step": 223350
    },
    {
      "epoch": 3.921417159711371,
      "grad_norm": 0.04121727165131206,
      "learning_rate": 2.4743501838113854e-05,
      "loss": 0.8408,
      "step": 223360
    },
    {
      "epoch": 3.9215927245913726,
      "grad_norm": 0.05243557372666204,
      "learning_rate": 2.473639579563115e-05,
      "loss": 0.842,
      "step": 223370
    },
    {
      "epoch": 3.921768289471374,
      "grad_norm": 0.046395350782784894,
      "learning_rate": 2.472929071942289e-05,
      "loss": 0.8387,
      "step": 223380
    },
    {
      "epoch": 3.9219438543513756,
      "grad_norm": 0.05562416432839406,
      "learning_rate": 2.472218660957918e-05,
      "loss": 0.8327,
      "step": 223390
    },
    {
      "epoch": 3.922119419231377,
      "grad_norm": 0.05249161707483656,
      "learning_rate": 2.4715083466189963e-05,
      "loss": 0.833,
      "step": 223400
    },
    {
      "epoch": 3.922294984111378,
      "grad_norm": 0.04924528856922924,
      "learning_rate": 2.4707981289345275e-05,
      "loss": 0.8384,
      "step": 223410
    },
    {
      "epoch": 3.92247054899138,
      "grad_norm": 0.05020663163262387,
      "learning_rate": 2.470088007913507e-05,
      "loss": 0.8503,
      "step": 223420
    },
    {
      "epoch": 3.922646113871381,
      "grad_norm": 0.05526747987591852,
      "learning_rate": 2.4693779835649322e-05,
      "loss": 0.8403,
      "step": 223430
    },
    {
      "epoch": 3.9228216787513825,
      "grad_norm": 0.054855693490470175,
      "learning_rate": 2.468668055897805e-05,
      "loss": 0.8386,
      "step": 223440
    },
    {
      "epoch": 3.922997243631384,
      "grad_norm": 0.11811092747384613,
      "learning_rate": 2.4679582249211127e-05,
      "loss": 0.8407,
      "step": 223450
    },
    {
      "epoch": 3.9231728085113855,
      "grad_norm": 0.03912785788399499,
      "learning_rate": 2.4672484906438513e-05,
      "loss": 0.843,
      "step": 223460
    },
    {
      "epoch": 3.923348373391387,
      "grad_norm": 0.04700597347386273,
      "learning_rate": 2.4665388530750187e-05,
      "loss": 0.8439,
      "step": 223470
    },
    {
      "epoch": 3.923523938271388,
      "grad_norm": 0.04839487041841482,
      "learning_rate": 2.465829312223599e-05,
      "loss": 0.8406,
      "step": 223480
    },
    {
      "epoch": 3.9236995031513895,
      "grad_norm": 0.06062724693202287,
      "learning_rate": 2.4651198680985846e-05,
      "loss": 0.8336,
      "step": 223490
    },
    {
      "epoch": 3.923875068031391,
      "grad_norm": 0.0629905031764298,
      "learning_rate": 2.4644105207089642e-05,
      "loss": 0.8386,
      "step": 223500
    },
    {
      "epoch": 3.9240506329113924,
      "grad_norm": 0.05038689536509044,
      "learning_rate": 2.463701270063729e-05,
      "loss": 0.8361,
      "step": 223510
    },
    {
      "epoch": 3.924226197791394,
      "grad_norm": 0.05271554142286564,
      "learning_rate": 2.4629921161718593e-05,
      "loss": 0.8457,
      "step": 223520
    },
    {
      "epoch": 3.924401762671395,
      "grad_norm": 0.04598280615078472,
      "learning_rate": 2.4622830590423428e-05,
      "loss": 0.8328,
      "step": 223530
    },
    {
      "epoch": 3.924577327551397,
      "grad_norm": 0.06775766193620684,
      "learning_rate": 2.461574098684167e-05,
      "loss": 0.8414,
      "step": 223540
    },
    {
      "epoch": 3.924752892431398,
      "grad_norm": 0.0499704203292704,
      "learning_rate": 2.4608652351063053e-05,
      "loss": 0.8386,
      "step": 223550
    },
    {
      "epoch": 3.9249284573113994,
      "grad_norm": 0.057802878623316706,
      "learning_rate": 2.4601564683177517e-05,
      "loss": 0.8375,
      "step": 223560
    },
    {
      "epoch": 3.925104022191401,
      "grad_norm": 0.05501947480566439,
      "learning_rate": 2.459447798327476e-05,
      "loss": 0.8388,
      "step": 223570
    },
    {
      "epoch": 3.9252795870714023,
      "grad_norm": 0.05105925218013196,
      "learning_rate": 2.4587392251444647e-05,
      "loss": 0.8394,
      "step": 223580
    },
    {
      "epoch": 3.925455151951404,
      "grad_norm": 0.04891881226928075,
      "learning_rate": 2.4580307487776877e-05,
      "loss": 0.8393,
      "step": 223590
    },
    {
      "epoch": 3.925630716831405,
      "grad_norm": 0.04393028668541129,
      "learning_rate": 2.4573223692361272e-05,
      "loss": 0.8381,
      "step": 223600
    },
    {
      "epoch": 3.9258062817114063,
      "grad_norm": 0.0703778171868106,
      "learning_rate": 2.4566140865287583e-05,
      "loss": 0.8386,
      "step": 223610
    },
    {
      "epoch": 3.925981846591408,
      "grad_norm": 0.0542423255277622,
      "learning_rate": 2.455905900664552e-05,
      "loss": 0.8367,
      "step": 223620
    },
    {
      "epoch": 3.9261574114714093,
      "grad_norm": 0.0883874779432791,
      "learning_rate": 2.455197811652482e-05,
      "loss": 0.8377,
      "step": 223630
    },
    {
      "epoch": 3.9263329763514108,
      "grad_norm": 0.05051554286958653,
      "learning_rate": 2.4544898195015245e-05,
      "loss": 0.8331,
      "step": 223640
    },
    {
      "epoch": 3.926508541231412,
      "grad_norm": 0.06461369541864205,
      "learning_rate": 2.4537819242206438e-05,
      "loss": 0.8448,
      "step": 223650
    },
    {
      "epoch": 3.9266841061114137,
      "grad_norm": 0.047449804428909184,
      "learning_rate": 2.45307412581881e-05,
      "loss": 0.8406,
      "step": 223660
    },
    {
      "epoch": 3.9268596709914148,
      "grad_norm": 0.07085143830001511,
      "learning_rate": 2.4523664243049946e-05,
      "loss": 0.8425,
      "step": 223670
    },
    {
      "epoch": 3.9270352358714162,
      "grad_norm": 0.047388456791911615,
      "learning_rate": 2.4516588196881633e-05,
      "loss": 0.8397,
      "step": 223680
    },
    {
      "epoch": 3.9272108007514177,
      "grad_norm": 0.14969775743968985,
      "learning_rate": 2.450951311977279e-05,
      "loss": 0.8494,
      "step": 223690
    },
    {
      "epoch": 3.927386365631419,
      "grad_norm": 0.046005296742803785,
      "learning_rate": 2.4502439011813065e-05,
      "loss": 0.8457,
      "step": 223700
    },
    {
      "epoch": 3.9275619305114207,
      "grad_norm": 0.06509808326749875,
      "learning_rate": 2.4495365873092136e-05,
      "loss": 0.8426,
      "step": 223710
    },
    {
      "epoch": 3.9277374953914217,
      "grad_norm": 0.05439897358312565,
      "learning_rate": 2.4488293703699514e-05,
      "loss": 0.8385,
      "step": 223720
    },
    {
      "epoch": 3.927913060271423,
      "grad_norm": 0.04938063441280698,
      "learning_rate": 2.448122250372494e-05,
      "loss": 0.8321,
      "step": 223730
    },
    {
      "epoch": 3.9280886251514247,
      "grad_norm": 0.06329698046664364,
      "learning_rate": 2.447415227325791e-05,
      "loss": 0.8368,
      "step": 223740
    },
    {
      "epoch": 3.928264190031426,
      "grad_norm": 0.06473332257704978,
      "learning_rate": 2.4467083012388072e-05,
      "loss": 0.8361,
      "step": 223750
    },
    {
      "epoch": 3.9284397549114276,
      "grad_norm": 0.1228807809432291,
      "learning_rate": 2.4460014721204927e-05,
      "loss": 0.8334,
      "step": 223760
    },
    {
      "epoch": 3.928615319791429,
      "grad_norm": 0.04672775553937674,
      "learning_rate": 2.445294739979806e-05,
      "loss": 0.8387,
      "step": 223770
    },
    {
      "epoch": 3.9287908846714306,
      "grad_norm": 0.05680709356047684,
      "learning_rate": 2.4445881048257057e-05,
      "loss": 0.8355,
      "step": 223780
    },
    {
      "epoch": 3.9289664495514316,
      "grad_norm": 0.07064693709782413,
      "learning_rate": 2.4438815666671377e-05,
      "loss": 0.8332,
      "step": 223790
    },
    {
      "epoch": 3.929142014431433,
      "grad_norm": 0.049232702142001795,
      "learning_rate": 2.4431751255130574e-05,
      "loss": 0.8392,
      "step": 223800
    },
    {
      "epoch": 3.9293175793114345,
      "grad_norm": 0.049183572633600785,
      "learning_rate": 2.442468781372419e-05,
      "loss": 0.8424,
      "step": 223810
    },
    {
      "epoch": 3.929493144191436,
      "grad_norm": 0.06007952204474032,
      "learning_rate": 2.441762534254166e-05,
      "loss": 0.8355,
      "step": 223820
    },
    {
      "epoch": 3.9296687090714375,
      "grad_norm": 0.05208452318674892,
      "learning_rate": 2.441056384167248e-05,
      "loss": 0.8394,
      "step": 223830
    },
    {
      "epoch": 3.9298442739514385,
      "grad_norm": 0.04986974777271606,
      "learning_rate": 2.4403503311206152e-05,
      "loss": 0.839,
      "step": 223840
    },
    {
      "epoch": 3.93001983883144,
      "grad_norm": 0.05326906435999666,
      "learning_rate": 2.439644375123214e-05,
      "loss": 0.8391,
      "step": 223850
    },
    {
      "epoch": 3.9301954037114415,
      "grad_norm": 0.048317015352379554,
      "learning_rate": 2.438938516183983e-05,
      "loss": 0.844,
      "step": 223860
    },
    {
      "epoch": 3.930370968591443,
      "grad_norm": 0.043541265757232224,
      "learning_rate": 2.4382327543118707e-05,
      "loss": 0.8408,
      "step": 223870
    },
    {
      "epoch": 3.9305465334714444,
      "grad_norm": 0.05457252810915343,
      "learning_rate": 2.4375270895158196e-05,
      "loss": 0.8426,
      "step": 223880
    },
    {
      "epoch": 3.930722098351446,
      "grad_norm": 0.06967359845156232,
      "learning_rate": 2.436821521804763e-05,
      "loss": 0.8372,
      "step": 223890
    },
    {
      "epoch": 3.9308976632314474,
      "grad_norm": 0.05559620310266747,
      "learning_rate": 2.436116051187653e-05,
      "loss": 0.8374,
      "step": 223900
    },
    {
      "epoch": 3.9310732281114484,
      "grad_norm": 0.04502612380973581,
      "learning_rate": 2.4354106776734174e-05,
      "loss": 0.843,
      "step": 223910
    },
    {
      "epoch": 3.93124879299145,
      "grad_norm": 0.058635889520780624,
      "learning_rate": 2.434705401271e-05,
      "loss": 0.8401,
      "step": 223920
    },
    {
      "epoch": 3.9314243578714514,
      "grad_norm": 0.04492718066042652,
      "learning_rate": 2.4340002219893334e-05,
      "loss": 0.8386,
      "step": 223930
    },
    {
      "epoch": 3.931599922751453,
      "grad_norm": 0.04355234583555915,
      "learning_rate": 2.4332951398373515e-05,
      "loss": 0.8304,
      "step": 223940
    },
    {
      "epoch": 3.9317754876314543,
      "grad_norm": 0.04444816717927564,
      "learning_rate": 2.4325901548239927e-05,
      "loss": 0.8432,
      "step": 223950
    },
    {
      "epoch": 3.9319510525114554,
      "grad_norm": 0.06308421271011419,
      "learning_rate": 2.4318852669581836e-05,
      "loss": 0.8363,
      "step": 223960
    },
    {
      "epoch": 3.932126617391457,
      "grad_norm": 0.0723679598270689,
      "learning_rate": 2.431180476248857e-05,
      "loss": 0.8388,
      "step": 223970
    },
    {
      "epoch": 3.9323021822714583,
      "grad_norm": 0.04954448325688299,
      "learning_rate": 2.430475782704947e-05,
      "loss": 0.834,
      "step": 223980
    },
    {
      "epoch": 3.93247774715146,
      "grad_norm": 0.04863550780569379,
      "learning_rate": 2.4297711863353757e-05,
      "loss": 0.8366,
      "step": 223990
    },
    {
      "epoch": 3.9326533120314613,
      "grad_norm": 0.05234502697807503,
      "learning_rate": 2.4290666871490737e-05,
      "loss": 0.8352,
      "step": 224000
    },
    {
      "epoch": 3.9328288769114628,
      "grad_norm": 0.05402449194254526,
      "learning_rate": 2.4283622851549665e-05,
      "loss": 0.8413,
      "step": 224010
    },
    {
      "epoch": 3.9330044417914642,
      "grad_norm": 0.04805542559108572,
      "learning_rate": 2.427657980361982e-05,
      "loss": 0.8442,
      "step": 224020
    },
    {
      "epoch": 3.9331800066714653,
      "grad_norm": 0.0518576905712402,
      "learning_rate": 2.426953772779039e-05,
      "loss": 0.8381,
      "step": 224030
    },
    {
      "epoch": 3.9333555715514668,
      "grad_norm": 0.05210529228102218,
      "learning_rate": 2.4262496624150623e-05,
      "loss": 0.8364,
      "step": 224040
    },
    {
      "epoch": 3.9335311364314682,
      "grad_norm": 0.06177317130231797,
      "learning_rate": 2.425545649278975e-05,
      "loss": 0.8385,
      "step": 224050
    },
    {
      "epoch": 3.9337067013114697,
      "grad_norm": 0.07529380083206047,
      "learning_rate": 2.4248417333796938e-05,
      "loss": 0.8346,
      "step": 224060
    },
    {
      "epoch": 3.933882266191471,
      "grad_norm": 0.0672219959658823,
      "learning_rate": 2.424137914726137e-05,
      "loss": 0.8335,
      "step": 224070
    },
    {
      "epoch": 3.9340578310714722,
      "grad_norm": 0.04438231908397419,
      "learning_rate": 2.4234341933272256e-05,
      "loss": 0.8422,
      "step": 224080
    },
    {
      "epoch": 3.934233395951474,
      "grad_norm": 0.07131112844252073,
      "learning_rate": 2.4227305691918753e-05,
      "loss": 0.8412,
      "step": 224090
    },
    {
      "epoch": 3.934408960831475,
      "grad_norm": 0.05185847460707317,
      "learning_rate": 2.4220270423289988e-05,
      "loss": 0.8411,
      "step": 224100
    },
    {
      "epoch": 3.9345845257114767,
      "grad_norm": 0.05092729042862933,
      "learning_rate": 2.42132361274751e-05,
      "loss": 0.8386,
      "step": 224110
    },
    {
      "epoch": 3.934760090591478,
      "grad_norm": 0.0469894209407835,
      "learning_rate": 2.420620280456326e-05,
      "loss": 0.847,
      "step": 224120
    },
    {
      "epoch": 3.9349356554714796,
      "grad_norm": 0.06310715101920093,
      "learning_rate": 2.4199170454643523e-05,
      "loss": 0.8363,
      "step": 224130
    },
    {
      "epoch": 3.935111220351481,
      "grad_norm": 0.053234016051740156,
      "learning_rate": 2.419213907780501e-05,
      "loss": 0.8474,
      "step": 224140
    },
    {
      "epoch": 3.935286785231482,
      "grad_norm": 0.056581789182372104,
      "learning_rate": 2.4185108674136845e-05,
      "loss": 0.8387,
      "step": 224150
    },
    {
      "epoch": 3.9354623501114836,
      "grad_norm": 0.0703410237965118,
      "learning_rate": 2.417807924372804e-05,
      "loss": 0.8308,
      "step": 224160
    },
    {
      "epoch": 3.935637914991485,
      "grad_norm": 0.044095131822475216,
      "learning_rate": 2.41710507866677e-05,
      "loss": 0.8445,
      "step": 224170
    },
    {
      "epoch": 3.9358134798714866,
      "grad_norm": 0.05128925858956376,
      "learning_rate": 2.4164023303044872e-05,
      "loss": 0.8452,
      "step": 224180
    },
    {
      "epoch": 3.935989044751488,
      "grad_norm": 0.05506190554991513,
      "learning_rate": 2.415699679294862e-05,
      "loss": 0.8372,
      "step": 224190
    },
    {
      "epoch": 3.936164609631489,
      "grad_norm": 0.05524985307123473,
      "learning_rate": 2.4149971256467917e-05,
      "loss": 0.8365,
      "step": 224200
    },
    {
      "epoch": 3.936340174511491,
      "grad_norm": 0.048725493016499186,
      "learning_rate": 2.4142946693691803e-05,
      "loss": 0.8483,
      "step": 224210
    },
    {
      "epoch": 3.936515739391492,
      "grad_norm": 0.040402586094626054,
      "learning_rate": 2.4135923104709325e-05,
      "loss": 0.843,
      "step": 224220
    },
    {
      "epoch": 3.9366913042714935,
      "grad_norm": 0.0644290195135227,
      "learning_rate": 2.412890048960939e-05,
      "loss": 0.8368,
      "step": 224230
    },
    {
      "epoch": 3.936866869151495,
      "grad_norm": 0.0553975343681094,
      "learning_rate": 2.4121878848481025e-05,
      "loss": 0.8362,
      "step": 224240
    },
    {
      "epoch": 3.9370424340314965,
      "grad_norm": 0.046753908509745534,
      "learning_rate": 2.411485818141319e-05,
      "loss": 0.8371,
      "step": 224250
    },
    {
      "epoch": 3.937217998911498,
      "grad_norm": 0.057076619384453395,
      "learning_rate": 2.4107838488494865e-05,
      "loss": 0.8411,
      "step": 224260
    },
    {
      "epoch": 3.937393563791499,
      "grad_norm": 0.12401009607673923,
      "learning_rate": 2.410081976981493e-05,
      "loss": 0.8429,
      "step": 224270
    },
    {
      "epoch": 3.9375691286715004,
      "grad_norm": 0.06687806341301054,
      "learning_rate": 2.4093802025462344e-05,
      "loss": 0.842,
      "step": 224280
    },
    {
      "epoch": 3.937744693551502,
      "grad_norm": 0.06059138225631249,
      "learning_rate": 2.4086785255526062e-05,
      "loss": 0.844,
      "step": 224290
    },
    {
      "epoch": 3.9379202584315034,
      "grad_norm": 0.05168650466024851,
      "learning_rate": 2.4079769460094912e-05,
      "loss": 0.8283,
      "step": 224300
    },
    {
      "epoch": 3.938095823311505,
      "grad_norm": 0.049303920423355775,
      "learning_rate": 2.4072754639257837e-05,
      "loss": 0.8371,
      "step": 224310
    },
    {
      "epoch": 3.938271388191506,
      "grad_norm": 0.05676106125255165,
      "learning_rate": 2.4065740793103727e-05,
      "loss": 0.8447,
      "step": 224320
    },
    {
      "epoch": 3.938446953071508,
      "grad_norm": 0.05198962897004642,
      "learning_rate": 2.4058727921721393e-05,
      "loss": 0.8424,
      "step": 224330
    },
    {
      "epoch": 3.938622517951509,
      "grad_norm": 0.04804131022060601,
      "learning_rate": 2.405171602519972e-05,
      "loss": 0.8382,
      "step": 224340
    },
    {
      "epoch": 3.9387980828315103,
      "grad_norm": 0.07439879210683698,
      "learning_rate": 2.404470510362756e-05,
      "loss": 0.8469,
      "step": 224350
    },
    {
      "epoch": 3.938973647711512,
      "grad_norm": 0.057092042861448665,
      "learning_rate": 2.4037695157093764e-05,
      "loss": 0.8519,
      "step": 224360
    },
    {
      "epoch": 3.9391492125915133,
      "grad_norm": 0.07076730944604544,
      "learning_rate": 2.403068618568709e-05,
      "loss": 0.8382,
      "step": 224370
    },
    {
      "epoch": 3.9393247774715148,
      "grad_norm": 0.04069287065214631,
      "learning_rate": 2.402367818949637e-05,
      "loss": 0.8395,
      "step": 224380
    },
    {
      "epoch": 3.939500342351516,
      "grad_norm": 0.05485048636218767,
      "learning_rate": 2.4016671168610427e-05,
      "loss": 0.8406,
      "step": 224390
    },
    {
      "epoch": 3.9396759072315173,
      "grad_norm": 0.04633185227112773,
      "learning_rate": 2.400966512311799e-05,
      "loss": 0.8441,
      "step": 224400
    },
    {
      "epoch": 3.9398514721115188,
      "grad_norm": 0.0557285575178913,
      "learning_rate": 2.400266005310785e-05,
      "loss": 0.8397,
      "step": 224410
    },
    {
      "epoch": 3.9400270369915202,
      "grad_norm": 0.046156037465429046,
      "learning_rate": 2.3995655958668757e-05,
      "loss": 0.8434,
      "step": 224420
    },
    {
      "epoch": 3.9402026018715217,
      "grad_norm": 0.059985935508809166,
      "learning_rate": 2.3988652839889493e-05,
      "loss": 0.8339,
      "step": 224430
    },
    {
      "epoch": 3.9403781667515227,
      "grad_norm": 0.05016818211120627,
      "learning_rate": 2.398165069685874e-05,
      "loss": 0.8374,
      "step": 224440
    },
    {
      "epoch": 3.9405537316315247,
      "grad_norm": 0.04968732113795546,
      "learning_rate": 2.3974649529665226e-05,
      "loss": 0.8428,
      "step": 224450
    },
    {
      "epoch": 3.9407292965115257,
      "grad_norm": 0.04317397619716404,
      "learning_rate": 2.3967649338397684e-05,
      "loss": 0.8358,
      "step": 224460
    },
    {
      "epoch": 3.940904861391527,
      "grad_norm": 0.04648251506646281,
      "learning_rate": 2.3960650123144766e-05,
      "loss": 0.8346,
      "step": 224470
    },
    {
      "epoch": 3.9410804262715287,
      "grad_norm": 0.06623392657144216,
      "learning_rate": 2.3953651883995175e-05,
      "loss": 0.8367,
      "step": 224480
    },
    {
      "epoch": 3.94125599115153,
      "grad_norm": 0.07177572560274494,
      "learning_rate": 2.394665462103761e-05,
      "loss": 0.8493,
      "step": 224490
    },
    {
      "epoch": 3.9414315560315316,
      "grad_norm": 0.048518738869775264,
      "learning_rate": 2.393965833436066e-05,
      "loss": 0.8437,
      "step": 224500
    },
    {
      "epoch": 3.9416071209115326,
      "grad_norm": 0.04831438592413749,
      "learning_rate": 2.3932663024053026e-05,
      "loss": 0.8406,
      "step": 224510
    },
    {
      "epoch": 3.941782685791534,
      "grad_norm": 0.05846614966278079,
      "learning_rate": 2.3925668690203316e-05,
      "loss": 0.8353,
      "step": 224520
    },
    {
      "epoch": 3.9419582506715356,
      "grad_norm": 0.068675400877912,
      "learning_rate": 2.391867533290017e-05,
      "loss": 0.8431,
      "step": 224530
    },
    {
      "epoch": 3.942133815551537,
      "grad_norm": 0.06320666224647879,
      "learning_rate": 2.3911682952232163e-05,
      "loss": 0.8399,
      "step": 224540
    },
    {
      "epoch": 3.9423093804315386,
      "grad_norm": 0.057210744658155226,
      "learning_rate": 2.3904691548287903e-05,
      "loss": 0.8399,
      "step": 224550
    },
    {
      "epoch": 3.94248494531154,
      "grad_norm": 0.05391059429535026,
      "learning_rate": 2.389770112115601e-05,
      "loss": 0.8406,
      "step": 224560
    },
    {
      "epoch": 3.9426605101915415,
      "grad_norm": 0.05395559766949972,
      "learning_rate": 2.389071167092499e-05,
      "loss": 0.8428,
      "step": 224570
    },
    {
      "epoch": 3.9428360750715425,
      "grad_norm": 0.046366601196112495,
      "learning_rate": 2.388372319768343e-05,
      "loss": 0.8401,
      "step": 224580
    },
    {
      "epoch": 3.943011639951544,
      "grad_norm": 0.06418005373784866,
      "learning_rate": 2.387673570151987e-05,
      "loss": 0.8401,
      "step": 224590
    },
    {
      "epoch": 3.9431872048315455,
      "grad_norm": 0.04845294425421254,
      "learning_rate": 2.386974918252288e-05,
      "loss": 0.8412,
      "step": 224600
    },
    {
      "epoch": 3.943362769711547,
      "grad_norm": 0.049253453459686916,
      "learning_rate": 2.3862763640780926e-05,
      "loss": 0.8458,
      "step": 224610
    },
    {
      "epoch": 3.9435383345915485,
      "grad_norm": 0.051672303299260064,
      "learning_rate": 2.3855779076382543e-05,
      "loss": 0.8401,
      "step": 224620
    },
    {
      "epoch": 3.9437138994715495,
      "grad_norm": 0.07562344674236351,
      "learning_rate": 2.3848795489416248e-05,
      "loss": 0.8367,
      "step": 224630
    },
    {
      "epoch": 3.943889464351551,
      "grad_norm": 0.048478626739089585,
      "learning_rate": 2.384181287997047e-05,
      "loss": 0.842,
      "step": 224640
    },
    {
      "epoch": 3.9440650292315524,
      "grad_norm": 0.06210288773760766,
      "learning_rate": 2.3834831248133707e-05,
      "loss": 0.838,
      "step": 224650
    },
    {
      "epoch": 3.944240594111554,
      "grad_norm": 0.04665486008408933,
      "learning_rate": 2.3827850593994457e-05,
      "loss": 0.8405,
      "step": 224660
    },
    {
      "epoch": 3.9444161589915554,
      "grad_norm": 0.061912702848381435,
      "learning_rate": 2.38208709176411e-05,
      "loss": 0.8399,
      "step": 224670
    },
    {
      "epoch": 3.944591723871557,
      "grad_norm": 0.043842179393565495,
      "learning_rate": 2.3813892219162105e-05,
      "loss": 0.8314,
      "step": 224680
    },
    {
      "epoch": 3.9447672887515584,
      "grad_norm": 0.04749931174912437,
      "learning_rate": 2.3806914498645873e-05,
      "loss": 0.8434,
      "step": 224690
    },
    {
      "epoch": 3.9449428536315594,
      "grad_norm": 0.055888787498836356,
      "learning_rate": 2.3799937756180874e-05,
      "loss": 0.8421,
      "step": 224700
    },
    {
      "epoch": 3.945118418511561,
      "grad_norm": 0.04921552296985843,
      "learning_rate": 2.379296199185542e-05,
      "loss": 0.8376,
      "step": 224710
    },
    {
      "epoch": 3.9452939833915623,
      "grad_norm": 0.04738319084311507,
      "learning_rate": 2.3785987205757937e-05,
      "loss": 0.8375,
      "step": 224720
    },
    {
      "epoch": 3.945469548271564,
      "grad_norm": 0.052032268133969856,
      "learning_rate": 2.3779013397976817e-05,
      "loss": 0.8393,
      "step": 224730
    },
    {
      "epoch": 3.9456451131515653,
      "grad_norm": 0.054983528357267185,
      "learning_rate": 2.377204056860037e-05,
      "loss": 0.8309,
      "step": 224740
    },
    {
      "epoch": 3.9458206780315663,
      "grad_norm": 0.06575384920093152,
      "learning_rate": 2.376506871771697e-05,
      "loss": 0.84,
      "step": 224750
    },
    {
      "epoch": 3.945996242911568,
      "grad_norm": 0.05502459380931436,
      "learning_rate": 2.3758097845414952e-05,
      "loss": 0.8418,
      "step": 224760
    },
    {
      "epoch": 3.9461718077915693,
      "grad_norm": 0.04698816824625595,
      "learning_rate": 2.3751127951782677e-05,
      "loss": 0.8429,
      "step": 224770
    },
    {
      "epoch": 3.9463473726715708,
      "grad_norm": 0.06837682643443004,
      "learning_rate": 2.3744159036908376e-05,
      "loss": 0.8384,
      "step": 224780
    },
    {
      "epoch": 3.9465229375515722,
      "grad_norm": 0.049644604669017614,
      "learning_rate": 2.373719110088038e-05,
      "loss": 0.8332,
      "step": 224790
    },
    {
      "epoch": 3.9466985024315737,
      "grad_norm": 0.05669255780967697,
      "learning_rate": 2.3730224143787022e-05,
      "loss": 0.8355,
      "step": 224800
    },
    {
      "epoch": 3.946874067311575,
      "grad_norm": 0.06456288139721122,
      "learning_rate": 2.37232581657165e-05,
      "loss": 0.8337,
      "step": 224810
    },
    {
      "epoch": 3.9470496321915762,
      "grad_norm": 0.06307850878788412,
      "learning_rate": 2.37162931667571e-05,
      "loss": 0.8391,
      "step": 224820
    },
    {
      "epoch": 3.9472251970715777,
      "grad_norm": 0.06912312947790923,
      "learning_rate": 2.370932914699712e-05,
      "loss": 0.8357,
      "step": 224830
    },
    {
      "epoch": 3.947400761951579,
      "grad_norm": 0.049213703689543115,
      "learning_rate": 2.3702366106524716e-05,
      "loss": 0.8438,
      "step": 224840
    },
    {
      "epoch": 3.9475763268315807,
      "grad_norm": 0.05753624121901695,
      "learning_rate": 2.3695404045428137e-05,
      "loss": 0.8439,
      "step": 224850
    },
    {
      "epoch": 3.947751891711582,
      "grad_norm": 0.048883732112584746,
      "learning_rate": 2.3688442963795616e-05,
      "loss": 0.8364,
      "step": 224860
    },
    {
      "epoch": 3.947927456591583,
      "grad_norm": 0.05227169470510754,
      "learning_rate": 2.3681482861715366e-05,
      "loss": 0.8339,
      "step": 224870
    },
    {
      "epoch": 3.948103021471585,
      "grad_norm": 0.04587854214649641,
      "learning_rate": 2.367452373927552e-05,
      "loss": 0.8343,
      "step": 224880
    },
    {
      "epoch": 3.948278586351586,
      "grad_norm": 0.049475846648960695,
      "learning_rate": 2.366756559656427e-05,
      "loss": 0.8418,
      "step": 224890
    },
    {
      "epoch": 3.9484541512315876,
      "grad_norm": 0.06701148413519502,
      "learning_rate": 2.3660608433669814e-05,
      "loss": 0.8423,
      "step": 224900
    },
    {
      "epoch": 3.948629716111589,
      "grad_norm": 0.06260232919526293,
      "learning_rate": 2.365365225068025e-05,
      "loss": 0.837,
      "step": 224910
    },
    {
      "epoch": 3.9488052809915906,
      "grad_norm": 0.06329780080211406,
      "learning_rate": 2.364669704768373e-05,
      "loss": 0.834,
      "step": 224920
    },
    {
      "epoch": 3.948980845871592,
      "grad_norm": 0.06754892195110392,
      "learning_rate": 2.3639742824768386e-05,
      "loss": 0.8516,
      "step": 224930
    },
    {
      "epoch": 3.949156410751593,
      "grad_norm": 0.05910715414148592,
      "learning_rate": 2.363278958202236e-05,
      "loss": 0.8405,
      "step": 224940
    },
    {
      "epoch": 3.9493319756315945,
      "grad_norm": 0.06515496981216616,
      "learning_rate": 2.3625837319533677e-05,
      "loss": 0.8309,
      "step": 224950
    },
    {
      "epoch": 3.949507540511596,
      "grad_norm": 0.041502391121396975,
      "learning_rate": 2.3618886037390474e-05,
      "loss": 0.8367,
      "step": 224960
    },
    {
      "epoch": 3.9496831053915975,
      "grad_norm": 0.05000038999335869,
      "learning_rate": 2.3611935735680836e-05,
      "loss": 0.8378,
      "step": 224970
    },
    {
      "epoch": 3.949858670271599,
      "grad_norm": 0.053879330042811216,
      "learning_rate": 2.3604986414492774e-05,
      "loss": 0.8378,
      "step": 224980
    },
    {
      "epoch": 3.9500342351516,
      "grad_norm": 0.07346493553644391,
      "learning_rate": 2.3598038073914376e-05,
      "loss": 0.8378,
      "step": 224990
    },
    {
      "epoch": 3.950209800031602,
      "grad_norm": 0.04840284641736829,
      "learning_rate": 2.3591090714033695e-05,
      "loss": 0.8289,
      "step": 225000
    },
    {
      "epoch": 3.950385364911603,
      "grad_norm": 0.04353469045987646,
      "learning_rate": 2.3584144334938695e-05,
      "loss": 0.8436,
      "step": 225010
    },
    {
      "epoch": 3.9505609297916044,
      "grad_norm": 0.060955053933933905,
      "learning_rate": 2.3577198936717432e-05,
      "loss": 0.8358,
      "step": 225020
    },
    {
      "epoch": 3.950736494671606,
      "grad_norm": 0.04854924015224883,
      "learning_rate": 2.3570254519457884e-05,
      "loss": 0.8335,
      "step": 225030
    },
    {
      "epoch": 3.9509120595516074,
      "grad_norm": 0.0447376968715446,
      "learning_rate": 2.3563311083248094e-05,
      "loss": 0.8342,
      "step": 225040
    },
    {
      "epoch": 3.951087624431609,
      "grad_norm": 0.051643653905525944,
      "learning_rate": 2.3556368628175954e-05,
      "loss": 0.8403,
      "step": 225050
    },
    {
      "epoch": 3.95126318931161,
      "grad_norm": 0.05049232012012966,
      "learning_rate": 2.3549427154329473e-05,
      "loss": 0.8417,
      "step": 225060
    },
    {
      "epoch": 3.9514387541916114,
      "grad_norm": 0.05695566957856484,
      "learning_rate": 2.354248666179662e-05,
      "loss": 0.8412,
      "step": 225070
    },
    {
      "epoch": 3.951614319071613,
      "grad_norm": 0.049236166393520436,
      "learning_rate": 2.3535547150665278e-05,
      "loss": 0.844,
      "step": 225080
    },
    {
      "epoch": 3.9517898839516143,
      "grad_norm": 0.06313357509540621,
      "learning_rate": 2.352860862102339e-05,
      "loss": 0.8346,
      "step": 225090
    },
    {
      "epoch": 3.951965448831616,
      "grad_norm": 0.05510044611014629,
      "learning_rate": 2.352167107295889e-05,
      "loss": 0.8383,
      "step": 225100
    },
    {
      "epoch": 3.952141013711617,
      "grad_norm": 0.06470712392658955,
      "learning_rate": 2.3514734506559685e-05,
      "loss": 0.8373,
      "step": 225110
    },
    {
      "epoch": 3.952316578591619,
      "grad_norm": 0.05203772168287974,
      "learning_rate": 2.3507798921913614e-05,
      "loss": 0.8375,
      "step": 225120
    },
    {
      "epoch": 3.95249214347162,
      "grad_norm": 0.05403772098405379,
      "learning_rate": 2.3500864319108593e-05,
      "loss": 0.8343,
      "step": 225130
    },
    {
      "epoch": 3.9526677083516213,
      "grad_norm": 0.045242162580193056,
      "learning_rate": 2.34939306982325e-05,
      "loss": 0.8435,
      "step": 225140
    },
    {
      "epoch": 3.9528432732316228,
      "grad_norm": 0.051923852448948815,
      "learning_rate": 2.3486998059373122e-05,
      "loss": 0.8344,
      "step": 225150
    },
    {
      "epoch": 3.9530188381116242,
      "grad_norm": 0.057886710015134187,
      "learning_rate": 2.348006640261834e-05,
      "loss": 0.837,
      "step": 225160
    },
    {
      "epoch": 3.9531944029916257,
      "grad_norm": 0.04705839178828937,
      "learning_rate": 2.3473135728056e-05,
      "loss": 0.8546,
      "step": 225170
    },
    {
      "epoch": 3.9533699678716268,
      "grad_norm": 0.04783437126401676,
      "learning_rate": 2.3466206035773866e-05,
      "loss": 0.8435,
      "step": 225180
    },
    {
      "epoch": 3.9535455327516282,
      "grad_norm": 0.07179909373520822,
      "learning_rate": 2.3459277325859753e-05,
      "loss": 0.8414,
      "step": 225190
    },
    {
      "epoch": 3.9537210976316297,
      "grad_norm": 0.04217915594854657,
      "learning_rate": 2.3452349598401465e-05,
      "loss": 0.8436,
      "step": 225200
    },
    {
      "epoch": 3.953896662511631,
      "grad_norm": 0.0638730192624945,
      "learning_rate": 2.3445422853486794e-05,
      "loss": 0.8439,
      "step": 225210
    },
    {
      "epoch": 3.9540722273916327,
      "grad_norm": 0.04682166556422792,
      "learning_rate": 2.3438497091203444e-05,
      "loss": 0.841,
      "step": 225220
    },
    {
      "epoch": 3.954247792271634,
      "grad_norm": 0.06905096161551508,
      "learning_rate": 2.3431572311639204e-05,
      "loss": 0.8331,
      "step": 225230
    },
    {
      "epoch": 3.9544233571516356,
      "grad_norm": 0.059520755992303935,
      "learning_rate": 2.342464851488184e-05,
      "loss": 0.8419,
      "step": 225240
    },
    {
      "epoch": 3.9545989220316367,
      "grad_norm": 0.05540182845579542,
      "learning_rate": 2.3417725701019015e-05,
      "loss": 0.8376,
      "step": 225250
    },
    {
      "epoch": 3.954774486911638,
      "grad_norm": 0.0420725183467415,
      "learning_rate": 2.3410803870138472e-05,
      "loss": 0.833,
      "step": 225260
    },
    {
      "epoch": 3.9549500517916396,
      "grad_norm": 0.06379193013100236,
      "learning_rate": 2.3403883022327926e-05,
      "loss": 0.8444,
      "step": 225270
    },
    {
      "epoch": 3.955125616671641,
      "grad_norm": 0.04924774012722095,
      "learning_rate": 2.3396963157675073e-05,
      "loss": 0.837,
      "step": 225280
    },
    {
      "epoch": 3.9553011815516426,
      "grad_norm": 0.0616076780021955,
      "learning_rate": 2.3390044276267543e-05,
      "loss": 0.8466,
      "step": 225290
    },
    {
      "epoch": 3.9554767464316436,
      "grad_norm": 0.051607132087945075,
      "learning_rate": 2.3383126378193015e-05,
      "loss": 0.8439,
      "step": 225300
    },
    {
      "epoch": 3.955652311311645,
      "grad_norm": 0.060178467303109946,
      "learning_rate": 2.337620946353919e-05,
      "loss": 0.8474,
      "step": 225310
    },
    {
      "epoch": 3.9558278761916466,
      "grad_norm": 0.04065689051682223,
      "learning_rate": 2.3369293532393642e-05,
      "loss": 0.8451,
      "step": 225320
    },
    {
      "epoch": 3.956003441071648,
      "grad_norm": 0.049374929055085656,
      "learning_rate": 2.3362378584844014e-05,
      "loss": 0.8431,
      "step": 225330
    },
    {
      "epoch": 3.9561790059516495,
      "grad_norm": 0.05401239693261681,
      "learning_rate": 2.335546462097797e-05,
      "loss": 0.834,
      "step": 225340
    },
    {
      "epoch": 3.956354570831651,
      "grad_norm": 0.04038480289425001,
      "learning_rate": 2.3348551640883032e-05,
      "loss": 0.8475,
      "step": 225350
    },
    {
      "epoch": 3.9565301357116525,
      "grad_norm": 0.051792679062727776,
      "learning_rate": 2.334163964464682e-05,
      "loss": 0.8346,
      "step": 225360
    },
    {
      "epoch": 3.9567057005916535,
      "grad_norm": 0.07478411283900097,
      "learning_rate": 2.3334728632356925e-05,
      "loss": 0.8435,
      "step": 225370
    },
    {
      "epoch": 3.956881265471655,
      "grad_norm": 0.06219660307306123,
      "learning_rate": 2.3327818604100933e-05,
      "loss": 0.8334,
      "step": 225380
    },
    {
      "epoch": 3.9570568303516565,
      "grad_norm": 0.042181317765587686,
      "learning_rate": 2.3320909559966333e-05,
      "loss": 0.8395,
      "step": 225390
    },
    {
      "epoch": 3.957232395231658,
      "grad_norm": 0.049165429798655604,
      "learning_rate": 2.3314001500040693e-05,
      "loss": 0.8317,
      "step": 225400
    },
    {
      "epoch": 3.9574079601116594,
      "grad_norm": 0.0593172415425077,
      "learning_rate": 2.3307094424411573e-05,
      "loss": 0.8369,
      "step": 225410
    },
    {
      "epoch": 3.9575835249916604,
      "grad_norm": 0.04505635264516648,
      "learning_rate": 2.3300188333166435e-05,
      "loss": 0.8498,
      "step": 225420
    },
    {
      "epoch": 3.957759089871662,
      "grad_norm": 0.05300548871726665,
      "learning_rate": 2.32932832263928e-05,
      "loss": 0.8424,
      "step": 225430
    },
    {
      "epoch": 3.9579346547516634,
      "grad_norm": 0.05111450859715031,
      "learning_rate": 2.328637910417815e-05,
      "loss": 0.8387,
      "step": 225440
    },
    {
      "epoch": 3.958110219631665,
      "grad_norm": 0.051273390350459025,
      "learning_rate": 2.327947596661001e-05,
      "loss": 0.8393,
      "step": 225450
    },
    {
      "epoch": 3.9582857845116663,
      "grad_norm": 0.04667941341002912,
      "learning_rate": 2.3272573813775765e-05,
      "loss": 0.8381,
      "step": 225460
    },
    {
      "epoch": 3.958461349391668,
      "grad_norm": 0.05198671493753532,
      "learning_rate": 2.3265672645762916e-05,
      "loss": 0.8307,
      "step": 225470
    },
    {
      "epoch": 3.9586369142716693,
      "grad_norm": 0.06288627013909785,
      "learning_rate": 2.325877246265891e-05,
      "loss": 0.8307,
      "step": 225480
    },
    {
      "epoch": 3.9588124791516703,
      "grad_norm": 0.05099434428653766,
      "learning_rate": 2.3251873264551144e-05,
      "loss": 0.8334,
      "step": 225490
    },
    {
      "epoch": 3.958988044031672,
      "grad_norm": 0.06215934327510436,
      "learning_rate": 2.3244975051527024e-05,
      "loss": 0.8401,
      "step": 225500
    },
    {
      "epoch": 3.9591636089116733,
      "grad_norm": 0.04317158311767044,
      "learning_rate": 2.3238077823674012e-05,
      "loss": 0.8406,
      "step": 225510
    },
    {
      "epoch": 3.9593391737916748,
      "grad_norm": 0.06692603823672869,
      "learning_rate": 2.323118158107943e-05,
      "loss": 0.8415,
      "step": 225520
    },
    {
      "epoch": 3.9595147386716762,
      "grad_norm": 0.0623700518599184,
      "learning_rate": 2.322428632383068e-05,
      "loss": 0.8376,
      "step": 225530
    },
    {
      "epoch": 3.9596903035516773,
      "grad_norm": 0.044918763334732846,
      "learning_rate": 2.321739205201513e-05,
      "loss": 0.8496,
      "step": 225540
    },
    {
      "epoch": 3.959865868431679,
      "grad_norm": 0.04524257608110994,
      "learning_rate": 2.3210498765720158e-05,
      "loss": 0.8289,
      "step": 225550
    },
    {
      "epoch": 3.9600414333116802,
      "grad_norm": 0.05476486599744425,
      "learning_rate": 2.320360646503303e-05,
      "loss": 0.8347,
      "step": 225560
    },
    {
      "epoch": 3.9602169981916817,
      "grad_norm": 0.05653042211268956,
      "learning_rate": 2.319671515004114e-05,
      "loss": 0.8322,
      "step": 225570
    },
    {
      "epoch": 3.960392563071683,
      "grad_norm": 0.06011296293197752,
      "learning_rate": 2.3189824820831797e-05,
      "loss": 0.8355,
      "step": 225580
    },
    {
      "epoch": 3.9605681279516847,
      "grad_norm": 0.048529690788469734,
      "learning_rate": 2.3182935477492256e-05,
      "loss": 0.8361,
      "step": 225590
    },
    {
      "epoch": 3.960743692831686,
      "grad_norm": 0.04369726379883062,
      "learning_rate": 2.3176047120109846e-05,
      "loss": 0.8399,
      "step": 225600
    },
    {
      "epoch": 3.960919257711687,
      "grad_norm": 0.05875493691745331,
      "learning_rate": 2.316915974877183e-05,
      "loss": 0.8418,
      "step": 225610
    },
    {
      "epoch": 3.9610948225916887,
      "grad_norm": 0.04850909848446647,
      "learning_rate": 2.3162273363565494e-05,
      "loss": 0.8416,
      "step": 225620
    },
    {
      "epoch": 3.96127038747169,
      "grad_norm": 0.05305197394927158,
      "learning_rate": 2.3155387964578057e-05,
      "loss": 0.8411,
      "step": 225630
    },
    {
      "epoch": 3.9614459523516916,
      "grad_norm": 0.05144439003435167,
      "learning_rate": 2.3148503551896763e-05,
      "loss": 0.8418,
      "step": 225640
    },
    {
      "epoch": 3.961621517231693,
      "grad_norm": 0.06314365854403091,
      "learning_rate": 2.3141620125608888e-05,
      "loss": 0.8404,
      "step": 225650
    },
    {
      "epoch": 3.961797082111694,
      "grad_norm": 0.05508293084541895,
      "learning_rate": 2.3134737685801576e-05,
      "loss": 0.8399,
      "step": 225660
    },
    {
      "epoch": 3.961972646991696,
      "grad_norm": 0.04644905192659217,
      "learning_rate": 2.3127856232562056e-05,
      "loss": 0.8362,
      "step": 225670
    },
    {
      "epoch": 3.962148211871697,
      "grad_norm": 0.04960314415503474,
      "learning_rate": 2.3120975765977555e-05,
      "loss": 0.847,
      "step": 225680
    },
    {
      "epoch": 3.9623237767516986,
      "grad_norm": 0.05440822991071069,
      "learning_rate": 2.3114096286135185e-05,
      "loss": 0.8342,
      "step": 225690
    },
    {
      "epoch": 3.9624993416317,
      "grad_norm": 0.04911120295128656,
      "learning_rate": 2.3107217793122134e-05,
      "loss": 0.8344,
      "step": 225700
    },
    {
      "epoch": 3.9626749065117015,
      "grad_norm": 0.054306748731680586,
      "learning_rate": 2.3100340287025567e-05,
      "loss": 0.8458,
      "step": 225710
    },
    {
      "epoch": 3.962850471391703,
      "grad_norm": 0.046142468181516724,
      "learning_rate": 2.309346376793265e-05,
      "loss": 0.8405,
      "step": 225720
    },
    {
      "epoch": 3.963026036271704,
      "grad_norm": 0.06213790586703322,
      "learning_rate": 2.3086588235930457e-05,
      "loss": 0.839,
      "step": 225730
    },
    {
      "epoch": 3.9632016011517055,
      "grad_norm": 0.05806237269574725,
      "learning_rate": 2.307971369110611e-05,
      "loss": 0.8384,
      "step": 225740
    },
    {
      "epoch": 3.963377166031707,
      "grad_norm": 0.05165416410915949,
      "learning_rate": 2.3072840133546755e-05,
      "loss": 0.8367,
      "step": 225750
    },
    {
      "epoch": 3.9635527309117085,
      "grad_norm": 0.04647062367936725,
      "learning_rate": 2.3065967563339427e-05,
      "loss": 0.8393,
      "step": 225760
    },
    {
      "epoch": 3.96372829579171,
      "grad_norm": 0.05269270487894632,
      "learning_rate": 2.3059095980571222e-05,
      "loss": 0.8409,
      "step": 225770
    },
    {
      "epoch": 3.963903860671711,
      "grad_norm": 0.04399231996183998,
      "learning_rate": 2.3052225385329214e-05,
      "loss": 0.8361,
      "step": 225780
    },
    {
      "epoch": 3.964079425551713,
      "grad_norm": 0.048290862896502744,
      "learning_rate": 2.3045355777700463e-05,
      "loss": 0.8326,
      "step": 225790
    },
    {
      "epoch": 3.964254990431714,
      "grad_norm": 0.043381131159501686,
      "learning_rate": 2.3038487157771986e-05,
      "loss": 0.8349,
      "step": 225800
    },
    {
      "epoch": 3.9644305553117154,
      "grad_norm": 0.04975475729327939,
      "learning_rate": 2.3031619525630802e-05,
      "loss": 0.8408,
      "step": 225810
    },
    {
      "epoch": 3.964606120191717,
      "grad_norm": 0.05145619193170061,
      "learning_rate": 2.302475288136398e-05,
      "loss": 0.8403,
      "step": 225820
    },
    {
      "epoch": 3.9647816850717184,
      "grad_norm": 0.0539233215433761,
      "learning_rate": 2.301788722505845e-05,
      "loss": 0.8427,
      "step": 225830
    },
    {
      "epoch": 3.96495724995172,
      "grad_norm": 0.05076041489299348,
      "learning_rate": 2.301102255680122e-05,
      "loss": 0.8494,
      "step": 225840
    },
    {
      "epoch": 3.965132814831721,
      "grad_norm": 0.06951154600721028,
      "learning_rate": 2.3004158876679335e-05,
      "loss": 0.842,
      "step": 225850
    },
    {
      "epoch": 3.9653083797117223,
      "grad_norm": 0.0568043682715977,
      "learning_rate": 2.2997296184779655e-05,
      "loss": 0.8347,
      "step": 225860
    },
    {
      "epoch": 3.965483944591724,
      "grad_norm": 0.0762659302176215,
      "learning_rate": 2.299043448118919e-05,
      "loss": 0.8368,
      "step": 225870
    },
    {
      "epoch": 3.9656595094717253,
      "grad_norm": 0.05090812348334133,
      "learning_rate": 2.2983573765994877e-05,
      "loss": 0.8367,
      "step": 225880
    },
    {
      "epoch": 3.9658350743517268,
      "grad_norm": 0.0744969642370656,
      "learning_rate": 2.2976714039283655e-05,
      "loss": 0.8316,
      "step": 225890
    },
    {
      "epoch": 3.966010639231728,
      "grad_norm": 0.04816313560943294,
      "learning_rate": 2.2969855301142397e-05,
      "loss": 0.8365,
      "step": 225900
    },
    {
      "epoch": 3.9661862041117297,
      "grad_norm": 0.04096872107683798,
      "learning_rate": 2.296299755165802e-05,
      "loss": 0.8473,
      "step": 225910
    },
    {
      "epoch": 3.9663617689917308,
      "grad_norm": 0.04760117685538072,
      "learning_rate": 2.295614079091746e-05,
      "loss": 0.8468,
      "step": 225920
    },
    {
      "epoch": 3.9665373338717322,
      "grad_norm": 0.0514970219977188,
      "learning_rate": 2.2949285019007517e-05,
      "loss": 0.8416,
      "step": 225930
    },
    {
      "epoch": 3.9667128987517337,
      "grad_norm": 0.05104448249231461,
      "learning_rate": 2.294243023601509e-05,
      "loss": 0.8345,
      "step": 225940
    },
    {
      "epoch": 3.966888463631735,
      "grad_norm": 0.04870001931343752,
      "learning_rate": 2.2935576442027026e-05,
      "loss": 0.8419,
      "step": 225950
    },
    {
      "epoch": 3.9670640285117367,
      "grad_norm": 0.04963396392310607,
      "learning_rate": 2.2928723637130206e-05,
      "loss": 0.8473,
      "step": 225960
    },
    {
      "epoch": 3.9672395933917377,
      "grad_norm": 0.037654005215487105,
      "learning_rate": 2.2921871821411392e-05,
      "loss": 0.8397,
      "step": 225970
    },
    {
      "epoch": 3.967415158271739,
      "grad_norm": 0.047628208381196024,
      "learning_rate": 2.2915020994957424e-05,
      "loss": 0.8414,
      "step": 225980
    },
    {
      "epoch": 3.9675907231517407,
      "grad_norm": 0.05544774623214029,
      "learning_rate": 2.2908171157855133e-05,
      "loss": 0.841,
      "step": 225990
    },
    {
      "epoch": 3.967766288031742,
      "grad_norm": 0.06092133986118265,
      "learning_rate": 2.2901322310191252e-05,
      "loss": 0.8434,
      "step": 226000
    },
    {
      "epoch": 3.9679418529117436,
      "grad_norm": 0.08808143597500004,
      "learning_rate": 2.2894474452052588e-05,
      "loss": 0.851,
      "step": 226010
    },
    {
      "epoch": 3.968117417791745,
      "grad_norm": 0.058044730309808536,
      "learning_rate": 2.288762758352592e-05,
      "loss": 0.8374,
      "step": 226020
    },
    {
      "epoch": 3.9682929826717466,
      "grad_norm": 0.049579162511288355,
      "learning_rate": 2.2880781704697962e-05,
      "loss": 0.8381,
      "step": 226030
    },
    {
      "epoch": 3.9684685475517476,
      "grad_norm": 0.05747987005939569,
      "learning_rate": 2.2873936815655462e-05,
      "loss": 0.8495,
      "step": 226040
    },
    {
      "epoch": 3.968644112431749,
      "grad_norm": 0.05321774459553546,
      "learning_rate": 2.286709291648516e-05,
      "loss": 0.8396,
      "step": 226050
    },
    {
      "epoch": 3.9688196773117506,
      "grad_norm": 0.05674652664605749,
      "learning_rate": 2.2860250007273792e-05,
      "loss": 0.8376,
      "step": 226060
    },
    {
      "epoch": 3.968995242191752,
      "grad_norm": 0.08130360691728586,
      "learning_rate": 2.2853408088108007e-05,
      "loss": 0.8414,
      "step": 226070
    },
    {
      "epoch": 3.9691708070717535,
      "grad_norm": 0.06112022776402533,
      "learning_rate": 2.284656715907452e-05,
      "loss": 0.8332,
      "step": 226080
    },
    {
      "epoch": 3.9693463719517545,
      "grad_norm": 0.04543150610007141,
      "learning_rate": 2.2839727220260033e-05,
      "loss": 0.8375,
      "step": 226090
    },
    {
      "epoch": 3.969521936831756,
      "grad_norm": 0.05779857467308036,
      "learning_rate": 2.2832888271751144e-05,
      "loss": 0.8392,
      "step": 226100
    },
    {
      "epoch": 3.9696975017117575,
      "grad_norm": 0.05821896444890764,
      "learning_rate": 2.2826050313634557e-05,
      "loss": 0.8414,
      "step": 226110
    },
    {
      "epoch": 3.969873066591759,
      "grad_norm": 0.05155330135743396,
      "learning_rate": 2.281921334599688e-05,
      "loss": 0.8478,
      "step": 226120
    },
    {
      "epoch": 3.9700486314717605,
      "grad_norm": 0.09931996245028926,
      "learning_rate": 2.281237736892479e-05,
      "loss": 0.8385,
      "step": 226130
    },
    {
      "epoch": 3.970224196351762,
      "grad_norm": 0.0583234613979553,
      "learning_rate": 2.280554238250483e-05,
      "loss": 0.8423,
      "step": 226140
    },
    {
      "epoch": 3.9703997612317634,
      "grad_norm": 0.06167940754758536,
      "learning_rate": 2.2798708386823638e-05,
      "loss": 0.8472,
      "step": 226150
    },
    {
      "epoch": 3.9705753261117644,
      "grad_norm": 0.04318904904847112,
      "learning_rate": 2.2791875381967816e-05,
      "loss": 0.8437,
      "step": 226160
    },
    {
      "epoch": 3.970750890991766,
      "grad_norm": 0.0526727922047294,
      "learning_rate": 2.2785043368023907e-05,
      "loss": 0.837,
      "step": 226170
    },
    {
      "epoch": 3.9709264558717674,
      "grad_norm": 0.04802089832016558,
      "learning_rate": 2.2778212345078486e-05,
      "loss": 0.8359,
      "step": 226180
    },
    {
      "epoch": 3.971102020751769,
      "grad_norm": 0.04099142111290917,
      "learning_rate": 2.2771382313218124e-05,
      "loss": 0.8408,
      "step": 226190
    },
    {
      "epoch": 3.9712775856317704,
      "grad_norm": 0.04415385326178758,
      "learning_rate": 2.2764553272529317e-05,
      "loss": 0.833,
      "step": 226200
    },
    {
      "epoch": 3.9714531505117714,
      "grad_norm": 0.06822338137746732,
      "learning_rate": 2.2757725223098614e-05,
      "loss": 0.8478,
      "step": 226210
    },
    {
      "epoch": 3.971628715391773,
      "grad_norm": 0.05602027772896421,
      "learning_rate": 2.275089816501254e-05,
      "loss": 0.845,
      "step": 226220
    },
    {
      "epoch": 3.9718042802717743,
      "grad_norm": 0.05760572681883167,
      "learning_rate": 2.2744072098357596e-05,
      "loss": 0.8423,
      "step": 226230
    },
    {
      "epoch": 3.971979845151776,
      "grad_norm": 0.04953970877146825,
      "learning_rate": 2.2737247023220227e-05,
      "loss": 0.8333,
      "step": 226240
    },
    {
      "epoch": 3.9721554100317773,
      "grad_norm": 0.09406755921083576,
      "learning_rate": 2.2730422939686942e-05,
      "loss": 0.8388,
      "step": 226250
    },
    {
      "epoch": 3.972330974911779,
      "grad_norm": 0.052307735572449776,
      "learning_rate": 2.2723599847844223e-05,
      "loss": 0.842,
      "step": 226260
    },
    {
      "epoch": 3.9725065397917803,
      "grad_norm": 0.059341669485236274,
      "learning_rate": 2.271677774777848e-05,
      "loss": 0.8421,
      "step": 226270
    },
    {
      "epoch": 3.9726821046717813,
      "grad_norm": 0.05514423131880389,
      "learning_rate": 2.270995663957615e-05,
      "loss": 0.8449,
      "step": 226280
    },
    {
      "epoch": 3.9728576695517828,
      "grad_norm": 0.07219590867987821,
      "learning_rate": 2.2703136523323685e-05,
      "loss": 0.8365,
      "step": 226290
    },
    {
      "epoch": 3.9730332344317842,
      "grad_norm": 0.06632044946837093,
      "learning_rate": 2.2696317399107506e-05,
      "loss": 0.8417,
      "step": 226300
    },
    {
      "epoch": 3.9732087993117857,
      "grad_norm": 0.051871066281613254,
      "learning_rate": 2.2689499267013968e-05,
      "loss": 0.8438,
      "step": 226310
    },
    {
      "epoch": 3.973384364191787,
      "grad_norm": 0.04339304581261615,
      "learning_rate": 2.268268212712949e-05,
      "loss": 0.8434,
      "step": 226320
    },
    {
      "epoch": 3.9735599290717882,
      "grad_norm": 0.06119972028314861,
      "learning_rate": 2.2675865979540456e-05,
      "loss": 0.848,
      "step": 226330
    },
    {
      "epoch": 3.97373549395179,
      "grad_norm": 0.04400076654777094,
      "learning_rate": 2.2669050824333185e-05,
      "loss": 0.8417,
      "step": 226340
    },
    {
      "epoch": 3.973911058831791,
      "grad_norm": 0.0525107264846198,
      "learning_rate": 2.2662236661594053e-05,
      "loss": 0.8404,
      "step": 226350
    },
    {
      "epoch": 3.9740866237117927,
      "grad_norm": 0.07176461329430332,
      "learning_rate": 2.2655423491409423e-05,
      "loss": 0.8378,
      "step": 226360
    },
    {
      "epoch": 3.974262188591794,
      "grad_norm": 0.04759042074798834,
      "learning_rate": 2.264861131386556e-05,
      "loss": 0.8339,
      "step": 226370
    },
    {
      "epoch": 3.9744377534717956,
      "grad_norm": 0.055375307348988954,
      "learning_rate": 2.2641800129048814e-05,
      "loss": 0.8397,
      "step": 226380
    },
    {
      "epoch": 3.974613318351797,
      "grad_norm": 0.05386799268699527,
      "learning_rate": 2.2634989937045474e-05,
      "loss": 0.8412,
      "step": 226390
    },
    {
      "epoch": 3.974788883231798,
      "grad_norm": 0.06752287827360776,
      "learning_rate": 2.2628180737941863e-05,
      "loss": 0.8412,
      "step": 226400
    },
    {
      "epoch": 3.9749644481117996,
      "grad_norm": 0.055117084391553106,
      "learning_rate": 2.262137253182419e-05,
      "loss": 0.8415,
      "step": 226410
    },
    {
      "epoch": 3.975140012991801,
      "grad_norm": 0.05647062233880191,
      "learning_rate": 2.2614565318778755e-05,
      "loss": 0.8437,
      "step": 226420
    },
    {
      "epoch": 3.9753155778718026,
      "grad_norm": 0.050731061063847546,
      "learning_rate": 2.2607759098891823e-05,
      "loss": 0.8369,
      "step": 226430
    },
    {
      "epoch": 3.975491142751804,
      "grad_norm": 0.0576610907283689,
      "learning_rate": 2.2600953872249576e-05,
      "loss": 0.84,
      "step": 226440
    },
    {
      "epoch": 3.975666707631805,
      "grad_norm": 0.05629774407810526,
      "learning_rate": 2.2594149638938265e-05,
      "loss": 0.8347,
      "step": 226450
    },
    {
      "epoch": 3.975842272511807,
      "grad_norm": 0.05692489711987001,
      "learning_rate": 2.2587346399044115e-05,
      "loss": 0.8353,
      "step": 226460
    },
    {
      "epoch": 3.976017837391808,
      "grad_norm": 0.052533202782194914,
      "learning_rate": 2.258054415265334e-05,
      "loss": 0.8387,
      "step": 226470
    },
    {
      "epoch": 3.9761934022718095,
      "grad_norm": 0.047580343551097755,
      "learning_rate": 2.2573742899852067e-05,
      "loss": 0.8489,
      "step": 226480
    },
    {
      "epoch": 3.976368967151811,
      "grad_norm": 0.05894713918098907,
      "learning_rate": 2.2566942640726503e-05,
      "loss": 0.8445,
      "step": 226490
    },
    {
      "epoch": 3.9765445320318125,
      "grad_norm": 0.06589652583978228,
      "learning_rate": 2.2560143375362836e-05,
      "loss": 0.8465,
      "step": 226500
    },
    {
      "epoch": 3.976720096911814,
      "grad_norm": 0.04912885202807653,
      "learning_rate": 2.2553345103847154e-05,
      "loss": 0.8377,
      "step": 226510
    },
    {
      "epoch": 3.976895661791815,
      "grad_norm": 0.04967598928977722,
      "learning_rate": 2.2546547826265634e-05,
      "loss": 0.8412,
      "step": 226520
    },
    {
      "epoch": 3.9770712266718165,
      "grad_norm": 0.053025090462624394,
      "learning_rate": 2.253975154270441e-05,
      "loss": 0.838,
      "step": 226530
    },
    {
      "epoch": 3.977246791551818,
      "grad_norm": 0.0733917367508938,
      "learning_rate": 2.2532956253249538e-05,
      "loss": 0.8395,
      "step": 226540
    },
    {
      "epoch": 3.9774223564318194,
      "grad_norm": 0.05804177859321806,
      "learning_rate": 2.2526161957987156e-05,
      "loss": 0.8415,
      "step": 226550
    },
    {
      "epoch": 3.977597921311821,
      "grad_norm": 0.04522171746689349,
      "learning_rate": 2.2519368657003337e-05,
      "loss": 0.8473,
      "step": 226560
    },
    {
      "epoch": 3.977773486191822,
      "grad_norm": 0.05263783458610326,
      "learning_rate": 2.251257635038419e-05,
      "loss": 0.8382,
      "step": 226570
    },
    {
      "epoch": 3.977949051071824,
      "grad_norm": 0.04030167210870562,
      "learning_rate": 2.2505785038215704e-05,
      "loss": 0.8424,
      "step": 226580
    },
    {
      "epoch": 3.978124615951825,
      "grad_norm": 0.050655501568373895,
      "learning_rate": 2.2498994720583975e-05,
      "loss": 0.8396,
      "step": 226590
    },
    {
      "epoch": 3.9783001808318263,
      "grad_norm": 0.05934574735300298,
      "learning_rate": 2.2492205397575054e-05,
      "loss": 0.8443,
      "step": 226600
    },
    {
      "epoch": 3.978475745711828,
      "grad_norm": 0.04607425069909458,
      "learning_rate": 2.2485417069274908e-05,
      "loss": 0.8412,
      "step": 226610
    },
    {
      "epoch": 3.9786513105918293,
      "grad_norm": 0.05017890199837621,
      "learning_rate": 2.247862973576958e-05,
      "loss": 0.8421,
      "step": 226620
    },
    {
      "epoch": 3.978826875471831,
      "grad_norm": 0.05915294043108734,
      "learning_rate": 2.2471843397145056e-05,
      "loss": 0.8419,
      "step": 226630
    },
    {
      "epoch": 3.979002440351832,
      "grad_norm": 0.042735548323021115,
      "learning_rate": 2.2465058053487354e-05,
      "loss": 0.8353,
      "step": 226640
    },
    {
      "epoch": 3.9791780052318333,
      "grad_norm": 0.04600409985834753,
      "learning_rate": 2.245827370488239e-05,
      "loss": 0.8402,
      "step": 226650
    },
    {
      "epoch": 3.9793535701118348,
      "grad_norm": 0.05762249195205805,
      "learning_rate": 2.2451490351416144e-05,
      "loss": 0.8394,
      "step": 226660
    },
    {
      "epoch": 3.9795291349918362,
      "grad_norm": 0.0538147471147669,
      "learning_rate": 2.2444707993174606e-05,
      "loss": 0.8421,
      "step": 226670
    },
    {
      "epoch": 3.9797046998718377,
      "grad_norm": 0.06489724446999166,
      "learning_rate": 2.243792663024365e-05,
      "loss": 0.8494,
      "step": 226680
    },
    {
      "epoch": 3.979880264751839,
      "grad_norm": 0.054526430257454156,
      "learning_rate": 2.2431146262709205e-05,
      "loss": 0.8327,
      "step": 226690
    },
    {
      "epoch": 3.9800558296318407,
      "grad_norm": 0.04935380854943934,
      "learning_rate": 2.2424366890657233e-05,
      "loss": 0.8431,
      "step": 226700
    },
    {
      "epoch": 3.9802313945118417,
      "grad_norm": 0.04886860390380545,
      "learning_rate": 2.2417588514173557e-05,
      "loss": 0.8445,
      "step": 226710
    },
    {
      "epoch": 3.980406959391843,
      "grad_norm": 0.05523251629182382,
      "learning_rate": 2.2410811133344103e-05,
      "loss": 0.8456,
      "step": 226720
    },
    {
      "epoch": 3.9805825242718447,
      "grad_norm": 0.07114654957068613,
      "learning_rate": 2.2404034748254727e-05,
      "loss": 0.84,
      "step": 226730
    },
    {
      "epoch": 3.980758089151846,
      "grad_norm": 0.04412216259962399,
      "learning_rate": 2.239725935899132e-05,
      "loss": 0.8463,
      "step": 226740
    },
    {
      "epoch": 3.9809336540318476,
      "grad_norm": 0.06477963054771592,
      "learning_rate": 2.2390484965639686e-05,
      "loss": 0.8341,
      "step": 226750
    },
    {
      "epoch": 3.9811092189118487,
      "grad_norm": 0.05905944682539789,
      "learning_rate": 2.238371156828566e-05,
      "loss": 0.8345,
      "step": 226760
    },
    {
      "epoch": 3.98128478379185,
      "grad_norm": 0.0585468497602008,
      "learning_rate": 2.2376939167015103e-05,
      "loss": 0.8333,
      "step": 226770
    },
    {
      "epoch": 3.9814603486718516,
      "grad_norm": 0.05293529518075925,
      "learning_rate": 2.2370167761913767e-05,
      "loss": 0.841,
      "step": 226780
    },
    {
      "epoch": 3.981635913551853,
      "grad_norm": 0.061130676721526195,
      "learning_rate": 2.2363397353067488e-05,
      "loss": 0.8388,
      "step": 226790
    },
    {
      "epoch": 3.9818114784318546,
      "grad_norm": 0.052330863669197286,
      "learning_rate": 2.2356627940562023e-05,
      "loss": 0.8413,
      "step": 226800
    },
    {
      "epoch": 3.981987043311856,
      "grad_norm": 0.048977060569110165,
      "learning_rate": 2.2349859524483183e-05,
      "loss": 0.8412,
      "step": 226810
    },
    {
      "epoch": 3.9821626081918575,
      "grad_norm": 0.050657880300437885,
      "learning_rate": 2.2343092104916673e-05,
      "loss": 0.8324,
      "step": 226820
    },
    {
      "epoch": 3.9823381730718586,
      "grad_norm": 0.05252651624162582,
      "learning_rate": 2.2336325681948255e-05,
      "loss": 0.8443,
      "step": 226830
    },
    {
      "epoch": 3.98251373795186,
      "grad_norm": 0.05309517342250724,
      "learning_rate": 2.23295602556637e-05,
      "loss": 0.8435,
      "step": 226840
    },
    {
      "epoch": 3.9826893028318615,
      "grad_norm": 0.0517466596639959,
      "learning_rate": 2.232279582614865e-05,
      "loss": 0.8347,
      "step": 226850
    },
    {
      "epoch": 3.982864867711863,
      "grad_norm": 0.047026939750811896,
      "learning_rate": 2.231603239348887e-05,
      "loss": 0.8401,
      "step": 226860
    },
    {
      "epoch": 3.9830404325918645,
      "grad_norm": 0.05190582402197581,
      "learning_rate": 2.2309269957770073e-05,
      "loss": 0.8356,
      "step": 226870
    },
    {
      "epoch": 3.9832159974718655,
      "grad_norm": 0.05491869208987791,
      "learning_rate": 2.230250851907786e-05,
      "loss": 0.8397,
      "step": 226880
    },
    {
      "epoch": 3.983391562351867,
      "grad_norm": 0.048200231425669246,
      "learning_rate": 2.2295748077497957e-05,
      "loss": 0.8401,
      "step": 226890
    },
    {
      "epoch": 3.9835671272318685,
      "grad_norm": 0.06086131852712872,
      "learning_rate": 2.2288988633116013e-05,
      "loss": 0.8407,
      "step": 226900
    },
    {
      "epoch": 3.98374269211187,
      "grad_norm": 0.04643107966836554,
      "learning_rate": 2.2282230186017695e-05,
      "loss": 0.8433,
      "step": 226910
    },
    {
      "epoch": 3.9839182569918714,
      "grad_norm": 0.0637420371047163,
      "learning_rate": 2.2275472736288577e-05,
      "loss": 0.8368,
      "step": 226920
    },
    {
      "epoch": 3.984093821871873,
      "grad_norm": 0.05394362726078167,
      "learning_rate": 2.2268716284014297e-05,
      "loss": 0.8335,
      "step": 226930
    },
    {
      "epoch": 3.9842693867518744,
      "grad_norm": 0.04842638468783668,
      "learning_rate": 2.2261960829280507e-05,
      "loss": 0.8344,
      "step": 226940
    },
    {
      "epoch": 3.9844449516318754,
      "grad_norm": 0.05564621547371898,
      "learning_rate": 2.225520637217274e-05,
      "loss": 0.8405,
      "step": 226950
    },
    {
      "epoch": 3.984620516511877,
      "grad_norm": 0.06042874191280068,
      "learning_rate": 2.22484529127766e-05,
      "loss": 0.8409,
      "step": 226960
    },
    {
      "epoch": 3.9847960813918784,
      "grad_norm": 0.0445711983724221,
      "learning_rate": 2.2241700451177647e-05,
      "loss": 0.8414,
      "step": 226970
    },
    {
      "epoch": 3.98497164627188,
      "grad_norm": 0.04991288363773975,
      "learning_rate": 2.223494898746147e-05,
      "loss": 0.8381,
      "step": 226980
    },
    {
      "epoch": 3.9851472111518813,
      "grad_norm": 0.0549724311428672,
      "learning_rate": 2.2228198521713565e-05,
      "loss": 0.8392,
      "step": 226990
    },
    {
      "epoch": 3.9853227760318823,
      "grad_norm": 0.05238576929733188,
      "learning_rate": 2.2221449054019473e-05,
      "loss": 0.837,
      "step": 227000
    },
    {
      "epoch": 3.9854983409118843,
      "grad_norm": 0.050892611974480126,
      "learning_rate": 2.2214700584464757e-05,
      "loss": 0.8433,
      "step": 227010
    },
    {
      "epoch": 3.9856739057918853,
      "grad_norm": 0.04712938905241488,
      "learning_rate": 2.2207953113134843e-05,
      "loss": 0.8379,
      "step": 227020
    },
    {
      "epoch": 3.9858494706718868,
      "grad_norm": 0.06144109925414211,
      "learning_rate": 2.220120664011527e-05,
      "loss": 0.8386,
      "step": 227030
    },
    {
      "epoch": 3.9860250355518883,
      "grad_norm": 0.051123138552844394,
      "learning_rate": 2.219446116549154e-05,
      "loss": 0.8412,
      "step": 227040
    },
    {
      "epoch": 3.9862006004318897,
      "grad_norm": 0.057301333050542545,
      "learning_rate": 2.2187716689349065e-05,
      "loss": 0.8388,
      "step": 227050
    },
    {
      "epoch": 3.986376165311891,
      "grad_norm": 0.04999942688027113,
      "learning_rate": 2.2180973211773313e-05,
      "loss": 0.8371,
      "step": 227060
    },
    {
      "epoch": 3.9865517301918922,
      "grad_norm": 0.041899314672286865,
      "learning_rate": 2.217423073284975e-05,
      "loss": 0.8347,
      "step": 227070
    },
    {
      "epoch": 3.9867272950718937,
      "grad_norm": 0.0678330494323735,
      "learning_rate": 2.2167489252663803e-05,
      "loss": 0.8382,
      "step": 227080
    },
    {
      "epoch": 3.986902859951895,
      "grad_norm": 0.05063874507699166,
      "learning_rate": 2.2160748771300857e-05,
      "loss": 0.8361,
      "step": 227090
    },
    {
      "epoch": 3.9870784248318967,
      "grad_norm": 0.056270212583534006,
      "learning_rate": 2.2154009288846336e-05,
      "loss": 0.8382,
      "step": 227100
    },
    {
      "epoch": 3.987253989711898,
      "grad_norm": 0.044495914589400026,
      "learning_rate": 2.2147270805385647e-05,
      "loss": 0.8479,
      "step": 227110
    },
    {
      "epoch": 3.987429554591899,
      "grad_norm": 0.06365743977912393,
      "learning_rate": 2.2140533321004134e-05,
      "loss": 0.8472,
      "step": 227120
    },
    {
      "epoch": 3.987605119471901,
      "grad_norm": 0.0587582558292326,
      "learning_rate": 2.213379683578717e-05,
      "loss": 0.8408,
      "step": 227130
    },
    {
      "epoch": 3.987780684351902,
      "grad_norm": 0.044106034482195404,
      "learning_rate": 2.2127061349820115e-05,
      "loss": 0.8305,
      "step": 227140
    },
    {
      "epoch": 3.9879562492319036,
      "grad_norm": 0.06985879722761215,
      "learning_rate": 2.2120326863188322e-05,
      "loss": 0.8357,
      "step": 227150
    },
    {
      "epoch": 3.988131814111905,
      "grad_norm": 0.04636394821963975,
      "learning_rate": 2.2113593375977094e-05,
      "loss": 0.8371,
      "step": 227160
    },
    {
      "epoch": 3.9883073789919066,
      "grad_norm": 0.05118090177600886,
      "learning_rate": 2.2106860888271753e-05,
      "loss": 0.8461,
      "step": 227170
    },
    {
      "epoch": 3.988482943871908,
      "grad_norm": 0.06809298143067441,
      "learning_rate": 2.2100129400157637e-05,
      "loss": 0.8351,
      "step": 227180
    },
    {
      "epoch": 3.988658508751909,
      "grad_norm": 0.0703820686601126,
      "learning_rate": 2.2093398911719965e-05,
      "loss": 0.8432,
      "step": 227190
    },
    {
      "epoch": 3.9888340736319106,
      "grad_norm": 0.04650253318459866,
      "learning_rate": 2.2086669423044058e-05,
      "loss": 0.8382,
      "step": 227200
    },
    {
      "epoch": 3.989009638511912,
      "grad_norm": 0.046454558814848945,
      "learning_rate": 2.2079940934215197e-05,
      "loss": 0.8422,
      "step": 227210
    },
    {
      "epoch": 3.9891852033919135,
      "grad_norm": 0.05696606857623079,
      "learning_rate": 2.2073213445318583e-05,
      "loss": 0.8355,
      "step": 227220
    },
    {
      "epoch": 3.989360768271915,
      "grad_norm": 0.05914474126213256,
      "learning_rate": 2.2066486956439485e-05,
      "loss": 0.841,
      "step": 227230
    },
    {
      "epoch": 3.989536333151916,
      "grad_norm": 0.06160147967250436,
      "learning_rate": 2.2059761467663114e-05,
      "loss": 0.8327,
      "step": 227240
    },
    {
      "epoch": 3.989711898031918,
      "grad_norm": 0.04738466682364862,
      "learning_rate": 2.205303697907473e-05,
      "loss": 0.8309,
      "step": 227250
    },
    {
      "epoch": 3.989887462911919,
      "grad_norm": 0.05865267959277675,
      "learning_rate": 2.204631349075947e-05,
      "loss": 0.8412,
      "step": 227260
    },
    {
      "epoch": 3.9900630277919205,
      "grad_norm": 0.04801719091928055,
      "learning_rate": 2.203959100280255e-05,
      "loss": 0.8431,
      "step": 227270
    },
    {
      "epoch": 3.990238592671922,
      "grad_norm": 0.05068138024911812,
      "learning_rate": 2.2032869515289172e-05,
      "loss": 0.8354,
      "step": 227280
    },
    {
      "epoch": 3.9904141575519234,
      "grad_norm": 0.05285638115743442,
      "learning_rate": 2.2026149028304443e-05,
      "loss": 0.8439,
      "step": 227290
    },
    {
      "epoch": 3.990589722431925,
      "grad_norm": 0.046168469860517494,
      "learning_rate": 2.2019429541933534e-05,
      "loss": 0.8397,
      "step": 227300
    },
    {
      "epoch": 3.990765287311926,
      "grad_norm": 0.049915204902594754,
      "learning_rate": 2.2012711056261602e-05,
      "loss": 0.836,
      "step": 227310
    },
    {
      "epoch": 3.9909408521919274,
      "grad_norm": 0.05225951905147399,
      "learning_rate": 2.2005993571373776e-05,
      "loss": 0.8353,
      "step": 227320
    },
    {
      "epoch": 3.991116417071929,
      "grad_norm": 0.050939780476679644,
      "learning_rate": 2.1999277087355126e-05,
      "loss": 0.8361,
      "step": 227330
    },
    {
      "epoch": 3.9912919819519304,
      "grad_norm": 0.05735679077132707,
      "learning_rate": 2.199256160429078e-05,
      "loss": 0.8414,
      "step": 227340
    },
    {
      "epoch": 3.991467546831932,
      "grad_norm": 0.05613003837982818,
      "learning_rate": 2.1985847122265846e-05,
      "loss": 0.8358,
      "step": 227350
    },
    {
      "epoch": 3.991643111711933,
      "grad_norm": 0.04670553408005558,
      "learning_rate": 2.1979133641365336e-05,
      "loss": 0.842,
      "step": 227360
    },
    {
      "epoch": 3.991818676591935,
      "grad_norm": 0.05468236048512556,
      "learning_rate": 2.197242116167436e-05,
      "loss": 0.8392,
      "step": 227370
    },
    {
      "epoch": 3.991994241471936,
      "grad_norm": 0.06431976723018325,
      "learning_rate": 2.1965709683277975e-05,
      "loss": 0.8393,
      "step": 227380
    },
    {
      "epoch": 3.9921698063519373,
      "grad_norm": 0.050171303876855616,
      "learning_rate": 2.1958999206261167e-05,
      "loss": 0.8346,
      "step": 227390
    },
    {
      "epoch": 3.9923453712319388,
      "grad_norm": 0.05342600629208988,
      "learning_rate": 2.1952289730708984e-05,
      "loss": 0.8403,
      "step": 227400
    },
    {
      "epoch": 3.9925209361119403,
      "grad_norm": 0.059078162549785256,
      "learning_rate": 2.1945581256706443e-05,
      "loss": 0.8362,
      "step": 227410
    },
    {
      "epoch": 3.9926965009919417,
      "grad_norm": 0.06994252526884759,
      "learning_rate": 2.193887378433856e-05,
      "loss": 0.8387,
      "step": 227420
    },
    {
      "epoch": 3.9928720658719428,
      "grad_norm": 0.05652007592176914,
      "learning_rate": 2.1932167313690286e-05,
      "loss": 0.8428,
      "step": 227430
    },
    {
      "epoch": 3.9930476307519442,
      "grad_norm": 0.053275299655313454,
      "learning_rate": 2.192546184484659e-05,
      "loss": 0.8412,
      "step": 227440
    },
    {
      "epoch": 3.9932231956319457,
      "grad_norm": 0.06141809818510227,
      "learning_rate": 2.1918757377892484e-05,
      "loss": 0.8386,
      "step": 227450
    },
    {
      "epoch": 3.993398760511947,
      "grad_norm": 0.04433531287015612,
      "learning_rate": 2.1912053912912844e-05,
      "loss": 0.848,
      "step": 227460
    },
    {
      "epoch": 3.9935743253919487,
      "grad_norm": 0.04751364168424958,
      "learning_rate": 2.190535144999264e-05,
      "loss": 0.8409,
      "step": 227470
    },
    {
      "epoch": 3.99374989027195,
      "grad_norm": 0.0750135911876498,
      "learning_rate": 2.1898649989216797e-05,
      "loss": 0.8424,
      "step": 227480
    },
    {
      "epoch": 3.9939254551519516,
      "grad_norm": 0.0500484918823922,
      "learning_rate": 2.1891949530670245e-05,
      "loss": 0.8373,
      "step": 227490
    },
    {
      "epoch": 3.9941010200319527,
      "grad_norm": 0.05247891712527428,
      "learning_rate": 2.188525007443782e-05,
      "loss": 0.8377,
      "step": 227500
    },
    {
      "epoch": 3.994276584911954,
      "grad_norm": 0.055226559510793756,
      "learning_rate": 2.1878551620604435e-05,
      "loss": 0.8426,
      "step": 227510
    },
    {
      "epoch": 3.9944521497919556,
      "grad_norm": 0.058196530672554495,
      "learning_rate": 2.1871854169255005e-05,
      "loss": 0.8398,
      "step": 227520
    },
    {
      "epoch": 3.994627714671957,
      "grad_norm": 0.04643825595916368,
      "learning_rate": 2.186515772047431e-05,
      "loss": 0.8476,
      "step": 227530
    },
    {
      "epoch": 3.9948032795519586,
      "grad_norm": 0.05002692232524761,
      "learning_rate": 2.185846227434724e-05,
      "loss": 0.8392,
      "step": 227540
    },
    {
      "epoch": 3.9949788444319596,
      "grad_norm": 0.059652802799637084,
      "learning_rate": 2.185176783095864e-05,
      "loss": 0.8388,
      "step": 227550
    },
    {
      "epoch": 3.995154409311961,
      "grad_norm": 0.04701987923405061,
      "learning_rate": 2.184507439039328e-05,
      "loss": 0.8427,
      "step": 227560
    },
    {
      "epoch": 3.9953299741919626,
      "grad_norm": 0.04754415936385225,
      "learning_rate": 2.183838195273601e-05,
      "loss": 0.8333,
      "step": 227570
    },
    {
      "epoch": 3.995505539071964,
      "grad_norm": 0.05763676965584185,
      "learning_rate": 2.1831690518071608e-05,
      "loss": 0.8354,
      "step": 227580
    },
    {
      "epoch": 3.9956811039519655,
      "grad_norm": 0.05689969035069112,
      "learning_rate": 2.1825000086484878e-05,
      "loss": 0.8404,
      "step": 227590
    },
    {
      "epoch": 3.995856668831967,
      "grad_norm": 0.05448919626840739,
      "learning_rate": 2.1818310658060552e-05,
      "loss": 0.8403,
      "step": 227600
    },
    {
      "epoch": 3.9960322337119685,
      "grad_norm": 0.04050177803148031,
      "learning_rate": 2.18116222328834e-05,
      "loss": 0.8446,
      "step": 227610
    },
    {
      "epoch": 3.9962077985919695,
      "grad_norm": 0.06555207188507584,
      "learning_rate": 2.18049348110382e-05,
      "loss": 0.8377,
      "step": 227620
    },
    {
      "epoch": 3.996383363471971,
      "grad_norm": 0.048013916196612404,
      "learning_rate": 2.1798248392609633e-05,
      "loss": 0.8482,
      "step": 227630
    },
    {
      "epoch": 3.9965589283519725,
      "grad_norm": 0.056765352553869754,
      "learning_rate": 2.179156297768243e-05,
      "loss": 0.8367,
      "step": 227640
    },
    {
      "epoch": 3.996734493231974,
      "grad_norm": 0.04808561064388354,
      "learning_rate": 2.1784878566341308e-05,
      "loss": 0.8398,
      "step": 227650
    },
    {
      "epoch": 3.9969100581119754,
      "grad_norm": 0.044499473038397755,
      "learning_rate": 2.1778195158670976e-05,
      "loss": 0.8313,
      "step": 227660
    },
    {
      "epoch": 3.9970856229919765,
      "grad_norm": 0.06031799051844324,
      "learning_rate": 2.1771512754756073e-05,
      "loss": 0.8396,
      "step": 227670
    },
    {
      "epoch": 3.997261187871978,
      "grad_norm": 0.08360284177449616,
      "learning_rate": 2.1764831354681294e-05,
      "loss": 0.8443,
      "step": 227680
    },
    {
      "epoch": 3.9974367527519794,
      "grad_norm": 0.05094004927161172,
      "learning_rate": 2.17581509585313e-05,
      "loss": 0.835,
      "step": 227690
    },
    {
      "epoch": 3.997612317631981,
      "grad_norm": 0.060297494807806176,
      "learning_rate": 2.175147156639071e-05,
      "loss": 0.8333,
      "step": 227700
    },
    {
      "epoch": 3.9977878825119824,
      "grad_norm": 0.05201927064065457,
      "learning_rate": 2.1744793178344156e-05,
      "loss": 0.836,
      "step": 227710
    },
    {
      "epoch": 3.997963447391984,
      "grad_norm": 0.04141621294754992,
      "learning_rate": 2.1738115794476276e-05,
      "loss": 0.8433,
      "step": 227720
    },
    {
      "epoch": 3.9981390122719853,
      "grad_norm": 0.08568285220978245,
      "learning_rate": 2.1731439414871652e-05,
      "loss": 0.8373,
      "step": 227730
    },
    {
      "epoch": 3.9983145771519863,
      "grad_norm": 0.049851209795767584,
      "learning_rate": 2.1724764039614864e-05,
      "loss": 0.8403,
      "step": 227740
    },
    {
      "epoch": 3.998490142031988,
      "grad_norm": 0.04993672049989771,
      "learning_rate": 2.1718089668790518e-05,
      "loss": 0.8394,
      "step": 227750
    },
    {
      "epoch": 3.9986657069119893,
      "grad_norm": 0.04795401766234504,
      "learning_rate": 2.1711416302483198e-05,
      "loss": 0.8408,
      "step": 227760
    },
    {
      "epoch": 3.998841271791991,
      "grad_norm": 0.05780452401840443,
      "learning_rate": 2.1704743940777404e-05,
      "loss": 0.8369,
      "step": 227770
    },
    {
      "epoch": 3.9990168366719923,
      "grad_norm": 0.05975472531800212,
      "learning_rate": 2.1698072583757697e-05,
      "loss": 0.8391,
      "step": 227780
    },
    {
      "epoch": 3.9991924015519933,
      "grad_norm": 0.053032980724063344,
      "learning_rate": 2.169140223150864e-05,
      "loss": 0.8282,
      "step": 227790
    },
    {
      "epoch": 3.999367966431995,
      "grad_norm": 0.054319920763426725,
      "learning_rate": 2.1684732884114688e-05,
      "loss": 0.8359,
      "step": 227800
    },
    {
      "epoch": 3.9995435313119962,
      "grad_norm": 0.049958317580873775,
      "learning_rate": 2.167806454166036e-05,
      "loss": 0.8335,
      "step": 227810
    },
    {
      "epoch": 3.9997190961919977,
      "grad_norm": 0.051806917227034066,
      "learning_rate": 2.1671397204230166e-05,
      "loss": 0.8361,
      "step": 227820
    },
    {
      "epoch": 3.999894661071999,
      "grad_norm": 0.046906767647533626,
      "learning_rate": 2.1664730871908602e-05,
      "loss": 0.8363,
      "step": 227830
    },
    {
      "epoch": 4.000070225952,
      "grad_norm": 0.051430352993023336,
      "learning_rate": 2.1658065544780068e-05,
      "loss": 0.8405,
      "step": 227840
    },
    {
      "epoch": 4.000245790832002,
      "grad_norm": 0.0576767638382457,
      "learning_rate": 2.1651401222929053e-05,
      "loss": 0.8392,
      "step": 227850
    },
    {
      "epoch": 4.000421355712003,
      "grad_norm": 0.055760159437616204,
      "learning_rate": 2.1644737906440024e-05,
      "loss": 0.834,
      "step": 227860
    },
    {
      "epoch": 4.000596920592005,
      "grad_norm": 0.04785713747947853,
      "learning_rate": 2.163807559539734e-05,
      "loss": 0.8366,
      "step": 227870
    },
    {
      "epoch": 4.000772485472006,
      "grad_norm": 0.05321898556249583,
      "learning_rate": 2.163141428988545e-05,
      "loss": 0.8324,
      "step": 227880
    },
    {
      "epoch": 4.000948050352007,
      "grad_norm": 0.052794632320891065,
      "learning_rate": 2.1624753989988776e-05,
      "loss": 0.8376,
      "step": 227890
    },
    {
      "epoch": 4.001123615232009,
      "grad_norm": 0.049229136786189835,
      "learning_rate": 2.1618094695791656e-05,
      "loss": 0.8365,
      "step": 227900
    },
    {
      "epoch": 4.00129918011201,
      "grad_norm": 0.054097480519453574,
      "learning_rate": 2.1611436407378493e-05,
      "loss": 0.8308,
      "step": 227910
    },
    {
      "epoch": 4.001474744992012,
      "grad_norm": 0.05695705889895112,
      "learning_rate": 2.1604779124833645e-05,
      "loss": 0.8379,
      "step": 227920
    },
    {
      "epoch": 4.001650309872013,
      "grad_norm": 0.03968928346594312,
      "learning_rate": 2.159812284824149e-05,
      "loss": 0.8352,
      "step": 227930
    },
    {
      "epoch": 4.001825874752015,
      "grad_norm": 0.058436739569295966,
      "learning_rate": 2.15914675776863e-05,
      "loss": 0.8422,
      "step": 227940
    },
    {
      "epoch": 4.002001439632016,
      "grad_norm": 0.05193673626605812,
      "learning_rate": 2.1584813313252438e-05,
      "loss": 0.8429,
      "step": 227950
    },
    {
      "epoch": 4.002177004512017,
      "grad_norm": 0.04539036144571544,
      "learning_rate": 2.157816005502425e-05,
      "loss": 0.8412,
      "step": 227960
    },
    {
      "epoch": 4.002352569392019,
      "grad_norm": 0.044633279325265514,
      "learning_rate": 2.1571507803085958e-05,
      "loss": 0.8347,
      "step": 227970
    },
    {
      "epoch": 4.00252813427202,
      "grad_norm": 0.06620418899545902,
      "learning_rate": 2.1564856557521877e-05,
      "loss": 0.8381,
      "step": 227980
    },
    {
      "epoch": 4.002703699152022,
      "grad_norm": 0.05065771949103179,
      "learning_rate": 2.1558206318416303e-05,
      "loss": 0.8404,
      "step": 227990
    },
    {
      "epoch": 4.002879264032023,
      "grad_norm": 0.044932618370699506,
      "learning_rate": 2.155155708585349e-05,
      "loss": 0.8416,
      "step": 228000
    },
    {
      "epoch": 4.003054828912024,
      "grad_norm": 0.05617844566079629,
      "learning_rate": 2.1544908859917663e-05,
      "loss": 0.8376,
      "step": 228010
    },
    {
      "epoch": 4.003230393792026,
      "grad_norm": 0.05420921302396657,
      "learning_rate": 2.1538261640693062e-05,
      "loss": 0.8325,
      "step": 228020
    },
    {
      "epoch": 4.003405958672027,
      "grad_norm": 0.053776786613921636,
      "learning_rate": 2.153161542826394e-05,
      "loss": 0.8299,
      "step": 228030
    },
    {
      "epoch": 4.003581523552029,
      "grad_norm": 0.05558587015981874,
      "learning_rate": 2.1524970222714455e-05,
      "loss": 0.8469,
      "step": 228040
    },
    {
      "epoch": 4.00375708843203,
      "grad_norm": 0.05382133791337714,
      "learning_rate": 2.1518326024128823e-05,
      "loss": 0.8357,
      "step": 228050
    },
    {
      "epoch": 4.003932653312032,
      "grad_norm": 0.04429267282932918,
      "learning_rate": 2.151168283259127e-05,
      "loss": 0.8331,
      "step": 228060
    },
    {
      "epoch": 4.004108218192033,
      "grad_norm": 0.05881223693866763,
      "learning_rate": 2.1505040648185896e-05,
      "loss": 0.8375,
      "step": 228070
    },
    {
      "epoch": 4.004283783072034,
      "grad_norm": 0.06304114584147148,
      "learning_rate": 2.1498399470996897e-05,
      "loss": 0.8387,
      "step": 228080
    },
    {
      "epoch": 4.004459347952036,
      "grad_norm": 0.11822737583314105,
      "learning_rate": 2.1491759301108413e-05,
      "loss": 0.8425,
      "step": 228090
    },
    {
      "epoch": 4.004634912832037,
      "grad_norm": 0.056233956238061086,
      "learning_rate": 2.14851201386046e-05,
      "loss": 0.8397,
      "step": 228100
    },
    {
      "epoch": 4.004810477712039,
      "grad_norm": 0.05198281883434514,
      "learning_rate": 2.1478481983569528e-05,
      "loss": 0.8381,
      "step": 228110
    },
    {
      "epoch": 4.00498604259204,
      "grad_norm": 0.050789157913089585,
      "learning_rate": 2.147184483608733e-05,
      "loss": 0.8403,
      "step": 228120
    },
    {
      "epoch": 4.005161607472041,
      "grad_norm": 0.047468325381398474,
      "learning_rate": 2.146520869624212e-05,
      "loss": 0.8414,
      "step": 228130
    },
    {
      "epoch": 4.005337172352043,
      "grad_norm": 0.0684045241733024,
      "learning_rate": 2.145857356411794e-05,
      "loss": 0.8445,
      "step": 228140
    },
    {
      "epoch": 4.005512737232044,
      "grad_norm": 0.04074191680106977,
      "learning_rate": 2.1451939439798873e-05,
      "loss": 0.8418,
      "step": 228150
    },
    {
      "epoch": 4.005688302112046,
      "grad_norm": 0.04274926323190166,
      "learning_rate": 2.1445306323368982e-05,
      "loss": 0.8359,
      "step": 228160
    },
    {
      "epoch": 4.005863866992047,
      "grad_norm": 0.054030238801262585,
      "learning_rate": 2.1438674214912324e-05,
      "loss": 0.8325,
      "step": 228170
    },
    {
      "epoch": 4.006039431872049,
      "grad_norm": 0.05288313714138526,
      "learning_rate": 2.1432043114512897e-05,
      "loss": 0.8312,
      "step": 228180
    },
    {
      "epoch": 4.00621499675205,
      "grad_norm": 0.04703872549835186,
      "learning_rate": 2.142541302225473e-05,
      "loss": 0.8343,
      "step": 228190
    },
    {
      "epoch": 4.006390561632051,
      "grad_norm": 0.042878462653139646,
      "learning_rate": 2.1418783938221857e-05,
      "loss": 0.8413,
      "step": 228200
    },
    {
      "epoch": 4.006566126512053,
      "grad_norm": 0.056131825493496795,
      "learning_rate": 2.1412155862498205e-05,
      "loss": 0.8366,
      "step": 228210
    },
    {
      "epoch": 4.006741691392054,
      "grad_norm": 0.07259725555726868,
      "learning_rate": 2.1405528795167802e-05,
      "loss": 0.8376,
      "step": 228220
    },
    {
      "epoch": 4.006917256272056,
      "grad_norm": 0.04912893715620019,
      "learning_rate": 2.1398902736314633e-05,
      "loss": 0.8394,
      "step": 228230
    },
    {
      "epoch": 4.007092821152057,
      "grad_norm": 0.05356669705135017,
      "learning_rate": 2.1392277686022584e-05,
      "loss": 0.842,
      "step": 228240
    },
    {
      "epoch": 4.007268386032058,
      "grad_norm": 0.07336891363176681,
      "learning_rate": 2.138565364437564e-05,
      "loss": 0.8425,
      "step": 228250
    },
    {
      "epoch": 4.00744395091206,
      "grad_norm": 0.06322305476642862,
      "learning_rate": 2.137903061145772e-05,
      "loss": 0.8381,
      "step": 228260
    },
    {
      "epoch": 4.007619515792061,
      "grad_norm": 0.04487674400774926,
      "learning_rate": 2.1372408587352775e-05,
      "loss": 0.8404,
      "step": 228270
    },
    {
      "epoch": 4.007795080672063,
      "grad_norm": 0.05593163457808573,
      "learning_rate": 2.1365787572144646e-05,
      "loss": 0.8411,
      "step": 228280
    },
    {
      "epoch": 4.007970645552064,
      "grad_norm": 0.0666449676803826,
      "learning_rate": 2.1359167565917237e-05,
      "loss": 0.8399,
      "step": 228290
    },
    {
      "epoch": 4.0081462104320655,
      "grad_norm": 0.050035597456899256,
      "learning_rate": 2.1352548568754483e-05,
      "loss": 0.8423,
      "step": 228300
    },
    {
      "epoch": 4.008321775312067,
      "grad_norm": 0.04431230228786136,
      "learning_rate": 2.134593058074016e-05,
      "loss": 0.8418,
      "step": 228310
    },
    {
      "epoch": 4.008497340192068,
      "grad_norm": 0.05154150147786471,
      "learning_rate": 2.1339313601958173e-05,
      "loss": 0.8444,
      "step": 228320
    },
    {
      "epoch": 4.0086729050720695,
      "grad_norm": 0.06364232171501465,
      "learning_rate": 2.1332697632492348e-05,
      "loss": 0.8359,
      "step": 228330
    },
    {
      "epoch": 4.008848469952071,
      "grad_norm": 0.049803764303391354,
      "learning_rate": 2.132608267242654e-05,
      "loss": 0.8458,
      "step": 228340
    },
    {
      "epoch": 4.0090240348320725,
      "grad_norm": 0.053056963271339755,
      "learning_rate": 2.1319468721844507e-05,
      "loss": 0.8417,
      "step": 228350
    },
    {
      "epoch": 4.0091995997120735,
      "grad_norm": 0.04697944963477992,
      "learning_rate": 2.1312855780830073e-05,
      "loss": 0.8418,
      "step": 228360
    },
    {
      "epoch": 4.009375164592075,
      "grad_norm": 0.05803210479643901,
      "learning_rate": 2.1306243849467053e-05,
      "loss": 0.8399,
      "step": 228370
    },
    {
      "epoch": 4.0095507294720765,
      "grad_norm": 0.05661515366311055,
      "learning_rate": 2.129963292783917e-05,
      "loss": 0.8294,
      "step": 228380
    },
    {
      "epoch": 4.0097262943520775,
      "grad_norm": 0.047304491595377,
      "learning_rate": 2.1293023016030214e-05,
      "loss": 0.8408,
      "step": 228390
    },
    {
      "epoch": 4.009901859232079,
      "grad_norm": 0.04292884025567959,
      "learning_rate": 2.1286414114123966e-05,
      "loss": 0.8361,
      "step": 228400
    },
    {
      "epoch": 4.0100774241120805,
      "grad_norm": 0.05472924179654737,
      "learning_rate": 2.12798062222041e-05,
      "loss": 0.8421,
      "step": 228410
    },
    {
      "epoch": 4.010252988992082,
      "grad_norm": 0.053983467934737496,
      "learning_rate": 2.1273199340354358e-05,
      "loss": 0.8386,
      "step": 228420
    },
    {
      "epoch": 4.010428553872083,
      "grad_norm": 0.046289193116781126,
      "learning_rate": 2.1266593468658466e-05,
      "loss": 0.8306,
      "step": 228430
    },
    {
      "epoch": 4.0106041187520844,
      "grad_norm": 0.05601504261186893,
      "learning_rate": 2.125998860720014e-05,
      "loss": 0.8398,
      "step": 228440
    },
    {
      "epoch": 4.010779683632086,
      "grad_norm": 0.054285550175944615,
      "learning_rate": 2.125338475606301e-05,
      "loss": 0.8425,
      "step": 228450
    },
    {
      "epoch": 4.010955248512087,
      "grad_norm": 0.07392760727142308,
      "learning_rate": 2.1246781915330782e-05,
      "loss": 0.8409,
      "step": 228460
    },
    {
      "epoch": 4.011130813392089,
      "grad_norm": 0.057907368671282954,
      "learning_rate": 2.1240180085087145e-05,
      "loss": 0.8395,
      "step": 228470
    },
    {
      "epoch": 4.01130637827209,
      "grad_norm": 0.05155061607121357,
      "learning_rate": 2.1233579265415682e-05,
      "loss": 0.8428,
      "step": 228480
    },
    {
      "epoch": 4.011481943152092,
      "grad_norm": 0.06077826723396317,
      "learning_rate": 2.1226979456400052e-05,
      "loss": 0.8364,
      "step": 228490
    },
    {
      "epoch": 4.011657508032093,
      "grad_norm": 0.07068341476732239,
      "learning_rate": 2.1220380658123888e-05,
      "loss": 0.8321,
      "step": 228500
    },
    {
      "epoch": 4.011833072912094,
      "grad_norm": 0.04661791102362994,
      "learning_rate": 2.1213782870670805e-05,
      "loss": 0.8415,
      "step": 228510
    },
    {
      "epoch": 4.012008637792096,
      "grad_norm": 0.0673298098082687,
      "learning_rate": 2.1207186094124372e-05,
      "loss": 0.8393,
      "step": 228520
    },
    {
      "epoch": 4.012184202672097,
      "grad_norm": 0.05359314444011101,
      "learning_rate": 2.1200590328568174e-05,
      "loss": 0.8348,
      "step": 228530
    },
    {
      "epoch": 4.012359767552099,
      "grad_norm": 0.056685701720829755,
      "learning_rate": 2.1193995574085816e-05,
      "loss": 0.8465,
      "step": 228540
    },
    {
      "epoch": 4.0125353324321,
      "grad_norm": 0.046000728534165376,
      "learning_rate": 2.1187401830760807e-05,
      "loss": 0.8382,
      "step": 228550
    },
    {
      "epoch": 4.012710897312101,
      "grad_norm": 0.04170575597062454,
      "learning_rate": 2.1180809098676716e-05,
      "loss": 0.8444,
      "step": 228560
    },
    {
      "epoch": 4.012886462192103,
      "grad_norm": 0.06877559088509268,
      "learning_rate": 2.117421737791709e-05,
      "loss": 0.8366,
      "step": 228570
    },
    {
      "epoch": 4.013062027072104,
      "grad_norm": 0.05102166288526095,
      "learning_rate": 2.1167626668565407e-05,
      "loss": 0.8398,
      "step": 228580
    },
    {
      "epoch": 4.013237591952106,
      "grad_norm": 0.05866404421887002,
      "learning_rate": 2.116103697070519e-05,
      "loss": 0.8362,
      "step": 228590
    },
    {
      "epoch": 4.013413156832107,
      "grad_norm": 0.051097366702653656,
      "learning_rate": 2.1154448284419935e-05,
      "loss": 0.8352,
      "step": 228600
    },
    {
      "epoch": 4.013588721712109,
      "grad_norm": 0.05332691387194018,
      "learning_rate": 2.1147860609793152e-05,
      "loss": 0.8349,
      "step": 228610
    },
    {
      "epoch": 4.01376428659211,
      "grad_norm": 0.0688926607579383,
      "learning_rate": 2.114127394690826e-05,
      "loss": 0.837,
      "step": 228620
    },
    {
      "epoch": 4.013939851472111,
      "grad_norm": 0.08593423603153874,
      "learning_rate": 2.1134688295848717e-05,
      "loss": 0.8369,
      "step": 228630
    },
    {
      "epoch": 4.014115416352113,
      "grad_norm": 0.043268393333409046,
      "learning_rate": 2.112810365669801e-05,
      "loss": 0.8347,
      "step": 228640
    },
    {
      "epoch": 4.014290981232114,
      "grad_norm": 0.07001389681159687,
      "learning_rate": 2.112152002953952e-05,
      "loss": 0.8417,
      "step": 228650
    },
    {
      "epoch": 4.014466546112116,
      "grad_norm": 0.046871567810866585,
      "learning_rate": 2.1114937414456668e-05,
      "loss": 0.8446,
      "step": 228660
    },
    {
      "epoch": 4.014642110992117,
      "grad_norm": 0.06935162948337731,
      "learning_rate": 2.1108355811532874e-05,
      "loss": 0.8313,
      "step": 228670
    },
    {
      "epoch": 4.014817675872118,
      "grad_norm": 0.046904488096247754,
      "learning_rate": 2.1101775220851556e-05,
      "loss": 0.8461,
      "step": 228680
    },
    {
      "epoch": 4.01499324075212,
      "grad_norm": 0.05080309763260092,
      "learning_rate": 2.1095195642496023e-05,
      "loss": 0.8333,
      "step": 228690
    },
    {
      "epoch": 4.015168805632121,
      "grad_norm": 0.048576151866846745,
      "learning_rate": 2.1088617076549674e-05,
      "loss": 0.8348,
      "step": 228700
    },
    {
      "epoch": 4.015344370512123,
      "grad_norm": 0.04706149783808083,
      "learning_rate": 2.1082039523095895e-05,
      "loss": 0.8441,
      "step": 228710
    },
    {
      "epoch": 4.015519935392124,
      "grad_norm": 0.046641459304434814,
      "learning_rate": 2.1075462982217963e-05,
      "loss": 0.8347,
      "step": 228720
    },
    {
      "epoch": 4.015695500272126,
      "grad_norm": 0.06544096926669513,
      "learning_rate": 2.1068887453999226e-05,
      "loss": 0.8439,
      "step": 228730
    },
    {
      "epoch": 4.015871065152127,
      "grad_norm": 0.06029203194313376,
      "learning_rate": 2.106231293852304e-05,
      "loss": 0.8348,
      "step": 228740
    },
    {
      "epoch": 4.016046630032128,
      "grad_norm": 0.06946006227116787,
      "learning_rate": 2.1055739435872634e-05,
      "loss": 0.8423,
      "step": 228750
    },
    {
      "epoch": 4.01622219491213,
      "grad_norm": 0.0556699317354452,
      "learning_rate": 2.1049166946131335e-05,
      "loss": 0.8397,
      "step": 228760
    },
    {
      "epoch": 4.016397759792131,
      "grad_norm": 0.05493212006811321,
      "learning_rate": 2.1042595469382417e-05,
      "loss": 0.8525,
      "step": 228770
    },
    {
      "epoch": 4.016573324672133,
      "grad_norm": 0.05001980939518627,
      "learning_rate": 2.103602500570916e-05,
      "loss": 0.8479,
      "step": 228780
    },
    {
      "epoch": 4.016748889552134,
      "grad_norm": 0.054303705657458144,
      "learning_rate": 2.1029455555194765e-05,
      "loss": 0.8393,
      "step": 228790
    },
    {
      "epoch": 4.016924454432135,
      "grad_norm": 0.055052570934754526,
      "learning_rate": 2.1022887117922486e-05,
      "loss": 0.8372,
      "step": 228800
    },
    {
      "epoch": 4.017100019312137,
      "grad_norm": 0.05373870906805272,
      "learning_rate": 2.1016319693975597e-05,
      "loss": 0.8377,
      "step": 228810
    },
    {
      "epoch": 4.017275584192138,
      "grad_norm": 0.05328153057326284,
      "learning_rate": 2.100975328343723e-05,
      "loss": 0.8378,
      "step": 228820
    },
    {
      "epoch": 4.01745114907214,
      "grad_norm": 0.052321304809117224,
      "learning_rate": 2.1003187886390625e-05,
      "loss": 0.8384,
      "step": 228830
    },
    {
      "epoch": 4.017626713952141,
      "grad_norm": 0.061335780381343505,
      "learning_rate": 2.0996623502918946e-05,
      "loss": 0.8362,
      "step": 228840
    },
    {
      "epoch": 4.017802278832143,
      "grad_norm": 0.04114487331837173,
      "learning_rate": 2.0990060133105422e-05,
      "loss": 0.8366,
      "step": 228850
    },
    {
      "epoch": 4.017977843712144,
      "grad_norm": 0.04617196205787872,
      "learning_rate": 2.0983497777033143e-05,
      "loss": 0.8319,
      "step": 228860
    },
    {
      "epoch": 4.018153408592145,
      "grad_norm": 0.04575493638678533,
      "learning_rate": 2.0976936434785275e-05,
      "loss": 0.8416,
      "step": 228870
    },
    {
      "epoch": 4.018328973472147,
      "grad_norm": 0.07639492808728432,
      "learning_rate": 2.097037610644499e-05,
      "loss": 0.8444,
      "step": 228880
    },
    {
      "epoch": 4.018504538352148,
      "grad_norm": 0.05285747748745995,
      "learning_rate": 2.0963816792095344e-05,
      "loss": 0.8375,
      "step": 228890
    },
    {
      "epoch": 4.01868010323215,
      "grad_norm": 0.052585179962595374,
      "learning_rate": 2.0957258491819484e-05,
      "loss": 0.8395,
      "step": 228900
    },
    {
      "epoch": 4.018855668112151,
      "grad_norm": 0.04465796165355224,
      "learning_rate": 2.0950701205700523e-05,
      "loss": 0.8452,
      "step": 228910
    },
    {
      "epoch": 4.019031232992152,
      "grad_norm": 0.044262827250961076,
      "learning_rate": 2.0944144933821504e-05,
      "loss": 0.8422,
      "step": 228920
    },
    {
      "epoch": 4.019206797872154,
      "grad_norm": 0.04681755782526959,
      "learning_rate": 2.0937589676265502e-05,
      "loss": 0.8395,
      "step": 228930
    },
    {
      "epoch": 4.019382362752155,
      "grad_norm": 0.05565973803937903,
      "learning_rate": 2.0931035433115582e-05,
      "loss": 0.8423,
      "step": 228940
    },
    {
      "epoch": 4.019557927632157,
      "grad_norm": 0.0548074844126927,
      "learning_rate": 2.0924482204454814e-05,
      "loss": 0.8495,
      "step": 228950
    },
    {
      "epoch": 4.019733492512158,
      "grad_norm": 0.048430253059878255,
      "learning_rate": 2.091792999036617e-05,
      "loss": 0.8398,
      "step": 228960
    },
    {
      "epoch": 4.01990905739216,
      "grad_norm": 0.05777988932159632,
      "learning_rate": 2.0911378790932716e-05,
      "loss": 0.8386,
      "step": 228970
    },
    {
      "epoch": 4.020084622272161,
      "grad_norm": 0.059178659428584024,
      "learning_rate": 2.090482860623745e-05,
      "loss": 0.8356,
      "step": 228980
    },
    {
      "epoch": 4.020260187152162,
      "grad_norm": 0.05119836354099496,
      "learning_rate": 2.0898279436363346e-05,
      "loss": 0.8403,
      "step": 228990
    },
    {
      "epoch": 4.020435752032164,
      "grad_norm": 0.04063604745279614,
      "learning_rate": 2.0891731281393385e-05,
      "loss": 0.8395,
      "step": 229000
    },
    {
      "epoch": 4.020611316912165,
      "grad_norm": 0.04188981306875498,
      "learning_rate": 2.0885184141410542e-05,
      "loss": 0.8414,
      "step": 229010
    },
    {
      "epoch": 4.020786881792167,
      "grad_norm": 0.0520519229039192,
      "learning_rate": 2.08786380164978e-05,
      "loss": 0.8319,
      "step": 229020
    },
    {
      "epoch": 4.020962446672168,
      "grad_norm": 0.05322617932147179,
      "learning_rate": 2.0872092906738038e-05,
      "loss": 0.8398,
      "step": 229030
    },
    {
      "epoch": 4.0211380115521695,
      "grad_norm": 0.05972623822650177,
      "learning_rate": 2.0865548812214215e-05,
      "loss": 0.8428,
      "step": 229040
    },
    {
      "epoch": 4.021313576432171,
      "grad_norm": 0.04614561226830447,
      "learning_rate": 2.0859005733009284e-05,
      "loss": 0.8392,
      "step": 229050
    },
    {
      "epoch": 4.021489141312172,
      "grad_norm": 0.054288280785504094,
      "learning_rate": 2.0852463669206067e-05,
      "loss": 0.8341,
      "step": 229060
    },
    {
      "epoch": 4.0216647061921735,
      "grad_norm": 0.054929403113677465,
      "learning_rate": 2.0845922620887503e-05,
      "loss": 0.8349,
      "step": 229070
    },
    {
      "epoch": 4.021840271072175,
      "grad_norm": 0.050251921852232984,
      "learning_rate": 2.0839382588136483e-05,
      "loss": 0.8437,
      "step": 229080
    },
    {
      "epoch": 4.0220158359521765,
      "grad_norm": 0.05684489863347725,
      "learning_rate": 2.0832843571035826e-05,
      "loss": 0.8379,
      "step": 229090
    },
    {
      "epoch": 4.0221914008321775,
      "grad_norm": 0.05980361390415074,
      "learning_rate": 2.08263055696684e-05,
      "loss": 0.8314,
      "step": 229100
    },
    {
      "epoch": 4.0223669657121786,
      "grad_norm": 0.05434443604606556,
      "learning_rate": 2.081976858411706e-05,
      "loss": 0.8314,
      "step": 229110
    },
    {
      "epoch": 4.0225425305921805,
      "grad_norm": 0.05395855289602521,
      "learning_rate": 2.0813232614464632e-05,
      "loss": 0.839,
      "step": 229120
    },
    {
      "epoch": 4.0227180954721815,
      "grad_norm": 0.05326344455832936,
      "learning_rate": 2.08066976607939e-05,
      "loss": 0.8402,
      "step": 229130
    },
    {
      "epoch": 4.022893660352183,
      "grad_norm": 0.05462504240120527,
      "learning_rate": 2.0800163723187667e-05,
      "loss": 0.8397,
      "step": 229140
    },
    {
      "epoch": 4.0230692252321845,
      "grad_norm": 0.05109738322699131,
      "learning_rate": 2.0793630801728766e-05,
      "loss": 0.8392,
      "step": 229150
    },
    {
      "epoch": 4.023244790112186,
      "grad_norm": 0.0781659077062586,
      "learning_rate": 2.0787098896499903e-05,
      "loss": 0.8342,
      "step": 229160
    },
    {
      "epoch": 4.023420354992187,
      "grad_norm": 0.04689314189005139,
      "learning_rate": 2.078056800758386e-05,
      "loss": 0.8451,
      "step": 229170
    },
    {
      "epoch": 4.0235959198721885,
      "grad_norm": 0.075037884509385,
      "learning_rate": 2.077403813506341e-05,
      "loss": 0.8305,
      "step": 229180
    },
    {
      "epoch": 4.02377148475219,
      "grad_norm": 0.04534386192080716,
      "learning_rate": 2.0767509279021295e-05,
      "loss": 0.8371,
      "step": 229190
    },
    {
      "epoch": 4.023947049632191,
      "grad_norm": 0.052122542147668725,
      "learning_rate": 2.076098143954018e-05,
      "loss": 0.8409,
      "step": 229200
    },
    {
      "epoch": 4.024122614512193,
      "grad_norm": 0.04731991775615175,
      "learning_rate": 2.0754454616702815e-05,
      "loss": 0.8376,
      "step": 229210
    },
    {
      "epoch": 4.024298179392194,
      "grad_norm": 0.07552432112392768,
      "learning_rate": 2.0747928810591908e-05,
      "loss": 0.8397,
      "step": 229220
    },
    {
      "epoch": 4.024473744272195,
      "grad_norm": 0.06688654622468823,
      "learning_rate": 2.07414040212901e-05,
      "loss": 0.8466,
      "step": 229230
    },
    {
      "epoch": 4.024649309152197,
      "grad_norm": 0.0595542574909324,
      "learning_rate": 2.0734880248880088e-05,
      "loss": 0.8358,
      "step": 229240
    },
    {
      "epoch": 4.024824874032198,
      "grad_norm": 0.05005220486992301,
      "learning_rate": 2.0728357493444555e-05,
      "loss": 0.8381,
      "step": 229250
    },
    {
      "epoch": 4.0250004389122,
      "grad_norm": 0.04514522852120793,
      "learning_rate": 2.072183575506608e-05,
      "loss": 0.8379,
      "step": 229260
    },
    {
      "epoch": 4.025176003792201,
      "grad_norm": 0.0654768182189137,
      "learning_rate": 2.071531503382733e-05,
      "loss": 0.8365,
      "step": 229270
    },
    {
      "epoch": 4.025351568672203,
      "grad_norm": 0.05395881682476808,
      "learning_rate": 2.070879532981094e-05,
      "loss": 0.8362,
      "step": 229280
    },
    {
      "epoch": 4.025527133552204,
      "grad_norm": 0.041633417103118665,
      "learning_rate": 2.0702276643099517e-05,
      "loss": 0.8513,
      "step": 229290
    },
    {
      "epoch": 4.025702698432205,
      "grad_norm": 0.05435898619722817,
      "learning_rate": 2.069575897377561e-05,
      "loss": 0.8423,
      "step": 229300
    },
    {
      "epoch": 4.025878263312207,
      "grad_norm": 0.04517912004343175,
      "learning_rate": 2.068924232192183e-05,
      "loss": 0.8424,
      "step": 229310
    },
    {
      "epoch": 4.026053828192208,
      "grad_norm": 0.04923447716152104,
      "learning_rate": 2.0682726687620766e-05,
      "loss": 0.8492,
      "step": 229320
    },
    {
      "epoch": 4.02622939307221,
      "grad_norm": 0.05167620374334098,
      "learning_rate": 2.067621207095493e-05,
      "loss": 0.8358,
      "step": 229330
    },
    {
      "epoch": 4.026404957952211,
      "grad_norm": 0.045581397119728424,
      "learning_rate": 2.0669698472006875e-05,
      "loss": 0.8435,
      "step": 229340
    },
    {
      "epoch": 4.026580522832212,
      "grad_norm": 0.05306434862809325,
      "learning_rate": 2.0663185890859135e-05,
      "loss": 0.8328,
      "step": 229350
    },
    {
      "epoch": 4.026756087712214,
      "grad_norm": 0.04940556528165706,
      "learning_rate": 2.065667432759425e-05,
      "loss": 0.8377,
      "step": 229360
    },
    {
      "epoch": 4.026931652592215,
      "grad_norm": 0.05481786993906389,
      "learning_rate": 2.0650163782294673e-05,
      "loss": 0.8396,
      "step": 229370
    },
    {
      "epoch": 4.027107217472217,
      "grad_norm": 0.04559542492497215,
      "learning_rate": 2.0643654255042926e-05,
      "loss": 0.8397,
      "step": 229380
    },
    {
      "epoch": 4.027282782352218,
      "grad_norm": 0.05170557756554042,
      "learning_rate": 2.0637145745921502e-05,
      "loss": 0.8307,
      "step": 229390
    },
    {
      "epoch": 4.02745834723222,
      "grad_norm": 0.052129619640951495,
      "learning_rate": 2.063063825501281e-05,
      "loss": 0.8402,
      "step": 229400
    },
    {
      "epoch": 4.027633912112221,
      "grad_norm": 0.05914574118171682,
      "learning_rate": 2.062413178239935e-05,
      "loss": 0.8431,
      "step": 229410
    },
    {
      "epoch": 4.027809476992222,
      "grad_norm": 0.05003056827678073,
      "learning_rate": 2.0617626328163554e-05,
      "loss": 0.8299,
      "step": 229420
    },
    {
      "epoch": 4.027985041872224,
      "grad_norm": 0.061301070887270846,
      "learning_rate": 2.0611121892387815e-05,
      "loss": 0.8397,
      "step": 229430
    },
    {
      "epoch": 4.028160606752225,
      "grad_norm": 0.04976110724961048,
      "learning_rate": 2.060461847515458e-05,
      "loss": 0.836,
      "step": 229440
    },
    {
      "epoch": 4.028336171632227,
      "grad_norm": 0.062433010457655244,
      "learning_rate": 2.0598116076546224e-05,
      "loss": 0.8385,
      "step": 229450
    },
    {
      "epoch": 4.028511736512228,
      "grad_norm": 0.05803487456801171,
      "learning_rate": 2.059161469664518e-05,
      "loss": 0.8362,
      "step": 229460
    },
    {
      "epoch": 4.028687301392229,
      "grad_norm": 0.05662005126322928,
      "learning_rate": 2.0585114335533757e-05,
      "loss": 0.8328,
      "step": 229470
    },
    {
      "epoch": 4.028862866272231,
      "grad_norm": 0.04094585809049682,
      "learning_rate": 2.057861499329435e-05,
      "loss": 0.839,
      "step": 229480
    },
    {
      "epoch": 4.029038431152232,
      "grad_norm": 0.04869585088897707,
      "learning_rate": 2.0572116670009325e-05,
      "loss": 0.8371,
      "step": 229490
    },
    {
      "epoch": 4.029213996032234,
      "grad_norm": 0.04839143705080867,
      "learning_rate": 2.056561936576097e-05,
      "loss": 0.8431,
      "step": 229500
    },
    {
      "epoch": 4.029389560912235,
      "grad_norm": 0.06038446338335348,
      "learning_rate": 2.055912308063164e-05,
      "loss": 0.8389,
      "step": 229510
    },
    {
      "epoch": 4.029565125792237,
      "grad_norm": 0.05759718577929173,
      "learning_rate": 2.0552627814703632e-05,
      "loss": 0.8373,
      "step": 229520
    },
    {
      "epoch": 4.029740690672238,
      "grad_norm": 0.05252545832095007,
      "learning_rate": 2.054613356805927e-05,
      "loss": 0.832,
      "step": 229530
    },
    {
      "epoch": 4.029916255552239,
      "grad_norm": 0.06055301627537975,
      "learning_rate": 2.05396403407808e-05,
      "loss": 0.8376,
      "step": 229540
    },
    {
      "epoch": 4.030091820432241,
      "grad_norm": 0.053585757271932835,
      "learning_rate": 2.0533148132950494e-05,
      "loss": 0.8347,
      "step": 229550
    },
    {
      "epoch": 4.030267385312242,
      "grad_norm": 0.04672589726295374,
      "learning_rate": 2.0526656944650656e-05,
      "loss": 0.8428,
      "step": 229560
    },
    {
      "epoch": 4.030442950192244,
      "grad_norm": 0.043816027513629896,
      "learning_rate": 2.052016677596347e-05,
      "loss": 0.8468,
      "step": 229570
    },
    {
      "epoch": 4.030618515072245,
      "grad_norm": 0.053921235018129864,
      "learning_rate": 2.05136776269712e-05,
      "loss": 0.8395,
      "step": 229580
    },
    {
      "epoch": 4.030794079952246,
      "grad_norm": 0.04754632932991002,
      "learning_rate": 2.050718949775607e-05,
      "loss": 0.8338,
      "step": 229590
    },
    {
      "epoch": 4.030969644832248,
      "grad_norm": 0.05915963759175247,
      "learning_rate": 2.050070238840026e-05,
      "loss": 0.8441,
      "step": 229600
    },
    {
      "epoch": 4.031145209712249,
      "grad_norm": 0.04625418285806889,
      "learning_rate": 2.0494216298985976e-05,
      "loss": 0.8338,
      "step": 229610
    },
    {
      "epoch": 4.031320774592251,
      "grad_norm": 0.06346793586062899,
      "learning_rate": 2.048773122959541e-05,
      "loss": 0.8425,
      "step": 229620
    },
    {
      "epoch": 4.031496339472252,
      "grad_norm": 0.04412300622396203,
      "learning_rate": 2.0481247180310738e-05,
      "loss": 0.8387,
      "step": 229630
    },
    {
      "epoch": 4.031671904352254,
      "grad_norm": 0.05812929567732477,
      "learning_rate": 2.047476415121406e-05,
      "loss": 0.8407,
      "step": 229640
    },
    {
      "epoch": 4.031847469232255,
      "grad_norm": 0.0601840727737345,
      "learning_rate": 2.046828214238757e-05,
      "loss": 0.8339,
      "step": 229650
    },
    {
      "epoch": 4.032023034112256,
      "grad_norm": 0.04279342620119814,
      "learning_rate": 2.0461801153913392e-05,
      "loss": 0.8394,
      "step": 229660
    },
    {
      "epoch": 4.032198598992258,
      "grad_norm": 0.053087433378207746,
      "learning_rate": 2.045532118587361e-05,
      "loss": 0.8354,
      "step": 229670
    },
    {
      "epoch": 4.032374163872259,
      "grad_norm": 0.05272274495130778,
      "learning_rate": 2.044884223835034e-05,
      "loss": 0.8426,
      "step": 229680
    },
    {
      "epoch": 4.032549728752261,
      "grad_norm": 0.0477275245628393,
      "learning_rate": 2.0442364311425677e-05,
      "loss": 0.8357,
      "step": 229690
    },
    {
      "epoch": 4.032725293632262,
      "grad_norm": 0.0695955560168129,
      "learning_rate": 2.043588740518172e-05,
      "loss": 0.8347,
      "step": 229700
    },
    {
      "epoch": 4.032900858512264,
      "grad_norm": 0.052668811750667455,
      "learning_rate": 2.0429411519700488e-05,
      "loss": 0.8422,
      "step": 229710
    },
    {
      "epoch": 4.033076423392265,
      "grad_norm": 0.04830158515900441,
      "learning_rate": 2.0422936655064055e-05,
      "loss": 0.8402,
      "step": 229720
    },
    {
      "epoch": 4.033251988272266,
      "grad_norm": 0.0514384340550545,
      "learning_rate": 2.041646281135449e-05,
      "loss": 0.837,
      "step": 229730
    },
    {
      "epoch": 4.033427553152268,
      "grad_norm": 0.04982723282642766,
      "learning_rate": 2.040998998865374e-05,
      "loss": 0.8386,
      "step": 229740
    },
    {
      "epoch": 4.033603118032269,
      "grad_norm": 0.05509410079303451,
      "learning_rate": 2.0403518187043883e-05,
      "loss": 0.8357,
      "step": 229750
    },
    {
      "epoch": 4.033778682912271,
      "grad_norm": 0.0768989395391401,
      "learning_rate": 2.0397047406606908e-05,
      "loss": 0.8407,
      "step": 229760
    },
    {
      "epoch": 4.033954247792272,
      "grad_norm": 0.06838177819649217,
      "learning_rate": 2.0390577647424773e-05,
      "loss": 0.8405,
      "step": 229770
    },
    {
      "epoch": 4.034129812672273,
      "grad_norm": 0.07068902917764515,
      "learning_rate": 2.038410890957946e-05,
      "loss": 0.8422,
      "step": 229780
    },
    {
      "epoch": 4.034305377552275,
      "grad_norm": 0.048590413878441624,
      "learning_rate": 2.037764119315294e-05,
      "loss": 0.8431,
      "step": 229790
    },
    {
      "epoch": 4.034480942432276,
      "grad_norm": 0.043272060083146865,
      "learning_rate": 2.0371174498227176e-05,
      "loss": 0.8349,
      "step": 229800
    },
    {
      "epoch": 4.0346565073122775,
      "grad_norm": 0.057516357336976714,
      "learning_rate": 2.0364708824884066e-05,
      "loss": 0.8377,
      "step": 229810
    },
    {
      "epoch": 4.034832072192279,
      "grad_norm": 0.04684869946905158,
      "learning_rate": 2.0358244173205557e-05,
      "loss": 0.8364,
      "step": 229820
    },
    {
      "epoch": 4.0350076370722805,
      "grad_norm": 0.053271513842218045,
      "learning_rate": 2.0351780543273562e-05,
      "loss": 0.8466,
      "step": 229830
    },
    {
      "epoch": 4.0351832019522815,
      "grad_norm": 0.07199149123656212,
      "learning_rate": 2.0345317935169944e-05,
      "loss": 0.8418,
      "step": 229840
    },
    {
      "epoch": 4.035358766832283,
      "grad_norm": 0.04472567682236191,
      "learning_rate": 2.0338856348976597e-05,
      "loss": 0.8303,
      "step": 229850
    },
    {
      "epoch": 4.0355343317122845,
      "grad_norm": 0.050332364208328526,
      "learning_rate": 2.03323957847754e-05,
      "loss": 0.8294,
      "step": 229860
    },
    {
      "epoch": 4.0357098965922855,
      "grad_norm": 0.06642933934758856,
      "learning_rate": 2.0325936242648247e-05,
      "loss": 0.8394,
      "step": 229870
    },
    {
      "epoch": 4.035885461472287,
      "grad_norm": 0.05359448357430296,
      "learning_rate": 2.031947772267691e-05,
      "loss": 0.8362,
      "step": 229880
    },
    {
      "epoch": 4.0360610263522885,
      "grad_norm": 0.0505188978734351,
      "learning_rate": 2.031302022494325e-05,
      "loss": 0.8403,
      "step": 229890
    },
    {
      "epoch": 4.0362365912322895,
      "grad_norm": 0.04318628101969457,
      "learning_rate": 2.0306563749529113e-05,
      "loss": 0.8348,
      "step": 229900
    },
    {
      "epoch": 4.036412156112291,
      "grad_norm": 0.055413898845277906,
      "learning_rate": 2.0300108296516247e-05,
      "loss": 0.8411,
      "step": 229910
    },
    {
      "epoch": 4.0365877209922925,
      "grad_norm": 0.05096788522882167,
      "learning_rate": 2.0293653865986485e-05,
      "loss": 0.8327,
      "step": 229920
    },
    {
      "epoch": 4.036763285872294,
      "grad_norm": 0.06667513818980107,
      "learning_rate": 2.028720045802162e-05,
      "loss": 0.8406,
      "step": 229930
    },
    {
      "epoch": 4.036938850752295,
      "grad_norm": 0.05396540024801403,
      "learning_rate": 2.0280748072703364e-05,
      "loss": 0.8375,
      "step": 229940
    },
    {
      "epoch": 4.037114415632297,
      "grad_norm": 0.06333452103054876,
      "learning_rate": 2.027429671011351e-05,
      "loss": 0.8469,
      "step": 229950
    },
    {
      "epoch": 4.037289980512298,
      "grad_norm": 0.04528334837384292,
      "learning_rate": 2.026784637033378e-05,
      "loss": 0.8361,
      "step": 229960
    },
    {
      "epoch": 4.037465545392299,
      "grad_norm": 0.04951291152002022,
      "learning_rate": 2.0261397053445926e-05,
      "loss": 0.8338,
      "step": 229970
    },
    {
      "epoch": 4.037641110272301,
      "grad_norm": 0.0618531626243401,
      "learning_rate": 2.0254948759531634e-05,
      "loss": 0.839,
      "step": 229980
    },
    {
      "epoch": 4.037816675152302,
      "grad_norm": 0.05484235075869022,
      "learning_rate": 2.0248501488672606e-05,
      "loss": 0.8355,
      "step": 229990
    },
    {
      "epoch": 4.037992240032304,
      "grad_norm": 0.06750076993132127,
      "learning_rate": 2.0242055240950567e-05,
      "loss": 0.8332,
      "step": 230000
    },
    {
      "epoch": 4.038167804912305,
      "grad_norm": 0.049873657401548185,
      "learning_rate": 2.023561001644714e-05,
      "loss": 0.8509,
      "step": 230010
    },
    {
      "epoch": 4.038343369792306,
      "grad_norm": 0.05452785793869996,
      "learning_rate": 2.022916581524402e-05,
      "loss": 0.8391,
      "step": 230020
    },
    {
      "epoch": 4.038518934672308,
      "grad_norm": 0.057166574503669285,
      "learning_rate": 2.0222722637422846e-05,
      "loss": 0.8415,
      "step": 230030
    },
    {
      "epoch": 4.038694499552309,
      "grad_norm": 0.046948715657761854,
      "learning_rate": 2.0216280483065275e-05,
      "loss": 0.8389,
      "step": 230040
    },
    {
      "epoch": 4.038870064432311,
      "grad_norm": 0.04258036543896514,
      "learning_rate": 2.020983935225289e-05,
      "loss": 0.8374,
      "step": 230050
    },
    {
      "epoch": 4.039045629312312,
      "grad_norm": 0.05209110647792407,
      "learning_rate": 2.0203399245067324e-05,
      "loss": 0.8373,
      "step": 230060
    },
    {
      "epoch": 4.039221194192314,
      "grad_norm": 0.048227514964742646,
      "learning_rate": 2.0196960161590197e-05,
      "loss": 0.8443,
      "step": 230070
    },
    {
      "epoch": 4.039396759072315,
      "grad_norm": 0.05012820101499754,
      "learning_rate": 2.019052210190306e-05,
      "loss": 0.8364,
      "step": 230080
    },
    {
      "epoch": 4.039572323952316,
      "grad_norm": 0.04181778123755861,
      "learning_rate": 2.0184085066087477e-05,
      "loss": 0.8431,
      "step": 230090
    },
    {
      "epoch": 4.039747888832318,
      "grad_norm": 0.05299582869710773,
      "learning_rate": 2.017764905422506e-05,
      "loss": 0.8466,
      "step": 230100
    },
    {
      "epoch": 4.039923453712319,
      "grad_norm": 0.05312111860646189,
      "learning_rate": 2.0171214066397286e-05,
      "loss": 0.8323,
      "step": 230110
    },
    {
      "epoch": 4.040099018592321,
      "grad_norm": 0.052189424024329197,
      "learning_rate": 2.0164780102685732e-05,
      "loss": 0.8305,
      "step": 230120
    },
    {
      "epoch": 4.040274583472322,
      "grad_norm": 0.048171353266782006,
      "learning_rate": 2.0158347163171897e-05,
      "loss": 0.8376,
      "step": 230130
    },
    {
      "epoch": 4.040450148352323,
      "grad_norm": 0.05319065647007432,
      "learning_rate": 2.0151915247937325e-05,
      "loss": 0.8378,
      "step": 230140
    },
    {
      "epoch": 4.040625713232325,
      "grad_norm": 0.06024830298609819,
      "learning_rate": 2.014548435706345e-05,
      "loss": 0.8444,
      "step": 230150
    },
    {
      "epoch": 4.040801278112326,
      "grad_norm": 0.0514173042108325,
      "learning_rate": 2.0139054490631786e-05,
      "loss": 0.8457,
      "step": 230160
    },
    {
      "epoch": 4.040976842992328,
      "grad_norm": 0.05107704039166751,
      "learning_rate": 2.0132625648723828e-05,
      "loss": 0.8362,
      "step": 230170
    },
    {
      "epoch": 4.041152407872329,
      "grad_norm": 0.05095917109593666,
      "learning_rate": 2.0126197831420977e-05,
      "loss": 0.8412,
      "step": 230180
    },
    {
      "epoch": 4.041327972752331,
      "grad_norm": 0.06418623251053185,
      "learning_rate": 2.0119771038804694e-05,
      "loss": 0.8371,
      "step": 230190
    },
    {
      "epoch": 4.041503537632332,
      "grad_norm": 0.050206475350025224,
      "learning_rate": 2.011334527095642e-05,
      "loss": 0.8406,
      "step": 230200
    },
    {
      "epoch": 4.041679102512333,
      "grad_norm": 0.054709607289363166,
      "learning_rate": 2.0106920527957588e-05,
      "loss": 0.8318,
      "step": 230210
    },
    {
      "epoch": 4.041854667392335,
      "grad_norm": 0.05724008255540709,
      "learning_rate": 2.0100496809889547e-05,
      "loss": 0.8429,
      "step": 230220
    },
    {
      "epoch": 4.042030232272336,
      "grad_norm": 0.055719897566878195,
      "learning_rate": 2.0094074116833714e-05,
      "loss": 0.8378,
      "step": 230230
    },
    {
      "epoch": 4.042205797152338,
      "grad_norm": 0.06624272850085407,
      "learning_rate": 2.0087652448871486e-05,
      "loss": 0.8441,
      "step": 230240
    },
    {
      "epoch": 4.042381362032339,
      "grad_norm": 0.05206164759595803,
      "learning_rate": 2.0081231806084195e-05,
      "loss": 0.8411,
      "step": 230250
    },
    {
      "epoch": 4.04255692691234,
      "grad_norm": 0.0528436280273677,
      "learning_rate": 2.0074812188553196e-05,
      "loss": 0.8329,
      "step": 230260
    },
    {
      "epoch": 4.042732491792342,
      "grad_norm": 0.04788627971280511,
      "learning_rate": 2.0068393596359872e-05,
      "loss": 0.8431,
      "step": 230270
    },
    {
      "epoch": 4.042908056672343,
      "grad_norm": 0.04940408867778806,
      "learning_rate": 2.0061976029585478e-05,
      "loss": 0.8388,
      "step": 230280
    },
    {
      "epoch": 4.043083621552345,
      "grad_norm": 0.054902354716712884,
      "learning_rate": 2.005555948831136e-05,
      "loss": 0.8368,
      "step": 230290
    },
    {
      "epoch": 4.043259186432346,
      "grad_norm": 0.06202236078802633,
      "learning_rate": 2.004914397261882e-05,
      "loss": 0.8421,
      "step": 230300
    },
    {
      "epoch": 4.043434751312348,
      "grad_norm": 0.045171609344725555,
      "learning_rate": 2.0042729482589164e-05,
      "loss": 0.8368,
      "step": 230310
    },
    {
      "epoch": 4.043610316192349,
      "grad_norm": 0.05482344779950296,
      "learning_rate": 2.003631601830361e-05,
      "loss": 0.8388,
      "step": 230320
    },
    {
      "epoch": 4.04378588107235,
      "grad_norm": 0.06530297545001046,
      "learning_rate": 2.0029903579843456e-05,
      "loss": 0.832,
      "step": 230330
    },
    {
      "epoch": 4.043961445952352,
      "grad_norm": 0.05697092816764783,
      "learning_rate": 2.0023492167289963e-05,
      "loss": 0.8281,
      "step": 230340
    },
    {
      "epoch": 4.044137010832353,
      "grad_norm": 0.0765800560707525,
      "learning_rate": 2.001708178072432e-05,
      "loss": 0.8413,
      "step": 230350
    },
    {
      "epoch": 4.044312575712355,
      "grad_norm": 0.045872073450713785,
      "learning_rate": 2.001067242022777e-05,
      "loss": 0.8385,
      "step": 230360
    },
    {
      "epoch": 4.044488140592356,
      "grad_norm": 0.047939326991458274,
      "learning_rate": 2.000426408588153e-05,
      "loss": 0.8386,
      "step": 230370
    },
    {
      "epoch": 4.044663705472357,
      "grad_norm": 0.049990038238962906,
      "learning_rate": 1.9997856777766798e-05,
      "loss": 0.8311,
      "step": 230380
    },
    {
      "epoch": 4.044839270352359,
      "grad_norm": 0.049917642562413024,
      "learning_rate": 1.9991450495964726e-05,
      "loss": 0.8387,
      "step": 230390
    },
    {
      "epoch": 4.04501483523236,
      "grad_norm": 0.04712547378928479,
      "learning_rate": 1.9985045240556506e-05,
      "loss": 0.8335,
      "step": 230400
    },
    {
      "epoch": 4.045190400112362,
      "grad_norm": 0.05082934902077995,
      "learning_rate": 1.9978641011623302e-05,
      "loss": 0.8478,
      "step": 230410
    },
    {
      "epoch": 4.045365964992363,
      "grad_norm": 0.04641030214395614,
      "learning_rate": 1.9972237809246228e-05,
      "loss": 0.8405,
      "step": 230420
    },
    {
      "epoch": 4.045541529872365,
      "grad_norm": 0.054738622603056536,
      "learning_rate": 1.9965835633506438e-05,
      "loss": 0.8414,
      "step": 230430
    },
    {
      "epoch": 4.045717094752366,
      "grad_norm": 0.07488425229863281,
      "learning_rate": 1.9959434484485055e-05,
      "loss": 0.8487,
      "step": 230440
    },
    {
      "epoch": 4.045892659632367,
      "grad_norm": 0.07982584572235056,
      "learning_rate": 1.9953034362263147e-05,
      "loss": 0.8386,
      "step": 230450
    },
    {
      "epoch": 4.046068224512369,
      "grad_norm": 0.04524325877420843,
      "learning_rate": 1.994663526692182e-05,
      "loss": 0.8297,
      "step": 230460
    },
    {
      "epoch": 4.04624378939237,
      "grad_norm": 0.04216529853922622,
      "learning_rate": 1.9940237198542165e-05,
      "loss": 0.8398,
      "step": 230470
    },
    {
      "epoch": 4.046419354272372,
      "grad_norm": 0.04192049900651561,
      "learning_rate": 1.9933840157205254e-05,
      "loss": 0.8377,
      "step": 230480
    },
    {
      "epoch": 4.046594919152373,
      "grad_norm": 0.048124354817627724,
      "learning_rate": 1.9927444142992103e-05,
      "loss": 0.8331,
      "step": 230490
    },
    {
      "epoch": 4.046770484032375,
      "grad_norm": 0.051050015380614856,
      "learning_rate": 1.992104915598378e-05,
      "loss": 0.8371,
      "step": 230500
    },
    {
      "epoch": 4.046946048912376,
      "grad_norm": 0.053698282029411806,
      "learning_rate": 1.9914655196261326e-05,
      "loss": 0.8451,
      "step": 230510
    },
    {
      "epoch": 4.047121613792377,
      "grad_norm": 0.05529156612040191,
      "learning_rate": 1.99082622639057e-05,
      "loss": 0.8345,
      "step": 230520
    },
    {
      "epoch": 4.047297178672379,
      "grad_norm": 0.06273283350336162,
      "learning_rate": 1.990187035899794e-05,
      "loss": 0.8454,
      "step": 230530
    },
    {
      "epoch": 4.04747274355238,
      "grad_norm": 0.055558940827344735,
      "learning_rate": 1.9895479481619026e-05,
      "loss": 0.8375,
      "step": 230540
    },
    {
      "epoch": 4.0476483084323815,
      "grad_norm": 0.047528202833406606,
      "learning_rate": 1.9889089631849942e-05,
      "loss": 0.831,
      "step": 230550
    },
    {
      "epoch": 4.047823873312383,
      "grad_norm": 0.14866635725108476,
      "learning_rate": 1.9882700809771628e-05,
      "loss": 0.8404,
      "step": 230560
    },
    {
      "epoch": 4.047999438192384,
      "grad_norm": 0.04182305810595356,
      "learning_rate": 1.987631301546503e-05,
      "loss": 0.8389,
      "step": 230570
    },
    {
      "epoch": 4.0481750030723855,
      "grad_norm": 0.061612582197164166,
      "learning_rate": 1.9869926249011117e-05,
      "loss": 0.8394,
      "step": 230580
    },
    {
      "epoch": 4.048350567952387,
      "grad_norm": 0.04732722722980612,
      "learning_rate": 1.986354051049077e-05,
      "loss": 0.8399,
      "step": 230590
    },
    {
      "epoch": 4.0485261328323885,
      "grad_norm": 0.04893108330147805,
      "learning_rate": 1.9857155799984913e-05,
      "loss": 0.8417,
      "step": 230600
    },
    {
      "epoch": 4.0487016977123895,
      "grad_norm": 0.051792623485462,
      "learning_rate": 1.985077211757447e-05,
      "loss": 0.8418,
      "step": 230610
    },
    {
      "epoch": 4.0488772625923914,
      "grad_norm": 0.04715118133839669,
      "learning_rate": 1.9844389463340266e-05,
      "loss": 0.842,
      "step": 230620
    },
    {
      "epoch": 4.0490528274723925,
      "grad_norm": 0.060536656146474556,
      "learning_rate": 1.9838007837363197e-05,
      "loss": 0.841,
      "step": 230630
    },
    {
      "epoch": 4.0492283923523935,
      "grad_norm": 0.04782490086695707,
      "learning_rate": 1.9831627239724136e-05,
      "loss": 0.8443,
      "step": 230640
    },
    {
      "epoch": 4.049403957232395,
      "grad_norm": 0.04754082076086008,
      "learning_rate": 1.9825247670503925e-05,
      "loss": 0.8382,
      "step": 230650
    },
    {
      "epoch": 4.0495795221123965,
      "grad_norm": 0.04948610023088782,
      "learning_rate": 1.981886912978337e-05,
      "loss": 0.8377,
      "step": 230660
    },
    {
      "epoch": 4.049755086992398,
      "grad_norm": 0.05000974842035047,
      "learning_rate": 1.9812491617643292e-05,
      "loss": 0.84,
      "step": 230670
    },
    {
      "epoch": 4.049930651872399,
      "grad_norm": 0.0451623428361975,
      "learning_rate": 1.9806115134164538e-05,
      "loss": 0.8349,
      "step": 230680
    },
    {
      "epoch": 4.0501062167524005,
      "grad_norm": 0.04658477130723671,
      "learning_rate": 1.9799739679427827e-05,
      "loss": 0.8342,
      "step": 230690
    },
    {
      "epoch": 4.050281781632402,
      "grad_norm": 0.05599991882704879,
      "learning_rate": 1.9793365253513997e-05,
      "loss": 0.846,
      "step": 230700
    },
    {
      "epoch": 4.050457346512403,
      "grad_norm": 0.04953797933931095,
      "learning_rate": 1.978699185650378e-05,
      "loss": 0.84,
      "step": 230710
    },
    {
      "epoch": 4.050632911392405,
      "grad_norm": 0.05662331545608743,
      "learning_rate": 1.9780619488477965e-05,
      "loss": 0.8406,
      "step": 230720
    },
    {
      "epoch": 4.050808476272406,
      "grad_norm": 0.05575493405496299,
      "learning_rate": 1.9774248149517245e-05,
      "loss": 0.8446,
      "step": 230730
    },
    {
      "epoch": 4.050984041152408,
      "grad_norm": 0.06226016212259846,
      "learning_rate": 1.9767877839702363e-05,
      "loss": 0.8345,
      "step": 230740
    },
    {
      "epoch": 4.051159606032409,
      "grad_norm": 0.05724124131392584,
      "learning_rate": 1.9761508559114067e-05,
      "loss": 0.8455,
      "step": 230750
    },
    {
      "epoch": 4.05133517091241,
      "grad_norm": 0.045820814214711396,
      "learning_rate": 1.9755140307833e-05,
      "loss": 0.8346,
      "step": 230760
    },
    {
      "epoch": 4.051510735792412,
      "grad_norm": 0.07422352317505747,
      "learning_rate": 1.9748773085939874e-05,
      "loss": 0.832,
      "step": 230770
    },
    {
      "epoch": 4.051686300672413,
      "grad_norm": 0.055134903917068726,
      "learning_rate": 1.974240689351539e-05,
      "loss": 0.8356,
      "step": 230780
    },
    {
      "epoch": 4.051861865552415,
      "grad_norm": 0.04688600175498848,
      "learning_rate": 1.973604173064017e-05,
      "loss": 0.8433,
      "step": 230790
    },
    {
      "epoch": 4.052037430432416,
      "grad_norm": 0.047104197131246084,
      "learning_rate": 1.972967759739486e-05,
      "loss": 0.8385,
      "step": 230800
    },
    {
      "epoch": 4.052212995312417,
      "grad_norm": 0.04483057720051681,
      "learning_rate": 1.9723314493860113e-05,
      "loss": 0.8434,
      "step": 230810
    },
    {
      "epoch": 4.052388560192419,
      "grad_norm": 0.04808688767881454,
      "learning_rate": 1.9716952420116585e-05,
      "loss": 0.8419,
      "step": 230820
    },
    {
      "epoch": 4.05256412507242,
      "grad_norm": 0.04845711607550383,
      "learning_rate": 1.9710591376244814e-05,
      "loss": 0.8434,
      "step": 230830
    },
    {
      "epoch": 4.052739689952422,
      "grad_norm": 0.05978333639268888,
      "learning_rate": 1.970423136232543e-05,
      "loss": 0.8436,
      "step": 230840
    },
    {
      "epoch": 4.052915254832423,
      "grad_norm": 0.04793370218551738,
      "learning_rate": 1.9697872378439027e-05,
      "loss": 0.8376,
      "step": 230850
    },
    {
      "epoch": 4.053090819712425,
      "grad_norm": 0.06817083177106423,
      "learning_rate": 1.9691514424666156e-05,
      "loss": 0.8408,
      "step": 230860
    },
    {
      "epoch": 4.053266384592426,
      "grad_norm": 0.042440060326242195,
      "learning_rate": 1.9685157501087366e-05,
      "loss": 0.8423,
      "step": 230870
    },
    {
      "epoch": 4.053441949472427,
      "grad_norm": 0.053302001603458264,
      "learning_rate": 1.9678801607783228e-05,
      "loss": 0.8348,
      "step": 230880
    },
    {
      "epoch": 4.053617514352429,
      "grad_norm": 0.05105966591051054,
      "learning_rate": 1.9672446744834276e-05,
      "loss": 0.8464,
      "step": 230890
    },
    {
      "epoch": 4.05379307923243,
      "grad_norm": 0.05536884241687365,
      "learning_rate": 1.966609291232099e-05,
      "loss": 0.8449,
      "step": 230900
    },
    {
      "epoch": 4.053968644112432,
      "grad_norm": 0.04680235134077853,
      "learning_rate": 1.965974011032389e-05,
      "loss": 0.8439,
      "step": 230910
    },
    {
      "epoch": 4.054144208992433,
      "grad_norm": 0.052827786558423184,
      "learning_rate": 1.9653388338923494e-05,
      "loss": 0.8377,
      "step": 230920
    },
    {
      "epoch": 4.054319773872434,
      "grad_norm": 0.05925802334627201,
      "learning_rate": 1.9647037598200236e-05,
      "loss": 0.8467,
      "step": 230930
    },
    {
      "epoch": 4.054495338752436,
      "grad_norm": 0.04734261710690397,
      "learning_rate": 1.964068788823461e-05,
      "loss": 0.8473,
      "step": 230940
    },
    {
      "epoch": 4.054670903632437,
      "grad_norm": 0.057233015521233443,
      "learning_rate": 1.9634339209107074e-05,
      "loss": 0.8433,
      "step": 230950
    },
    {
      "epoch": 4.054846468512439,
      "grad_norm": 0.050791406646814384,
      "learning_rate": 1.9627991560898033e-05,
      "loss": 0.8336,
      "step": 230960
    },
    {
      "epoch": 4.05502203339244,
      "grad_norm": 0.0511118396481289,
      "learning_rate": 1.9621644943687937e-05,
      "loss": 0.8347,
      "step": 230970
    },
    {
      "epoch": 4.055197598272442,
      "grad_norm": 0.07123364151921885,
      "learning_rate": 1.961529935755719e-05,
      "loss": 0.8365,
      "step": 230980
    },
    {
      "epoch": 4.055373163152443,
      "grad_norm": 0.04283127142495337,
      "learning_rate": 1.9608954802586224e-05,
      "loss": 0.8349,
      "step": 230990
    },
    {
      "epoch": 4.055548728032444,
      "grad_norm": 0.04446109402751642,
      "learning_rate": 1.9602611278855384e-05,
      "loss": 0.8376,
      "step": 231000
    },
    {
      "epoch": 4.055724292912446,
      "grad_norm": 0.04441492431452363,
      "learning_rate": 1.959626878644504e-05,
      "loss": 0.8401,
      "step": 231010
    },
    {
      "epoch": 4.055899857792447,
      "grad_norm": 0.048440088970209856,
      "learning_rate": 1.9589927325435616e-05,
      "loss": 0.8297,
      "step": 231020
    },
    {
      "epoch": 4.056075422672449,
      "grad_norm": 0.05341169988061649,
      "learning_rate": 1.9583586895907377e-05,
      "loss": 0.843,
      "step": 231030
    },
    {
      "epoch": 4.05625098755245,
      "grad_norm": 0.05490835406372275,
      "learning_rate": 1.95772474979407e-05,
      "loss": 0.8429,
      "step": 231040
    },
    {
      "epoch": 4.056426552432451,
      "grad_norm": 0.07448919092952341,
      "learning_rate": 1.9570909131615907e-05,
      "loss": 0.8439,
      "step": 231050
    },
    {
      "epoch": 4.056602117312453,
      "grad_norm": 0.05444306030800364,
      "learning_rate": 1.956457179701332e-05,
      "loss": 0.8443,
      "step": 231060
    },
    {
      "epoch": 4.056777682192454,
      "grad_norm": 0.0527080771243837,
      "learning_rate": 1.955823549421319e-05,
      "loss": 0.8458,
      "step": 231070
    },
    {
      "epoch": 4.056953247072456,
      "grad_norm": 0.053211752840372786,
      "learning_rate": 1.955190022329582e-05,
      "loss": 0.8483,
      "step": 231080
    },
    {
      "epoch": 4.057128811952457,
      "grad_norm": 0.05544859020631584,
      "learning_rate": 1.9545565984341513e-05,
      "loss": 0.8384,
      "step": 231090
    },
    {
      "epoch": 4.057304376832459,
      "grad_norm": 0.07032927052062285,
      "learning_rate": 1.9539232777430476e-05,
      "loss": 0.8377,
      "step": 231100
    },
    {
      "epoch": 4.05747994171246,
      "grad_norm": 0.05242504132075124,
      "learning_rate": 1.9532900602642967e-05,
      "loss": 0.8443,
      "step": 231110
    },
    {
      "epoch": 4.057655506592461,
      "grad_norm": 0.05028043223345739,
      "learning_rate": 1.952656946005925e-05,
      "loss": 0.8395,
      "step": 231120
    },
    {
      "epoch": 4.057831071472463,
      "grad_norm": 0.052988688941566865,
      "learning_rate": 1.952023934975948e-05,
      "loss": 0.8321,
      "step": 231130
    },
    {
      "epoch": 4.058006636352464,
      "grad_norm": 0.04472317757276596,
      "learning_rate": 1.951391027182389e-05,
      "loss": 0.8398,
      "step": 231140
    },
    {
      "epoch": 4.058182201232466,
      "grad_norm": 0.04369705893361105,
      "learning_rate": 1.950758222633268e-05,
      "loss": 0.8495,
      "step": 231150
    },
    {
      "epoch": 4.058357766112467,
      "grad_norm": 0.06636777723131143,
      "learning_rate": 1.9501255213366047e-05,
      "loss": 0.8369,
      "step": 231160
    },
    {
      "epoch": 4.058533330992468,
      "grad_norm": 0.049957364971525545,
      "learning_rate": 1.9494929233004105e-05,
      "loss": 0.8379,
      "step": 231170
    },
    {
      "epoch": 4.05870889587247,
      "grad_norm": 0.04753568075900819,
      "learning_rate": 1.948860428532703e-05,
      "loss": 0.8354,
      "step": 231180
    },
    {
      "epoch": 4.058884460752471,
      "grad_norm": 0.055880241524895374,
      "learning_rate": 1.948228037041499e-05,
      "loss": 0.8439,
      "step": 231190
    },
    {
      "epoch": 4.059060025632473,
      "grad_norm": 0.047787024129376146,
      "learning_rate": 1.9475957488348044e-05,
      "loss": 0.8361,
      "step": 231200
    },
    {
      "epoch": 4.059235590512474,
      "grad_norm": 0.044636138014736884,
      "learning_rate": 1.9469635639206342e-05,
      "loss": 0.8384,
      "step": 231210
    },
    {
      "epoch": 4.059411155392476,
      "grad_norm": 0.06093169862835277,
      "learning_rate": 1.9463314823069988e-05,
      "loss": 0.8349,
      "step": 231220
    },
    {
      "epoch": 4.059586720272477,
      "grad_norm": 0.05032044317949286,
      "learning_rate": 1.945699504001908e-05,
      "loss": 0.8387,
      "step": 231230
    },
    {
      "epoch": 4.059762285152478,
      "grad_norm": 0.05113714224357784,
      "learning_rate": 1.9450676290133648e-05,
      "loss": 0.8394,
      "step": 231240
    },
    {
      "epoch": 4.05993785003248,
      "grad_norm": 0.06132724645688432,
      "learning_rate": 1.944435857349379e-05,
      "loss": 0.8377,
      "step": 231250
    },
    {
      "epoch": 4.060113414912481,
      "grad_norm": 0.05012891896340847,
      "learning_rate": 1.943804189017955e-05,
      "loss": 0.848,
      "step": 231260
    },
    {
      "epoch": 4.060288979792483,
      "grad_norm": 0.05507952314200843,
      "learning_rate": 1.9431726240270928e-05,
      "loss": 0.8421,
      "step": 231270
    },
    {
      "epoch": 4.060464544672484,
      "grad_norm": 0.03815857092978236,
      "learning_rate": 1.9425411623847967e-05,
      "loss": 0.8427,
      "step": 231280
    },
    {
      "epoch": 4.0606401095524856,
      "grad_norm": 0.062066269186314006,
      "learning_rate": 1.94190980409907e-05,
      "loss": 0.8395,
      "step": 231290
    },
    {
      "epoch": 4.060815674432487,
      "grad_norm": 0.05819735507152214,
      "learning_rate": 1.9412785491779073e-05,
      "loss": 0.8331,
      "step": 231300
    },
    {
      "epoch": 4.060991239312488,
      "grad_norm": 0.05064363702879113,
      "learning_rate": 1.940647397629309e-05,
      "loss": 0.8319,
      "step": 231310
    },
    {
      "epoch": 4.0611668041924895,
      "grad_norm": 0.057143877336356785,
      "learning_rate": 1.9400163494612716e-05,
      "loss": 0.836,
      "step": 231320
    },
    {
      "epoch": 4.061342369072491,
      "grad_norm": 0.05699978276106688,
      "learning_rate": 1.9393854046817932e-05,
      "loss": 0.8397,
      "step": 231330
    },
    {
      "epoch": 4.0615179339524925,
      "grad_norm": 0.0486315362176169,
      "learning_rate": 1.9387545632988636e-05,
      "loss": 0.843,
      "step": 231340
    },
    {
      "epoch": 4.0616934988324935,
      "grad_norm": 0.046700756035497,
      "learning_rate": 1.9381238253204783e-05,
      "loss": 0.8452,
      "step": 231350
    },
    {
      "epoch": 4.061869063712495,
      "grad_norm": 0.06905920618879362,
      "learning_rate": 1.9374931907546304e-05,
      "loss": 0.8438,
      "step": 231360
    },
    {
      "epoch": 4.0620446285924965,
      "grad_norm": 0.05266336593372966,
      "learning_rate": 1.936862659609305e-05,
      "loss": 0.8466,
      "step": 231370
    },
    {
      "epoch": 4.0622201934724975,
      "grad_norm": 0.04652467494411541,
      "learning_rate": 1.9362322318924943e-05,
      "loss": 0.8335,
      "step": 231380
    },
    {
      "epoch": 4.062395758352499,
      "grad_norm": 0.06348950727008061,
      "learning_rate": 1.935601907612187e-05,
      "loss": 0.8306,
      "step": 231390
    },
    {
      "epoch": 4.0625713232325005,
      "grad_norm": 0.051892074666541346,
      "learning_rate": 1.93497168677637e-05,
      "loss": 0.8433,
      "step": 231400
    },
    {
      "epoch": 4.062746888112502,
      "grad_norm": 0.05069974333451811,
      "learning_rate": 1.934341569393023e-05,
      "loss": 0.8394,
      "step": 231410
    },
    {
      "epoch": 4.062922452992503,
      "grad_norm": 0.04401721339856393,
      "learning_rate": 1.9337115554701348e-05,
      "loss": 0.8343,
      "step": 231420
    },
    {
      "epoch": 4.0630980178725045,
      "grad_norm": 0.059747418821036444,
      "learning_rate": 1.933081645015688e-05,
      "loss": 0.842,
      "step": 231430
    },
    {
      "epoch": 4.063273582752506,
      "grad_norm": 0.04549335603673088,
      "learning_rate": 1.9324518380376598e-05,
      "loss": 0.8393,
      "step": 231440
    },
    {
      "epoch": 4.063449147632507,
      "grad_norm": 0.041985999640050925,
      "learning_rate": 1.931822134544032e-05,
      "loss": 0.835,
      "step": 231450
    },
    {
      "epoch": 4.063624712512509,
      "grad_norm": 0.05718866927110598,
      "learning_rate": 1.9311925345427844e-05,
      "loss": 0.8389,
      "step": 231460
    },
    {
      "epoch": 4.06380027739251,
      "grad_norm": 0.051040306491554305,
      "learning_rate": 1.9305630380418925e-05,
      "loss": 0.8393,
      "step": 231470
    },
    {
      "epoch": 4.063975842272511,
      "grad_norm": 0.04285964090931558,
      "learning_rate": 1.9299336450493313e-05,
      "loss": 0.8318,
      "step": 231480
    },
    {
      "epoch": 4.064151407152513,
      "grad_norm": 0.07980636467391183,
      "learning_rate": 1.929304355573077e-05,
      "loss": 0.8342,
      "step": 231490
    },
    {
      "epoch": 4.064326972032514,
      "grad_norm": 0.0491790337690851,
      "learning_rate": 1.9286751696211052e-05,
      "loss": 0.8477,
      "step": 231500
    },
    {
      "epoch": 4.064502536912516,
      "grad_norm": 0.0478867762682142,
      "learning_rate": 1.928046087201382e-05,
      "loss": 0.8527,
      "step": 231510
    },
    {
      "epoch": 4.064678101792517,
      "grad_norm": 0.06223320771924356,
      "learning_rate": 1.927417108321881e-05,
      "loss": 0.8422,
      "step": 231520
    },
    {
      "epoch": 4.064853666672519,
      "grad_norm": 0.036126257059412734,
      "learning_rate": 1.926788232990574e-05,
      "loss": 0.8429,
      "step": 231530
    },
    {
      "epoch": 4.06502923155252,
      "grad_norm": 0.05354549850779721,
      "learning_rate": 1.9261594612154247e-05,
      "loss": 0.8369,
      "step": 231540
    },
    {
      "epoch": 4.065204796432521,
      "grad_norm": 0.062409852748057174,
      "learning_rate": 1.925530793004401e-05,
      "loss": 0.8464,
      "step": 231550
    },
    {
      "epoch": 4.065380361312523,
      "grad_norm": 0.06266499684185736,
      "learning_rate": 1.924902228365469e-05,
      "loss": 0.8463,
      "step": 231560
    },
    {
      "epoch": 4.065555926192524,
      "grad_norm": 0.056758186035428815,
      "learning_rate": 1.9242737673065945e-05,
      "loss": 0.8348,
      "step": 231570
    },
    {
      "epoch": 4.065731491072526,
      "grad_norm": 0.05953266491353704,
      "learning_rate": 1.9236454098357363e-05,
      "loss": 0.8405,
      "step": 231580
    },
    {
      "epoch": 4.065907055952527,
      "grad_norm": 0.04979564114508449,
      "learning_rate": 1.9230171559608578e-05,
      "loss": 0.8437,
      "step": 231590
    },
    {
      "epoch": 4.066082620832528,
      "grad_norm": 0.06348101298652463,
      "learning_rate": 1.922389005689921e-05,
      "loss": 0.8335,
      "step": 231600
    },
    {
      "epoch": 4.06625818571253,
      "grad_norm": 0.04679999363020437,
      "learning_rate": 1.9217609590308803e-05,
      "loss": 0.8334,
      "step": 231610
    },
    {
      "epoch": 4.066433750592531,
      "grad_norm": 0.054355880006412345,
      "learning_rate": 1.9211330159916963e-05,
      "loss": 0.8369,
      "step": 231620
    },
    {
      "epoch": 4.066609315472533,
      "grad_norm": 0.050046379272353385,
      "learning_rate": 1.920505176580327e-05,
      "loss": 0.8357,
      "step": 231630
    },
    {
      "epoch": 4.066784880352534,
      "grad_norm": 0.06168276997051324,
      "learning_rate": 1.9198774408047208e-05,
      "loss": 0.8363,
      "step": 231640
    },
    {
      "epoch": 4.066960445232536,
      "grad_norm": 0.05875192513641352,
      "learning_rate": 1.919249808672837e-05,
      "loss": 0.8341,
      "step": 231650
    },
    {
      "epoch": 4.067136010112537,
      "grad_norm": 0.059584091945962585,
      "learning_rate": 1.9186222801926237e-05,
      "loss": 0.8396,
      "step": 231660
    },
    {
      "epoch": 4.067311574992538,
      "grad_norm": 0.05115839356702313,
      "learning_rate": 1.9179948553720383e-05,
      "loss": 0.8516,
      "step": 231670
    },
    {
      "epoch": 4.06748713987254,
      "grad_norm": 0.07347945363678789,
      "learning_rate": 1.9173675342190236e-05,
      "loss": 0.8432,
      "step": 231680
    },
    {
      "epoch": 4.067662704752541,
      "grad_norm": 0.05432548354276119,
      "learning_rate": 1.9167403167415297e-05,
      "loss": 0.837,
      "step": 231690
    },
    {
      "epoch": 4.067838269632543,
      "grad_norm": 0.07735568184152238,
      "learning_rate": 1.9161132029475076e-05,
      "loss": 0.8323,
      "step": 231700
    },
    {
      "epoch": 4.068013834512544,
      "grad_norm": 0.04785642859154197,
      "learning_rate": 1.915486192844896e-05,
      "loss": 0.8372,
      "step": 231710
    },
    {
      "epoch": 4.068189399392545,
      "grad_norm": 0.05019688843516327,
      "learning_rate": 1.9148592864416445e-05,
      "loss": 0.8464,
      "step": 231720
    },
    {
      "epoch": 4.068364964272547,
      "grad_norm": 0.06852023757778283,
      "learning_rate": 1.9142324837456935e-05,
      "loss": 0.8432,
      "step": 231730
    },
    {
      "epoch": 4.068540529152548,
      "grad_norm": 0.05383735275889387,
      "learning_rate": 1.9136057847649876e-05,
      "loss": 0.8367,
      "step": 231740
    },
    {
      "epoch": 4.06871609403255,
      "grad_norm": 0.06776904770809017,
      "learning_rate": 1.9129791895074636e-05,
      "loss": 0.8405,
      "step": 231750
    },
    {
      "epoch": 4.068891658912551,
      "grad_norm": 0.07389485918866771,
      "learning_rate": 1.912352697981063e-05,
      "loss": 0.835,
      "step": 231760
    },
    {
      "epoch": 4.069067223792553,
      "grad_norm": 0.06272097563954862,
      "learning_rate": 1.911726310193724e-05,
      "loss": 0.8431,
      "step": 231770
    },
    {
      "epoch": 4.069242788672554,
      "grad_norm": 0.04394470434029819,
      "learning_rate": 1.91110002615338e-05,
      "loss": 0.8383,
      "step": 231780
    },
    {
      "epoch": 4.069418353552555,
      "grad_norm": 0.05066793154899464,
      "learning_rate": 1.910473845867969e-05,
      "loss": 0.8397,
      "step": 231790
    },
    {
      "epoch": 4.069593918432557,
      "grad_norm": 0.04458892626304208,
      "learning_rate": 1.9098477693454246e-05,
      "loss": 0.8441,
      "step": 231800
    },
    {
      "epoch": 4.069769483312558,
      "grad_norm": 0.06388760762603107,
      "learning_rate": 1.9092217965936772e-05,
      "loss": 0.8379,
      "step": 231810
    },
    {
      "epoch": 4.06994504819256,
      "grad_norm": 0.04925793942506397,
      "learning_rate": 1.9085959276206593e-05,
      "loss": 0.8397,
      "step": 231820
    },
    {
      "epoch": 4.070120613072561,
      "grad_norm": 0.05650332209594922,
      "learning_rate": 1.907970162434301e-05,
      "loss": 0.8418,
      "step": 231830
    },
    {
      "epoch": 4.070296177952562,
      "grad_norm": 0.04598957616999553,
      "learning_rate": 1.9073445010425335e-05,
      "loss": 0.835,
      "step": 231840
    },
    {
      "epoch": 4.070471742832564,
      "grad_norm": 0.05757685177633966,
      "learning_rate": 1.9067189434532777e-05,
      "loss": 0.8433,
      "step": 231850
    },
    {
      "epoch": 4.070647307712565,
      "grad_norm": 0.06181082364671672,
      "learning_rate": 1.9060934896744632e-05,
      "loss": 0.8423,
      "step": 231860
    },
    {
      "epoch": 4.070822872592567,
      "grad_norm": 0.05721278379345325,
      "learning_rate": 1.9054681397140176e-05,
      "loss": 0.8397,
      "step": 231870
    },
    {
      "epoch": 4.070998437472568,
      "grad_norm": 0.05229045461944193,
      "learning_rate": 1.9048428935798594e-05,
      "loss": 0.8402,
      "step": 231880
    },
    {
      "epoch": 4.07117400235257,
      "grad_norm": 0.05003819099273758,
      "learning_rate": 1.904217751279911e-05,
      "loss": 0.8462,
      "step": 231890
    },
    {
      "epoch": 4.071349567232571,
      "grad_norm": 0.042698616241057766,
      "learning_rate": 1.9035927128220953e-05,
      "loss": 0.838,
      "step": 231900
    },
    {
      "epoch": 4.071525132112572,
      "grad_norm": 0.0619597233991457,
      "learning_rate": 1.9029677782143324e-05,
      "loss": 0.8362,
      "step": 231910
    },
    {
      "epoch": 4.071700696992574,
      "grad_norm": 0.052630695938754676,
      "learning_rate": 1.9023429474645375e-05,
      "loss": 0.8313,
      "step": 231920
    },
    {
      "epoch": 4.071876261872575,
      "grad_norm": 0.07442551294712521,
      "learning_rate": 1.9017182205806267e-05,
      "loss": 0.8409,
      "step": 231930
    },
    {
      "epoch": 4.072051826752577,
      "grad_norm": 0.059212676720308055,
      "learning_rate": 1.9010935975705214e-05,
      "loss": 0.8378,
      "step": 231940
    },
    {
      "epoch": 4.072227391632578,
      "grad_norm": 0.05104594867061404,
      "learning_rate": 1.9004690784421285e-05,
      "loss": 0.8383,
      "step": 231950
    },
    {
      "epoch": 4.07240295651258,
      "grad_norm": 0.05902508270622011,
      "learning_rate": 1.899844663203363e-05,
      "loss": 0.8466,
      "step": 231960
    },
    {
      "epoch": 4.072578521392581,
      "grad_norm": 0.05933065086260977,
      "learning_rate": 1.899220351862141e-05,
      "loss": 0.8463,
      "step": 231970
    },
    {
      "epoch": 4.072754086272582,
      "grad_norm": 0.05802793542237467,
      "learning_rate": 1.8985961444263655e-05,
      "loss": 0.842,
      "step": 231980
    },
    {
      "epoch": 4.072929651152584,
      "grad_norm": 0.057240913729545734,
      "learning_rate": 1.8979720409039487e-05,
      "loss": 0.8473,
      "step": 231990
    },
    {
      "epoch": 4.073105216032585,
      "grad_norm": 0.06780564322840385,
      "learning_rate": 1.8973480413027982e-05,
      "loss": 0.8372,
      "step": 232000
    },
    {
      "epoch": 4.073280780912587,
      "grad_norm": 0.046020965337144594,
      "learning_rate": 1.896724145630823e-05,
      "loss": 0.8401,
      "step": 232010
    },
    {
      "epoch": 4.073456345792588,
      "grad_norm": 0.05337883014248958,
      "learning_rate": 1.8961003538959223e-05,
      "loss": 0.8295,
      "step": 232020
    },
    {
      "epoch": 4.073631910672589,
      "grad_norm": 0.04651664968927143,
      "learning_rate": 1.8954766661060024e-05,
      "loss": 0.84,
      "step": 232030
    },
    {
      "epoch": 4.073807475552591,
      "grad_norm": 0.04571866301597431,
      "learning_rate": 1.894853082268967e-05,
      "loss": 0.8413,
      "step": 232040
    },
    {
      "epoch": 4.073983040432592,
      "grad_norm": 0.055881728330920796,
      "learning_rate": 1.8942296023927138e-05,
      "loss": 0.8448,
      "step": 232050
    },
    {
      "epoch": 4.0741586053125936,
      "grad_norm": 0.060752750090542484,
      "learning_rate": 1.8936062264851444e-05,
      "loss": 0.8336,
      "step": 232060
    },
    {
      "epoch": 4.074334170192595,
      "grad_norm": 0.05917656847292841,
      "learning_rate": 1.8929829545541564e-05,
      "loss": 0.8359,
      "step": 232070
    },
    {
      "epoch": 4.0745097350725965,
      "grad_norm": 0.052800191312936236,
      "learning_rate": 1.8923597866076506e-05,
      "loss": 0.8444,
      "step": 232080
    },
    {
      "epoch": 4.0746852999525975,
      "grad_norm": 0.048801409945758274,
      "learning_rate": 1.8917367226535156e-05,
      "loss": 0.8439,
      "step": 232090
    },
    {
      "epoch": 4.074860864832599,
      "grad_norm": 0.04632996164114891,
      "learning_rate": 1.89111376269965e-05,
      "loss": 0.8396,
      "step": 232100
    },
    {
      "epoch": 4.0750364297126005,
      "grad_norm": 0.045791933478224356,
      "learning_rate": 1.8904909067539487e-05,
      "loss": 0.8355,
      "step": 232110
    },
    {
      "epoch": 4.0752119945926015,
      "grad_norm": 0.05753970192290261,
      "learning_rate": 1.8898681548242987e-05,
      "loss": 0.8409,
      "step": 232120
    },
    {
      "epoch": 4.0753875594726034,
      "grad_norm": 0.06164038044609515,
      "learning_rate": 1.889245506918593e-05,
      "loss": 0.8369,
      "step": 232130
    },
    {
      "epoch": 4.0755631243526045,
      "grad_norm": 0.058160504381626225,
      "learning_rate": 1.888622963044722e-05,
      "loss": 0.8425,
      "step": 232140
    },
    {
      "epoch": 4.0757386892326055,
      "grad_norm": 0.05174820248214816,
      "learning_rate": 1.8880005232105695e-05,
      "loss": 0.8418,
      "step": 232150
    },
    {
      "epoch": 4.075914254112607,
      "grad_norm": 0.04717376238371022,
      "learning_rate": 1.887378187424024e-05,
      "loss": 0.8446,
      "step": 232160
    },
    {
      "epoch": 4.0760898189926085,
      "grad_norm": 0.04291939495297339,
      "learning_rate": 1.8867559556929712e-05,
      "loss": 0.8353,
      "step": 232170
    },
    {
      "epoch": 4.07626538387261,
      "grad_norm": 0.05839313676528621,
      "learning_rate": 1.8861338280252972e-05,
      "loss": 0.8342,
      "step": 232180
    },
    {
      "epoch": 4.076440948752611,
      "grad_norm": 0.060954301816119674,
      "learning_rate": 1.8855118044288792e-05,
      "loss": 0.8378,
      "step": 232190
    },
    {
      "epoch": 4.076616513632613,
      "grad_norm": 0.06137332011720221,
      "learning_rate": 1.8848898849115997e-05,
      "loss": 0.8445,
      "step": 232200
    },
    {
      "epoch": 4.076792078512614,
      "grad_norm": 0.06421750133589402,
      "learning_rate": 1.8842680694813433e-05,
      "loss": 0.8385,
      "step": 232210
    },
    {
      "epoch": 4.076967643392615,
      "grad_norm": 0.057943836089584984,
      "learning_rate": 1.8836463581459828e-05,
      "loss": 0.8409,
      "step": 232220
    },
    {
      "epoch": 4.077143208272617,
      "grad_norm": 0.05570686029253389,
      "learning_rate": 1.8830247509133974e-05,
      "loss": 0.8366,
      "step": 232230
    },
    {
      "epoch": 4.077318773152618,
      "grad_norm": 0.0868491835881463,
      "learning_rate": 1.8824032477914626e-05,
      "loss": 0.8477,
      "step": 232240
    },
    {
      "epoch": 4.07749433803262,
      "grad_norm": 0.04862829005910427,
      "learning_rate": 1.881781848788057e-05,
      "loss": 0.8466,
      "step": 232250
    },
    {
      "epoch": 4.077669902912621,
      "grad_norm": 0.0594013112575181,
      "learning_rate": 1.8811605539110472e-05,
      "loss": 0.8392,
      "step": 232260
    },
    {
      "epoch": 4.077845467792622,
      "grad_norm": 0.058358232688048,
      "learning_rate": 1.880539363168309e-05,
      "loss": 0.8422,
      "step": 232270
    },
    {
      "epoch": 4.078021032672624,
      "grad_norm": 0.04362042424506637,
      "learning_rate": 1.8799182765677142e-05,
      "loss": 0.8422,
      "step": 232280
    },
    {
      "epoch": 4.078196597552625,
      "grad_norm": 0.04689758451234852,
      "learning_rate": 1.8792972941171275e-05,
      "loss": 0.8382,
      "step": 232290
    },
    {
      "epoch": 4.078372162432627,
      "grad_norm": 0.05005749640259704,
      "learning_rate": 1.8786764158244195e-05,
      "loss": 0.8376,
      "step": 232300
    },
    {
      "epoch": 4.078547727312628,
      "grad_norm": 0.053901677462497695,
      "learning_rate": 1.8780556416974602e-05,
      "loss": 0.8418,
      "step": 232310
    },
    {
      "epoch": 4.07872329219263,
      "grad_norm": 0.054742945416316996,
      "learning_rate": 1.877434971744109e-05,
      "loss": 0.843,
      "step": 232320
    },
    {
      "epoch": 4.078898857072631,
      "grad_norm": 0.05518309368364387,
      "learning_rate": 1.8768144059722322e-05,
      "loss": 0.8401,
      "step": 232330
    },
    {
      "epoch": 4.079074421952632,
      "grad_norm": 0.07732217486196705,
      "learning_rate": 1.8761939443896924e-05,
      "loss": 0.8303,
      "step": 232340
    },
    {
      "epoch": 4.079249986832634,
      "grad_norm": 0.05507932173267617,
      "learning_rate": 1.8755735870043542e-05,
      "loss": 0.8412,
      "step": 232350
    },
    {
      "epoch": 4.079425551712635,
      "grad_norm": 0.05008629649150695,
      "learning_rate": 1.8749533338240728e-05,
      "loss": 0.8386,
      "step": 232360
    },
    {
      "epoch": 4.079601116592637,
      "grad_norm": 0.0479942088328177,
      "learning_rate": 1.8743331848567092e-05,
      "loss": 0.8336,
      "step": 232370
    },
    {
      "epoch": 4.079776681472638,
      "grad_norm": 0.039527362526217386,
      "learning_rate": 1.8737131401101226e-05,
      "loss": 0.8358,
      "step": 232380
    },
    {
      "epoch": 4.079952246352639,
      "grad_norm": 0.052550855386714744,
      "learning_rate": 1.8730931995921643e-05,
      "loss": 0.846,
      "step": 232390
    },
    {
      "epoch": 4.080127811232641,
      "grad_norm": 0.06623227345928479,
      "learning_rate": 1.8724733633106927e-05,
      "loss": 0.8433,
      "step": 232400
    },
    {
      "epoch": 4.080303376112642,
      "grad_norm": 0.06683389561589564,
      "learning_rate": 1.8718536312735598e-05,
      "loss": 0.8419,
      "step": 232410
    },
    {
      "epoch": 4.080478940992644,
      "grad_norm": 0.06261289145682652,
      "learning_rate": 1.8712340034886207e-05,
      "loss": 0.8467,
      "step": 232420
    },
    {
      "epoch": 4.080654505872645,
      "grad_norm": 0.050628349357312856,
      "learning_rate": 1.870614479963722e-05,
      "loss": 0.848,
      "step": 232430
    },
    {
      "epoch": 4.080830070752647,
      "grad_norm": 0.058978914059611046,
      "learning_rate": 1.8699950607067138e-05,
      "loss": 0.8346,
      "step": 232440
    },
    {
      "epoch": 4.081005635632648,
      "grad_norm": 0.05828792841724674,
      "learning_rate": 1.869375745725449e-05,
      "loss": 0.8465,
      "step": 232450
    },
    {
      "epoch": 4.081181200512649,
      "grad_norm": 0.057343245315436095,
      "learning_rate": 1.868756535027768e-05,
      "loss": 0.8401,
      "step": 232460
    },
    {
      "epoch": 4.081356765392651,
      "grad_norm": 0.07598158295073791,
      "learning_rate": 1.868137428621519e-05,
      "loss": 0.8355,
      "step": 232470
    },
    {
      "epoch": 4.081532330272652,
      "grad_norm": 0.05277949078841818,
      "learning_rate": 1.8675184265145497e-05,
      "loss": 0.8389,
      "step": 232480
    },
    {
      "epoch": 4.081707895152654,
      "grad_norm": 0.061109132932524395,
      "learning_rate": 1.866899528714696e-05,
      "loss": 0.8428,
      "step": 232490
    },
    {
      "epoch": 4.081883460032655,
      "grad_norm": 0.0646947099650042,
      "learning_rate": 1.8662807352298044e-05,
      "loss": 0.8422,
      "step": 232500
    },
    {
      "epoch": 4.082059024912656,
      "grad_norm": 0.04574593908837088,
      "learning_rate": 1.8656620460677125e-05,
      "loss": 0.8423,
      "step": 232510
    },
    {
      "epoch": 4.082234589792658,
      "grad_norm": 0.05243665555481325,
      "learning_rate": 1.8650434612362645e-05,
      "loss": 0.8389,
      "step": 232520
    },
    {
      "epoch": 4.082410154672659,
      "grad_norm": 0.06353797709150275,
      "learning_rate": 1.8644249807432908e-05,
      "loss": 0.8359,
      "step": 232530
    },
    {
      "epoch": 4.082585719552661,
      "grad_norm": 0.046949574881012666,
      "learning_rate": 1.8638066045966314e-05,
      "loss": 0.8387,
      "step": 232540
    },
    {
      "epoch": 4.082761284432662,
      "grad_norm": 0.04382628422790846,
      "learning_rate": 1.8631883328041228e-05,
      "loss": 0.8346,
      "step": 232550
    },
    {
      "epoch": 4.082936849312664,
      "grad_norm": 0.06842379076502902,
      "learning_rate": 1.8625701653735942e-05,
      "loss": 0.8481,
      "step": 232560
    },
    {
      "epoch": 4.083112414192665,
      "grad_norm": 0.05815154628723508,
      "learning_rate": 1.86195210231288e-05,
      "loss": 0.8367,
      "step": 232570
    },
    {
      "epoch": 4.083287979072666,
      "grad_norm": 0.05337918638360118,
      "learning_rate": 1.861334143629811e-05,
      "loss": 0.8348,
      "step": 232580
    },
    {
      "epoch": 4.083463543952668,
      "grad_norm": 0.049806083314725386,
      "learning_rate": 1.860716289332221e-05,
      "loss": 0.8508,
      "step": 232590
    },
    {
      "epoch": 4.083639108832669,
      "grad_norm": 0.04587333145666694,
      "learning_rate": 1.860098539427932e-05,
      "loss": 0.8363,
      "step": 232600
    },
    {
      "epoch": 4.083814673712671,
      "grad_norm": 0.04681498222881848,
      "learning_rate": 1.859480893924774e-05,
      "loss": 0.8394,
      "step": 232610
    },
    {
      "epoch": 4.083990238592672,
      "grad_norm": 0.04701641783079446,
      "learning_rate": 1.8588633528305742e-05,
      "loss": 0.834,
      "step": 232620
    },
    {
      "epoch": 4.084165803472674,
      "grad_norm": 0.04461893529935489,
      "learning_rate": 1.8582459161531526e-05,
      "loss": 0.8361,
      "step": 232630
    },
    {
      "epoch": 4.084341368352675,
      "grad_norm": 0.06908147253317377,
      "learning_rate": 1.8576285839003352e-05,
      "loss": 0.8423,
      "step": 232640
    },
    {
      "epoch": 4.084516933232676,
      "grad_norm": 0.05594422573152666,
      "learning_rate": 1.857011356079945e-05,
      "loss": 0.8492,
      "step": 232650
    },
    {
      "epoch": 4.084692498112678,
      "grad_norm": 0.04824984270621321,
      "learning_rate": 1.8563942326997986e-05,
      "loss": 0.8494,
      "step": 232660
    },
    {
      "epoch": 4.084868062992679,
      "grad_norm": 0.045889138582903714,
      "learning_rate": 1.8557772137677176e-05,
      "loss": 0.8406,
      "step": 232670
    },
    {
      "epoch": 4.085043627872681,
      "grad_norm": 0.04115440940883448,
      "learning_rate": 1.8551602992915188e-05,
      "loss": 0.8368,
      "step": 232680
    },
    {
      "epoch": 4.085219192752682,
      "grad_norm": 0.049671463394112,
      "learning_rate": 1.8545434892790223e-05,
      "loss": 0.8464,
      "step": 232690
    },
    {
      "epoch": 4.085394757632683,
      "grad_norm": 0.05321915096616634,
      "learning_rate": 1.8539267837380367e-05,
      "loss": 0.8399,
      "step": 232700
    },
    {
      "epoch": 4.085570322512685,
      "grad_norm": 0.059352757062620665,
      "learning_rate": 1.8533101826763798e-05,
      "loss": 0.8393,
      "step": 232710
    },
    {
      "epoch": 4.085745887392686,
      "grad_norm": 0.04195227087016098,
      "learning_rate": 1.8526936861018653e-05,
      "loss": 0.8413,
      "step": 232720
    },
    {
      "epoch": 4.085921452272688,
      "grad_norm": 0.046384972702589164,
      "learning_rate": 1.8520772940223e-05,
      "loss": 0.8382,
      "step": 232730
    },
    {
      "epoch": 4.086097017152689,
      "grad_norm": 0.06193230963292315,
      "learning_rate": 1.8514610064454953e-05,
      "loss": 0.8351,
      "step": 232740
    },
    {
      "epoch": 4.086272582032691,
      "grad_norm": 0.053881072799145745,
      "learning_rate": 1.8508448233792622e-05,
      "loss": 0.8422,
      "step": 232750
    },
    {
      "epoch": 4.086448146912692,
      "grad_norm": 0.06667844116391637,
      "learning_rate": 1.8502287448314073e-05,
      "loss": 0.837,
      "step": 232760
    },
    {
      "epoch": 4.086623711792693,
      "grad_norm": 0.0406441501081758,
      "learning_rate": 1.8496127708097333e-05,
      "loss": 0.8405,
      "step": 232770
    },
    {
      "epoch": 4.086799276672695,
      "grad_norm": 0.050888615194887056,
      "learning_rate": 1.848996901322047e-05,
      "loss": 0.8409,
      "step": 232780
    },
    {
      "epoch": 4.086974841552696,
      "grad_norm": 0.05202353192285021,
      "learning_rate": 1.848381136376153e-05,
      "loss": 0.8412,
      "step": 232790
    },
    {
      "epoch": 4.087150406432698,
      "grad_norm": 0.050114849978260045,
      "learning_rate": 1.8477654759798496e-05,
      "loss": 0.8459,
      "step": 232800
    },
    {
      "epoch": 4.087325971312699,
      "grad_norm": 0.04268633354325176,
      "learning_rate": 1.847149920140939e-05,
      "loss": 0.8355,
      "step": 232810
    },
    {
      "epoch": 4.0875015361927,
      "grad_norm": 0.05794187224054888,
      "learning_rate": 1.8465344688672224e-05,
      "loss": 0.8281,
      "step": 232820
    },
    {
      "epoch": 4.0876771010727015,
      "grad_norm": 0.04589682445696792,
      "learning_rate": 1.845919122166494e-05,
      "loss": 0.8334,
      "step": 232830
    },
    {
      "epoch": 4.087852665952703,
      "grad_norm": 0.04977059520610898,
      "learning_rate": 1.845303880046552e-05,
      "loss": 0.8442,
      "step": 232840
    },
    {
      "epoch": 4.0880282308327045,
      "grad_norm": 0.049849714828336264,
      "learning_rate": 1.8446887425151914e-05,
      "loss": 0.8412,
      "step": 232850
    },
    {
      "epoch": 4.0882037957127055,
      "grad_norm": 0.04256013520777272,
      "learning_rate": 1.8440737095802102e-05,
      "loss": 0.8374,
      "step": 232860
    },
    {
      "epoch": 4.0883793605927075,
      "grad_norm": 0.054472712948363894,
      "learning_rate": 1.8434587812493938e-05,
      "loss": 0.827,
      "step": 232870
    },
    {
      "epoch": 4.0885549254727085,
      "grad_norm": 0.0436622389598755,
      "learning_rate": 1.842843957530538e-05,
      "loss": 0.8402,
      "step": 232880
    },
    {
      "epoch": 4.0887304903527095,
      "grad_norm": 0.048621296147271925,
      "learning_rate": 1.842229238431433e-05,
      "loss": 0.8387,
      "step": 232890
    },
    {
      "epoch": 4.0889060552327114,
      "grad_norm": 0.0695110546441136,
      "learning_rate": 1.8416146239598638e-05,
      "loss": 0.8333,
      "step": 232900
    },
    {
      "epoch": 4.0890816201127125,
      "grad_norm": 0.042853287202203785,
      "learning_rate": 1.8410001141236213e-05,
      "loss": 0.8397,
      "step": 232910
    },
    {
      "epoch": 4.089257184992714,
      "grad_norm": 0.050323678413430814,
      "learning_rate": 1.8403857089304887e-05,
      "loss": 0.8442,
      "step": 232920
    },
    {
      "epoch": 4.089432749872715,
      "grad_norm": 0.052115623178527325,
      "learning_rate": 1.8397714083882554e-05,
      "loss": 0.8383,
      "step": 232930
    },
    {
      "epoch": 4.0896083147527165,
      "grad_norm": 0.0413142443974431,
      "learning_rate": 1.8391572125046985e-05,
      "loss": 0.8435,
      "step": 232940
    },
    {
      "epoch": 4.089783879632718,
      "grad_norm": 0.048707530023586014,
      "learning_rate": 1.8385431212876032e-05,
      "loss": 0.8365,
      "step": 232950
    },
    {
      "epoch": 4.089959444512719,
      "grad_norm": 0.050119646423522936,
      "learning_rate": 1.837929134744753e-05,
      "loss": 0.8352,
      "step": 232960
    },
    {
      "epoch": 4.090135009392721,
      "grad_norm": 0.0468437683838989,
      "learning_rate": 1.837315252883921e-05,
      "loss": 0.8387,
      "step": 232970
    },
    {
      "epoch": 4.090310574272722,
      "grad_norm": 0.05024785641494346,
      "learning_rate": 1.8367014757128884e-05,
      "loss": 0.8457,
      "step": 232980
    },
    {
      "epoch": 4.090486139152724,
      "grad_norm": 0.043589268848240256,
      "learning_rate": 1.8360878032394337e-05,
      "loss": 0.8319,
      "step": 232990
    },
    {
      "epoch": 4.090661704032725,
      "grad_norm": 0.048507727923426024,
      "learning_rate": 1.8354742354713275e-05,
      "loss": 0.8396,
      "step": 233000
    },
    {
      "epoch": 4.090837268912726,
      "grad_norm": 0.05299516331342044,
      "learning_rate": 1.8348607724163474e-05,
      "loss": 0.8373,
      "step": 233010
    },
    {
      "epoch": 4.091012833792728,
      "grad_norm": 0.04996041171477614,
      "learning_rate": 1.834247414082266e-05,
      "loss": 0.8353,
      "step": 233020
    },
    {
      "epoch": 4.091188398672729,
      "grad_norm": 0.04704166398782162,
      "learning_rate": 1.8336341604768556e-05,
      "loss": 0.8356,
      "step": 233030
    },
    {
      "epoch": 4.091363963552731,
      "grad_norm": 0.041295761164709274,
      "learning_rate": 1.8330210116078815e-05,
      "loss": 0.8451,
      "step": 233040
    },
    {
      "epoch": 4.091539528432732,
      "grad_norm": 0.05391981670961067,
      "learning_rate": 1.832407967483116e-05,
      "loss": 0.8445,
      "step": 233050
    },
    {
      "epoch": 4.091715093312733,
      "grad_norm": 0.08687075077461112,
      "learning_rate": 1.83179502811033e-05,
      "loss": 0.8421,
      "step": 233060
    },
    {
      "epoch": 4.091890658192735,
      "grad_norm": 0.047478101534727306,
      "learning_rate": 1.8311821934972813e-05,
      "loss": 0.8309,
      "step": 233070
    },
    {
      "epoch": 4.092066223072736,
      "grad_norm": 0.05650383025850332,
      "learning_rate": 1.8305694636517393e-05,
      "loss": 0.8433,
      "step": 233080
    },
    {
      "epoch": 4.092241787952738,
      "grad_norm": 0.050116071210993575,
      "learning_rate": 1.8299568385814677e-05,
      "loss": 0.844,
      "step": 233090
    },
    {
      "epoch": 4.092417352832739,
      "grad_norm": 0.0583513054486585,
      "learning_rate": 1.8293443182942296e-05,
      "loss": 0.8389,
      "step": 233100
    },
    {
      "epoch": 4.092592917712741,
      "grad_norm": 0.048428652444198764,
      "learning_rate": 1.8287319027977826e-05,
      "loss": 0.8322,
      "step": 233110
    },
    {
      "epoch": 4.092768482592742,
      "grad_norm": 0.04560649728051521,
      "learning_rate": 1.8281195920998863e-05,
      "loss": 0.8347,
      "step": 233120
    },
    {
      "epoch": 4.092944047472743,
      "grad_norm": 0.059023708291281646,
      "learning_rate": 1.8275073862083035e-05,
      "loss": 0.8437,
      "step": 233130
    },
    {
      "epoch": 4.093119612352745,
      "grad_norm": 0.04514433272381797,
      "learning_rate": 1.8268952851307847e-05,
      "loss": 0.8448,
      "step": 233140
    },
    {
      "epoch": 4.093295177232746,
      "grad_norm": 0.061180280336695765,
      "learning_rate": 1.8262832888750883e-05,
      "loss": 0.8383,
      "step": 233150
    },
    {
      "epoch": 4.093470742112748,
      "grad_norm": 0.047405310107196195,
      "learning_rate": 1.8256713974489698e-05,
      "loss": 0.8423,
      "step": 233160
    },
    {
      "epoch": 4.093646306992749,
      "grad_norm": 0.0644385639552445,
      "learning_rate": 1.8250596108601787e-05,
      "loss": 0.8343,
      "step": 233170
    },
    {
      "epoch": 4.09382187187275,
      "grad_norm": 0.06077066463067004,
      "learning_rate": 1.8244479291164686e-05,
      "loss": 0.8316,
      "step": 233180
    },
    {
      "epoch": 4.093997436752752,
      "grad_norm": 0.0668470530430684,
      "learning_rate": 1.8238363522255897e-05,
      "loss": 0.8433,
      "step": 233190
    },
    {
      "epoch": 4.094173001632753,
      "grad_norm": 0.05844814558590147,
      "learning_rate": 1.823224880195291e-05,
      "loss": 0.8389,
      "step": 233200
    },
    {
      "epoch": 4.094348566512755,
      "grad_norm": 0.05430920236008408,
      "learning_rate": 1.8226135130333177e-05,
      "loss": 0.843,
      "step": 233210
    },
    {
      "epoch": 4.094524131392756,
      "grad_norm": 0.0545037678603426,
      "learning_rate": 1.822002250747418e-05,
      "loss": 0.8421,
      "step": 233220
    },
    {
      "epoch": 4.094699696272758,
      "grad_norm": 0.055962272566363734,
      "learning_rate": 1.8213910933453384e-05,
      "loss": 0.8337,
      "step": 233230
    },
    {
      "epoch": 4.094875261152759,
      "grad_norm": 0.06411045179098769,
      "learning_rate": 1.820780040834818e-05,
      "loss": 0.8332,
      "step": 233240
    },
    {
      "epoch": 4.09505082603276,
      "grad_norm": 0.05588211793486126,
      "learning_rate": 1.8201690932236013e-05,
      "loss": 0.8432,
      "step": 233250
    },
    {
      "epoch": 4.095226390912762,
      "grad_norm": 0.050076013843710736,
      "learning_rate": 1.8195582505194285e-05,
      "loss": 0.8463,
      "step": 233260
    },
    {
      "epoch": 4.095401955792763,
      "grad_norm": 0.05973457974885507,
      "learning_rate": 1.8189475127300427e-05,
      "loss": 0.8486,
      "step": 233270
    },
    {
      "epoch": 4.095577520672765,
      "grad_norm": 0.0690430328011517,
      "learning_rate": 1.8183368798631752e-05,
      "loss": 0.8379,
      "step": 233280
    },
    {
      "epoch": 4.095753085552766,
      "grad_norm": 0.05396149575681845,
      "learning_rate": 1.8177263519265686e-05,
      "loss": 0.8357,
      "step": 233290
    },
    {
      "epoch": 4.095928650432767,
      "grad_norm": 0.055837088435569504,
      "learning_rate": 1.8171159289279567e-05,
      "loss": 0.8454,
      "step": 233300
    },
    {
      "epoch": 4.096104215312769,
      "grad_norm": 0.06239318670214508,
      "learning_rate": 1.816505610875073e-05,
      "loss": 0.8399,
      "step": 233310
    },
    {
      "epoch": 4.09627978019277,
      "grad_norm": 0.05925679994062814,
      "learning_rate": 1.8158953977756492e-05,
      "loss": 0.8403,
      "step": 233320
    },
    {
      "epoch": 4.096455345072772,
      "grad_norm": 0.05620066996172539,
      "learning_rate": 1.815285289637421e-05,
      "loss": 0.8358,
      "step": 233330
    },
    {
      "epoch": 4.096630909952773,
      "grad_norm": 0.05733214838478868,
      "learning_rate": 1.8146752864681142e-05,
      "loss": 0.8399,
      "step": 233340
    },
    {
      "epoch": 4.096806474832775,
      "grad_norm": 0.06833194199768243,
      "learning_rate": 1.8140653882754592e-05,
      "loss": 0.8435,
      "step": 233350
    },
    {
      "epoch": 4.096982039712776,
      "grad_norm": 0.05346964513928917,
      "learning_rate": 1.8134555950671837e-05,
      "loss": 0.8327,
      "step": 233360
    },
    {
      "epoch": 4.097157604592777,
      "grad_norm": 0.060917166625946643,
      "learning_rate": 1.812845906851016e-05,
      "loss": 0.8451,
      "step": 233370
    },
    {
      "epoch": 4.097333169472779,
      "grad_norm": 0.04689440849062854,
      "learning_rate": 1.812236323634677e-05,
      "loss": 0.8441,
      "step": 233380
    },
    {
      "epoch": 4.09750873435278,
      "grad_norm": 0.05435368174059268,
      "learning_rate": 1.8116268454258918e-05,
      "loss": 0.8371,
      "step": 233390
    },
    {
      "epoch": 4.097684299232782,
      "grad_norm": 0.04252032563831401,
      "learning_rate": 1.8110174722323857e-05,
      "loss": 0.8398,
      "step": 233400
    },
    {
      "epoch": 4.097859864112783,
      "grad_norm": 0.05048437645936816,
      "learning_rate": 1.8104082040618732e-05,
      "loss": 0.8414,
      "step": 233410
    },
    {
      "epoch": 4.098035428992784,
      "grad_norm": 0.052211778602426456,
      "learning_rate": 1.8097990409220785e-05,
      "loss": 0.8401,
      "step": 233420
    },
    {
      "epoch": 4.098210993872786,
      "grad_norm": 0.0470390501502205,
      "learning_rate": 1.8091899828207182e-05,
      "loss": 0.8387,
      "step": 233430
    },
    {
      "epoch": 4.098386558752787,
      "grad_norm": 0.049902327940477664,
      "learning_rate": 1.8085810297655122e-05,
      "loss": 0.8333,
      "step": 233440
    },
    {
      "epoch": 4.098562123632789,
      "grad_norm": 0.04350708021272303,
      "learning_rate": 1.8079721817641717e-05,
      "loss": 0.8462,
      "step": 233450
    },
    {
      "epoch": 4.09873768851279,
      "grad_norm": 0.04791241245860213,
      "learning_rate": 1.8073634388244123e-05,
      "loss": 0.8388,
      "step": 233460
    },
    {
      "epoch": 4.098913253392792,
      "grad_norm": 0.046086679511281065,
      "learning_rate": 1.8067548009539503e-05,
      "loss": 0.8392,
      "step": 233470
    },
    {
      "epoch": 4.099088818272793,
      "grad_norm": 0.04332022657037043,
      "learning_rate": 1.8061462681604915e-05,
      "loss": 0.846,
      "step": 233480
    },
    {
      "epoch": 4.099264383152794,
      "grad_norm": 0.07218755134070327,
      "learning_rate": 1.8055378404517494e-05,
      "loss": 0.8346,
      "step": 233490
    },
    {
      "epoch": 4.099439948032796,
      "grad_norm": 0.059148540645545485,
      "learning_rate": 1.8049295178354354e-05,
      "loss": 0.8446,
      "step": 233500
    },
    {
      "epoch": 4.099615512912797,
      "grad_norm": 0.08229480191953949,
      "learning_rate": 1.8043213003192507e-05,
      "loss": 0.831,
      "step": 233510
    },
    {
      "epoch": 4.099791077792799,
      "grad_norm": 0.06175842020667579,
      "learning_rate": 1.8037131879109054e-05,
      "loss": 0.8481,
      "step": 233520
    },
    {
      "epoch": 4.0999666426728,
      "grad_norm": 0.05066399644918555,
      "learning_rate": 1.8031051806181036e-05,
      "loss": 0.8412,
      "step": 233530
    },
    {
      "epoch": 4.100142207552802,
      "grad_norm": 0.05056323547815906,
      "learning_rate": 1.8024972784485518e-05,
      "loss": 0.8303,
      "step": 233540
    },
    {
      "epoch": 4.100317772432803,
      "grad_norm": 0.04707043362493295,
      "learning_rate": 1.801889481409948e-05,
      "loss": 0.837,
      "step": 233550
    },
    {
      "epoch": 4.100493337312804,
      "grad_norm": 0.05751873262031128,
      "learning_rate": 1.8012817895099944e-05,
      "loss": 0.8422,
      "step": 233560
    },
    {
      "epoch": 4.1006689021928056,
      "grad_norm": 0.044505001094348705,
      "learning_rate": 1.8006742027563933e-05,
      "loss": 0.8416,
      "step": 233570
    },
    {
      "epoch": 4.100844467072807,
      "grad_norm": 0.05060607732185944,
      "learning_rate": 1.8000667211568378e-05,
      "loss": 0.8405,
      "step": 233580
    },
    {
      "epoch": 4.1010200319528085,
      "grad_norm": 0.048849565985029646,
      "learning_rate": 1.799459344719028e-05,
      "loss": 0.8346,
      "step": 233590
    },
    {
      "epoch": 4.1011955968328095,
      "grad_norm": 0.05188911678409261,
      "learning_rate": 1.7988520734506592e-05,
      "loss": 0.8436,
      "step": 233600
    },
    {
      "epoch": 4.101371161712811,
      "grad_norm": 0.05417765173245984,
      "learning_rate": 1.7982449073594278e-05,
      "loss": 0.8426,
      "step": 233610
    },
    {
      "epoch": 4.1015467265928125,
      "grad_norm": 0.05405200495902855,
      "learning_rate": 1.7976378464530225e-05,
      "loss": 0.8327,
      "step": 233620
    },
    {
      "epoch": 4.1017222914728135,
      "grad_norm": 0.04295404103962365,
      "learning_rate": 1.7970308907391365e-05,
      "loss": 0.8554,
      "step": 233630
    },
    {
      "epoch": 4.1018978563528155,
      "grad_norm": 0.05797015858269802,
      "learning_rate": 1.796424040225462e-05,
      "loss": 0.8371,
      "step": 233640
    },
    {
      "epoch": 4.1020734212328165,
      "grad_norm": 0.06056425478223552,
      "learning_rate": 1.7958172949196847e-05,
      "loss": 0.8308,
      "step": 233650
    },
    {
      "epoch": 4.102248986112818,
      "grad_norm": 0.05716513569091293,
      "learning_rate": 1.795210654829494e-05,
      "loss": 0.8416,
      "step": 233660
    },
    {
      "epoch": 4.102424550992819,
      "grad_norm": 0.07438481216184158,
      "learning_rate": 1.794604119962578e-05,
      "loss": 0.843,
      "step": 233670
    },
    {
      "epoch": 4.1026001158728205,
      "grad_norm": 0.05543868750553875,
      "learning_rate": 1.7939976903266172e-05,
      "loss": 0.8385,
      "step": 233680
    },
    {
      "epoch": 4.102775680752822,
      "grad_norm": 0.06460047541232235,
      "learning_rate": 1.7933913659292967e-05,
      "loss": 0.8462,
      "step": 233690
    },
    {
      "epoch": 4.102951245632823,
      "grad_norm": 0.05365922563709247,
      "learning_rate": 1.792785146778301e-05,
      "loss": 0.8349,
      "step": 233700
    },
    {
      "epoch": 4.103126810512825,
      "grad_norm": 0.05513468112755364,
      "learning_rate": 1.7921790328813106e-05,
      "loss": 0.8331,
      "step": 233710
    },
    {
      "epoch": 4.103302375392826,
      "grad_norm": 0.054282590587864464,
      "learning_rate": 1.791573024246002e-05,
      "loss": 0.8384,
      "step": 233720
    },
    {
      "epoch": 4.103477940272827,
      "grad_norm": 0.048216211684415304,
      "learning_rate": 1.790967120880056e-05,
      "loss": 0.8406,
      "step": 233730
    },
    {
      "epoch": 4.103653505152829,
      "grad_norm": 0.06014352041070765,
      "learning_rate": 1.7903613227911506e-05,
      "loss": 0.8389,
      "step": 233740
    },
    {
      "epoch": 4.10382907003283,
      "grad_norm": 0.0507279258358342,
      "learning_rate": 1.7897556299869585e-05,
      "loss": 0.8423,
      "step": 233750
    },
    {
      "epoch": 4.104004634912832,
      "grad_norm": 0.09378649066607059,
      "learning_rate": 1.789150042475155e-05,
      "loss": 0.8394,
      "step": 233760
    },
    {
      "epoch": 4.104180199792833,
      "grad_norm": 0.052986146349155765,
      "learning_rate": 1.7885445602634123e-05,
      "loss": 0.8439,
      "step": 233770
    },
    {
      "epoch": 4.104355764672835,
      "grad_norm": 0.045694626288392264,
      "learning_rate": 1.7879391833594057e-05,
      "loss": 0.8433,
      "step": 233780
    },
    {
      "epoch": 4.104531329552836,
      "grad_norm": 0.05664677560046383,
      "learning_rate": 1.7873339117708002e-05,
      "loss": 0.8383,
      "step": 233790
    },
    {
      "epoch": 4.104706894432837,
      "grad_norm": 0.05997197434309478,
      "learning_rate": 1.786728745505267e-05,
      "loss": 0.8417,
      "step": 233800
    },
    {
      "epoch": 4.104882459312839,
      "grad_norm": 0.049627609131123046,
      "learning_rate": 1.786123684570477e-05,
      "loss": 0.8439,
      "step": 233810
    },
    {
      "epoch": 4.10505802419284,
      "grad_norm": 0.066192475515816,
      "learning_rate": 1.78551872897409e-05,
      "loss": 0.8274,
      "step": 233820
    },
    {
      "epoch": 4.105233589072842,
      "grad_norm": 0.0544530215305201,
      "learning_rate": 1.7849138787237746e-05,
      "loss": 0.8362,
      "step": 233830
    },
    {
      "epoch": 4.105409153952843,
      "grad_norm": 0.05465602413923256,
      "learning_rate": 1.7843091338271966e-05,
      "loss": 0.8394,
      "step": 233840
    },
    {
      "epoch": 4.105584718832844,
      "grad_norm": 0.05220597570596629,
      "learning_rate": 1.7837044942920125e-05,
      "loss": 0.8332,
      "step": 233850
    },
    {
      "epoch": 4.105760283712846,
      "grad_norm": 0.06344808978619919,
      "learning_rate": 1.783099960125887e-05,
      "loss": 0.8263,
      "step": 233860
    },
    {
      "epoch": 4.105935848592847,
      "grad_norm": 0.06459812064634493,
      "learning_rate": 1.782495531336479e-05,
      "loss": 0.8375,
      "step": 233870
    },
    {
      "epoch": 4.106111413472849,
      "grad_norm": 0.04546656478326062,
      "learning_rate": 1.7818912079314477e-05,
      "loss": 0.8501,
      "step": 233880
    },
    {
      "epoch": 4.10628697835285,
      "grad_norm": 0.06021879491841797,
      "learning_rate": 1.7812869899184483e-05,
      "loss": 0.8361,
      "step": 233890
    },
    {
      "epoch": 4.106462543232852,
      "grad_norm": 0.04581336406265435,
      "learning_rate": 1.7806828773051355e-05,
      "loss": 0.8456,
      "step": 233900
    },
    {
      "epoch": 4.106638108112853,
      "grad_norm": 0.044714838561385986,
      "learning_rate": 1.7800788700991685e-05,
      "loss": 0.8414,
      "step": 233910
    },
    {
      "epoch": 4.106813672992854,
      "grad_norm": 0.05861089296421893,
      "learning_rate": 1.7794749683081943e-05,
      "loss": 0.8499,
      "step": 233920
    },
    {
      "epoch": 4.106989237872856,
      "grad_norm": 0.05955066406818617,
      "learning_rate": 1.778871171939866e-05,
      "loss": 0.8423,
      "step": 233930
    },
    {
      "epoch": 4.107164802752857,
      "grad_norm": 0.05133327553972502,
      "learning_rate": 1.7782674810018353e-05,
      "loss": 0.8327,
      "step": 233940
    },
    {
      "epoch": 4.107340367632859,
      "grad_norm": 0.06403013990216798,
      "learning_rate": 1.7776638955017528e-05,
      "loss": 0.8408,
      "step": 233950
    },
    {
      "epoch": 4.10751593251286,
      "grad_norm": 0.04812748674418775,
      "learning_rate": 1.777060415447261e-05,
      "loss": 0.8507,
      "step": 233960
    },
    {
      "epoch": 4.107691497392861,
      "grad_norm": 0.07134291456895304,
      "learning_rate": 1.776457040846009e-05,
      "loss": 0.838,
      "step": 233970
    },
    {
      "epoch": 4.107867062272863,
      "grad_norm": 0.0411324117555695,
      "learning_rate": 1.7758537717056447e-05,
      "loss": 0.8425,
      "step": 233980
    },
    {
      "epoch": 4.108042627152864,
      "grad_norm": 0.04225511073102169,
      "learning_rate": 1.7752506080338046e-05,
      "loss": 0.8353,
      "step": 233990
    },
    {
      "epoch": 4.108218192032866,
      "grad_norm": 0.06173432864353543,
      "learning_rate": 1.7746475498381364e-05,
      "loss": 0.8409,
      "step": 234000
    },
    {
      "epoch": 4.108393756912867,
      "grad_norm": 0.04523980915434045,
      "learning_rate": 1.77404459712628e-05,
      "loss": 0.8329,
      "step": 234010
    },
    {
      "epoch": 4.108569321792869,
      "grad_norm": 0.06390683259745612,
      "learning_rate": 1.7734417499058735e-05,
      "loss": 0.832,
      "step": 234020
    },
    {
      "epoch": 4.10874488667287,
      "grad_norm": 0.05314354979162781,
      "learning_rate": 1.772839008184554e-05,
      "loss": 0.8342,
      "step": 234030
    },
    {
      "epoch": 4.108920451552871,
      "grad_norm": 0.05189840103745334,
      "learning_rate": 1.7722363719699613e-05,
      "loss": 0.8356,
      "step": 234040
    },
    {
      "epoch": 4.109096016432873,
      "grad_norm": 0.04960757241282705,
      "learning_rate": 1.7716338412697314e-05,
      "loss": 0.8446,
      "step": 234050
    },
    {
      "epoch": 4.109271581312874,
      "grad_norm": 0.05154251328881215,
      "learning_rate": 1.771031416091496e-05,
      "loss": 0.8427,
      "step": 234060
    },
    {
      "epoch": 4.109447146192876,
      "grad_norm": 0.050989653230786545,
      "learning_rate": 1.770429096442887e-05,
      "loss": 0.8426,
      "step": 234070
    },
    {
      "epoch": 4.109622711072877,
      "grad_norm": 0.05183205824271167,
      "learning_rate": 1.769826882331541e-05,
      "loss": 0.8395,
      "step": 234080
    },
    {
      "epoch": 4.109798275952878,
      "grad_norm": 0.05700269545847513,
      "learning_rate": 1.7692247737650817e-05,
      "loss": 0.8389,
      "step": 234090
    },
    {
      "epoch": 4.10997384083288,
      "grad_norm": 0.04536213258798704,
      "learning_rate": 1.768622770751143e-05,
      "loss": 0.8381,
      "step": 234100
    },
    {
      "epoch": 4.110149405712881,
      "grad_norm": 0.05705601102265365,
      "learning_rate": 1.768020873297349e-05,
      "loss": 0.8413,
      "step": 234110
    },
    {
      "epoch": 4.110324970592883,
      "grad_norm": 0.06203536652009816,
      "learning_rate": 1.7674190814113297e-05,
      "loss": 0.8344,
      "step": 234120
    },
    {
      "epoch": 4.110500535472884,
      "grad_norm": 0.046688362270438946,
      "learning_rate": 1.766817395100705e-05,
      "loss": 0.8363,
      "step": 234130
    },
    {
      "epoch": 4.110676100352886,
      "grad_norm": 0.0636845408943228,
      "learning_rate": 1.7662158143731015e-05,
      "loss": 0.8448,
      "step": 234140
    },
    {
      "epoch": 4.110851665232887,
      "grad_norm": 0.053320385749331484,
      "learning_rate": 1.7656143392361432e-05,
      "loss": 0.8322,
      "step": 234150
    },
    {
      "epoch": 4.111027230112888,
      "grad_norm": 0.04568045631908342,
      "learning_rate": 1.765012969697446e-05,
      "loss": 0.8423,
      "step": 234160
    },
    {
      "epoch": 4.11120279499289,
      "grad_norm": 0.049795361241490636,
      "learning_rate": 1.7644117057646324e-05,
      "loss": 0.8417,
      "step": 234170
    },
    {
      "epoch": 4.111378359872891,
      "grad_norm": 0.04883126281315454,
      "learning_rate": 1.7638105474453213e-05,
      "loss": 0.8322,
      "step": 234180
    },
    {
      "epoch": 4.111553924752893,
      "grad_norm": 0.07636779491731244,
      "learning_rate": 1.7632094947471265e-05,
      "loss": 0.8454,
      "step": 234190
    },
    {
      "epoch": 4.111729489632894,
      "grad_norm": 0.06237136436762392,
      "learning_rate": 1.7626085476776648e-05,
      "loss": 0.8363,
      "step": 234200
    },
    {
      "epoch": 4.111905054512896,
      "grad_norm": 0.051231858687043406,
      "learning_rate": 1.762007706244551e-05,
      "loss": 0.8394,
      "step": 234210
    },
    {
      "epoch": 4.112080619392897,
      "grad_norm": 0.04741821374448961,
      "learning_rate": 1.7614069704554003e-05,
      "loss": 0.8453,
      "step": 234220
    },
    {
      "epoch": 4.112256184272898,
      "grad_norm": 0.04320414358673438,
      "learning_rate": 1.7608063403178186e-05,
      "loss": 0.8392,
      "step": 234230
    },
    {
      "epoch": 4.1124317491529,
      "grad_norm": 0.06259767405206428,
      "learning_rate": 1.760205815839418e-05,
      "loss": 0.8339,
      "step": 234240
    },
    {
      "epoch": 4.112607314032901,
      "grad_norm": 0.06112520899029145,
      "learning_rate": 1.759605397027811e-05,
      "loss": 0.8401,
      "step": 234250
    },
    {
      "epoch": 4.112782878912903,
      "grad_norm": 0.06361678946433064,
      "learning_rate": 1.7590050838906e-05,
      "loss": 0.8364,
      "step": 234260
    },
    {
      "epoch": 4.112958443792904,
      "grad_norm": 0.04923098733482669,
      "learning_rate": 1.7584048764353923e-05,
      "loss": 0.8393,
      "step": 234270
    },
    {
      "epoch": 4.113134008672905,
      "grad_norm": 0.05317518465812162,
      "learning_rate": 1.757804774669795e-05,
      "loss": 0.8396,
      "step": 234280
    },
    {
      "epoch": 4.113309573552907,
      "grad_norm": 0.0564742776975935,
      "learning_rate": 1.757204778601411e-05,
      "loss": 0.844,
      "step": 234290
    },
    {
      "epoch": 4.113485138432908,
      "grad_norm": 0.04447742852524688,
      "learning_rate": 1.7566048882378394e-05,
      "loss": 0.8417,
      "step": 234300
    },
    {
      "epoch": 4.11366070331291,
      "grad_norm": 0.06251673754193275,
      "learning_rate": 1.7560051035866825e-05,
      "loss": 0.8371,
      "step": 234310
    },
    {
      "epoch": 4.113836268192911,
      "grad_norm": 0.05707089925164556,
      "learning_rate": 1.755405424655543e-05,
      "loss": 0.846,
      "step": 234320
    },
    {
      "epoch": 4.1140118330729125,
      "grad_norm": 0.04883766353868733,
      "learning_rate": 1.7548058514520136e-05,
      "loss": 0.8356,
      "step": 234330
    },
    {
      "epoch": 4.1141873979529135,
      "grad_norm": 0.074588182953143,
      "learning_rate": 1.7542063839836923e-05,
      "loss": 0.8321,
      "step": 234340
    },
    {
      "epoch": 4.114362962832915,
      "grad_norm": 0.04886197662497424,
      "learning_rate": 1.7536070222581787e-05,
      "loss": 0.8477,
      "step": 234350
    },
    {
      "epoch": 4.1145385277129165,
      "grad_norm": 0.05149736686885737,
      "learning_rate": 1.7530077662830624e-05,
      "loss": 0.845,
      "step": 234360
    },
    {
      "epoch": 4.1147140925929175,
      "grad_norm": 0.07031067814553756,
      "learning_rate": 1.7524086160659357e-05,
      "loss": 0.839,
      "step": 234370
    },
    {
      "epoch": 4.1148896574729195,
      "grad_norm": 0.05983683438461239,
      "learning_rate": 1.7518095716143925e-05,
      "loss": 0.8412,
      "step": 234380
    },
    {
      "epoch": 4.1150652223529205,
      "grad_norm": 0.05093948577966753,
      "learning_rate": 1.7512106329360245e-05,
      "loss": 0.8415,
      "step": 234390
    },
    {
      "epoch": 4.1152407872329215,
      "grad_norm": 0.06099773981430376,
      "learning_rate": 1.7506118000384144e-05,
      "loss": 0.8367,
      "step": 234400
    },
    {
      "epoch": 4.1154163521129234,
      "grad_norm": 0.055041589226949336,
      "learning_rate": 1.7500130729291543e-05,
      "loss": 0.8373,
      "step": 234410
    },
    {
      "epoch": 4.1155919169929245,
      "grad_norm": 0.052742809543159755,
      "learning_rate": 1.749414451615831e-05,
      "loss": 0.8444,
      "step": 234420
    },
    {
      "epoch": 4.115767481872926,
      "grad_norm": 0.04600618092886495,
      "learning_rate": 1.7488159361060243e-05,
      "loss": 0.84,
      "step": 234430
    },
    {
      "epoch": 4.115943046752927,
      "grad_norm": 0.05186259818202852,
      "learning_rate": 1.7482175264073196e-05,
      "loss": 0.8406,
      "step": 234440
    },
    {
      "epoch": 4.116118611632929,
      "grad_norm": 0.053999639017999444,
      "learning_rate": 1.7476192225273e-05,
      "loss": 0.8385,
      "step": 234450
    },
    {
      "epoch": 4.11629417651293,
      "grad_norm": 0.04945080919736853,
      "learning_rate": 1.747021024473548e-05,
      "loss": 0.8456,
      "step": 234460
    },
    {
      "epoch": 4.116469741392931,
      "grad_norm": 0.041614555424759656,
      "learning_rate": 1.7464229322536378e-05,
      "loss": 0.8358,
      "step": 234470
    },
    {
      "epoch": 4.116645306272933,
      "grad_norm": 0.05125609536613273,
      "learning_rate": 1.7458249458751498e-05,
      "loss": 0.8417,
      "step": 234480
    },
    {
      "epoch": 4.116820871152934,
      "grad_norm": 0.04255487748945043,
      "learning_rate": 1.745227065345664e-05,
      "loss": 0.84,
      "step": 234490
    },
    {
      "epoch": 4.116996436032936,
      "grad_norm": 0.054202744389663736,
      "learning_rate": 1.7446292906727497e-05,
      "loss": 0.8407,
      "step": 234500
    },
    {
      "epoch": 4.117172000912937,
      "grad_norm": 0.05288665338138002,
      "learning_rate": 1.7440316218639838e-05,
      "loss": 0.8402,
      "step": 234510
    },
    {
      "epoch": 4.117347565792938,
      "grad_norm": 0.06038664446387986,
      "learning_rate": 1.7434340589269403e-05,
      "loss": 0.8411,
      "step": 234520
    },
    {
      "epoch": 4.11752313067294,
      "grad_norm": 0.04814584081758831,
      "learning_rate": 1.742836601869187e-05,
      "loss": 0.8405,
      "step": 234530
    },
    {
      "epoch": 4.117698695552941,
      "grad_norm": 0.05396248182411174,
      "learning_rate": 1.742239250698296e-05,
      "loss": 0.8362,
      "step": 234540
    },
    {
      "epoch": 4.117874260432943,
      "grad_norm": 0.05382661332812177,
      "learning_rate": 1.7416420054218345e-05,
      "loss": 0.8417,
      "step": 234550
    },
    {
      "epoch": 4.118049825312944,
      "grad_norm": 0.0414453055648091,
      "learning_rate": 1.7410448660473737e-05,
      "loss": 0.8406,
      "step": 234560
    },
    {
      "epoch": 4.118225390192946,
      "grad_norm": 0.04987369774080971,
      "learning_rate": 1.7404478325824737e-05,
      "loss": 0.8343,
      "step": 234570
    },
    {
      "epoch": 4.118400955072947,
      "grad_norm": 0.04399735346327648,
      "learning_rate": 1.739850905034702e-05,
      "loss": 0.8455,
      "step": 234580
    },
    {
      "epoch": 4.118576519952948,
      "grad_norm": 0.056959679684240705,
      "learning_rate": 1.7392540834116236e-05,
      "loss": 0.8397,
      "step": 234590
    },
    {
      "epoch": 4.11875208483295,
      "grad_norm": 0.07239583776098774,
      "learning_rate": 1.738657367720796e-05,
      "loss": 0.8436,
      "step": 234600
    },
    {
      "epoch": 4.118927649712951,
      "grad_norm": 0.053675120037024224,
      "learning_rate": 1.7380607579697826e-05,
      "loss": 0.8407,
      "step": 234610
    },
    {
      "epoch": 4.119103214592953,
      "grad_norm": 0.05311120623634274,
      "learning_rate": 1.737464254166142e-05,
      "loss": 0.8304,
      "step": 234620
    },
    {
      "epoch": 4.119278779472954,
      "grad_norm": 0.054967653288015964,
      "learning_rate": 1.7368678563174336e-05,
      "loss": 0.8389,
      "step": 234630
    },
    {
      "epoch": 4.119454344352955,
      "grad_norm": 0.04794764088383234,
      "learning_rate": 1.73627156443121e-05,
      "loss": 0.8398,
      "step": 234640
    },
    {
      "epoch": 4.119629909232957,
      "grad_norm": 0.05564567894472691,
      "learning_rate": 1.7356753785150293e-05,
      "loss": 0.8414,
      "step": 234650
    },
    {
      "epoch": 4.119805474112958,
      "grad_norm": 0.04879054866690539,
      "learning_rate": 1.735079298576446e-05,
      "loss": 0.8321,
      "step": 234660
    },
    {
      "epoch": 4.11998103899296,
      "grad_norm": 0.05191277534873428,
      "learning_rate": 1.734483324623009e-05,
      "loss": 0.8422,
      "step": 234670
    },
    {
      "epoch": 4.120156603872961,
      "grad_norm": 0.05310921042825396,
      "learning_rate": 1.7338874566622714e-05,
      "loss": 0.8411,
      "step": 234680
    },
    {
      "epoch": 4.120332168752963,
      "grad_norm": 0.06275827900395828,
      "learning_rate": 1.733291694701786e-05,
      "loss": 0.8383,
      "step": 234690
    },
    {
      "epoch": 4.120507733632964,
      "grad_norm": 0.06344737019212822,
      "learning_rate": 1.7326960387490952e-05,
      "loss": 0.8411,
      "step": 234700
    },
    {
      "epoch": 4.120683298512965,
      "grad_norm": 0.09488362145822753,
      "learning_rate": 1.73210048881175e-05,
      "loss": 0.8417,
      "step": 234710
    },
    {
      "epoch": 4.120858863392967,
      "grad_norm": 0.04987990973960136,
      "learning_rate": 1.731505044897295e-05,
      "loss": 0.841,
      "step": 234720
    },
    {
      "epoch": 4.121034428272968,
      "grad_norm": 0.05084155618949037,
      "learning_rate": 1.7309097070132778e-05,
      "loss": 0.8412,
      "step": 234730
    },
    {
      "epoch": 4.12120999315297,
      "grad_norm": 0.04455123554308123,
      "learning_rate": 1.7303144751672346e-05,
      "loss": 0.8396,
      "step": 234740
    },
    {
      "epoch": 4.121385558032971,
      "grad_norm": 0.04346811713998301,
      "learning_rate": 1.729719349366712e-05,
      "loss": 0.8378,
      "step": 234750
    },
    {
      "epoch": 4.121561122912972,
      "grad_norm": 0.05001197070497517,
      "learning_rate": 1.729124329619252e-05,
      "loss": 0.8478,
      "step": 234760
    },
    {
      "epoch": 4.121736687792974,
      "grad_norm": 0.05299884212328996,
      "learning_rate": 1.7285294159323888e-05,
      "loss": 0.8354,
      "step": 234770
    },
    {
      "epoch": 4.121912252672975,
      "grad_norm": 0.057647671594959635,
      "learning_rate": 1.7279346083136614e-05,
      "loss": 0.8432,
      "step": 234780
    },
    {
      "epoch": 4.122087817552977,
      "grad_norm": 0.06195874734413342,
      "learning_rate": 1.7273399067706078e-05,
      "loss": 0.8327,
      "step": 234790
    },
    {
      "epoch": 4.122263382432978,
      "grad_norm": 0.05750256029646597,
      "learning_rate": 1.7267453113107642e-05,
      "loss": 0.8387,
      "step": 234800
    },
    {
      "epoch": 4.12243894731298,
      "grad_norm": 0.05250991722348684,
      "learning_rate": 1.7261508219416603e-05,
      "loss": 0.8383,
      "step": 234810
    },
    {
      "epoch": 4.122614512192981,
      "grad_norm": 0.05515349795894327,
      "learning_rate": 1.7255564386708294e-05,
      "loss": 0.8414,
      "step": 234820
    },
    {
      "epoch": 4.122790077072982,
      "grad_norm": 0.05827120208085201,
      "learning_rate": 1.7249621615058063e-05,
      "loss": 0.8433,
      "step": 234830
    },
    {
      "epoch": 4.122965641952984,
      "grad_norm": 0.0464705873558169,
      "learning_rate": 1.7243679904541153e-05,
      "loss": 0.845,
      "step": 234840
    },
    {
      "epoch": 4.123141206832985,
      "grad_norm": 0.04497342180548477,
      "learning_rate": 1.723773925523287e-05,
      "loss": 0.8418,
      "step": 234850
    },
    {
      "epoch": 4.123316771712987,
      "grad_norm": 0.05152269147502093,
      "learning_rate": 1.7231799667208507e-05,
      "loss": 0.837,
      "step": 234860
    },
    {
      "epoch": 4.123492336592988,
      "grad_norm": 0.0546796501331538,
      "learning_rate": 1.722586114054328e-05,
      "loss": 0.8463,
      "step": 234870
    },
    {
      "epoch": 4.12366790147299,
      "grad_norm": 0.060523403138548725,
      "learning_rate": 1.7219923675312437e-05,
      "loss": 0.8388,
      "step": 234880
    },
    {
      "epoch": 4.123843466352991,
      "grad_norm": 0.0674122183476116,
      "learning_rate": 1.7213987271591224e-05,
      "loss": 0.8281,
      "step": 234890
    },
    {
      "epoch": 4.124019031232992,
      "grad_norm": 0.0697767245929838,
      "learning_rate": 1.720805192945487e-05,
      "loss": 0.8383,
      "step": 234900
    },
    {
      "epoch": 4.124194596112994,
      "grad_norm": 0.05267889483543525,
      "learning_rate": 1.7202117648978527e-05,
      "loss": 0.8423,
      "step": 234910
    },
    {
      "epoch": 4.124370160992995,
      "grad_norm": 0.04500525839719952,
      "learning_rate": 1.719618443023742e-05,
      "loss": 0.8419,
      "step": 234920
    },
    {
      "epoch": 4.124545725872997,
      "grad_norm": 0.054034297720997285,
      "learning_rate": 1.7190252273306747e-05,
      "loss": 0.8342,
      "step": 234930
    },
    {
      "epoch": 4.124721290752998,
      "grad_norm": 0.048646665372742044,
      "learning_rate": 1.7184321178261612e-05,
      "loss": 0.8389,
      "step": 234940
    },
    {
      "epoch": 4.124896855632999,
      "grad_norm": 0.057179047025789194,
      "learning_rate": 1.7178391145177185e-05,
      "loss": 0.8366,
      "step": 234950
    },
    {
      "epoch": 4.125072420513001,
      "grad_norm": 0.050954832213349434,
      "learning_rate": 1.7172462174128613e-05,
      "loss": 0.8401,
      "step": 234960
    },
    {
      "epoch": 4.125247985393002,
      "grad_norm": 0.06950686316195245,
      "learning_rate": 1.716653426519103e-05,
      "loss": 0.8362,
      "step": 234970
    },
    {
      "epoch": 4.125423550273004,
      "grad_norm": 0.04164361234012297,
      "learning_rate": 1.7160607418439506e-05,
      "loss": 0.8362,
      "step": 234980
    },
    {
      "epoch": 4.125599115153005,
      "grad_norm": 0.04969816147983404,
      "learning_rate": 1.7154681633949146e-05,
      "loss": 0.8368,
      "step": 234990
    },
    {
      "epoch": 4.125774680033007,
      "grad_norm": 0.05559943769587121,
      "learning_rate": 1.7148756911795064e-05,
      "loss": 0.8418,
      "step": 235000
    },
    {
      "epoch": 4.125950244913008,
      "grad_norm": 0.04573122592216604,
      "learning_rate": 1.7142833252052283e-05,
      "loss": 0.8417,
      "step": 235010
    },
    {
      "epoch": 4.126125809793009,
      "grad_norm": 0.07315233313308833,
      "learning_rate": 1.7136910654795875e-05,
      "loss": 0.8358,
      "step": 235020
    },
    {
      "epoch": 4.126301374673011,
      "grad_norm": 0.07279770267905203,
      "learning_rate": 1.7130989120100902e-05,
      "loss": 0.8384,
      "step": 235030
    },
    {
      "epoch": 4.126476939553012,
      "grad_norm": 0.05722818153628665,
      "learning_rate": 1.7125068648042352e-05,
      "loss": 0.8441,
      "step": 235040
    },
    {
      "epoch": 4.126652504433014,
      "grad_norm": 0.04724630685113138,
      "learning_rate": 1.7119149238695254e-05,
      "loss": 0.8389,
      "step": 235050
    },
    {
      "epoch": 4.126828069313015,
      "grad_norm": 0.046272910811731946,
      "learning_rate": 1.711323089213461e-05,
      "loss": 0.8379,
      "step": 235060
    },
    {
      "epoch": 4.127003634193016,
      "grad_norm": 0.0453612340908399,
      "learning_rate": 1.7107313608435425e-05,
      "loss": 0.8408,
      "step": 235070
    },
    {
      "epoch": 4.127179199073018,
      "grad_norm": 0.0689793439772486,
      "learning_rate": 1.7101397387672635e-05,
      "loss": 0.8446,
      "step": 235080
    },
    {
      "epoch": 4.127354763953019,
      "grad_norm": 0.06240268830373271,
      "learning_rate": 1.7095482229921225e-05,
      "loss": 0.845,
      "step": 235090
    },
    {
      "epoch": 4.1275303288330205,
      "grad_norm": 0.05276254878899509,
      "learning_rate": 1.7089568135256155e-05,
      "loss": 0.8335,
      "step": 235100
    },
    {
      "epoch": 4.1277058937130215,
      "grad_norm": 0.05108152621859422,
      "learning_rate": 1.7083655103752318e-05,
      "loss": 0.8384,
      "step": 235110
    },
    {
      "epoch": 4.1278814585930235,
      "grad_norm": 0.04965126181644683,
      "learning_rate": 1.707774313548465e-05,
      "loss": 0.8445,
      "step": 235120
    },
    {
      "epoch": 4.1280570234730245,
      "grad_norm": 0.06289319520830705,
      "learning_rate": 1.707183223052806e-05,
      "loss": 0.8391,
      "step": 235130
    },
    {
      "epoch": 4.1282325883530255,
      "grad_norm": 0.04750486372806142,
      "learning_rate": 1.706592238895746e-05,
      "loss": 0.8345,
      "step": 235140
    },
    {
      "epoch": 4.1284081532330275,
      "grad_norm": 0.04228958733517577,
      "learning_rate": 1.7060013610847695e-05,
      "loss": 0.8398,
      "step": 235150
    },
    {
      "epoch": 4.1285837181130285,
      "grad_norm": 0.06886201901120413,
      "learning_rate": 1.7054105896273632e-05,
      "loss": 0.8443,
      "step": 235160
    },
    {
      "epoch": 4.12875928299303,
      "grad_norm": 0.04883282222432168,
      "learning_rate": 1.7048199245310165e-05,
      "loss": 0.8331,
      "step": 235170
    },
    {
      "epoch": 4.128934847873031,
      "grad_norm": 0.04527842336934617,
      "learning_rate": 1.7042293658032082e-05,
      "loss": 0.8372,
      "step": 235180
    },
    {
      "epoch": 4.1291104127530325,
      "grad_norm": 0.050596089523548944,
      "learning_rate": 1.703638913451423e-05,
      "loss": 0.844,
      "step": 235190
    },
    {
      "epoch": 4.129285977633034,
      "grad_norm": 0.05700427590244759,
      "learning_rate": 1.703048567483144e-05,
      "loss": 0.8374,
      "step": 235200
    },
    {
      "epoch": 4.129461542513035,
      "grad_norm": 0.04847299816745394,
      "learning_rate": 1.7024583279058467e-05,
      "loss": 0.8429,
      "step": 235210
    },
    {
      "epoch": 4.129637107393037,
      "grad_norm": 0.05358846185542368,
      "learning_rate": 1.7018681947270116e-05,
      "loss": 0.8437,
      "step": 235220
    },
    {
      "epoch": 4.129812672273038,
      "grad_norm": 0.052335119158798356,
      "learning_rate": 1.7012781679541153e-05,
      "loss": 0.85,
      "step": 235230
    },
    {
      "epoch": 4.12998823715304,
      "grad_norm": 0.04650760144617285,
      "learning_rate": 1.700688247594638e-05,
      "loss": 0.8367,
      "step": 235240
    },
    {
      "epoch": 4.130163802033041,
      "grad_norm": 0.05134971733082181,
      "learning_rate": 1.7000984336560473e-05,
      "loss": 0.8381,
      "step": 235250
    },
    {
      "epoch": 4.130339366913042,
      "grad_norm": 0.05810557649206399,
      "learning_rate": 1.699508726145819e-05,
      "loss": 0.849,
      "step": 235260
    },
    {
      "epoch": 4.130514931793044,
      "grad_norm": 0.06055077128009558,
      "learning_rate": 1.698919125071428e-05,
      "loss": 0.8313,
      "step": 235270
    },
    {
      "epoch": 4.130690496673045,
      "grad_norm": 0.06918774925685399,
      "learning_rate": 1.6983296304403387e-05,
      "loss": 0.8382,
      "step": 235280
    },
    {
      "epoch": 4.130866061553047,
      "grad_norm": 0.06615788086079645,
      "learning_rate": 1.6977402422600235e-05,
      "loss": 0.8415,
      "step": 235290
    },
    {
      "epoch": 4.131041626433048,
      "grad_norm": 0.046203195242998076,
      "learning_rate": 1.6971509605379506e-05,
      "loss": 0.8371,
      "step": 235300
    },
    {
      "epoch": 4.131217191313049,
      "grad_norm": 0.04651898132111819,
      "learning_rate": 1.6965617852815867e-05,
      "loss": 0.8417,
      "step": 235310
    },
    {
      "epoch": 4.131392756193051,
      "grad_norm": 0.05890475580927434,
      "learning_rate": 1.6959727164983926e-05,
      "loss": 0.8458,
      "step": 235320
    },
    {
      "epoch": 4.131568321073052,
      "grad_norm": 0.041264092471387966,
      "learning_rate": 1.6953837541958363e-05,
      "loss": 0.8342,
      "step": 235330
    },
    {
      "epoch": 4.131743885953054,
      "grad_norm": 0.045375160281706836,
      "learning_rate": 1.694794898381379e-05,
      "loss": 0.8457,
      "step": 235340
    },
    {
      "epoch": 4.131919450833055,
      "grad_norm": 0.05925223592052818,
      "learning_rate": 1.6942061490624802e-05,
      "loss": 0.8424,
      "step": 235350
    },
    {
      "epoch": 4.132095015713057,
      "grad_norm": 0.04854567134132715,
      "learning_rate": 1.6936175062466004e-05,
      "loss": 0.8349,
      "step": 235360
    },
    {
      "epoch": 4.132270580593058,
      "grad_norm": 0.09361928160062316,
      "learning_rate": 1.6930289699412e-05,
      "loss": 0.8429,
      "step": 235370
    },
    {
      "epoch": 4.132446145473059,
      "grad_norm": 0.04741882243031577,
      "learning_rate": 1.6924405401537318e-05,
      "loss": 0.8382,
      "step": 235380
    },
    {
      "epoch": 4.132621710353061,
      "grad_norm": 0.050010070683402484,
      "learning_rate": 1.6918522168916538e-05,
      "loss": 0.8469,
      "step": 235390
    },
    {
      "epoch": 4.132797275233062,
      "grad_norm": 0.05365008111230403,
      "learning_rate": 1.691264000162418e-05,
      "loss": 0.8426,
      "step": 235400
    },
    {
      "epoch": 4.132972840113064,
      "grad_norm": 0.049387743320496925,
      "learning_rate": 1.6906758899734823e-05,
      "loss": 0.8408,
      "step": 235410
    },
    {
      "epoch": 4.133148404993065,
      "grad_norm": 0.05324731514706297,
      "learning_rate": 1.690087886332292e-05,
      "loss": 0.8359,
      "step": 235420
    },
    {
      "epoch": 4.133323969873066,
      "grad_norm": 0.0675510109568414,
      "learning_rate": 1.6894999892462994e-05,
      "loss": 0.832,
      "step": 235430
    },
    {
      "epoch": 4.133499534753068,
      "grad_norm": 0.045571343731171034,
      "learning_rate": 1.6889121987229573e-05,
      "loss": 0.8489,
      "step": 235440
    },
    {
      "epoch": 4.133675099633069,
      "grad_norm": 0.059493831642811155,
      "learning_rate": 1.6883245147697068e-05,
      "loss": 0.8523,
      "step": 235450
    },
    {
      "epoch": 4.133850664513071,
      "grad_norm": 0.06869912102086842,
      "learning_rate": 1.6877369373939968e-05,
      "loss": 0.8445,
      "step": 235460
    },
    {
      "epoch": 4.134026229393072,
      "grad_norm": 0.04941765215386544,
      "learning_rate": 1.6871494666032712e-05,
      "loss": 0.8448,
      "step": 235470
    },
    {
      "epoch": 4.134201794273074,
      "grad_norm": 0.05441435699541412,
      "learning_rate": 1.686562102404977e-05,
      "loss": 0.8387,
      "step": 235480
    },
    {
      "epoch": 4.134377359153075,
      "grad_norm": 0.06000864308006045,
      "learning_rate": 1.685974844806551e-05,
      "loss": 0.8424,
      "step": 235490
    },
    {
      "epoch": 4.134552924033076,
      "grad_norm": 0.05493608513535751,
      "learning_rate": 1.6853876938154372e-05,
      "loss": 0.8383,
      "step": 235500
    },
    {
      "epoch": 4.134728488913078,
      "grad_norm": 0.04540740597754138,
      "learning_rate": 1.6848006494390754e-05,
      "loss": 0.8343,
      "step": 235510
    },
    {
      "epoch": 4.134904053793079,
      "grad_norm": 0.08109308998861921,
      "learning_rate": 1.684213711684901e-05,
      "loss": 0.84,
      "step": 235520
    },
    {
      "epoch": 4.135079618673081,
      "grad_norm": 0.052185776680238334,
      "learning_rate": 1.68362688056035e-05,
      "loss": 0.833,
      "step": 235530
    },
    {
      "epoch": 4.135255183553082,
      "grad_norm": 0.04118586517499772,
      "learning_rate": 1.683040156072863e-05,
      "loss": 0.8413,
      "step": 235540
    },
    {
      "epoch": 4.135430748433084,
      "grad_norm": 0.07381308690221731,
      "learning_rate": 1.682453538229867e-05,
      "loss": 0.8471,
      "step": 235550
    },
    {
      "epoch": 4.135606313313085,
      "grad_norm": 0.0454057346183786,
      "learning_rate": 1.6818670270387984e-05,
      "loss": 0.8342,
      "step": 235560
    },
    {
      "epoch": 4.135781878193086,
      "grad_norm": 0.05204877679000117,
      "learning_rate": 1.681280622507088e-05,
      "loss": 0.8439,
      "step": 235570
    },
    {
      "epoch": 4.135957443073088,
      "grad_norm": 0.06600689195419857,
      "learning_rate": 1.6806943246421675e-05,
      "loss": 0.8403,
      "step": 235580
    },
    {
      "epoch": 4.136133007953089,
      "grad_norm": 0.05843261857898632,
      "learning_rate": 1.6801081334514618e-05,
      "loss": 0.8453,
      "step": 235590
    },
    {
      "epoch": 4.136308572833091,
      "grad_norm": 0.061291363878105735,
      "learning_rate": 1.6795220489423988e-05,
      "loss": 0.8394,
      "step": 235600
    },
    {
      "epoch": 4.136484137713092,
      "grad_norm": 0.05330032101929443,
      "learning_rate": 1.6789360711224082e-05,
      "loss": 0.8233,
      "step": 235610
    },
    {
      "epoch": 4.136659702593093,
      "grad_norm": 0.046081767634715526,
      "learning_rate": 1.6783501999989092e-05,
      "loss": 0.8439,
      "step": 235620
    },
    {
      "epoch": 4.136835267473095,
      "grad_norm": 0.06624259233382424,
      "learning_rate": 1.6777644355793263e-05,
      "loss": 0.8378,
      "step": 235630
    },
    {
      "epoch": 4.137010832353096,
      "grad_norm": 0.0478726082989011,
      "learning_rate": 1.677178777871083e-05,
      "loss": 0.8332,
      "step": 235640
    },
    {
      "epoch": 4.137186397233098,
      "grad_norm": 0.0563228169572709,
      "learning_rate": 1.676593226881601e-05,
      "loss": 0.8384,
      "step": 235650
    },
    {
      "epoch": 4.137361962113099,
      "grad_norm": 0.05230080852919149,
      "learning_rate": 1.6760077826182946e-05,
      "loss": 0.8321,
      "step": 235660
    },
    {
      "epoch": 4.1375375269931,
      "grad_norm": 0.062399143360944376,
      "learning_rate": 1.6754224450885848e-05,
      "loss": 0.8401,
      "step": 235670
    },
    {
      "epoch": 4.137713091873102,
      "grad_norm": 0.052180105891082816,
      "learning_rate": 1.6748372142998888e-05,
      "loss": 0.8416,
      "step": 235680
    },
    {
      "epoch": 4.137888656753103,
      "grad_norm": 0.05376964345806113,
      "learning_rate": 1.674252090259618e-05,
      "loss": 0.844,
      "step": 235690
    },
    {
      "epoch": 4.138064221633105,
      "grad_norm": 0.04158725937805825,
      "learning_rate": 1.6736670729751867e-05,
      "loss": 0.8471,
      "step": 235700
    },
    {
      "epoch": 4.138239786513106,
      "grad_norm": 0.04870385641231815,
      "learning_rate": 1.6730821624540123e-05,
      "loss": 0.8376,
      "step": 235710
    },
    {
      "epoch": 4.138415351393108,
      "grad_norm": 0.050256387700856456,
      "learning_rate": 1.672497358703499e-05,
      "loss": 0.8368,
      "step": 235720
    },
    {
      "epoch": 4.138590916273109,
      "grad_norm": 0.04999031492596745,
      "learning_rate": 1.67191266173106e-05,
      "loss": 0.8288,
      "step": 235730
    },
    {
      "epoch": 4.13876648115311,
      "grad_norm": 0.04947243159349324,
      "learning_rate": 1.671328071544102e-05,
      "loss": 0.8453,
      "step": 235740
    },
    {
      "epoch": 4.138942046033112,
      "grad_norm": 0.04431970440815433,
      "learning_rate": 1.6707435881500355e-05,
      "loss": 0.8398,
      "step": 235750
    },
    {
      "epoch": 4.139117610913113,
      "grad_norm": 0.07136447270315502,
      "learning_rate": 1.6701592115562597e-05,
      "loss": 0.839,
      "step": 235760
    },
    {
      "epoch": 4.139293175793115,
      "grad_norm": 0.05132152009182859,
      "learning_rate": 1.669574941770184e-05,
      "loss": 0.8321,
      "step": 235770
    },
    {
      "epoch": 4.139468740673116,
      "grad_norm": 0.04273271768582594,
      "learning_rate": 1.6689907787992108e-05,
      "loss": 0.8383,
      "step": 235780
    },
    {
      "epoch": 4.139644305553118,
      "grad_norm": 0.04032692357559119,
      "learning_rate": 1.6684067226507386e-05,
      "loss": 0.8361,
      "step": 235790
    },
    {
      "epoch": 4.139819870433119,
      "grad_norm": 0.05513283079877782,
      "learning_rate": 1.6678227733321677e-05,
      "loss": 0.8287,
      "step": 235800
    },
    {
      "epoch": 4.13999543531312,
      "grad_norm": 0.04904884588816031,
      "learning_rate": 1.6672389308508993e-05,
      "loss": 0.839,
      "step": 235810
    },
    {
      "epoch": 4.140171000193122,
      "grad_norm": 0.06833366917736175,
      "learning_rate": 1.6666551952143322e-05,
      "loss": 0.8424,
      "step": 235820
    },
    {
      "epoch": 4.140346565073123,
      "grad_norm": 0.05396863601377881,
      "learning_rate": 1.6660715664298575e-05,
      "loss": 0.8405,
      "step": 235830
    },
    {
      "epoch": 4.1405221299531245,
      "grad_norm": 0.044545584588989426,
      "learning_rate": 1.6654880445048723e-05,
      "loss": 0.8405,
      "step": 235840
    },
    {
      "epoch": 4.1406976948331256,
      "grad_norm": 0.06215820668274176,
      "learning_rate": 1.6649046294467726e-05,
      "loss": 0.8458,
      "step": 235850
    },
    {
      "epoch": 4.140873259713127,
      "grad_norm": 0.05422308441147754,
      "learning_rate": 1.664321321262945e-05,
      "loss": 0.8249,
      "step": 235860
    },
    {
      "epoch": 4.1410488245931285,
      "grad_norm": 0.044219291078601404,
      "learning_rate": 1.6637381199607836e-05,
      "loss": 0.8371,
      "step": 235870
    },
    {
      "epoch": 4.1412243894731295,
      "grad_norm": 0.04648765215742854,
      "learning_rate": 1.6631550255476797e-05,
      "loss": 0.8396,
      "step": 235880
    },
    {
      "epoch": 4.1413999543531315,
      "grad_norm": 0.045275227925774444,
      "learning_rate": 1.6625720380310147e-05,
      "loss": 0.8415,
      "step": 235890
    },
    {
      "epoch": 4.1415755192331325,
      "grad_norm": 0.05714498450499077,
      "learning_rate": 1.66198915741818e-05,
      "loss": 0.8369,
      "step": 235900
    },
    {
      "epoch": 4.141751084113134,
      "grad_norm": 0.05166985608833274,
      "learning_rate": 1.66140638371656e-05,
      "loss": 0.8346,
      "step": 235910
    },
    {
      "epoch": 4.1419266489931355,
      "grad_norm": 0.04714350163504745,
      "learning_rate": 1.6608237169335407e-05,
      "loss": 0.8391,
      "step": 235920
    },
    {
      "epoch": 4.1421022138731365,
      "grad_norm": 0.04798245387275625,
      "learning_rate": 1.6602411570765e-05,
      "loss": 0.8332,
      "step": 235930
    },
    {
      "epoch": 4.142277778753138,
      "grad_norm": 0.07181919278847415,
      "learning_rate": 1.6596587041528207e-05,
      "loss": 0.8403,
      "step": 235940
    },
    {
      "epoch": 4.142453343633139,
      "grad_norm": 0.05482673155595358,
      "learning_rate": 1.6590763581698853e-05,
      "loss": 0.8406,
      "step": 235950
    },
    {
      "epoch": 4.142628908513141,
      "grad_norm": 0.0550404319581507,
      "learning_rate": 1.6584941191350682e-05,
      "loss": 0.8348,
      "step": 235960
    },
    {
      "epoch": 4.142804473393142,
      "grad_norm": 0.05260776581451765,
      "learning_rate": 1.657911987055749e-05,
      "loss": 0.8338,
      "step": 235970
    },
    {
      "epoch": 4.142980038273143,
      "grad_norm": 0.04671229219606289,
      "learning_rate": 1.6573299619393022e-05,
      "loss": 0.8376,
      "step": 235980
    },
    {
      "epoch": 4.143155603153145,
      "grad_norm": 0.0524820160021772,
      "learning_rate": 1.6567480437931052e-05,
      "loss": 0.8346,
      "step": 235990
    },
    {
      "epoch": 4.143331168033146,
      "grad_norm": 0.05267774162636728,
      "learning_rate": 1.656166232624526e-05,
      "loss": 0.8404,
      "step": 236000
    },
    {
      "epoch": 4.143506732913148,
      "grad_norm": 0.04734261012256279,
      "learning_rate": 1.6555845284409393e-05,
      "loss": 0.8363,
      "step": 236010
    },
    {
      "epoch": 4.143682297793149,
      "grad_norm": 0.05093617439921944,
      "learning_rate": 1.6550029312497178e-05,
      "loss": 0.8482,
      "step": 236020
    },
    {
      "epoch": 4.143857862673151,
      "grad_norm": 0.07323958456687325,
      "learning_rate": 1.654421441058224e-05,
      "loss": 0.8428,
      "step": 236030
    },
    {
      "epoch": 4.144033427553152,
      "grad_norm": 0.05239952731467768,
      "learning_rate": 1.6538400578738305e-05,
      "loss": 0.8336,
      "step": 236040
    },
    {
      "epoch": 4.144208992433153,
      "grad_norm": 0.049296880847542975,
      "learning_rate": 1.653258781703904e-05,
      "loss": 0.8371,
      "step": 236050
    },
    {
      "epoch": 4.144384557313155,
      "grad_norm": 0.05769792878830423,
      "learning_rate": 1.6526776125558045e-05,
      "loss": 0.835,
      "step": 236060
    },
    {
      "epoch": 4.144560122193156,
      "grad_norm": 0.04116416752032641,
      "learning_rate": 1.6520965504368995e-05,
      "loss": 0.8462,
      "step": 236070
    },
    {
      "epoch": 4.144735687073158,
      "grad_norm": 0.05289467534712351,
      "learning_rate": 1.6515155953545498e-05,
      "loss": 0.8375,
      "step": 236080
    },
    {
      "epoch": 4.144911251953159,
      "grad_norm": 0.05660666153263503,
      "learning_rate": 1.6509347473161193e-05,
      "loss": 0.8414,
      "step": 236090
    },
    {
      "epoch": 4.14508681683316,
      "grad_norm": 0.054083937621200597,
      "learning_rate": 1.6503540063289626e-05,
      "loss": 0.8474,
      "step": 236100
    },
    {
      "epoch": 4.145262381713162,
      "grad_norm": 0.060056138612166986,
      "learning_rate": 1.6497733724004396e-05,
      "loss": 0.8354,
      "step": 236110
    },
    {
      "epoch": 4.145437946593163,
      "grad_norm": 0.04230713840816596,
      "learning_rate": 1.6491928455379102e-05,
      "loss": 0.8392,
      "step": 236120
    },
    {
      "epoch": 4.145613511473165,
      "grad_norm": 0.04511979245407741,
      "learning_rate": 1.6486124257487242e-05,
      "loss": 0.8503,
      "step": 236130
    },
    {
      "epoch": 4.145789076353166,
      "grad_norm": 0.07202310892109016,
      "learning_rate": 1.648032113040238e-05,
      "loss": 0.8481,
      "step": 236140
    },
    {
      "epoch": 4.145964641233168,
      "grad_norm": 0.07178529382405721,
      "learning_rate": 1.6474519074198058e-05,
      "loss": 0.8462,
      "step": 236150
    },
    {
      "epoch": 4.146140206113169,
      "grad_norm": 0.05183880292308848,
      "learning_rate": 1.6468718088947793e-05,
      "loss": 0.8415,
      "step": 236160
    },
    {
      "epoch": 4.14631577099317,
      "grad_norm": 0.046965253943769686,
      "learning_rate": 1.646291817472506e-05,
      "loss": 0.844,
      "step": 236170
    },
    {
      "epoch": 4.146491335873172,
      "grad_norm": 0.04388312515821237,
      "learning_rate": 1.645711933160334e-05,
      "loss": 0.8382,
      "step": 236180
    },
    {
      "epoch": 4.146666900753173,
      "grad_norm": 0.0466001946899459,
      "learning_rate": 1.6451321559656153e-05,
      "loss": 0.8459,
      "step": 236190
    },
    {
      "epoch": 4.146842465633175,
      "grad_norm": 0.06064388529568784,
      "learning_rate": 1.64455248589569e-05,
      "loss": 0.8363,
      "step": 236200
    },
    {
      "epoch": 4.147018030513176,
      "grad_norm": 0.055734733046907695,
      "learning_rate": 1.643972922957905e-05,
      "loss": 0.8412,
      "step": 236210
    },
    {
      "epoch": 4.147193595393178,
      "grad_norm": 0.04862710298526778,
      "learning_rate": 1.6433934671596068e-05,
      "loss": 0.8349,
      "step": 236220
    },
    {
      "epoch": 4.147369160273179,
      "grad_norm": 0.046437366471734784,
      "learning_rate": 1.642814118508131e-05,
      "loss": 0.8438,
      "step": 236230
    },
    {
      "epoch": 4.14754472515318,
      "grad_norm": 0.05732199042906949,
      "learning_rate": 1.642234877010822e-05,
      "loss": 0.8436,
      "step": 236240
    },
    {
      "epoch": 4.147720290033182,
      "grad_norm": 0.054608412367273075,
      "learning_rate": 1.6416557426750168e-05,
      "loss": 0.8386,
      "step": 236250
    },
    {
      "epoch": 4.147895854913183,
      "grad_norm": 0.04967602616705417,
      "learning_rate": 1.6410767155080574e-05,
      "loss": 0.838,
      "step": 236260
    },
    {
      "epoch": 4.148071419793185,
      "grad_norm": 0.04950897904272935,
      "learning_rate": 1.6404977955172742e-05,
      "loss": 0.8427,
      "step": 236270
    },
    {
      "epoch": 4.148246984673186,
      "grad_norm": 0.051482928242632404,
      "learning_rate": 1.6399189827100057e-05,
      "loss": 0.8308,
      "step": 236280
    },
    {
      "epoch": 4.148422549553187,
      "grad_norm": 0.16924808896800042,
      "learning_rate": 1.6393402770935867e-05,
      "loss": 0.8383,
      "step": 236290
    },
    {
      "epoch": 4.148598114433189,
      "grad_norm": 0.053831059090021546,
      "learning_rate": 1.6387616786753464e-05,
      "loss": 0.8375,
      "step": 236300
    },
    {
      "epoch": 4.14877367931319,
      "grad_norm": 0.06887628203703657,
      "learning_rate": 1.6381831874626175e-05,
      "loss": 0.8444,
      "step": 236310
    },
    {
      "epoch": 4.148949244193192,
      "grad_norm": 0.04751683187422008,
      "learning_rate": 1.637604803462728e-05,
      "loss": 0.8438,
      "step": 236320
    },
    {
      "epoch": 4.149124809073193,
      "grad_norm": 0.06049073846120699,
      "learning_rate": 1.637026526683011e-05,
      "loss": 0.8417,
      "step": 236330
    },
    {
      "epoch": 4.149300373953194,
      "grad_norm": 0.04162189838705405,
      "learning_rate": 1.6364483571307872e-05,
      "loss": 0.8424,
      "step": 236340
    },
    {
      "epoch": 4.149475938833196,
      "grad_norm": 0.04583285658284482,
      "learning_rate": 1.6358702948133846e-05,
      "loss": 0.8415,
      "step": 236350
    },
    {
      "epoch": 4.149651503713197,
      "grad_norm": 0.045116540655720155,
      "learning_rate": 1.6352923397381293e-05,
      "loss": 0.8455,
      "step": 236360
    },
    {
      "epoch": 4.149827068593199,
      "grad_norm": 0.04454238913467225,
      "learning_rate": 1.6347144919123405e-05,
      "loss": 0.8365,
      "step": 236370
    },
    {
      "epoch": 4.1500026334732,
      "grad_norm": 0.04998332115527799,
      "learning_rate": 1.634136751343341e-05,
      "loss": 0.8394,
      "step": 236380
    },
    {
      "epoch": 4.150178198353202,
      "grad_norm": 0.0482208698769195,
      "learning_rate": 1.6335591180384545e-05,
      "loss": 0.8408,
      "step": 236390
    },
    {
      "epoch": 4.150353763233203,
      "grad_norm": 0.05935776303785838,
      "learning_rate": 1.6329815920049947e-05,
      "loss": 0.8387,
      "step": 236400
    },
    {
      "epoch": 4.150529328113204,
      "grad_norm": 0.046425093503225046,
      "learning_rate": 1.6324041732502797e-05,
      "loss": 0.8483,
      "step": 236410
    },
    {
      "epoch": 4.150704892993206,
      "grad_norm": 0.05477139883367371,
      "learning_rate": 1.6318268617816272e-05,
      "loss": 0.8451,
      "step": 236420
    },
    {
      "epoch": 4.150880457873207,
      "grad_norm": 0.04902494497142857,
      "learning_rate": 1.6312496576063525e-05,
      "loss": 0.843,
      "step": 236430
    },
    {
      "epoch": 4.151056022753209,
      "grad_norm": 0.046023690259519585,
      "learning_rate": 1.6306725607317662e-05,
      "loss": 0.8434,
      "step": 236440
    },
    {
      "epoch": 4.15123158763321,
      "grad_norm": 0.065943493164625,
      "learning_rate": 1.6300955711651816e-05,
      "loss": 0.8371,
      "step": 236450
    },
    {
      "epoch": 4.151407152513212,
      "grad_norm": 0.046992222846045074,
      "learning_rate": 1.6295186889139126e-05,
      "loss": 0.8429,
      "step": 236460
    },
    {
      "epoch": 4.151582717393213,
      "grad_norm": 0.04273502380859984,
      "learning_rate": 1.628941913985261e-05,
      "loss": 0.8401,
      "step": 236470
    },
    {
      "epoch": 4.151758282273214,
      "grad_norm": 0.05098981613691101,
      "learning_rate": 1.62836524638654e-05,
      "loss": 0.839,
      "step": 236480
    },
    {
      "epoch": 4.151933847153216,
      "grad_norm": 0.053892519393925004,
      "learning_rate": 1.6277886861250555e-05,
      "loss": 0.8395,
      "step": 236490
    },
    {
      "epoch": 4.152109412033217,
      "grad_norm": 0.04865832622274255,
      "learning_rate": 1.6272122332081127e-05,
      "loss": 0.8499,
      "step": 236500
    },
    {
      "epoch": 4.152284976913219,
      "grad_norm": 0.046905171768748734,
      "learning_rate": 1.6266358876430136e-05,
      "loss": 0.8474,
      "step": 236510
    },
    {
      "epoch": 4.15246054179322,
      "grad_norm": 0.050033143153869246,
      "learning_rate": 1.6260596494370614e-05,
      "loss": 0.8264,
      "step": 236520
    },
    {
      "epoch": 4.152636106673221,
      "grad_norm": 0.05698828588424838,
      "learning_rate": 1.62548351859756e-05,
      "loss": 0.8386,
      "step": 236530
    },
    {
      "epoch": 4.152811671553223,
      "grad_norm": 0.06563010639159209,
      "learning_rate": 1.6249074951318037e-05,
      "loss": 0.8339,
      "step": 236540
    },
    {
      "epoch": 4.152987236433224,
      "grad_norm": 0.06883984108177145,
      "learning_rate": 1.6243315790470942e-05,
      "loss": 0.8315,
      "step": 236550
    },
    {
      "epoch": 4.153162801313226,
      "grad_norm": 0.07728611662406826,
      "learning_rate": 1.62375577035073e-05,
      "loss": 0.8342,
      "step": 236560
    },
    {
      "epoch": 4.153338366193227,
      "grad_norm": 0.05486314943498038,
      "learning_rate": 1.6231800690500028e-05,
      "loss": 0.8397,
      "step": 236570
    },
    {
      "epoch": 4.1535139310732285,
      "grad_norm": 0.053826608600699516,
      "learning_rate": 1.6226044751522082e-05,
      "loss": 0.8345,
      "step": 236580
    },
    {
      "epoch": 4.15368949595323,
      "grad_norm": 0.11458636235548834,
      "learning_rate": 1.6220289886646396e-05,
      "loss": 0.841,
      "step": 236590
    },
    {
      "epoch": 4.153865060833231,
      "grad_norm": 0.053723628479885224,
      "learning_rate": 1.621453609594591e-05,
      "loss": 0.8443,
      "step": 236600
    },
    {
      "epoch": 4.1540406257132325,
      "grad_norm": 0.042047771751866865,
      "learning_rate": 1.6208783379493478e-05,
      "loss": 0.8403,
      "step": 236610
    },
    {
      "epoch": 4.1542161905932335,
      "grad_norm": 0.04806760671409712,
      "learning_rate": 1.6203031737362003e-05,
      "loss": 0.8341,
      "step": 236620
    },
    {
      "epoch": 4.1543917554732355,
      "grad_norm": 0.04580662825110685,
      "learning_rate": 1.6197281169624392e-05,
      "loss": 0.8477,
      "step": 236630
    },
    {
      "epoch": 4.1545673203532365,
      "grad_norm": 0.056300080288815016,
      "learning_rate": 1.619153167635346e-05,
      "loss": 0.8395,
      "step": 236640
    },
    {
      "epoch": 4.1547428852332375,
      "grad_norm": 0.05451166660784311,
      "learning_rate": 1.618578325762208e-05,
      "loss": 0.842,
      "step": 236650
    },
    {
      "epoch": 4.1549184501132395,
      "grad_norm": 0.05468351209334234,
      "learning_rate": 1.6180035913503082e-05,
      "loss": 0.8343,
      "step": 236660
    },
    {
      "epoch": 4.1550940149932405,
      "grad_norm": 0.05493109022947597,
      "learning_rate": 1.617428964406931e-05,
      "loss": 0.8359,
      "step": 236670
    },
    {
      "epoch": 4.155269579873242,
      "grad_norm": 0.05544214683071618,
      "learning_rate": 1.616854444939351e-05,
      "loss": 0.8307,
      "step": 236680
    },
    {
      "epoch": 4.1554451447532434,
      "grad_norm": 0.04459873923791056,
      "learning_rate": 1.6162800329548523e-05,
      "loss": 0.8407,
      "step": 236690
    },
    {
      "epoch": 4.155620709633245,
      "grad_norm": 0.04676636056660861,
      "learning_rate": 1.6157057284607126e-05,
      "loss": 0.8349,
      "step": 236700
    },
    {
      "epoch": 4.155796274513246,
      "grad_norm": 0.0589630738133615,
      "learning_rate": 1.6151315314642054e-05,
      "loss": 0.839,
      "step": 236710
    },
    {
      "epoch": 4.155971839393247,
      "grad_norm": 0.04612836240794525,
      "learning_rate": 1.614557441972607e-05,
      "loss": 0.8353,
      "step": 236720
    },
    {
      "epoch": 4.156147404273249,
      "grad_norm": 0.05478908569954184,
      "learning_rate": 1.6139834599931956e-05,
      "loss": 0.8369,
      "step": 236730
    },
    {
      "epoch": 4.15632296915325,
      "grad_norm": 0.04852720758487893,
      "learning_rate": 1.613409585533238e-05,
      "loss": 0.8279,
      "step": 236740
    },
    {
      "epoch": 4.156498534033252,
      "grad_norm": 0.07508255153893044,
      "learning_rate": 1.6128358186000065e-05,
      "loss": 0.834,
      "step": 236750
    },
    {
      "epoch": 4.156674098913253,
      "grad_norm": 0.040427576470996915,
      "learning_rate": 1.612262159200772e-05,
      "loss": 0.8435,
      "step": 236760
    },
    {
      "epoch": 4.156849663793254,
      "grad_norm": 0.044259710695286635,
      "learning_rate": 1.6116886073428046e-05,
      "loss": 0.844,
      "step": 236770
    },
    {
      "epoch": 4.157025228673256,
      "grad_norm": 0.0561574071613516,
      "learning_rate": 1.6111151630333674e-05,
      "loss": 0.8326,
      "step": 236780
    },
    {
      "epoch": 4.157200793553257,
      "grad_norm": 0.0460077361208732,
      "learning_rate": 1.610541826279729e-05,
      "loss": 0.8347,
      "step": 236790
    },
    {
      "epoch": 4.157376358433259,
      "grad_norm": 0.05606647031732127,
      "learning_rate": 1.609968597089153e-05,
      "loss": 0.8328,
      "step": 236800
    },
    {
      "epoch": 4.15755192331326,
      "grad_norm": 0.04590145262521,
      "learning_rate": 1.6093954754689015e-05,
      "loss": 0.8433,
      "step": 236810
    },
    {
      "epoch": 4.157727488193262,
      "grad_norm": 0.056963302655967475,
      "learning_rate": 1.6088224614262352e-05,
      "loss": 0.8299,
      "step": 236820
    },
    {
      "epoch": 4.157903053073263,
      "grad_norm": 0.06070149308009766,
      "learning_rate": 1.6082495549684166e-05,
      "loss": 0.8343,
      "step": 236830
    },
    {
      "epoch": 4.158078617953264,
      "grad_norm": 0.04691397168672473,
      "learning_rate": 1.6076767561027046e-05,
      "loss": 0.8357,
      "step": 236840
    },
    {
      "epoch": 4.158254182833266,
      "grad_norm": 0.054236265878029835,
      "learning_rate": 1.6071040648363536e-05,
      "loss": 0.8457,
      "step": 236850
    },
    {
      "epoch": 4.158429747713267,
      "grad_norm": 0.05124961035247888,
      "learning_rate": 1.6065314811766227e-05,
      "loss": 0.8415,
      "step": 236860
    },
    {
      "epoch": 4.158605312593269,
      "grad_norm": 0.07765043641203366,
      "learning_rate": 1.6059590051307676e-05,
      "loss": 0.8425,
      "step": 236870
    },
    {
      "epoch": 4.15878087747327,
      "grad_norm": 0.07064221166643189,
      "learning_rate": 1.6053866367060375e-05,
      "loss": 0.8418,
      "step": 236880
    },
    {
      "epoch": 4.158956442353272,
      "grad_norm": 0.06201970215749725,
      "learning_rate": 1.604814375909686e-05,
      "loss": 0.8392,
      "step": 236890
    },
    {
      "epoch": 4.159132007233273,
      "grad_norm": 0.06128914719511327,
      "learning_rate": 1.6042422227489683e-05,
      "loss": 0.8426,
      "step": 236900
    },
    {
      "epoch": 4.159307572113274,
      "grad_norm": 0.06959942277781714,
      "learning_rate": 1.603670177231127e-05,
      "loss": 0.8405,
      "step": 236910
    },
    {
      "epoch": 4.159483136993276,
      "grad_norm": 0.0476633003433106,
      "learning_rate": 1.6030982393634114e-05,
      "loss": 0.837,
      "step": 236920
    },
    {
      "epoch": 4.159658701873277,
      "grad_norm": 0.051979062890963275,
      "learning_rate": 1.602526409153071e-05,
      "loss": 0.8354,
      "step": 236930
    },
    {
      "epoch": 4.159834266753279,
      "grad_norm": 0.04492333516414296,
      "learning_rate": 1.601954686607352e-05,
      "loss": 0.8456,
      "step": 236940
    },
    {
      "epoch": 4.16000983163328,
      "grad_norm": 0.053928238339920415,
      "learning_rate": 1.601383071733492e-05,
      "loss": 0.8412,
      "step": 236950
    },
    {
      "epoch": 4.160185396513281,
      "grad_norm": 0.04417052735165964,
      "learning_rate": 1.600811564538739e-05,
      "loss": 0.8351,
      "step": 236960
    },
    {
      "epoch": 4.160360961393283,
      "grad_norm": 0.04890983767434255,
      "learning_rate": 1.6002401650303335e-05,
      "loss": 0.8378,
      "step": 236970
    },
    {
      "epoch": 4.160536526273284,
      "grad_norm": 0.05537320622626246,
      "learning_rate": 1.5996688732155126e-05,
      "loss": 0.8388,
      "step": 236980
    },
    {
      "epoch": 4.160712091153286,
      "grad_norm": 0.058094114262345994,
      "learning_rate": 1.5990976891015147e-05,
      "loss": 0.8354,
      "step": 236990
    },
    {
      "epoch": 4.160887656033287,
      "grad_norm": 0.04848185044570661,
      "learning_rate": 1.59852661269558e-05,
      "loss": 0.8426,
      "step": 237000
    },
    {
      "epoch": 4.161063220913288,
      "grad_norm": 0.06291936243096229,
      "learning_rate": 1.5979556440049437e-05,
      "loss": 0.8354,
      "step": 237010
    },
    {
      "epoch": 4.16123878579329,
      "grad_norm": 0.044586652081064264,
      "learning_rate": 1.597384783036836e-05,
      "loss": 0.8355,
      "step": 237020
    },
    {
      "epoch": 4.161414350673291,
      "grad_norm": 0.061842581919870195,
      "learning_rate": 1.5968140297984934e-05,
      "loss": 0.8352,
      "step": 237030
    },
    {
      "epoch": 4.161589915553293,
      "grad_norm": 0.054080246400189894,
      "learning_rate": 1.5962433842971497e-05,
      "loss": 0.8367,
      "step": 237040
    },
    {
      "epoch": 4.161765480433294,
      "grad_norm": 0.04854651497868882,
      "learning_rate": 1.595672846540029e-05,
      "loss": 0.8399,
      "step": 237050
    },
    {
      "epoch": 4.161941045313296,
      "grad_norm": 0.04830017186921659,
      "learning_rate": 1.5951024165343633e-05,
      "loss": 0.8379,
      "step": 237060
    },
    {
      "epoch": 4.162116610193297,
      "grad_norm": 0.06103574203836695,
      "learning_rate": 1.5945320942873817e-05,
      "loss": 0.8345,
      "step": 237070
    },
    {
      "epoch": 4.162292175073298,
      "grad_norm": 0.051547795610638596,
      "learning_rate": 1.593961879806307e-05,
      "loss": 0.8472,
      "step": 237080
    },
    {
      "epoch": 4.1624677399533,
      "grad_norm": 0.04719569858024964,
      "learning_rate": 1.593391773098365e-05,
      "loss": 0.8416,
      "step": 237090
    },
    {
      "epoch": 4.162643304833301,
      "grad_norm": 0.0472460152102647,
      "learning_rate": 1.592821774170779e-05,
      "loss": 0.8415,
      "step": 237100
    },
    {
      "epoch": 4.162818869713303,
      "grad_norm": 0.04805976251699496,
      "learning_rate": 1.592251883030775e-05,
      "loss": 0.8341,
      "step": 237110
    },
    {
      "epoch": 4.162994434593304,
      "grad_norm": 0.06615494413622988,
      "learning_rate": 1.591682099685567e-05,
      "loss": 0.8338,
      "step": 237120
    },
    {
      "epoch": 4.163169999473306,
      "grad_norm": 0.06304057210236624,
      "learning_rate": 1.591112424142378e-05,
      "loss": 0.8427,
      "step": 237130
    },
    {
      "epoch": 4.163345564353307,
      "grad_norm": 0.056519530910250415,
      "learning_rate": 1.5905428564084273e-05,
      "loss": 0.8334,
      "step": 237140
    },
    {
      "epoch": 4.163521129233308,
      "grad_norm": 0.057614687349324685,
      "learning_rate": 1.589973396490926e-05,
      "loss": 0.838,
      "step": 237150
    },
    {
      "epoch": 4.16369669411331,
      "grad_norm": 0.058331300911747735,
      "learning_rate": 1.5894040443970947e-05,
      "loss": 0.8403,
      "step": 237160
    },
    {
      "epoch": 4.163872258993311,
      "grad_norm": 0.05596665658772206,
      "learning_rate": 1.5888348001341442e-05,
      "loss": 0.8401,
      "step": 237170
    },
    {
      "epoch": 4.164047823873313,
      "grad_norm": 0.05631186987269469,
      "learning_rate": 1.5882656637092907e-05,
      "loss": 0.8389,
      "step": 237180
    },
    {
      "epoch": 4.164223388753314,
      "grad_norm": 0.05183086383617692,
      "learning_rate": 1.5876966351297395e-05,
      "loss": 0.836,
      "step": 237190
    },
    {
      "epoch": 4.164398953633315,
      "grad_norm": 0.04839112153845961,
      "learning_rate": 1.5871277144027035e-05,
      "loss": 0.8399,
      "step": 237200
    },
    {
      "epoch": 4.164574518513317,
      "grad_norm": 0.05096228441292146,
      "learning_rate": 1.5865589015353925e-05,
      "loss": 0.8418,
      "step": 237210
    },
    {
      "epoch": 4.164750083393318,
      "grad_norm": 0.05070805929886248,
      "learning_rate": 1.585990196535011e-05,
      "loss": 0.8412,
      "step": 237220
    },
    {
      "epoch": 4.16492564827332,
      "grad_norm": 0.04771064365845065,
      "learning_rate": 1.5854215994087648e-05,
      "loss": 0.8457,
      "step": 237230
    },
    {
      "epoch": 4.165101213153321,
      "grad_norm": 0.05821420046596766,
      "learning_rate": 1.5848531101638615e-05,
      "loss": 0.8346,
      "step": 237240
    },
    {
      "epoch": 4.165276778033323,
      "grad_norm": 0.050380196459206904,
      "learning_rate": 1.584284728807498e-05,
      "loss": 0.8385,
      "step": 237250
    },
    {
      "epoch": 4.165452342913324,
      "grad_norm": 0.05986865887273319,
      "learning_rate": 1.583716455346879e-05,
      "loss": 0.8304,
      "step": 237260
    },
    {
      "epoch": 4.165627907793325,
      "grad_norm": 0.047667895467381816,
      "learning_rate": 1.583148289789205e-05,
      "loss": 0.8368,
      "step": 237270
    },
    {
      "epoch": 4.165803472673327,
      "grad_norm": 0.07328197298065893,
      "learning_rate": 1.582580232141677e-05,
      "loss": 0.8415,
      "step": 237280
    },
    {
      "epoch": 4.165979037553328,
      "grad_norm": 0.05816886738409154,
      "learning_rate": 1.5820122824114873e-05,
      "loss": 0.8406,
      "step": 237290
    },
    {
      "epoch": 4.16615460243333,
      "grad_norm": 0.0415941708577262,
      "learning_rate": 1.5814444406058345e-05,
      "loss": 0.8341,
      "step": 237300
    },
    {
      "epoch": 4.166330167313331,
      "grad_norm": 0.05268399529138067,
      "learning_rate": 1.5808767067319156e-05,
      "loss": 0.8412,
      "step": 237310
    },
    {
      "epoch": 4.166505732193332,
      "grad_norm": 0.07599920427770494,
      "learning_rate": 1.5803090807969194e-05,
      "loss": 0.8405,
      "step": 237320
    },
    {
      "epoch": 4.166681297073334,
      "grad_norm": 0.054540809654889416,
      "learning_rate": 1.5797415628080395e-05,
      "loss": 0.8363,
      "step": 237330
    },
    {
      "epoch": 4.166856861953335,
      "grad_norm": 0.061538662327750314,
      "learning_rate": 1.5791741527724673e-05,
      "loss": 0.835,
      "step": 237340
    },
    {
      "epoch": 4.1670324268333365,
      "grad_norm": 0.06493535824630037,
      "learning_rate": 1.578606850697393e-05,
      "loss": 0.8481,
      "step": 237350
    },
    {
      "epoch": 4.167207991713338,
      "grad_norm": 0.05139497308999755,
      "learning_rate": 1.5780396565900016e-05,
      "loss": 0.8409,
      "step": 237360
    },
    {
      "epoch": 4.1673835565933395,
      "grad_norm": 0.063509201220524,
      "learning_rate": 1.5774725704574804e-05,
      "loss": 0.837,
      "step": 237370
    },
    {
      "epoch": 4.1675591214733405,
      "grad_norm": 0.0626869586152941,
      "learning_rate": 1.576905592307018e-05,
      "loss": 0.8392,
      "step": 237380
    },
    {
      "epoch": 4.1677346863533415,
      "grad_norm": 0.048931749244859164,
      "learning_rate": 1.5763387221457926e-05,
      "loss": 0.8348,
      "step": 237390
    },
    {
      "epoch": 4.1679102512333435,
      "grad_norm": 0.047471507267144684,
      "learning_rate": 1.575771959980989e-05,
      "loss": 0.827,
      "step": 237400
    },
    {
      "epoch": 4.1680858161133445,
      "grad_norm": 0.04811257292690325,
      "learning_rate": 1.575205305819791e-05,
      "loss": 0.8505,
      "step": 237410
    },
    {
      "epoch": 4.168261380993346,
      "grad_norm": 0.05539784344065212,
      "learning_rate": 1.5746387596693722e-05,
      "loss": 0.8409,
      "step": 237420
    },
    {
      "epoch": 4.1684369458733475,
      "grad_norm": 0.05175957261514728,
      "learning_rate": 1.5740723215369148e-05,
      "loss": 0.8357,
      "step": 237430
    },
    {
      "epoch": 4.1686125107533485,
      "grad_norm": 0.04940206573455357,
      "learning_rate": 1.573505991429595e-05,
      "loss": 0.8494,
      "step": 237440
    },
    {
      "epoch": 4.16878807563335,
      "grad_norm": 0.0465056445409293,
      "learning_rate": 1.5729397693545912e-05,
      "loss": 0.832,
      "step": 237450
    },
    {
      "epoch": 4.168963640513351,
      "grad_norm": 0.05257347918348014,
      "learning_rate": 1.5723736553190728e-05,
      "loss": 0.8368,
      "step": 237460
    },
    {
      "epoch": 4.169139205393353,
      "grad_norm": 0.05267459279620543,
      "learning_rate": 1.5718076493302135e-05,
      "loss": 0.8383,
      "step": 237470
    },
    {
      "epoch": 4.169314770273354,
      "grad_norm": 0.058759991617948364,
      "learning_rate": 1.5712417513951883e-05,
      "loss": 0.8373,
      "step": 237480
    },
    {
      "epoch": 4.169490335153356,
      "grad_norm": 0.05212115295936904,
      "learning_rate": 1.5706759615211632e-05,
      "loss": 0.8403,
      "step": 237490
    },
    {
      "epoch": 4.169665900033357,
      "grad_norm": 0.046726293820642695,
      "learning_rate": 1.5701102797153068e-05,
      "loss": 0.8347,
      "step": 237500
    },
    {
      "epoch": 4.169841464913358,
      "grad_norm": 0.05068644019056657,
      "learning_rate": 1.569544705984789e-05,
      "loss": 0.8393,
      "step": 237510
    },
    {
      "epoch": 4.17001702979336,
      "grad_norm": 0.06302071799726015,
      "learning_rate": 1.5689792403367764e-05,
      "loss": 0.8406,
      "step": 237520
    },
    {
      "epoch": 4.170192594673361,
      "grad_norm": 0.052858039205676666,
      "learning_rate": 1.5684138827784295e-05,
      "loss": 0.8355,
      "step": 237530
    },
    {
      "epoch": 4.170368159553363,
      "grad_norm": 0.07099572686054266,
      "learning_rate": 1.5678486333169135e-05,
      "loss": 0.8375,
      "step": 237540
    },
    {
      "epoch": 4.170543724433364,
      "grad_norm": 0.07666764848604417,
      "learning_rate": 1.5672834919593933e-05,
      "loss": 0.8447,
      "step": 237550
    },
    {
      "epoch": 4.170719289313365,
      "grad_norm": 0.0558475366459478,
      "learning_rate": 1.5667184587130236e-05,
      "loss": 0.8394,
      "step": 237560
    },
    {
      "epoch": 4.170894854193367,
      "grad_norm": 0.044683290110733326,
      "learning_rate": 1.566153533584967e-05,
      "loss": 0.8371,
      "step": 237570
    },
    {
      "epoch": 4.171070419073368,
      "grad_norm": 0.061291423840917986,
      "learning_rate": 1.5655887165823835e-05,
      "loss": 0.8376,
      "step": 237580
    },
    {
      "epoch": 4.17124598395337,
      "grad_norm": 0.09024125421251072,
      "learning_rate": 1.5650240077124234e-05,
      "loss": 0.8315,
      "step": 237590
    },
    {
      "epoch": 4.171421548833371,
      "grad_norm": 0.05236051773235789,
      "learning_rate": 1.5644594069822447e-05,
      "loss": 0.843,
      "step": 237600
    },
    {
      "epoch": 4.171597113713373,
      "grad_norm": 0.06834745671328477,
      "learning_rate": 1.5638949143990015e-05,
      "loss": 0.8419,
      "step": 237610
    },
    {
      "epoch": 4.171772678593374,
      "grad_norm": 0.057630257263934184,
      "learning_rate": 1.563330529969849e-05,
      "loss": 0.8386,
      "step": 237620
    },
    {
      "epoch": 4.171948243473375,
      "grad_norm": 0.05600033755598047,
      "learning_rate": 1.5627662537019314e-05,
      "loss": 0.8302,
      "step": 237630
    },
    {
      "epoch": 4.172123808353377,
      "grad_norm": 0.05917292644649656,
      "learning_rate": 1.5622020856024004e-05,
      "loss": 0.8416,
      "step": 237640
    },
    {
      "epoch": 4.172299373233378,
      "grad_norm": 0.05863402948987593,
      "learning_rate": 1.561638025678409e-05,
      "loss": 0.8485,
      "step": 237650
    },
    {
      "epoch": 4.17247493811338,
      "grad_norm": 0.05362645081248306,
      "learning_rate": 1.561074073937098e-05,
      "loss": 0.832,
      "step": 237660
    },
    {
      "epoch": 4.172650502993381,
      "grad_norm": 0.09079964418284694,
      "learning_rate": 1.560510230385615e-05,
      "loss": 0.8399,
      "step": 237670
    },
    {
      "epoch": 4.172826067873382,
      "grad_norm": 0.05378746473615646,
      "learning_rate": 1.559946495031103e-05,
      "loss": 0.8377,
      "step": 237680
    },
    {
      "epoch": 4.173001632753384,
      "grad_norm": 0.04365943146091295,
      "learning_rate": 1.559382867880708e-05,
      "loss": 0.8365,
      "step": 237690
    },
    {
      "epoch": 4.173177197633385,
      "grad_norm": 0.05543504337218526,
      "learning_rate": 1.5588193489415673e-05,
      "loss": 0.8356,
      "step": 237700
    },
    {
      "epoch": 4.173352762513387,
      "grad_norm": 0.047253386912819315,
      "learning_rate": 1.5582559382208215e-05,
      "loss": 0.834,
      "step": 237710
    },
    {
      "epoch": 4.173528327393388,
      "grad_norm": 0.04496636398156135,
      "learning_rate": 1.5576926357256137e-05,
      "loss": 0.8457,
      "step": 237720
    },
    {
      "epoch": 4.17370389227339,
      "grad_norm": 0.04860665429584265,
      "learning_rate": 1.5571294414630737e-05,
      "loss": 0.8394,
      "step": 237730
    },
    {
      "epoch": 4.173879457153391,
      "grad_norm": 0.051118056731141955,
      "learning_rate": 1.5565663554403414e-05,
      "loss": 0.8371,
      "step": 237740
    },
    {
      "epoch": 4.174055022033392,
      "grad_norm": 0.0518730428114326,
      "learning_rate": 1.5560033776645526e-05,
      "loss": 0.8332,
      "step": 237750
    },
    {
      "epoch": 4.174230586913394,
      "grad_norm": 0.048891551937543354,
      "learning_rate": 1.5554405081428372e-05,
      "loss": 0.8517,
      "step": 237760
    },
    {
      "epoch": 4.174406151793395,
      "grad_norm": 0.0391635186800861,
      "learning_rate": 1.5548777468823273e-05,
      "loss": 0.8437,
      "step": 237770
    },
    {
      "epoch": 4.174581716673397,
      "grad_norm": 0.0539687883273859,
      "learning_rate": 1.5543150938901538e-05,
      "loss": 0.8333,
      "step": 237780
    },
    {
      "epoch": 4.174757281553398,
      "grad_norm": 0.06151670818098311,
      "learning_rate": 1.553752549173449e-05,
      "loss": 0.8375,
      "step": 237790
    },
    {
      "epoch": 4.1749328464334,
      "grad_norm": 0.0549945364949087,
      "learning_rate": 1.5531901127393344e-05,
      "loss": 0.8487,
      "step": 237800
    },
    {
      "epoch": 4.175108411313401,
      "grad_norm": 0.05462192648645926,
      "learning_rate": 1.552627784594939e-05,
      "loss": 0.8411,
      "step": 237810
    },
    {
      "epoch": 4.175283976193402,
      "grad_norm": 0.04614672456225115,
      "learning_rate": 1.5520655647473898e-05,
      "loss": 0.8398,
      "step": 237820
    },
    {
      "epoch": 4.175459541073404,
      "grad_norm": 0.0734560295569145,
      "learning_rate": 1.5515034532038063e-05,
      "loss": 0.8398,
      "step": 237830
    },
    {
      "epoch": 4.175635105953405,
      "grad_norm": 0.065757225490644,
      "learning_rate": 1.550941449971312e-05,
      "loss": 0.8353,
      "step": 237840
    },
    {
      "epoch": 4.175810670833407,
      "grad_norm": 0.0662112726089326,
      "learning_rate": 1.5503795550570283e-05,
      "loss": 0.8296,
      "step": 237850
    },
    {
      "epoch": 4.175986235713408,
      "grad_norm": 0.04530865926134366,
      "learning_rate": 1.5498177684680757e-05,
      "loss": 0.8419,
      "step": 237860
    },
    {
      "epoch": 4.176161800593409,
      "grad_norm": 0.057765958238872026,
      "learning_rate": 1.5492560902115695e-05,
      "loss": 0.8353,
      "step": 237870
    },
    {
      "epoch": 4.176337365473411,
      "grad_norm": 0.04622158848437269,
      "learning_rate": 1.548694520294627e-05,
      "loss": 0.8357,
      "step": 237880
    },
    {
      "epoch": 4.176512930353412,
      "grad_norm": 0.05829067144167094,
      "learning_rate": 1.5481330587243655e-05,
      "loss": 0.8415,
      "step": 237890
    },
    {
      "epoch": 4.176688495233414,
      "grad_norm": 0.04575188789061447,
      "learning_rate": 1.547571705507895e-05,
      "loss": 0.8481,
      "step": 237900
    },
    {
      "epoch": 4.176864060113415,
      "grad_norm": 0.0456266517685205,
      "learning_rate": 1.5470104606523302e-05,
      "loss": 0.853,
      "step": 237910
    },
    {
      "epoch": 4.177039624993417,
      "grad_norm": 0.06799142098790438,
      "learning_rate": 1.546449324164785e-05,
      "loss": 0.8431,
      "step": 237920
    },
    {
      "epoch": 4.177215189873418,
      "grad_norm": 0.04801579457301374,
      "learning_rate": 1.545888296052363e-05,
      "loss": 0.8352,
      "step": 237930
    },
    {
      "epoch": 4.177390754753419,
      "grad_norm": 0.05298234524790623,
      "learning_rate": 1.545327376322175e-05,
      "loss": 0.8431,
      "step": 237940
    },
    {
      "epoch": 4.177566319633421,
      "grad_norm": 0.06028313896875768,
      "learning_rate": 1.5447665649813298e-05,
      "loss": 0.8468,
      "step": 237950
    },
    {
      "epoch": 4.177741884513422,
      "grad_norm": 0.08595093063632316,
      "learning_rate": 1.5442058620369344e-05,
      "loss": 0.8381,
      "step": 237960
    },
    {
      "epoch": 4.177917449393424,
      "grad_norm": 0.053457935053715724,
      "learning_rate": 1.543645267496087e-05,
      "loss": 0.8362,
      "step": 237970
    },
    {
      "epoch": 4.178093014273425,
      "grad_norm": 0.061695570243840855,
      "learning_rate": 1.543084781365895e-05,
      "loss": 0.8397,
      "step": 237980
    },
    {
      "epoch": 4.178268579153426,
      "grad_norm": 0.0588664177801311,
      "learning_rate": 1.5425244036534604e-05,
      "loss": 0.8417,
      "step": 237990
    },
    {
      "epoch": 4.178444144033428,
      "grad_norm": 0.059200242156609566,
      "learning_rate": 1.5419641343658786e-05,
      "loss": 0.8446,
      "step": 238000
    },
    {
      "epoch": 4.178619708913429,
      "grad_norm": 0.052073721332885944,
      "learning_rate": 1.541403973510253e-05,
      "loss": 0.8456,
      "step": 238010
    },
    {
      "epoch": 4.178795273793431,
      "grad_norm": 0.053001376720151476,
      "learning_rate": 1.5408439210936777e-05,
      "loss": 0.8428,
      "step": 238020
    },
    {
      "epoch": 4.178970838673432,
      "grad_norm": 0.05679707690995328,
      "learning_rate": 1.5402839771232532e-05,
      "loss": 0.8416,
      "step": 238030
    },
    {
      "epoch": 4.179146403553434,
      "grad_norm": 0.05432236250435067,
      "learning_rate": 1.5397241416060694e-05,
      "loss": 0.8488,
      "step": 238040
    },
    {
      "epoch": 4.179321968433435,
      "grad_norm": 0.05266965896415285,
      "learning_rate": 1.5391644145492214e-05,
      "loss": 0.8357,
      "step": 238050
    },
    {
      "epoch": 4.179497533313436,
      "grad_norm": 0.05344514242241278,
      "learning_rate": 1.5386047959598022e-05,
      "loss": 0.8425,
      "step": 238060
    },
    {
      "epoch": 4.179673098193438,
      "grad_norm": 0.053132226976448996,
      "learning_rate": 1.5380452858448998e-05,
      "loss": 0.8295,
      "step": 238070
    },
    {
      "epoch": 4.179848663073439,
      "grad_norm": 0.05410261630711112,
      "learning_rate": 1.5374858842116032e-05,
      "loss": 0.8379,
      "step": 238080
    },
    {
      "epoch": 4.1800242279534405,
      "grad_norm": 0.046992945798786843,
      "learning_rate": 1.536926591067005e-05,
      "loss": 0.8336,
      "step": 238090
    },
    {
      "epoch": 4.180199792833442,
      "grad_norm": 0.05624715944763452,
      "learning_rate": 1.536367406418185e-05,
      "loss": 0.8325,
      "step": 238100
    },
    {
      "epoch": 4.180375357713443,
      "grad_norm": 0.04987329351990984,
      "learning_rate": 1.5358083302722316e-05,
      "loss": 0.8387,
      "step": 238110
    },
    {
      "epoch": 4.1805509225934445,
      "grad_norm": 0.05591762346623597,
      "learning_rate": 1.5352493626362277e-05,
      "loss": 0.8313,
      "step": 238120
    },
    {
      "epoch": 4.1807264874734456,
      "grad_norm": 0.06251201261890421,
      "learning_rate": 1.5346905035172583e-05,
      "loss": 0.8356,
      "step": 238130
    },
    {
      "epoch": 4.1809020523534475,
      "grad_norm": 0.054127012745713673,
      "learning_rate": 1.5341317529223998e-05,
      "loss": 0.8434,
      "step": 238140
    },
    {
      "epoch": 4.1810776172334485,
      "grad_norm": 0.08327994273071626,
      "learning_rate": 1.5335731108587324e-05,
      "loss": 0.8448,
      "step": 238150
    },
    {
      "epoch": 4.18125318211345,
      "grad_norm": 0.056472425914585574,
      "learning_rate": 1.5330145773333394e-05,
      "loss": 0.8326,
      "step": 238160
    },
    {
      "epoch": 4.1814287469934515,
      "grad_norm": 0.08418211993514242,
      "learning_rate": 1.53245615235329e-05,
      "loss": 0.8335,
      "step": 238170
    },
    {
      "epoch": 4.1816043118734525,
      "grad_norm": 0.05103268395533753,
      "learning_rate": 1.5318978359256634e-05,
      "loss": 0.8345,
      "step": 238180
    },
    {
      "epoch": 4.181779876753454,
      "grad_norm": 0.05470066149117324,
      "learning_rate": 1.5313396280575327e-05,
      "loss": 0.8374,
      "step": 238190
    },
    {
      "epoch": 4.1819554416334555,
      "grad_norm": 0.057853097571775244,
      "learning_rate": 1.5307815287559744e-05,
      "loss": 0.8423,
      "step": 238200
    },
    {
      "epoch": 4.182131006513457,
      "grad_norm": 0.05315194395924367,
      "learning_rate": 1.530223538028054e-05,
      "loss": 0.8427,
      "step": 238210
    },
    {
      "epoch": 4.182306571393458,
      "grad_norm": 0.051635923520867315,
      "learning_rate": 1.529665655880844e-05,
      "loss": 0.8392,
      "step": 238220
    },
    {
      "epoch": 4.182482136273459,
      "grad_norm": 0.04540957503456549,
      "learning_rate": 1.5291078823214143e-05,
      "loss": 0.8445,
      "step": 238230
    },
    {
      "epoch": 4.182657701153461,
      "grad_norm": 0.04710786587161601,
      "learning_rate": 1.5285502173568283e-05,
      "loss": 0.8373,
      "step": 238240
    },
    {
      "epoch": 4.182833266033462,
      "grad_norm": 0.061006359790733745,
      "learning_rate": 1.5279926609941534e-05,
      "loss": 0.8386,
      "step": 238250
    },
    {
      "epoch": 4.183008830913464,
      "grad_norm": 0.057740611063476464,
      "learning_rate": 1.5274352132404568e-05,
      "loss": 0.85,
      "step": 238260
    },
    {
      "epoch": 4.183184395793465,
      "grad_norm": 0.054706696408251834,
      "learning_rate": 1.5268778741027968e-05,
      "loss": 0.8379,
      "step": 238270
    },
    {
      "epoch": 4.183359960673467,
      "grad_norm": 0.04964071688304258,
      "learning_rate": 1.5263206435882368e-05,
      "loss": 0.8412,
      "step": 238280
    },
    {
      "epoch": 4.183535525553468,
      "grad_norm": 0.04826037468763159,
      "learning_rate": 1.5257635217038374e-05,
      "loss": 0.8385,
      "step": 238290
    },
    {
      "epoch": 4.183711090433469,
      "grad_norm": 0.0517257219026431,
      "learning_rate": 1.5252065084566606e-05,
      "loss": 0.8426,
      "step": 238300
    },
    {
      "epoch": 4.183886655313471,
      "grad_norm": 0.046895920357134614,
      "learning_rate": 1.5246496038537578e-05,
      "loss": 0.8482,
      "step": 238310
    },
    {
      "epoch": 4.184062220193472,
      "grad_norm": 0.04931874082693212,
      "learning_rate": 1.5240928079021882e-05,
      "loss": 0.8401,
      "step": 238320
    },
    {
      "epoch": 4.184237785073474,
      "grad_norm": 0.048411562807018416,
      "learning_rate": 1.5235361206090077e-05,
      "loss": 0.8385,
      "step": 238330
    },
    {
      "epoch": 4.184413349953475,
      "grad_norm": 0.05111557248148754,
      "learning_rate": 1.5229795419812675e-05,
      "loss": 0.8319,
      "step": 238340
    },
    {
      "epoch": 4.184588914833476,
      "grad_norm": 0.047961582454427186,
      "learning_rate": 1.5224230720260194e-05,
      "loss": 0.8409,
      "step": 238350
    },
    {
      "epoch": 4.184764479713478,
      "grad_norm": 0.06208081091379162,
      "learning_rate": 1.521866710750314e-05,
      "loss": 0.8417,
      "step": 238360
    },
    {
      "epoch": 4.184940044593479,
      "grad_norm": 0.05203853841828185,
      "learning_rate": 1.521310458161205e-05,
      "loss": 0.8383,
      "step": 238370
    },
    {
      "epoch": 4.185115609473481,
      "grad_norm": 0.06759735805714266,
      "learning_rate": 1.5207543142657337e-05,
      "loss": 0.8416,
      "step": 238380
    },
    {
      "epoch": 4.185291174353482,
      "grad_norm": 0.05611924766516319,
      "learning_rate": 1.5201982790709495e-05,
      "loss": 0.837,
      "step": 238390
    },
    {
      "epoch": 4.185466739233484,
      "grad_norm": 0.0634211908584395,
      "learning_rate": 1.5196423525838996e-05,
      "loss": 0.8413,
      "step": 238400
    },
    {
      "epoch": 4.185642304113485,
      "grad_norm": 0.061739196937488326,
      "learning_rate": 1.5190865348116236e-05,
      "loss": 0.8359,
      "step": 238410
    },
    {
      "epoch": 4.185817868993486,
      "grad_norm": 0.06297684707926873,
      "learning_rate": 1.5185308257611653e-05,
      "loss": 0.8392,
      "step": 238420
    },
    {
      "epoch": 4.185993433873488,
      "grad_norm": 0.06281775791925857,
      "learning_rate": 1.5179752254395682e-05,
      "loss": 0.8399,
      "step": 238430
    },
    {
      "epoch": 4.186168998753489,
      "grad_norm": 0.042584733783266424,
      "learning_rate": 1.5174197338538684e-05,
      "loss": 0.8356,
      "step": 238440
    },
    {
      "epoch": 4.186344563633491,
      "grad_norm": 0.05077358454852594,
      "learning_rate": 1.5168643510111047e-05,
      "loss": 0.8457,
      "step": 238450
    },
    {
      "epoch": 4.186520128513492,
      "grad_norm": 0.05464533757142158,
      "learning_rate": 1.5163090769183152e-05,
      "loss": 0.8368,
      "step": 238460
    },
    {
      "epoch": 4.186695693393494,
      "grad_norm": 0.051602188373002636,
      "learning_rate": 1.5157539115825362e-05,
      "loss": 0.8408,
      "step": 238470
    },
    {
      "epoch": 4.186871258273495,
      "grad_norm": 0.04942576612999989,
      "learning_rate": 1.5151988550107982e-05,
      "loss": 0.8366,
      "step": 238480
    },
    {
      "epoch": 4.187046823153496,
      "grad_norm": 0.059212400463179556,
      "learning_rate": 1.5146439072101368e-05,
      "loss": 0.8334,
      "step": 238490
    },
    {
      "epoch": 4.187222388033498,
      "grad_norm": 0.04668449345944676,
      "learning_rate": 1.5140890681875841e-05,
      "loss": 0.8337,
      "step": 238500
    },
    {
      "epoch": 4.187397952913499,
      "grad_norm": 0.06368415265119988,
      "learning_rate": 1.513534337950166e-05,
      "loss": 0.8434,
      "step": 238510
    },
    {
      "epoch": 4.187573517793501,
      "grad_norm": 0.055549351584762596,
      "learning_rate": 1.512979716504913e-05,
      "loss": 0.8392,
      "step": 238520
    },
    {
      "epoch": 4.187749082673502,
      "grad_norm": 0.06573308595289327,
      "learning_rate": 1.5124252038588538e-05,
      "loss": 0.8401,
      "step": 238530
    },
    {
      "epoch": 4.187924647553503,
      "grad_norm": 0.05313618164342479,
      "learning_rate": 1.5118708000190144e-05,
      "loss": 0.8408,
      "step": 238540
    },
    {
      "epoch": 4.188100212433505,
      "grad_norm": 0.04443712513462778,
      "learning_rate": 1.5113165049924158e-05,
      "loss": 0.8396,
      "step": 238550
    },
    {
      "epoch": 4.188275777313506,
      "grad_norm": 0.08258350336148287,
      "learning_rate": 1.510762318786083e-05,
      "loss": 0.8437,
      "step": 238560
    },
    {
      "epoch": 4.188451342193508,
      "grad_norm": 0.0472689925518072,
      "learning_rate": 1.5102082414070404e-05,
      "loss": 0.8322,
      "step": 238570
    },
    {
      "epoch": 4.188626907073509,
      "grad_norm": 0.06684211566884615,
      "learning_rate": 1.5096542728623034e-05,
      "loss": 0.8409,
      "step": 238580
    },
    {
      "epoch": 4.18880247195351,
      "grad_norm": 0.046224175795940375,
      "learning_rate": 1.5091004131588934e-05,
      "loss": 0.8408,
      "step": 238590
    },
    {
      "epoch": 4.188978036833512,
      "grad_norm": 0.041109895002831624,
      "learning_rate": 1.5085466623038295e-05,
      "loss": 0.836,
      "step": 238600
    },
    {
      "epoch": 4.189153601713513,
      "grad_norm": 0.06351291250265909,
      "learning_rate": 1.5079930203041238e-05,
      "loss": 0.8408,
      "step": 238610
    },
    {
      "epoch": 4.189329166593515,
      "grad_norm": 0.05287773736716652,
      "learning_rate": 1.507439487166792e-05,
      "loss": 0.8354,
      "step": 238620
    },
    {
      "epoch": 4.189504731473516,
      "grad_norm": 0.051011986095251254,
      "learning_rate": 1.50688606289885e-05,
      "loss": 0.8302,
      "step": 238630
    },
    {
      "epoch": 4.189680296353518,
      "grad_norm": 0.051753119312690096,
      "learning_rate": 1.5063327475073102e-05,
      "loss": 0.8415,
      "step": 238640
    },
    {
      "epoch": 4.189855861233519,
      "grad_norm": 0.06020016985127818,
      "learning_rate": 1.5057795409991793e-05,
      "loss": 0.8327,
      "step": 238650
    },
    {
      "epoch": 4.19003142611352,
      "grad_norm": 0.04974031002722971,
      "learning_rate": 1.5052264433814686e-05,
      "loss": 0.8285,
      "step": 238660
    },
    {
      "epoch": 4.190206990993522,
      "grad_norm": 0.05084526883883265,
      "learning_rate": 1.5046734546611885e-05,
      "loss": 0.8363,
      "step": 238670
    },
    {
      "epoch": 4.190382555873523,
      "grad_norm": 0.049117185458053936,
      "learning_rate": 1.5041205748453413e-05,
      "loss": 0.8335,
      "step": 238680
    },
    {
      "epoch": 4.190558120753525,
      "grad_norm": 0.043426739301558714,
      "learning_rate": 1.5035678039409328e-05,
      "loss": 0.8304,
      "step": 238690
    },
    {
      "epoch": 4.190733685633526,
      "grad_norm": 0.04694061840316071,
      "learning_rate": 1.5030151419549701e-05,
      "loss": 0.8388,
      "step": 238700
    },
    {
      "epoch": 4.190909250513528,
      "grad_norm": 0.1151666515540686,
      "learning_rate": 1.502462588894451e-05,
      "loss": 0.8318,
      "step": 238710
    },
    {
      "epoch": 4.191084815393529,
      "grad_norm": 0.058596476309650744,
      "learning_rate": 1.5019101447663795e-05,
      "loss": 0.8353,
      "step": 238720
    },
    {
      "epoch": 4.19126038027353,
      "grad_norm": 0.06004457338589939,
      "learning_rate": 1.501357809577753e-05,
      "loss": 0.8392,
      "step": 238730
    },
    {
      "epoch": 4.191435945153532,
      "grad_norm": 0.05209936406949446,
      "learning_rate": 1.5008055833355738e-05,
      "loss": 0.8361,
      "step": 238740
    },
    {
      "epoch": 4.191611510033533,
      "grad_norm": 0.05118218317308762,
      "learning_rate": 1.500253466046834e-05,
      "loss": 0.8403,
      "step": 238750
    },
    {
      "epoch": 4.191787074913535,
      "grad_norm": 0.048328573583636396,
      "learning_rate": 1.4997014577185307e-05,
      "loss": 0.8377,
      "step": 238760
    },
    {
      "epoch": 4.191962639793536,
      "grad_norm": 0.047682331677322054,
      "learning_rate": 1.4991495583576598e-05,
      "loss": 0.8327,
      "step": 238770
    },
    {
      "epoch": 4.192138204673537,
      "grad_norm": 0.04861208611219997,
      "learning_rate": 1.4985977679712104e-05,
      "loss": 0.8408,
      "step": 238780
    },
    {
      "epoch": 4.192313769553539,
      "grad_norm": 0.05597187846579625,
      "learning_rate": 1.4980460865661764e-05,
      "loss": 0.8444,
      "step": 238790
    },
    {
      "epoch": 4.19248933443354,
      "grad_norm": 0.04655114850839296,
      "learning_rate": 1.497494514149546e-05,
      "loss": 0.8363,
      "step": 238800
    },
    {
      "epoch": 4.192664899313542,
      "grad_norm": 0.06118203707475685,
      "learning_rate": 1.4969430507283119e-05,
      "loss": 0.8384,
      "step": 238810
    },
    {
      "epoch": 4.192840464193543,
      "grad_norm": 0.08836443897200208,
      "learning_rate": 1.4963916963094557e-05,
      "loss": 0.8412,
      "step": 238820
    },
    {
      "epoch": 4.1930160290735445,
      "grad_norm": 0.04726217725189252,
      "learning_rate": 1.4958404508999643e-05,
      "loss": 0.8287,
      "step": 238830
    },
    {
      "epoch": 4.193191593953546,
      "grad_norm": 0.06154243401955423,
      "learning_rate": 1.4952893145068273e-05,
      "loss": 0.8375,
      "step": 238840
    },
    {
      "epoch": 4.193367158833547,
      "grad_norm": 0.05242484017947446,
      "learning_rate": 1.4947382871370206e-05,
      "loss": 0.8425,
      "step": 238850
    },
    {
      "epoch": 4.1935427237135485,
      "grad_norm": 0.05206156723495848,
      "learning_rate": 1.494187368797529e-05,
      "loss": 0.8291,
      "step": 238860
    },
    {
      "epoch": 4.19371828859355,
      "grad_norm": 0.058354860026841565,
      "learning_rate": 1.4936365594953356e-05,
      "loss": 0.8337,
      "step": 238870
    },
    {
      "epoch": 4.1938938534735515,
      "grad_norm": 0.040664504757747945,
      "learning_rate": 1.4930858592374141e-05,
      "loss": 0.838,
      "step": 238880
    },
    {
      "epoch": 4.1940694183535525,
      "grad_norm": 0.04431804307988354,
      "learning_rate": 1.4925352680307439e-05,
      "loss": 0.8435,
      "step": 238890
    },
    {
      "epoch": 4.1942449832335535,
      "grad_norm": 0.04157011012163535,
      "learning_rate": 1.4919847858823022e-05,
      "loss": 0.8422,
      "step": 238900
    },
    {
      "epoch": 4.1944205481135555,
      "grad_norm": 0.05467649292908454,
      "learning_rate": 1.491434412799064e-05,
      "loss": 0.8479,
      "step": 238910
    },
    {
      "epoch": 4.1945961129935565,
      "grad_norm": 0.07201497248590844,
      "learning_rate": 1.4908841487880001e-05,
      "loss": 0.8359,
      "step": 238920
    },
    {
      "epoch": 4.194771677873558,
      "grad_norm": 0.0454261056214724,
      "learning_rate": 1.4903339938560834e-05,
      "loss": 0.8334,
      "step": 238930
    },
    {
      "epoch": 4.1949472427535595,
      "grad_norm": 0.06381271636284838,
      "learning_rate": 1.4897839480102884e-05,
      "loss": 0.8385,
      "step": 238940
    },
    {
      "epoch": 4.195122807633561,
      "grad_norm": 0.05110857931001891,
      "learning_rate": 1.4892340112575775e-05,
      "loss": 0.8401,
      "step": 238950
    },
    {
      "epoch": 4.195298372513562,
      "grad_norm": 0.050699325390513,
      "learning_rate": 1.488684183604923e-05,
      "loss": 0.8483,
      "step": 238960
    },
    {
      "epoch": 4.1954739373935634,
      "grad_norm": 0.09506957458651517,
      "learning_rate": 1.4881344650592896e-05,
      "loss": 0.8462,
      "step": 238970
    },
    {
      "epoch": 4.195649502273565,
      "grad_norm": 0.04977119563810642,
      "learning_rate": 1.487584855627645e-05,
      "loss": 0.833,
      "step": 238980
    },
    {
      "epoch": 4.195825067153566,
      "grad_norm": 0.051269208447805394,
      "learning_rate": 1.4870353553169483e-05,
      "loss": 0.8355,
      "step": 238990
    },
    {
      "epoch": 4.196000632033568,
      "grad_norm": 0.05439355174028603,
      "learning_rate": 1.4864859641341655e-05,
      "loss": 0.8344,
      "step": 239000
    },
    {
      "epoch": 4.196176196913569,
      "grad_norm": 0.08036506277291794,
      "learning_rate": 1.4859366820862583e-05,
      "loss": 0.8394,
      "step": 239010
    },
    {
      "epoch": 4.19635176179357,
      "grad_norm": 0.050670245958855824,
      "learning_rate": 1.485387509180181e-05,
      "loss": 0.8449,
      "step": 239020
    },
    {
      "epoch": 4.196527326673572,
      "grad_norm": 0.060054114104270785,
      "learning_rate": 1.4848384454228956e-05,
      "loss": 0.8333,
      "step": 239030
    },
    {
      "epoch": 4.196702891553573,
      "grad_norm": 0.059192208346806455,
      "learning_rate": 1.4842894908213603e-05,
      "loss": 0.8325,
      "step": 239040
    },
    {
      "epoch": 4.196878456433575,
      "grad_norm": 0.055375145143179115,
      "learning_rate": 1.4837406453825258e-05,
      "loss": 0.8449,
      "step": 239050
    },
    {
      "epoch": 4.197054021313576,
      "grad_norm": 0.06777053761179155,
      "learning_rate": 1.4831919091133493e-05,
      "loss": 0.8514,
      "step": 239060
    },
    {
      "epoch": 4.197229586193578,
      "grad_norm": 0.05112911154956292,
      "learning_rate": 1.4826432820207827e-05,
      "loss": 0.8402,
      "step": 239070
    },
    {
      "epoch": 4.197405151073579,
      "grad_norm": 0.0532879719998436,
      "learning_rate": 1.4820947641117789e-05,
      "loss": 0.8382,
      "step": 239080
    },
    {
      "epoch": 4.19758071595358,
      "grad_norm": 0.06795008024360363,
      "learning_rate": 1.4815463553932848e-05,
      "loss": 0.8295,
      "step": 239090
    },
    {
      "epoch": 4.197756280833582,
      "grad_norm": 0.04642792416122419,
      "learning_rate": 1.480998055872249e-05,
      "loss": 0.8338,
      "step": 239100
    },
    {
      "epoch": 4.197931845713583,
      "grad_norm": 0.05441607737636711,
      "learning_rate": 1.480449865555622e-05,
      "loss": 0.8378,
      "step": 239110
    },
    {
      "epoch": 4.198107410593585,
      "grad_norm": 0.05229518661667282,
      "learning_rate": 1.4799017844503454e-05,
      "loss": 0.8402,
      "step": 239120
    },
    {
      "epoch": 4.198282975473586,
      "grad_norm": 0.057068406491435096,
      "learning_rate": 1.4793538125633655e-05,
      "loss": 0.8385,
      "step": 239130
    },
    {
      "epoch": 4.198458540353588,
      "grad_norm": 0.05615783779173883,
      "learning_rate": 1.4788059499016243e-05,
      "loss": 0.8362,
      "step": 239140
    },
    {
      "epoch": 4.198634105233589,
      "grad_norm": 0.05521186971599593,
      "learning_rate": 1.4782581964720663e-05,
      "loss": 0.8348,
      "step": 239150
    },
    {
      "epoch": 4.19880967011359,
      "grad_norm": 0.04982772194462838,
      "learning_rate": 1.4777105522816267e-05,
      "loss": 0.8399,
      "step": 239160
    },
    {
      "epoch": 4.198985234993592,
      "grad_norm": 0.05075504042842896,
      "learning_rate": 1.4771630173372477e-05,
      "loss": 0.8416,
      "step": 239170
    },
    {
      "epoch": 4.199160799873593,
      "grad_norm": 0.05551057015734395,
      "learning_rate": 1.476615591645867e-05,
      "loss": 0.8364,
      "step": 239180
    },
    {
      "epoch": 4.199336364753595,
      "grad_norm": 0.05562441039203986,
      "learning_rate": 1.4760682752144181e-05,
      "loss": 0.832,
      "step": 239190
    },
    {
      "epoch": 4.199511929633596,
      "grad_norm": 0.04861344602947894,
      "learning_rate": 1.4755210680498372e-05,
      "loss": 0.8435,
      "step": 239200
    },
    {
      "epoch": 4.199687494513597,
      "grad_norm": 0.057115163388982815,
      "learning_rate": 1.4749739701590584e-05,
      "loss": 0.833,
      "step": 239210
    },
    {
      "epoch": 4.199863059393599,
      "grad_norm": 0.06944293835484647,
      "learning_rate": 1.4744269815490108e-05,
      "loss": 0.8376,
      "step": 239220
    },
    {
      "epoch": 4.2000386242736,
      "grad_norm": 0.066101515900532,
      "learning_rate": 1.4738801022266258e-05,
      "loss": 0.8332,
      "step": 239230
    },
    {
      "epoch": 4.200214189153602,
      "grad_norm": 0.054092544799123184,
      "learning_rate": 1.473333332198834e-05,
      "loss": 0.8395,
      "step": 239240
    },
    {
      "epoch": 4.200389754033603,
      "grad_norm": 0.05283988460160457,
      "learning_rate": 1.4727866714725636e-05,
      "loss": 0.8387,
      "step": 239250
    },
    {
      "epoch": 4.200565318913604,
      "grad_norm": 0.04564682519469561,
      "learning_rate": 1.4722401200547376e-05,
      "loss": 0.8412,
      "step": 239260
    },
    {
      "epoch": 4.200740883793606,
      "grad_norm": 0.04564044989829147,
      "learning_rate": 1.4716936779522825e-05,
      "loss": 0.8437,
      "step": 239270
    },
    {
      "epoch": 4.200916448673607,
      "grad_norm": 0.0437978729227878,
      "learning_rate": 1.4711473451721255e-05,
      "loss": 0.8475,
      "step": 239280
    },
    {
      "epoch": 4.201092013553609,
      "grad_norm": 0.0559277399925842,
      "learning_rate": 1.4706011217211817e-05,
      "loss": 0.8344,
      "step": 239290
    },
    {
      "epoch": 4.20126757843361,
      "grad_norm": 0.05082950275858786,
      "learning_rate": 1.4700550076063763e-05,
      "loss": 0.831,
      "step": 239300
    },
    {
      "epoch": 4.201443143313612,
      "grad_norm": 0.047734592882326614,
      "learning_rate": 1.4695090028346273e-05,
      "loss": 0.8423,
      "step": 239310
    },
    {
      "epoch": 4.201618708193613,
      "grad_norm": 0.04742461018060397,
      "learning_rate": 1.4689631074128565e-05,
      "loss": 0.842,
      "step": 239320
    },
    {
      "epoch": 4.201794273073614,
      "grad_norm": 0.06667239370607894,
      "learning_rate": 1.4684173213479743e-05,
      "loss": 0.8332,
      "step": 239330
    },
    {
      "epoch": 4.201969837953616,
      "grad_norm": 0.058742184058261766,
      "learning_rate": 1.467871644646899e-05,
      "loss": 0.8344,
      "step": 239340
    },
    {
      "epoch": 4.202145402833617,
      "grad_norm": 0.06134110593429998,
      "learning_rate": 1.4673260773165465e-05,
      "loss": 0.8305,
      "step": 239350
    },
    {
      "epoch": 4.202320967713619,
      "grad_norm": 0.04359962469914271,
      "learning_rate": 1.4667806193638239e-05,
      "loss": 0.8433,
      "step": 239360
    },
    {
      "epoch": 4.20249653259362,
      "grad_norm": 0.048332450896433864,
      "learning_rate": 1.4662352707956467e-05,
      "loss": 0.8409,
      "step": 239370
    },
    {
      "epoch": 4.202672097473622,
      "grad_norm": 0.05246657147172225,
      "learning_rate": 1.4656900316189251e-05,
      "loss": 0.8377,
      "step": 239380
    },
    {
      "epoch": 4.202847662353623,
      "grad_norm": 0.04915531275091929,
      "learning_rate": 1.4651449018405627e-05,
      "loss": 0.8302,
      "step": 239390
    },
    {
      "epoch": 4.203023227233624,
      "grad_norm": 0.055099345760648594,
      "learning_rate": 1.4645998814674708e-05,
      "loss": 0.8428,
      "step": 239400
    },
    {
      "epoch": 4.203198792113626,
      "grad_norm": 0.0672299639432927,
      "learning_rate": 1.4640549705065518e-05,
      "loss": 0.8385,
      "step": 239410
    },
    {
      "epoch": 4.203374356993627,
      "grad_norm": 0.06094646922522416,
      "learning_rate": 1.4635101689647145e-05,
      "loss": 0.8377,
      "step": 239420
    },
    {
      "epoch": 4.203549921873629,
      "grad_norm": 0.049247914204005844,
      "learning_rate": 1.4629654768488565e-05,
      "loss": 0.8428,
      "step": 239430
    },
    {
      "epoch": 4.20372548675363,
      "grad_norm": 0.04481555266259817,
      "learning_rate": 1.46242089416588e-05,
      "loss": 0.8403,
      "step": 239440
    },
    {
      "epoch": 4.203901051633631,
      "grad_norm": 0.057079552484352536,
      "learning_rate": 1.4618764209226894e-05,
      "loss": 0.8373,
      "step": 239450
    },
    {
      "epoch": 4.204076616513633,
      "grad_norm": 0.047648313625265525,
      "learning_rate": 1.4613320571261777e-05,
      "loss": 0.8409,
      "step": 239460
    },
    {
      "epoch": 4.204252181393634,
      "grad_norm": 0.04216260142589827,
      "learning_rate": 1.4607878027832436e-05,
      "loss": 0.8406,
      "step": 239470
    },
    {
      "epoch": 4.204427746273636,
      "grad_norm": 0.04311285402295743,
      "learning_rate": 1.4602436579007848e-05,
      "loss": 0.84,
      "step": 239480
    },
    {
      "epoch": 4.204603311153637,
      "grad_norm": 0.04910500524783706,
      "learning_rate": 1.4596996224856964e-05,
      "loss": 0.8379,
      "step": 239490
    },
    {
      "epoch": 4.204778876033639,
      "grad_norm": 0.04664365335233457,
      "learning_rate": 1.4591556965448682e-05,
      "loss": 0.838,
      "step": 239500
    },
    {
      "epoch": 4.20495444091364,
      "grad_norm": 0.046546588757555206,
      "learning_rate": 1.4586118800851928e-05,
      "loss": 0.8419,
      "step": 239510
    },
    {
      "epoch": 4.205130005793641,
      "grad_norm": 0.05320306832080392,
      "learning_rate": 1.4580681731135625e-05,
      "loss": 0.8364,
      "step": 239520
    },
    {
      "epoch": 4.205305570673643,
      "grad_norm": 0.04640300647198697,
      "learning_rate": 1.4575245756368633e-05,
      "loss": 0.8352,
      "step": 239530
    },
    {
      "epoch": 4.205481135553644,
      "grad_norm": 0.0626675928885754,
      "learning_rate": 1.4569810876619838e-05,
      "loss": 0.8456,
      "step": 239540
    },
    {
      "epoch": 4.205656700433646,
      "grad_norm": 0.05394861355639748,
      "learning_rate": 1.4564377091958128e-05,
      "loss": 0.8399,
      "step": 239550
    },
    {
      "epoch": 4.205832265313647,
      "grad_norm": 0.13375287537607627,
      "learning_rate": 1.4558944402452298e-05,
      "loss": 0.8353,
      "step": 239560
    },
    {
      "epoch": 4.206007830193648,
      "grad_norm": 0.04707292953863844,
      "learning_rate": 1.455351280817121e-05,
      "loss": 0.8284,
      "step": 239570
    },
    {
      "epoch": 4.20618339507365,
      "grad_norm": 0.049319210607717025,
      "learning_rate": 1.454808230918368e-05,
      "loss": 0.8424,
      "step": 239580
    },
    {
      "epoch": 4.206358959953651,
      "grad_norm": 0.049159034782633754,
      "learning_rate": 1.4542652905558546e-05,
      "loss": 0.8434,
      "step": 239590
    },
    {
      "epoch": 4.2065345248336525,
      "grad_norm": 0.05243876094898421,
      "learning_rate": 1.4537224597364544e-05,
      "loss": 0.8462,
      "step": 239600
    },
    {
      "epoch": 4.206710089713654,
      "grad_norm": 0.062094705677215964,
      "learning_rate": 1.453179738467048e-05,
      "loss": 0.8351,
      "step": 239610
    },
    {
      "epoch": 4.2068856545936555,
      "grad_norm": 0.06311591305295802,
      "learning_rate": 1.4526371267545125e-05,
      "loss": 0.8328,
      "step": 239620
    },
    {
      "epoch": 4.2070612194736565,
      "grad_norm": 0.06572120926269463,
      "learning_rate": 1.4520946246057202e-05,
      "loss": 0.8402,
      "step": 239630
    },
    {
      "epoch": 4.2072367843536576,
      "grad_norm": 0.048506757815291686,
      "learning_rate": 1.4515522320275465e-05,
      "loss": 0.8339,
      "step": 239640
    },
    {
      "epoch": 4.2074123492336595,
      "grad_norm": 0.06202590339290872,
      "learning_rate": 1.4510099490268634e-05,
      "loss": 0.8388,
      "step": 239650
    },
    {
      "epoch": 4.2075879141136605,
      "grad_norm": 0.044937578249730284,
      "learning_rate": 1.4504677756105447e-05,
      "loss": 0.8328,
      "step": 239660
    },
    {
      "epoch": 4.207763478993662,
      "grad_norm": 0.053245773704435735,
      "learning_rate": 1.4499257117854537e-05,
      "loss": 0.8347,
      "step": 239670
    },
    {
      "epoch": 4.2079390438736635,
      "grad_norm": 0.046593603774677674,
      "learning_rate": 1.4493837575584629e-05,
      "loss": 0.8363,
      "step": 239680
    },
    {
      "epoch": 4.2081146087536645,
      "grad_norm": 0.04154043578838674,
      "learning_rate": 1.4488419129364395e-05,
      "loss": 0.839,
      "step": 239690
    },
    {
      "epoch": 4.208290173633666,
      "grad_norm": 0.05632678181856761,
      "learning_rate": 1.448300177926244e-05,
      "loss": 0.8304,
      "step": 239700
    },
    {
      "epoch": 4.2084657385136675,
      "grad_norm": 0.04888436198780514,
      "learning_rate": 1.4477585525347443e-05,
      "loss": 0.8438,
      "step": 239710
    },
    {
      "epoch": 4.208641303393669,
      "grad_norm": 0.04612268225629976,
      "learning_rate": 1.4472170367688042e-05,
      "loss": 0.8416,
      "step": 239720
    },
    {
      "epoch": 4.20881686827367,
      "grad_norm": 0.05862468909100746,
      "learning_rate": 1.4466756306352798e-05,
      "loss": 0.84,
      "step": 239730
    },
    {
      "epoch": 4.208992433153672,
      "grad_norm": 0.053176819738471895,
      "learning_rate": 1.446134334141033e-05,
      "loss": 0.8435,
      "step": 239740
    },
    {
      "epoch": 4.209167998033673,
      "grad_norm": 0.05347546392726441,
      "learning_rate": 1.4455931472929238e-05,
      "loss": 0.8419,
      "step": 239750
    },
    {
      "epoch": 4.209343562913674,
      "grad_norm": 0.055170817331794256,
      "learning_rate": 1.4450520700978102e-05,
      "loss": 0.8438,
      "step": 239760
    },
    {
      "epoch": 4.209519127793676,
      "grad_norm": 0.047892550815512594,
      "learning_rate": 1.444511102562543e-05,
      "loss": 0.8473,
      "step": 239770
    },
    {
      "epoch": 4.209694692673677,
      "grad_norm": 0.04542913732760854,
      "learning_rate": 1.4439702446939788e-05,
      "loss": 0.836,
      "step": 239780
    },
    {
      "epoch": 4.209870257553679,
      "grad_norm": 0.05408377941717164,
      "learning_rate": 1.4434294964989727e-05,
      "loss": 0.8439,
      "step": 239790
    },
    {
      "epoch": 4.21004582243368,
      "grad_norm": 0.050882143312471415,
      "learning_rate": 1.4428888579843725e-05,
      "loss": 0.8275,
      "step": 239800
    },
    {
      "epoch": 4.210221387313682,
      "grad_norm": 0.06243959297127503,
      "learning_rate": 1.4423483291570286e-05,
      "loss": 0.8364,
      "step": 239810
    },
    {
      "epoch": 4.210396952193683,
      "grad_norm": 0.05661655305043809,
      "learning_rate": 1.441807910023791e-05,
      "loss": 0.841,
      "step": 239820
    },
    {
      "epoch": 4.210572517073684,
      "grad_norm": 0.05660857478523464,
      "learning_rate": 1.4412676005915093e-05,
      "loss": 0.8344,
      "step": 239830
    },
    {
      "epoch": 4.210748081953686,
      "grad_norm": 0.04767426704510583,
      "learning_rate": 1.4407274008670232e-05,
      "loss": 0.8424,
      "step": 239840
    },
    {
      "epoch": 4.210923646833687,
      "grad_norm": 0.07176836486653354,
      "learning_rate": 1.4401873108571817e-05,
      "loss": 0.8408,
      "step": 239850
    },
    {
      "epoch": 4.211099211713689,
      "grad_norm": 0.04844068680834538,
      "learning_rate": 1.4396473305688275e-05,
      "loss": 0.8397,
      "step": 239860
    },
    {
      "epoch": 4.21127477659369,
      "grad_norm": 0.0567201208637584,
      "learning_rate": 1.4391074600087995e-05,
      "loss": 0.8353,
      "step": 239870
    },
    {
      "epoch": 4.211450341473691,
      "grad_norm": 0.05183923497762161,
      "learning_rate": 1.438567699183941e-05,
      "loss": 0.8445,
      "step": 239880
    },
    {
      "epoch": 4.211625906353693,
      "grad_norm": 0.061980424769381245,
      "learning_rate": 1.4380280481010906e-05,
      "loss": 0.8386,
      "step": 239890
    },
    {
      "epoch": 4.211801471233694,
      "grad_norm": 0.047036437569523123,
      "learning_rate": 1.4374885067670837e-05,
      "loss": 0.8444,
      "step": 239900
    },
    {
      "epoch": 4.211977036113696,
      "grad_norm": 0.08694814501790638,
      "learning_rate": 1.4369490751887569e-05,
      "loss": 0.8353,
      "step": 239910
    },
    {
      "epoch": 4.212152600993697,
      "grad_norm": 0.04468459914659829,
      "learning_rate": 1.4364097533729453e-05,
      "loss": 0.8353,
      "step": 239920
    },
    {
      "epoch": 4.212328165873698,
      "grad_norm": 0.09953494315730398,
      "learning_rate": 1.4358705413264848e-05,
      "loss": 0.8385,
      "step": 239930
    },
    {
      "epoch": 4.2125037307537,
      "grad_norm": 0.04862348740876168,
      "learning_rate": 1.4353314390562029e-05,
      "loss": 0.8426,
      "step": 239940
    },
    {
      "epoch": 4.212679295633701,
      "grad_norm": 0.045509243176999475,
      "learning_rate": 1.434792446568933e-05,
      "loss": 0.833,
      "step": 239950
    },
    {
      "epoch": 4.212854860513703,
      "grad_norm": 0.05819063060678466,
      "learning_rate": 1.4342535638715046e-05,
      "loss": 0.835,
      "step": 239960
    },
    {
      "epoch": 4.213030425393704,
      "grad_norm": 0.05098627665511443,
      "learning_rate": 1.4337147909707426e-05,
      "loss": 0.8337,
      "step": 239970
    },
    {
      "epoch": 4.213205990273706,
      "grad_norm": 0.05256090373784981,
      "learning_rate": 1.4331761278734759e-05,
      "loss": 0.8475,
      "step": 239980
    },
    {
      "epoch": 4.213381555153707,
      "grad_norm": 0.04977669171837467,
      "learning_rate": 1.4326375745865276e-05,
      "loss": 0.8447,
      "step": 239990
    },
    {
      "epoch": 4.213557120033708,
      "grad_norm": 0.0588241158797664,
      "learning_rate": 1.4320991311167253e-05,
      "loss": 0.8388,
      "step": 240000
    },
    {
      "epoch": 4.21373268491371,
      "grad_norm": 0.0691371800545454,
      "learning_rate": 1.431560797470885e-05,
      "loss": 0.8397,
      "step": 240010
    },
    {
      "epoch": 4.213908249793711,
      "grad_norm": 0.08658492739864736,
      "learning_rate": 1.431022573655832e-05,
      "loss": 0.8404,
      "step": 240020
    },
    {
      "epoch": 4.214083814673713,
      "grad_norm": 0.050659362197159975,
      "learning_rate": 1.4304844596783854e-05,
      "loss": 0.8384,
      "step": 240030
    },
    {
      "epoch": 4.214259379553714,
      "grad_norm": 0.04314502990934232,
      "learning_rate": 1.4299464555453606e-05,
      "loss": 0.8376,
      "step": 240040
    },
    {
      "epoch": 4.214434944433716,
      "grad_norm": 0.05785387930395185,
      "learning_rate": 1.4294085612635754e-05,
      "loss": 0.8403,
      "step": 240050
    },
    {
      "epoch": 4.214610509313717,
      "grad_norm": 0.051471994312663895,
      "learning_rate": 1.4288707768398483e-05,
      "loss": 0.836,
      "step": 240060
    },
    {
      "epoch": 4.214786074193718,
      "grad_norm": 0.041762554227087634,
      "learning_rate": 1.4283331022809876e-05,
      "loss": 0.8416,
      "step": 240070
    },
    {
      "epoch": 4.21496163907372,
      "grad_norm": 0.04746770824948644,
      "learning_rate": 1.4277955375938084e-05,
      "loss": 0.8416,
      "step": 240080
    },
    {
      "epoch": 4.215137203953721,
      "grad_norm": 0.08345358592668715,
      "learning_rate": 1.4272580827851215e-05,
      "loss": 0.8414,
      "step": 240090
    },
    {
      "epoch": 4.215312768833723,
      "grad_norm": 0.05417783968134573,
      "learning_rate": 1.4267207378617395e-05,
      "loss": 0.8299,
      "step": 240100
    },
    {
      "epoch": 4.215488333713724,
      "grad_norm": 0.04744012312150704,
      "learning_rate": 1.426183502830466e-05,
      "loss": 0.8375,
      "step": 240110
    },
    {
      "epoch": 4.215663898593725,
      "grad_norm": 0.05021094069534072,
      "learning_rate": 1.4256463776981088e-05,
      "loss": 0.8486,
      "step": 240120
    },
    {
      "epoch": 4.215839463473727,
      "grad_norm": 0.05218723167536544,
      "learning_rate": 1.425109362471477e-05,
      "loss": 0.8447,
      "step": 240130
    },
    {
      "epoch": 4.216015028353728,
      "grad_norm": 0.05336545987882545,
      "learning_rate": 1.424572457157371e-05,
      "loss": 0.8403,
      "step": 240140
    },
    {
      "epoch": 4.21619059323373,
      "grad_norm": 0.049889094040252206,
      "learning_rate": 1.4240356617625942e-05,
      "loss": 0.8431,
      "step": 240150
    },
    {
      "epoch": 4.216366158113731,
      "grad_norm": 0.05714527300062633,
      "learning_rate": 1.4234989762939477e-05,
      "loss": 0.8429,
      "step": 240160
    },
    {
      "epoch": 4.216541722993733,
      "grad_norm": 0.050431916769047104,
      "learning_rate": 1.4229624007582335e-05,
      "loss": 0.8346,
      "step": 240170
    },
    {
      "epoch": 4.216717287873734,
      "grad_norm": 0.0553296566216928,
      "learning_rate": 1.4224259351622482e-05,
      "loss": 0.8403,
      "step": 240180
    },
    {
      "epoch": 4.216892852753735,
      "grad_norm": 0.05257262810740638,
      "learning_rate": 1.4218895795127888e-05,
      "loss": 0.8418,
      "step": 240190
    },
    {
      "epoch": 4.217068417633737,
      "grad_norm": 0.06062336036857276,
      "learning_rate": 1.4213533338166533e-05,
      "loss": 0.8425,
      "step": 240200
    },
    {
      "epoch": 4.217243982513738,
      "grad_norm": 0.054374654983159264,
      "learning_rate": 1.4208171980806317e-05,
      "loss": 0.8378,
      "step": 240210
    },
    {
      "epoch": 4.21741954739374,
      "grad_norm": 0.046218201194533876,
      "learning_rate": 1.420281172311521e-05,
      "loss": 0.8376,
      "step": 240220
    },
    {
      "epoch": 4.217595112273741,
      "grad_norm": 0.047402336224244716,
      "learning_rate": 1.4197452565161124e-05,
      "loss": 0.8384,
      "step": 240230
    },
    {
      "epoch": 4.217770677153742,
      "grad_norm": 0.06303333659762887,
      "learning_rate": 1.419209450701193e-05,
      "loss": 0.8442,
      "step": 240240
    },
    {
      "epoch": 4.217946242033744,
      "grad_norm": 0.043595678834713734,
      "learning_rate": 1.4186737548735535e-05,
      "loss": 0.844,
      "step": 240250
    },
    {
      "epoch": 4.218121806913745,
      "grad_norm": 0.07179739657347596,
      "learning_rate": 1.4181381690399814e-05,
      "loss": 0.8404,
      "step": 240260
    },
    {
      "epoch": 4.218297371793747,
      "grad_norm": 0.036985896921491866,
      "learning_rate": 1.4176026932072645e-05,
      "loss": 0.8377,
      "step": 240270
    },
    {
      "epoch": 4.218472936673748,
      "grad_norm": 0.043087754280461216,
      "learning_rate": 1.4170673273821838e-05,
      "loss": 0.8362,
      "step": 240280
    },
    {
      "epoch": 4.21864850155375,
      "grad_norm": 0.05084117062269771,
      "learning_rate": 1.4165320715715235e-05,
      "loss": 0.8471,
      "step": 240290
    },
    {
      "epoch": 4.218824066433751,
      "grad_norm": 0.053270904105779886,
      "learning_rate": 1.4159969257820681e-05,
      "loss": 0.8417,
      "step": 240300
    },
    {
      "epoch": 4.218999631313752,
      "grad_norm": 0.050441131056717985,
      "learning_rate": 1.4154618900205943e-05,
      "loss": 0.8385,
      "step": 240310
    },
    {
      "epoch": 4.219175196193754,
      "grad_norm": 0.053592334077166064,
      "learning_rate": 1.4149269642938828e-05,
      "loss": 0.8453,
      "step": 240320
    },
    {
      "epoch": 4.219350761073755,
      "grad_norm": 0.07871029781810085,
      "learning_rate": 1.4143921486087105e-05,
      "loss": 0.8377,
      "step": 240330
    },
    {
      "epoch": 4.2195263259537565,
      "grad_norm": 0.05194243840494294,
      "learning_rate": 1.413857442971857e-05,
      "loss": 0.8313,
      "step": 240340
    },
    {
      "epoch": 4.219701890833758,
      "grad_norm": 0.05485290938898338,
      "learning_rate": 1.4133228473900931e-05,
      "loss": 0.8368,
      "step": 240350
    },
    {
      "epoch": 4.219877455713759,
      "grad_norm": 0.053053207872248064,
      "learning_rate": 1.412788361870192e-05,
      "loss": 0.8368,
      "step": 240360
    },
    {
      "epoch": 4.2200530205937605,
      "grad_norm": 0.05628992490649282,
      "learning_rate": 1.4122539864189294e-05,
      "loss": 0.8407,
      "step": 240370
    },
    {
      "epoch": 4.220228585473762,
      "grad_norm": 0.04524397306460336,
      "learning_rate": 1.4117197210430722e-05,
      "loss": 0.8381,
      "step": 240380
    },
    {
      "epoch": 4.2204041503537635,
      "grad_norm": 0.04696154424649314,
      "learning_rate": 1.4111855657493923e-05,
      "loss": 0.84,
      "step": 240390
    },
    {
      "epoch": 4.2205797152337645,
      "grad_norm": 0.05301748565743717,
      "learning_rate": 1.4106515205446566e-05,
      "loss": 0.84,
      "step": 240400
    },
    {
      "epoch": 4.220755280113766,
      "grad_norm": 0.06306794049682851,
      "learning_rate": 1.4101175854356307e-05,
      "loss": 0.8381,
      "step": 240410
    },
    {
      "epoch": 4.2209308449937675,
      "grad_norm": 0.046748362126222576,
      "learning_rate": 1.4095837604290826e-05,
      "loss": 0.8389,
      "step": 240420
    },
    {
      "epoch": 4.2211064098737685,
      "grad_norm": 0.0547410516824761,
      "learning_rate": 1.4090500455317687e-05,
      "loss": 0.8406,
      "step": 240430
    },
    {
      "epoch": 4.22128197475377,
      "grad_norm": 0.056219745924112656,
      "learning_rate": 1.4085164407504608e-05,
      "loss": 0.8365,
      "step": 240440
    },
    {
      "epoch": 4.2214575396337715,
      "grad_norm": 0.05777817915772739,
      "learning_rate": 1.4079829460919127e-05,
      "loss": 0.8461,
      "step": 240450
    },
    {
      "epoch": 4.221633104513773,
      "grad_norm": 0.05958441483850585,
      "learning_rate": 1.4074495615628879e-05,
      "loss": 0.8457,
      "step": 240460
    },
    {
      "epoch": 4.221808669393774,
      "grad_norm": 0.05518231555715791,
      "learning_rate": 1.4069162871701444e-05,
      "loss": 0.8323,
      "step": 240470
    },
    {
      "epoch": 4.2219842342737754,
      "grad_norm": 0.05577944563249251,
      "learning_rate": 1.406383122920435e-05,
      "loss": 0.8354,
      "step": 240480
    },
    {
      "epoch": 4.222159799153777,
      "grad_norm": 0.04828805365640686,
      "learning_rate": 1.4058500688205175e-05,
      "loss": 0.8362,
      "step": 240490
    },
    {
      "epoch": 4.222335364033778,
      "grad_norm": 0.047802613108813646,
      "learning_rate": 1.4053171248771471e-05,
      "loss": 0.8436,
      "step": 240500
    },
    {
      "epoch": 4.22251092891378,
      "grad_norm": 0.05320671800272749,
      "learning_rate": 1.4047842910970754e-05,
      "loss": 0.8395,
      "step": 240510
    },
    {
      "epoch": 4.222686493793781,
      "grad_norm": 0.04654654936526097,
      "learning_rate": 1.4042515674870527e-05,
      "loss": 0.8428,
      "step": 240520
    },
    {
      "epoch": 4.222862058673783,
      "grad_norm": 0.06206292906587966,
      "learning_rate": 1.403718954053829e-05,
      "loss": 0.8493,
      "step": 240530
    },
    {
      "epoch": 4.223037623553784,
      "grad_norm": 0.044050355725322975,
      "learning_rate": 1.4031864508041544e-05,
      "loss": 0.8404,
      "step": 240540
    },
    {
      "epoch": 4.223213188433785,
      "grad_norm": 0.05820797378277896,
      "learning_rate": 1.4026540577447722e-05,
      "loss": 0.839,
      "step": 240550
    },
    {
      "epoch": 4.223388753313787,
      "grad_norm": 0.0522229121803316,
      "learning_rate": 1.4021217748824295e-05,
      "loss": 0.8463,
      "step": 240560
    },
    {
      "epoch": 4.223564318193788,
      "grad_norm": 0.04695277183924568,
      "learning_rate": 1.4015896022238728e-05,
      "loss": 0.8433,
      "step": 240570
    },
    {
      "epoch": 4.22373988307379,
      "grad_norm": 0.0456566881710039,
      "learning_rate": 1.4010575397758413e-05,
      "loss": 0.8432,
      "step": 240580
    },
    {
      "epoch": 4.223915447953791,
      "grad_norm": 0.06104497596843356,
      "learning_rate": 1.4005255875450808e-05,
      "loss": 0.8331,
      "step": 240590
    },
    {
      "epoch": 4.224091012833792,
      "grad_norm": 0.04638142712089257,
      "learning_rate": 1.3999937455383224e-05,
      "loss": 0.8437,
      "step": 240600
    },
    {
      "epoch": 4.224266577713794,
      "grad_norm": 0.05002820889684861,
      "learning_rate": 1.399462013762316e-05,
      "loss": 0.8425,
      "step": 240610
    },
    {
      "epoch": 4.224442142593795,
      "grad_norm": 0.09255004555767461,
      "learning_rate": 1.3989303922237914e-05,
      "loss": 0.8393,
      "step": 240620
    },
    {
      "epoch": 4.224617707473797,
      "grad_norm": 0.08894637088355647,
      "learning_rate": 1.3983988809294854e-05,
      "loss": 0.8496,
      "step": 240630
    },
    {
      "epoch": 4.224793272353798,
      "grad_norm": 0.05029105431129263,
      "learning_rate": 1.3978674798861363e-05,
      "loss": 0.8346,
      "step": 240640
    },
    {
      "epoch": 4.2249688372338,
      "grad_norm": 0.05047268795774555,
      "learning_rate": 1.3973361891004707e-05,
      "loss": 0.8432,
      "step": 240650
    },
    {
      "epoch": 4.225144402113801,
      "grad_norm": 0.04528177602203799,
      "learning_rate": 1.3968050085792245e-05,
      "loss": 0.8433,
      "step": 240660
    },
    {
      "epoch": 4.225319966993802,
      "grad_norm": 0.053971114611789904,
      "learning_rate": 1.396273938329126e-05,
      "loss": 0.8308,
      "step": 240670
    },
    {
      "epoch": 4.225495531873804,
      "grad_norm": 0.044148444454506414,
      "learning_rate": 1.3957429783569075e-05,
      "loss": 0.8357,
      "step": 240680
    },
    {
      "epoch": 4.225671096753805,
      "grad_norm": 0.04691361856806417,
      "learning_rate": 1.3952121286692915e-05,
      "loss": 0.8334,
      "step": 240690
    },
    {
      "epoch": 4.225846661633807,
      "grad_norm": 0.049931625949768375,
      "learning_rate": 1.3946813892730053e-05,
      "loss": 0.8393,
      "step": 240700
    },
    {
      "epoch": 4.226022226513808,
      "grad_norm": 0.049982279863026356,
      "learning_rate": 1.394150760174778e-05,
      "loss": 0.838,
      "step": 240710
    },
    {
      "epoch": 4.22619779139381,
      "grad_norm": 0.066776172043513,
      "learning_rate": 1.3936202413813248e-05,
      "loss": 0.8322,
      "step": 240720
    },
    {
      "epoch": 4.226373356273811,
      "grad_norm": 0.04837134120172976,
      "learning_rate": 1.393089832899373e-05,
      "loss": 0.8412,
      "step": 240730
    },
    {
      "epoch": 4.226548921153812,
      "grad_norm": 0.05809968415623063,
      "learning_rate": 1.3925595347356444e-05,
      "loss": 0.8335,
      "step": 240740
    },
    {
      "epoch": 4.226724486033814,
      "grad_norm": 0.0687266888138864,
      "learning_rate": 1.3920293468968523e-05,
      "loss": 0.8409,
      "step": 240750
    },
    {
      "epoch": 4.226900050913815,
      "grad_norm": 0.04815078801560053,
      "learning_rate": 1.3914992693897191e-05,
      "loss": 0.8397,
      "step": 240760
    },
    {
      "epoch": 4.227075615793817,
      "grad_norm": 0.059629883773512596,
      "learning_rate": 1.3909693022209552e-05,
      "loss": 0.8484,
      "step": 240770
    },
    {
      "epoch": 4.227251180673818,
      "grad_norm": 0.055217984386785796,
      "learning_rate": 1.3904394453972836e-05,
      "loss": 0.8411,
      "step": 240780
    },
    {
      "epoch": 4.227426745553819,
      "grad_norm": 0.05128201216686496,
      "learning_rate": 1.389909698925411e-05,
      "loss": 0.8407,
      "step": 240790
    },
    {
      "epoch": 4.227602310433821,
      "grad_norm": 0.046602934455631455,
      "learning_rate": 1.3893800628120523e-05,
      "loss": 0.8365,
      "step": 240800
    },
    {
      "epoch": 4.227777875313822,
      "grad_norm": 0.06894930359639355,
      "learning_rate": 1.3888505370639196e-05,
      "loss": 0.837,
      "step": 240810
    },
    {
      "epoch": 4.227953440193824,
      "grad_norm": 0.0429988954065823,
      "learning_rate": 1.3883211216877177e-05,
      "loss": 0.8431,
      "step": 240820
    },
    {
      "epoch": 4.228129005073825,
      "grad_norm": 0.05739031139367823,
      "learning_rate": 1.3877918166901573e-05,
      "loss": 0.8384,
      "step": 240830
    },
    {
      "epoch": 4.228304569953827,
      "grad_norm": 0.05670047754792792,
      "learning_rate": 1.3872626220779439e-05,
      "loss": 0.8428,
      "step": 240840
    },
    {
      "epoch": 4.228480134833828,
      "grad_norm": 0.04657814276538687,
      "learning_rate": 1.386733537857785e-05,
      "loss": 0.8419,
      "step": 240850
    },
    {
      "epoch": 4.228655699713829,
      "grad_norm": 0.055545684425114164,
      "learning_rate": 1.3862045640363796e-05,
      "loss": 0.8355,
      "step": 240860
    },
    {
      "epoch": 4.228831264593831,
      "grad_norm": 0.05899241344619098,
      "learning_rate": 1.3856757006204323e-05,
      "loss": 0.8354,
      "step": 240870
    },
    {
      "epoch": 4.229006829473832,
      "grad_norm": 0.03928037243323132,
      "learning_rate": 1.3851469476166473e-05,
      "loss": 0.844,
      "step": 240880
    },
    {
      "epoch": 4.229182394353834,
      "grad_norm": 0.05520586236153388,
      "learning_rate": 1.3846183050317174e-05,
      "loss": 0.8381,
      "step": 240890
    },
    {
      "epoch": 4.229357959233835,
      "grad_norm": 0.06010785961894963,
      "learning_rate": 1.3840897728723443e-05,
      "loss": 0.8326,
      "step": 240900
    },
    {
      "epoch": 4.229533524113836,
      "grad_norm": 0.053308526762104354,
      "learning_rate": 1.3835613511452262e-05,
      "loss": 0.8384,
      "step": 240910
    },
    {
      "epoch": 4.229709088993838,
      "grad_norm": 0.04631867079015561,
      "learning_rate": 1.3830330398570543e-05,
      "loss": 0.8416,
      "step": 240920
    },
    {
      "epoch": 4.229884653873839,
      "grad_norm": 0.07720090397818026,
      "learning_rate": 1.3825048390145256e-05,
      "loss": 0.8348,
      "step": 240930
    },
    {
      "epoch": 4.230060218753841,
      "grad_norm": 0.05217843192680856,
      "learning_rate": 1.3819767486243282e-05,
      "loss": 0.8401,
      "step": 240940
    },
    {
      "epoch": 4.230235783633842,
      "grad_norm": 0.06353922630971842,
      "learning_rate": 1.3814487686931593e-05,
      "loss": 0.839,
      "step": 240950
    },
    {
      "epoch": 4.230411348513844,
      "grad_norm": 0.05644149516119645,
      "learning_rate": 1.3809208992277048e-05,
      "loss": 0.8333,
      "step": 240960
    },
    {
      "epoch": 4.230586913393845,
      "grad_norm": 0.06049510675073393,
      "learning_rate": 1.3803931402346523e-05,
      "loss": 0.8444,
      "step": 240970
    },
    {
      "epoch": 4.230762478273846,
      "grad_norm": 0.05616164899380814,
      "learning_rate": 1.379865491720693e-05,
      "loss": 0.8438,
      "step": 240980
    },
    {
      "epoch": 4.230938043153848,
      "grad_norm": 0.055454130400046424,
      "learning_rate": 1.3793379536925056e-05,
      "loss": 0.837,
      "step": 240990
    },
    {
      "epoch": 4.231113608033849,
      "grad_norm": 0.04562442800242614,
      "learning_rate": 1.3788105261567785e-05,
      "loss": 0.8431,
      "step": 241000
    },
    {
      "epoch": 4.231289172913851,
      "grad_norm": 0.04735038968235897,
      "learning_rate": 1.3782832091201934e-05,
      "loss": 0.8381,
      "step": 241010
    },
    {
      "epoch": 4.231464737793852,
      "grad_norm": 0.05580196258080762,
      "learning_rate": 1.3777560025894337e-05,
      "loss": 0.84,
      "step": 241020
    },
    {
      "epoch": 4.231640302673853,
      "grad_norm": 0.05621500040058437,
      "learning_rate": 1.3772289065711744e-05,
      "loss": 0.8319,
      "step": 241030
    },
    {
      "epoch": 4.231815867553855,
      "grad_norm": 0.042686868769861386,
      "learning_rate": 1.3767019210720975e-05,
      "loss": 0.8385,
      "step": 241040
    },
    {
      "epoch": 4.231991432433856,
      "grad_norm": 0.053606467223112374,
      "learning_rate": 1.3761750460988809e-05,
      "loss": 0.8372,
      "step": 241050
    },
    {
      "epoch": 4.232166997313858,
      "grad_norm": 0.05685169774326579,
      "learning_rate": 1.3756482816581962e-05,
      "loss": 0.8415,
      "step": 241060
    },
    {
      "epoch": 4.232342562193859,
      "grad_norm": 0.050604534088638825,
      "learning_rate": 1.3751216277567196e-05,
      "loss": 0.8318,
      "step": 241070
    },
    {
      "epoch": 4.2325181270738605,
      "grad_norm": 0.04520950670625645,
      "learning_rate": 1.3745950844011262e-05,
      "loss": 0.8354,
      "step": 241080
    },
    {
      "epoch": 4.232693691953862,
      "grad_norm": 0.065133991179496,
      "learning_rate": 1.374068651598083e-05,
      "loss": 0.8418,
      "step": 241090
    },
    {
      "epoch": 4.232869256833863,
      "grad_norm": 0.0641231266871208,
      "learning_rate": 1.3735423293542641e-05,
      "loss": 0.8358,
      "step": 241100
    },
    {
      "epoch": 4.2330448217138645,
      "grad_norm": 0.0450861981607851,
      "learning_rate": 1.3730161176763334e-05,
      "loss": 0.8411,
      "step": 241110
    },
    {
      "epoch": 4.233220386593866,
      "grad_norm": 0.056462589954999455,
      "learning_rate": 1.3724900165709644e-05,
      "loss": 0.8438,
      "step": 241120
    },
    {
      "epoch": 4.2333959514738675,
      "grad_norm": 0.0661697314151564,
      "learning_rate": 1.3719640260448172e-05,
      "loss": 0.8383,
      "step": 241130
    },
    {
      "epoch": 4.2335715163538685,
      "grad_norm": 0.05177688953728711,
      "learning_rate": 1.371438146104559e-05,
      "loss": 0.8342,
      "step": 241140
    },
    {
      "epoch": 4.23374708123387,
      "grad_norm": 0.05528178848510979,
      "learning_rate": 1.3709123767568543e-05,
      "loss": 0.8402,
      "step": 241150
    },
    {
      "epoch": 4.2339226461138715,
      "grad_norm": 0.04946842067642066,
      "learning_rate": 1.3703867180083601e-05,
      "loss": 0.8408,
      "step": 241160
    },
    {
      "epoch": 4.2340982109938725,
      "grad_norm": 0.052757891718693294,
      "learning_rate": 1.36986116986574e-05,
      "loss": 0.8411,
      "step": 241170
    },
    {
      "epoch": 4.234273775873874,
      "grad_norm": 0.07054115415634576,
      "learning_rate": 1.3693357323356512e-05,
      "loss": 0.8436,
      "step": 241180
    },
    {
      "epoch": 4.2344493407538755,
      "grad_norm": 0.056808358404189574,
      "learning_rate": 1.368810405424755e-05,
      "loss": 0.8377,
      "step": 241190
    },
    {
      "epoch": 4.234624905633877,
      "grad_norm": 0.08088815300851024,
      "learning_rate": 1.3682851891397014e-05,
      "loss": 0.8336,
      "step": 241200
    },
    {
      "epoch": 4.234800470513878,
      "grad_norm": 0.047856268531734684,
      "learning_rate": 1.3677600834871473e-05,
      "loss": 0.8417,
      "step": 241210
    },
    {
      "epoch": 4.2349760353938795,
      "grad_norm": 0.05776089537884784,
      "learning_rate": 1.3672350884737497e-05,
      "loss": 0.8379,
      "step": 241220
    },
    {
      "epoch": 4.235151600273881,
      "grad_norm": 0.08728035332027381,
      "learning_rate": 1.3667102041061542e-05,
      "loss": 0.8408,
      "step": 241230
    },
    {
      "epoch": 4.235327165153882,
      "grad_norm": 0.057246920152760686,
      "learning_rate": 1.3661854303910146e-05,
      "loss": 0.8432,
      "step": 241240
    },
    {
      "epoch": 4.235502730033884,
      "grad_norm": 0.05379176737311252,
      "learning_rate": 1.3656607673349819e-05,
      "loss": 0.8383,
      "step": 241250
    },
    {
      "epoch": 4.235678294913885,
      "grad_norm": 0.054842426030088216,
      "learning_rate": 1.3651362149446989e-05,
      "loss": 0.8314,
      "step": 241260
    },
    {
      "epoch": 4.235853859793886,
      "grad_norm": 0.0691559803782567,
      "learning_rate": 1.3646117732268155e-05,
      "loss": 0.8335,
      "step": 241270
    },
    {
      "epoch": 4.236029424673888,
      "grad_norm": 0.04717467703286169,
      "learning_rate": 1.3640874421879715e-05,
      "loss": 0.8396,
      "step": 241280
    },
    {
      "epoch": 4.236204989553889,
      "grad_norm": 0.04390337480728649,
      "learning_rate": 1.3635632218348185e-05,
      "loss": 0.8351,
      "step": 241290
    },
    {
      "epoch": 4.236380554433891,
      "grad_norm": 0.07006956524728941,
      "learning_rate": 1.3630391121739913e-05,
      "loss": 0.8329,
      "step": 241300
    },
    {
      "epoch": 4.236556119313892,
      "grad_norm": 0.05060373249719249,
      "learning_rate": 1.3625151132121325e-05,
      "loss": 0.8459,
      "step": 241310
    },
    {
      "epoch": 4.236731684193894,
      "grad_norm": 0.05153749019459427,
      "learning_rate": 1.3619912249558842e-05,
      "loss": 0.8323,
      "step": 241320
    },
    {
      "epoch": 4.236907249073895,
      "grad_norm": 0.05963429071528505,
      "learning_rate": 1.3614674474118799e-05,
      "loss": 0.844,
      "step": 241330
    },
    {
      "epoch": 4.237082813953896,
      "grad_norm": 0.056053586583545895,
      "learning_rate": 1.3609437805867566e-05,
      "loss": 0.8341,
      "step": 241340
    },
    {
      "epoch": 4.237258378833898,
      "grad_norm": 0.06639020968076868,
      "learning_rate": 1.3604202244871517e-05,
      "loss": 0.8251,
      "step": 241350
    },
    {
      "epoch": 4.237433943713899,
      "grad_norm": 0.04472469274322239,
      "learning_rate": 1.3598967791196996e-05,
      "loss": 0.8385,
      "step": 241360
    },
    {
      "epoch": 4.237609508593901,
      "grad_norm": 0.0700762583481631,
      "learning_rate": 1.3593734444910274e-05,
      "loss": 0.8286,
      "step": 241370
    },
    {
      "epoch": 4.237785073473902,
      "grad_norm": 0.060761300054985305,
      "learning_rate": 1.3588502206077696e-05,
      "loss": 0.8329,
      "step": 241380
    },
    {
      "epoch": 4.237960638353904,
      "grad_norm": 0.04791565911846247,
      "learning_rate": 1.3583271074765569e-05,
      "loss": 0.8335,
      "step": 241390
    },
    {
      "epoch": 4.238136203233905,
      "grad_norm": 0.07902045392038887,
      "learning_rate": 1.3578041051040125e-05,
      "loss": 0.8427,
      "step": 241400
    },
    {
      "epoch": 4.238311768113906,
      "grad_norm": 0.05537030605835103,
      "learning_rate": 1.3572812134967652e-05,
      "loss": 0.8297,
      "step": 241410
    },
    {
      "epoch": 4.238487332993908,
      "grad_norm": 0.0529074222800093,
      "learning_rate": 1.356758432661443e-05,
      "loss": 0.8369,
      "step": 241420
    },
    {
      "epoch": 4.238662897873909,
      "grad_norm": 0.05572481285708759,
      "learning_rate": 1.3562357626046645e-05,
      "loss": 0.8407,
      "step": 241430
    },
    {
      "epoch": 4.238838462753911,
      "grad_norm": 0.04327310847845957,
      "learning_rate": 1.3557132033330577e-05,
      "loss": 0.8402,
      "step": 241440
    },
    {
      "epoch": 4.239014027633912,
      "grad_norm": 0.05329100753581048,
      "learning_rate": 1.3551907548532347e-05,
      "loss": 0.8365,
      "step": 241450
    },
    {
      "epoch": 4.239189592513913,
      "grad_norm": 0.04757568173762572,
      "learning_rate": 1.3546684171718266e-05,
      "loss": 0.8364,
      "step": 241460
    },
    {
      "epoch": 4.239365157393915,
      "grad_norm": 0.05326145461173357,
      "learning_rate": 1.3541461902954432e-05,
      "loss": 0.8453,
      "step": 241470
    },
    {
      "epoch": 4.239540722273916,
      "grad_norm": 0.046160820900204624,
      "learning_rate": 1.3536240742307029e-05,
      "loss": 0.842,
      "step": 241480
    },
    {
      "epoch": 4.239716287153918,
      "grad_norm": 0.05215502272625961,
      "learning_rate": 1.3531020689842241e-05,
      "loss": 0.8455,
      "step": 241490
    },
    {
      "epoch": 4.239891852033919,
      "grad_norm": 0.055413404394274654,
      "learning_rate": 1.352580174562617e-05,
      "loss": 0.8374,
      "step": 241500
    },
    {
      "epoch": 4.24006741691392,
      "grad_norm": 0.06289976867026521,
      "learning_rate": 1.3520583909724955e-05,
      "loss": 0.8385,
      "step": 241510
    },
    {
      "epoch": 4.240242981793922,
      "grad_norm": 0.04710476060465758,
      "learning_rate": 1.3515367182204699e-05,
      "loss": 0.8369,
      "step": 241520
    },
    {
      "epoch": 4.240418546673923,
      "grad_norm": 0.07120314039183764,
      "learning_rate": 1.3510151563131543e-05,
      "loss": 0.8338,
      "step": 241530
    },
    {
      "epoch": 4.240594111553925,
      "grad_norm": 0.04699338177894847,
      "learning_rate": 1.3504937052571501e-05,
      "loss": 0.8355,
      "step": 241540
    },
    {
      "epoch": 4.240769676433926,
      "grad_norm": 0.055008831673032825,
      "learning_rate": 1.3499723650590679e-05,
      "loss": 0.8393,
      "step": 241550
    },
    {
      "epoch": 4.240945241313928,
      "grad_norm": 0.04340659028650438,
      "learning_rate": 1.3494511357255149e-05,
      "loss": 0.8351,
      "step": 241560
    },
    {
      "epoch": 4.241120806193929,
      "grad_norm": 0.04524228874865247,
      "learning_rate": 1.34893001726309e-05,
      "loss": 0.8463,
      "step": 241570
    },
    {
      "epoch": 4.24129637107393,
      "grad_norm": 0.06601588060408757,
      "learning_rate": 1.3484090096783993e-05,
      "loss": 0.8406,
      "step": 241580
    },
    {
      "epoch": 4.241471935953932,
      "grad_norm": 0.05117474832007829,
      "learning_rate": 1.3478881129780462e-05,
      "loss": 0.8434,
      "step": 241590
    },
    {
      "epoch": 4.241647500833933,
      "grad_norm": 0.04271839031463982,
      "learning_rate": 1.3473673271686266e-05,
      "loss": 0.8401,
      "step": 241600
    },
    {
      "epoch": 4.241823065713935,
      "grad_norm": 0.049168451551456785,
      "learning_rate": 1.3468466522567414e-05,
      "loss": 0.8362,
      "step": 241610
    },
    {
      "epoch": 4.241998630593936,
      "grad_norm": 0.04647121969705215,
      "learning_rate": 1.3463260882489834e-05,
      "loss": 0.8372,
      "step": 241620
    },
    {
      "epoch": 4.242174195473938,
      "grad_norm": 0.05493191534055402,
      "learning_rate": 1.345805635151955e-05,
      "loss": 0.8371,
      "step": 241630
    },
    {
      "epoch": 4.242349760353939,
      "grad_norm": 0.05069818333730023,
      "learning_rate": 1.3452852929722453e-05,
      "loss": 0.8373,
      "step": 241640
    },
    {
      "epoch": 4.24252532523394,
      "grad_norm": 0.054294577751108346,
      "learning_rate": 1.3447650617164482e-05,
      "loss": 0.8409,
      "step": 241650
    },
    {
      "epoch": 4.242700890113942,
      "grad_norm": 0.05028978979192163,
      "learning_rate": 1.3442449413911582e-05,
      "loss": 0.8341,
      "step": 241660
    },
    {
      "epoch": 4.242876454993943,
      "grad_norm": 0.06698424135856194,
      "learning_rate": 1.3437249320029601e-05,
      "loss": 0.8383,
      "step": 241670
    },
    {
      "epoch": 4.243052019873945,
      "grad_norm": 0.043619779832724206,
      "learning_rate": 1.3432050335584458e-05,
      "loss": 0.8472,
      "step": 241680
    },
    {
      "epoch": 4.243227584753946,
      "grad_norm": 0.05055235389471156,
      "learning_rate": 1.3426852460642016e-05,
      "loss": 0.8414,
      "step": 241690
    },
    {
      "epoch": 4.243403149633947,
      "grad_norm": 0.045345716873456676,
      "learning_rate": 1.3421655695268164e-05,
      "loss": 0.8394,
      "step": 241700
    },
    {
      "epoch": 4.243578714513949,
      "grad_norm": 0.049473932433064714,
      "learning_rate": 1.3416460039528688e-05,
      "loss": 0.8457,
      "step": 241710
    },
    {
      "epoch": 4.24375427939395,
      "grad_norm": 0.045044453442434976,
      "learning_rate": 1.3411265493489463e-05,
      "loss": 0.8376,
      "step": 241720
    },
    {
      "epoch": 4.243929844273952,
      "grad_norm": 0.05530692626906091,
      "learning_rate": 1.3406072057216303e-05,
      "loss": 0.841,
      "step": 241730
    },
    {
      "epoch": 4.244105409153953,
      "grad_norm": 0.07602651293921141,
      "learning_rate": 1.3400879730774968e-05,
      "loss": 0.8406,
      "step": 241740
    },
    {
      "epoch": 4.244280974033955,
      "grad_norm": 0.05802822238993708,
      "learning_rate": 1.3395688514231287e-05,
      "loss": 0.8358,
      "step": 241750
    },
    {
      "epoch": 4.244456538913956,
      "grad_norm": 0.05814539089112582,
      "learning_rate": 1.3390498407651036e-05,
      "loss": 0.8459,
      "step": 241760
    },
    {
      "epoch": 4.244632103793957,
      "grad_norm": 0.04798526272991988,
      "learning_rate": 1.338530941109995e-05,
      "loss": 0.8365,
      "step": 241770
    },
    {
      "epoch": 4.244807668673959,
      "grad_norm": 0.04238992162764868,
      "learning_rate": 1.3380121524643796e-05,
      "loss": 0.8422,
      "step": 241780
    },
    {
      "epoch": 4.24498323355396,
      "grad_norm": 0.04576127237619247,
      "learning_rate": 1.3374934748348258e-05,
      "loss": 0.8361,
      "step": 241790
    },
    {
      "epoch": 4.245158798433962,
      "grad_norm": 0.05026444726832254,
      "learning_rate": 1.3369749082279131e-05,
      "loss": 0.8398,
      "step": 241800
    },
    {
      "epoch": 4.245334363313963,
      "grad_norm": 0.04747487455059191,
      "learning_rate": 1.3364564526502052e-05,
      "loss": 0.8497,
      "step": 241810
    },
    {
      "epoch": 4.245509928193964,
      "grad_norm": 0.04854943535687436,
      "learning_rate": 1.335938108108274e-05,
      "loss": 0.8355,
      "step": 241820
    },
    {
      "epoch": 4.245685493073966,
      "grad_norm": 0.05761305444321205,
      "learning_rate": 1.3354198746086887e-05,
      "loss": 0.8376,
      "step": 241830
    },
    {
      "epoch": 4.245861057953967,
      "grad_norm": 0.0545071224398786,
      "learning_rate": 1.3349017521580112e-05,
      "loss": 0.8458,
      "step": 241840
    },
    {
      "epoch": 4.2460366228339685,
      "grad_norm": 0.04773444746289833,
      "learning_rate": 1.3343837407628076e-05,
      "loss": 0.8345,
      "step": 241850
    },
    {
      "epoch": 4.24621218771397,
      "grad_norm": 0.04122744429185112,
      "learning_rate": 1.333865840429643e-05,
      "loss": 0.8403,
      "step": 241860
    },
    {
      "epoch": 4.2463877525939715,
      "grad_norm": 0.04899755247983229,
      "learning_rate": 1.3333480511650793e-05,
      "loss": 0.8369,
      "step": 241870
    },
    {
      "epoch": 4.2465633174739725,
      "grad_norm": 0.04453043987676115,
      "learning_rate": 1.3328303729756744e-05,
      "loss": 0.8392,
      "step": 241880
    },
    {
      "epoch": 4.246738882353974,
      "grad_norm": 0.05597401235032556,
      "learning_rate": 1.3323128058679896e-05,
      "loss": 0.8353,
      "step": 241890
    },
    {
      "epoch": 4.2469144472339755,
      "grad_norm": 0.05977092863334185,
      "learning_rate": 1.331795349848583e-05,
      "loss": 0.8355,
      "step": 241900
    },
    {
      "epoch": 4.2470900121139765,
      "grad_norm": 0.06648765200072392,
      "learning_rate": 1.3312780049240076e-05,
      "loss": 0.828,
      "step": 241910
    },
    {
      "epoch": 4.247265576993978,
      "grad_norm": 0.05691198781292603,
      "learning_rate": 1.3307607711008208e-05,
      "loss": 0.8378,
      "step": 241920
    },
    {
      "epoch": 4.2474411418739795,
      "grad_norm": 0.044902025117654835,
      "learning_rate": 1.3302436483855777e-05,
      "loss": 0.8406,
      "step": 241930
    },
    {
      "epoch": 4.2476167067539805,
      "grad_norm": 0.06328920593771623,
      "learning_rate": 1.3297266367848258e-05,
      "loss": 0.836,
      "step": 241940
    },
    {
      "epoch": 4.247792271633982,
      "grad_norm": 0.06493181293253505,
      "learning_rate": 1.3292097363051204e-05,
      "loss": 0.8366,
      "step": 241950
    },
    {
      "epoch": 4.2479678365139835,
      "grad_norm": 0.05924680715192928,
      "learning_rate": 1.3286929469530055e-05,
      "loss": 0.8415,
      "step": 241960
    },
    {
      "epoch": 4.248143401393985,
      "grad_norm": 0.04917902992132883,
      "learning_rate": 1.3281762687350348e-05,
      "loss": 0.8391,
      "step": 241970
    },
    {
      "epoch": 4.248318966273986,
      "grad_norm": 0.04990182963370488,
      "learning_rate": 1.32765970165775e-05,
      "loss": 0.8396,
      "step": 241980
    },
    {
      "epoch": 4.248494531153988,
      "grad_norm": 0.06389512409173101,
      "learning_rate": 1.327143245727698e-05,
      "loss": 0.8386,
      "step": 241990
    },
    {
      "epoch": 4.248670096033989,
      "grad_norm": 0.05978384949998291,
      "learning_rate": 1.3266269009514254e-05,
      "loss": 0.8304,
      "step": 242000
    },
    {
      "epoch": 4.24884566091399,
      "grad_norm": 0.07304571696560898,
      "learning_rate": 1.3261106673354685e-05,
      "loss": 0.8352,
      "step": 242010
    },
    {
      "epoch": 4.249021225793992,
      "grad_norm": 0.05759227297358204,
      "learning_rate": 1.3255945448863702e-05,
      "loss": 0.846,
      "step": 242020
    },
    {
      "epoch": 4.249196790673993,
      "grad_norm": 0.05271365190010736,
      "learning_rate": 1.325078533610672e-05,
      "loss": 0.8336,
      "step": 242030
    },
    {
      "epoch": 4.249372355553995,
      "grad_norm": 0.03825835536314534,
      "learning_rate": 1.3245626335149116e-05,
      "loss": 0.8415,
      "step": 242040
    },
    {
      "epoch": 4.249547920433996,
      "grad_norm": 0.048913933679321134,
      "learning_rate": 1.3240468446056231e-05,
      "loss": 0.8327,
      "step": 242050
    },
    {
      "epoch": 4.249723485313998,
      "grad_norm": 0.05629054836584174,
      "learning_rate": 1.3235311668893425e-05,
      "loss": 0.8414,
      "step": 242060
    },
    {
      "epoch": 4.249899050193999,
      "grad_norm": 0.06005281795549607,
      "learning_rate": 1.3230156003726063e-05,
      "loss": 0.8431,
      "step": 242070
    },
    {
      "epoch": 4.250074615074,
      "grad_norm": 0.047164499174835096,
      "learning_rate": 1.3225001450619426e-05,
      "loss": 0.8468,
      "step": 242080
    },
    {
      "epoch": 4.250250179954002,
      "grad_norm": 0.058636207105191525,
      "learning_rate": 1.3219848009638838e-05,
      "loss": 0.8359,
      "step": 242090
    },
    {
      "epoch": 4.250425744834003,
      "grad_norm": 0.04671129157323506,
      "learning_rate": 1.3214695680849625e-05,
      "loss": 0.8367,
      "step": 242100
    },
    {
      "epoch": 4.250601309714005,
      "grad_norm": 0.06201569893386411,
      "learning_rate": 1.3209544464317019e-05,
      "loss": 0.8282,
      "step": 242110
    },
    {
      "epoch": 4.250776874594006,
      "grad_norm": 0.0462313319673513,
      "learning_rate": 1.320439436010633e-05,
      "loss": 0.8347,
      "step": 242120
    },
    {
      "epoch": 4.250952439474007,
      "grad_norm": 0.04549540082111642,
      "learning_rate": 1.3199245368282748e-05,
      "loss": 0.832,
      "step": 242130
    },
    {
      "epoch": 4.251128004354009,
      "grad_norm": 0.06206307774208999,
      "learning_rate": 1.3194097488911591e-05,
      "loss": 0.8347,
      "step": 242140
    },
    {
      "epoch": 4.25130356923401,
      "grad_norm": 0.04957044994972902,
      "learning_rate": 1.3188950722058039e-05,
      "loss": 0.8376,
      "step": 242150
    },
    {
      "epoch": 4.251479134114012,
      "grad_norm": 0.04402436530631118,
      "learning_rate": 1.3183805067787299e-05,
      "loss": 0.8371,
      "step": 242160
    },
    {
      "epoch": 4.251654698994013,
      "grad_norm": 0.049600885773311704,
      "learning_rate": 1.317866052616461e-05,
      "loss": 0.8422,
      "step": 242170
    },
    {
      "epoch": 4.251830263874014,
      "grad_norm": 0.04567852643834656,
      "learning_rate": 1.317351709725509e-05,
      "loss": 0.8346,
      "step": 242180
    },
    {
      "epoch": 4.252005828754016,
      "grad_norm": 0.04504928519518124,
      "learning_rate": 1.3168374781123941e-05,
      "loss": 0.8357,
      "step": 242190
    },
    {
      "epoch": 4.252181393634017,
      "grad_norm": 0.04602975418105225,
      "learning_rate": 1.3163233577836329e-05,
      "loss": 0.8401,
      "step": 242200
    },
    {
      "epoch": 4.252356958514019,
      "grad_norm": 0.056364236838173196,
      "learning_rate": 1.3158093487457388e-05,
      "loss": 0.8388,
      "step": 242210
    },
    {
      "epoch": 4.25253252339402,
      "grad_norm": 0.06238030855983602,
      "learning_rate": 1.315295451005223e-05,
      "loss": 0.8443,
      "step": 242220
    },
    {
      "epoch": 4.252708088274022,
      "grad_norm": 0.047276020684696225,
      "learning_rate": 1.3147816645685965e-05,
      "loss": 0.8385,
      "step": 242230
    },
    {
      "epoch": 4.252883653154023,
      "grad_norm": 0.05531504773045428,
      "learning_rate": 1.314267989442372e-05,
      "loss": 0.8385,
      "step": 242240
    },
    {
      "epoch": 4.253059218034024,
      "grad_norm": 0.048641261269313826,
      "learning_rate": 1.313754425633054e-05,
      "loss": 0.8393,
      "step": 242250
    },
    {
      "epoch": 4.253234782914026,
      "grad_norm": 0.04671435683780637,
      "learning_rate": 1.3132409731471512e-05,
      "loss": 0.8348,
      "step": 242260
    },
    {
      "epoch": 4.253410347794027,
      "grad_norm": 0.05932493716049996,
      "learning_rate": 1.3127276319911718e-05,
      "loss": 0.8323,
      "step": 242270
    },
    {
      "epoch": 4.253585912674029,
      "grad_norm": 0.04876420050760672,
      "learning_rate": 1.3122144021716149e-05,
      "loss": 0.846,
      "step": 242280
    },
    {
      "epoch": 4.25376147755403,
      "grad_norm": 0.05942522786461549,
      "learning_rate": 1.3117012836949868e-05,
      "loss": 0.8357,
      "step": 242290
    },
    {
      "epoch": 4.253937042434032,
      "grad_norm": 0.045163574419054296,
      "learning_rate": 1.3111882765677851e-05,
      "loss": 0.8337,
      "step": 242300
    },
    {
      "epoch": 4.254112607314033,
      "grad_norm": 0.05830146330711567,
      "learning_rate": 1.3106753807965161e-05,
      "loss": 0.8423,
      "step": 242310
    },
    {
      "epoch": 4.254288172194034,
      "grad_norm": 0.060918598390278744,
      "learning_rate": 1.3101625963876726e-05,
      "loss": 0.8372,
      "step": 242320
    },
    {
      "epoch": 4.254463737074036,
      "grad_norm": 0.045268733816303384,
      "learning_rate": 1.3096499233477525e-05,
      "loss": 0.8269,
      "step": 242330
    },
    {
      "epoch": 4.254639301954037,
      "grad_norm": 0.06593371227131385,
      "learning_rate": 1.3091373616832558e-05,
      "loss": 0.8429,
      "step": 242340
    },
    {
      "epoch": 4.254814866834039,
      "grad_norm": 0.05964018144875091,
      "learning_rate": 1.308624911400671e-05,
      "loss": 0.8278,
      "step": 242350
    },
    {
      "epoch": 4.25499043171404,
      "grad_norm": 0.0486012769594161,
      "learning_rate": 1.3081125725064939e-05,
      "loss": 0.8366,
      "step": 242360
    },
    {
      "epoch": 4.255165996594041,
      "grad_norm": 0.04942617462369631,
      "learning_rate": 1.3076003450072142e-05,
      "loss": 0.8373,
      "step": 242370
    },
    {
      "epoch": 4.255341561474043,
      "grad_norm": 0.0483091695798849,
      "learning_rate": 1.307088228909326e-05,
      "loss": 0.8381,
      "step": 242380
    },
    {
      "epoch": 4.255517126354044,
      "grad_norm": 0.05307991426135474,
      "learning_rate": 1.306576224219313e-05,
      "loss": 0.8402,
      "step": 242390
    },
    {
      "epoch": 4.255692691234046,
      "grad_norm": 0.05357365413304437,
      "learning_rate": 1.3060643309436642e-05,
      "loss": 0.8316,
      "step": 242400
    },
    {
      "epoch": 4.255868256114047,
      "grad_norm": 0.05124660657985264,
      "learning_rate": 1.3055525490888679e-05,
      "loss": 0.8425,
      "step": 242410
    },
    {
      "epoch": 4.256043820994049,
      "grad_norm": 0.05397031066064715,
      "learning_rate": 1.3050408786614046e-05,
      "loss": 0.8447,
      "step": 242420
    },
    {
      "epoch": 4.25621938587405,
      "grad_norm": 0.05336289546046423,
      "learning_rate": 1.3045293196677589e-05,
      "loss": 0.8349,
      "step": 242430
    },
    {
      "epoch": 4.256394950754051,
      "grad_norm": 0.053558733494403606,
      "learning_rate": 1.3040178721144145e-05,
      "loss": 0.8377,
      "step": 242440
    },
    {
      "epoch": 4.256570515634053,
      "grad_norm": 0.052784358740563704,
      "learning_rate": 1.3035065360078465e-05,
      "loss": 0.8429,
      "step": 242450
    },
    {
      "epoch": 4.256746080514054,
      "grad_norm": 0.05197527841371333,
      "learning_rate": 1.3029953113545395e-05,
      "loss": 0.837,
      "step": 242460
    },
    {
      "epoch": 4.256921645394056,
      "grad_norm": 0.05475921408677017,
      "learning_rate": 1.3024841981609632e-05,
      "loss": 0.8428,
      "step": 242470
    },
    {
      "epoch": 4.257097210274057,
      "grad_norm": 0.04625936133120867,
      "learning_rate": 1.3019731964336021e-05,
      "loss": 0.8381,
      "step": 242480
    },
    {
      "epoch": 4.257272775154058,
      "grad_norm": 0.04777383582683556,
      "learning_rate": 1.301462306178925e-05,
      "loss": 0.8405,
      "step": 242490
    },
    {
      "epoch": 4.25744834003406,
      "grad_norm": 0.05550459479838766,
      "learning_rate": 1.3009515274034062e-05,
      "loss": 0.8475,
      "step": 242500
    },
    {
      "epoch": 4.257623904914061,
      "grad_norm": 0.05764701501427514,
      "learning_rate": 1.3004408601135209e-05,
      "loss": 0.8445,
      "step": 242510
    },
    {
      "epoch": 4.257799469794063,
      "grad_norm": 0.05406282735311401,
      "learning_rate": 1.2999303043157332e-05,
      "loss": 0.8435,
      "step": 242520
    },
    {
      "epoch": 4.257975034674064,
      "grad_norm": 0.05034356781407459,
      "learning_rate": 1.2994198600165159e-05,
      "loss": 0.8395,
      "step": 242530
    },
    {
      "epoch": 4.258150599554066,
      "grad_norm": 0.05185165816759088,
      "learning_rate": 1.2989095272223355e-05,
      "loss": 0.8384,
      "step": 242540
    },
    {
      "epoch": 4.258326164434067,
      "grad_norm": 0.05187329814364592,
      "learning_rate": 1.2983993059396602e-05,
      "loss": 0.8363,
      "step": 242550
    },
    {
      "epoch": 4.258501729314068,
      "grad_norm": 0.05105091941912696,
      "learning_rate": 1.2978891961749498e-05,
      "loss": 0.8421,
      "step": 242560
    },
    {
      "epoch": 4.25867729419407,
      "grad_norm": 0.04412348172468183,
      "learning_rate": 1.2973791979346708e-05,
      "loss": 0.837,
      "step": 242570
    },
    {
      "epoch": 4.258852859074071,
      "grad_norm": 0.07642861086201984,
      "learning_rate": 1.2968693112252861e-05,
      "loss": 0.8389,
      "step": 242580
    },
    {
      "epoch": 4.2590284239540726,
      "grad_norm": 0.057771429975473465,
      "learning_rate": 1.2963595360532532e-05,
      "loss": 0.8445,
      "step": 242590
    },
    {
      "epoch": 4.259203988834074,
      "grad_norm": 0.061122447140581716,
      "learning_rate": 1.2958498724250312e-05,
      "loss": 0.8361,
      "step": 242600
    },
    {
      "epoch": 4.259379553714075,
      "grad_norm": 0.051610133860869455,
      "learning_rate": 1.2953403203470807e-05,
      "loss": 0.8446,
      "step": 242610
    },
    {
      "epoch": 4.2595551185940765,
      "grad_norm": 0.04637698986518553,
      "learning_rate": 1.294830879825854e-05,
      "loss": 0.8392,
      "step": 242620
    },
    {
      "epoch": 4.259730683474078,
      "grad_norm": 0.04508035272149031,
      "learning_rate": 1.2943215508678094e-05,
      "loss": 0.8341,
      "step": 242630
    },
    {
      "epoch": 4.2599062483540795,
      "grad_norm": 0.06369010037538243,
      "learning_rate": 1.293812333479395e-05,
      "loss": 0.8351,
      "step": 242640
    },
    {
      "epoch": 4.2600818132340805,
      "grad_norm": 0.05047318521902229,
      "learning_rate": 1.2933032276670699e-05,
      "loss": 0.8342,
      "step": 242650
    },
    {
      "epoch": 4.2602573781140824,
      "grad_norm": 0.057972212119803496,
      "learning_rate": 1.2927942334372778e-05,
      "loss": 0.8375,
      "step": 242660
    },
    {
      "epoch": 4.2604329429940835,
      "grad_norm": 0.04732270922310417,
      "learning_rate": 1.2922853507964716e-05,
      "loss": 0.839,
      "step": 242670
    },
    {
      "epoch": 4.2606085078740845,
      "grad_norm": 0.047418389798708604,
      "learning_rate": 1.2917765797510991e-05,
      "loss": 0.8278,
      "step": 242680
    },
    {
      "epoch": 4.260784072754086,
      "grad_norm": 0.04654095024428527,
      "learning_rate": 1.2912679203076042e-05,
      "loss": 0.8323,
      "step": 242690
    },
    {
      "epoch": 4.2609596376340875,
      "grad_norm": 0.05229678445684744,
      "learning_rate": 1.2907593724724321e-05,
      "loss": 0.8401,
      "step": 242700
    },
    {
      "epoch": 4.261135202514089,
      "grad_norm": 0.06505310922639307,
      "learning_rate": 1.2902509362520283e-05,
      "loss": 0.8392,
      "step": 242710
    },
    {
      "epoch": 4.26131076739409,
      "grad_norm": 0.0664282002661711,
      "learning_rate": 1.2897426116528344e-05,
      "loss": 0.836,
      "step": 242720
    },
    {
      "epoch": 4.261486332274092,
      "grad_norm": 0.04643481174810188,
      "learning_rate": 1.289234398681288e-05,
      "loss": 0.8479,
      "step": 242730
    },
    {
      "epoch": 4.261661897154093,
      "grad_norm": 0.05652757230192746,
      "learning_rate": 1.2887262973438314e-05,
      "loss": 0.845,
      "step": 242740
    },
    {
      "epoch": 4.261837462034094,
      "grad_norm": 0.05041272405530305,
      "learning_rate": 1.288218307646902e-05,
      "loss": 0.8367,
      "step": 242750
    },
    {
      "epoch": 4.262013026914096,
      "grad_norm": 0.05194966438662301,
      "learning_rate": 1.2877104295969348e-05,
      "loss": 0.8342,
      "step": 242760
    },
    {
      "epoch": 4.262188591794097,
      "grad_norm": 0.04653447164401356,
      "learning_rate": 1.2872026632003645e-05,
      "loss": 0.8425,
      "step": 242770
    },
    {
      "epoch": 4.262364156674099,
      "grad_norm": 0.05005771931292351,
      "learning_rate": 1.2866950084636273e-05,
      "loss": 0.8423,
      "step": 242780
    },
    {
      "epoch": 4.2625397215541,
      "grad_norm": 0.06754659458606181,
      "learning_rate": 1.2861874653931521e-05,
      "loss": 0.8391,
      "step": 242790
    },
    {
      "epoch": 4.262715286434101,
      "grad_norm": 0.04833763500340026,
      "learning_rate": 1.2856800339953721e-05,
      "loss": 0.8376,
      "step": 242800
    },
    {
      "epoch": 4.262890851314103,
      "grad_norm": 0.052926923892620534,
      "learning_rate": 1.2851727142767115e-05,
      "loss": 0.8392,
      "step": 242810
    },
    {
      "epoch": 4.263066416194104,
      "grad_norm": 0.05682119001807977,
      "learning_rate": 1.2846655062436066e-05,
      "loss": 0.8405,
      "step": 242820
    },
    {
      "epoch": 4.263241981074106,
      "grad_norm": 0.06545723221609097,
      "learning_rate": 1.2841584099024767e-05,
      "loss": 0.8388,
      "step": 242830
    },
    {
      "epoch": 4.263417545954107,
      "grad_norm": 0.06430139432264231,
      "learning_rate": 1.283651425259749e-05,
      "loss": 0.8469,
      "step": 242840
    },
    {
      "epoch": 4.263593110834108,
      "grad_norm": 0.0626513494107096,
      "learning_rate": 1.2831445523218494e-05,
      "loss": 0.8376,
      "step": 242850
    },
    {
      "epoch": 4.26376867571411,
      "grad_norm": 0.060045696033337745,
      "learning_rate": 1.2826377910951964e-05,
      "loss": 0.8337,
      "step": 242860
    },
    {
      "epoch": 4.263944240594111,
      "grad_norm": 0.05243271989026457,
      "learning_rate": 1.2821311415862125e-05,
      "loss": 0.8384,
      "step": 242870
    },
    {
      "epoch": 4.264119805474113,
      "grad_norm": 0.07299048044256033,
      "learning_rate": 1.2816246038013172e-05,
      "loss": 0.8425,
      "step": 242880
    },
    {
      "epoch": 4.264295370354114,
      "grad_norm": 0.054988534903137425,
      "learning_rate": 1.2811181777469299e-05,
      "loss": 0.8357,
      "step": 242890
    },
    {
      "epoch": 4.264470935234116,
      "grad_norm": 0.06476739278926044,
      "learning_rate": 1.2806118634294626e-05,
      "loss": 0.8349,
      "step": 242900
    },
    {
      "epoch": 4.264646500114117,
      "grad_norm": 0.05215845308270696,
      "learning_rate": 1.2801056608553341e-05,
      "loss": 0.8469,
      "step": 242910
    },
    {
      "epoch": 4.264822064994118,
      "grad_norm": 0.046057595320923767,
      "learning_rate": 1.2795995700309587e-05,
      "loss": 0.8466,
      "step": 242920
    },
    {
      "epoch": 4.26499762987412,
      "grad_norm": 0.04923370350567672,
      "learning_rate": 1.2790935909627461e-05,
      "loss": 0.8379,
      "step": 242930
    },
    {
      "epoch": 4.265173194754121,
      "grad_norm": 0.05832881166686537,
      "learning_rate": 1.2785877236571079e-05,
      "loss": 0.8364,
      "step": 242940
    },
    {
      "epoch": 4.265348759634123,
      "grad_norm": 0.047161344442871074,
      "learning_rate": 1.2780819681204554e-05,
      "loss": 0.8349,
      "step": 242950
    },
    {
      "epoch": 4.265524324514124,
      "grad_norm": 0.05273534510922233,
      "learning_rate": 1.2775763243591932e-05,
      "loss": 0.8399,
      "step": 242960
    },
    {
      "epoch": 4.265699889394126,
      "grad_norm": 0.08333769304411634,
      "learning_rate": 1.2770707923797326e-05,
      "loss": 0.8411,
      "step": 242970
    },
    {
      "epoch": 4.265875454274127,
      "grad_norm": 0.05628447166174564,
      "learning_rate": 1.2765653721884704e-05,
      "loss": 0.8405,
      "step": 242980
    },
    {
      "epoch": 4.266051019154128,
      "grad_norm": 0.08744901385820075,
      "learning_rate": 1.2760600637918213e-05,
      "loss": 0.8375,
      "step": 242990
    },
    {
      "epoch": 4.26622658403413,
      "grad_norm": 0.054992318144549045,
      "learning_rate": 1.2755548671961794e-05,
      "loss": 0.8359,
      "step": 243000
    },
    {
      "epoch": 4.266402148914131,
      "grad_norm": 0.04011887718502325,
      "learning_rate": 1.2750497824079496e-05,
      "loss": 0.8359,
      "step": 243010
    },
    {
      "epoch": 4.266577713794133,
      "grad_norm": 0.057137715415160495,
      "learning_rate": 1.2745448094335313e-05,
      "loss": 0.8458,
      "step": 243020
    },
    {
      "epoch": 4.266753278674134,
      "grad_norm": 0.046452995244857284,
      "learning_rate": 1.27403994827932e-05,
      "loss": 0.8351,
      "step": 243030
    },
    {
      "epoch": 4.266928843554135,
      "grad_norm": 0.05181529693900153,
      "learning_rate": 1.2735351989517144e-05,
      "loss": 0.8382,
      "step": 243040
    },
    {
      "epoch": 4.267104408434137,
      "grad_norm": 0.0492889891166414,
      "learning_rate": 1.2730305614571088e-05,
      "loss": 0.8402,
      "step": 243050
    },
    {
      "epoch": 4.267279973314138,
      "grad_norm": 0.0566121258382044,
      "learning_rate": 1.2725260358018994e-05,
      "loss": 0.8419,
      "step": 243060
    },
    {
      "epoch": 4.26745553819414,
      "grad_norm": 0.0566407722717939,
      "learning_rate": 1.2720216219924764e-05,
      "loss": 0.8398,
      "step": 243070
    },
    {
      "epoch": 4.267631103074141,
      "grad_norm": 0.044445557952637806,
      "learning_rate": 1.2715173200352299e-05,
      "loss": 0.8414,
      "step": 243080
    },
    {
      "epoch": 4.267806667954142,
      "grad_norm": 0.05582752629034151,
      "learning_rate": 1.2710131299365532e-05,
      "loss": 0.8375,
      "step": 243090
    },
    {
      "epoch": 4.267982232834144,
      "grad_norm": 0.04707676102808554,
      "learning_rate": 1.2705090517028309e-05,
      "loss": 0.8342,
      "step": 243100
    },
    {
      "epoch": 4.268157797714145,
      "grad_norm": 0.05801833245749176,
      "learning_rate": 1.2700050853404499e-05,
      "loss": 0.8346,
      "step": 243110
    },
    {
      "epoch": 4.268333362594147,
      "grad_norm": 0.04425973051190119,
      "learning_rate": 1.2695012308557996e-05,
      "loss": 0.8419,
      "step": 243120
    },
    {
      "epoch": 4.268508927474148,
      "grad_norm": 0.06062811152420002,
      "learning_rate": 1.268997488255259e-05,
      "loss": 0.8387,
      "step": 243130
    },
    {
      "epoch": 4.26868449235415,
      "grad_norm": 0.04953693403392331,
      "learning_rate": 1.2684938575452146e-05,
      "loss": 0.8406,
      "step": 243140
    },
    {
      "epoch": 4.268860057234151,
      "grad_norm": 0.04941619131262898,
      "learning_rate": 1.2679903387320417e-05,
      "loss": 0.8455,
      "step": 243150
    },
    {
      "epoch": 4.269035622114152,
      "grad_norm": 0.05662076719839458,
      "learning_rate": 1.2674869318221274e-05,
      "loss": 0.84,
      "step": 243160
    },
    {
      "epoch": 4.269211186994154,
      "grad_norm": 0.051611473473385995,
      "learning_rate": 1.2669836368218446e-05,
      "loss": 0.8368,
      "step": 243170
    },
    {
      "epoch": 4.269386751874155,
      "grad_norm": 0.049678377370191276,
      "learning_rate": 1.2664804537375727e-05,
      "loss": 0.8385,
      "step": 243180
    },
    {
      "epoch": 4.269562316754157,
      "grad_norm": 0.0497058090007426,
      "learning_rate": 1.2659773825756877e-05,
      "loss": 0.8391,
      "step": 243190
    },
    {
      "epoch": 4.269737881634158,
      "grad_norm": 0.05156175076359754,
      "learning_rate": 1.2654744233425609e-05,
      "loss": 0.8367,
      "step": 243200
    },
    {
      "epoch": 4.26991344651416,
      "grad_norm": 0.08325740255104307,
      "learning_rate": 1.264971576044565e-05,
      "loss": 0.8399,
      "step": 243210
    },
    {
      "epoch": 4.270089011394161,
      "grad_norm": 0.08941909449304974,
      "learning_rate": 1.2644688406880731e-05,
      "loss": 0.8348,
      "step": 243220
    },
    {
      "epoch": 4.270264576274162,
      "grad_norm": 0.04486953615558507,
      "learning_rate": 1.2639662172794563e-05,
      "loss": 0.8398,
      "step": 243230
    },
    {
      "epoch": 4.270440141154164,
      "grad_norm": 0.06703572065971837,
      "learning_rate": 1.263463705825079e-05,
      "loss": 0.8422,
      "step": 243240
    },
    {
      "epoch": 4.270615706034165,
      "grad_norm": 0.07180279406068314,
      "learning_rate": 1.2629613063313103e-05,
      "loss": 0.8405,
      "step": 243250
    },
    {
      "epoch": 4.270791270914167,
      "grad_norm": 0.04764637589725297,
      "learning_rate": 1.2624590188045177e-05,
      "loss": 0.8435,
      "step": 243260
    },
    {
      "epoch": 4.270966835794168,
      "grad_norm": 0.04983073441931259,
      "learning_rate": 1.2619568432510603e-05,
      "loss": 0.8436,
      "step": 243270
    },
    {
      "epoch": 4.271142400674169,
      "grad_norm": 0.052205973676732007,
      "learning_rate": 1.2614547796773044e-05,
      "loss": 0.8321,
      "step": 243280
    },
    {
      "epoch": 4.271317965554171,
      "grad_norm": 0.06340742876882399,
      "learning_rate": 1.2609528280896107e-05,
      "loss": 0.8399,
      "step": 243290
    },
    {
      "epoch": 4.271493530434172,
      "grad_norm": 0.048725781619105374,
      "learning_rate": 1.2604509884943386e-05,
      "loss": 0.8325,
      "step": 243300
    },
    {
      "epoch": 4.271669095314174,
      "grad_norm": 0.05042177939892984,
      "learning_rate": 1.2599492608978478e-05,
      "loss": 0.8362,
      "step": 243310
    },
    {
      "epoch": 4.271844660194175,
      "grad_norm": 0.04866308406717237,
      "learning_rate": 1.2594476453064897e-05,
      "loss": 0.8361,
      "step": 243320
    },
    {
      "epoch": 4.272020225074177,
      "grad_norm": 0.054357421457897126,
      "learning_rate": 1.2589461417266286e-05,
      "loss": 0.8331,
      "step": 243330
    },
    {
      "epoch": 4.272195789954178,
      "grad_norm": 0.06580800503690988,
      "learning_rate": 1.2584447501646118e-05,
      "loss": 0.8388,
      "step": 243340
    },
    {
      "epoch": 4.272371354834179,
      "grad_norm": 0.052975265761652734,
      "learning_rate": 1.2579434706267945e-05,
      "loss": 0.84,
      "step": 243350
    },
    {
      "epoch": 4.2725469197141805,
      "grad_norm": 0.056365970003588145,
      "learning_rate": 1.2574423031195304e-05,
      "loss": 0.842,
      "step": 243360
    },
    {
      "epoch": 4.272722484594182,
      "grad_norm": 0.06056484754443047,
      "learning_rate": 1.2569412476491652e-05,
      "loss": 0.8351,
      "step": 243370
    },
    {
      "epoch": 4.2728980494741835,
      "grad_norm": 0.05079591387074029,
      "learning_rate": 1.2564403042220494e-05,
      "loss": 0.8393,
      "step": 243380
    },
    {
      "epoch": 4.2730736143541845,
      "grad_norm": 0.07508006807313614,
      "learning_rate": 1.255939472844529e-05,
      "loss": 0.8348,
      "step": 243390
    },
    {
      "epoch": 4.2732491792341865,
      "grad_norm": 0.05153144173683568,
      "learning_rate": 1.255438753522952e-05,
      "loss": 0.8299,
      "step": 243400
    },
    {
      "epoch": 4.2734247441141875,
      "grad_norm": 0.07374664230872349,
      "learning_rate": 1.25493814626366e-05,
      "loss": 0.8422,
      "step": 243410
    },
    {
      "epoch": 4.2736003089941885,
      "grad_norm": 0.04567364423263618,
      "learning_rate": 1.2544376510729972e-05,
      "loss": 0.8325,
      "step": 243420
    },
    {
      "epoch": 4.2737758738741904,
      "grad_norm": 0.058243286435696956,
      "learning_rate": 1.2539372679573052e-05,
      "loss": 0.8418,
      "step": 243430
    },
    {
      "epoch": 4.2739514387541915,
      "grad_norm": 0.04998344615254459,
      "learning_rate": 1.2534369969229224e-05,
      "loss": 0.8434,
      "step": 243440
    },
    {
      "epoch": 4.274127003634193,
      "grad_norm": 0.0641621688812465,
      "learning_rate": 1.2529368379761874e-05,
      "loss": 0.842,
      "step": 243450
    },
    {
      "epoch": 4.274302568514194,
      "grad_norm": 0.04724240979184091,
      "learning_rate": 1.2524367911234409e-05,
      "loss": 0.8343,
      "step": 243460
    },
    {
      "epoch": 4.2744781333941955,
      "grad_norm": 0.046323521757375315,
      "learning_rate": 1.2519368563710143e-05,
      "loss": 0.8485,
      "step": 243470
    },
    {
      "epoch": 4.274653698274197,
      "grad_norm": 0.042076019244133525,
      "learning_rate": 1.2514370337252454e-05,
      "loss": 0.8378,
      "step": 243480
    },
    {
      "epoch": 4.274829263154198,
      "grad_norm": 0.060162432442483815,
      "learning_rate": 1.2509373231924606e-05,
      "loss": 0.8379,
      "step": 243490
    },
    {
      "epoch": 4.2750048280342,
      "grad_norm": 0.044117710586214755,
      "learning_rate": 1.250437724779e-05,
      "loss": 0.8377,
      "step": 243500
    },
    {
      "epoch": 4.275180392914201,
      "grad_norm": 0.045737364504768126,
      "learning_rate": 1.2499382384911865e-05,
      "loss": 0.8426,
      "step": 243510
    },
    {
      "epoch": 4.275355957794202,
      "grad_norm": 0.04617984633480924,
      "learning_rate": 1.249438864335353e-05,
      "loss": 0.8416,
      "step": 243520
    },
    {
      "epoch": 4.275531522674204,
      "grad_norm": 0.06145220714159467,
      "learning_rate": 1.2489396023178266e-05,
      "loss": 0.8352,
      "step": 243530
    },
    {
      "epoch": 4.275707087554205,
      "grad_norm": 0.04939218493859776,
      "learning_rate": 1.2484404524449304e-05,
      "loss": 0.8385,
      "step": 243540
    },
    {
      "epoch": 4.275882652434207,
      "grad_norm": 0.05160087658782202,
      "learning_rate": 1.2479414147229896e-05,
      "loss": 0.8474,
      "step": 243550
    },
    {
      "epoch": 4.276058217314208,
      "grad_norm": 0.04938219240322841,
      "learning_rate": 1.2474424891583273e-05,
      "loss": 0.8375,
      "step": 243560
    },
    {
      "epoch": 4.27623378219421,
      "grad_norm": 0.06766591135845429,
      "learning_rate": 1.2469436757572673e-05,
      "loss": 0.8455,
      "step": 243570
    },
    {
      "epoch": 4.276409347074211,
      "grad_norm": 0.06488732805771705,
      "learning_rate": 1.2464449745261259e-05,
      "loss": 0.832,
      "step": 243580
    },
    {
      "epoch": 4.276584911954212,
      "grad_norm": 0.0474081851827153,
      "learning_rate": 1.2459463854712231e-05,
      "loss": 0.8365,
      "step": 243590
    },
    {
      "epoch": 4.276760476834214,
      "grad_norm": 0.05213510029850534,
      "learning_rate": 1.2454479085988787e-05,
      "loss": 0.8288,
      "step": 243600
    },
    {
      "epoch": 4.276936041714215,
      "grad_norm": 0.050651299567227394,
      "learning_rate": 1.2449495439154041e-05,
      "loss": 0.8401,
      "step": 243610
    },
    {
      "epoch": 4.277111606594217,
      "grad_norm": 0.04915316261878375,
      "learning_rate": 1.2444512914271158e-05,
      "loss": 0.8472,
      "step": 243620
    },
    {
      "epoch": 4.277287171474218,
      "grad_norm": 0.050877953736880255,
      "learning_rate": 1.2439531511403286e-05,
      "loss": 0.8357,
      "step": 243630
    },
    {
      "epoch": 4.27746273635422,
      "grad_norm": 0.08391780105072462,
      "learning_rate": 1.243455123061351e-05,
      "loss": 0.8362,
      "step": 243640
    },
    {
      "epoch": 4.277638301234221,
      "grad_norm": 0.05135827770875194,
      "learning_rate": 1.2429572071964962e-05,
      "loss": 0.8383,
      "step": 243650
    },
    {
      "epoch": 4.277813866114222,
      "grad_norm": 0.05407052655003994,
      "learning_rate": 1.242459403552067e-05,
      "loss": 0.8455,
      "step": 243660
    },
    {
      "epoch": 4.277989430994224,
      "grad_norm": 0.048508402678191564,
      "learning_rate": 1.2419617121343777e-05,
      "loss": 0.8476,
      "step": 243670
    },
    {
      "epoch": 4.278164995874225,
      "grad_norm": 0.054196432852187906,
      "learning_rate": 1.2414641329497306e-05,
      "loss": 0.8365,
      "step": 243680
    },
    {
      "epoch": 4.278340560754227,
      "grad_norm": 0.05496555080497052,
      "learning_rate": 1.2409666660044294e-05,
      "loss": 0.8339,
      "step": 243690
    },
    {
      "epoch": 4.278516125634228,
      "grad_norm": 0.04270982395789623,
      "learning_rate": 1.2404693113047808e-05,
      "loss": 0.8401,
      "step": 243700
    },
    {
      "epoch": 4.278691690514229,
      "grad_norm": 0.03883944820608379,
      "learning_rate": 1.2399720688570824e-05,
      "loss": 0.8365,
      "step": 243710
    },
    {
      "epoch": 4.278867255394231,
      "grad_norm": 0.04854759632841802,
      "learning_rate": 1.2394749386676355e-05,
      "loss": 0.8411,
      "step": 243720
    },
    {
      "epoch": 4.279042820274232,
      "grad_norm": 0.05427797683607138,
      "learning_rate": 1.2389779207427397e-05,
      "loss": 0.8332,
      "step": 243730
    },
    {
      "epoch": 4.279218385154234,
      "grad_norm": 0.0472257753493364,
      "learning_rate": 1.2384810150886935e-05,
      "loss": 0.8361,
      "step": 243740
    },
    {
      "epoch": 4.279393950034235,
      "grad_norm": 0.050253822948359325,
      "learning_rate": 1.2379842217117888e-05,
      "loss": 0.8388,
      "step": 243750
    },
    {
      "epoch": 4.279569514914236,
      "grad_norm": 0.06059474493561488,
      "learning_rate": 1.2374875406183229e-05,
      "loss": 0.8394,
      "step": 243760
    },
    {
      "epoch": 4.279745079794238,
      "grad_norm": 0.05263274809284093,
      "learning_rate": 1.2369909718145898e-05,
      "loss": 0.8362,
      "step": 243770
    },
    {
      "epoch": 4.279920644674239,
      "grad_norm": 0.05525979734134049,
      "learning_rate": 1.2364945153068785e-05,
      "loss": 0.8417,
      "step": 243780
    },
    {
      "epoch": 4.280096209554241,
      "grad_norm": 0.056888199345068974,
      "learning_rate": 1.2359981711014796e-05,
      "loss": 0.8445,
      "step": 243790
    },
    {
      "epoch": 4.280271774434242,
      "grad_norm": 0.046249360809967645,
      "learning_rate": 1.2355019392046851e-05,
      "loss": 0.8387,
      "step": 243800
    },
    {
      "epoch": 4.280447339314244,
      "grad_norm": 0.05735332603008036,
      "learning_rate": 1.2350058196227778e-05,
      "loss": 0.8363,
      "step": 243810
    },
    {
      "epoch": 4.280622904194245,
      "grad_norm": 0.0645610975812796,
      "learning_rate": 1.2345098123620482e-05,
      "loss": 0.8318,
      "step": 243820
    },
    {
      "epoch": 4.280798469074246,
      "grad_norm": 0.040216104688058994,
      "learning_rate": 1.2340139174287741e-05,
      "loss": 0.8386,
      "step": 243830
    },
    {
      "epoch": 4.280974033954248,
      "grad_norm": 0.06042383455798125,
      "learning_rate": 1.2335181348292457e-05,
      "loss": 0.8401,
      "step": 243840
    },
    {
      "epoch": 4.281149598834249,
      "grad_norm": 0.04745865846991535,
      "learning_rate": 1.2330224645697405e-05,
      "loss": 0.8469,
      "step": 243850
    },
    {
      "epoch": 4.281325163714251,
      "grad_norm": 0.048553339624766295,
      "learning_rate": 1.232526906656541e-05,
      "loss": 0.8404,
      "step": 243860
    },
    {
      "epoch": 4.281500728594252,
      "grad_norm": 0.05233218673761235,
      "learning_rate": 1.232031461095926e-05,
      "loss": 0.8424,
      "step": 243870
    },
    {
      "epoch": 4.281676293474254,
      "grad_norm": 0.05740765680617244,
      "learning_rate": 1.2315361278941698e-05,
      "loss": 0.8346,
      "step": 243880
    },
    {
      "epoch": 4.281851858354255,
      "grad_norm": 0.05031566371650933,
      "learning_rate": 1.2310409070575515e-05,
      "loss": 0.8344,
      "step": 243890
    },
    {
      "epoch": 4.282027423234256,
      "grad_norm": 0.05562503482460219,
      "learning_rate": 1.2305457985923438e-05,
      "loss": 0.833,
      "step": 243900
    },
    {
      "epoch": 4.282202988114258,
      "grad_norm": 0.04205336889276104,
      "learning_rate": 1.2300508025048223e-05,
      "loss": 0.8484,
      "step": 243910
    },
    {
      "epoch": 4.282378552994259,
      "grad_norm": 0.06153711526015306,
      "learning_rate": 1.2295559188012562e-05,
      "loss": 0.8405,
      "step": 243920
    },
    {
      "epoch": 4.282554117874261,
      "grad_norm": 0.049053204596296016,
      "learning_rate": 1.2290611474879163e-05,
      "loss": 0.8426,
      "step": 243930
    },
    {
      "epoch": 4.282729682754262,
      "grad_norm": 0.044307363271849975,
      "learning_rate": 1.2285664885710731e-05,
      "loss": 0.8357,
      "step": 243940
    },
    {
      "epoch": 4.282905247634263,
      "grad_norm": 0.041510616772437066,
      "learning_rate": 1.2280719420569908e-05,
      "loss": 0.8403,
      "step": 243950
    },
    {
      "epoch": 4.283080812514265,
      "grad_norm": 0.058839105276302844,
      "learning_rate": 1.2275775079519377e-05,
      "loss": 0.8404,
      "step": 243960
    },
    {
      "epoch": 4.283256377394266,
      "grad_norm": 0.06109807562893681,
      "learning_rate": 1.2270831862621798e-05,
      "loss": 0.8315,
      "step": 243970
    },
    {
      "epoch": 4.283431942274268,
      "grad_norm": 0.061233678325721776,
      "learning_rate": 1.2265889769939765e-05,
      "loss": 0.8326,
      "step": 243980
    },
    {
      "epoch": 4.283607507154269,
      "grad_norm": 0.057797339204325934,
      "learning_rate": 1.226094880153594e-05,
      "loss": 0.8335,
      "step": 243990
    },
    {
      "epoch": 4.283783072034271,
      "grad_norm": 0.05876496182166458,
      "learning_rate": 1.225600895747285e-05,
      "loss": 0.8392,
      "step": 244000
    },
    {
      "epoch": 4.283958636914272,
      "grad_norm": 0.07178437421614126,
      "learning_rate": 1.225107023781318e-05,
      "loss": 0.8403,
      "step": 244010
    },
    {
      "epoch": 4.284134201794273,
      "grad_norm": 0.055981957244261145,
      "learning_rate": 1.2246132642619437e-05,
      "loss": 0.8405,
      "step": 244020
    },
    {
      "epoch": 4.284309766674275,
      "grad_norm": 0.049854149502988276,
      "learning_rate": 1.2241196171954205e-05,
      "loss": 0.8437,
      "step": 244030
    },
    {
      "epoch": 4.284485331554276,
      "grad_norm": 0.05919760452852691,
      "learning_rate": 1.2236260825880045e-05,
      "loss": 0.8396,
      "step": 244040
    },
    {
      "epoch": 4.284660896434278,
      "grad_norm": 0.049802380117870006,
      "learning_rate": 1.223132660445945e-05,
      "loss": 0.8427,
      "step": 244050
    },
    {
      "epoch": 4.284836461314279,
      "grad_norm": 0.05328056722470482,
      "learning_rate": 1.2226393507754969e-05,
      "loss": 0.8354,
      "step": 244060
    },
    {
      "epoch": 4.285012026194281,
      "grad_norm": 0.047962797685726914,
      "learning_rate": 1.2221461535829087e-05,
      "loss": 0.8339,
      "step": 244070
    },
    {
      "epoch": 4.285187591074282,
      "grad_norm": 0.05770526882285774,
      "learning_rate": 1.2216530688744319e-05,
      "loss": 0.833,
      "step": 244080
    },
    {
      "epoch": 4.285363155954283,
      "grad_norm": 0.06018105635588434,
      "learning_rate": 1.2211600966563113e-05,
      "loss": 0.8365,
      "step": 244090
    },
    {
      "epoch": 4.2855387208342846,
      "grad_norm": 0.04529728553208439,
      "learning_rate": 1.2206672369347935e-05,
      "loss": 0.8344,
      "step": 244100
    },
    {
      "epoch": 4.285714285714286,
      "grad_norm": 0.042049017873484605,
      "learning_rate": 1.2201744897161244e-05,
      "loss": 0.8509,
      "step": 244110
    },
    {
      "epoch": 4.2858898505942875,
      "grad_norm": 0.048939733438797664,
      "learning_rate": 1.2196818550065463e-05,
      "loss": 0.8404,
      "step": 244120
    },
    {
      "epoch": 4.2860654154742885,
      "grad_norm": 0.04281931902607773,
      "learning_rate": 1.2191893328123e-05,
      "loss": 0.8356,
      "step": 244130
    },
    {
      "epoch": 4.28624098035429,
      "grad_norm": 0.05186885760579349,
      "learning_rate": 1.2186969231396281e-05,
      "loss": 0.8378,
      "step": 244140
    },
    {
      "epoch": 4.2864165452342915,
      "grad_norm": 0.06022206209170929,
      "learning_rate": 1.2182046259947675e-05,
      "loss": 0.8308,
      "step": 244150
    },
    {
      "epoch": 4.2865921101142925,
      "grad_norm": 0.06510160525969833,
      "learning_rate": 1.2177124413839574e-05,
      "loss": 0.8472,
      "step": 244160
    },
    {
      "epoch": 4.2867676749942945,
      "grad_norm": 0.06106186038622757,
      "learning_rate": 1.2172203693134301e-05,
      "loss": 0.8378,
      "step": 244170
    },
    {
      "epoch": 4.2869432398742955,
      "grad_norm": 0.04491228226120275,
      "learning_rate": 1.2167284097894267e-05,
      "loss": 0.8419,
      "step": 244180
    },
    {
      "epoch": 4.2871188047542965,
      "grad_norm": 0.047031480994144344,
      "learning_rate": 1.2162365628181743e-05,
      "loss": 0.8391,
      "step": 244190
    },
    {
      "epoch": 4.287294369634298,
      "grad_norm": 0.07452425075593842,
      "learning_rate": 1.2157448284059073e-05,
      "loss": 0.8356,
      "step": 244200
    },
    {
      "epoch": 4.2874699345142995,
      "grad_norm": 0.047679999457830434,
      "learning_rate": 1.215253206558859e-05,
      "loss": 0.8399,
      "step": 244210
    },
    {
      "epoch": 4.287645499394301,
      "grad_norm": 0.042947432672802856,
      "learning_rate": 1.2147616972832523e-05,
      "loss": 0.8356,
      "step": 244220
    },
    {
      "epoch": 4.287821064274302,
      "grad_norm": 0.047582918424767126,
      "learning_rate": 1.2142703005853184e-05,
      "loss": 0.8351,
      "step": 244230
    },
    {
      "epoch": 4.287996629154304,
      "grad_norm": 0.05247348038271087,
      "learning_rate": 1.2137790164712814e-05,
      "loss": 0.8357,
      "step": 244240
    },
    {
      "epoch": 4.288172194034305,
      "grad_norm": 0.04751592854969255,
      "learning_rate": 1.2132878449473704e-05,
      "loss": 0.8339,
      "step": 244250
    },
    {
      "epoch": 4.288347758914306,
      "grad_norm": 0.04389219572185419,
      "learning_rate": 1.2127967860198036e-05,
      "loss": 0.8383,
      "step": 244260
    },
    {
      "epoch": 4.288523323794308,
      "grad_norm": 0.05636238057894353,
      "learning_rate": 1.2123058396948048e-05,
      "loss": 0.8441,
      "step": 244270
    },
    {
      "epoch": 4.288698888674309,
      "grad_norm": 0.06381700937898929,
      "learning_rate": 1.2118150059785957e-05,
      "loss": 0.8402,
      "step": 244280
    },
    {
      "epoch": 4.288874453554311,
      "grad_norm": 0.08597102125226226,
      "learning_rate": 1.2113242848773923e-05,
      "loss": 0.8361,
      "step": 244290
    },
    {
      "epoch": 4.289050018434312,
      "grad_norm": 0.05059543778871967,
      "learning_rate": 1.210833676397414e-05,
      "loss": 0.8379,
      "step": 244300
    },
    {
      "epoch": 4.289225583314314,
      "grad_norm": 0.06561227426414713,
      "learning_rate": 1.2103431805448778e-05,
      "loss": 0.8375,
      "step": 244310
    },
    {
      "epoch": 4.289401148194315,
      "grad_norm": 0.04445657732310863,
      "learning_rate": 1.2098527973259957e-05,
      "loss": 0.8408,
      "step": 244320
    },
    {
      "epoch": 4.289576713074316,
      "grad_norm": 0.058941198482326776,
      "learning_rate": 1.2093625267469846e-05,
      "loss": 0.8367,
      "step": 244330
    },
    {
      "epoch": 4.289752277954318,
      "grad_norm": 0.07700066389434689,
      "learning_rate": 1.20887236881405e-05,
      "loss": 0.8451,
      "step": 244340
    },
    {
      "epoch": 4.289927842834319,
      "grad_norm": 0.04336136946572611,
      "learning_rate": 1.2083823235334105e-05,
      "loss": 0.8421,
      "step": 244350
    },
    {
      "epoch": 4.290103407714321,
      "grad_norm": 0.04683432933984994,
      "learning_rate": 1.2078923909112697e-05,
      "loss": 0.8389,
      "step": 244360
    },
    {
      "epoch": 4.290278972594322,
      "grad_norm": 0.04870168248029764,
      "learning_rate": 1.2074025709538356e-05,
      "loss": 0.8343,
      "step": 244370
    },
    {
      "epoch": 4.290454537474323,
      "grad_norm": 0.05420940425705515,
      "learning_rate": 1.2069128636673168e-05,
      "loss": 0.8414,
      "step": 244380
    },
    {
      "epoch": 4.290630102354325,
      "grad_norm": 0.04779517429713064,
      "learning_rate": 1.206423269057915e-05,
      "loss": 0.8427,
      "step": 244390
    },
    {
      "epoch": 4.290805667234326,
      "grad_norm": 0.050010359086229354,
      "learning_rate": 1.2059337871318355e-05,
      "loss": 0.8403,
      "step": 244400
    },
    {
      "epoch": 4.290981232114328,
      "grad_norm": 0.047812748664901476,
      "learning_rate": 1.2054444178952784e-05,
      "loss": 0.8436,
      "step": 244410
    },
    {
      "epoch": 4.291156796994329,
      "grad_norm": 0.04486403045550754,
      "learning_rate": 1.2049551613544471e-05,
      "loss": 0.8382,
      "step": 244420
    },
    {
      "epoch": 4.29133236187433,
      "grad_norm": 0.044516154820135406,
      "learning_rate": 1.2044660175155367e-05,
      "loss": 0.8359,
      "step": 244430
    },
    {
      "epoch": 4.291507926754332,
      "grad_norm": 0.04461820890790119,
      "learning_rate": 1.2039769863847481e-05,
      "loss": 0.8456,
      "step": 244440
    },
    {
      "epoch": 4.291683491634333,
      "grad_norm": 0.05209887859408612,
      "learning_rate": 1.2034880679682758e-05,
      "loss": 0.8317,
      "step": 244450
    },
    {
      "epoch": 4.291859056514335,
      "grad_norm": 0.08843145846260526,
      "learning_rate": 1.2029992622723137e-05,
      "loss": 0.8337,
      "step": 244460
    },
    {
      "epoch": 4.292034621394336,
      "grad_norm": 0.059214530072757496,
      "learning_rate": 1.2025105693030563e-05,
      "loss": 0.839,
      "step": 244470
    },
    {
      "epoch": 4.292210186274338,
      "grad_norm": 0.04979769824776254,
      "learning_rate": 1.202021989066697e-05,
      "loss": 0.8413,
      "step": 244480
    },
    {
      "epoch": 4.292385751154339,
      "grad_norm": 0.042705671856737605,
      "learning_rate": 1.2015335215694229e-05,
      "loss": 0.8427,
      "step": 244490
    },
    {
      "epoch": 4.29256131603434,
      "grad_norm": 0.053710309926316424,
      "learning_rate": 1.2010451668174255e-05,
      "loss": 0.8303,
      "step": 244500
    },
    {
      "epoch": 4.292736880914342,
      "grad_norm": 0.048456267184408264,
      "learning_rate": 1.2005569248168879e-05,
      "loss": 0.8405,
      "step": 244510
    },
    {
      "epoch": 4.292912445794343,
      "grad_norm": 0.05784210572377025,
      "learning_rate": 1.2000687955740042e-05,
      "loss": 0.8368,
      "step": 244520
    },
    {
      "epoch": 4.293088010674345,
      "grad_norm": 0.05336795441289424,
      "learning_rate": 1.1995807790949516e-05,
      "loss": 0.8376,
      "step": 244530
    },
    {
      "epoch": 4.293263575554346,
      "grad_norm": 0.05726096718251255,
      "learning_rate": 1.1990928753859163e-05,
      "loss": 0.8418,
      "step": 244540
    },
    {
      "epoch": 4.293439140434348,
      "grad_norm": 0.06431257811598978,
      "learning_rate": 1.1986050844530824e-05,
      "loss": 0.8424,
      "step": 244550
    },
    {
      "epoch": 4.293614705314349,
      "grad_norm": 0.07161094238151122,
      "learning_rate": 1.198117406302626e-05,
      "loss": 0.8397,
      "step": 244560
    },
    {
      "epoch": 4.29379027019435,
      "grad_norm": 0.06980730206888551,
      "learning_rate": 1.1976298409407284e-05,
      "loss": 0.8425,
      "step": 244570
    },
    {
      "epoch": 4.293965835074352,
      "grad_norm": 0.05341142964196267,
      "learning_rate": 1.1971423883735664e-05,
      "loss": 0.8345,
      "step": 244580
    },
    {
      "epoch": 4.294141399954353,
      "grad_norm": 0.05739001195714825,
      "learning_rate": 1.1966550486073185e-05,
      "loss": 0.8368,
      "step": 244590
    },
    {
      "epoch": 4.294316964834355,
      "grad_norm": 0.05161584704525461,
      "learning_rate": 1.1961678216481551e-05,
      "loss": 0.8393,
      "step": 244600
    },
    {
      "epoch": 4.294492529714356,
      "grad_norm": 0.04138233396747195,
      "learning_rate": 1.1956807075022533e-05,
      "loss": 0.8394,
      "step": 244610
    },
    {
      "epoch": 4.294668094594357,
      "grad_norm": 0.054821080133501496,
      "learning_rate": 1.1951937061757834e-05,
      "loss": 0.8335,
      "step": 244620
    },
    {
      "epoch": 4.294843659474359,
      "grad_norm": 0.05021169979814272,
      "learning_rate": 1.194706817674915e-05,
      "loss": 0.8436,
      "step": 244630
    },
    {
      "epoch": 4.29501922435436,
      "grad_norm": 0.04433458797325256,
      "learning_rate": 1.1942200420058183e-05,
      "loss": 0.8454,
      "step": 244640
    },
    {
      "epoch": 4.295194789234362,
      "grad_norm": 0.047970805217978175,
      "learning_rate": 1.193733379174662e-05,
      "loss": 0.8412,
      "step": 244650
    },
    {
      "epoch": 4.295370354114363,
      "grad_norm": 0.0624123001053848,
      "learning_rate": 1.1932468291876082e-05,
      "loss": 0.8378,
      "step": 244660
    },
    {
      "epoch": 4.295545918994365,
      "grad_norm": 0.13142262994387713,
      "learning_rate": 1.1927603920508269e-05,
      "loss": 0.8317,
      "step": 244670
    },
    {
      "epoch": 4.295721483874366,
      "grad_norm": 0.060142868420599,
      "learning_rate": 1.1922740677704747e-05,
      "loss": 0.8485,
      "step": 244680
    },
    {
      "epoch": 4.295897048754367,
      "grad_norm": 0.06098260564595067,
      "learning_rate": 1.1917878563527204e-05,
      "loss": 0.8337,
      "step": 244690
    },
    {
      "epoch": 4.296072613634369,
      "grad_norm": 0.0439301758572402,
      "learning_rate": 1.1913017578037192e-05,
      "loss": 0.8416,
      "step": 244700
    },
    {
      "epoch": 4.29624817851437,
      "grad_norm": 0.05150725746110822,
      "learning_rate": 1.1908157721296323e-05,
      "loss": 0.8365,
      "step": 244710
    },
    {
      "epoch": 4.296423743394372,
      "grad_norm": 0.05683223346221046,
      "learning_rate": 1.1903298993366186e-05,
      "loss": 0.8343,
      "step": 244720
    },
    {
      "epoch": 4.296599308274373,
      "grad_norm": 0.07657865853627895,
      "learning_rate": 1.1898441394308313e-05,
      "loss": 0.8272,
      "step": 244730
    },
    {
      "epoch": 4.296774873154374,
      "grad_norm": 0.05867583535768245,
      "learning_rate": 1.1893584924184263e-05,
      "loss": 0.842,
      "step": 244740
    },
    {
      "epoch": 4.296950438034376,
      "grad_norm": 0.05619390364604521,
      "learning_rate": 1.188872958305557e-05,
      "loss": 0.8356,
      "step": 244750
    },
    {
      "epoch": 4.297126002914377,
      "grad_norm": 0.053905479259493023,
      "learning_rate": 1.1883875370983766e-05,
      "loss": 0.8382,
      "step": 244760
    },
    {
      "epoch": 4.297301567794379,
      "grad_norm": 0.04140738071273563,
      "learning_rate": 1.1879022288030323e-05,
      "loss": 0.8421,
      "step": 244770
    },
    {
      "epoch": 4.29747713267438,
      "grad_norm": 0.054064051725794794,
      "learning_rate": 1.187417033425674e-05,
      "loss": 0.8337,
      "step": 244780
    },
    {
      "epoch": 4.297652697554382,
      "grad_norm": 0.06361684316238882,
      "learning_rate": 1.186931950972453e-05,
      "loss": 0.8399,
      "step": 244790
    },
    {
      "epoch": 4.297828262434383,
      "grad_norm": 0.054400890811550494,
      "learning_rate": 1.1864469814495101e-05,
      "loss": 0.8337,
      "step": 244800
    },
    {
      "epoch": 4.298003827314384,
      "grad_norm": 0.05954778250865471,
      "learning_rate": 1.1859621248629923e-05,
      "loss": 0.8342,
      "step": 244810
    },
    {
      "epoch": 4.298179392194386,
      "grad_norm": 0.05032897457704196,
      "learning_rate": 1.1854773812190442e-05,
      "loss": 0.8373,
      "step": 244820
    },
    {
      "epoch": 4.298354957074387,
      "grad_norm": 0.05139830820780887,
      "learning_rate": 1.1849927505238043e-05,
      "loss": 0.8426,
      "step": 244830
    },
    {
      "epoch": 4.298530521954389,
      "grad_norm": 0.049824423174182834,
      "learning_rate": 1.184508232783417e-05,
      "loss": 0.8479,
      "step": 244840
    },
    {
      "epoch": 4.29870608683439,
      "grad_norm": 0.03996273977131417,
      "learning_rate": 1.1840238280040152e-05,
      "loss": 0.8504,
      "step": 244850
    },
    {
      "epoch": 4.298881651714391,
      "grad_norm": 0.08172327167555041,
      "learning_rate": 1.1835395361917442e-05,
      "loss": 0.8376,
      "step": 244860
    },
    {
      "epoch": 4.2990572165943925,
      "grad_norm": 0.06594967981822622,
      "learning_rate": 1.1830553573527336e-05,
      "loss": 0.8478,
      "step": 244870
    },
    {
      "epoch": 4.299232781474394,
      "grad_norm": 0.05557580843759508,
      "learning_rate": 1.1825712914931214e-05,
      "loss": 0.8378,
      "step": 244880
    },
    {
      "epoch": 4.2994083463543955,
      "grad_norm": 0.054667818135308716,
      "learning_rate": 1.1820873386190424e-05,
      "loss": 0.8393,
      "step": 244890
    },
    {
      "epoch": 4.2995839112343965,
      "grad_norm": 0.057441578236777,
      "learning_rate": 1.1816034987366228e-05,
      "loss": 0.8326,
      "step": 244900
    },
    {
      "epoch": 4.2997594761143985,
      "grad_norm": 0.05380601850319926,
      "learning_rate": 1.1811197718519953e-05,
      "loss": 0.8404,
      "step": 244910
    },
    {
      "epoch": 4.2999350409943995,
      "grad_norm": 0.061545844638529454,
      "learning_rate": 1.1806361579712918e-05,
      "loss": 0.8383,
      "step": 244920
    },
    {
      "epoch": 4.3001106058744005,
      "grad_norm": 0.04584977686480838,
      "learning_rate": 1.1801526571006375e-05,
      "loss": 0.8343,
      "step": 244930
    },
    {
      "epoch": 4.3002861707544024,
      "grad_norm": 0.04900715950556405,
      "learning_rate": 1.1796692692461576e-05,
      "loss": 0.8275,
      "step": 244940
    },
    {
      "epoch": 4.3004617356344035,
      "grad_norm": 0.04004644341141746,
      "learning_rate": 1.179185994413978e-05,
      "loss": 0.8445,
      "step": 244950
    },
    {
      "epoch": 4.300637300514405,
      "grad_norm": 0.06069050558636753,
      "learning_rate": 1.1787028326102224e-05,
      "loss": 0.839,
      "step": 244960
    },
    {
      "epoch": 4.300812865394406,
      "grad_norm": 0.07190191801390593,
      "learning_rate": 1.17821978384101e-05,
      "loss": 0.84,
      "step": 244970
    },
    {
      "epoch": 4.300988430274408,
      "grad_norm": 0.04212830601291455,
      "learning_rate": 1.1777368481124629e-05,
      "loss": 0.8465,
      "step": 244980
    },
    {
      "epoch": 4.301163995154409,
      "grad_norm": 0.07155635379873759,
      "learning_rate": 1.1772540254307022e-05,
      "loss": 0.8379,
      "step": 244990
    },
    {
      "epoch": 4.30133956003441,
      "grad_norm": 0.043870891887915583,
      "learning_rate": 1.1767713158018406e-05,
      "loss": 0.8259,
      "step": 245000
    },
    {
      "epoch": 4.301515124914412,
      "grad_norm": 0.05058865515048825,
      "learning_rate": 1.1762887192319988e-05,
      "loss": 0.8429,
      "step": 245010
    },
    {
      "epoch": 4.301690689794413,
      "grad_norm": 0.05385236640474205,
      "learning_rate": 1.1758062357272854e-05,
      "loss": 0.8299,
      "step": 245020
    },
    {
      "epoch": 4.301866254674415,
      "grad_norm": 0.06154945049895782,
      "learning_rate": 1.1753238652938212e-05,
      "loss": 0.8399,
      "step": 245030
    },
    {
      "epoch": 4.302041819554416,
      "grad_norm": 0.048673559789500394,
      "learning_rate": 1.1748416079377119e-05,
      "loss": 0.8493,
      "step": 245040
    },
    {
      "epoch": 4.302217384434417,
      "grad_norm": 0.04804337326424673,
      "learning_rate": 1.1743594636650701e-05,
      "loss": 0.8344,
      "step": 245050
    },
    {
      "epoch": 4.302392949314419,
      "grad_norm": 0.06578787998663972,
      "learning_rate": 1.1738774324820061e-05,
      "loss": 0.8368,
      "step": 245060
    },
    {
      "epoch": 4.30256851419442,
      "grad_norm": 0.04538952960993355,
      "learning_rate": 1.1733955143946243e-05,
      "loss": 0.8338,
      "step": 245070
    },
    {
      "epoch": 4.302744079074422,
      "grad_norm": 0.04723023669017851,
      "learning_rate": 1.1729137094090309e-05,
      "loss": 0.8327,
      "step": 245080
    },
    {
      "epoch": 4.302919643954423,
      "grad_norm": 0.05020374107220653,
      "learning_rate": 1.1724320175313325e-05,
      "loss": 0.8375,
      "step": 245090
    },
    {
      "epoch": 4.303095208834424,
      "grad_norm": 0.052560736988129915,
      "learning_rate": 1.1719504387676336e-05,
      "loss": 0.8437,
      "step": 245100
    },
    {
      "epoch": 4.303270773714426,
      "grad_norm": 0.04936403507502994,
      "learning_rate": 1.1714689731240314e-05,
      "loss": 0.8385,
      "step": 245110
    },
    {
      "epoch": 4.303446338594427,
      "grad_norm": 0.05041446251194637,
      "learning_rate": 1.1709876206066293e-05,
      "loss": 0.8382,
      "step": 245120
    },
    {
      "epoch": 4.303621903474429,
      "grad_norm": 0.05553838402557932,
      "learning_rate": 1.1705063812215258e-05,
      "loss": 0.8364,
      "step": 245130
    },
    {
      "epoch": 4.30379746835443,
      "grad_norm": 0.04649232961564507,
      "learning_rate": 1.170025254974817e-05,
      "loss": 0.8496,
      "step": 245140
    },
    {
      "epoch": 4.303973033234432,
      "grad_norm": 0.049938812925953986,
      "learning_rate": 1.1695442418725985e-05,
      "loss": 0.8398,
      "step": 245150
    },
    {
      "epoch": 4.304148598114433,
      "grad_norm": 0.046526860298201834,
      "learning_rate": 1.1690633419209685e-05,
      "loss": 0.8302,
      "step": 245160
    },
    {
      "epoch": 4.304324162994434,
      "grad_norm": 0.04704129099535935,
      "learning_rate": 1.168582555126016e-05,
      "loss": 0.8436,
      "step": 245170
    },
    {
      "epoch": 4.304499727874436,
      "grad_norm": 0.05093876760221799,
      "learning_rate": 1.1681018814938352e-05,
      "loss": 0.839,
      "step": 245180
    },
    {
      "epoch": 4.304675292754437,
      "grad_norm": 0.0711765274780771,
      "learning_rate": 1.1676213210305125e-05,
      "loss": 0.8413,
      "step": 245190
    },
    {
      "epoch": 4.304850857634439,
      "grad_norm": 0.043965545015763156,
      "learning_rate": 1.1671408737421436e-05,
      "loss": 0.8306,
      "step": 245200
    },
    {
      "epoch": 4.30502642251444,
      "grad_norm": 0.050452756241773304,
      "learning_rate": 1.1666605396348085e-05,
      "loss": 0.8441,
      "step": 245210
    },
    {
      "epoch": 4.305201987394442,
      "grad_norm": 0.07116607183796357,
      "learning_rate": 1.1661803187145976e-05,
      "loss": 0.8433,
      "step": 245220
    },
    {
      "epoch": 4.305377552274443,
      "grad_norm": 0.05053123597153073,
      "learning_rate": 1.1657002109875953e-05,
      "loss": 0.8428,
      "step": 245230
    },
    {
      "epoch": 4.305553117154444,
      "grad_norm": 0.05060172310495273,
      "learning_rate": 1.1652202164598824e-05,
      "loss": 0.8362,
      "step": 245240
    },
    {
      "epoch": 4.305728682034446,
      "grad_norm": 0.05582020383401215,
      "learning_rate": 1.1647403351375416e-05,
      "loss": 0.8354,
      "step": 245250
    },
    {
      "epoch": 4.305904246914447,
      "grad_norm": 0.049964122501737485,
      "learning_rate": 1.1642605670266527e-05,
      "loss": 0.8337,
      "step": 245260
    },
    {
      "epoch": 4.306079811794449,
      "grad_norm": 0.05332377230219553,
      "learning_rate": 1.163780912133297e-05,
      "loss": 0.8387,
      "step": 245270
    },
    {
      "epoch": 4.30625537667445,
      "grad_norm": 0.11333517702182735,
      "learning_rate": 1.163301370463549e-05,
      "loss": 0.8337,
      "step": 245280
    },
    {
      "epoch": 4.306430941554451,
      "grad_norm": 0.05635449189900751,
      "learning_rate": 1.1628219420234843e-05,
      "loss": 0.8362,
      "step": 245290
    },
    {
      "epoch": 4.306606506434453,
      "grad_norm": 0.05146107119301413,
      "learning_rate": 1.1623426268191796e-05,
      "loss": 0.8462,
      "step": 245300
    },
    {
      "epoch": 4.306782071314454,
      "grad_norm": 0.06593791671290132,
      "learning_rate": 1.1618634248567057e-05,
      "loss": 0.8403,
      "step": 245310
    },
    {
      "epoch": 4.306957636194456,
      "grad_norm": 0.05097764203100669,
      "learning_rate": 1.1613843361421354e-05,
      "loss": 0.8308,
      "step": 245320
    },
    {
      "epoch": 4.307133201074457,
      "grad_norm": 0.04000073956058394,
      "learning_rate": 1.1609053606815406e-05,
      "loss": 0.8378,
      "step": 245330
    },
    {
      "epoch": 4.307308765954459,
      "grad_norm": 0.04542219497467321,
      "learning_rate": 1.1604264984809849e-05,
      "loss": 0.841,
      "step": 245340
    },
    {
      "epoch": 4.30748433083446,
      "grad_norm": 0.047297998394782076,
      "learning_rate": 1.1599477495465416e-05,
      "loss": 0.8413,
      "step": 245350
    },
    {
      "epoch": 4.307659895714461,
      "grad_norm": 0.043103117572115386,
      "learning_rate": 1.1594691138842703e-05,
      "loss": 0.8393,
      "step": 245360
    },
    {
      "epoch": 4.307835460594463,
      "grad_norm": 0.053223814130104,
      "learning_rate": 1.1589905915002424e-05,
      "loss": 0.8423,
      "step": 245370
    },
    {
      "epoch": 4.308011025474464,
      "grad_norm": 0.046327531610138084,
      "learning_rate": 1.1585121824005141e-05,
      "loss": 0.8449,
      "step": 245380
    },
    {
      "epoch": 4.308186590354466,
      "grad_norm": 0.043936658117397706,
      "learning_rate": 1.1580338865911516e-05,
      "loss": 0.8401,
      "step": 245390
    },
    {
      "epoch": 4.308362155234467,
      "grad_norm": 0.053265140330648116,
      "learning_rate": 1.1575557040782142e-05,
      "loss": 0.8399,
      "step": 245400
    },
    {
      "epoch": 4.308537720114468,
      "grad_norm": 0.04534178750566319,
      "learning_rate": 1.1570776348677593e-05,
      "loss": 0.8386,
      "step": 245410
    },
    {
      "epoch": 4.30871328499447,
      "grad_norm": 0.047724339597722605,
      "learning_rate": 1.1565996789658428e-05,
      "loss": 0.8437,
      "step": 245420
    },
    {
      "epoch": 4.308888849874471,
      "grad_norm": 0.0556544273913987,
      "learning_rate": 1.1561218363785243e-05,
      "loss": 0.8411,
      "step": 245430
    },
    {
      "epoch": 4.309064414754473,
      "grad_norm": 0.06107082268735241,
      "learning_rate": 1.1556441071118567e-05,
      "loss": 0.8303,
      "step": 245440
    },
    {
      "epoch": 4.309239979634474,
      "grad_norm": 0.05201781316174373,
      "learning_rate": 1.1551664911718907e-05,
      "loss": 0.8392,
      "step": 245450
    },
    {
      "epoch": 4.309415544514476,
      "grad_norm": 0.05802487428795362,
      "learning_rate": 1.15468898856468e-05,
      "loss": 0.8366,
      "step": 245460
    },
    {
      "epoch": 4.309591109394477,
      "grad_norm": 0.05443425954780417,
      "learning_rate": 1.1542115992962769e-05,
      "loss": 0.8357,
      "step": 245470
    },
    {
      "epoch": 4.309766674274478,
      "grad_norm": 0.048033804130247,
      "learning_rate": 1.1537343233727235e-05,
      "loss": 0.8346,
      "step": 245480
    },
    {
      "epoch": 4.30994223915448,
      "grad_norm": 0.0578134773629109,
      "learning_rate": 1.1532571608000725e-05,
      "loss": 0.8296,
      "step": 245490
    },
    {
      "epoch": 4.310117804034481,
      "grad_norm": 0.05347676335260744,
      "learning_rate": 1.1527801115843685e-05,
      "loss": 0.8386,
      "step": 245500
    },
    {
      "epoch": 4.310293368914483,
      "grad_norm": 0.05176060807033332,
      "learning_rate": 1.1523031757316543e-05,
      "loss": 0.8336,
      "step": 245510
    },
    {
      "epoch": 4.310468933794484,
      "grad_norm": 0.05100448724346923,
      "learning_rate": 1.1518263532479747e-05,
      "loss": 0.836,
      "step": 245520
    },
    {
      "epoch": 4.310644498674485,
      "grad_norm": 0.0546506104409918,
      "learning_rate": 1.1513496441393665e-05,
      "loss": 0.8407,
      "step": 245530
    },
    {
      "epoch": 4.310820063554487,
      "grad_norm": 0.04632124770484475,
      "learning_rate": 1.1508730484118773e-05,
      "loss": 0.8444,
      "step": 245540
    },
    {
      "epoch": 4.310995628434488,
      "grad_norm": 0.04812895952692338,
      "learning_rate": 1.1503965660715402e-05,
      "loss": 0.8388,
      "step": 245550
    },
    {
      "epoch": 4.31117119331449,
      "grad_norm": 0.05538867998470552,
      "learning_rate": 1.149920197124394e-05,
      "loss": 0.8369,
      "step": 245560
    },
    {
      "epoch": 4.311346758194491,
      "grad_norm": 0.06117540188164554,
      "learning_rate": 1.1494439415764743e-05,
      "loss": 0.8413,
      "step": 245570
    },
    {
      "epoch": 4.311522323074493,
      "grad_norm": 0.04374815998504964,
      "learning_rate": 1.148967799433815e-05,
      "loss": 0.832,
      "step": 245580
    },
    {
      "epoch": 4.311697887954494,
      "grad_norm": 0.04716904052808959,
      "learning_rate": 1.1484917707024487e-05,
      "loss": 0.8407,
      "step": 245590
    },
    {
      "epoch": 4.311873452834495,
      "grad_norm": 0.04631750071267555,
      "learning_rate": 1.1480158553884076e-05,
      "loss": 0.8392,
      "step": 245600
    },
    {
      "epoch": 4.312049017714497,
      "grad_norm": 0.07910281987274533,
      "learning_rate": 1.1475400534977226e-05,
      "loss": 0.8409,
      "step": 245610
    },
    {
      "epoch": 4.312224582594498,
      "grad_norm": 0.053538726836829954,
      "learning_rate": 1.147064365036419e-05,
      "loss": 0.8422,
      "step": 245620
    },
    {
      "epoch": 4.3124001474744995,
      "grad_norm": 0.03910904133532887,
      "learning_rate": 1.1465887900105254e-05,
      "loss": 0.8489,
      "step": 245630
    },
    {
      "epoch": 4.3125757123545005,
      "grad_norm": 0.04639232600582583,
      "learning_rate": 1.1461133284260706e-05,
      "loss": 0.8386,
      "step": 245640
    },
    {
      "epoch": 4.3127512772345025,
      "grad_norm": 0.04757575092060136,
      "learning_rate": 1.1456379802890732e-05,
      "loss": 0.838,
      "step": 245650
    },
    {
      "epoch": 4.3129268421145035,
      "grad_norm": 0.06307347335514871,
      "learning_rate": 1.1451627456055586e-05,
      "loss": 0.8407,
      "step": 245660
    },
    {
      "epoch": 4.3131024069945045,
      "grad_norm": 0.05123451234815867,
      "learning_rate": 1.1446876243815505e-05,
      "loss": 0.8363,
      "step": 245670
    },
    {
      "epoch": 4.3132779718745065,
      "grad_norm": 0.047281794257342906,
      "learning_rate": 1.1442126166230646e-05,
      "loss": 0.8321,
      "step": 245680
    },
    {
      "epoch": 4.3134535367545075,
      "grad_norm": 0.04662344887414292,
      "learning_rate": 1.1437377223361227e-05,
      "loss": 0.8388,
      "step": 245690
    },
    {
      "epoch": 4.313629101634509,
      "grad_norm": 0.06220551376782061,
      "learning_rate": 1.1432629415267376e-05,
      "loss": 0.8395,
      "step": 245700
    },
    {
      "epoch": 4.3138046665145104,
      "grad_norm": 0.0473934379307805,
      "learning_rate": 1.1427882742009305e-05,
      "loss": 0.8399,
      "step": 245710
    },
    {
      "epoch": 4.3139802313945115,
      "grad_norm": 0.05445678040843425,
      "learning_rate": 1.1423137203647102e-05,
      "loss": 0.8422,
      "step": 245720
    },
    {
      "epoch": 4.314155796274513,
      "grad_norm": 0.05158136010326667,
      "learning_rate": 1.1418392800240935e-05,
      "loss": 0.8308,
      "step": 245730
    },
    {
      "epoch": 4.314331361154514,
      "grad_norm": 0.08153596834331182,
      "learning_rate": 1.1413649531850898e-05,
      "loss": 0.8432,
      "step": 245740
    },
    {
      "epoch": 4.314506926034516,
      "grad_norm": 0.05755127471958991,
      "learning_rate": 1.1408907398537087e-05,
      "loss": 0.8378,
      "step": 245750
    },
    {
      "epoch": 4.314682490914517,
      "grad_norm": 0.05719940036906305,
      "learning_rate": 1.1404166400359586e-05,
      "loss": 0.8424,
      "step": 245760
    },
    {
      "epoch": 4.314858055794518,
      "grad_norm": 0.04789681212576469,
      "learning_rate": 1.1399426537378467e-05,
      "loss": 0.8363,
      "step": 245770
    },
    {
      "epoch": 4.31503362067452,
      "grad_norm": 0.04922862543755677,
      "learning_rate": 1.1394687809653804e-05,
      "loss": 0.8346,
      "step": 245780
    },
    {
      "epoch": 4.315209185554521,
      "grad_norm": 0.08075315158298212,
      "learning_rate": 1.1389950217245595e-05,
      "loss": 0.8464,
      "step": 245790
    },
    {
      "epoch": 4.315384750434523,
      "grad_norm": 0.06149364827369889,
      "learning_rate": 1.1385213760213893e-05,
      "loss": 0.8393,
      "step": 245800
    },
    {
      "epoch": 4.315560315314524,
      "grad_norm": 0.07473195562622234,
      "learning_rate": 1.1380478438618728e-05,
      "loss": 0.8278,
      "step": 245810
    },
    {
      "epoch": 4.315735880194526,
      "grad_norm": 0.06559473606312984,
      "learning_rate": 1.1375744252520055e-05,
      "loss": 0.8362,
      "step": 245820
    },
    {
      "epoch": 4.315911445074527,
      "grad_norm": 0.05335887612945319,
      "learning_rate": 1.1371011201977888e-05,
      "loss": 0.8401,
      "step": 245830
    },
    {
      "epoch": 4.316087009954528,
      "grad_norm": 0.0490505162542878,
      "learning_rate": 1.1366279287052196e-05,
      "loss": 0.834,
      "step": 245840
    },
    {
      "epoch": 4.31626257483453,
      "grad_norm": 0.04320766174986809,
      "learning_rate": 1.136154850780291e-05,
      "loss": 0.8445,
      "step": 245850
    },
    {
      "epoch": 4.316438139714531,
      "grad_norm": 0.1456196749008855,
      "learning_rate": 1.1356818864290004e-05,
      "loss": 0.8419,
      "step": 245860
    },
    {
      "epoch": 4.316613704594533,
      "grad_norm": 0.05239227394381047,
      "learning_rate": 1.135209035657333e-05,
      "loss": 0.8377,
      "step": 245870
    },
    {
      "epoch": 4.316789269474534,
      "grad_norm": 0.04647257805340461,
      "learning_rate": 1.1347362984712907e-05,
      "loss": 0.8395,
      "step": 245880
    },
    {
      "epoch": 4.316964834354536,
      "grad_norm": 0.06044983363456469,
      "learning_rate": 1.1342636748768554e-05,
      "loss": 0.8381,
      "step": 245890
    },
    {
      "epoch": 4.317140399234537,
      "grad_norm": 0.05945416853515307,
      "learning_rate": 1.1337911648800174e-05,
      "loss": 0.8322,
      "step": 245900
    },
    {
      "epoch": 4.317315964114538,
      "grad_norm": 0.051621776609643365,
      "learning_rate": 1.1333187684867649e-05,
      "loss": 0.8431,
      "step": 245910
    },
    {
      "epoch": 4.31749152899454,
      "grad_norm": 0.05282641626439428,
      "learning_rate": 1.1328464857030816e-05,
      "loss": 0.8378,
      "step": 245920
    },
    {
      "epoch": 4.317667093874541,
      "grad_norm": 0.0456798777507656,
      "learning_rate": 1.1323743165349531e-05,
      "loss": 0.8324,
      "step": 245930
    },
    {
      "epoch": 4.317842658754543,
      "grad_norm": 0.05254010839732587,
      "learning_rate": 1.1319022609883564e-05,
      "loss": 0.8477,
      "step": 245940
    },
    {
      "epoch": 4.318018223634544,
      "grad_norm": 0.04465380912773114,
      "learning_rate": 1.131430319069281e-05,
      "loss": 0.8439,
      "step": 245950
    },
    {
      "epoch": 4.318193788514545,
      "grad_norm": 0.051579788709676125,
      "learning_rate": 1.1309584907837001e-05,
      "loss": 0.8355,
      "step": 245960
    },
    {
      "epoch": 4.318369353394547,
      "grad_norm": 0.05601585830296319,
      "learning_rate": 1.130486776137594e-05,
      "loss": 0.8326,
      "step": 245970
    },
    {
      "epoch": 4.318544918274548,
      "grad_norm": 0.044449206320163526,
      "learning_rate": 1.1300151751369426e-05,
      "loss": 0.8373,
      "step": 245980
    },
    {
      "epoch": 4.31872048315455,
      "grad_norm": 0.04664832950030553,
      "learning_rate": 1.1295436877877144e-05,
      "loss": 0.8456,
      "step": 245990
    },
    {
      "epoch": 4.318896048034551,
      "grad_norm": 0.057237619979073326,
      "learning_rate": 1.1290723140958881e-05,
      "loss": 0.8392,
      "step": 246000
    },
    {
      "epoch": 4.319071612914552,
      "grad_norm": 0.05466263174698412,
      "learning_rate": 1.128601054067436e-05,
      "loss": 0.8429,
      "step": 246010
    },
    {
      "epoch": 4.319247177794554,
      "grad_norm": 0.05104641756973391,
      "learning_rate": 1.1281299077083265e-05,
      "loss": 0.8367,
      "step": 246020
    },
    {
      "epoch": 4.319422742674555,
      "grad_norm": 0.04545992288778748,
      "learning_rate": 1.127658875024532e-05,
      "loss": 0.8422,
      "step": 246030
    },
    {
      "epoch": 4.319598307554557,
      "grad_norm": 0.052182944405072866,
      "learning_rate": 1.1271879560220155e-05,
      "loss": 0.8392,
      "step": 246040
    },
    {
      "epoch": 4.319773872434558,
      "grad_norm": 0.052450246182211155,
      "learning_rate": 1.1267171507067509e-05,
      "loss": 0.8456,
      "step": 246050
    },
    {
      "epoch": 4.31994943731456,
      "grad_norm": 0.050292689467526595,
      "learning_rate": 1.126246459084698e-05,
      "loss": 0.8351,
      "step": 246060
    },
    {
      "epoch": 4.320125002194561,
      "grad_norm": 0.05640001242792479,
      "learning_rate": 1.1257758811618225e-05,
      "loss": 0.8396,
      "step": 246070
    },
    {
      "epoch": 4.320300567074562,
      "grad_norm": 0.06082219362775022,
      "learning_rate": 1.1253054169440877e-05,
      "loss": 0.8373,
      "step": 246080
    },
    {
      "epoch": 4.320476131954564,
      "grad_norm": 0.08200972198943803,
      "learning_rate": 1.1248350664374521e-05,
      "loss": 0.8395,
      "step": 246090
    },
    {
      "epoch": 4.320651696834565,
      "grad_norm": 0.053436975112367543,
      "learning_rate": 1.1243648296478775e-05,
      "loss": 0.8392,
      "step": 246100
    },
    {
      "epoch": 4.320827261714567,
      "grad_norm": 0.040252176739707476,
      "learning_rate": 1.1238947065813176e-05,
      "loss": 0.8336,
      "step": 246110
    },
    {
      "epoch": 4.321002826594568,
      "grad_norm": 0.05574740923586585,
      "learning_rate": 1.1234246972437353e-05,
      "loss": 0.8324,
      "step": 246120
    },
    {
      "epoch": 4.32117839147457,
      "grad_norm": 0.06164237906156451,
      "learning_rate": 1.1229548016410812e-05,
      "loss": 0.8422,
      "step": 246130
    },
    {
      "epoch": 4.321353956354571,
      "grad_norm": 0.04994522710747393,
      "learning_rate": 1.1224850197793094e-05,
      "loss": 0.8339,
      "step": 246140
    },
    {
      "epoch": 4.321529521234572,
      "grad_norm": 0.0541850197846064,
      "learning_rate": 1.1220153516643756e-05,
      "loss": 0.844,
      "step": 246150
    },
    {
      "epoch": 4.321705086114574,
      "grad_norm": 0.05875738818884981,
      "learning_rate": 1.1215457973022253e-05,
      "loss": 0.8354,
      "step": 246160
    },
    {
      "epoch": 4.321880650994575,
      "grad_norm": 0.04629261844241304,
      "learning_rate": 1.12107635669881e-05,
      "loss": 0.8486,
      "step": 246170
    },
    {
      "epoch": 4.322056215874577,
      "grad_norm": 0.07028599515154274,
      "learning_rate": 1.1206070298600815e-05,
      "loss": 0.8371,
      "step": 246180
    },
    {
      "epoch": 4.322231780754578,
      "grad_norm": 0.05976276228139678,
      "learning_rate": 1.1201378167919798e-05,
      "loss": 0.8387,
      "step": 246190
    },
    {
      "epoch": 4.322407345634579,
      "grad_norm": 0.063456416936515,
      "learning_rate": 1.119668717500455e-05,
      "loss": 0.839,
      "step": 246200
    },
    {
      "epoch": 4.322582910514581,
      "grad_norm": 0.04257637850094369,
      "learning_rate": 1.1191997319914452e-05,
      "loss": 0.8407,
      "step": 246210
    },
    {
      "epoch": 4.322758475394582,
      "grad_norm": 0.0462726173011611,
      "learning_rate": 1.118730860270899e-05,
      "loss": 0.8377,
      "step": 246220
    },
    {
      "epoch": 4.322934040274584,
      "grad_norm": 0.04312337420808283,
      "learning_rate": 1.1182621023447526e-05,
      "loss": 0.846,
      "step": 246230
    },
    {
      "epoch": 4.323109605154585,
      "grad_norm": 0.043391924320128526,
      "learning_rate": 1.1177934582189473e-05,
      "loss": 0.8363,
      "step": 246240
    },
    {
      "epoch": 4.323285170034587,
      "grad_norm": 0.0572035603327061,
      "learning_rate": 1.1173249278994219e-05,
      "loss": 0.8389,
      "step": 246250
    },
    {
      "epoch": 4.323460734914588,
      "grad_norm": 0.04374176197534241,
      "learning_rate": 1.1168565113921092e-05,
      "loss": 0.8446,
      "step": 246260
    },
    {
      "epoch": 4.323636299794589,
      "grad_norm": 0.04344135765151653,
      "learning_rate": 1.116388208702947e-05,
      "loss": 0.8346,
      "step": 246270
    },
    {
      "epoch": 4.323811864674591,
      "grad_norm": 0.04213466003144027,
      "learning_rate": 1.1159200198378658e-05,
      "loss": 0.8435,
      "step": 246280
    },
    {
      "epoch": 4.323987429554592,
      "grad_norm": 0.0526740748439592,
      "learning_rate": 1.1154519448028023e-05,
      "loss": 0.8343,
      "step": 246290
    },
    {
      "epoch": 4.324162994434594,
      "grad_norm": 0.04575707032887137,
      "learning_rate": 1.1149839836036838e-05,
      "loss": 0.84,
      "step": 246300
    },
    {
      "epoch": 4.324338559314595,
      "grad_norm": 0.06610153288354238,
      "learning_rate": 1.1145161362464388e-05,
      "loss": 0.832,
      "step": 246310
    },
    {
      "epoch": 4.324514124194597,
      "grad_norm": 0.07356751373780998,
      "learning_rate": 1.1140484027369985e-05,
      "loss": 0.8353,
      "step": 246320
    },
    {
      "epoch": 4.324689689074598,
      "grad_norm": 0.053096331138369895,
      "learning_rate": 1.1135807830812861e-05,
      "loss": 0.8348,
      "step": 246330
    },
    {
      "epoch": 4.324865253954599,
      "grad_norm": 0.039503922097329155,
      "learning_rate": 1.1131132772852268e-05,
      "loss": 0.8451,
      "step": 246340
    },
    {
      "epoch": 4.325040818834601,
      "grad_norm": 0.04861484543432536,
      "learning_rate": 1.112645885354747e-05,
      "loss": 0.8368,
      "step": 246350
    },
    {
      "epoch": 4.325216383714602,
      "grad_norm": 0.07231507751235747,
      "learning_rate": 1.1121786072957644e-05,
      "loss": 0.8419,
      "step": 246360
    },
    {
      "epoch": 4.3253919485946035,
      "grad_norm": 0.06261723763640438,
      "learning_rate": 1.111711443114203e-05,
      "loss": 0.8429,
      "step": 246370
    },
    {
      "epoch": 4.3255675134746046,
      "grad_norm": 0.055858348460470615,
      "learning_rate": 1.1112443928159757e-05,
      "loss": 0.8316,
      "step": 246380
    },
    {
      "epoch": 4.325743078354606,
      "grad_norm": 0.05723107836725781,
      "learning_rate": 1.1107774564070095e-05,
      "loss": 0.8468,
      "step": 246390
    },
    {
      "epoch": 4.3259186432346075,
      "grad_norm": 0.046026867140160604,
      "learning_rate": 1.1103106338932135e-05,
      "loss": 0.842,
      "step": 246400
    },
    {
      "epoch": 4.3260942081146085,
      "grad_norm": 0.05988624665884387,
      "learning_rate": 1.1098439252805037e-05,
      "loss": 0.8336,
      "step": 246410
    },
    {
      "epoch": 4.3262697729946105,
      "grad_norm": 0.05938465471094208,
      "learning_rate": 1.1093773305747975e-05,
      "loss": 0.8436,
      "step": 246420
    },
    {
      "epoch": 4.3264453378746115,
      "grad_norm": 0.127978270634578,
      "learning_rate": 1.1089108497820014e-05,
      "loss": 0.8359,
      "step": 246430
    },
    {
      "epoch": 4.3266209027546125,
      "grad_norm": 0.04207254928046134,
      "learning_rate": 1.1084444829080293e-05,
      "loss": 0.8352,
      "step": 246440
    },
    {
      "epoch": 4.3267964676346145,
      "grad_norm": 0.06912948848247473,
      "learning_rate": 1.107978229958786e-05,
      "loss": 0.8413,
      "step": 246450
    },
    {
      "epoch": 4.3269720325146155,
      "grad_norm": 0.04369068777483732,
      "learning_rate": 1.1075120909401845e-05,
      "loss": 0.8374,
      "step": 246460
    },
    {
      "epoch": 4.327147597394617,
      "grad_norm": 0.09760934002587227,
      "learning_rate": 1.1070460658581267e-05,
      "loss": 0.8462,
      "step": 246470
    },
    {
      "epoch": 4.327323162274618,
      "grad_norm": 0.05518121994044258,
      "learning_rate": 1.1065801547185194e-05,
      "loss": 0.8424,
      "step": 246480
    },
    {
      "epoch": 4.32749872715462,
      "grad_norm": 0.05719942409781246,
      "learning_rate": 1.1061143575272656e-05,
      "loss": 0.8422,
      "step": 246490
    },
    {
      "epoch": 4.327674292034621,
      "grad_norm": 0.06183008418350007,
      "learning_rate": 1.105648674290266e-05,
      "loss": 0.8304,
      "step": 246500
    },
    {
      "epoch": 4.327849856914622,
      "grad_norm": 0.06627478174728471,
      "learning_rate": 1.1051831050134213e-05,
      "loss": 0.8343,
      "step": 246510
    },
    {
      "epoch": 4.328025421794624,
      "grad_norm": 0.0547292771935589,
      "learning_rate": 1.104717649702632e-05,
      "loss": 0.832,
      "step": 246520
    },
    {
      "epoch": 4.328200986674625,
      "grad_norm": 0.05503781333067508,
      "learning_rate": 1.1042523083637921e-05,
      "loss": 0.8435,
      "step": 246530
    },
    {
      "epoch": 4.328376551554627,
      "grad_norm": 0.0778686073287856,
      "learning_rate": 1.1037870810028012e-05,
      "loss": 0.8331,
      "step": 246540
    },
    {
      "epoch": 4.328552116434628,
      "grad_norm": 0.053326634554785035,
      "learning_rate": 1.1033219676255487e-05,
      "loss": 0.834,
      "step": 246550
    },
    {
      "epoch": 4.32872768131463,
      "grad_norm": 0.057467174570761434,
      "learning_rate": 1.1028569682379348e-05,
      "loss": 0.8405,
      "step": 246560
    },
    {
      "epoch": 4.328903246194631,
      "grad_norm": 0.047283480772804656,
      "learning_rate": 1.1023920828458451e-05,
      "loss": 0.8295,
      "step": 246570
    },
    {
      "epoch": 4.329078811074632,
      "grad_norm": 0.05564013596966547,
      "learning_rate": 1.1019273114551729e-05,
      "loss": 0.839,
      "step": 246580
    },
    {
      "epoch": 4.329254375954634,
      "grad_norm": 0.050810796512307715,
      "learning_rate": 1.1014626540718075e-05,
      "loss": 0.8432,
      "step": 246590
    },
    {
      "epoch": 4.329429940834635,
      "grad_norm": 0.05529550316422678,
      "learning_rate": 1.1009981107016326e-05,
      "loss": 0.8436,
      "step": 246600
    },
    {
      "epoch": 4.329605505714637,
      "grad_norm": 0.04280166827208792,
      "learning_rate": 1.1005336813505396e-05,
      "loss": 0.8425,
      "step": 246610
    },
    {
      "epoch": 4.329781070594638,
      "grad_norm": 0.0478666075915732,
      "learning_rate": 1.100069366024405e-05,
      "loss": 0.841,
      "step": 246620
    },
    {
      "epoch": 4.329956635474639,
      "grad_norm": 0.0540065439158734,
      "learning_rate": 1.0996051647291191e-05,
      "loss": 0.8323,
      "step": 246630
    },
    {
      "epoch": 4.330132200354641,
      "grad_norm": 0.04794469113166784,
      "learning_rate": 1.0991410774705594e-05,
      "loss": 0.8437,
      "step": 246640
    },
    {
      "epoch": 4.330307765234642,
      "grad_norm": 0.05790242885633262,
      "learning_rate": 1.0986771042546079e-05,
      "loss": 0.8312,
      "step": 246650
    },
    {
      "epoch": 4.330483330114644,
      "grad_norm": 0.0481492951252245,
      "learning_rate": 1.098213245087144e-05,
      "loss": 0.8376,
      "step": 246660
    },
    {
      "epoch": 4.330658894994645,
      "grad_norm": 0.06837519826671988,
      "learning_rate": 1.0977494999740426e-05,
      "loss": 0.8366,
      "step": 246670
    },
    {
      "epoch": 4.330834459874646,
      "grad_norm": 0.05911164997572352,
      "learning_rate": 1.0972858689211793e-05,
      "loss": 0.835,
      "step": 246680
    },
    {
      "epoch": 4.331010024754648,
      "grad_norm": 0.03990117410925253,
      "learning_rate": 1.0968223519344312e-05,
      "loss": 0.8405,
      "step": 246690
    },
    {
      "epoch": 4.331185589634649,
      "grad_norm": 0.05356470024666078,
      "learning_rate": 1.0963589490196677e-05,
      "loss": 0.8375,
      "step": 246700
    },
    {
      "epoch": 4.331361154514651,
      "grad_norm": 0.05043129020550859,
      "learning_rate": 1.0958956601827635e-05,
      "loss": 0.8379,
      "step": 246710
    },
    {
      "epoch": 4.331536719394652,
      "grad_norm": 0.06245407765999437,
      "learning_rate": 1.0954324854295836e-05,
      "loss": 0.8369,
      "step": 246720
    },
    {
      "epoch": 4.331712284274654,
      "grad_norm": 0.05786650095282733,
      "learning_rate": 1.0949694247660028e-05,
      "loss": 0.8421,
      "step": 246730
    },
    {
      "epoch": 4.331887849154655,
      "grad_norm": 0.04402620309703318,
      "learning_rate": 1.094506478197884e-05,
      "loss": 0.8417,
      "step": 246740
    },
    {
      "epoch": 4.332063414034656,
      "grad_norm": 0.06647996766992324,
      "learning_rate": 1.0940436457310938e-05,
      "loss": 0.847,
      "step": 246750
    },
    {
      "epoch": 4.332238978914658,
      "grad_norm": 0.048918721126191965,
      "learning_rate": 1.0935809273714986e-05,
      "loss": 0.8377,
      "step": 246760
    },
    {
      "epoch": 4.332414543794659,
      "grad_norm": 0.04258599759705257,
      "learning_rate": 1.0931183231249576e-05,
      "loss": 0.8467,
      "step": 246770
    },
    {
      "epoch": 4.332590108674661,
      "grad_norm": 0.05183150207789099,
      "learning_rate": 1.0926558329973346e-05,
      "loss": 0.8375,
      "step": 246780
    },
    {
      "epoch": 4.332765673554662,
      "grad_norm": 0.06186298077224962,
      "learning_rate": 1.092193456994485e-05,
      "loss": 0.8407,
      "step": 246790
    },
    {
      "epoch": 4.332941238434664,
      "grad_norm": 0.0491263429306468,
      "learning_rate": 1.0917311951222745e-05,
      "loss": 0.8334,
      "step": 246800
    },
    {
      "epoch": 4.333116803314665,
      "grad_norm": 0.06189438258472078,
      "learning_rate": 1.0912690473865548e-05,
      "loss": 0.8357,
      "step": 246810
    },
    {
      "epoch": 4.333292368194666,
      "grad_norm": 0.04657725806537658,
      "learning_rate": 1.0908070137931822e-05,
      "loss": 0.8399,
      "step": 246820
    },
    {
      "epoch": 4.333467933074668,
      "grad_norm": 0.06420233514350097,
      "learning_rate": 1.0903450943480137e-05,
      "loss": 0.8424,
      "step": 246830
    },
    {
      "epoch": 4.333643497954669,
      "grad_norm": 0.05007879164212958,
      "learning_rate": 1.0898832890568966e-05,
      "loss": 0.8289,
      "step": 246840
    },
    {
      "epoch": 4.333819062834671,
      "grad_norm": 0.05611724632921337,
      "learning_rate": 1.0894215979256861e-05,
      "loss": 0.8412,
      "step": 246850
    },
    {
      "epoch": 4.333994627714672,
      "grad_norm": 0.06347556228322218,
      "learning_rate": 1.0889600209602316e-05,
      "loss": 0.8465,
      "step": 246860
    },
    {
      "epoch": 4.334170192594673,
      "grad_norm": 0.050112205303067994,
      "learning_rate": 1.0884985581663795e-05,
      "loss": 0.84,
      "step": 246870
    },
    {
      "epoch": 4.334345757474675,
      "grad_norm": 0.07002613979632667,
      "learning_rate": 1.0880372095499802e-05,
      "loss": 0.8399,
      "step": 246880
    },
    {
      "epoch": 4.334521322354676,
      "grad_norm": 0.04754482543186891,
      "learning_rate": 1.0875759751168724e-05,
      "loss": 0.839,
      "step": 246890
    },
    {
      "epoch": 4.334696887234678,
      "grad_norm": 0.05362356482925689,
      "learning_rate": 1.0871148548729076e-05,
      "loss": 0.8318,
      "step": 246900
    },
    {
      "epoch": 4.334872452114679,
      "grad_norm": 0.04842190130608221,
      "learning_rate": 1.0866538488239226e-05,
      "loss": 0.8426,
      "step": 246910
    },
    {
      "epoch": 4.335048016994681,
      "grad_norm": 0.054089228168250646,
      "learning_rate": 1.0861929569757618e-05,
      "loss": 0.8339,
      "step": 246920
    },
    {
      "epoch": 4.335223581874682,
      "grad_norm": 0.04833851451945741,
      "learning_rate": 1.0857321793342654e-05,
      "loss": 0.8413,
      "step": 246930
    },
    {
      "epoch": 4.335399146754683,
      "grad_norm": 0.051076096585777934,
      "learning_rate": 1.0852715159052673e-05,
      "loss": 0.839,
      "step": 246940
    },
    {
      "epoch": 4.335574711634685,
      "grad_norm": 0.04828870907572849,
      "learning_rate": 1.0848109666946085e-05,
      "loss": 0.8381,
      "step": 246950
    },
    {
      "epoch": 4.335750276514686,
      "grad_norm": 0.04701197133690719,
      "learning_rate": 1.0843505317081192e-05,
      "loss": 0.8449,
      "step": 246960
    },
    {
      "epoch": 4.335925841394688,
      "grad_norm": 0.04686151001621272,
      "learning_rate": 1.0838902109516411e-05,
      "loss": 0.84,
      "step": 246970
    },
    {
      "epoch": 4.336101406274689,
      "grad_norm": 0.04898871907550736,
      "learning_rate": 1.0834300044309997e-05,
      "loss": 0.8443,
      "step": 246980
    },
    {
      "epoch": 4.336276971154691,
      "grad_norm": 0.048046759794959336,
      "learning_rate": 1.0829699121520273e-05,
      "loss": 0.8378,
      "step": 246990
    },
    {
      "epoch": 4.336452536034692,
      "grad_norm": 0.05444105085838679,
      "learning_rate": 1.082509934120556e-05,
      "loss": 0.8434,
      "step": 247000
    },
    {
      "epoch": 4.336628100914693,
      "grad_norm": 0.05684948773481905,
      "learning_rate": 1.082050070342411e-05,
      "loss": 0.8389,
      "step": 247010
    },
    {
      "epoch": 4.336803665794695,
      "grad_norm": 0.05843160146340864,
      "learning_rate": 1.08159032082342e-05,
      "loss": 0.8463,
      "step": 247020
    },
    {
      "epoch": 4.336979230674696,
      "grad_norm": 0.05712792669522365,
      "learning_rate": 1.0811306855694089e-05,
      "loss": 0.8401,
      "step": 247030
    },
    {
      "epoch": 4.337154795554698,
      "grad_norm": 0.07449278470442873,
      "learning_rate": 1.0806711645861992e-05,
      "loss": 0.8396,
      "step": 247040
    },
    {
      "epoch": 4.337330360434699,
      "grad_norm": 0.062063274822815304,
      "learning_rate": 1.0802117578796162e-05,
      "loss": 0.8361,
      "step": 247050
    },
    {
      "epoch": 4.3375059253147,
      "grad_norm": 0.05279793972870181,
      "learning_rate": 1.0797524654554761e-05,
      "loss": 0.8386,
      "step": 247060
    },
    {
      "epoch": 4.337681490194702,
      "grad_norm": 0.04763312099154702,
      "learning_rate": 1.0792932873196035e-05,
      "loss": 0.8389,
      "step": 247070
    },
    {
      "epoch": 4.337857055074703,
      "grad_norm": 0.05519361864753907,
      "learning_rate": 1.0788342234778125e-05,
      "loss": 0.851,
      "step": 247080
    },
    {
      "epoch": 4.338032619954705,
      "grad_norm": 0.09092976029530582,
      "learning_rate": 1.0783752739359198e-05,
      "loss": 0.847,
      "step": 247090
    },
    {
      "epoch": 4.338208184834706,
      "grad_norm": 0.05110667296969216,
      "learning_rate": 1.077916438699744e-05,
      "loss": 0.8387,
      "step": 247100
    },
    {
      "epoch": 4.338383749714707,
      "grad_norm": 0.04407679898662703,
      "learning_rate": 1.0774577177750942e-05,
      "loss": 0.8391,
      "step": 247110
    },
    {
      "epoch": 4.338559314594709,
      "grad_norm": 0.04564855410729714,
      "learning_rate": 1.0769991111677854e-05,
      "loss": 0.8424,
      "step": 247120
    },
    {
      "epoch": 4.33873487947471,
      "grad_norm": 0.0659091102575813,
      "learning_rate": 1.076540618883625e-05,
      "loss": 0.8394,
      "step": 247130
    },
    {
      "epoch": 4.3389104443547115,
      "grad_norm": 0.048936976824179815,
      "learning_rate": 1.0760822409284266e-05,
      "loss": 0.8333,
      "step": 247140
    },
    {
      "epoch": 4.3390860092347125,
      "grad_norm": 0.06167454242062714,
      "learning_rate": 1.0756239773079941e-05,
      "loss": 0.8402,
      "step": 247150
    },
    {
      "epoch": 4.3392615741147145,
      "grad_norm": 0.04950138699663164,
      "learning_rate": 1.0751658280281357e-05,
      "loss": 0.8355,
      "step": 247160
    },
    {
      "epoch": 4.3394371389947155,
      "grad_norm": 0.059812072913810896,
      "learning_rate": 1.0747077930946565e-05,
      "loss": 0.8434,
      "step": 247170
    },
    {
      "epoch": 4.3396127038747165,
      "grad_norm": 0.049532160188277656,
      "learning_rate": 1.0742498725133585e-05,
      "loss": 0.8453,
      "step": 247180
    },
    {
      "epoch": 4.3397882687547185,
      "grad_norm": 0.0469355692947796,
      "learning_rate": 1.0737920662900442e-05,
      "loss": 0.8475,
      "step": 247190
    },
    {
      "epoch": 4.3399638336347195,
      "grad_norm": 0.045484326001113955,
      "learning_rate": 1.0733343744305163e-05,
      "loss": 0.84,
      "step": 247200
    },
    {
      "epoch": 4.340139398514721,
      "grad_norm": 0.053639670514234954,
      "learning_rate": 1.07287679694057e-05,
      "loss": 0.8335,
      "step": 247210
    },
    {
      "epoch": 4.3403149633947224,
      "grad_norm": 0.056180447572302304,
      "learning_rate": 1.0724193338260064e-05,
      "loss": 0.8501,
      "step": 247220
    },
    {
      "epoch": 4.340490528274724,
      "grad_norm": 0.06230455596011744,
      "learning_rate": 1.0719619850926167e-05,
      "loss": 0.841,
      "step": 247230
    },
    {
      "epoch": 4.340666093154725,
      "grad_norm": 0.04273900688371017,
      "learning_rate": 1.0715047507462018e-05,
      "loss": 0.8349,
      "step": 247240
    },
    {
      "epoch": 4.340841658034726,
      "grad_norm": 0.056924041198447586,
      "learning_rate": 1.0710476307925499e-05,
      "loss": 0.835,
      "step": 247250
    },
    {
      "epoch": 4.341017222914728,
      "grad_norm": 0.05107736751474325,
      "learning_rate": 1.0705906252374554e-05,
      "loss": 0.8399,
      "step": 247260
    },
    {
      "epoch": 4.341192787794729,
      "grad_norm": 0.05663667603502806,
      "learning_rate": 1.070133734086711e-05,
      "loss": 0.8413,
      "step": 247270
    },
    {
      "epoch": 4.341368352674731,
      "grad_norm": 0.0527925204318732,
      "learning_rate": 1.0696769573461e-05,
      "loss": 0.8406,
      "step": 247280
    },
    {
      "epoch": 4.341543917554732,
      "grad_norm": 0.05407713319039168,
      "learning_rate": 1.0692202950214136e-05,
      "loss": 0.8359,
      "step": 247290
    },
    {
      "epoch": 4.341719482434733,
      "grad_norm": 0.05628223671673813,
      "learning_rate": 1.0687637471184344e-05,
      "loss": 0.8385,
      "step": 247300
    },
    {
      "epoch": 4.341895047314735,
      "grad_norm": 0.05281536307407832,
      "learning_rate": 1.0683073136429527e-05,
      "loss": 0.8332,
      "step": 247310
    },
    {
      "epoch": 4.342070612194736,
      "grad_norm": 0.04757572461229528,
      "learning_rate": 1.0678509946007467e-05,
      "loss": 0.8348,
      "step": 247320
    },
    {
      "epoch": 4.342246177074738,
      "grad_norm": 0.04893681033145574,
      "learning_rate": 1.0673947899975984e-05,
      "loss": 0.84,
      "step": 247330
    },
    {
      "epoch": 4.342421741954739,
      "grad_norm": 0.05286417861486481,
      "learning_rate": 1.0669386998392925e-05,
      "loss": 0.8443,
      "step": 247340
    },
    {
      "epoch": 4.34259730683474,
      "grad_norm": 0.051704083329076,
      "learning_rate": 1.066482724131602e-05,
      "loss": 0.8461,
      "step": 247350
    },
    {
      "epoch": 4.342772871714742,
      "grad_norm": 0.04654090643555636,
      "learning_rate": 1.0660268628803058e-05,
      "loss": 0.8349,
      "step": 247360
    },
    {
      "epoch": 4.342948436594743,
      "grad_norm": 0.052469854772400924,
      "learning_rate": 1.0655711160911832e-05,
      "loss": 0.8385,
      "step": 247370
    },
    {
      "epoch": 4.343124001474745,
      "grad_norm": 0.06969684141723473,
      "learning_rate": 1.0651154837700045e-05,
      "loss": 0.8331,
      "step": 247380
    },
    {
      "epoch": 4.343299566354746,
      "grad_norm": 0.07220358730270393,
      "learning_rate": 1.0646599659225456e-05,
      "loss": 0.8339,
      "step": 247390
    },
    {
      "epoch": 4.343475131234748,
      "grad_norm": 0.047696123842767,
      "learning_rate": 1.0642045625545735e-05,
      "loss": 0.8313,
      "step": 247400
    },
    {
      "epoch": 4.343650696114749,
      "grad_norm": 0.0538883079162919,
      "learning_rate": 1.0637492736718643e-05,
      "loss": 0.8397,
      "step": 247410
    },
    {
      "epoch": 4.34382626099475,
      "grad_norm": 0.05170058592431217,
      "learning_rate": 1.0632940992801817e-05,
      "loss": 0.836,
      "step": 247420
    },
    {
      "epoch": 4.344001825874752,
      "grad_norm": 0.05022168130150661,
      "learning_rate": 1.062839039385295e-05,
      "loss": 0.8358,
      "step": 247430
    },
    {
      "epoch": 4.344177390754753,
      "grad_norm": 0.04672470034711838,
      "learning_rate": 1.0623840939929711e-05,
      "loss": 0.834,
      "step": 247440
    },
    {
      "epoch": 4.344352955634755,
      "grad_norm": 0.06644501616367558,
      "learning_rate": 1.0619292631089717e-05,
      "loss": 0.834,
      "step": 247450
    },
    {
      "epoch": 4.344528520514756,
      "grad_norm": 0.04492369377434546,
      "learning_rate": 1.0614745467390615e-05,
      "loss": 0.8472,
      "step": 247460
    },
    {
      "epoch": 4.344704085394758,
      "grad_norm": 0.06677609000791049,
      "learning_rate": 1.0610199448889982e-05,
      "loss": 0.8422,
      "step": 247470
    },
    {
      "epoch": 4.344879650274759,
      "grad_norm": 0.05835753118693348,
      "learning_rate": 1.0605654575645485e-05,
      "loss": 0.8431,
      "step": 247480
    },
    {
      "epoch": 4.34505521515476,
      "grad_norm": 0.04580593013546882,
      "learning_rate": 1.0601110847714643e-05,
      "loss": 0.8392,
      "step": 247490
    },
    {
      "epoch": 4.345230780034762,
      "grad_norm": 0.05270133606018113,
      "learning_rate": 1.0596568265155048e-05,
      "loss": 0.8335,
      "step": 247500
    },
    {
      "epoch": 4.345406344914763,
      "grad_norm": 0.05158037393316766,
      "learning_rate": 1.0592026828024285e-05,
      "loss": 0.8387,
      "step": 247510
    },
    {
      "epoch": 4.345581909794765,
      "grad_norm": 0.07770085544286182,
      "learning_rate": 1.0587486536379846e-05,
      "loss": 0.844,
      "step": 247520
    },
    {
      "epoch": 4.345757474674766,
      "grad_norm": 0.054507614878273304,
      "learning_rate": 1.0582947390279282e-05,
      "loss": 0.8348,
      "step": 247530
    },
    {
      "epoch": 4.345933039554767,
      "grad_norm": 0.060450586292299394,
      "learning_rate": 1.0578409389780113e-05,
      "loss": 0.8375,
      "step": 247540
    },
    {
      "epoch": 4.346108604434769,
      "grad_norm": 0.05366989459582905,
      "learning_rate": 1.0573872534939821e-05,
      "loss": 0.8404,
      "step": 247550
    },
    {
      "epoch": 4.34628416931477,
      "grad_norm": 0.057234028573235625,
      "learning_rate": 1.056933682581591e-05,
      "loss": 0.8406,
      "step": 247560
    },
    {
      "epoch": 4.346459734194772,
      "grad_norm": 0.055334799915985104,
      "learning_rate": 1.056480226246579e-05,
      "loss": 0.8432,
      "step": 247570
    },
    {
      "epoch": 4.346635299074773,
      "grad_norm": 0.06986775410998694,
      "learning_rate": 1.0560268844946997e-05,
      "loss": 0.8429,
      "step": 247580
    },
    {
      "epoch": 4.346810863954775,
      "grad_norm": 0.06073338657550239,
      "learning_rate": 1.055573657331691e-05,
      "loss": 0.8433,
      "step": 247590
    },
    {
      "epoch": 4.346986428834776,
      "grad_norm": 0.05578398808854327,
      "learning_rate": 1.0551205447632976e-05,
      "loss": 0.8399,
      "step": 247600
    },
    {
      "epoch": 4.347161993714777,
      "grad_norm": 0.04817940906201229,
      "learning_rate": 1.0546675467952622e-05,
      "loss": 0.837,
      "step": 247610
    },
    {
      "epoch": 4.347337558594779,
      "grad_norm": 0.05841089064428632,
      "learning_rate": 1.0542146634333206e-05,
      "loss": 0.8329,
      "step": 247620
    },
    {
      "epoch": 4.34751312347478,
      "grad_norm": 0.05551715172882545,
      "learning_rate": 1.0537618946832143e-05,
      "loss": 0.8306,
      "step": 247630
    },
    {
      "epoch": 4.347688688354782,
      "grad_norm": 0.05902304870166148,
      "learning_rate": 1.053309240550675e-05,
      "loss": 0.839,
      "step": 247640
    },
    {
      "epoch": 4.347864253234783,
      "grad_norm": 0.06171582198645077,
      "learning_rate": 1.052856701041445e-05,
      "loss": 0.8343,
      "step": 247650
    },
    {
      "epoch": 4.348039818114784,
      "grad_norm": 0.054104103373196016,
      "learning_rate": 1.0524042761612537e-05,
      "loss": 0.8474,
      "step": 247660
    },
    {
      "epoch": 4.348215382994786,
      "grad_norm": 0.06415949368301029,
      "learning_rate": 1.0519519659158333e-05,
      "loss": 0.8415,
      "step": 247670
    },
    {
      "epoch": 4.348390947874787,
      "grad_norm": 0.06397336789263386,
      "learning_rate": 1.0514997703109175e-05,
      "loss": 0.8406,
      "step": 247680
    },
    {
      "epoch": 4.348566512754789,
      "grad_norm": 0.04352273759347458,
      "learning_rate": 1.0510476893522316e-05,
      "loss": 0.8348,
      "step": 247690
    },
    {
      "epoch": 4.34874207763479,
      "grad_norm": 0.23529036128416236,
      "learning_rate": 1.0505957230455057e-05,
      "loss": 0.8396,
      "step": 247700
    },
    {
      "epoch": 4.348917642514792,
      "grad_norm": 0.04672847735605027,
      "learning_rate": 1.0501438713964696e-05,
      "loss": 0.84,
      "step": 247710
    },
    {
      "epoch": 4.349093207394793,
      "grad_norm": 0.053544058196758115,
      "learning_rate": 1.0496921344108412e-05,
      "loss": 0.8453,
      "step": 247720
    },
    {
      "epoch": 4.349268772274794,
      "grad_norm": 0.05876012443051975,
      "learning_rate": 1.0492405120943513e-05,
      "loss": 0.8377,
      "step": 247730
    },
    {
      "epoch": 4.349444337154796,
      "grad_norm": 0.04874382382400339,
      "learning_rate": 1.0487890044527136e-05,
      "loss": 0.833,
      "step": 247740
    },
    {
      "epoch": 4.349619902034797,
      "grad_norm": 0.04688489029225544,
      "learning_rate": 1.0483376114916587e-05,
      "loss": 0.8341,
      "step": 247750
    },
    {
      "epoch": 4.349795466914799,
      "grad_norm": 0.057506881517610936,
      "learning_rate": 1.0478863332168984e-05,
      "loss": 0.8252,
      "step": 247760
    },
    {
      "epoch": 4.3499710317948,
      "grad_norm": 0.046669235543395665,
      "learning_rate": 1.0474351696341539e-05,
      "loss": 0.8383,
      "step": 247770
    },
    {
      "epoch": 4.350146596674801,
      "grad_norm": 0.060049167073318636,
      "learning_rate": 1.0469841207491413e-05,
      "loss": 0.8389,
      "step": 247780
    },
    {
      "epoch": 4.350322161554803,
      "grad_norm": 0.052804423134517005,
      "learning_rate": 1.0465331865675745e-05,
      "loss": 0.8443,
      "step": 247790
    },
    {
      "epoch": 4.350497726434804,
      "grad_norm": 0.05624406797707193,
      "learning_rate": 1.046082367095168e-05,
      "loss": 0.8341,
      "step": 247800
    },
    {
      "epoch": 4.350673291314806,
      "grad_norm": 0.05996664596104661,
      "learning_rate": 1.0456316623376309e-05,
      "loss": 0.8291,
      "step": 247810
    },
    {
      "epoch": 4.350848856194807,
      "grad_norm": 0.11346253769494401,
      "learning_rate": 1.0451810723006778e-05,
      "loss": 0.8413,
      "step": 247820
    },
    {
      "epoch": 4.351024421074809,
      "grad_norm": 0.047423873788799385,
      "learning_rate": 1.0447305969900151e-05,
      "loss": 0.8403,
      "step": 247830
    },
    {
      "epoch": 4.35119998595481,
      "grad_norm": 0.053328702500530234,
      "learning_rate": 1.0442802364113502e-05,
      "loss": 0.8389,
      "step": 247840
    },
    {
      "epoch": 4.351375550834811,
      "grad_norm": 0.04887173842295668,
      "learning_rate": 1.0438299905703926e-05,
      "loss": 0.8326,
      "step": 247850
    },
    {
      "epoch": 4.351551115714813,
      "grad_norm": 0.052764923990648485,
      "learning_rate": 1.0433798594728434e-05,
      "loss": 0.8326,
      "step": 247860
    },
    {
      "epoch": 4.351726680594814,
      "grad_norm": 0.08116676628433284,
      "learning_rate": 1.0429298431244062e-05,
      "loss": 0.8396,
      "step": 247870
    },
    {
      "epoch": 4.3519022454748155,
      "grad_norm": 0.06662988678372961,
      "learning_rate": 1.0424799415307865e-05,
      "loss": 0.837,
      "step": 247880
    },
    {
      "epoch": 4.352077810354817,
      "grad_norm": 0.04847472515560618,
      "learning_rate": 1.0420301546976793e-05,
      "loss": 0.8461,
      "step": 247890
    },
    {
      "epoch": 4.3522533752348185,
      "grad_norm": 0.05229557450522811,
      "learning_rate": 1.041580482630788e-05,
      "loss": 0.8443,
      "step": 247900
    },
    {
      "epoch": 4.3524289401148195,
      "grad_norm": 0.06124256201457836,
      "learning_rate": 1.0411309253358042e-05,
      "loss": 0.8368,
      "step": 247910
    },
    {
      "epoch": 4.3526045049948205,
      "grad_norm": 0.04888830399024857,
      "learning_rate": 1.0406814828184308e-05,
      "loss": 0.8389,
      "step": 247920
    },
    {
      "epoch": 4.3527800698748225,
      "grad_norm": 0.05183078901593183,
      "learning_rate": 1.0402321550843589e-05,
      "loss": 0.8336,
      "step": 247930
    },
    {
      "epoch": 4.3529556347548235,
      "grad_norm": 0.05587531757395444,
      "learning_rate": 1.0397829421392808e-05,
      "loss": 0.8374,
      "step": 247940
    },
    {
      "epoch": 4.353131199634825,
      "grad_norm": 0.05771440425391081,
      "learning_rate": 1.039333843988891e-05,
      "loss": 0.8399,
      "step": 247950
    },
    {
      "epoch": 4.3533067645148265,
      "grad_norm": 0.048295444853711164,
      "learning_rate": 1.0388848606388765e-05,
      "loss": 0.8376,
      "step": 247960
    },
    {
      "epoch": 4.3534823293948275,
      "grad_norm": 0.046846489367724856,
      "learning_rate": 1.0384359920949282e-05,
      "loss": 0.8436,
      "step": 247970
    },
    {
      "epoch": 4.353657894274829,
      "grad_norm": 0.04807611438363218,
      "learning_rate": 1.0379872383627304e-05,
      "loss": 0.8401,
      "step": 247980
    },
    {
      "epoch": 4.35383345915483,
      "grad_norm": 0.04629529238700462,
      "learning_rate": 1.0375385994479733e-05,
      "loss": 0.8364,
      "step": 247990
    },
    {
      "epoch": 4.354009024034832,
      "grad_norm": 0.06460084136640042,
      "learning_rate": 1.0370900753563381e-05,
      "loss": 0.8314,
      "step": 248000
    },
    {
      "epoch": 4.354184588914833,
      "grad_norm": 0.06727189488232657,
      "learning_rate": 1.036641666093508e-05,
      "loss": 0.8343,
      "step": 248010
    },
    {
      "epoch": 4.354360153794834,
      "grad_norm": 0.0700292966412189,
      "learning_rate": 1.0361933716651673e-05,
      "loss": 0.8394,
      "step": 248020
    },
    {
      "epoch": 4.354535718674836,
      "grad_norm": 0.04157539762181821,
      "learning_rate": 1.0357451920769911e-05,
      "loss": 0.8304,
      "step": 248030
    },
    {
      "epoch": 4.354711283554837,
      "grad_norm": 0.04812858393298939,
      "learning_rate": 1.035297127334661e-05,
      "loss": 0.8396,
      "step": 248040
    },
    {
      "epoch": 4.354886848434839,
      "grad_norm": 0.12660719170533888,
      "learning_rate": 1.0348491774438557e-05,
      "loss": 0.8383,
      "step": 248050
    },
    {
      "epoch": 4.35506241331484,
      "grad_norm": 0.04675807454898014,
      "learning_rate": 1.0344013424102458e-05,
      "loss": 0.8359,
      "step": 248060
    },
    {
      "epoch": 4.355237978194842,
      "grad_norm": 0.040414460646019106,
      "learning_rate": 1.0339536222395105e-05,
      "loss": 0.8352,
      "step": 248070
    },
    {
      "epoch": 4.355413543074843,
      "grad_norm": 0.05139404357536068,
      "learning_rate": 1.0335060169373168e-05,
      "loss": 0.8379,
      "step": 248080
    },
    {
      "epoch": 4.355589107954844,
      "grad_norm": 0.04118243034615932,
      "learning_rate": 1.0330585265093436e-05,
      "loss": 0.8433,
      "step": 248090
    },
    {
      "epoch": 4.355764672834846,
      "grad_norm": 0.04634328061899146,
      "learning_rate": 1.032611150961254e-05,
      "loss": 0.8337,
      "step": 248100
    },
    {
      "epoch": 4.355940237714847,
      "grad_norm": 0.05989227479439736,
      "learning_rate": 1.032163890298719e-05,
      "loss": 0.8354,
      "step": 248110
    },
    {
      "epoch": 4.356115802594849,
      "grad_norm": 0.04273307687358923,
      "learning_rate": 1.0317167445274075e-05,
      "loss": 0.8463,
      "step": 248120
    },
    {
      "epoch": 4.35629136747485,
      "grad_norm": 0.06927850967946549,
      "learning_rate": 1.0312697136529814e-05,
      "loss": 0.8343,
      "step": 248130
    },
    {
      "epoch": 4.356466932354852,
      "grad_norm": 0.05196514140429685,
      "learning_rate": 1.030822797681107e-05,
      "loss": 0.8389,
      "step": 248140
    },
    {
      "epoch": 4.356642497234853,
      "grad_norm": 0.07622986650535693,
      "learning_rate": 1.0303759966174425e-05,
      "loss": 0.8342,
      "step": 248150
    },
    {
      "epoch": 4.356818062114854,
      "grad_norm": 0.05678150618857258,
      "learning_rate": 1.0299293104676553e-05,
      "loss": 0.8366,
      "step": 248160
    },
    {
      "epoch": 4.356993626994856,
      "grad_norm": 0.04706023335039374,
      "learning_rate": 1.029482739237401e-05,
      "loss": 0.8375,
      "step": 248170
    },
    {
      "epoch": 4.357169191874857,
      "grad_norm": 0.04902464985652531,
      "learning_rate": 1.029036282932339e-05,
      "loss": 0.8413,
      "step": 248180
    },
    {
      "epoch": 4.357344756754859,
      "grad_norm": 0.04699657898456082,
      "learning_rate": 1.0285899415581269e-05,
      "loss": 0.8436,
      "step": 248190
    },
    {
      "epoch": 4.35752032163486,
      "grad_norm": 0.0511597232892209,
      "learning_rate": 1.028143715120417e-05,
      "loss": 0.8408,
      "step": 248200
    },
    {
      "epoch": 4.357695886514861,
      "grad_norm": 0.06183189334610382,
      "learning_rate": 1.0276976036248646e-05,
      "loss": 0.8401,
      "step": 248210
    },
    {
      "epoch": 4.357871451394863,
      "grad_norm": 0.05335398428608222,
      "learning_rate": 1.0272516070771245e-05,
      "loss": 0.8372,
      "step": 248220
    },
    {
      "epoch": 4.358047016274864,
      "grad_norm": 0.05237487927704034,
      "learning_rate": 1.0268057254828442e-05,
      "loss": 0.8373,
      "step": 248230
    },
    {
      "epoch": 4.358222581154866,
      "grad_norm": 0.05315674556357546,
      "learning_rate": 1.026359958847675e-05,
      "loss": 0.8309,
      "step": 248240
    },
    {
      "epoch": 4.358398146034867,
      "grad_norm": 0.050284301206024645,
      "learning_rate": 1.0259143071772604e-05,
      "loss": 0.8424,
      "step": 248250
    },
    {
      "epoch": 4.358573710914869,
      "grad_norm": 0.05679402617691081,
      "learning_rate": 1.0254687704772545e-05,
      "loss": 0.8319,
      "step": 248260
    },
    {
      "epoch": 4.35874927579487,
      "grad_norm": 0.0511883134477251,
      "learning_rate": 1.0250233487532965e-05,
      "loss": 0.8408,
      "step": 248270
    },
    {
      "epoch": 4.358924840674871,
      "grad_norm": 0.053856124526375844,
      "learning_rate": 1.0245780420110323e-05,
      "loss": 0.835,
      "step": 248280
    },
    {
      "epoch": 4.359100405554873,
      "grad_norm": 0.04395722420969056,
      "learning_rate": 1.0241328502561051e-05,
      "loss": 0.8331,
      "step": 248290
    },
    {
      "epoch": 4.359275970434874,
      "grad_norm": 0.0485601728746751,
      "learning_rate": 1.023687773494153e-05,
      "loss": 0.839,
      "step": 248300
    },
    {
      "epoch": 4.359451535314876,
      "grad_norm": 0.05775614338668411,
      "learning_rate": 1.0232428117308168e-05,
      "loss": 0.8339,
      "step": 248310
    },
    {
      "epoch": 4.359627100194877,
      "grad_norm": 0.046382438993263095,
      "learning_rate": 1.0227979649717307e-05,
      "loss": 0.8343,
      "step": 248320
    },
    {
      "epoch": 4.359802665074878,
      "grad_norm": 0.06150496253714156,
      "learning_rate": 1.0223532332225374e-05,
      "loss": 0.8313,
      "step": 248330
    },
    {
      "epoch": 4.35997822995488,
      "grad_norm": 0.051111961672807335,
      "learning_rate": 1.0219086164888667e-05,
      "loss": 0.8376,
      "step": 248340
    },
    {
      "epoch": 4.360153794834881,
      "grad_norm": 0.049302046013316565,
      "learning_rate": 1.021464114776354e-05,
      "loss": 0.8358,
      "step": 248350
    },
    {
      "epoch": 4.360329359714883,
      "grad_norm": 0.05423480325895099,
      "learning_rate": 1.021019728090633e-05,
      "loss": 0.8312,
      "step": 248360
    },
    {
      "epoch": 4.360504924594884,
      "grad_norm": 0.05903042191828325,
      "learning_rate": 1.0205754564373293e-05,
      "loss": 0.8328,
      "step": 248370
    },
    {
      "epoch": 4.360680489474886,
      "grad_norm": 0.04636917884180159,
      "learning_rate": 1.0201312998220744e-05,
      "loss": 0.8341,
      "step": 248380
    },
    {
      "epoch": 4.360856054354887,
      "grad_norm": 0.04737016722973275,
      "learning_rate": 1.0196872582504987e-05,
      "loss": 0.8428,
      "step": 248390
    },
    {
      "epoch": 4.361031619234888,
      "grad_norm": 0.04217218441151889,
      "learning_rate": 1.0192433317282234e-05,
      "loss": 0.8387,
      "step": 248400
    },
    {
      "epoch": 4.36120718411489,
      "grad_norm": 0.04845909688459904,
      "learning_rate": 1.0187995202608778e-05,
      "loss": 0.8398,
      "step": 248410
    },
    {
      "epoch": 4.361382748994891,
      "grad_norm": 0.05869441688206428,
      "learning_rate": 1.018355823854079e-05,
      "loss": 0.833,
      "step": 248420
    },
    {
      "epoch": 4.361558313874893,
      "grad_norm": 0.052931981364087666,
      "learning_rate": 1.017912242513455e-05,
      "loss": 0.8417,
      "step": 248430
    },
    {
      "epoch": 4.361733878754894,
      "grad_norm": 0.058290062912634236,
      "learning_rate": 1.0174687762446228e-05,
      "loss": 0.8394,
      "step": 248440
    },
    {
      "epoch": 4.361909443634895,
      "grad_norm": 0.04433505543113629,
      "learning_rate": 1.0170254250532013e-05,
      "loss": 0.8453,
      "step": 248450
    },
    {
      "epoch": 4.362085008514897,
      "grad_norm": 0.048805727625056174,
      "learning_rate": 1.0165821889448108e-05,
      "loss": 0.8385,
      "step": 248460
    },
    {
      "epoch": 4.362260573394898,
      "grad_norm": 0.06040740629445998,
      "learning_rate": 1.0161390679250615e-05,
      "loss": 0.8442,
      "step": 248470
    },
    {
      "epoch": 4.3624361382749,
      "grad_norm": 0.05069631021974266,
      "learning_rate": 1.0156960619995747e-05,
      "loss": 0.844,
      "step": 248480
    },
    {
      "epoch": 4.362611703154901,
      "grad_norm": 0.05437775903719531,
      "learning_rate": 1.0152531711739552e-05,
      "loss": 0.8495,
      "step": 248490
    },
    {
      "epoch": 4.362787268034903,
      "grad_norm": 0.054323861051807615,
      "learning_rate": 1.0148103954538232e-05,
      "loss": 0.8406,
      "step": 248500
    },
    {
      "epoch": 4.362962832914904,
      "grad_norm": 0.05374788573138218,
      "learning_rate": 1.0143677348447828e-05,
      "loss": 0.8343,
      "step": 248510
    },
    {
      "epoch": 4.363138397794905,
      "grad_norm": 0.04888992673500921,
      "learning_rate": 1.0139251893524443e-05,
      "loss": 0.8411,
      "step": 248520
    },
    {
      "epoch": 4.363313962674907,
      "grad_norm": 0.05397941825256283,
      "learning_rate": 1.0134827589824156e-05,
      "loss": 0.8484,
      "step": 248530
    },
    {
      "epoch": 4.363489527554908,
      "grad_norm": 0.04349010104439979,
      "learning_rate": 1.0130404437403013e-05,
      "loss": 0.8368,
      "step": 248540
    },
    {
      "epoch": 4.36366509243491,
      "grad_norm": 0.04430518502816031,
      "learning_rate": 1.012598243631707e-05,
      "loss": 0.8357,
      "step": 248550
    },
    {
      "epoch": 4.363840657314911,
      "grad_norm": 0.05410065528765288,
      "learning_rate": 1.0121561586622344e-05,
      "loss": 0.8419,
      "step": 248560
    },
    {
      "epoch": 4.364016222194913,
      "grad_norm": 0.04958426891636123,
      "learning_rate": 1.011714188837484e-05,
      "loss": 0.8383,
      "step": 248570
    },
    {
      "epoch": 4.364191787074914,
      "grad_norm": 0.04845747155118376,
      "learning_rate": 1.011272334163059e-05,
      "loss": 0.831,
      "step": 248580
    },
    {
      "epoch": 4.364367351954915,
      "grad_norm": 0.05652436297000436,
      "learning_rate": 1.0108305946445504e-05,
      "loss": 0.8368,
      "step": 248590
    },
    {
      "epoch": 4.364542916834917,
      "grad_norm": 0.043899562235199266,
      "learning_rate": 1.010388970287564e-05,
      "loss": 0.837,
      "step": 248600
    },
    {
      "epoch": 4.364718481714918,
      "grad_norm": 0.05382468631558322,
      "learning_rate": 1.00994746109769e-05,
      "loss": 0.8326,
      "step": 248610
    },
    {
      "epoch": 4.3648940465949195,
      "grad_norm": 0.04789280842271871,
      "learning_rate": 1.009506067080524e-05,
      "loss": 0.8387,
      "step": 248620
    },
    {
      "epoch": 4.365069611474921,
      "grad_norm": 0.06411270570866912,
      "learning_rate": 1.0090647882416602e-05,
      "loss": 0.8343,
      "step": 248630
    },
    {
      "epoch": 4.365245176354922,
      "grad_norm": 0.06358708809467144,
      "learning_rate": 1.0086236245866856e-05,
      "loss": 0.8374,
      "step": 248640
    },
    {
      "epoch": 4.3654207412349235,
      "grad_norm": 0.054522354527278584,
      "learning_rate": 1.0081825761211936e-05,
      "loss": 0.8438,
      "step": 248650
    },
    {
      "epoch": 4.3655963061149246,
      "grad_norm": 0.049164295645291844,
      "learning_rate": 1.007741642850768e-05,
      "loss": 0.8367,
      "step": 248660
    },
    {
      "epoch": 4.3657718709949265,
      "grad_norm": 0.053033872848738325,
      "learning_rate": 1.0073008247810018e-05,
      "loss": 0.8403,
      "step": 248670
    },
    {
      "epoch": 4.3659474358749275,
      "grad_norm": 0.07118601328830425,
      "learning_rate": 1.0068601219174745e-05,
      "loss": 0.8414,
      "step": 248680
    },
    {
      "epoch": 4.3661230007549285,
      "grad_norm": 0.05198133701618133,
      "learning_rate": 1.0064195342657715e-05,
      "loss": 0.842,
      "step": 248690
    },
    {
      "epoch": 4.3662985656349305,
      "grad_norm": 0.05881656206127965,
      "learning_rate": 1.0059790618314787e-05,
      "loss": 0.8382,
      "step": 248700
    },
    {
      "epoch": 4.3664741305149315,
      "grad_norm": 0.04899867239000686,
      "learning_rate": 1.0055387046201722e-05,
      "loss": 0.8411,
      "step": 248710
    },
    {
      "epoch": 4.366649695394933,
      "grad_norm": 0.04678959433178068,
      "learning_rate": 1.005098462637433e-05,
      "loss": 0.8414,
      "step": 248720
    },
    {
      "epoch": 4.3668252602749345,
      "grad_norm": 0.04902744814546622,
      "learning_rate": 1.0046583358888403e-05,
      "loss": 0.8451,
      "step": 248730
    },
    {
      "epoch": 4.367000825154936,
      "grad_norm": 0.04503327431226676,
      "learning_rate": 1.0042183243799688e-05,
      "loss": 0.8445,
      "step": 248740
    },
    {
      "epoch": 4.367176390034937,
      "grad_norm": 0.05532853523455059,
      "learning_rate": 1.0037784281163956e-05,
      "loss": 0.8428,
      "step": 248750
    },
    {
      "epoch": 4.367351954914938,
      "grad_norm": 0.05400787769014786,
      "learning_rate": 1.0033386471036899e-05,
      "loss": 0.8427,
      "step": 248760
    },
    {
      "epoch": 4.36752751979494,
      "grad_norm": 0.06659401838344604,
      "learning_rate": 1.0028989813474307e-05,
      "loss": 0.8386,
      "step": 248770
    },
    {
      "epoch": 4.367703084674941,
      "grad_norm": 0.04301939148825867,
      "learning_rate": 1.0024594308531818e-05,
      "loss": 0.8375,
      "step": 248780
    },
    {
      "epoch": 4.367878649554943,
      "grad_norm": 0.05529329513610902,
      "learning_rate": 1.002019995626517e-05,
      "loss": 0.8394,
      "step": 248790
    },
    {
      "epoch": 4.368054214434944,
      "grad_norm": 0.046920977119141206,
      "learning_rate": 1.001580675673003e-05,
      "loss": 0.8422,
      "step": 248800
    },
    {
      "epoch": 4.368229779314946,
      "grad_norm": 0.0526843885764775,
      "learning_rate": 1.001141470998204e-05,
      "loss": 0.8442,
      "step": 248810
    },
    {
      "epoch": 4.368405344194947,
      "grad_norm": 0.06517069551379079,
      "learning_rate": 1.0007023816076889e-05,
      "loss": 0.828,
      "step": 248820
    },
    {
      "epoch": 4.368580909074948,
      "grad_norm": 0.04272013999133828,
      "learning_rate": 1.000263407507014e-05,
      "loss": 0.8353,
      "step": 248830
    },
    {
      "epoch": 4.36875647395495,
      "grad_norm": 0.04994292946155752,
      "learning_rate": 9.998245487017505e-06,
      "loss": 0.8293,
      "step": 248840
    },
    {
      "epoch": 4.368932038834951,
      "grad_norm": 0.05037269718392068,
      "learning_rate": 9.993858051974512e-06,
      "loss": 0.8424,
      "step": 248850
    },
    {
      "epoch": 4.369107603714953,
      "grad_norm": 0.05508911553231758,
      "learning_rate": 9.989471769996778e-06,
      "loss": 0.8376,
      "step": 248860
    },
    {
      "epoch": 4.369283168594954,
      "grad_norm": 0.047645231762443466,
      "learning_rate": 9.985086641139896e-06,
      "loss": 0.836,
      "step": 248870
    },
    {
      "epoch": 4.369458733474955,
      "grad_norm": 0.05125741602709451,
      "learning_rate": 9.980702665459405e-06,
      "loss": 0.8353,
      "step": 248880
    },
    {
      "epoch": 4.369634298354957,
      "grad_norm": 0.045805949071411625,
      "learning_rate": 9.976319843010842e-06,
      "loss": 0.8343,
      "step": 248890
    },
    {
      "epoch": 4.369809863234958,
      "grad_norm": 0.047410626761658595,
      "learning_rate": 9.971938173849771e-06,
      "loss": 0.8461,
      "step": 248900
    },
    {
      "epoch": 4.36998542811496,
      "grad_norm": 0.060443327849524595,
      "learning_rate": 9.96755765803167e-06,
      "loss": 0.8448,
      "step": 248910
    },
    {
      "epoch": 4.370160992994961,
      "grad_norm": 0.06761175278998749,
      "learning_rate": 9.963178295612082e-06,
      "loss": 0.841,
      "step": 248920
    },
    {
      "epoch": 4.370336557874962,
      "grad_norm": 0.050922801872098605,
      "learning_rate": 9.958800086646433e-06,
      "loss": 0.8381,
      "step": 248930
    },
    {
      "epoch": 4.370512122754964,
      "grad_norm": 0.057496415112045014,
      "learning_rate": 9.954423031190278e-06,
      "loss": 0.8409,
      "step": 248940
    },
    {
      "epoch": 4.370687687634965,
      "grad_norm": 0.05042822394517842,
      "learning_rate": 9.950047129299003e-06,
      "loss": 0.8375,
      "step": 248950
    },
    {
      "epoch": 4.370863252514967,
      "grad_norm": 0.04094578052307378,
      "learning_rate": 9.945672381028078e-06,
      "loss": 0.8334,
      "step": 248960
    },
    {
      "epoch": 4.371038817394968,
      "grad_norm": 0.05495139728502668,
      "learning_rate": 9.941298786432961e-06,
      "loss": 0.837,
      "step": 248970
    },
    {
      "epoch": 4.37121438227497,
      "grad_norm": 0.05547866299881619,
      "learning_rate": 9.936926345569028e-06,
      "loss": 0.8314,
      "step": 248980
    },
    {
      "epoch": 4.371389947154971,
      "grad_norm": 0.04385621453952655,
      "learning_rate": 9.932555058491693e-06,
      "loss": 0.835,
      "step": 248990
    },
    {
      "epoch": 4.371565512034972,
      "grad_norm": 0.05123200790157125,
      "learning_rate": 9.928184925256315e-06,
      "loss": 0.8362,
      "step": 249000
    },
    {
      "epoch": 4.371741076914974,
      "grad_norm": 0.050349521462646867,
      "learning_rate": 9.923815945918314e-06,
      "loss": 0.8261,
      "step": 249010
    },
    {
      "epoch": 4.371916641794975,
      "grad_norm": 0.07386852341109845,
      "learning_rate": 9.919448120533012e-06,
      "loss": 0.8304,
      "step": 249020
    },
    {
      "epoch": 4.372092206674977,
      "grad_norm": 0.05990999491210293,
      "learning_rate": 9.91508144915575e-06,
      "loss": 0.8415,
      "step": 249030
    },
    {
      "epoch": 4.372267771554978,
      "grad_norm": 0.07068731608349914,
      "learning_rate": 9.91071593184188e-06,
      "loss": 0.8414,
      "step": 249040
    },
    {
      "epoch": 4.37244333643498,
      "grad_norm": 0.04388061970148544,
      "learning_rate": 9.90635156864668e-06,
      "loss": 0.8357,
      "step": 249050
    },
    {
      "epoch": 4.372618901314981,
      "grad_norm": 0.04469423030665372,
      "learning_rate": 9.901988359625471e-06,
      "loss": 0.8416,
      "step": 249060
    },
    {
      "epoch": 4.372794466194982,
      "grad_norm": 0.05612687730248978,
      "learning_rate": 9.897626304833538e-06,
      "loss": 0.8405,
      "step": 249070
    },
    {
      "epoch": 4.372970031074984,
      "grad_norm": 0.05309558548799841,
      "learning_rate": 9.893265404326124e-06,
      "loss": 0.8376,
      "step": 249080
    },
    {
      "epoch": 4.373145595954985,
      "grad_norm": 0.053504712179967155,
      "learning_rate": 9.888905658158511e-06,
      "loss": 0.8381,
      "step": 249090
    },
    {
      "epoch": 4.373321160834987,
      "grad_norm": 0.04677908749013473,
      "learning_rate": 9.884547066385896e-06,
      "loss": 0.8411,
      "step": 249100
    },
    {
      "epoch": 4.373496725714988,
      "grad_norm": 0.04711099364470261,
      "learning_rate": 9.880189629063562e-06,
      "loss": 0.8338,
      "step": 249110
    },
    {
      "epoch": 4.373672290594989,
      "grad_norm": 0.05092884572323222,
      "learning_rate": 9.875833346246677e-06,
      "loss": 0.8414,
      "step": 249120
    },
    {
      "epoch": 4.373847855474991,
      "grad_norm": 0.04188982661386521,
      "learning_rate": 9.871478217990446e-06,
      "loss": 0.836,
      "step": 249130
    },
    {
      "epoch": 4.374023420354992,
      "grad_norm": 0.06602741312002793,
      "learning_rate": 9.867124244350059e-06,
      "loss": 0.8491,
      "step": 249140
    },
    {
      "epoch": 4.374198985234994,
      "grad_norm": 0.045975871553402055,
      "learning_rate": 9.86277142538067e-06,
      "loss": 0.8343,
      "step": 249150
    },
    {
      "epoch": 4.374374550114995,
      "grad_norm": 0.050867968533257334,
      "learning_rate": 9.85841976113744e-06,
      "loss": 0.8424,
      "step": 249160
    },
    {
      "epoch": 4.374550114994997,
      "grad_norm": 0.04929012442184749,
      "learning_rate": 9.854069251675464e-06,
      "loss": 0.8328,
      "step": 249170
    },
    {
      "epoch": 4.374725679874998,
      "grad_norm": 0.04484563046332468,
      "learning_rate": 9.849719897049941e-06,
      "loss": 0.8492,
      "step": 249180
    },
    {
      "epoch": 4.374901244754999,
      "grad_norm": 0.04541211533379785,
      "learning_rate": 9.845371697315911e-06,
      "loss": 0.8418,
      "step": 249190
    },
    {
      "epoch": 4.375076809635001,
      "grad_norm": 0.05675361950669918,
      "learning_rate": 9.841024652528494e-06,
      "loss": 0.8408,
      "step": 249200
    },
    {
      "epoch": 4.375252374515002,
      "grad_norm": 0.05159887359531993,
      "learning_rate": 9.836678762742791e-06,
      "loss": 0.8408,
      "step": 249210
    },
    {
      "epoch": 4.375427939395004,
      "grad_norm": 0.0580475837877503,
      "learning_rate": 9.832334028013806e-06,
      "loss": 0.8351,
      "step": 249220
    },
    {
      "epoch": 4.375603504275005,
      "grad_norm": 0.04693996391331549,
      "learning_rate": 9.827990448396632e-06,
      "loss": 0.8396,
      "step": 249230
    },
    {
      "epoch": 4.375779069155007,
      "grad_norm": 0.04810508873558979,
      "learning_rate": 9.823648023946302e-06,
      "loss": 0.8366,
      "step": 249240
    },
    {
      "epoch": 4.375954634035008,
      "grad_norm": 0.06942495450433706,
      "learning_rate": 9.819306754717817e-06,
      "loss": 0.8455,
      "step": 249250
    },
    {
      "epoch": 4.376130198915009,
      "grad_norm": 0.06797876989431985,
      "learning_rate": 9.814966640766205e-06,
      "loss": 0.8299,
      "step": 249260
    },
    {
      "epoch": 4.376305763795011,
      "grad_norm": 0.060183940164738096,
      "learning_rate": 9.810627682146406e-06,
      "loss": 0.8434,
      "step": 249270
    },
    {
      "epoch": 4.376481328675012,
      "grad_norm": 0.04382848299436312,
      "learning_rate": 9.806289878913458e-06,
      "loss": 0.8412,
      "step": 249280
    },
    {
      "epoch": 4.376656893555014,
      "grad_norm": 0.04803939193556529,
      "learning_rate": 9.801953231122291e-06,
      "loss": 0.8389,
      "step": 249290
    },
    {
      "epoch": 4.376832458435015,
      "grad_norm": 0.05107956372197884,
      "learning_rate": 9.797617738827843e-06,
      "loss": 0.8418,
      "step": 249300
    },
    {
      "epoch": 4.377008023315016,
      "grad_norm": 0.056996064877260666,
      "learning_rate": 9.793283402085073e-06,
      "loss": 0.846,
      "step": 249310
    },
    {
      "epoch": 4.377183588195018,
      "grad_norm": 0.05106064550309327,
      "learning_rate": 9.788950220948875e-06,
      "loss": 0.839,
      "step": 249320
    },
    {
      "epoch": 4.377359153075019,
      "grad_norm": 0.05365400363357463,
      "learning_rate": 9.784618195474166e-06,
      "loss": 0.8419,
      "step": 249330
    },
    {
      "epoch": 4.377534717955021,
      "grad_norm": 0.05413463702819866,
      "learning_rate": 9.780287325715806e-06,
      "loss": 0.8316,
      "step": 249340
    },
    {
      "epoch": 4.377710282835022,
      "grad_norm": 0.04880922330598473,
      "learning_rate": 9.775957611728709e-06,
      "loss": 0.8359,
      "step": 249350
    },
    {
      "epoch": 4.377885847715023,
      "grad_norm": 0.04819266191724967,
      "learning_rate": 9.771629053567694e-06,
      "loss": 0.8307,
      "step": 249360
    },
    {
      "epoch": 4.378061412595025,
      "grad_norm": 0.048593549884073106,
      "learning_rate": 9.767301651287633e-06,
      "loss": 0.8374,
      "step": 249370
    },
    {
      "epoch": 4.378236977475026,
      "grad_norm": 0.04149940613072813,
      "learning_rate": 9.762975404943348e-06,
      "loss": 0.8425,
      "step": 249380
    },
    {
      "epoch": 4.3784125423550275,
      "grad_norm": 0.0451300986216177,
      "learning_rate": 9.758650314589643e-06,
      "loss": 0.838,
      "step": 249390
    },
    {
      "epoch": 4.378588107235029,
      "grad_norm": 0.06260368552253591,
      "learning_rate": 9.75432638028132e-06,
      "loss": 0.8446,
      "step": 249400
    },
    {
      "epoch": 4.3787636721150305,
      "grad_norm": 0.04558345513743233,
      "learning_rate": 9.75000360207319e-06,
      "loss": 0.8378,
      "step": 249410
    },
    {
      "epoch": 4.3789392369950315,
      "grad_norm": 0.0638939995086517,
      "learning_rate": 9.745681980019983e-06,
      "loss": 0.8321,
      "step": 249420
    },
    {
      "epoch": 4.3791148018750325,
      "grad_norm": 0.04574029824322572,
      "learning_rate": 9.741361514176495e-06,
      "loss": 0.8454,
      "step": 249430
    },
    {
      "epoch": 4.3792903667550345,
      "grad_norm": 0.04656280552222453,
      "learning_rate": 9.73704220459741e-06,
      "loss": 0.838,
      "step": 249440
    },
    {
      "epoch": 4.3794659316350355,
      "grad_norm": 0.04606637910787882,
      "learning_rate": 9.732724051337522e-06,
      "loss": 0.8466,
      "step": 249450
    },
    {
      "epoch": 4.379641496515037,
      "grad_norm": 0.046618406022645525,
      "learning_rate": 9.728407054451497e-06,
      "loss": 0.8379,
      "step": 249460
    },
    {
      "epoch": 4.3798170613950385,
      "grad_norm": 0.05279606552166028,
      "learning_rate": 9.72409121399404e-06,
      "loss": 0.8294,
      "step": 249470
    },
    {
      "epoch": 4.37999262627504,
      "grad_norm": 0.04281514694205284,
      "learning_rate": 9.719776530019864e-06,
      "loss": 0.8427,
      "step": 249480
    },
    {
      "epoch": 4.380168191155041,
      "grad_norm": 0.0463030553260527,
      "learning_rate": 9.7154630025836e-06,
      "loss": 0.8373,
      "step": 249490
    },
    {
      "epoch": 4.3803437560350424,
      "grad_norm": 0.05282623071838824,
      "learning_rate": 9.711150631739921e-06,
      "loss": 0.8435,
      "step": 249500
    },
    {
      "epoch": 4.380519320915044,
      "grad_norm": 0.04986738699618986,
      "learning_rate": 9.70683941754344e-06,
      "loss": 0.8373,
      "step": 249510
    },
    {
      "epoch": 4.380694885795045,
      "grad_norm": 0.053025171455297775,
      "learning_rate": 9.70252936004882e-06,
      "loss": 0.8427,
      "step": 249520
    },
    {
      "epoch": 4.380870450675047,
      "grad_norm": 0.0476116917242871,
      "learning_rate": 9.698220459310639e-06,
      "loss": 0.8423,
      "step": 249530
    },
    {
      "epoch": 4.381046015555048,
      "grad_norm": 0.06064420330627581,
      "learning_rate": 9.693912715383497e-06,
      "loss": 0.8329,
      "step": 249540
    },
    {
      "epoch": 4.381221580435049,
      "grad_norm": 0.05544931057352584,
      "learning_rate": 9.68960612832201e-06,
      "loss": 0.8339,
      "step": 249550
    },
    {
      "epoch": 4.381397145315051,
      "grad_norm": 0.04767928425932525,
      "learning_rate": 9.685300698180683e-06,
      "loss": 0.8259,
      "step": 249560
    },
    {
      "epoch": 4.381572710195052,
      "grad_norm": 0.06326129435367762,
      "learning_rate": 9.680996425014097e-06,
      "loss": 0.8384,
      "step": 249570
    },
    {
      "epoch": 4.381748275075054,
      "grad_norm": 0.05109216377911851,
      "learning_rate": 9.676693308876806e-06,
      "loss": 0.8302,
      "step": 249580
    },
    {
      "epoch": 4.381923839955055,
      "grad_norm": 0.046966653559451677,
      "learning_rate": 9.6723913498233e-06,
      "loss": 0.8293,
      "step": 249590
    },
    {
      "epoch": 4.382099404835056,
      "grad_norm": 0.050840743978369934,
      "learning_rate": 9.668090547908113e-06,
      "loss": 0.8381,
      "step": 249600
    },
    {
      "epoch": 4.382274969715058,
      "grad_norm": 0.07286403870882209,
      "learning_rate": 9.663790903185685e-06,
      "loss": 0.8335,
      "step": 249610
    },
    {
      "epoch": 4.382450534595059,
      "grad_norm": 0.0566738180969571,
      "learning_rate": 9.659492415710575e-06,
      "loss": 0.8358,
      "step": 249620
    },
    {
      "epoch": 4.382626099475061,
      "grad_norm": 0.04873184728229524,
      "learning_rate": 9.655195085537172e-06,
      "loss": 0.8385,
      "step": 249630
    },
    {
      "epoch": 4.382801664355062,
      "grad_norm": 0.043052134553500696,
      "learning_rate": 9.65089891271996e-06,
      "loss": 0.8362,
      "step": 249640
    },
    {
      "epoch": 4.382977229235064,
      "grad_norm": 0.04649703934174705,
      "learning_rate": 9.646603897313385e-06,
      "loss": 0.8407,
      "step": 249650
    },
    {
      "epoch": 4.383152794115065,
      "grad_norm": 0.057107164217081036,
      "learning_rate": 9.642310039371824e-06,
      "loss": 0.8418,
      "step": 249660
    },
    {
      "epoch": 4.383328358995066,
      "grad_norm": 0.05558345502075509,
      "learning_rate": 9.63801733894972e-06,
      "loss": 0.8369,
      "step": 249670
    },
    {
      "epoch": 4.383503923875068,
      "grad_norm": 0.050291869535012995,
      "learning_rate": 9.633725796101423e-06,
      "loss": 0.8396,
      "step": 249680
    },
    {
      "epoch": 4.383679488755069,
      "grad_norm": 0.05659088899670341,
      "learning_rate": 9.629435410881351e-06,
      "loss": 0.8342,
      "step": 249690
    },
    {
      "epoch": 4.383855053635071,
      "grad_norm": 0.05133232817966617,
      "learning_rate": 9.625146183343837e-06,
      "loss": 0.8363,
      "step": 249700
    },
    {
      "epoch": 4.384030618515072,
      "grad_norm": 0.06246089536617769,
      "learning_rate": 9.620858113543226e-06,
      "loss": 0.838,
      "step": 249710
    },
    {
      "epoch": 4.384206183395074,
      "grad_norm": 0.05221195431082349,
      "learning_rate": 9.616571201533874e-06,
      "loss": 0.8352,
      "step": 249720
    },
    {
      "epoch": 4.384381748275075,
      "grad_norm": 0.06369491495788297,
      "learning_rate": 9.612285447370057e-06,
      "loss": 0.8385,
      "step": 249730
    },
    {
      "epoch": 4.384557313155076,
      "grad_norm": 0.05931581284978873,
      "learning_rate": 9.608000851106097e-06,
      "loss": 0.8391,
      "step": 249740
    },
    {
      "epoch": 4.384732878035078,
      "grad_norm": 0.04336253270558309,
      "learning_rate": 9.603717412796308e-06,
      "loss": 0.8364,
      "step": 249750
    },
    {
      "epoch": 4.384908442915079,
      "grad_norm": 0.05440838234697337,
      "learning_rate": 9.59943513249491e-06,
      "loss": 0.8357,
      "step": 249760
    },
    {
      "epoch": 4.385084007795081,
      "grad_norm": 0.04820377421643382,
      "learning_rate": 9.59515401025621e-06,
      "loss": 0.8388,
      "step": 249770
    },
    {
      "epoch": 4.385259572675082,
      "grad_norm": 0.06918884086134826,
      "learning_rate": 9.59087404613439e-06,
      "loss": 0.8397,
      "step": 249780
    },
    {
      "epoch": 4.385435137555083,
      "grad_norm": 0.047684505913767765,
      "learning_rate": 9.586595240183741e-06,
      "loss": 0.8373,
      "step": 249790
    },
    {
      "epoch": 4.385610702435085,
      "grad_norm": 0.05632367136069801,
      "learning_rate": 9.582317592458448e-06,
      "loss": 0.8372,
      "step": 249800
    },
    {
      "epoch": 4.385786267315086,
      "grad_norm": 0.06549391167029689,
      "learning_rate": 9.578041103012694e-06,
      "loss": 0.8414,
      "step": 249810
    },
    {
      "epoch": 4.385961832195088,
      "grad_norm": 0.05572149662692996,
      "learning_rate": 9.573765771900705e-06,
      "loss": 0.8339,
      "step": 249820
    },
    {
      "epoch": 4.386137397075089,
      "grad_norm": 0.05104797188395551,
      "learning_rate": 9.569491599176608e-06,
      "loss": 0.8362,
      "step": 249830
    },
    {
      "epoch": 4.386312961955091,
      "grad_norm": 0.05234409930026774,
      "learning_rate": 9.565218584894592e-06,
      "loss": 0.8409,
      "step": 249840
    },
    {
      "epoch": 4.386488526835092,
      "grad_norm": 0.04791358172149713,
      "learning_rate": 9.560946729108744e-06,
      "loss": 0.838,
      "step": 249850
    },
    {
      "epoch": 4.386664091715093,
      "grad_norm": 0.07210082493946582,
      "learning_rate": 9.556676031873262e-06,
      "loss": 0.8429,
      "step": 249860
    },
    {
      "epoch": 4.386839656595095,
      "grad_norm": 0.05184235710627151,
      "learning_rate": 9.552406493242212e-06,
      "loss": 0.8447,
      "step": 249870
    },
    {
      "epoch": 4.387015221475096,
      "grad_norm": 0.051841697607396764,
      "learning_rate": 9.54813811326968e-06,
      "loss": 0.8431,
      "step": 249880
    },
    {
      "epoch": 4.387190786355098,
      "grad_norm": 0.048820849151274055,
      "learning_rate": 9.543870892009787e-06,
      "loss": 0.8397,
      "step": 249890
    },
    {
      "epoch": 4.387366351235099,
      "grad_norm": 0.06741945580192654,
      "learning_rate": 9.539604829516569e-06,
      "loss": 0.8404,
      "step": 249900
    },
    {
      "epoch": 4.387541916115101,
      "grad_norm": 0.05035439713417252,
      "learning_rate": 9.535339925844078e-06,
      "loss": 0.8351,
      "step": 249910
    },
    {
      "epoch": 4.387717480995102,
      "grad_norm": 0.0749932983915975,
      "learning_rate": 9.531076181046378e-06,
      "loss": 0.8403,
      "step": 249920
    },
    {
      "epoch": 4.387893045875103,
      "grad_norm": 0.05448498126717617,
      "learning_rate": 9.52681359517746e-06,
      "loss": 0.8346,
      "step": 249930
    },
    {
      "epoch": 4.388068610755105,
      "grad_norm": 0.05890090855179807,
      "learning_rate": 9.52255216829136e-06,
      "loss": 0.8378,
      "step": 249940
    },
    {
      "epoch": 4.388244175635106,
      "grad_norm": 0.054794255058811824,
      "learning_rate": 9.518291900442017e-06,
      "loss": 0.8434,
      "step": 249950
    },
    {
      "epoch": 4.388419740515108,
      "grad_norm": 0.061555507450643705,
      "learning_rate": 9.514032791683478e-06,
      "loss": 0.8424,
      "step": 249960
    },
    {
      "epoch": 4.388595305395109,
      "grad_norm": 0.04701596096427494,
      "learning_rate": 9.509774842069674e-06,
      "loss": 0.8375,
      "step": 249970
    },
    {
      "epoch": 4.38877087027511,
      "grad_norm": 0.06811457420020534,
      "learning_rate": 9.505518051654544e-06,
      "loss": 0.8348,
      "step": 249980
    },
    {
      "epoch": 4.388946435155112,
      "grad_norm": 0.05331561342773622,
      "learning_rate": 9.50126242049206e-06,
      "loss": 0.841,
      "step": 249990
    },
    {
      "epoch": 4.389122000035113,
      "grad_norm": 0.05417008611807753,
      "learning_rate": 9.497007948636098e-06,
      "loss": 0.8327,
      "step": 250000
    },
    {
      "epoch": 4.389297564915115,
      "grad_norm": 0.042284562515361,
      "learning_rate": 9.49275463614059e-06,
      "loss": 0.8436,
      "step": 250010
    },
    {
      "epoch": 4.389473129795116,
      "grad_norm": 0.0814357559531662,
      "learning_rate": 9.488502483059398e-06,
      "loss": 0.8288,
      "step": 250020
    },
    {
      "epoch": 4.389648694675117,
      "grad_norm": 0.05009718484998311,
      "learning_rate": 9.484251489446441e-06,
      "loss": 0.8353,
      "step": 250030
    },
    {
      "epoch": 4.389824259555119,
      "grad_norm": 0.05803697273643302,
      "learning_rate": 9.480001655355541e-06,
      "loss": 0.8357,
      "step": 250040
    },
    {
      "epoch": 4.38999982443512,
      "grad_norm": 0.06157386359891623,
      "learning_rate": 9.475752980840563e-06,
      "loss": 0.8344,
      "step": 250050
    },
    {
      "epoch": 4.390175389315122,
      "grad_norm": 0.04522178356332587,
      "learning_rate": 9.471505465955351e-06,
      "loss": 0.8406,
      "step": 250060
    },
    {
      "epoch": 4.390350954195123,
      "grad_norm": 0.06825843813453555,
      "learning_rate": 9.467259110753691e-06,
      "loss": 0.8302,
      "step": 250070
    },
    {
      "epoch": 4.390526519075125,
      "grad_norm": 0.046538173268722685,
      "learning_rate": 9.463013915289401e-06,
      "loss": 0.8406,
      "step": 250080
    },
    {
      "epoch": 4.390702083955126,
      "grad_norm": 0.051211949185671395,
      "learning_rate": 9.458769879616275e-06,
      "loss": 0.842,
      "step": 250090
    },
    {
      "epoch": 4.390877648835127,
      "grad_norm": 0.06167265185366056,
      "learning_rate": 9.45452700378807e-06,
      "loss": 0.8318,
      "step": 250100
    },
    {
      "epoch": 4.391053213715129,
      "grad_norm": 0.09607699682312983,
      "learning_rate": 9.450285287858576e-06,
      "loss": 0.8307,
      "step": 250110
    },
    {
      "epoch": 4.39122877859513,
      "grad_norm": 0.05541184409244226,
      "learning_rate": 9.446044731881465e-06,
      "loss": 0.8443,
      "step": 250120
    },
    {
      "epoch": 4.3914043434751315,
      "grad_norm": 0.0483929692292693,
      "learning_rate": 9.441805335910562e-06,
      "loss": 0.8359,
      "step": 250130
    },
    {
      "epoch": 4.391579908355133,
      "grad_norm": 0.04770473435787869,
      "learning_rate": 9.437567099999497e-06,
      "loss": 0.8353,
      "step": 250140
    },
    {
      "epoch": 4.3917554732351345,
      "grad_norm": 0.06482300047564389,
      "learning_rate": 9.433330024202017e-06,
      "loss": 0.8427,
      "step": 250150
    },
    {
      "epoch": 4.3919310381151355,
      "grad_norm": 0.0786678969411667,
      "learning_rate": 9.429094108571815e-06,
      "loss": 0.834,
      "step": 250160
    },
    {
      "epoch": 4.3921066029951366,
      "grad_norm": 0.05023753599923858,
      "learning_rate": 9.424859353162513e-06,
      "loss": 0.8391,
      "step": 250170
    },
    {
      "epoch": 4.3922821678751385,
      "grad_norm": 0.046284866609783475,
      "learning_rate": 9.42062575802781e-06,
      "loss": 0.8421,
      "step": 250180
    },
    {
      "epoch": 4.3924577327551395,
      "grad_norm": 0.05663136283114535,
      "learning_rate": 9.4163933232213e-06,
      "loss": 0.8349,
      "step": 250190
    },
    {
      "epoch": 4.392633297635141,
      "grad_norm": 0.05812924745765406,
      "learning_rate": 9.412162048796678e-06,
      "loss": 0.8354,
      "step": 250200
    },
    {
      "epoch": 4.3928088625151425,
      "grad_norm": 0.05997327533841313,
      "learning_rate": 9.407931934807487e-06,
      "loss": 0.8413,
      "step": 250210
    },
    {
      "epoch": 4.3929844273951435,
      "grad_norm": 0.047690855827685855,
      "learning_rate": 9.403702981307361e-06,
      "loss": 0.8432,
      "step": 250220
    },
    {
      "epoch": 4.393159992275145,
      "grad_norm": 0.048969100456241976,
      "learning_rate": 9.399475188349885e-06,
      "loss": 0.8467,
      "step": 250230
    },
    {
      "epoch": 4.3933355571551465,
      "grad_norm": 0.05398518551574716,
      "learning_rate": 9.395248555988604e-06,
      "loss": 0.8416,
      "step": 250240
    },
    {
      "epoch": 4.393511122035148,
      "grad_norm": 0.0534478227695198,
      "learning_rate": 9.391023084277076e-06,
      "loss": 0.8338,
      "step": 250250
    },
    {
      "epoch": 4.393686686915149,
      "grad_norm": 0.04650662607507607,
      "learning_rate": 9.386798773268857e-06,
      "loss": 0.8371,
      "step": 250260
    },
    {
      "epoch": 4.39386225179515,
      "grad_norm": 0.05409485482161247,
      "learning_rate": 9.382575623017436e-06,
      "loss": 0.8382,
      "step": 250270
    },
    {
      "epoch": 4.394037816675152,
      "grad_norm": 0.04247891006657534,
      "learning_rate": 9.378353633576358e-06,
      "loss": 0.8405,
      "step": 250280
    },
    {
      "epoch": 4.394213381555153,
      "grad_norm": 0.047471596658553224,
      "learning_rate": 9.374132804999067e-06,
      "loss": 0.8418,
      "step": 250290
    },
    {
      "epoch": 4.394388946435155,
      "grad_norm": 0.05309671804936861,
      "learning_rate": 9.369913137339108e-06,
      "loss": 0.8371,
      "step": 250300
    },
    {
      "epoch": 4.394564511315156,
      "grad_norm": 0.05831194080912037,
      "learning_rate": 9.365694630649888e-06,
      "loss": 0.8401,
      "step": 250310
    },
    {
      "epoch": 4.394740076195158,
      "grad_norm": 0.043197974061099195,
      "learning_rate": 9.36147728498489e-06,
      "loss": 0.8365,
      "step": 250320
    },
    {
      "epoch": 4.394915641075159,
      "grad_norm": 0.054614841436262426,
      "learning_rate": 9.357261100397546e-06,
      "loss": 0.8416,
      "step": 250330
    },
    {
      "epoch": 4.39509120595516,
      "grad_norm": 0.05874929162024638,
      "learning_rate": 9.353046076941247e-06,
      "loss": 0.8431,
      "step": 250340
    },
    {
      "epoch": 4.395266770835162,
      "grad_norm": 0.06289145956018707,
      "learning_rate": 9.348832214669442e-06,
      "loss": 0.8387,
      "step": 250350
    },
    {
      "epoch": 4.395442335715163,
      "grad_norm": 0.0792231000777063,
      "learning_rate": 9.344619513635467e-06,
      "loss": 0.8408,
      "step": 250360
    },
    {
      "epoch": 4.395617900595165,
      "grad_norm": 0.09119388050056848,
      "learning_rate": 9.340407973892754e-06,
      "loss": 0.8282,
      "step": 250370
    },
    {
      "epoch": 4.395793465475166,
      "grad_norm": 0.05664302502587843,
      "learning_rate": 9.336197595494625e-06,
      "loss": 0.8445,
      "step": 250380
    },
    {
      "epoch": 4.395969030355168,
      "grad_norm": 0.05840583862911107,
      "learning_rate": 9.331988378494433e-06,
      "loss": 0.8397,
      "step": 250390
    },
    {
      "epoch": 4.396144595235169,
      "grad_norm": 0.04790376198998712,
      "learning_rate": 9.327780322945547e-06,
      "loss": 0.8362,
      "step": 250400
    },
    {
      "epoch": 4.39632016011517,
      "grad_norm": 0.048447659804067206,
      "learning_rate": 9.323573428901222e-06,
      "loss": 0.8356,
      "step": 250410
    },
    {
      "epoch": 4.396495724995172,
      "grad_norm": 0.059187451335722494,
      "learning_rate": 9.319367696414796e-06,
      "loss": 0.8421,
      "step": 250420
    },
    {
      "epoch": 4.396671289875173,
      "grad_norm": 0.04054083854162119,
      "learning_rate": 9.315163125539569e-06,
      "loss": 0.8389,
      "step": 250430
    },
    {
      "epoch": 4.396846854755175,
      "grad_norm": 0.045853746624040084,
      "learning_rate": 9.310959716328783e-06,
      "loss": 0.8382,
      "step": 250440
    },
    {
      "epoch": 4.397022419635176,
      "grad_norm": 0.045449790375425986,
      "learning_rate": 9.30675746883573e-06,
      "loss": 0.8474,
      "step": 250450
    },
    {
      "epoch": 4.397197984515177,
      "grad_norm": 0.051347549599226876,
      "learning_rate": 9.302556383113586e-06,
      "loss": 0.8404,
      "step": 250460
    },
    {
      "epoch": 4.397373549395179,
      "grad_norm": 0.05587685825611571,
      "learning_rate": 9.298356459215674e-06,
      "loss": 0.8363,
      "step": 250470
    },
    {
      "epoch": 4.39754911427518,
      "grad_norm": 0.0547363942821251,
      "learning_rate": 9.29415769719514e-06,
      "loss": 0.8347,
      "step": 250480
    },
    {
      "epoch": 4.397724679155182,
      "grad_norm": 0.05061586967606174,
      "learning_rate": 9.289960097105199e-06,
      "loss": 0.8309,
      "step": 250490
    },
    {
      "epoch": 4.397900244035183,
      "grad_norm": 0.05061706467583482,
      "learning_rate": 9.285763658999059e-06,
      "loss": 0.8323,
      "step": 250500
    },
    {
      "epoch": 4.398075808915185,
      "grad_norm": 0.046453744345627726,
      "learning_rate": 9.281568382929848e-06,
      "loss": 0.8412,
      "step": 250510
    },
    {
      "epoch": 4.398251373795186,
      "grad_norm": 0.06982495410436416,
      "learning_rate": 9.277374268950765e-06,
      "loss": 0.8408,
      "step": 250520
    },
    {
      "epoch": 4.398426938675187,
      "grad_norm": 0.053875969663511464,
      "learning_rate": 9.273181317114898e-06,
      "loss": 0.8422,
      "step": 250530
    },
    {
      "epoch": 4.398602503555189,
      "grad_norm": 0.0452138741024222,
      "learning_rate": 9.268989527475433e-06,
      "loss": 0.8435,
      "step": 250540
    },
    {
      "epoch": 4.39877806843519,
      "grad_norm": 0.06655546253937995,
      "learning_rate": 9.264798900085436e-06,
      "loss": 0.8383,
      "step": 250550
    },
    {
      "epoch": 4.398953633315192,
      "grad_norm": 0.04115028634110336,
      "learning_rate": 9.260609434998016e-06,
      "loss": 0.842,
      "step": 250560
    },
    {
      "epoch": 4.399129198195193,
      "grad_norm": 0.05164531486345813,
      "learning_rate": 9.25642113226627e-06,
      "loss": 0.8418,
      "step": 250570
    },
    {
      "epoch": 4.399304763075194,
      "grad_norm": 0.08134644177554483,
      "learning_rate": 9.252233991943235e-06,
      "loss": 0.8362,
      "step": 250580
    },
    {
      "epoch": 4.399480327955196,
      "grad_norm": 0.043615399643811524,
      "learning_rate": 9.248048014081984e-06,
      "loss": 0.8429,
      "step": 250590
    },
    {
      "epoch": 4.399655892835197,
      "grad_norm": 0.04253542569649958,
      "learning_rate": 9.243863198735558e-06,
      "loss": 0.8456,
      "step": 250600
    },
    {
      "epoch": 4.399831457715199,
      "grad_norm": 0.05081785709755138,
      "learning_rate": 9.239679545956965e-06,
      "loss": 0.8466,
      "step": 250610
    },
    {
      "epoch": 4.4000070225952,
      "grad_norm": 0.047447318350175266,
      "learning_rate": 9.235497055799219e-06,
      "loss": 0.841,
      "step": 250620
    },
    {
      "epoch": 4.400182587475202,
      "grad_norm": 0.04255660607970881,
      "learning_rate": 9.23131572831529e-06,
      "loss": 0.8325,
      "step": 250630
    },
    {
      "epoch": 4.400358152355203,
      "grad_norm": 0.0693748193590077,
      "learning_rate": 9.227135563558209e-06,
      "loss": 0.8333,
      "step": 250640
    },
    {
      "epoch": 4.400533717235204,
      "grad_norm": 0.05028377627719966,
      "learning_rate": 9.22295656158089e-06,
      "loss": 0.8406,
      "step": 250650
    },
    {
      "epoch": 4.400709282115206,
      "grad_norm": 0.11458139517851433,
      "learning_rate": 9.218778722436309e-06,
      "loss": 0.8327,
      "step": 250660
    },
    {
      "epoch": 4.400884846995207,
      "grad_norm": 0.07552394374709752,
      "learning_rate": 9.214602046177397e-06,
      "loss": 0.8338,
      "step": 250670
    },
    {
      "epoch": 4.401060411875209,
      "grad_norm": 0.05522796081784795,
      "learning_rate": 9.210426532857062e-06,
      "loss": 0.8406,
      "step": 250680
    },
    {
      "epoch": 4.40123597675521,
      "grad_norm": 0.040635968429315605,
      "learning_rate": 9.206252182528218e-06,
      "loss": 0.8304,
      "step": 250690
    },
    {
      "epoch": 4.401411541635211,
      "grad_norm": 0.06256419799613347,
      "learning_rate": 9.202078995243726e-06,
      "loss": 0.835,
      "step": 250700
    },
    {
      "epoch": 4.401587106515213,
      "grad_norm": 0.05401589264608262,
      "learning_rate": 9.197906971056511e-06,
      "loss": 0.8379,
      "step": 250710
    },
    {
      "epoch": 4.401762671395214,
      "grad_norm": 0.052940574630773354,
      "learning_rate": 9.193736110019405e-06,
      "loss": 0.843,
      "step": 250720
    },
    {
      "epoch": 4.401938236275216,
      "grad_norm": 0.049280352618761236,
      "learning_rate": 9.18956641218524e-06,
      "loss": 0.8355,
      "step": 250730
    },
    {
      "epoch": 4.402113801155217,
      "grad_norm": 0.04737286516987363,
      "learning_rate": 9.1853978776069e-06,
      "loss": 0.8361,
      "step": 250740
    },
    {
      "epoch": 4.402289366035219,
      "grad_norm": 0.05228560508200319,
      "learning_rate": 9.181230506337129e-06,
      "loss": 0.8474,
      "step": 250750
    },
    {
      "epoch": 4.40246493091522,
      "grad_norm": 0.0570075127177841,
      "learning_rate": 9.177064298428786e-06,
      "loss": 0.8426,
      "step": 250760
    },
    {
      "epoch": 4.402640495795221,
      "grad_norm": 0.0533948452880883,
      "learning_rate": 9.172899253934635e-06,
      "loss": 0.8376,
      "step": 250770
    },
    {
      "epoch": 4.402816060675223,
      "grad_norm": 0.06281344388471158,
      "learning_rate": 9.168735372907439e-06,
      "loss": 0.831,
      "step": 250780
    },
    {
      "epoch": 4.402991625555224,
      "grad_norm": 0.047499250839111974,
      "learning_rate": 9.164572655399987e-06,
      "loss": 0.8305,
      "step": 250790
    },
    {
      "epoch": 4.403167190435226,
      "grad_norm": 0.05628907634073075,
      "learning_rate": 9.160411101464955e-06,
      "loss": 0.8354,
      "step": 250800
    },
    {
      "epoch": 4.403342755315227,
      "grad_norm": 0.05270616298185229,
      "learning_rate": 9.156250711155153e-06,
      "loss": 0.8352,
      "step": 250810
    },
    {
      "epoch": 4.403518320195229,
      "grad_norm": 0.05155737909221415,
      "learning_rate": 9.152091484523248e-06,
      "loss": 0.8466,
      "step": 250820
    },
    {
      "epoch": 4.40369388507523,
      "grad_norm": 0.04803206858922459,
      "learning_rate": 9.147933421621938e-06,
      "loss": 0.8458,
      "step": 250830
    },
    {
      "epoch": 4.403869449955231,
      "grad_norm": 0.045403594853264354,
      "learning_rate": 9.143776522503934e-06,
      "loss": 0.841,
      "step": 250840
    },
    {
      "epoch": 4.404045014835233,
      "grad_norm": 0.05463179467931895,
      "learning_rate": 9.139620787221865e-06,
      "loss": 0.8486,
      "step": 250850
    },
    {
      "epoch": 4.404220579715234,
      "grad_norm": 0.05190133594890208,
      "learning_rate": 9.135466215828428e-06,
      "loss": 0.8407,
      "step": 250860
    },
    {
      "epoch": 4.4043961445952355,
      "grad_norm": 0.06213622440896758,
      "learning_rate": 9.131312808376205e-06,
      "loss": 0.8341,
      "step": 250870
    },
    {
      "epoch": 4.404571709475237,
      "grad_norm": 0.051869308386611065,
      "learning_rate": 9.127160564917884e-06,
      "loss": 0.8404,
      "step": 250880
    },
    {
      "epoch": 4.404747274355238,
      "grad_norm": 0.054987315083329776,
      "learning_rate": 9.123009485506026e-06,
      "loss": 0.8307,
      "step": 250890
    },
    {
      "epoch": 4.4049228392352395,
      "grad_norm": 0.06728792733659232,
      "learning_rate": 9.11885957019326e-06,
      "loss": 0.8461,
      "step": 250900
    },
    {
      "epoch": 4.405098404115241,
      "grad_norm": 0.05266077721785263,
      "learning_rate": 9.114710819032157e-06,
      "loss": 0.8456,
      "step": 250910
    },
    {
      "epoch": 4.4052739689952425,
      "grad_norm": 0.05246813984553654,
      "learning_rate": 9.110563232075273e-06,
      "loss": 0.8384,
      "step": 250920
    },
    {
      "epoch": 4.4054495338752435,
      "grad_norm": 0.0620600457761443,
      "learning_rate": 9.106416809375148e-06,
      "loss": 0.841,
      "step": 250930
    },
    {
      "epoch": 4.4056250987552446,
      "grad_norm": 0.049865388690591456,
      "learning_rate": 9.102271550984365e-06,
      "loss": 0.8312,
      "step": 250940
    },
    {
      "epoch": 4.4058006636352465,
      "grad_norm": 0.06352329589441828,
      "learning_rate": 9.098127456955382e-06,
      "loss": 0.8402,
      "step": 250950
    },
    {
      "epoch": 4.4059762285152475,
      "grad_norm": 0.048500845964938574,
      "learning_rate": 9.093984527340762e-06,
      "loss": 0.8379,
      "step": 250960
    },
    {
      "epoch": 4.406151793395249,
      "grad_norm": 0.05523597085967654,
      "learning_rate": 9.089842762192944e-06,
      "loss": 0.8401,
      "step": 250970
    },
    {
      "epoch": 4.4063273582752505,
      "grad_norm": 0.060733640701600325,
      "learning_rate": 9.085702161564452e-06,
      "loss": 0.8289,
      "step": 250980
    },
    {
      "epoch": 4.406502923155252,
      "grad_norm": 0.06507300974761854,
      "learning_rate": 9.081562725507728e-06,
      "loss": 0.8414,
      "step": 250990
    },
    {
      "epoch": 4.406678488035253,
      "grad_norm": 0.05573289072215501,
      "learning_rate": 9.07742445407521e-06,
      "loss": 0.8353,
      "step": 251000
    },
    {
      "epoch": 4.4068540529152544,
      "grad_norm": 0.06860637408053208,
      "learning_rate": 9.07328734731936e-06,
      "loss": 0.8414,
      "step": 251010
    },
    {
      "epoch": 4.407029617795256,
      "grad_norm": 0.06414352137800788,
      "learning_rate": 9.069151405292559e-06,
      "loss": 0.8347,
      "step": 251020
    },
    {
      "epoch": 4.407205182675257,
      "grad_norm": 0.060075595866169264,
      "learning_rate": 9.065016628047249e-06,
      "loss": 0.8362,
      "step": 251030
    },
    {
      "epoch": 4.407380747555259,
      "grad_norm": 0.0556313919266062,
      "learning_rate": 9.06088301563576e-06,
      "loss": 0.8319,
      "step": 251040
    },
    {
      "epoch": 4.40755631243526,
      "grad_norm": 0.05303463859324361,
      "learning_rate": 9.056750568110538e-06,
      "loss": 0.8356,
      "step": 251050
    },
    {
      "epoch": 4.407731877315262,
      "grad_norm": 0.04792812865201051,
      "learning_rate": 9.052619285523894e-06,
      "loss": 0.8381,
      "step": 251060
    },
    {
      "epoch": 4.407907442195263,
      "grad_norm": 0.05271272341608069,
      "learning_rate": 9.048489167928178e-06,
      "loss": 0.8457,
      "step": 251070
    },
    {
      "epoch": 4.408083007075264,
      "grad_norm": 0.0600123883195301,
      "learning_rate": 9.044360215375752e-06,
      "loss": 0.8384,
      "step": 251080
    },
    {
      "epoch": 4.408258571955266,
      "grad_norm": 0.05332563534039346,
      "learning_rate": 9.04023242791888e-06,
      "loss": 0.8374,
      "step": 251090
    },
    {
      "epoch": 4.408434136835267,
      "grad_norm": 0.04441127987842643,
      "learning_rate": 9.036105805609887e-06,
      "loss": 0.8346,
      "step": 251100
    },
    {
      "epoch": 4.408609701715269,
      "grad_norm": 0.05377857330478821,
      "learning_rate": 9.031980348501084e-06,
      "loss": 0.8377,
      "step": 251110
    },
    {
      "epoch": 4.40878526659527,
      "grad_norm": 0.04730254703811243,
      "learning_rate": 9.027856056644693e-06,
      "loss": 0.8438,
      "step": 251120
    },
    {
      "epoch": 4.408960831475271,
      "grad_norm": 0.045141515929253986,
      "learning_rate": 9.023732930093016e-06,
      "loss": 0.8412,
      "step": 251130
    },
    {
      "epoch": 4.409136396355273,
      "grad_norm": 0.054100859724552725,
      "learning_rate": 9.019610968898228e-06,
      "loss": 0.8356,
      "step": 251140
    },
    {
      "epoch": 4.409311961235274,
      "grad_norm": 0.04987230710490107,
      "learning_rate": 9.01549017311263e-06,
      "loss": 0.8345,
      "step": 251150
    },
    {
      "epoch": 4.409487526115276,
      "grad_norm": 0.0448410485306441,
      "learning_rate": 9.01137054278838e-06,
      "loss": 0.8345,
      "step": 251160
    },
    {
      "epoch": 4.409663090995277,
      "grad_norm": 0.05833462427810838,
      "learning_rate": 9.007252077977704e-06,
      "loss": 0.8455,
      "step": 251170
    },
    {
      "epoch": 4.409838655875279,
      "grad_norm": 0.05321595583403733,
      "learning_rate": 9.003134778732794e-06,
      "loss": 0.8485,
      "step": 251180
    },
    {
      "epoch": 4.41001422075528,
      "grad_norm": 0.04978655192413645,
      "learning_rate": 8.999018645105772e-06,
      "loss": 0.8494,
      "step": 251190
    },
    {
      "epoch": 4.410189785635281,
      "grad_norm": 0.052656998099387405,
      "learning_rate": 8.994903677148835e-06,
      "loss": 0.8431,
      "step": 251200
    },
    {
      "epoch": 4.410365350515283,
      "grad_norm": 0.050370540041383915,
      "learning_rate": 8.990789874914064e-06,
      "loss": 0.835,
      "step": 251210
    },
    {
      "epoch": 4.410540915395284,
      "grad_norm": 0.051769176676633145,
      "learning_rate": 8.986677238453658e-06,
      "loss": 0.8407,
      "step": 251220
    },
    {
      "epoch": 4.410716480275286,
      "grad_norm": 0.04931536097796973,
      "learning_rate": 8.98256576781967e-06,
      "loss": 0.8433,
      "step": 251230
    },
    {
      "epoch": 4.410892045155287,
      "grad_norm": 0.054548427385838406,
      "learning_rate": 8.9784554630642e-06,
      "loss": 0.8364,
      "step": 251240
    },
    {
      "epoch": 4.411067610035288,
      "grad_norm": 0.03875270932224459,
      "learning_rate": 8.974346324239366e-06,
      "loss": 0.8384,
      "step": 251250
    },
    {
      "epoch": 4.41124317491529,
      "grad_norm": 0.051216818684988986,
      "learning_rate": 8.970238351397173e-06,
      "loss": 0.829,
      "step": 251260
    },
    {
      "epoch": 4.411418739795291,
      "grad_norm": 0.06593571959538541,
      "learning_rate": 8.966131544589698e-06,
      "loss": 0.8292,
      "step": 251270
    },
    {
      "epoch": 4.411594304675293,
      "grad_norm": 0.04498355306963626,
      "learning_rate": 8.962025903868995e-06,
      "loss": 0.846,
      "step": 251280
    },
    {
      "epoch": 4.411769869555294,
      "grad_norm": 0.04902382023259568,
      "learning_rate": 8.957921429287045e-06,
      "loss": 0.8403,
      "step": 251290
    },
    {
      "epoch": 4.411945434435296,
      "grad_norm": 0.05746897745920257,
      "learning_rate": 8.953818120895887e-06,
      "loss": 0.8292,
      "step": 251300
    },
    {
      "epoch": 4.412120999315297,
      "grad_norm": 0.04934598784254338,
      "learning_rate": 8.949715978747458e-06,
      "loss": 0.8339,
      "step": 251310
    },
    {
      "epoch": 4.412296564195298,
      "grad_norm": 0.0517498659136083,
      "learning_rate": 8.945615002893812e-06,
      "loss": 0.835,
      "step": 251320
    },
    {
      "epoch": 4.4124721290753,
      "grad_norm": 0.05961630280019871,
      "learning_rate": 8.941515193386834e-06,
      "loss": 0.8443,
      "step": 251330
    },
    {
      "epoch": 4.412647693955301,
      "grad_norm": 0.04227121323368595,
      "learning_rate": 8.937416550278519e-06,
      "loss": 0.8385,
      "step": 251340
    },
    {
      "epoch": 4.412823258835303,
      "grad_norm": 0.05287794267871087,
      "learning_rate": 8.933319073620783e-06,
      "loss": 0.8359,
      "step": 251350
    },
    {
      "epoch": 4.412998823715304,
      "grad_norm": 0.048533852312024964,
      "learning_rate": 8.92922276346552e-06,
      "loss": 0.839,
      "step": 251360
    },
    {
      "epoch": 4.413174388595305,
      "grad_norm": 0.0552187263178662,
      "learning_rate": 8.925127619864673e-06,
      "loss": 0.8271,
      "step": 251370
    },
    {
      "epoch": 4.413349953475307,
      "grad_norm": 0.044268045270449696,
      "learning_rate": 8.921033642870075e-06,
      "loss": 0.8446,
      "step": 251380
    },
    {
      "epoch": 4.413525518355308,
      "grad_norm": 0.07310894368391559,
      "learning_rate": 8.916940832533651e-06,
      "loss": 0.8375,
      "step": 251390
    },
    {
      "epoch": 4.41370108323531,
      "grad_norm": 0.04560157202071827,
      "learning_rate": 8.912849188907227e-06,
      "loss": 0.8375,
      "step": 251400
    },
    {
      "epoch": 4.413876648115311,
      "grad_norm": 0.04414807882275178,
      "learning_rate": 8.908758712042653e-06,
      "loss": 0.842,
      "step": 251410
    },
    {
      "epoch": 4.414052212995313,
      "grad_norm": 0.04696289133411016,
      "learning_rate": 8.904669401991767e-06,
      "loss": 0.8346,
      "step": 251420
    },
    {
      "epoch": 4.414227777875314,
      "grad_norm": 0.05235938758337804,
      "learning_rate": 8.900581258806359e-06,
      "loss": 0.8347,
      "step": 251430
    },
    {
      "epoch": 4.414403342755315,
      "grad_norm": 0.05606102366884861,
      "learning_rate": 8.896494282538242e-06,
      "loss": 0.8369,
      "step": 251440
    },
    {
      "epoch": 4.414578907635317,
      "grad_norm": 0.04597399400516327,
      "learning_rate": 8.892408473239183e-06,
      "loss": 0.8395,
      "step": 251450
    },
    {
      "epoch": 4.414754472515318,
      "grad_norm": 0.06160858525140655,
      "learning_rate": 8.888323830960965e-06,
      "loss": 0.8398,
      "step": 251460
    },
    {
      "epoch": 4.41493003739532,
      "grad_norm": 0.04411418834233668,
      "learning_rate": 8.884240355755356e-06,
      "loss": 0.8332,
      "step": 251470
    },
    {
      "epoch": 4.415105602275321,
      "grad_norm": 0.05888073664369715,
      "learning_rate": 8.880158047674032e-06,
      "loss": 0.8376,
      "step": 251480
    },
    {
      "epoch": 4.415281167155323,
      "grad_norm": 0.04864807574005934,
      "learning_rate": 8.8760769067688e-06,
      "loss": 0.8357,
      "step": 251490
    },
    {
      "epoch": 4.415456732035324,
      "grad_norm": 0.05317471315552484,
      "learning_rate": 8.871996933091315e-06,
      "loss": 0.8384,
      "step": 251500
    },
    {
      "epoch": 4.415632296915325,
      "grad_norm": 0.051397538733654255,
      "learning_rate": 8.86791812669328e-06,
      "loss": 0.837,
      "step": 251510
    },
    {
      "epoch": 4.415807861795327,
      "grad_norm": 0.054892923631296234,
      "learning_rate": 8.863840487626401e-06,
      "loss": 0.8351,
      "step": 251520
    },
    {
      "epoch": 4.415983426675328,
      "grad_norm": 0.08783052872432914,
      "learning_rate": 8.859764015942301e-06,
      "loss": 0.8376,
      "step": 251530
    },
    {
      "epoch": 4.41615899155533,
      "grad_norm": 0.044715855441787515,
      "learning_rate": 8.855688711692662e-06,
      "loss": 0.8351,
      "step": 251540
    },
    {
      "epoch": 4.416334556435331,
      "grad_norm": 0.046102603167139315,
      "learning_rate": 8.851614574929073e-06,
      "loss": 0.843,
      "step": 251550
    },
    {
      "epoch": 4.416510121315332,
      "grad_norm": 0.04563031417203102,
      "learning_rate": 8.84754160570323e-06,
      "loss": 0.8452,
      "step": 251560
    },
    {
      "epoch": 4.416685686195334,
      "grad_norm": 0.06962421442073607,
      "learning_rate": 8.84346980406667e-06,
      "loss": 0.8426,
      "step": 251570
    },
    {
      "epoch": 4.416861251075335,
      "grad_norm": 0.05444024408337803,
      "learning_rate": 8.839399170071017e-06,
      "loss": 0.8405,
      "step": 251580
    },
    {
      "epoch": 4.417036815955337,
      "grad_norm": 0.05944259066278874,
      "learning_rate": 8.835329703767852e-06,
      "loss": 0.8359,
      "step": 251590
    },
    {
      "epoch": 4.417212380835338,
      "grad_norm": 0.06292754162256531,
      "learning_rate": 8.831261405208706e-06,
      "loss": 0.8379,
      "step": 251600
    },
    {
      "epoch": 4.417387945715339,
      "grad_norm": 0.055236424624510706,
      "learning_rate": 8.827194274445157e-06,
      "loss": 0.8269,
      "step": 251610
    },
    {
      "epoch": 4.417563510595341,
      "grad_norm": 0.04324221345194529,
      "learning_rate": 8.823128311528703e-06,
      "loss": 0.8397,
      "step": 251620
    },
    {
      "epoch": 4.417739075475342,
      "grad_norm": 0.05139592973938702,
      "learning_rate": 8.819063516510886e-06,
      "loss": 0.8393,
      "step": 251630
    },
    {
      "epoch": 4.4179146403553435,
      "grad_norm": 0.051553792232305565,
      "learning_rate": 8.814999889443218e-06,
      "loss": 0.8414,
      "step": 251640
    },
    {
      "epoch": 4.418090205235345,
      "grad_norm": 0.050881031404733254,
      "learning_rate": 8.81093743037713e-06,
      "loss": 0.8381,
      "step": 251650
    },
    {
      "epoch": 4.4182657701153465,
      "grad_norm": 0.05680941677627339,
      "learning_rate": 8.80687613936417e-06,
      "loss": 0.8408,
      "step": 251660
    },
    {
      "epoch": 4.4184413349953475,
      "grad_norm": 0.05530049609087751,
      "learning_rate": 8.802816016455735e-06,
      "loss": 0.8349,
      "step": 251670
    },
    {
      "epoch": 4.418616899875349,
      "grad_norm": 0.05517278942549323,
      "learning_rate": 8.7987570617033e-06,
      "loss": 0.8358,
      "step": 251680
    },
    {
      "epoch": 4.4187924647553505,
      "grad_norm": 0.05797983664105514,
      "learning_rate": 8.794699275158292e-06,
      "loss": 0.8372,
      "step": 251690
    },
    {
      "epoch": 4.4189680296353515,
      "grad_norm": 0.04401031852165657,
      "learning_rate": 8.790642656872094e-06,
      "loss": 0.834,
      "step": 251700
    },
    {
      "epoch": 4.419143594515353,
      "grad_norm": 0.055224980443944374,
      "learning_rate": 8.786587206896152e-06,
      "loss": 0.8413,
      "step": 251710
    },
    {
      "epoch": 4.4193191593953545,
      "grad_norm": 0.044958952063427246,
      "learning_rate": 8.782532925281787e-06,
      "loss": 0.8404,
      "step": 251720
    },
    {
      "epoch": 4.419494724275356,
      "grad_norm": 0.045163662470338525,
      "learning_rate": 8.778479812080438e-06,
      "loss": 0.8332,
      "step": 251730
    },
    {
      "epoch": 4.419670289155357,
      "grad_norm": 0.05190870029321653,
      "learning_rate": 8.774427867343404e-06,
      "loss": 0.8407,
      "step": 251740
    },
    {
      "epoch": 4.4198458540353585,
      "grad_norm": 0.04942702603536956,
      "learning_rate": 8.770377091122036e-06,
      "loss": 0.8448,
      "step": 251750
    },
    {
      "epoch": 4.42002141891536,
      "grad_norm": 0.04709180780120835,
      "learning_rate": 8.766327483467695e-06,
      "loss": 0.8331,
      "step": 251760
    },
    {
      "epoch": 4.420196983795361,
      "grad_norm": 0.0677516297263376,
      "learning_rate": 8.762279044431645e-06,
      "loss": 0.8328,
      "step": 251770
    },
    {
      "epoch": 4.420372548675363,
      "grad_norm": 0.04981151470405238,
      "learning_rate": 8.758231774065204e-06,
      "loss": 0.8346,
      "step": 251780
    },
    {
      "epoch": 4.420548113555364,
      "grad_norm": 0.05826909785983269,
      "learning_rate": 8.754185672419632e-06,
      "loss": 0.8398,
      "step": 251790
    },
    {
      "epoch": 4.420723678435365,
      "grad_norm": 0.06393251374575598,
      "learning_rate": 8.750140739546217e-06,
      "loss": 0.836,
      "step": 251800
    },
    {
      "epoch": 4.420899243315367,
      "grad_norm": 0.058944257329536495,
      "learning_rate": 8.746096975496198e-06,
      "loss": 0.8328,
      "step": 251810
    },
    {
      "epoch": 4.421074808195368,
      "grad_norm": 0.04601100366134911,
      "learning_rate": 8.742054380320785e-06,
      "loss": 0.8409,
      "step": 251820
    },
    {
      "epoch": 4.42125037307537,
      "grad_norm": 0.04192501850107229,
      "learning_rate": 8.738012954071262e-06,
      "loss": 0.8385,
      "step": 251830
    },
    {
      "epoch": 4.421425937955371,
      "grad_norm": 0.05624022573971229,
      "learning_rate": 8.733972696798772e-06,
      "loss": 0.8447,
      "step": 251840
    },
    {
      "epoch": 4.421601502835372,
      "grad_norm": 0.0475725121826854,
      "learning_rate": 8.729933608554544e-06,
      "loss": 0.8332,
      "step": 251850
    },
    {
      "epoch": 4.421777067715374,
      "grad_norm": 0.05008280076692091,
      "learning_rate": 8.72589568938974e-06,
      "loss": 0.8405,
      "step": 251860
    },
    {
      "epoch": 4.421952632595375,
      "grad_norm": 0.08583636041742021,
      "learning_rate": 8.721858939355517e-06,
      "loss": 0.8413,
      "step": 251870
    },
    {
      "epoch": 4.422128197475377,
      "grad_norm": 0.04865840772970111,
      "learning_rate": 8.717823358503038e-06,
      "loss": 0.8334,
      "step": 251880
    },
    {
      "epoch": 4.422303762355378,
      "grad_norm": 0.06952862057464201,
      "learning_rate": 8.713788946883388e-06,
      "loss": 0.8386,
      "step": 251890
    },
    {
      "epoch": 4.42247932723538,
      "grad_norm": 0.05331566804049456,
      "learning_rate": 8.709755704547756e-06,
      "loss": 0.8463,
      "step": 251900
    },
    {
      "epoch": 4.422654892115381,
      "grad_norm": 0.052312412136635685,
      "learning_rate": 8.705723631547186e-06,
      "loss": 0.8352,
      "step": 251910
    },
    {
      "epoch": 4.422830456995382,
      "grad_norm": 0.05521291363013707,
      "learning_rate": 8.701692727932791e-06,
      "loss": 0.833,
      "step": 251920
    },
    {
      "epoch": 4.423006021875384,
      "grad_norm": 0.04497438881243429,
      "learning_rate": 8.697662993755656e-06,
      "loss": 0.8387,
      "step": 251930
    },
    {
      "epoch": 4.423181586755385,
      "grad_norm": 0.04518610913393563,
      "learning_rate": 8.693634429066797e-06,
      "loss": 0.8316,
      "step": 251940
    },
    {
      "epoch": 4.423357151635387,
      "grad_norm": 0.050322263249580064,
      "learning_rate": 8.689607033917301e-06,
      "loss": 0.8401,
      "step": 251950
    },
    {
      "epoch": 4.423532716515388,
      "grad_norm": 0.06030631477987046,
      "learning_rate": 8.685580808358145e-06,
      "loss": 0.8312,
      "step": 251960
    },
    {
      "epoch": 4.42370828139539,
      "grad_norm": 0.059231921653843334,
      "learning_rate": 8.68155575244039e-06,
      "loss": 0.8434,
      "step": 251970
    },
    {
      "epoch": 4.423883846275391,
      "grad_norm": 0.054596649662896235,
      "learning_rate": 8.677531866215027e-06,
      "loss": 0.8327,
      "step": 251980
    },
    {
      "epoch": 4.424059411155392,
      "grad_norm": 0.06512656208369046,
      "learning_rate": 8.673509149733003e-06,
      "loss": 0.8367,
      "step": 251990
    },
    {
      "epoch": 4.424234976035394,
      "grad_norm": 0.053701944443584046,
      "learning_rate": 8.669487603045329e-06,
      "loss": 0.8389,
      "step": 252000
    },
    {
      "epoch": 4.424410540915395,
      "grad_norm": 0.0511458459626452,
      "learning_rate": 8.665467226202937e-06,
      "loss": 0.8393,
      "step": 252010
    },
    {
      "epoch": 4.424586105795397,
      "grad_norm": 0.059184262206251136,
      "learning_rate": 8.661448019256772e-06,
      "loss": 0.8383,
      "step": 252020
    },
    {
      "epoch": 4.424761670675398,
      "grad_norm": 0.06487204329370354,
      "learning_rate": 8.657429982257767e-06,
      "loss": 0.836,
      "step": 252030
    },
    {
      "epoch": 4.424937235555399,
      "grad_norm": 0.0597499763188485,
      "learning_rate": 8.653413115256811e-06,
      "loss": 0.839,
      "step": 252040
    },
    {
      "epoch": 4.425112800435401,
      "grad_norm": 0.06848915176703438,
      "learning_rate": 8.649397418304816e-06,
      "loss": 0.8399,
      "step": 252050
    },
    {
      "epoch": 4.425288365315402,
      "grad_norm": 0.046773243112396876,
      "learning_rate": 8.64538289145263e-06,
      "loss": 0.8411,
      "step": 252060
    },
    {
      "epoch": 4.425463930195404,
      "grad_norm": 0.06774700383312032,
      "learning_rate": 8.641369534751182e-06,
      "loss": 0.8299,
      "step": 252070
    },
    {
      "epoch": 4.425639495075405,
      "grad_norm": 0.04456170265039534,
      "learning_rate": 8.637357348251249e-06,
      "loss": 0.8413,
      "step": 252080
    },
    {
      "epoch": 4.425815059955407,
      "grad_norm": 0.043731182127319454,
      "learning_rate": 8.633346332003717e-06,
      "loss": 0.8344,
      "step": 252090
    },
    {
      "epoch": 4.425990624835408,
      "grad_norm": 0.04881236622933115,
      "learning_rate": 8.629336486059392e-06,
      "loss": 0.8385,
      "step": 252100
    },
    {
      "epoch": 4.426166189715409,
      "grad_norm": 0.12687663890704864,
      "learning_rate": 8.62532781046907e-06,
      "loss": 0.8359,
      "step": 252110
    },
    {
      "epoch": 4.426341754595411,
      "grad_norm": 0.0539627998023148,
      "learning_rate": 8.621320305283556e-06,
      "loss": 0.8499,
      "step": 252120
    },
    {
      "epoch": 4.426517319475412,
      "grad_norm": 0.09576170241984366,
      "learning_rate": 8.617313970553604e-06,
      "loss": 0.834,
      "step": 252130
    },
    {
      "epoch": 4.426692884355414,
      "grad_norm": 0.062163497085160196,
      "learning_rate": 8.613308806329998e-06,
      "loss": 0.8383,
      "step": 252140
    },
    {
      "epoch": 4.426868449235415,
      "grad_norm": 0.05396401705053233,
      "learning_rate": 8.609304812663477e-06,
      "loss": 0.8371,
      "step": 252150
    },
    {
      "epoch": 4.427044014115417,
      "grad_norm": 0.06247378073206681,
      "learning_rate": 8.605301989604742e-06,
      "loss": 0.8378,
      "step": 252160
    },
    {
      "epoch": 4.427219578995418,
      "grad_norm": 0.047742519129201924,
      "learning_rate": 8.601300337204574e-06,
      "loss": 0.8415,
      "step": 252170
    },
    {
      "epoch": 4.427395143875419,
      "grad_norm": 0.06004412002888419,
      "learning_rate": 8.597299855513621e-06,
      "loss": 0.8397,
      "step": 252180
    },
    {
      "epoch": 4.427570708755421,
      "grad_norm": 0.05255523218735327,
      "learning_rate": 8.59330054458259e-06,
      "loss": 0.8406,
      "step": 252190
    },
    {
      "epoch": 4.427746273635422,
      "grad_norm": 0.05233183734842703,
      "learning_rate": 8.589302404462174e-06,
      "loss": 0.8413,
      "step": 252200
    },
    {
      "epoch": 4.427921838515424,
      "grad_norm": 0.05975808951973075,
      "learning_rate": 8.585305435202976e-06,
      "loss": 0.8421,
      "step": 252210
    },
    {
      "epoch": 4.428097403395425,
      "grad_norm": 0.06477441588830757,
      "learning_rate": 8.581309636855692e-06,
      "loss": 0.8297,
      "step": 252220
    },
    {
      "epoch": 4.428272968275426,
      "grad_norm": 0.04873000160133851,
      "learning_rate": 8.577315009470902e-06,
      "loss": 0.8377,
      "step": 252230
    },
    {
      "epoch": 4.428448533155428,
      "grad_norm": 0.052631652055762895,
      "learning_rate": 8.573321553099264e-06,
      "loss": 0.8377,
      "step": 252240
    },
    {
      "epoch": 4.428624098035429,
      "grad_norm": 0.04705929305807769,
      "learning_rate": 8.569329267791342e-06,
      "loss": 0.8431,
      "step": 252250
    },
    {
      "epoch": 4.428799662915431,
      "grad_norm": 0.049140232847169785,
      "learning_rate": 8.565338153597736e-06,
      "loss": 0.8301,
      "step": 252260
    },
    {
      "epoch": 4.428975227795432,
      "grad_norm": 0.04476562374844736,
      "learning_rate": 8.561348210569026e-06,
      "loss": 0.8348,
      "step": 252270
    },
    {
      "epoch": 4.429150792675433,
      "grad_norm": 0.05126280704887249,
      "learning_rate": 8.557359438755733e-06,
      "loss": 0.8332,
      "step": 252280
    },
    {
      "epoch": 4.429326357555435,
      "grad_norm": 0.058864152385961724,
      "learning_rate": 8.553371838208422e-06,
      "loss": 0.8399,
      "step": 252290
    },
    {
      "epoch": 4.429501922435436,
      "grad_norm": 0.05082660345233886,
      "learning_rate": 8.5493854089776e-06,
      "loss": 0.8386,
      "step": 252300
    },
    {
      "epoch": 4.429677487315438,
      "grad_norm": 0.0569126718822101,
      "learning_rate": 8.545400151113783e-06,
      "loss": 0.8343,
      "step": 252310
    },
    {
      "epoch": 4.429853052195439,
      "grad_norm": 0.0484925268089862,
      "learning_rate": 8.541416064667474e-06,
      "loss": 0.8375,
      "step": 252320
    },
    {
      "epoch": 4.430028617075441,
      "grad_norm": 0.05972126122178655,
      "learning_rate": 8.537433149689125e-06,
      "loss": 0.84,
      "step": 252330
    },
    {
      "epoch": 4.430204181955442,
      "grad_norm": 0.05213718705264309,
      "learning_rate": 8.533451406229237e-06,
      "loss": 0.83,
      "step": 252340
    },
    {
      "epoch": 4.430379746835443,
      "grad_norm": 0.0777019268910168,
      "learning_rate": 8.529470834338229e-06,
      "loss": 0.8426,
      "step": 252350
    },
    {
      "epoch": 4.430555311715445,
      "grad_norm": 0.05115141541965036,
      "learning_rate": 8.525491434066544e-06,
      "loss": 0.8367,
      "step": 252360
    },
    {
      "epoch": 4.430730876595446,
      "grad_norm": 0.04164235234598252,
      "learning_rate": 8.521513205464623e-06,
      "loss": 0.8447,
      "step": 252370
    },
    {
      "epoch": 4.4309064414754475,
      "grad_norm": 0.052694905724934724,
      "learning_rate": 8.517536148582845e-06,
      "loss": 0.8326,
      "step": 252380
    },
    {
      "epoch": 4.431082006355449,
      "grad_norm": 0.06258813541203853,
      "learning_rate": 8.513560263471612e-06,
      "loss": 0.8326,
      "step": 252390
    },
    {
      "epoch": 4.4312575712354505,
      "grad_norm": 0.05960451788082379,
      "learning_rate": 8.509585550181265e-06,
      "loss": 0.8407,
      "step": 252400
    },
    {
      "epoch": 4.4314331361154515,
      "grad_norm": 0.052791810737246474,
      "learning_rate": 8.505612008762234e-06,
      "loss": 0.8322,
      "step": 252410
    },
    {
      "epoch": 4.431608700995453,
      "grad_norm": 0.0430270024490982,
      "learning_rate": 8.50163963926481e-06,
      "loss": 0.8389,
      "step": 252420
    },
    {
      "epoch": 4.4317842658754545,
      "grad_norm": 0.047253606594707974,
      "learning_rate": 8.497668441739326e-06,
      "loss": 0.8346,
      "step": 252430
    },
    {
      "epoch": 4.4319598307554555,
      "grad_norm": 0.048608924693918254,
      "learning_rate": 8.493698416236143e-06,
      "loss": 0.8409,
      "step": 252440
    },
    {
      "epoch": 4.432135395635457,
      "grad_norm": 0.05598225466948425,
      "learning_rate": 8.489729562805501e-06,
      "loss": 0.8313,
      "step": 252450
    },
    {
      "epoch": 4.4323109605154585,
      "grad_norm": 0.050493214374802205,
      "learning_rate": 8.48576188149774e-06,
      "loss": 0.8406,
      "step": 252460
    },
    {
      "epoch": 4.4324865253954595,
      "grad_norm": 0.04887000404814448,
      "learning_rate": 8.481795372363089e-06,
      "loss": 0.8357,
      "step": 252470
    },
    {
      "epoch": 4.432662090275461,
      "grad_norm": 0.04494159069759292,
      "learning_rate": 8.47783003545182e-06,
      "loss": 0.8438,
      "step": 252480
    },
    {
      "epoch": 4.4328376551554625,
      "grad_norm": 0.050018982083394246,
      "learning_rate": 8.473865870814191e-06,
      "loss": 0.8393,
      "step": 252490
    },
    {
      "epoch": 4.433013220035464,
      "grad_norm": 0.04943171663339999,
      "learning_rate": 8.469902878500383e-06,
      "loss": 0.841,
      "step": 252500
    },
    {
      "epoch": 4.433188784915465,
      "grad_norm": 0.0397991778076104,
      "learning_rate": 8.46594105856067e-06,
      "loss": 0.8415,
      "step": 252510
    },
    {
      "epoch": 4.4333643497954665,
      "grad_norm": 0.05760699132088902,
      "learning_rate": 8.461980411045203e-06,
      "loss": 0.8421,
      "step": 252520
    },
    {
      "epoch": 4.433539914675468,
      "grad_norm": 0.04173019095152896,
      "learning_rate": 8.458020936004174e-06,
      "loss": 0.8461,
      "step": 252530
    },
    {
      "epoch": 4.433715479555469,
      "grad_norm": 0.036691106472172476,
      "learning_rate": 8.454062633487778e-06,
      "loss": 0.8359,
      "step": 252540
    },
    {
      "epoch": 4.433891044435471,
      "grad_norm": 0.05659934985376422,
      "learning_rate": 8.45010550354613e-06,
      "loss": 0.8393,
      "step": 252550
    },
    {
      "epoch": 4.434066609315472,
      "grad_norm": 0.04906409790239427,
      "learning_rate": 8.446149546229387e-06,
      "loss": 0.8389,
      "step": 252560
    },
    {
      "epoch": 4.434242174195474,
      "grad_norm": 0.05264551938433608,
      "learning_rate": 8.442194761587645e-06,
      "loss": 0.8372,
      "step": 252570
    },
    {
      "epoch": 4.434417739075475,
      "grad_norm": 0.05048190062831621,
      "learning_rate": 8.438241149671073e-06,
      "loss": 0.843,
      "step": 252580
    },
    {
      "epoch": 4.434593303955476,
      "grad_norm": 0.05325235102453735,
      "learning_rate": 8.434288710529695e-06,
      "loss": 0.8375,
      "step": 252590
    },
    {
      "epoch": 4.434768868835478,
      "grad_norm": 0.06651956533772455,
      "learning_rate": 8.430337444213623e-06,
      "loss": 0.8366,
      "step": 252600
    },
    {
      "epoch": 4.434944433715479,
      "grad_norm": 0.05015455683368762,
      "learning_rate": 8.426387350772937e-06,
      "loss": 0.8422,
      "step": 252610
    },
    {
      "epoch": 4.435119998595481,
      "grad_norm": 0.040168010768565084,
      "learning_rate": 8.422438430257643e-06,
      "loss": 0.8376,
      "step": 252620
    },
    {
      "epoch": 4.435295563475482,
      "grad_norm": 0.05265180023654487,
      "learning_rate": 8.418490682717817e-06,
      "loss": 0.8424,
      "step": 252630
    },
    {
      "epoch": 4.435471128355484,
      "grad_norm": 0.044897074663661864,
      "learning_rate": 8.414544108203437e-06,
      "loss": 0.8396,
      "step": 252640
    },
    {
      "epoch": 4.435646693235485,
      "grad_norm": 0.0587263931524389,
      "learning_rate": 8.410598706764525e-06,
      "loss": 0.8397,
      "step": 252650
    },
    {
      "epoch": 4.435822258115486,
      "grad_norm": 0.04579971368935492,
      "learning_rate": 8.406654478451102e-06,
      "loss": 0.84,
      "step": 252660
    },
    {
      "epoch": 4.435997822995488,
      "grad_norm": 0.04835866559293575,
      "learning_rate": 8.402711423313062e-06,
      "loss": 0.8286,
      "step": 252670
    },
    {
      "epoch": 4.436173387875489,
      "grad_norm": 0.04829381120191266,
      "learning_rate": 8.398769541400456e-06,
      "loss": 0.8396,
      "step": 252680
    },
    {
      "epoch": 4.436348952755491,
      "grad_norm": 0.062017361389420884,
      "learning_rate": 8.394828832763169e-06,
      "loss": 0.8373,
      "step": 252690
    },
    {
      "epoch": 4.436524517635492,
      "grad_norm": 0.0530129165291228,
      "learning_rate": 8.390889297451157e-06,
      "loss": 0.8407,
      "step": 252700
    },
    {
      "epoch": 4.436700082515493,
      "grad_norm": 0.043426773851002874,
      "learning_rate": 8.386950935514332e-06,
      "loss": 0.8452,
      "step": 252710
    },
    {
      "epoch": 4.436875647395495,
      "grad_norm": 0.05361314660364481,
      "learning_rate": 8.383013747002569e-06,
      "loss": 0.8485,
      "step": 252720
    },
    {
      "epoch": 4.437051212275496,
      "grad_norm": 0.049950750221300404,
      "learning_rate": 8.379077731965796e-06,
      "loss": 0.8386,
      "step": 252730
    },
    {
      "epoch": 4.437226777155498,
      "grad_norm": 0.04755479596816508,
      "learning_rate": 8.37514289045383e-06,
      "loss": 0.8386,
      "step": 252740
    },
    {
      "epoch": 4.437402342035499,
      "grad_norm": 0.0511913280570267,
      "learning_rate": 8.371209222516572e-06,
      "loss": 0.8428,
      "step": 252750
    },
    {
      "epoch": 4.437577906915501,
      "grad_norm": 0.045271877110248285,
      "learning_rate": 8.367276728203848e-06,
      "loss": 0.8423,
      "step": 252760
    },
    {
      "epoch": 4.437753471795502,
      "grad_norm": 0.055855419191929,
      "learning_rate": 8.363345407565467e-06,
      "loss": 0.8326,
      "step": 252770
    },
    {
      "epoch": 4.437929036675503,
      "grad_norm": 0.057117335921622714,
      "learning_rate": 8.359415260651268e-06,
      "loss": 0.8305,
      "step": 252780
    },
    {
      "epoch": 4.438104601555505,
      "grad_norm": 0.06915939177231492,
      "learning_rate": 8.355486287511024e-06,
      "loss": 0.8435,
      "step": 252790
    },
    {
      "epoch": 4.438280166435506,
      "grad_norm": 0.05858125913631465,
      "learning_rate": 8.35155848819454e-06,
      "loss": 0.8374,
      "step": 252800
    },
    {
      "epoch": 4.438455731315508,
      "grad_norm": 0.05820110215708391,
      "learning_rate": 8.347631862751542e-06,
      "loss": 0.833,
      "step": 252810
    },
    {
      "epoch": 4.438631296195509,
      "grad_norm": 0.05018260336805941,
      "learning_rate": 8.343706411231817e-06,
      "loss": 0.8333,
      "step": 252820
    },
    {
      "epoch": 4.438806861075511,
      "grad_norm": 0.04940950759425424,
      "learning_rate": 8.339782133685098e-06,
      "loss": 0.8369,
      "step": 252830
    },
    {
      "epoch": 4.438982425955512,
      "grad_norm": 0.048678496439097,
      "learning_rate": 8.33585903016107e-06,
      "loss": 0.8416,
      "step": 252840
    },
    {
      "epoch": 4.439157990835513,
      "grad_norm": 0.04791758201478931,
      "learning_rate": 8.331937100709488e-06,
      "loss": 0.8328,
      "step": 252850
    },
    {
      "epoch": 4.439333555715515,
      "grad_norm": 0.04614173672598014,
      "learning_rate": 8.328016345380021e-06,
      "loss": 0.8468,
      "step": 252860
    },
    {
      "epoch": 4.439509120595516,
      "grad_norm": 0.04897909292366216,
      "learning_rate": 8.324096764222339e-06,
      "loss": 0.8362,
      "step": 252870
    },
    {
      "epoch": 4.439684685475518,
      "grad_norm": 0.0453455886252759,
      "learning_rate": 8.320178357286133e-06,
      "loss": 0.8397,
      "step": 252880
    },
    {
      "epoch": 4.439860250355519,
      "grad_norm": 0.03846464490985724,
      "learning_rate": 8.316261124621005e-06,
      "loss": 0.84,
      "step": 252890
    },
    {
      "epoch": 4.44003581523552,
      "grad_norm": 0.04488780462271407,
      "learning_rate": 8.312345066276628e-06,
      "loss": 0.8308,
      "step": 252900
    },
    {
      "epoch": 4.440211380115522,
      "grad_norm": 0.07241250336317415,
      "learning_rate": 8.308430182302582e-06,
      "loss": 0.8331,
      "step": 252910
    },
    {
      "epoch": 4.440386944995523,
      "grad_norm": 0.0465909713946117,
      "learning_rate": 8.30451647274851e-06,
      "loss": 0.8441,
      "step": 252920
    },
    {
      "epoch": 4.440562509875525,
      "grad_norm": 0.051052336951995986,
      "learning_rate": 8.300603937663975e-06,
      "loss": 0.8367,
      "step": 252930
    },
    {
      "epoch": 4.440738074755526,
      "grad_norm": 0.054674620939189685,
      "learning_rate": 8.296692577098549e-06,
      "loss": 0.8357,
      "step": 252940
    },
    {
      "epoch": 4.440913639635527,
      "grad_norm": 0.05392716597715143,
      "learning_rate": 8.2927823911018e-06,
      "loss": 0.8402,
      "step": 252950
    },
    {
      "epoch": 4.441089204515529,
      "grad_norm": 0.04513381752561494,
      "learning_rate": 8.288873379723263e-06,
      "loss": 0.8487,
      "step": 252960
    },
    {
      "epoch": 4.44126476939553,
      "grad_norm": 0.1179500438852326,
      "learning_rate": 8.284965543012482e-06,
      "loss": 0.8452,
      "step": 252970
    },
    {
      "epoch": 4.441440334275532,
      "grad_norm": 0.04547282660596662,
      "learning_rate": 8.281058881018932e-06,
      "loss": 0.8431,
      "step": 252980
    },
    {
      "epoch": 4.441615899155533,
      "grad_norm": 0.049829303656315424,
      "learning_rate": 8.277153393792147e-06,
      "loss": 0.8359,
      "step": 252990
    },
    {
      "epoch": 4.441791464035535,
      "grad_norm": 0.053116792847342525,
      "learning_rate": 8.2732490813816e-06,
      "loss": 0.8425,
      "step": 253000
    },
    {
      "epoch": 4.441967028915536,
      "grad_norm": 0.05445839316961897,
      "learning_rate": 8.26934594383674e-06,
      "loss": 0.8402,
      "step": 253010
    },
    {
      "epoch": 4.442142593795537,
      "grad_norm": 0.054964048092426616,
      "learning_rate": 8.265443981207064e-06,
      "loss": 0.8411,
      "step": 253020
    },
    {
      "epoch": 4.442318158675539,
      "grad_norm": 0.05458702421229574,
      "learning_rate": 8.261543193541968e-06,
      "loss": 0.8304,
      "step": 253030
    },
    {
      "epoch": 4.44249372355554,
      "grad_norm": 0.06390618096835546,
      "learning_rate": 8.257643580890904e-06,
      "loss": 0.8278,
      "step": 253040
    },
    {
      "epoch": 4.442669288435542,
      "grad_norm": 0.055144928957749376,
      "learning_rate": 8.25374514330327e-06,
      "loss": 0.85,
      "step": 253050
    },
    {
      "epoch": 4.442844853315543,
      "grad_norm": 0.10617738964115878,
      "learning_rate": 8.24984788082845e-06,
      "loss": 0.8328,
      "step": 253060
    },
    {
      "epoch": 4.443020418195545,
      "grad_norm": 0.05266719538229584,
      "learning_rate": 8.245951793515852e-06,
      "loss": 0.8391,
      "step": 253070
    },
    {
      "epoch": 4.443195983075546,
      "grad_norm": 0.05181205066873956,
      "learning_rate": 8.242056881414787e-06,
      "loss": 0.8383,
      "step": 253080
    },
    {
      "epoch": 4.443371547955547,
      "grad_norm": 0.05116172601419715,
      "learning_rate": 8.238163144574679e-06,
      "loss": 0.8436,
      "step": 253090
    },
    {
      "epoch": 4.443547112835549,
      "grad_norm": 0.046016586555812894,
      "learning_rate": 8.234270583044792e-06,
      "loss": 0.8395,
      "step": 253100
    },
    {
      "epoch": 4.44372267771555,
      "grad_norm": 0.04643887209908536,
      "learning_rate": 8.230379196874476e-06,
      "loss": 0.8362,
      "step": 253110
    },
    {
      "epoch": 4.4438982425955516,
      "grad_norm": 0.04941678050234471,
      "learning_rate": 8.226488986113062e-06,
      "loss": 0.8316,
      "step": 253120
    },
    {
      "epoch": 4.444073807475553,
      "grad_norm": 0.05820255664293819,
      "learning_rate": 8.222599950809786e-06,
      "loss": 0.8372,
      "step": 253130
    },
    {
      "epoch": 4.444249372355554,
      "grad_norm": 0.05412807515830896,
      "learning_rate": 8.218712091013972e-06,
      "loss": 0.8405,
      "step": 253140
    },
    {
      "epoch": 4.4444249372355555,
      "grad_norm": 0.045457937416361324,
      "learning_rate": 8.214825406774833e-06,
      "loss": 0.8424,
      "step": 253150
    },
    {
      "epoch": 4.444600502115557,
      "grad_norm": 0.04923724711260702,
      "learning_rate": 8.210939898141646e-06,
      "loss": 0.8369,
      "step": 253160
    },
    {
      "epoch": 4.4447760669955585,
      "grad_norm": 0.04857174026790331,
      "learning_rate": 8.207055565163649e-06,
      "loss": 0.8341,
      "step": 253170
    },
    {
      "epoch": 4.4449516318755595,
      "grad_norm": 0.059499086457694444,
      "learning_rate": 8.203172407890006e-06,
      "loss": 0.8311,
      "step": 253180
    },
    {
      "epoch": 4.445127196755561,
      "grad_norm": 0.14822179587027492,
      "learning_rate": 8.199290426369993e-06,
      "loss": 0.8397,
      "step": 253190
    },
    {
      "epoch": 4.4453027616355625,
      "grad_norm": 0.055472280291511224,
      "learning_rate": 8.195409620652727e-06,
      "loss": 0.8377,
      "step": 253200
    },
    {
      "epoch": 4.4454783265155635,
      "grad_norm": 0.04791118523063821,
      "learning_rate": 8.191529990787419e-06,
      "loss": 0.831,
      "step": 253210
    },
    {
      "epoch": 4.445653891395565,
      "grad_norm": 0.06110373366320234,
      "learning_rate": 8.187651536823216e-06,
      "loss": 0.8431,
      "step": 253220
    },
    {
      "epoch": 4.4458294562755665,
      "grad_norm": 0.055368615493865174,
      "learning_rate": 8.183774258809243e-06,
      "loss": 0.8396,
      "step": 253230
    },
    {
      "epoch": 4.446005021155568,
      "grad_norm": 0.0476741737366664,
      "learning_rate": 8.179898156794662e-06,
      "loss": 0.8416,
      "step": 253240
    },
    {
      "epoch": 4.446180586035569,
      "grad_norm": 0.059775098788216144,
      "learning_rate": 8.176023230828529e-06,
      "loss": 0.8309,
      "step": 253250
    },
    {
      "epoch": 4.4463561509155705,
      "grad_norm": 0.06331207516245135,
      "learning_rate": 8.17214948095999e-06,
      "loss": 0.8412,
      "step": 253260
    },
    {
      "epoch": 4.446531715795572,
      "grad_norm": 0.055230781479085236,
      "learning_rate": 8.168276907238105e-06,
      "loss": 0.8424,
      "step": 253270
    },
    {
      "epoch": 4.446707280675573,
      "grad_norm": 0.059389775356306976,
      "learning_rate": 8.164405509711937e-06,
      "loss": 0.8385,
      "step": 253280
    },
    {
      "epoch": 4.446882845555575,
      "grad_norm": 0.045297220468078235,
      "learning_rate": 8.160535288430563e-06,
      "loss": 0.8431,
      "step": 253290
    },
    {
      "epoch": 4.447058410435576,
      "grad_norm": 0.05332083173527712,
      "learning_rate": 8.156666243442978e-06,
      "loss": 0.8368,
      "step": 253300
    },
    {
      "epoch": 4.447233975315578,
      "grad_norm": 0.0482515273672711,
      "learning_rate": 8.152798374798239e-06,
      "loss": 0.8392,
      "step": 253310
    },
    {
      "epoch": 4.447409540195579,
      "grad_norm": 0.042731923727394754,
      "learning_rate": 8.148931682545333e-06,
      "loss": 0.844,
      "step": 253320
    },
    {
      "epoch": 4.44758510507558,
      "grad_norm": 0.052132700994759945,
      "learning_rate": 8.145066166733248e-06,
      "loss": 0.8399,
      "step": 253330
    },
    {
      "epoch": 4.447760669955582,
      "grad_norm": 0.06141558138925998,
      "learning_rate": 8.141201827410992e-06,
      "loss": 0.8315,
      "step": 253340
    },
    {
      "epoch": 4.447936234835583,
      "grad_norm": 0.05157409400234576,
      "learning_rate": 8.137338664627466e-06,
      "loss": 0.8414,
      "step": 253350
    },
    {
      "epoch": 4.448111799715585,
      "grad_norm": 0.06101725485466882,
      "learning_rate": 8.133476678431695e-06,
      "loss": 0.8338,
      "step": 253360
    },
    {
      "epoch": 4.448287364595586,
      "grad_norm": 0.04827084949321106,
      "learning_rate": 8.12961586887255e-06,
      "loss": 0.8398,
      "step": 253370
    },
    {
      "epoch": 4.448462929475587,
      "grad_norm": 0.05589793441619098,
      "learning_rate": 8.12575623599898e-06,
      "loss": 0.8343,
      "step": 253380
    },
    {
      "epoch": 4.448638494355589,
      "grad_norm": 0.06459981341385668,
      "learning_rate": 8.121897779859881e-06,
      "loss": 0.8372,
      "step": 253390
    },
    {
      "epoch": 4.44881405923559,
      "grad_norm": 0.047644218711099556,
      "learning_rate": 8.118040500504134e-06,
      "loss": 0.8342,
      "step": 253400
    },
    {
      "epoch": 4.448989624115592,
      "grad_norm": 0.04988771952197799,
      "learning_rate": 8.114184397980626e-06,
      "loss": 0.8389,
      "step": 253410
    },
    {
      "epoch": 4.449165188995593,
      "grad_norm": 0.06508353985604418,
      "learning_rate": 8.11032947233817e-06,
      "loss": 0.8512,
      "step": 253420
    },
    {
      "epoch": 4.449340753875595,
      "grad_norm": 0.051086824338451735,
      "learning_rate": 8.106475723625664e-06,
      "loss": 0.8382,
      "step": 253430
    },
    {
      "epoch": 4.449516318755596,
      "grad_norm": 0.05505812923507727,
      "learning_rate": 8.102623151891914e-06,
      "loss": 0.8372,
      "step": 253440
    },
    {
      "epoch": 4.449691883635597,
      "grad_norm": 0.05176933707280488,
      "learning_rate": 8.098771757185725e-06,
      "loss": 0.8476,
      "step": 253450
    },
    {
      "epoch": 4.449867448515599,
      "grad_norm": 0.0547178764172615,
      "learning_rate": 8.094921539555913e-06,
      "loss": 0.8425,
      "step": 253460
    },
    {
      "epoch": 4.4500430133956,
      "grad_norm": 0.054543163193648554,
      "learning_rate": 8.091072499051248e-06,
      "loss": 0.8281,
      "step": 253470
    },
    {
      "epoch": 4.450218578275602,
      "grad_norm": 0.05724473494578292,
      "learning_rate": 8.087224635720502e-06,
      "loss": 0.8455,
      "step": 253480
    },
    {
      "epoch": 4.450394143155603,
      "grad_norm": 0.059174847770914786,
      "learning_rate": 8.083377949612417e-06,
      "loss": 0.8451,
      "step": 253490
    },
    {
      "epoch": 4.450569708035604,
      "grad_norm": 0.055770713103994576,
      "learning_rate": 8.079532440775748e-06,
      "loss": 0.8386,
      "step": 253500
    },
    {
      "epoch": 4.450745272915606,
      "grad_norm": 0.04236396922273869,
      "learning_rate": 8.075688109259217e-06,
      "loss": 0.8393,
      "step": 253510
    },
    {
      "epoch": 4.450920837795607,
      "grad_norm": 0.04947024236040556,
      "learning_rate": 8.071844955111495e-06,
      "loss": 0.8402,
      "step": 253520
    },
    {
      "epoch": 4.451096402675609,
      "grad_norm": 0.04138167583861395,
      "learning_rate": 8.068002978381349e-06,
      "loss": 0.8362,
      "step": 253530
    },
    {
      "epoch": 4.45127196755561,
      "grad_norm": 0.0549308777121658,
      "learning_rate": 8.064162179117383e-06,
      "loss": 0.8373,
      "step": 253540
    },
    {
      "epoch": 4.451447532435612,
      "grad_norm": 0.06198457496529774,
      "learning_rate": 8.06032255736831e-06,
      "loss": 0.8368,
      "step": 253550
    },
    {
      "epoch": 4.451623097315613,
      "grad_norm": 0.048692960729082566,
      "learning_rate": 8.056484113182763e-06,
      "loss": 0.8365,
      "step": 253560
    },
    {
      "epoch": 4.451798662195614,
      "grad_norm": 0.06041423796825397,
      "learning_rate": 8.052646846609374e-06,
      "loss": 0.839,
      "step": 253570
    },
    {
      "epoch": 4.451974227075616,
      "grad_norm": 0.0852688626785839,
      "learning_rate": 8.048810757696773e-06,
      "loss": 0.8411,
      "step": 253580
    },
    {
      "epoch": 4.452149791955617,
      "grad_norm": 0.04919059618678376,
      "learning_rate": 8.04497584649352e-06,
      "loss": 0.8348,
      "step": 253590
    },
    {
      "epoch": 4.452325356835619,
      "grad_norm": 0.05630567402871224,
      "learning_rate": 8.041142113048265e-06,
      "loss": 0.8383,
      "step": 253600
    },
    {
      "epoch": 4.45250092171562,
      "grad_norm": 0.052039007033881306,
      "learning_rate": 8.037309557409541e-06,
      "loss": 0.8438,
      "step": 253610
    },
    {
      "epoch": 4.452676486595621,
      "grad_norm": 0.05574571774647606,
      "learning_rate": 8.033478179625922e-06,
      "loss": 0.8346,
      "step": 253620
    },
    {
      "epoch": 4.452852051475623,
      "grad_norm": 0.043972624481849916,
      "learning_rate": 8.029647979745977e-06,
      "loss": 0.8416,
      "step": 253630
    },
    {
      "epoch": 4.453027616355624,
      "grad_norm": 0.05688775843067392,
      "learning_rate": 8.025818957818175e-06,
      "loss": 0.838,
      "step": 253640
    },
    {
      "epoch": 4.453203181235626,
      "grad_norm": 0.04644066711347084,
      "learning_rate": 8.0219911138911e-06,
      "loss": 0.8532,
      "step": 253650
    },
    {
      "epoch": 4.453378746115627,
      "grad_norm": 0.051726798886964284,
      "learning_rate": 8.018164448013189e-06,
      "loss": 0.8343,
      "step": 253660
    },
    {
      "epoch": 4.453554310995629,
      "grad_norm": 0.04006401692182455,
      "learning_rate": 8.014338960232959e-06,
      "loss": 0.8328,
      "step": 253670
    },
    {
      "epoch": 4.45372987587563,
      "grad_norm": 0.053863293662990905,
      "learning_rate": 8.010514650598889e-06,
      "loss": 0.8386,
      "step": 253680
    },
    {
      "epoch": 4.453905440755631,
      "grad_norm": 0.052369016239497336,
      "learning_rate": 8.006691519159386e-06,
      "loss": 0.8432,
      "step": 253690
    },
    {
      "epoch": 4.454081005635633,
      "grad_norm": 0.05489301568503847,
      "learning_rate": 8.002869565962964e-06,
      "loss": 0.8473,
      "step": 253700
    },
    {
      "epoch": 4.454256570515634,
      "grad_norm": 0.055192559433276035,
      "learning_rate": 7.999048791057988e-06,
      "loss": 0.8374,
      "step": 253710
    },
    {
      "epoch": 4.454432135395636,
      "grad_norm": 0.048221739777525696,
      "learning_rate": 7.995229194492882e-06,
      "loss": 0.8328,
      "step": 253720
    },
    {
      "epoch": 4.454607700275637,
      "grad_norm": 0.06389700473010076,
      "learning_rate": 7.991410776316062e-06,
      "loss": 0.8324,
      "step": 253730
    },
    {
      "epoch": 4.454783265155639,
      "grad_norm": 0.05312736320049464,
      "learning_rate": 7.987593536575879e-06,
      "loss": 0.8439,
      "step": 253740
    },
    {
      "epoch": 4.45495883003564,
      "grad_norm": 0.04896230923543457,
      "learning_rate": 7.983777475320728e-06,
      "loss": 0.8409,
      "step": 253750
    },
    {
      "epoch": 4.455134394915641,
      "grad_norm": 0.05168414843814896,
      "learning_rate": 7.979962592598916e-06,
      "loss": 0.841,
      "step": 253760
    },
    {
      "epoch": 4.455309959795643,
      "grad_norm": 0.05541396790048837,
      "learning_rate": 7.976148888458834e-06,
      "loss": 0.8412,
      "step": 253770
    },
    {
      "epoch": 4.455485524675644,
      "grad_norm": 0.04346581417227673,
      "learning_rate": 7.97233636294875e-06,
      "loss": 0.8407,
      "step": 253780
    },
    {
      "epoch": 4.455661089555646,
      "grad_norm": 0.04703002568446316,
      "learning_rate": 7.968525016116997e-06,
      "loss": 0.837,
      "step": 253790
    },
    {
      "epoch": 4.455836654435647,
      "grad_norm": 0.0594934048509442,
      "learning_rate": 7.964714848011885e-06,
      "loss": 0.8369,
      "step": 253800
    },
    {
      "epoch": 4.456012219315648,
      "grad_norm": 0.052955061000342814,
      "learning_rate": 7.960905858681643e-06,
      "loss": 0.8351,
      "step": 253810
    },
    {
      "epoch": 4.45618778419565,
      "grad_norm": 0.048876992907327776,
      "learning_rate": 7.957098048174577e-06,
      "loss": 0.8433,
      "step": 253820
    },
    {
      "epoch": 4.456363349075651,
      "grad_norm": 0.05492853889005068,
      "learning_rate": 7.953291416538883e-06,
      "loss": 0.8399,
      "step": 253830
    },
    {
      "epoch": 4.456538913955653,
      "grad_norm": 0.042994349682831244,
      "learning_rate": 7.949485963822825e-06,
      "loss": 0.8394,
      "step": 253840
    },
    {
      "epoch": 4.456714478835654,
      "grad_norm": 0.057276780436262935,
      "learning_rate": 7.94568169007463e-06,
      "loss": 0.8357,
      "step": 253850
    },
    {
      "epoch": 4.456890043715655,
      "grad_norm": 0.04012895721559447,
      "learning_rate": 7.941878595342453e-06,
      "loss": 0.8373,
      "step": 253860
    },
    {
      "epoch": 4.457065608595657,
      "grad_norm": 0.06418185329992672,
      "learning_rate": 7.938076679674533e-06,
      "loss": 0.8527,
      "step": 253870
    },
    {
      "epoch": 4.457241173475658,
      "grad_norm": 0.05788198822139475,
      "learning_rate": 7.93427594311901e-06,
      "loss": 0.8479,
      "step": 253880
    },
    {
      "epoch": 4.4574167383556595,
      "grad_norm": 0.04207821420534781,
      "learning_rate": 7.930476385724043e-06,
      "loss": 0.8443,
      "step": 253890
    },
    {
      "epoch": 4.457592303235661,
      "grad_norm": 0.060905498150800576,
      "learning_rate": 7.926678007537785e-06,
      "loss": 0.8414,
      "step": 253900
    },
    {
      "epoch": 4.4577678681156625,
      "grad_norm": 0.14902456383508908,
      "learning_rate": 7.922880808608343e-06,
      "loss": 0.8384,
      "step": 253910
    },
    {
      "epoch": 4.4579434329956635,
      "grad_norm": 0.04809189455593536,
      "learning_rate": 7.919084788983853e-06,
      "loss": 0.8359,
      "step": 253920
    },
    {
      "epoch": 4.458118997875665,
      "grad_norm": 0.0603285470948903,
      "learning_rate": 7.91528994871237e-06,
      "loss": 0.8402,
      "step": 253930
    },
    {
      "epoch": 4.4582945627556665,
      "grad_norm": 0.04468239535240472,
      "learning_rate": 7.911496287842037e-06,
      "loss": 0.8491,
      "step": 253940
    },
    {
      "epoch": 4.4584701276356675,
      "grad_norm": 0.054815539911039504,
      "learning_rate": 7.907703806420872e-06,
      "loss": 0.8379,
      "step": 253950
    },
    {
      "epoch": 4.4586456925156694,
      "grad_norm": 0.06638031433099234,
      "learning_rate": 7.90391250449693e-06,
      "loss": 0.8397,
      "step": 253960
    },
    {
      "epoch": 4.4588212573956705,
      "grad_norm": 0.04495647904352887,
      "learning_rate": 7.900122382118288e-06,
      "loss": 0.8409,
      "step": 253970
    },
    {
      "epoch": 4.458996822275672,
      "grad_norm": 0.04538052214090267,
      "learning_rate": 7.89633343933291e-06,
      "loss": 0.8351,
      "step": 253980
    },
    {
      "epoch": 4.459172387155673,
      "grad_norm": 0.07298584972438256,
      "learning_rate": 7.892545676188847e-06,
      "loss": 0.8356,
      "step": 253990
    },
    {
      "epoch": 4.4593479520356745,
      "grad_norm": 0.10549250938980217,
      "learning_rate": 7.88875909273407e-06,
      "loss": 0.8384,
      "step": 254000
    },
    {
      "epoch": 4.459523516915676,
      "grad_norm": 0.04631578068089807,
      "learning_rate": 7.88497368901655e-06,
      "loss": 0.8359,
      "step": 254010
    },
    {
      "epoch": 4.459699081795677,
      "grad_norm": 0.04601402322724891,
      "learning_rate": 7.88118946508428e-06,
      "loss": 0.834,
      "step": 254020
    },
    {
      "epoch": 4.459874646675679,
      "grad_norm": 0.052822487467334533,
      "learning_rate": 7.877406420985144e-06,
      "loss": 0.8389,
      "step": 254030
    },
    {
      "epoch": 4.46005021155568,
      "grad_norm": 0.049600438143670746,
      "learning_rate": 7.873624556767156e-06,
      "loss": 0.8358,
      "step": 254040
    },
    {
      "epoch": 4.460225776435681,
      "grad_norm": 0.05448336542191109,
      "learning_rate": 7.869843872478181e-06,
      "loss": 0.834,
      "step": 254050
    },
    {
      "epoch": 4.460401341315683,
      "grad_norm": 0.03703100035262663,
      "learning_rate": 7.866064368166127e-06,
      "loss": 0.8456,
      "step": 254060
    },
    {
      "epoch": 4.460576906195684,
      "grad_norm": 0.0542857532079856,
      "learning_rate": 7.862286043878896e-06,
      "loss": 0.8343,
      "step": 254070
    },
    {
      "epoch": 4.460752471075686,
      "grad_norm": 0.0519679122898802,
      "learning_rate": 7.858508899664347e-06,
      "loss": 0.8442,
      "step": 254080
    },
    {
      "epoch": 4.460928035955687,
      "grad_norm": 0.05926567340877031,
      "learning_rate": 7.854732935570339e-06,
      "loss": 0.8377,
      "step": 254090
    },
    {
      "epoch": 4.461103600835689,
      "grad_norm": 0.05879898235358202,
      "learning_rate": 7.850958151644702e-06,
      "loss": 0.8358,
      "step": 254100
    },
    {
      "epoch": 4.46127916571569,
      "grad_norm": 0.06473967846003433,
      "learning_rate": 7.847184547935296e-06,
      "loss": 0.8372,
      "step": 254110
    },
    {
      "epoch": 4.461454730595691,
      "grad_norm": 0.048171953737021704,
      "learning_rate": 7.843412124489907e-06,
      "loss": 0.8372,
      "step": 254120
    },
    {
      "epoch": 4.461630295475693,
      "grad_norm": 0.05935363141194245,
      "learning_rate": 7.839640881356342e-06,
      "loss": 0.8441,
      "step": 254130
    },
    {
      "epoch": 4.461805860355694,
      "grad_norm": 0.059069066389590544,
      "learning_rate": 7.835870818582392e-06,
      "loss": 0.8371,
      "step": 254140
    },
    {
      "epoch": 4.461981425235696,
      "grad_norm": 0.05920948742293633,
      "learning_rate": 7.832101936215793e-06,
      "loss": 0.8315,
      "step": 254150
    },
    {
      "epoch": 4.462156990115697,
      "grad_norm": 0.04630016091246819,
      "learning_rate": 7.828334234304344e-06,
      "loss": 0.8537,
      "step": 254160
    },
    {
      "epoch": 4.462332554995698,
      "grad_norm": 0.04784439137970437,
      "learning_rate": 7.824567712895737e-06,
      "loss": 0.8412,
      "step": 254170
    },
    {
      "epoch": 4.4625081198757,
      "grad_norm": 0.0440660019940757,
      "learning_rate": 7.820802372037704e-06,
      "loss": 0.8394,
      "step": 254180
    },
    {
      "epoch": 4.462683684755701,
      "grad_norm": 0.0484652309084411,
      "learning_rate": 7.817038211777994e-06,
      "loss": 0.8344,
      "step": 254190
    },
    {
      "epoch": 4.462859249635703,
      "grad_norm": 0.051024580571682264,
      "learning_rate": 7.813275232164233e-06,
      "loss": 0.841,
      "step": 254200
    },
    {
      "epoch": 4.463034814515704,
      "grad_norm": 0.07024351353297857,
      "learning_rate": 7.809513433244152e-06,
      "loss": 0.8291,
      "step": 254210
    },
    {
      "epoch": 4.463210379395706,
      "grad_norm": 0.05390775461720687,
      "learning_rate": 7.80575281506539e-06,
      "loss": 0.8415,
      "step": 254220
    },
    {
      "epoch": 4.463385944275707,
      "grad_norm": 0.052189061021326165,
      "learning_rate": 7.801993377675601e-06,
      "loss": 0.8441,
      "step": 254230
    },
    {
      "epoch": 4.463561509155708,
      "grad_norm": 0.08306393701580858,
      "learning_rate": 7.798235121122437e-06,
      "loss": 0.836,
      "step": 254240
    },
    {
      "epoch": 4.46373707403571,
      "grad_norm": 0.0478594611149379,
      "learning_rate": 7.794478045453476e-06,
      "loss": 0.839,
      "step": 254250
    },
    {
      "epoch": 4.463912638915711,
      "grad_norm": 0.06244442585642008,
      "learning_rate": 7.790722150716353e-06,
      "loss": 0.8363,
      "step": 254260
    },
    {
      "epoch": 4.464088203795713,
      "grad_norm": 0.05180210370247561,
      "learning_rate": 7.78696743695862e-06,
      "loss": 0.8423,
      "step": 254270
    },
    {
      "epoch": 4.464263768675714,
      "grad_norm": 0.058940682850389184,
      "learning_rate": 7.783213904227896e-06,
      "loss": 0.8384,
      "step": 254280
    },
    {
      "epoch": 4.464439333555715,
      "grad_norm": 0.04620292695825995,
      "learning_rate": 7.779461552571712e-06,
      "loss": 0.8361,
      "step": 254290
    },
    {
      "epoch": 4.464614898435717,
      "grad_norm": 0.06749595924047568,
      "learning_rate": 7.775710382037604e-06,
      "loss": 0.8354,
      "step": 254300
    },
    {
      "epoch": 4.464790463315718,
      "grad_norm": 0.06280812916085882,
      "learning_rate": 7.77196039267314e-06,
      "loss": 0.8381,
      "step": 254310
    },
    {
      "epoch": 4.46496602819572,
      "grad_norm": 0.04265591537173002,
      "learning_rate": 7.76821158452578e-06,
      "loss": 0.8304,
      "step": 254320
    },
    {
      "epoch": 4.465141593075721,
      "grad_norm": 0.04957870483249826,
      "learning_rate": 7.764463957643062e-06,
      "loss": 0.8387,
      "step": 254330
    },
    {
      "epoch": 4.465317157955723,
      "grad_norm": 0.04220859628809577,
      "learning_rate": 7.760717512072447e-06,
      "loss": 0.8358,
      "step": 254340
    },
    {
      "epoch": 4.465492722835724,
      "grad_norm": 0.048275048728328857,
      "learning_rate": 7.756972247861406e-06,
      "loss": 0.8476,
      "step": 254350
    },
    {
      "epoch": 4.465668287715725,
      "grad_norm": 0.05643791290999846,
      "learning_rate": 7.753228165057408e-06,
      "loss": 0.8383,
      "step": 254360
    },
    {
      "epoch": 4.465843852595727,
      "grad_norm": 0.08182084066815044,
      "learning_rate": 7.749485263707852e-06,
      "loss": 0.8364,
      "step": 254370
    },
    {
      "epoch": 4.466019417475728,
      "grad_norm": 0.05274199476959374,
      "learning_rate": 7.745743543860222e-06,
      "loss": 0.8368,
      "step": 254380
    },
    {
      "epoch": 4.46619498235573,
      "grad_norm": 0.04928010993099302,
      "learning_rate": 7.742003005561877e-06,
      "loss": 0.8405,
      "step": 254390
    },
    {
      "epoch": 4.466370547235731,
      "grad_norm": 0.056194257551149755,
      "learning_rate": 7.738263648860227e-06,
      "loss": 0.843,
      "step": 254400
    },
    {
      "epoch": 4.466546112115733,
      "grad_norm": 0.049498852901331505,
      "learning_rate": 7.734525473802658e-06,
      "loss": 0.8439,
      "step": 254410
    },
    {
      "epoch": 4.466721676995734,
      "grad_norm": 0.04699841958760405,
      "learning_rate": 7.730788480436521e-06,
      "loss": 0.8476,
      "step": 254420
    },
    {
      "epoch": 4.466897241875735,
      "grad_norm": 0.05731494508784165,
      "learning_rate": 7.727052668809175e-06,
      "loss": 0.8426,
      "step": 254430
    },
    {
      "epoch": 4.467072806755737,
      "grad_norm": 0.053011495284430145,
      "learning_rate": 7.723318038967927e-06,
      "loss": 0.8391,
      "step": 254440
    },
    {
      "epoch": 4.467248371635738,
      "grad_norm": 0.04312337912236825,
      "learning_rate": 7.719584590960138e-06,
      "loss": 0.8407,
      "step": 254450
    },
    {
      "epoch": 4.46742393651574,
      "grad_norm": 0.045295759484497725,
      "learning_rate": 7.715852324833082e-06,
      "loss": 0.8395,
      "step": 254460
    },
    {
      "epoch": 4.467599501395741,
      "grad_norm": 0.0601482097493204,
      "learning_rate": 7.712121240634042e-06,
      "loss": 0.8422,
      "step": 254470
    },
    {
      "epoch": 4.467775066275742,
      "grad_norm": 0.051742626955988566,
      "learning_rate": 7.708391338410336e-06,
      "loss": 0.8378,
      "step": 254480
    },
    {
      "epoch": 4.467950631155744,
      "grad_norm": 0.06343848379078056,
      "learning_rate": 7.704662618209159e-06,
      "loss": 0.8366,
      "step": 254490
    },
    {
      "epoch": 4.468126196035745,
      "grad_norm": 0.04716398392354567,
      "learning_rate": 7.700935080077806e-06,
      "loss": 0.8363,
      "step": 254500
    },
    {
      "epoch": 4.468301760915747,
      "grad_norm": 0.05128188363420445,
      "learning_rate": 7.697208724063475e-06,
      "loss": 0.8366,
      "step": 254510
    },
    {
      "epoch": 4.468477325795748,
      "grad_norm": 0.0404099532543392,
      "learning_rate": 7.693483550213397e-06,
      "loss": 0.8408,
      "step": 254520
    },
    {
      "epoch": 4.468652890675749,
      "grad_norm": 0.06589654884758527,
      "learning_rate": 7.689759558574774e-06,
      "loss": 0.8448,
      "step": 254530
    },
    {
      "epoch": 4.468828455555751,
      "grad_norm": 0.04719089672857979,
      "learning_rate": 7.686036749194752e-06,
      "loss": 0.8452,
      "step": 254540
    },
    {
      "epoch": 4.469004020435752,
      "grad_norm": 0.041584216833093185,
      "learning_rate": 7.682315122120557e-06,
      "loss": 0.8441,
      "step": 254550
    },
    {
      "epoch": 4.469179585315754,
      "grad_norm": 0.06166676754229799,
      "learning_rate": 7.678594677399299e-06,
      "loss": 0.8311,
      "step": 254560
    },
    {
      "epoch": 4.469355150195755,
      "grad_norm": 0.04973056863631194,
      "learning_rate": 7.67487541507813e-06,
      "loss": 0.8402,
      "step": 254570
    },
    {
      "epoch": 4.469530715075757,
      "grad_norm": 0.05066015192234843,
      "learning_rate": 7.67115733520418e-06,
      "loss": 0.8329,
      "step": 254580
    },
    {
      "epoch": 4.469706279955758,
      "grad_norm": 0.047129951309531765,
      "learning_rate": 7.667440437824557e-06,
      "loss": 0.8356,
      "step": 254590
    },
    {
      "epoch": 4.469881844835759,
      "grad_norm": 0.06626886701802695,
      "learning_rate": 7.66372472298635e-06,
      "loss": 0.8465,
      "step": 254600
    },
    {
      "epoch": 4.470057409715761,
      "grad_norm": 0.059540565649613124,
      "learning_rate": 7.660010190736616e-06,
      "loss": 0.8408,
      "step": 254610
    },
    {
      "epoch": 4.470232974595762,
      "grad_norm": 0.05241816160942367,
      "learning_rate": 7.656296841122474e-06,
      "loss": 0.8355,
      "step": 254620
    },
    {
      "epoch": 4.4704085394757636,
      "grad_norm": 0.06861513042550292,
      "learning_rate": 7.652584674190924e-06,
      "loss": 0.8364,
      "step": 254630
    },
    {
      "epoch": 4.470584104355765,
      "grad_norm": 0.0779868395926562,
      "learning_rate": 7.648873689989014e-06,
      "loss": 0.8456,
      "step": 254640
    },
    {
      "epoch": 4.4707596692357665,
      "grad_norm": 0.04310760822069281,
      "learning_rate": 7.645163888563772e-06,
      "loss": 0.843,
      "step": 254650
    },
    {
      "epoch": 4.4709352341157675,
      "grad_norm": 0.053634464764067476,
      "learning_rate": 7.641455269962189e-06,
      "loss": 0.8388,
      "step": 254660
    },
    {
      "epoch": 4.471110798995769,
      "grad_norm": 0.052118958698317105,
      "learning_rate": 7.637747834231272e-06,
      "loss": 0.8385,
      "step": 254670
    },
    {
      "epoch": 4.4712863638757705,
      "grad_norm": 0.05924436970875331,
      "learning_rate": 7.634041581417974e-06,
      "loss": 0.8344,
      "step": 254680
    },
    {
      "epoch": 4.4714619287557715,
      "grad_norm": 0.06919418811809365,
      "learning_rate": 7.630336511569256e-06,
      "loss": 0.8432,
      "step": 254690
    },
    {
      "epoch": 4.4716374936357735,
      "grad_norm": 0.04782795548943559,
      "learning_rate": 7.626632624732086e-06,
      "loss": 0.8432,
      "step": 254700
    },
    {
      "epoch": 4.4718130585157745,
      "grad_norm": 0.04418823679817901,
      "learning_rate": 7.622929920953345e-06,
      "loss": 0.8361,
      "step": 254710
    },
    {
      "epoch": 4.4719886233957755,
      "grad_norm": 0.0572327535518679,
      "learning_rate": 7.619228400280011e-06,
      "loss": 0.8332,
      "step": 254720
    },
    {
      "epoch": 4.472164188275777,
      "grad_norm": 0.048763449070432806,
      "learning_rate": 7.615528062758925e-06,
      "loss": 0.8388,
      "step": 254730
    },
    {
      "epoch": 4.4723397531557785,
      "grad_norm": 0.07420942525092306,
      "learning_rate": 7.611828908437002e-06,
      "loss": 0.8389,
      "step": 254740
    },
    {
      "epoch": 4.47251531803578,
      "grad_norm": 0.053060907124893156,
      "learning_rate": 7.608130937361128e-06,
      "loss": 0.84,
      "step": 254750
    },
    {
      "epoch": 4.472690882915781,
      "grad_norm": 0.0621892941526554,
      "learning_rate": 7.604434149578103e-06,
      "loss": 0.8373,
      "step": 254760
    },
    {
      "epoch": 4.4728664477957825,
      "grad_norm": 0.05250844579056338,
      "learning_rate": 7.600738545134817e-06,
      "loss": 0.8414,
      "step": 254770
    },
    {
      "epoch": 4.473042012675784,
      "grad_norm": 0.049170066925130596,
      "learning_rate": 7.597044124078047e-06,
      "loss": 0.8411,
      "step": 254780
    },
    {
      "epoch": 4.473217577555785,
      "grad_norm": 0.051947615848269485,
      "learning_rate": 7.59335088645466e-06,
      "loss": 0.8408,
      "step": 254790
    },
    {
      "epoch": 4.473393142435787,
      "grad_norm": 0.04619302549727149,
      "learning_rate": 7.5896588323114025e-06,
      "loss": 0.8441,
      "step": 254800
    },
    {
      "epoch": 4.473568707315788,
      "grad_norm": 0.0455508762012442,
      "learning_rate": 7.585967961695063e-06,
      "loss": 0.8447,
      "step": 254810
    },
    {
      "epoch": 4.47374427219579,
      "grad_norm": 0.03993761752180329,
      "learning_rate": 7.582278274652432e-06,
      "loss": 0.8417,
      "step": 254820
    },
    {
      "epoch": 4.473919837075791,
      "grad_norm": 0.05224660370523509,
      "learning_rate": 7.578589771230221e-06,
      "loss": 0.8373,
      "step": 254830
    },
    {
      "epoch": 4.474095401955792,
      "grad_norm": 0.044344446625762875,
      "learning_rate": 7.5749024514752e-06,
      "loss": 0.8383,
      "step": 254840
    },
    {
      "epoch": 4.474270966835794,
      "grad_norm": 0.05719184697952422,
      "learning_rate": 7.571216315434057e-06,
      "loss": 0.8383,
      "step": 254850
    },
    {
      "epoch": 4.474446531715795,
      "grad_norm": 0.07104062733701597,
      "learning_rate": 7.567531363153505e-06,
      "loss": 0.8414,
      "step": 254860
    },
    {
      "epoch": 4.474622096595797,
      "grad_norm": 0.04558334638073091,
      "learning_rate": 7.563847594680245e-06,
      "loss": 0.8303,
      "step": 254870
    },
    {
      "epoch": 4.474797661475798,
      "grad_norm": 0.05759443615785554,
      "learning_rate": 7.560165010060927e-06,
      "loss": 0.8365,
      "step": 254880
    },
    {
      "epoch": 4.4749732263558,
      "grad_norm": 0.0491343450027992,
      "learning_rate": 7.556483609342249e-06,
      "loss": 0.8313,
      "step": 254890
    },
    {
      "epoch": 4.475148791235801,
      "grad_norm": 0.056769853226541156,
      "learning_rate": 7.5528033925708176e-06,
      "loss": 0.8415,
      "step": 254900
    },
    {
      "epoch": 4.475324356115802,
      "grad_norm": 0.06066874930463641,
      "learning_rate": 7.5491243597932745e-06,
      "loss": 0.8343,
      "step": 254910
    },
    {
      "epoch": 4.475499920995804,
      "grad_norm": 0.06437935103633036,
      "learning_rate": 7.545446511056248e-06,
      "loss": 0.8451,
      "step": 254920
    },
    {
      "epoch": 4.475675485875805,
      "grad_norm": 0.05545811362174651,
      "learning_rate": 7.541769846406319e-06,
      "loss": 0.8416,
      "step": 254930
    },
    {
      "epoch": 4.475851050755807,
      "grad_norm": 0.049657140929998934,
      "learning_rate": 7.538094365890078e-06,
      "loss": 0.8336,
      "step": 254940
    },
    {
      "epoch": 4.476026615635808,
      "grad_norm": 0.05973230390329226,
      "learning_rate": 7.5344200695540714e-06,
      "loss": 0.8449,
      "step": 254950
    },
    {
      "epoch": 4.476202180515809,
      "grad_norm": 0.052225820834596315,
      "learning_rate": 7.530746957444915e-06,
      "loss": 0.8378,
      "step": 254960
    },
    {
      "epoch": 4.476377745395811,
      "grad_norm": 0.06065782676768428,
      "learning_rate": 7.527075029609079e-06,
      "loss": 0.8251,
      "step": 254970
    },
    {
      "epoch": 4.476553310275812,
      "grad_norm": 0.05914358696276536,
      "learning_rate": 7.523404286093122e-06,
      "loss": 0.8386,
      "step": 254980
    },
    {
      "epoch": 4.476728875155814,
      "grad_norm": 0.05552500868340997,
      "learning_rate": 7.5197347269435584e-06,
      "loss": 0.8292,
      "step": 254990
    },
    {
      "epoch": 4.476904440035815,
      "grad_norm": 0.04826672228372833,
      "learning_rate": 7.5160663522068605e-06,
      "loss": 0.8356,
      "step": 255000
    },
    {
      "epoch": 4.477080004915817,
      "grad_norm": 0.04569417570607637,
      "learning_rate": 7.5123991619295315e-06,
      "loss": 0.8383,
      "step": 255010
    },
    {
      "epoch": 4.477255569795818,
      "grad_norm": 0.06090843828911807,
      "learning_rate": 7.508733156158007e-06,
      "loss": 0.8322,
      "step": 255020
    },
    {
      "epoch": 4.477431134675819,
      "grad_norm": 0.04114157636009874,
      "learning_rate": 7.5050683349387506e-06,
      "loss": 0.8343,
      "step": 255030
    },
    {
      "epoch": 4.477606699555821,
      "grad_norm": 0.05419820967203716,
      "learning_rate": 7.50140469831821e-06,
      "loss": 0.8328,
      "step": 255040
    },
    {
      "epoch": 4.477782264435822,
      "grad_norm": 0.04454911257545706,
      "learning_rate": 7.497742246342765e-06,
      "loss": 0.8404,
      "step": 255050
    },
    {
      "epoch": 4.477957829315824,
      "grad_norm": 0.06237324970619175,
      "learning_rate": 7.494080979058879e-06,
      "loss": 0.8421,
      "step": 255060
    },
    {
      "epoch": 4.478133394195825,
      "grad_norm": 0.04793649211583569,
      "learning_rate": 7.490420896512878e-06,
      "loss": 0.8381,
      "step": 255070
    },
    {
      "epoch": 4.478308959075827,
      "grad_norm": 0.047572472367698126,
      "learning_rate": 7.486761998751182e-06,
      "loss": 0.8423,
      "step": 255080
    },
    {
      "epoch": 4.478484523955828,
      "grad_norm": 0.05393240974228496,
      "learning_rate": 7.483104285820134e-06,
      "loss": 0.8422,
      "step": 255090
    },
    {
      "epoch": 4.478660088835829,
      "grad_norm": 0.04328795310952077,
      "learning_rate": 7.479447757766067e-06,
      "loss": 0.8359,
      "step": 255100
    },
    {
      "epoch": 4.478835653715831,
      "grad_norm": 0.046103197596077165,
      "learning_rate": 7.475792414635341e-06,
      "loss": 0.8327,
      "step": 255110
    },
    {
      "epoch": 4.479011218595832,
      "grad_norm": 0.06866609180353592,
      "learning_rate": 7.472138256474216e-06,
      "loss": 0.838,
      "step": 255120
    },
    {
      "epoch": 4.479186783475834,
      "grad_norm": 0.05535717263391965,
      "learning_rate": 7.468485283329043e-06,
      "loss": 0.8329,
      "step": 255130
    },
    {
      "epoch": 4.479362348355835,
      "grad_norm": 0.05054502574334718,
      "learning_rate": 7.464833495246088e-06,
      "loss": 0.8351,
      "step": 255140
    },
    {
      "epoch": 4.479537913235836,
      "grad_norm": 0.05036789939910729,
      "learning_rate": 7.461182892271605e-06,
      "loss": 0.8313,
      "step": 255150
    },
    {
      "epoch": 4.479713478115838,
      "grad_norm": 0.045986389542916936,
      "learning_rate": 7.457533474451883e-06,
      "loss": 0.8362,
      "step": 255160
    },
    {
      "epoch": 4.479889042995839,
      "grad_norm": 0.055569411893894416,
      "learning_rate": 7.453885241833131e-06,
      "loss": 0.8336,
      "step": 255170
    },
    {
      "epoch": 4.480064607875841,
      "grad_norm": 0.044487491445766414,
      "learning_rate": 7.450238194461584e-06,
      "loss": 0.8329,
      "step": 255180
    },
    {
      "epoch": 4.480240172755842,
      "grad_norm": 0.0665461487244328,
      "learning_rate": 7.446592332383442e-06,
      "loss": 0.8374,
      "step": 255190
    },
    {
      "epoch": 4.480415737635843,
      "grad_norm": 0.05051503162753059,
      "learning_rate": 7.442947655644898e-06,
      "loss": 0.8495,
      "step": 255200
    },
    {
      "epoch": 4.480591302515845,
      "grad_norm": 0.061300787824120795,
      "learning_rate": 7.4393041642921505e-06,
      "loss": 0.8396,
      "step": 255210
    },
    {
      "epoch": 4.480766867395846,
      "grad_norm": 0.04962279232018603,
      "learning_rate": 7.435661858371329e-06,
      "loss": 0.843,
      "step": 255220
    },
    {
      "epoch": 4.480942432275848,
      "grad_norm": 0.05082861687538463,
      "learning_rate": 7.432020737928627e-06,
      "loss": 0.8424,
      "step": 255230
    },
    {
      "epoch": 4.481117997155849,
      "grad_norm": 0.04943769915642633,
      "learning_rate": 7.428380803010146e-06,
      "loss": 0.8352,
      "step": 255240
    },
    {
      "epoch": 4.481293562035851,
      "grad_norm": 0.052502983779781444,
      "learning_rate": 7.424742053662002e-06,
      "loss": 0.84,
      "step": 255250
    },
    {
      "epoch": 4.481469126915852,
      "grad_norm": 0.04795331893837094,
      "learning_rate": 7.4211044899303374e-06,
      "loss": 0.8465,
      "step": 255260
    },
    {
      "epoch": 4.481644691795853,
      "grad_norm": 0.04658234366212989,
      "learning_rate": 7.417468111861184e-06,
      "loss": 0.8286,
      "step": 255270
    },
    {
      "epoch": 4.481820256675855,
      "grad_norm": 0.04289715128108351,
      "learning_rate": 7.413832919500668e-06,
      "loss": 0.8406,
      "step": 255280
    },
    {
      "epoch": 4.481995821555856,
      "grad_norm": 0.04824998492350425,
      "learning_rate": 7.410198912894792e-06,
      "loss": 0.8376,
      "step": 255290
    },
    {
      "epoch": 4.482171386435858,
      "grad_norm": 0.056891014159961076,
      "learning_rate": 7.406566092089665e-06,
      "loss": 0.8467,
      "step": 255300
    },
    {
      "epoch": 4.482346951315859,
      "grad_norm": 0.05505759467178569,
      "learning_rate": 7.402934457131268e-06,
      "loss": 0.8392,
      "step": 255310
    },
    {
      "epoch": 4.482522516195861,
      "grad_norm": 0.04958214682778057,
      "learning_rate": 7.399304008065627e-06,
      "loss": 0.839,
      "step": 255320
    },
    {
      "epoch": 4.482698081075862,
      "grad_norm": 0.05021894091492862,
      "learning_rate": 7.395674744938761e-06,
      "loss": 0.8375,
      "step": 255330
    },
    {
      "epoch": 4.482873645955863,
      "grad_norm": 0.05823853498363734,
      "learning_rate": 7.392046667796621e-06,
      "loss": 0.8343,
      "step": 255340
    },
    {
      "epoch": 4.483049210835865,
      "grad_norm": 0.05296831453951765,
      "learning_rate": 7.388419776685197e-06,
      "loss": 0.8396,
      "step": 255350
    },
    {
      "epoch": 4.483224775715866,
      "grad_norm": 0.05773851110370642,
      "learning_rate": 7.384794071650422e-06,
      "loss": 0.837,
      "step": 255360
    },
    {
      "epoch": 4.483400340595868,
      "grad_norm": 0.05338566651961475,
      "learning_rate": 7.38116955273826e-06,
      "loss": 0.8408,
      "step": 255370
    },
    {
      "epoch": 4.483575905475869,
      "grad_norm": 0.051773996754601974,
      "learning_rate": 7.377546219994622e-06,
      "loss": 0.8313,
      "step": 255380
    },
    {
      "epoch": 4.48375147035587,
      "grad_norm": 0.04472560531842852,
      "learning_rate": 7.373924073465394e-06,
      "loss": 0.8458,
      "step": 255390
    },
    {
      "epoch": 4.4839270352358715,
      "grad_norm": 0.05023178753786063,
      "learning_rate": 7.370303113196511e-06,
      "loss": 0.8388,
      "step": 255400
    },
    {
      "epoch": 4.484102600115873,
      "grad_norm": 0.11519708502795915,
      "learning_rate": 7.366683339233827e-06,
      "loss": 0.8343,
      "step": 255410
    },
    {
      "epoch": 4.4842781649958745,
      "grad_norm": 0.039773597837798394,
      "learning_rate": 7.363064751623208e-06,
      "loss": 0.8318,
      "step": 255420
    },
    {
      "epoch": 4.4844537298758755,
      "grad_norm": 0.049762467144542875,
      "learning_rate": 7.359447350410511e-06,
      "loss": 0.8467,
      "step": 255430
    },
    {
      "epoch": 4.484629294755877,
      "grad_norm": 0.0626546995706494,
      "learning_rate": 7.355831135641557e-06,
      "loss": 0.8426,
      "step": 255440
    },
    {
      "epoch": 4.4848048596358785,
      "grad_norm": 0.04558441728764477,
      "learning_rate": 7.352216107362179e-06,
      "loss": 0.8319,
      "step": 255450
    },
    {
      "epoch": 4.4849804245158795,
      "grad_norm": 0.051464664709201684,
      "learning_rate": 7.3486022656181455e-06,
      "loss": 0.8424,
      "step": 255460
    },
    {
      "epoch": 4.4851559893958814,
      "grad_norm": 0.04606643469924806,
      "learning_rate": 7.344989610455301e-06,
      "loss": 0.8402,
      "step": 255470
    },
    {
      "epoch": 4.4853315542758825,
      "grad_norm": 0.05492175872110588,
      "learning_rate": 7.341378141919367e-06,
      "loss": 0.8413,
      "step": 255480
    },
    {
      "epoch": 4.485507119155884,
      "grad_norm": 0.061908878440358174,
      "learning_rate": 7.3377678600561246e-06,
      "loss": 0.8447,
      "step": 255490
    },
    {
      "epoch": 4.485682684035885,
      "grad_norm": 0.04327613454515917,
      "learning_rate": 7.334158764911316e-06,
      "loss": 0.8434,
      "step": 255500
    },
    {
      "epoch": 4.4858582489158865,
      "grad_norm": 0.053215428826216216,
      "learning_rate": 7.330550856530667e-06,
      "loss": 0.8383,
      "step": 255510
    },
    {
      "epoch": 4.486033813795888,
      "grad_norm": 0.06568867083857424,
      "learning_rate": 7.326944134959901e-06,
      "loss": 0.8383,
      "step": 255520
    },
    {
      "epoch": 4.486209378675889,
      "grad_norm": 0.05054331436279004,
      "learning_rate": 7.323338600244683e-06,
      "loss": 0.8407,
      "step": 255530
    },
    {
      "epoch": 4.486384943555891,
      "grad_norm": 0.057141060643684016,
      "learning_rate": 7.319734252430732e-06,
      "loss": 0.8361,
      "step": 255540
    },
    {
      "epoch": 4.486560508435892,
      "grad_norm": 0.04818481912382552,
      "learning_rate": 7.31613109156371e-06,
      "loss": 0.8378,
      "step": 255550
    },
    {
      "epoch": 4.486736073315894,
      "grad_norm": 0.05723591141491616,
      "learning_rate": 7.312529117689225e-06,
      "loss": 0.8384,
      "step": 255560
    },
    {
      "epoch": 4.486911638195895,
      "grad_norm": 0.05754925302362289,
      "learning_rate": 7.308928330852985e-06,
      "loss": 0.8382,
      "step": 255570
    },
    {
      "epoch": 4.487087203075896,
      "grad_norm": 0.06549926957991233,
      "learning_rate": 7.305328731100572e-06,
      "loss": 0.8385,
      "step": 255580
    },
    {
      "epoch": 4.487262767955898,
      "grad_norm": 0.0423055545875055,
      "learning_rate": 7.30173031847759e-06,
      "loss": 0.8366,
      "step": 255590
    },
    {
      "epoch": 4.487438332835899,
      "grad_norm": 0.05785470601671392,
      "learning_rate": 7.298133093029661e-06,
      "loss": 0.8396,
      "step": 255600
    },
    {
      "epoch": 4.487613897715901,
      "grad_norm": 0.0460079433049,
      "learning_rate": 7.294537054802335e-06,
      "loss": 0.8372,
      "step": 255610
    },
    {
      "epoch": 4.487789462595902,
      "grad_norm": 0.05923979070029029,
      "learning_rate": 7.290942203841191e-06,
      "loss": 0.8374,
      "step": 255620
    },
    {
      "epoch": 4.487965027475903,
      "grad_norm": 0.058885527150874264,
      "learning_rate": 7.287348540191733e-06,
      "loss": 0.8342,
      "step": 255630
    },
    {
      "epoch": 4.488140592355905,
      "grad_norm": 0.053229197821959465,
      "learning_rate": 7.283756063899573e-06,
      "loss": 0.8347,
      "step": 255640
    },
    {
      "epoch": 4.488316157235906,
      "grad_norm": 0.04770885119615419,
      "learning_rate": 7.280164775010162e-06,
      "loss": 0.8358,
      "step": 255650
    },
    {
      "epoch": 4.488491722115908,
      "grad_norm": 0.049742767255494424,
      "learning_rate": 7.2765746735690256e-06,
      "loss": 0.8337,
      "step": 255660
    },
    {
      "epoch": 4.488667286995909,
      "grad_norm": 0.0612008300224235,
      "learning_rate": 7.272985759621655e-06,
      "loss": 0.8429,
      "step": 255670
    },
    {
      "epoch": 4.488842851875911,
      "grad_norm": 0.051156290467532714,
      "learning_rate": 7.269398033213509e-06,
      "loss": 0.8336,
      "step": 255680
    },
    {
      "epoch": 4.489018416755912,
      "grad_norm": 0.05778683967356671,
      "learning_rate": 7.2658114943900734e-06,
      "loss": 0.846,
      "step": 255690
    },
    {
      "epoch": 4.489193981635913,
      "grad_norm": 0.05964509638903737,
      "learning_rate": 7.26222614319676e-06,
      "loss": 0.8265,
      "step": 255700
    },
    {
      "epoch": 4.489369546515915,
      "grad_norm": 0.051774304564097436,
      "learning_rate": 7.258641979678995e-06,
      "loss": 0.8385,
      "step": 255710
    },
    {
      "epoch": 4.489545111395916,
      "grad_norm": 0.050392410788030965,
      "learning_rate": 7.2550590038822185e-06,
      "loss": 0.8261,
      "step": 255720
    },
    {
      "epoch": 4.489720676275918,
      "grad_norm": 0.044864083193910914,
      "learning_rate": 7.251477215851779e-06,
      "loss": 0.8384,
      "step": 255730
    },
    {
      "epoch": 4.489896241155919,
      "grad_norm": 0.054470403374418545,
      "learning_rate": 7.247896615633126e-06,
      "loss": 0.8438,
      "step": 255740
    },
    {
      "epoch": 4.490071806035921,
      "grad_norm": 0.04708838241789794,
      "learning_rate": 7.244317203271575e-06,
      "loss": 0.8441,
      "step": 255750
    },
    {
      "epoch": 4.490247370915922,
      "grad_norm": 0.05650908326730408,
      "learning_rate": 7.240738978812489e-06,
      "loss": 0.8297,
      "step": 255760
    },
    {
      "epoch": 4.490422935795923,
      "grad_norm": 0.04161941036077219,
      "learning_rate": 7.237161942301225e-06,
      "loss": 0.8406,
      "step": 255770
    },
    {
      "epoch": 4.490598500675925,
      "grad_norm": 0.050185504585723774,
      "learning_rate": 7.233586093783082e-06,
      "loss": 0.8392,
      "step": 255780
    },
    {
      "epoch": 4.490774065555926,
      "grad_norm": 0.057389684592258794,
      "learning_rate": 7.230011433303371e-06,
      "loss": 0.837,
      "step": 255790
    },
    {
      "epoch": 4.490949630435928,
      "grad_norm": 0.046990165657880445,
      "learning_rate": 7.226437960907373e-06,
      "loss": 0.8416,
      "step": 255800
    },
    {
      "epoch": 4.491125195315929,
      "grad_norm": 0.0606186292556797,
      "learning_rate": 7.222865676640397e-06,
      "loss": 0.8341,
      "step": 255810
    },
    {
      "epoch": 4.49130076019593,
      "grad_norm": 0.051649262310980486,
      "learning_rate": 7.219294580547675e-06,
      "loss": 0.8287,
      "step": 255820
    },
    {
      "epoch": 4.491476325075932,
      "grad_norm": 0.05138523538232476,
      "learning_rate": 7.215724672674459e-06,
      "loss": 0.8485,
      "step": 255830
    },
    {
      "epoch": 4.491651889955933,
      "grad_norm": 0.05226174916804882,
      "learning_rate": 7.21215595306601e-06,
      "loss": 0.8363,
      "step": 255840
    },
    {
      "epoch": 4.491827454835935,
      "grad_norm": 0.050133683877151405,
      "learning_rate": 7.208588421767501e-06,
      "loss": 0.8359,
      "step": 255850
    },
    {
      "epoch": 4.492003019715936,
      "grad_norm": 0.04887552434030501,
      "learning_rate": 7.205022078824162e-06,
      "loss": 0.8474,
      "step": 255860
    },
    {
      "epoch": 4.492178584595937,
      "grad_norm": 0.04379196863365765,
      "learning_rate": 7.201456924281153e-06,
      "loss": 0.8391,
      "step": 255870
    },
    {
      "epoch": 4.492354149475939,
      "grad_norm": 0.05925854767449463,
      "learning_rate": 7.197892958183663e-06,
      "loss": 0.8336,
      "step": 255880
    },
    {
      "epoch": 4.49252971435594,
      "grad_norm": 0.0512911476588675,
      "learning_rate": 7.194330180576864e-06,
      "loss": 0.8306,
      "step": 255890
    },
    {
      "epoch": 4.492705279235942,
      "grad_norm": 0.045400874318134395,
      "learning_rate": 7.1907685915058505e-06,
      "loss": 0.8487,
      "step": 255900
    },
    {
      "epoch": 4.492880844115943,
      "grad_norm": 0.05542115891179121,
      "learning_rate": 7.1872081910158016e-06,
      "loss": 0.8306,
      "step": 255910
    },
    {
      "epoch": 4.493056408995945,
      "grad_norm": 0.06926026678510917,
      "learning_rate": 7.183648979151797e-06,
      "loss": 0.8441,
      "step": 255920
    },
    {
      "epoch": 4.493231973875946,
      "grad_norm": 0.05275086816365971,
      "learning_rate": 7.180090955958928e-06,
      "loss": 0.8439,
      "step": 255930
    },
    {
      "epoch": 4.493407538755947,
      "grad_norm": 0.048170021603165664,
      "learning_rate": 7.176534121482312e-06,
      "loss": 0.8362,
      "step": 255940
    },
    {
      "epoch": 4.493583103635949,
      "grad_norm": 0.054536695784744356,
      "learning_rate": 7.172978475766978e-06,
      "loss": 0.8326,
      "step": 255950
    },
    {
      "epoch": 4.49375866851595,
      "grad_norm": 0.06815143197472791,
      "learning_rate": 7.169424018857991e-06,
      "loss": 0.8408,
      "step": 255960
    },
    {
      "epoch": 4.493934233395952,
      "grad_norm": 0.052135613879880485,
      "learning_rate": 7.165870750800362e-06,
      "loss": 0.8432,
      "step": 255970
    },
    {
      "epoch": 4.494109798275953,
      "grad_norm": 0.060537115023781875,
      "learning_rate": 7.162318671639162e-06,
      "loss": 0.8394,
      "step": 255980
    },
    {
      "epoch": 4.494285363155955,
      "grad_norm": 0.053544698758368224,
      "learning_rate": 7.158767781419351e-06,
      "loss": 0.8432,
      "step": 255990
    },
    {
      "epoch": 4.494460928035956,
      "grad_norm": 0.05326287148683299,
      "learning_rate": 7.1552180801859424e-06,
      "loss": 0.8364,
      "step": 256000
    },
    {
      "epoch": 4.494636492915957,
      "grad_norm": 0.042435971676999494,
      "learning_rate": 7.151669567983915e-06,
      "loss": 0.8434,
      "step": 256010
    },
    {
      "epoch": 4.494812057795959,
      "grad_norm": 0.05423564773219382,
      "learning_rate": 7.148122244858213e-06,
      "loss": 0.8404,
      "step": 256020
    },
    {
      "epoch": 4.49498762267596,
      "grad_norm": 0.05647859325718329,
      "learning_rate": 7.144576110853801e-06,
      "loss": 0.8345,
      "step": 256030
    },
    {
      "epoch": 4.495163187555962,
      "grad_norm": 0.0719610189803458,
      "learning_rate": 7.1410311660155774e-06,
      "loss": 0.8371,
      "step": 256040
    },
    {
      "epoch": 4.495338752435963,
      "grad_norm": 0.045628010120614335,
      "learning_rate": 7.1374874103884744e-06,
      "loss": 0.8457,
      "step": 256050
    },
    {
      "epoch": 4.495514317315964,
      "grad_norm": 0.05707883602332175,
      "learning_rate": 7.1339448440174235e-06,
      "loss": 0.8453,
      "step": 256060
    },
    {
      "epoch": 4.495689882195966,
      "grad_norm": 0.06225395815917709,
      "learning_rate": 7.130403466947236e-06,
      "loss": 0.8331,
      "step": 256070
    },
    {
      "epoch": 4.495865447075967,
      "grad_norm": 0.04740149036214051,
      "learning_rate": 7.126863279222866e-06,
      "loss": 0.8404,
      "step": 256080
    },
    {
      "epoch": 4.496041011955969,
      "grad_norm": 0.05279562726626463,
      "learning_rate": 7.123324280889104e-06,
      "loss": 0.8305,
      "step": 256090
    },
    {
      "epoch": 4.49621657683597,
      "grad_norm": 0.08789035149046412,
      "learning_rate": 7.119786471990826e-06,
      "loss": 0.8407,
      "step": 256100
    },
    {
      "epoch": 4.496392141715971,
      "grad_norm": 0.05264396270661553,
      "learning_rate": 7.116249852572854e-06,
      "loss": 0.8414,
      "step": 256110
    },
    {
      "epoch": 4.496567706595973,
      "grad_norm": 0.050110183968291776,
      "learning_rate": 7.112714422679978e-06,
      "loss": 0.8395,
      "step": 256120
    },
    {
      "epoch": 4.496743271475974,
      "grad_norm": 0.06477057860517081,
      "learning_rate": 7.109180182357018e-06,
      "loss": 0.8377,
      "step": 256130
    },
    {
      "epoch": 4.496918836355976,
      "grad_norm": 0.053923089096688484,
      "learning_rate": 7.105647131648711e-06,
      "loss": 0.8366,
      "step": 256140
    },
    {
      "epoch": 4.497094401235977,
      "grad_norm": 0.054245120974975164,
      "learning_rate": 7.102115270599867e-06,
      "loss": 0.8345,
      "step": 256150
    },
    {
      "epoch": 4.4972699661159785,
      "grad_norm": 0.046856269149141296,
      "learning_rate": 7.09858459925522e-06,
      "loss": 0.842,
      "step": 256160
    },
    {
      "epoch": 4.4974455309959795,
      "grad_norm": 0.05950618345127812,
      "learning_rate": 7.095055117659494e-06,
      "loss": 0.8366,
      "step": 256170
    },
    {
      "epoch": 4.497621095875981,
      "grad_norm": 0.05208864017073413,
      "learning_rate": 7.091526825857434e-06,
      "loss": 0.8473,
      "step": 256180
    },
    {
      "epoch": 4.4977966607559825,
      "grad_norm": 0.049049744768569506,
      "learning_rate": 7.087999723893706e-06,
      "loss": 0.8415,
      "step": 256190
    },
    {
      "epoch": 4.4979722256359835,
      "grad_norm": 0.0480585542284058,
      "learning_rate": 7.084473811813036e-06,
      "loss": 0.8406,
      "step": 256200
    },
    {
      "epoch": 4.4981477905159855,
      "grad_norm": 0.05938560044637286,
      "learning_rate": 7.080949089660069e-06,
      "loss": 0.8362,
      "step": 256210
    },
    {
      "epoch": 4.4983233553959865,
      "grad_norm": 0.049384722219456,
      "learning_rate": 7.077425557479474e-06,
      "loss": 0.8444,
      "step": 256220
    },
    {
      "epoch": 4.498498920275988,
      "grad_norm": 0.053059751106242226,
      "learning_rate": 7.073903215315919e-06,
      "loss": 0.8396,
      "step": 256230
    },
    {
      "epoch": 4.4986744851559894,
      "grad_norm": 0.06895449764633586,
      "learning_rate": 7.070382063213973e-06,
      "loss": 0.8301,
      "step": 256240
    },
    {
      "epoch": 4.4988500500359905,
      "grad_norm": 0.045128134865926685,
      "learning_rate": 7.066862101218326e-06,
      "loss": 0.8414,
      "step": 256250
    },
    {
      "epoch": 4.499025614915992,
      "grad_norm": 0.046689966038680815,
      "learning_rate": 7.063343329373515e-06,
      "loss": 0.8399,
      "step": 256260
    },
    {
      "epoch": 4.499201179795993,
      "grad_norm": 0.0557676315796469,
      "learning_rate": 7.059825747724154e-06,
      "loss": 0.8436,
      "step": 256270
    },
    {
      "epoch": 4.499376744675995,
      "grad_norm": 0.05922666965314246,
      "learning_rate": 7.056309356314811e-06,
      "loss": 0.8308,
      "step": 256280
    },
    {
      "epoch": 4.499552309555996,
      "grad_norm": 0.04995490163472403,
      "learning_rate": 7.052794155190023e-06,
      "loss": 0.8419,
      "step": 256290
    },
    {
      "epoch": 4.499727874435997,
      "grad_norm": 0.044885034597582404,
      "learning_rate": 7.04928014439436e-06,
      "loss": 0.8479,
      "step": 256300
    },
    {
      "epoch": 4.499903439315999,
      "grad_norm": 0.061715946372538945,
      "learning_rate": 7.04576732397228e-06,
      "loss": 0.8357,
      "step": 256310
    },
    {
      "epoch": 4.500079004196,
      "grad_norm": 0.05146210297199584,
      "learning_rate": 7.042255693968375e-06,
      "loss": 0.8384,
      "step": 256320
    },
    {
      "epoch": 4.500254569076002,
      "grad_norm": 0.05263534762469842,
      "learning_rate": 7.0387452544270716e-06,
      "loss": 0.8392,
      "step": 256330
    },
    {
      "epoch": 4.500430133956003,
      "grad_norm": 0.0534365899421023,
      "learning_rate": 7.035236005392884e-06,
      "loss": 0.8373,
      "step": 256340
    },
    {
      "epoch": 4.500605698836004,
      "grad_norm": 0.04757211941145551,
      "learning_rate": 7.031727946910272e-06,
      "loss": 0.842,
      "step": 256350
    },
    {
      "epoch": 4.500781263716006,
      "grad_norm": 0.061775595939195176,
      "learning_rate": 7.028221079023673e-06,
      "loss": 0.8422,
      "step": 256360
    },
    {
      "epoch": 4.500956828596007,
      "grad_norm": 0.05768431926189581,
      "learning_rate": 7.024715401777524e-06,
      "loss": 0.8404,
      "step": 256370
    },
    {
      "epoch": 4.501132393476009,
      "grad_norm": 0.05301977199770704,
      "learning_rate": 7.021210915216228e-06,
      "loss": 0.8403,
      "step": 256380
    },
    {
      "epoch": 4.50130795835601,
      "grad_norm": 0.04954471800416144,
      "learning_rate": 7.017707619384205e-06,
      "loss": 0.8349,
      "step": 256390
    },
    {
      "epoch": 4.501483523236012,
      "grad_norm": 0.04992691896497576,
      "learning_rate": 7.014205514325844e-06,
      "loss": 0.838,
      "step": 256400
    },
    {
      "epoch": 4.501659088116013,
      "grad_norm": 0.04230333265162784,
      "learning_rate": 7.010704600085496e-06,
      "loss": 0.8424,
      "step": 256410
    },
    {
      "epoch": 4.501834652996015,
      "grad_norm": 0.049443188377669166,
      "learning_rate": 7.007204876707557e-06,
      "loss": 0.8289,
      "step": 256420
    },
    {
      "epoch": 4.502010217876016,
      "grad_norm": 0.044395483391339204,
      "learning_rate": 7.003706344236327e-06,
      "loss": 0.8389,
      "step": 256430
    },
    {
      "epoch": 4.502185782756017,
      "grad_norm": 0.06956235684668577,
      "learning_rate": 7.000209002716161e-06,
      "loss": 0.8404,
      "step": 256440
    },
    {
      "epoch": 4.502361347636019,
      "grad_norm": 0.05119786093623953,
      "learning_rate": 6.9967128521913616e-06,
      "loss": 0.8335,
      "step": 256450
    },
    {
      "epoch": 4.50253691251602,
      "grad_norm": 0.04903869541373508,
      "learning_rate": 6.993217892706225e-06,
      "loss": 0.8383,
      "step": 256460
    },
    {
      "epoch": 4.502712477396022,
      "grad_norm": 0.046745633567219017,
      "learning_rate": 6.9897241243050445e-06,
      "loss": 0.8347,
      "step": 256470
    },
    {
      "epoch": 4.502888042276023,
      "grad_norm": 0.0529948217363045,
      "learning_rate": 6.986231547032048e-06,
      "loss": 0.84,
      "step": 256480
    },
    {
      "epoch": 4.503063607156024,
      "grad_norm": 0.049175746402470084,
      "learning_rate": 6.9827401609315415e-06,
      "loss": 0.8395,
      "step": 256490
    },
    {
      "epoch": 4.503239172036026,
      "grad_norm": 0.043785128824621766,
      "learning_rate": 6.979249966047724e-06,
      "loss": 0.8399,
      "step": 256500
    },
    {
      "epoch": 4.503414736916027,
      "grad_norm": 0.04050696563359873,
      "learning_rate": 6.975760962424829e-06,
      "loss": 0.8475,
      "step": 256510
    },
    {
      "epoch": 4.503590301796029,
      "grad_norm": 0.05925516788017181,
      "learning_rate": 6.9722731501070696e-06,
      "loss": 0.8371,
      "step": 256520
    },
    {
      "epoch": 4.50376586667603,
      "grad_norm": 0.0660211057077375,
      "learning_rate": 6.968786529138627e-06,
      "loss": 0.8397,
      "step": 256530
    },
    {
      "epoch": 4.503941431556031,
      "grad_norm": 0.059159057246673176,
      "learning_rate": 6.965301099563684e-06,
      "loss": 0.8426,
      "step": 256540
    },
    {
      "epoch": 4.504116996436033,
      "grad_norm": 0.07019331874694139,
      "learning_rate": 6.961816861426386e-06,
      "loss": 0.8426,
      "step": 256550
    },
    {
      "epoch": 4.504292561316034,
      "grad_norm": 0.08141681240617951,
      "learning_rate": 6.958333814770892e-06,
      "loss": 0.8415,
      "step": 256560
    },
    {
      "epoch": 4.504468126196036,
      "grad_norm": 0.06151150996389767,
      "learning_rate": 6.954851959641345e-06,
      "loss": 0.8388,
      "step": 256570
    },
    {
      "epoch": 4.504643691076037,
      "grad_norm": 0.06006422209439365,
      "learning_rate": 6.951371296081828e-06,
      "loss": 0.8359,
      "step": 256580
    },
    {
      "epoch": 4.504819255956039,
      "grad_norm": 0.04776295514034343,
      "learning_rate": 6.947891824136474e-06,
      "loss": 0.8403,
      "step": 256590
    },
    {
      "epoch": 4.50499482083604,
      "grad_norm": 0.059378217080155886,
      "learning_rate": 6.944413543849354e-06,
      "loss": 0.8375,
      "step": 256600
    },
    {
      "epoch": 4.505170385716041,
      "grad_norm": 0.0844133505663742,
      "learning_rate": 6.940936455264535e-06,
      "loss": 0.8314,
      "step": 256610
    },
    {
      "epoch": 4.505345950596043,
      "grad_norm": 0.05527674088318252,
      "learning_rate": 6.937460558426091e-06,
      "loss": 0.8335,
      "step": 256620
    },
    {
      "epoch": 4.505521515476044,
      "grad_norm": 0.04909761436415814,
      "learning_rate": 6.93398585337803e-06,
      "loss": 0.8379,
      "step": 256630
    },
    {
      "epoch": 4.505697080356046,
      "grad_norm": 0.06106570454410188,
      "learning_rate": 6.930512340164416e-06,
      "loss": 0.8418,
      "step": 256640
    },
    {
      "epoch": 4.505872645236047,
      "grad_norm": 0.051847763459794756,
      "learning_rate": 6.927040018829215e-06,
      "loss": 0.8376,
      "step": 256650
    },
    {
      "epoch": 4.506048210116049,
      "grad_norm": 0.07496532487970543,
      "learning_rate": 6.923568889416468e-06,
      "loss": 0.8331,
      "step": 256660
    },
    {
      "epoch": 4.50622377499605,
      "grad_norm": 0.05375275256876201,
      "learning_rate": 6.920098951970116e-06,
      "loss": 0.8343,
      "step": 256670
    },
    {
      "epoch": 4.506399339876051,
      "grad_norm": 0.040282080953007,
      "learning_rate": 6.916630206534147e-06,
      "loss": 0.843,
      "step": 256680
    },
    {
      "epoch": 4.506574904756053,
      "grad_norm": 0.0640866204213749,
      "learning_rate": 6.913162653152516e-06,
      "loss": 0.8301,
      "step": 256690
    },
    {
      "epoch": 4.506750469636054,
      "grad_norm": 0.05039941170067501,
      "learning_rate": 6.909696291869133e-06,
      "loss": 0.8329,
      "step": 256700
    },
    {
      "epoch": 4.506926034516056,
      "grad_norm": 0.05537971561636846,
      "learning_rate": 6.906231122727938e-06,
      "loss": 0.8358,
      "step": 256710
    },
    {
      "epoch": 4.507101599396057,
      "grad_norm": 0.058537759614135446,
      "learning_rate": 6.9027671457728235e-06,
      "loss": 0.8365,
      "step": 256720
    },
    {
      "epoch": 4.507277164276058,
      "grad_norm": 0.053353671921219224,
      "learning_rate": 6.899304361047673e-06,
      "loss": 0.8367,
      "step": 256730
    },
    {
      "epoch": 4.50745272915606,
      "grad_norm": 0.052032759038481785,
      "learning_rate": 6.895842768596388e-06,
      "loss": 0.8426,
      "step": 256740
    },
    {
      "epoch": 4.507628294036061,
      "grad_norm": 0.040754222099338475,
      "learning_rate": 6.892382368462788e-06,
      "loss": 0.834,
      "step": 256750
    },
    {
      "epoch": 4.507803858916063,
      "grad_norm": 0.05564717563513395,
      "learning_rate": 6.888923160690763e-06,
      "loss": 0.8365,
      "step": 256760
    },
    {
      "epoch": 4.507979423796064,
      "grad_norm": 0.048164149708132,
      "learning_rate": 6.885465145324102e-06,
      "loss": 0.8438,
      "step": 256770
    },
    {
      "epoch": 4.508154988676065,
      "grad_norm": 0.050788744976370044,
      "learning_rate": 6.882008322406638e-06,
      "loss": 0.8448,
      "step": 256780
    },
    {
      "epoch": 4.508330553556067,
      "grad_norm": 0.04428173638226006,
      "learning_rate": 6.878552691982182e-06,
      "loss": 0.8333,
      "step": 256790
    },
    {
      "epoch": 4.508506118436068,
      "grad_norm": 0.054384271765745264,
      "learning_rate": 6.875098254094475e-06,
      "loss": 0.8356,
      "step": 256800
    },
    {
      "epoch": 4.50868168331607,
      "grad_norm": 0.053358887498011834,
      "learning_rate": 6.871645008787333e-06,
      "loss": 0.8394,
      "step": 256810
    },
    {
      "epoch": 4.508857248196071,
      "grad_norm": 0.05844411111161257,
      "learning_rate": 6.868192956104469e-06,
      "loss": 0.8358,
      "step": 256820
    },
    {
      "epoch": 4.509032813076073,
      "grad_norm": 0.047908378462212936,
      "learning_rate": 6.864742096089655e-06,
      "loss": 0.8476,
      "step": 256830
    },
    {
      "epoch": 4.509208377956074,
      "grad_norm": 0.0639151005872408,
      "learning_rate": 6.861292428786598e-06,
      "loss": 0.8332,
      "step": 256840
    },
    {
      "epoch": 4.509383942836075,
      "grad_norm": 0.04719328024580165,
      "learning_rate": 6.8578439542390045e-06,
      "loss": 0.8392,
      "step": 256850
    },
    {
      "epoch": 4.509559507716077,
      "grad_norm": 0.05185762621399299,
      "learning_rate": 6.854396672490579e-06,
      "loss": 0.8445,
      "step": 256860
    },
    {
      "epoch": 4.509735072596078,
      "grad_norm": 0.06711378173695523,
      "learning_rate": 6.85095058358498e-06,
      "loss": 0.8394,
      "step": 256870
    },
    {
      "epoch": 4.50991063747608,
      "grad_norm": 0.048119284824122976,
      "learning_rate": 6.847505687565894e-06,
      "loss": 0.8396,
      "step": 256880
    },
    {
      "epoch": 4.510086202356081,
      "grad_norm": 0.08937671928987448,
      "learning_rate": 6.844061984476935e-06,
      "loss": 0.8352,
      "step": 256890
    },
    {
      "epoch": 4.5102617672360825,
      "grad_norm": 0.057293764221349644,
      "learning_rate": 6.840619474361762e-06,
      "loss": 0.8403,
      "step": 256900
    },
    {
      "epoch": 4.5104373321160836,
      "grad_norm": 0.05570860406316803,
      "learning_rate": 6.83717815726401e-06,
      "loss": 0.8409,
      "step": 256910
    },
    {
      "epoch": 4.510612896996085,
      "grad_norm": 0.06601556289840636,
      "learning_rate": 6.83373803322721e-06,
      "loss": 0.832,
      "step": 256920
    },
    {
      "epoch": 4.5107884618760865,
      "grad_norm": 0.05055567197684911,
      "learning_rate": 6.830299102295038e-06,
      "loss": 0.8381,
      "step": 256930
    },
    {
      "epoch": 4.5109640267560875,
      "grad_norm": 0.043570886346580476,
      "learning_rate": 6.826861364511003e-06,
      "loss": 0.8344,
      "step": 256940
    },
    {
      "epoch": 4.5111395916360895,
      "grad_norm": 0.05829616468258409,
      "learning_rate": 6.823424819918697e-06,
      "loss": 0.8322,
      "step": 256950
    },
    {
      "epoch": 4.5113151565160905,
      "grad_norm": 0.05266682140643785,
      "learning_rate": 6.819989468561636e-06,
      "loss": 0.8356,
      "step": 256960
    },
    {
      "epoch": 4.5114907213960915,
      "grad_norm": 0.057372281646934926,
      "learning_rate": 6.816555310483353e-06,
      "loss": 0.8404,
      "step": 256970
    },
    {
      "epoch": 4.5116662862760935,
      "grad_norm": 0.054823612973248656,
      "learning_rate": 6.813122345727388e-06,
      "loss": 0.8333,
      "step": 256980
    },
    {
      "epoch": 4.5118418511560945,
      "grad_norm": 0.0475331810358645,
      "learning_rate": 6.809690574337174e-06,
      "loss": 0.8368,
      "step": 256990
    },
    {
      "epoch": 4.512017416036096,
      "grad_norm": 0.06620111258697498,
      "learning_rate": 6.806259996356274e-06,
      "loss": 0.8312,
      "step": 257000
    },
    {
      "epoch": 4.512192980916097,
      "grad_norm": 0.042129526829830316,
      "learning_rate": 6.802830611828087e-06,
      "loss": 0.8352,
      "step": 257010
    },
    {
      "epoch": 4.5123685457960985,
      "grad_norm": 0.047385438993787124,
      "learning_rate": 6.799402420796086e-06,
      "loss": 0.8374,
      "step": 257020
    },
    {
      "epoch": 4.5125441106761,
      "grad_norm": 0.05635656122161265,
      "learning_rate": 6.795975423303733e-06,
      "loss": 0.8343,
      "step": 257030
    },
    {
      "epoch": 4.512719675556101,
      "grad_norm": 0.05219243525502161,
      "learning_rate": 6.792549619394403e-06,
      "loss": 0.8405,
      "step": 257040
    },
    {
      "epoch": 4.512895240436103,
      "grad_norm": 0.05184508716211594,
      "learning_rate": 6.789125009111541e-06,
      "loss": 0.8398,
      "step": 257050
    },
    {
      "epoch": 4.513070805316104,
      "grad_norm": 0.05786951720183779,
      "learning_rate": 6.785701592498514e-06,
      "loss": 0.8378,
      "step": 257060
    },
    {
      "epoch": 4.513246370196106,
      "grad_norm": 0.05431527187184887,
      "learning_rate": 6.782279369598694e-06,
      "loss": 0.8372,
      "step": 257070
    },
    {
      "epoch": 4.513421935076107,
      "grad_norm": 0.05769669503440229,
      "learning_rate": 6.7788583404554755e-06,
      "loss": 0.8386,
      "step": 257080
    },
    {
      "epoch": 4.513597499956109,
      "grad_norm": 0.05214133555965386,
      "learning_rate": 6.77543850511214e-06,
      "loss": 0.8283,
      "step": 257090
    },
    {
      "epoch": 4.51377306483611,
      "grad_norm": 0.06514690304522829,
      "learning_rate": 6.772019863612093e-06,
      "loss": 0.8357,
      "step": 257100
    },
    {
      "epoch": 4.513948629716111,
      "grad_norm": 0.06964475488286119,
      "learning_rate": 6.768602415998596e-06,
      "loss": 0.8387,
      "step": 257110
    },
    {
      "epoch": 4.514124194596113,
      "grad_norm": 0.05325422971143329,
      "learning_rate": 6.7651861623149764e-06,
      "loss": 0.8439,
      "step": 257120
    },
    {
      "epoch": 4.514299759476114,
      "grad_norm": 0.04369324085136749,
      "learning_rate": 6.761771102604505e-06,
      "loss": 0.8385,
      "step": 257130
    },
    {
      "epoch": 4.514475324356116,
      "grad_norm": 0.059455090352139535,
      "learning_rate": 6.758357236910445e-06,
      "loss": 0.8349,
      "step": 257140
    },
    {
      "epoch": 4.514650889236117,
      "grad_norm": 0.04542073094222972,
      "learning_rate": 6.754944565276078e-06,
      "loss": 0.8426,
      "step": 257150
    },
    {
      "epoch": 4.514826454116118,
      "grad_norm": 0.044609823499298075,
      "learning_rate": 6.751533087744591e-06,
      "loss": 0.8345,
      "step": 257160
    },
    {
      "epoch": 4.51500201899612,
      "grad_norm": 0.05954310838409795,
      "learning_rate": 6.7481228043592774e-06,
      "loss": 0.8417,
      "step": 257170
    },
    {
      "epoch": 4.515177583876121,
      "grad_norm": 0.049744313113648755,
      "learning_rate": 6.744713715163298e-06,
      "loss": 0.8294,
      "step": 257180
    },
    {
      "epoch": 4.515353148756123,
      "grad_norm": 0.042360209957226946,
      "learning_rate": 6.741305820199849e-06,
      "loss": 0.84,
      "step": 257190
    },
    {
      "epoch": 4.515528713636124,
      "grad_norm": 0.0737666854592852,
      "learning_rate": 6.737899119512138e-06,
      "loss": 0.8461,
      "step": 257200
    },
    {
      "epoch": 4.515704278516125,
      "grad_norm": 0.05296903038627075,
      "learning_rate": 6.734493613143305e-06,
      "loss": 0.8434,
      "step": 257210
    },
    {
      "epoch": 4.515879843396127,
      "grad_norm": 0.058798247421625725,
      "learning_rate": 6.731089301136512e-06,
      "loss": 0.8371,
      "step": 257220
    },
    {
      "epoch": 4.516055408276128,
      "grad_norm": 0.052831977675073284,
      "learning_rate": 6.727686183534868e-06,
      "loss": 0.8308,
      "step": 257230
    },
    {
      "epoch": 4.51623097315613,
      "grad_norm": 0.054728737212779315,
      "learning_rate": 6.724284260381511e-06,
      "loss": 0.832,
      "step": 257240
    },
    {
      "epoch": 4.516406538036131,
      "grad_norm": 0.0413277229994161,
      "learning_rate": 6.720883531719549e-06,
      "loss": 0.8385,
      "step": 257250
    },
    {
      "epoch": 4.516582102916132,
      "grad_norm": 0.05875599614554673,
      "learning_rate": 6.717483997592047e-06,
      "loss": 0.8375,
      "step": 257260
    },
    {
      "epoch": 4.516757667796134,
      "grad_norm": 0.05219769531794767,
      "learning_rate": 6.714085658042102e-06,
      "loss": 0.8325,
      "step": 257270
    },
    {
      "epoch": 4.516933232676135,
      "grad_norm": 0.0520128095188777,
      "learning_rate": 6.710688513112765e-06,
      "loss": 0.8312,
      "step": 257280
    },
    {
      "epoch": 4.517108797556137,
      "grad_norm": 0.047658410494105394,
      "learning_rate": 6.707292562847075e-06,
      "loss": 0.8398,
      "step": 257290
    },
    {
      "epoch": 4.517284362436138,
      "grad_norm": 0.05889772302999428,
      "learning_rate": 6.7038978072880485e-06,
      "loss": 0.8409,
      "step": 257300
    },
    {
      "epoch": 4.51745992731614,
      "grad_norm": 0.060531284547080136,
      "learning_rate": 6.700504246478709e-06,
      "loss": 0.8405,
      "step": 257310
    },
    {
      "epoch": 4.517635492196141,
      "grad_norm": 0.05879089860924705,
      "learning_rate": 6.697111880462066e-06,
      "loss": 0.8352,
      "step": 257320
    },
    {
      "epoch": 4.517811057076143,
      "grad_norm": 0.04611361438385633,
      "learning_rate": 6.693720709281065e-06,
      "loss": 0.8361,
      "step": 257330
    },
    {
      "epoch": 4.517986621956144,
      "grad_norm": 0.07308626006044867,
      "learning_rate": 6.690330732978724e-06,
      "loss": 0.8379,
      "step": 257340
    },
    {
      "epoch": 4.518162186836145,
      "grad_norm": 0.049087449113620396,
      "learning_rate": 6.6869419515979524e-06,
      "loss": 0.8416,
      "step": 257350
    },
    {
      "epoch": 4.518337751716147,
      "grad_norm": 0.05071235587671142,
      "learning_rate": 6.683554365181693e-06,
      "loss": 0.8336,
      "step": 257360
    },
    {
      "epoch": 4.518513316596148,
      "grad_norm": 0.06349673123232821,
      "learning_rate": 6.6801679737728995e-06,
      "loss": 0.8349,
      "step": 257370
    },
    {
      "epoch": 4.51868888147615,
      "grad_norm": 0.05067338666005322,
      "learning_rate": 6.676782777414436e-06,
      "loss": 0.8411,
      "step": 257380
    },
    {
      "epoch": 4.518864446356151,
      "grad_norm": 0.05031431390344126,
      "learning_rate": 6.673398776149229e-06,
      "loss": 0.8428,
      "step": 257390
    },
    {
      "epoch": 4.519040011236152,
      "grad_norm": 0.05434962535691771,
      "learning_rate": 6.670015970020116e-06,
      "loss": 0.8309,
      "step": 257400
    },
    {
      "epoch": 4.519215576116154,
      "grad_norm": 0.05098508186924281,
      "learning_rate": 6.666634359069986e-06,
      "loss": 0.8328,
      "step": 257410
    },
    {
      "epoch": 4.519391140996155,
      "grad_norm": 0.04973129032069178,
      "learning_rate": 6.663253943341694e-06,
      "loss": 0.8416,
      "step": 257420
    },
    {
      "epoch": 4.519566705876157,
      "grad_norm": 0.04738427604911563,
      "learning_rate": 6.659874722878021e-06,
      "loss": 0.8375,
      "step": 257430
    },
    {
      "epoch": 4.519742270756158,
      "grad_norm": 0.06336267220885693,
      "learning_rate": 6.656496697721838e-06,
      "loss": 0.8345,
      "step": 257440
    },
    {
      "epoch": 4.519917835636159,
      "grad_norm": 0.056824047372132425,
      "learning_rate": 6.653119867915916e-06,
      "loss": 0.8355,
      "step": 257450
    },
    {
      "epoch": 4.520093400516161,
      "grad_norm": 0.05531923979808199,
      "learning_rate": 6.649744233503065e-06,
      "loss": 0.8373,
      "step": 257460
    },
    {
      "epoch": 4.520268965396162,
      "grad_norm": 0.052778704760012944,
      "learning_rate": 6.646369794526018e-06,
      "loss": 0.831,
      "step": 257470
    },
    {
      "epoch": 4.520444530276164,
      "grad_norm": 0.04843023038586308,
      "learning_rate": 6.642996551027542e-06,
      "loss": 0.8305,
      "step": 257480
    },
    {
      "epoch": 4.520620095156165,
      "grad_norm": 0.0486834365878468,
      "learning_rate": 6.639624503050404e-06,
      "loss": 0.8457,
      "step": 257490
    },
    {
      "epoch": 4.520795660036167,
      "grad_norm": 0.0461806464361591,
      "learning_rate": 6.6362536506372835e-06,
      "loss": 0.8382,
      "step": 257500
    },
    {
      "epoch": 4.520971224916168,
      "grad_norm": 0.07134296950141081,
      "learning_rate": 6.632883993830935e-06,
      "loss": 0.8411,
      "step": 257510
    },
    {
      "epoch": 4.521146789796169,
      "grad_norm": 0.05695764214658916,
      "learning_rate": 6.629515532674016e-06,
      "loss": 0.836,
      "step": 257520
    },
    {
      "epoch": 4.521322354676171,
      "grad_norm": 0.045295798499162425,
      "learning_rate": 6.6261482672092254e-06,
      "loss": 0.8444,
      "step": 257530
    },
    {
      "epoch": 4.521497919556172,
      "grad_norm": 0.0555792302044756,
      "learning_rate": 6.622782197479235e-06,
      "loss": 0.8404,
      "step": 257540
    },
    {
      "epoch": 4.521673484436174,
      "grad_norm": 0.04968261866294143,
      "learning_rate": 6.619417323526666e-06,
      "loss": 0.8338,
      "step": 257550
    },
    {
      "epoch": 4.521849049316175,
      "grad_norm": 0.046881744731230926,
      "learning_rate": 6.616053645394188e-06,
      "loss": 0.837,
      "step": 257560
    },
    {
      "epoch": 4.522024614196177,
      "grad_norm": 0.044770261223333614,
      "learning_rate": 6.61269116312438e-06,
      "loss": 0.8424,
      "step": 257570
    },
    {
      "epoch": 4.522200179076178,
      "grad_norm": 0.04595529214594434,
      "learning_rate": 6.609329876759865e-06,
      "loss": 0.8418,
      "step": 257580
    },
    {
      "epoch": 4.522375743956179,
      "grad_norm": 0.048547534646282905,
      "learning_rate": 6.605969786343245e-06,
      "loss": 0.838,
      "step": 257590
    },
    {
      "epoch": 4.522551308836181,
      "grad_norm": 0.05031621094665321,
      "learning_rate": 6.602610891917047e-06,
      "loss": 0.8413,
      "step": 257600
    },
    {
      "epoch": 4.522726873716182,
      "grad_norm": 0.06899088035946838,
      "learning_rate": 6.599253193523894e-06,
      "loss": 0.8364,
      "step": 257610
    },
    {
      "epoch": 4.522902438596184,
      "grad_norm": 0.06301881039037725,
      "learning_rate": 6.595896691206275e-06,
      "loss": 0.837,
      "step": 257620
    },
    {
      "epoch": 4.523078003476185,
      "grad_norm": 0.06519954251667955,
      "learning_rate": 6.592541385006754e-06,
      "loss": 0.8245,
      "step": 257630
    },
    {
      "epoch": 4.523253568356186,
      "grad_norm": 0.04488714594150446,
      "learning_rate": 6.589187274967817e-06,
      "loss": 0.826,
      "step": 257640
    },
    {
      "epoch": 4.523429133236188,
      "grad_norm": 0.06303144160185384,
      "learning_rate": 6.585834361131959e-06,
      "loss": 0.8353,
      "step": 257650
    },
    {
      "epoch": 4.523604698116189,
      "grad_norm": 0.059338296901475206,
      "learning_rate": 6.5824826435417054e-06,
      "loss": 0.8351,
      "step": 257660
    },
    {
      "epoch": 4.5237802629961905,
      "grad_norm": 0.05235425426686765,
      "learning_rate": 6.579132122239448e-06,
      "loss": 0.8343,
      "step": 257670
    },
    {
      "epoch": 4.5239558278761915,
      "grad_norm": 0.049108009296135166,
      "learning_rate": 6.575782797267724e-06,
      "loss": 0.8391,
      "step": 257680
    },
    {
      "epoch": 4.524131392756193,
      "grad_norm": 0.06121734928926586,
      "learning_rate": 6.572434668668903e-06,
      "loss": 0.8347,
      "step": 257690
    },
    {
      "epoch": 4.5243069576361945,
      "grad_norm": 0.05195754765625715,
      "learning_rate": 6.569087736485433e-06,
      "loss": 0.8455,
      "step": 257700
    },
    {
      "epoch": 4.5244825225161955,
      "grad_norm": 0.05128502220680411,
      "learning_rate": 6.5657420007597305e-06,
      "loss": 0.8335,
      "step": 257710
    },
    {
      "epoch": 4.5246580873961975,
      "grad_norm": 0.04564897239077919,
      "learning_rate": 6.562397461534167e-06,
      "loss": 0.8396,
      "step": 257720
    },
    {
      "epoch": 4.5248336522761985,
      "grad_norm": 0.06327999605675556,
      "learning_rate": 6.559054118851143e-06,
      "loss": 0.8398,
      "step": 257730
    },
    {
      "epoch": 4.5250092171562,
      "grad_norm": 0.051049478144168936,
      "learning_rate": 6.5557119727529915e-06,
      "loss": 0.8455,
      "step": 257740
    },
    {
      "epoch": 4.5251847820362014,
      "grad_norm": 0.04493944649687119,
      "learning_rate": 6.552371023282056e-06,
      "loss": 0.8351,
      "step": 257750
    },
    {
      "epoch": 4.525360346916203,
      "grad_norm": 0.06298421911376387,
      "learning_rate": 6.5490312704807e-06,
      "loss": 0.8376,
      "step": 257760
    },
    {
      "epoch": 4.525535911796204,
      "grad_norm": 0.04707535355462161,
      "learning_rate": 6.545692714391194e-06,
      "loss": 0.8394,
      "step": 257770
    },
    {
      "epoch": 4.525711476676205,
      "grad_norm": 0.0506834229274037,
      "learning_rate": 6.5423553550558984e-06,
      "loss": 0.8409,
      "step": 257780
    },
    {
      "epoch": 4.525887041556207,
      "grad_norm": 0.05177421850613353,
      "learning_rate": 6.539019192517043e-06,
      "loss": 0.8373,
      "step": 257790
    },
    {
      "epoch": 4.526062606436208,
      "grad_norm": 0.05364285231899406,
      "learning_rate": 6.535684226816921e-06,
      "loss": 0.8381,
      "step": 257800
    },
    {
      "epoch": 4.52623817131621,
      "grad_norm": 0.07329205593110029,
      "learning_rate": 6.532350457997781e-06,
      "loss": 0.8386,
      "step": 257810
    },
    {
      "epoch": 4.526413736196211,
      "grad_norm": 0.04143325726331699,
      "learning_rate": 6.529017886101864e-06,
      "loss": 0.8373,
      "step": 257820
    },
    {
      "epoch": 4.526589301076212,
      "grad_norm": 0.05623834207623918,
      "learning_rate": 6.525686511171408e-06,
      "loss": 0.8395,
      "step": 257830
    },
    {
      "epoch": 4.526764865956214,
      "grad_norm": 0.06893938801104793,
      "learning_rate": 6.522356333248588e-06,
      "loss": 0.8403,
      "step": 257840
    },
    {
      "epoch": 4.526940430836215,
      "grad_norm": 0.050634238695703965,
      "learning_rate": 6.519027352375642e-06,
      "loss": 0.8361,
      "step": 257850
    },
    {
      "epoch": 4.527115995716217,
      "grad_norm": 0.06811512578099445,
      "learning_rate": 6.5156995685947106e-06,
      "loss": 0.83,
      "step": 257860
    },
    {
      "epoch": 4.527291560596218,
      "grad_norm": 0.07703054372732084,
      "learning_rate": 6.512372981947979e-06,
      "loss": 0.8453,
      "step": 257870
    },
    {
      "epoch": 4.527467125476219,
      "grad_norm": 0.05006134035127573,
      "learning_rate": 6.509047592477595e-06,
      "loss": 0.8454,
      "step": 257880
    },
    {
      "epoch": 4.527642690356221,
      "grad_norm": 0.054204231354204935,
      "learning_rate": 6.50572340022568e-06,
      "loss": 0.8399,
      "step": 257890
    },
    {
      "epoch": 4.527818255236222,
      "grad_norm": 0.05956319179409437,
      "learning_rate": 6.502400405234374e-06,
      "loss": 0.8427,
      "step": 257900
    },
    {
      "epoch": 4.527993820116224,
      "grad_norm": 0.07473085770623586,
      "learning_rate": 6.499078607545753e-06,
      "loss": 0.8283,
      "step": 257910
    },
    {
      "epoch": 4.528169384996225,
      "grad_norm": 0.054851272385107806,
      "learning_rate": 6.495758007201919e-06,
      "loss": 0.8403,
      "step": 257920
    },
    {
      "epoch": 4.528344949876226,
      "grad_norm": 0.047839352727130906,
      "learning_rate": 6.492438604244952e-06,
      "loss": 0.8384,
      "step": 257930
    },
    {
      "epoch": 4.528520514756228,
      "grad_norm": 0.05158213641647173,
      "learning_rate": 6.489120398716869e-06,
      "loss": 0.8369,
      "step": 257940
    },
    {
      "epoch": 4.528696079636229,
      "grad_norm": 0.055634069980786634,
      "learning_rate": 6.485803390659776e-06,
      "loss": 0.8427,
      "step": 257950
    },
    {
      "epoch": 4.528871644516231,
      "grad_norm": 0.051049315460870985,
      "learning_rate": 6.482487580115651e-06,
      "loss": 0.8378,
      "step": 257960
    },
    {
      "epoch": 4.529047209396232,
      "grad_norm": 0.03956171145846347,
      "learning_rate": 6.479172967126533e-06,
      "loss": 0.8385,
      "step": 257970
    },
    {
      "epoch": 4.529222774276234,
      "grad_norm": 0.056459770295316174,
      "learning_rate": 6.475859551734399e-06,
      "loss": 0.839,
      "step": 257980
    },
    {
      "epoch": 4.529398339156235,
      "grad_norm": 0.049409455084437205,
      "learning_rate": 6.472547333981246e-06,
      "loss": 0.8374,
      "step": 257990
    },
    {
      "epoch": 4.529573904036237,
      "grad_norm": 0.05478855116889959,
      "learning_rate": 6.469236313909038e-06,
      "loss": 0.8258,
      "step": 258000
    },
    {
      "epoch": 4.529749468916238,
      "grad_norm": 0.04112849010977086,
      "learning_rate": 6.465926491559697e-06,
      "loss": 0.8328,
      "step": 258010
    },
    {
      "epoch": 4.529925033796239,
      "grad_norm": 0.044113158199913245,
      "learning_rate": 6.4626178669752195e-06,
      "loss": 0.8379,
      "step": 258020
    },
    {
      "epoch": 4.530100598676241,
      "grad_norm": 0.05582904087357936,
      "learning_rate": 6.4593104401974695e-06,
      "loss": 0.8452,
      "step": 258030
    },
    {
      "epoch": 4.530276163556242,
      "grad_norm": 0.061652707091803004,
      "learning_rate": 6.4560042112683805e-06,
      "loss": 0.8432,
      "step": 258040
    },
    {
      "epoch": 4.530451728436244,
      "grad_norm": 0.05191184298141779,
      "learning_rate": 6.45269918022985e-06,
      "loss": 0.8338,
      "step": 258050
    },
    {
      "epoch": 4.530627293316245,
      "grad_norm": 0.04274364686212786,
      "learning_rate": 6.449395347123721e-06,
      "loss": 0.8474,
      "step": 258060
    },
    {
      "epoch": 4.530802858196246,
      "grad_norm": 0.06093273660454719,
      "learning_rate": 6.446092711991894e-06,
      "loss": 0.8365,
      "step": 258070
    },
    {
      "epoch": 4.530978423076248,
      "grad_norm": 0.06526108097574959,
      "learning_rate": 6.442791274876178e-06,
      "loss": 0.8318,
      "step": 258080
    },
    {
      "epoch": 4.531153987956249,
      "grad_norm": 0.043380871103582214,
      "learning_rate": 6.439491035818418e-06,
      "loss": 0.8306,
      "step": 258090
    },
    {
      "epoch": 4.531329552836251,
      "grad_norm": 0.046505622584656986,
      "learning_rate": 6.436191994860444e-06,
      "loss": 0.8407,
      "step": 258100
    },
    {
      "epoch": 4.531505117716252,
      "grad_norm": 0.06484022361763504,
      "learning_rate": 6.4328941520440125e-06,
      "loss": 0.8404,
      "step": 258110
    },
    {
      "epoch": 4.531680682596253,
      "grad_norm": 0.056458781243714985,
      "learning_rate": 6.429597507410957e-06,
      "loss": 0.8475,
      "step": 258120
    },
    {
      "epoch": 4.531856247476255,
      "grad_norm": 0.05132554904543629,
      "learning_rate": 6.426302061003023e-06,
      "loss": 0.8373,
      "step": 258130
    },
    {
      "epoch": 4.532031812356256,
      "grad_norm": 0.051470419299153615,
      "learning_rate": 6.423007812861972e-06,
      "loss": 0.8386,
      "step": 258140
    },
    {
      "epoch": 4.532207377236258,
      "grad_norm": 0.04699862189222076,
      "learning_rate": 6.419714763029534e-06,
      "loss": 0.8395,
      "step": 258150
    },
    {
      "epoch": 4.532382942116259,
      "grad_norm": 0.0524439446787785,
      "learning_rate": 6.416422911547426e-06,
      "loss": 0.8403,
      "step": 258160
    },
    {
      "epoch": 4.532558506996261,
      "grad_norm": 0.06308370229325212,
      "learning_rate": 6.4131322584573855e-06,
      "loss": 0.8374,
      "step": 258170
    },
    {
      "epoch": 4.532734071876262,
      "grad_norm": 0.04442638358957028,
      "learning_rate": 6.4098428038010555e-06,
      "loss": 0.8459,
      "step": 258180
    },
    {
      "epoch": 4.532909636756263,
      "grad_norm": 0.05781057244538682,
      "learning_rate": 6.40655454762017e-06,
      "loss": 0.8412,
      "step": 258190
    },
    {
      "epoch": 4.533085201636265,
      "grad_norm": 0.055437823976226194,
      "learning_rate": 6.403267489956363e-06,
      "loss": 0.8394,
      "step": 258200
    },
    {
      "epoch": 4.533260766516266,
      "grad_norm": 0.04644450546432996,
      "learning_rate": 6.399981630851283e-06,
      "loss": 0.8352,
      "step": 258210
    },
    {
      "epoch": 4.533436331396268,
      "grad_norm": 0.05023993562057638,
      "learning_rate": 6.3966969703465744e-06,
      "loss": 0.8415,
      "step": 258220
    },
    {
      "epoch": 4.533611896276269,
      "grad_norm": 0.055741658363992364,
      "learning_rate": 6.393413508483836e-06,
      "loss": 0.8446,
      "step": 258230
    },
    {
      "epoch": 4.533787461156271,
      "grad_norm": 0.04910165263183902,
      "learning_rate": 6.390131245304695e-06,
      "loss": 0.8391,
      "step": 258240
    },
    {
      "epoch": 4.533963026036272,
      "grad_norm": 0.07297516685599904,
      "learning_rate": 6.386850180850698e-06,
      "loss": 0.8384,
      "step": 258250
    },
    {
      "epoch": 4.534138590916273,
      "grad_norm": 0.06158198495685985,
      "learning_rate": 6.383570315163456e-06,
      "loss": 0.832,
      "step": 258260
    },
    {
      "epoch": 4.534314155796275,
      "grad_norm": 0.04739337567926968,
      "learning_rate": 6.3802916482845174e-06,
      "loss": 0.8398,
      "step": 258270
    },
    {
      "epoch": 4.534489720676276,
      "grad_norm": 0.04644365541254171,
      "learning_rate": 6.377014180255394e-06,
      "loss": 0.8362,
      "step": 258280
    },
    {
      "epoch": 4.534665285556278,
      "grad_norm": 0.04847315748847859,
      "learning_rate": 6.3737379111176584e-06,
      "loss": 0.8306,
      "step": 258290
    },
    {
      "epoch": 4.534840850436279,
      "grad_norm": 0.052753523855847195,
      "learning_rate": 6.370462840912797e-06,
      "loss": 0.8368,
      "step": 258300
    },
    {
      "epoch": 4.53501641531628,
      "grad_norm": 0.06872268249568514,
      "learning_rate": 6.367188969682305e-06,
      "loss": 0.8426,
      "step": 258310
    },
    {
      "epoch": 4.535191980196282,
      "grad_norm": 0.05190077769447328,
      "learning_rate": 6.363916297467665e-06,
      "loss": 0.8288,
      "step": 258320
    },
    {
      "epoch": 4.535367545076283,
      "grad_norm": 0.05659877477786072,
      "learning_rate": 6.36064482431033e-06,
      "loss": 0.8486,
      "step": 258330
    },
    {
      "epoch": 4.535543109956285,
      "grad_norm": 0.05489044538146064,
      "learning_rate": 6.357374550251775e-06,
      "loss": 0.8397,
      "step": 258340
    },
    {
      "epoch": 4.535718674836286,
      "grad_norm": 0.042737012831216434,
      "learning_rate": 6.354105475333414e-06,
      "loss": 0.8429,
      "step": 258350
    },
    {
      "epoch": 4.535894239716287,
      "grad_norm": 0.051165063547542515,
      "learning_rate": 6.350837599596682e-06,
      "loss": 0.8382,
      "step": 258360
    },
    {
      "epoch": 4.536069804596289,
      "grad_norm": 0.08014737001530298,
      "learning_rate": 6.347570923082971e-06,
      "loss": 0.838,
      "step": 258370
    },
    {
      "epoch": 4.53624536947629,
      "grad_norm": 0.07663645659132963,
      "learning_rate": 6.344305445833678e-06,
      "loss": 0.8436,
      "step": 258380
    },
    {
      "epoch": 4.536420934356292,
      "grad_norm": 0.05455254978203809,
      "learning_rate": 6.341041167890193e-06,
      "loss": 0.8325,
      "step": 258390
    },
    {
      "epoch": 4.536596499236293,
      "grad_norm": 0.05414107602183725,
      "learning_rate": 6.337778089293833e-06,
      "loss": 0.8362,
      "step": 258400
    },
    {
      "epoch": 4.5367720641162945,
      "grad_norm": 0.04424690985420161,
      "learning_rate": 6.33451621008599e-06,
      "loss": 0.842,
      "step": 258410
    },
    {
      "epoch": 4.536947628996296,
      "grad_norm": 0.055224547377774005,
      "learning_rate": 6.331255530307958e-06,
      "loss": 0.8381,
      "step": 258420
    },
    {
      "epoch": 4.5371231938762975,
      "grad_norm": 0.05006852295295286,
      "learning_rate": 6.327996050001056e-06,
      "loss": 0.8356,
      "step": 258430
    },
    {
      "epoch": 4.5372987587562985,
      "grad_norm": 0.05109983569842612,
      "learning_rate": 6.3247377692065954e-06,
      "loss": 0.8421,
      "step": 258440
    },
    {
      "epoch": 4.5374743236362995,
      "grad_norm": 0.04881460185788697,
      "learning_rate": 6.321480687965828e-06,
      "loss": 0.835,
      "step": 258450
    },
    {
      "epoch": 4.5376498885163015,
      "grad_norm": 0.05873756913646965,
      "learning_rate": 6.3182248063200805e-06,
      "loss": 0.8382,
      "step": 258460
    },
    {
      "epoch": 4.5378254533963025,
      "grad_norm": 0.053844294816562825,
      "learning_rate": 6.31497012431055e-06,
      "loss": 0.8405,
      "step": 258470
    },
    {
      "epoch": 4.538001018276304,
      "grad_norm": 0.04919332965110138,
      "learning_rate": 6.311716641978506e-06,
      "loss": 0.8384,
      "step": 258480
    },
    {
      "epoch": 4.5381765831563055,
      "grad_norm": 0.05179570062971542,
      "learning_rate": 6.308464359365154e-06,
      "loss": 0.843,
      "step": 258490
    },
    {
      "epoch": 4.5383521480363065,
      "grad_norm": 0.04434693391872347,
      "learning_rate": 6.305213276511704e-06,
      "loss": 0.8312,
      "step": 258500
    },
    {
      "epoch": 4.538527712916308,
      "grad_norm": 0.05076746840655468,
      "learning_rate": 6.3019633934593575e-06,
      "loss": 0.837,
      "step": 258510
    },
    {
      "epoch": 4.538703277796309,
      "grad_norm": 0.04870904003935724,
      "learning_rate": 6.298714710249258e-06,
      "loss": 0.8382,
      "step": 258520
    },
    {
      "epoch": 4.538878842676311,
      "grad_norm": 0.06668367478618586,
      "learning_rate": 6.29546722692262e-06,
      "loss": 0.8372,
      "step": 258530
    },
    {
      "epoch": 4.539054407556312,
      "grad_norm": 0.053036512408864867,
      "learning_rate": 6.292220943520552e-06,
      "loss": 0.8407,
      "step": 258540
    },
    {
      "epoch": 4.539229972436313,
      "grad_norm": 0.08281141935591595,
      "learning_rate": 6.2889758600841954e-06,
      "loss": 0.8346,
      "step": 258550
    },
    {
      "epoch": 4.539405537316315,
      "grad_norm": 0.04809843195497083,
      "learning_rate": 6.285731976654674e-06,
      "loss": 0.841,
      "step": 258560
    },
    {
      "epoch": 4.539581102196316,
      "grad_norm": 0.05392355154352116,
      "learning_rate": 6.2824892932730695e-06,
      "loss": 0.8352,
      "step": 258570
    },
    {
      "epoch": 4.539756667076318,
      "grad_norm": 0.04770645568317571,
      "learning_rate": 6.279247809980493e-06,
      "loss": 0.8357,
      "step": 258580
    },
    {
      "epoch": 4.539932231956319,
      "grad_norm": 0.05531420595169237,
      "learning_rate": 6.276007526817991e-06,
      "loss": 0.8393,
      "step": 258590
    },
    {
      "epoch": 4.54010779683632,
      "grad_norm": 0.04882301701710628,
      "learning_rate": 6.272768443826613e-06,
      "loss": 0.8412,
      "step": 258600
    },
    {
      "epoch": 4.540283361716322,
      "grad_norm": 0.06346319805107963,
      "learning_rate": 6.269530561047444e-06,
      "loss": 0.8347,
      "step": 258610
    },
    {
      "epoch": 4.540458926596323,
      "grad_norm": 0.044801562065619624,
      "learning_rate": 6.266293878521438e-06,
      "loss": 0.8412,
      "step": 258620
    },
    {
      "epoch": 4.540634491476325,
      "grad_norm": 0.04919694380774686,
      "learning_rate": 6.263058396289681e-06,
      "loss": 0.8437,
      "step": 258630
    },
    {
      "epoch": 4.540810056356326,
      "grad_norm": 0.047283620493168206,
      "learning_rate": 6.259824114393102e-06,
      "loss": 0.8431,
      "step": 258640
    },
    {
      "epoch": 4.540985621236328,
      "grad_norm": 0.05516595888685987,
      "learning_rate": 6.256591032872733e-06,
      "loss": 0.8414,
      "step": 258650
    },
    {
      "epoch": 4.541161186116329,
      "grad_norm": 0.05716566974476635,
      "learning_rate": 6.253359151769506e-06,
      "loss": 0.8416,
      "step": 258660
    },
    {
      "epoch": 4.541336750996331,
      "grad_norm": 0.055147490383936966,
      "learning_rate": 6.25012847112436e-06,
      "loss": 0.8393,
      "step": 258670
    },
    {
      "epoch": 4.541512315876332,
      "grad_norm": 0.0611267836572989,
      "learning_rate": 6.2468989909782724e-06,
      "loss": 0.8378,
      "step": 258680
    },
    {
      "epoch": 4.541687880756333,
      "grad_norm": 0.0690722910826504,
      "learning_rate": 6.2436707113720975e-06,
      "loss": 0.8395,
      "step": 258690
    },
    {
      "epoch": 4.541863445636335,
      "grad_norm": 0.04783580549515059,
      "learning_rate": 6.24044363234681e-06,
      "loss": 0.8357,
      "step": 258700
    },
    {
      "epoch": 4.542039010516336,
      "grad_norm": 0.04892832847340393,
      "learning_rate": 6.237217753943244e-06,
      "loss": 0.8407,
      "step": 258710
    },
    {
      "epoch": 4.542214575396338,
      "grad_norm": 0.05739701599485136,
      "learning_rate": 6.233993076202295e-06,
      "loss": 0.8416,
      "step": 258720
    },
    {
      "epoch": 4.542390140276339,
      "grad_norm": 0.04997391924325357,
      "learning_rate": 6.230769599164818e-06,
      "loss": 0.8467,
      "step": 258730
    },
    {
      "epoch": 4.54256570515634,
      "grad_norm": 0.056242343897485676,
      "learning_rate": 6.227547322871647e-06,
      "loss": 0.8379,
      "step": 258740
    },
    {
      "epoch": 4.542741270036342,
      "grad_norm": 0.04747605977614506,
      "learning_rate": 6.224326247363625e-06,
      "loss": 0.8403,
      "step": 258750
    },
    {
      "epoch": 4.542916834916343,
      "grad_norm": 0.06948122107070676,
      "learning_rate": 6.221106372681549e-06,
      "loss": 0.8312,
      "step": 258760
    },
    {
      "epoch": 4.543092399796345,
      "grad_norm": 0.058444218659340345,
      "learning_rate": 6.217887698866211e-06,
      "loss": 0.8333,
      "step": 258770
    },
    {
      "epoch": 4.543267964676346,
      "grad_norm": 0.06803311508451354,
      "learning_rate": 6.2146702259584185e-06,
      "loss": 0.8375,
      "step": 258780
    },
    {
      "epoch": 4.543443529556347,
      "grad_norm": 0.06047064063841801,
      "learning_rate": 6.211453953998883e-06,
      "loss": 0.841,
      "step": 258790
    },
    {
      "epoch": 4.543619094436349,
      "grad_norm": 0.04695808763661924,
      "learning_rate": 6.208238883028428e-06,
      "loss": 0.844,
      "step": 258800
    },
    {
      "epoch": 4.54379465931635,
      "grad_norm": 0.04642524793803961,
      "learning_rate": 6.205025013087741e-06,
      "loss": 0.8386,
      "step": 258810
    },
    {
      "epoch": 4.543970224196352,
      "grad_norm": 0.05048710834300065,
      "learning_rate": 6.201812344217556e-06,
      "loss": 0.8373,
      "step": 258820
    },
    {
      "epoch": 4.544145789076353,
      "grad_norm": 0.06689854134867647,
      "learning_rate": 6.198600876458573e-06,
      "loss": 0.8376,
      "step": 258830
    },
    {
      "epoch": 4.544321353956355,
      "grad_norm": 0.07432313989923793,
      "learning_rate": 6.195390609851483e-06,
      "loss": 0.8314,
      "step": 258840
    },
    {
      "epoch": 4.544496918836356,
      "grad_norm": 0.05006862727867589,
      "learning_rate": 6.192181544436985e-06,
      "loss": 0.8283,
      "step": 258850
    },
    {
      "epoch": 4.544672483716357,
      "grad_norm": 0.05383384917774438,
      "learning_rate": 6.188973680255681e-06,
      "loss": 0.8287,
      "step": 258860
    },
    {
      "epoch": 4.544848048596359,
      "grad_norm": 0.04598121662673226,
      "learning_rate": 6.1857670173482816e-06,
      "loss": 0.8396,
      "step": 258870
    },
    {
      "epoch": 4.54502361347636,
      "grad_norm": 0.05670985155003879,
      "learning_rate": 6.182561555755368e-06,
      "loss": 0.8497,
      "step": 258880
    },
    {
      "epoch": 4.545199178356362,
      "grad_norm": 0.04642859242453754,
      "learning_rate": 6.179357295517574e-06,
      "loss": 0.8347,
      "step": 258890
    },
    {
      "epoch": 4.545374743236363,
      "grad_norm": 0.051149482097598196,
      "learning_rate": 6.1761542366755e-06,
      "loss": 0.8368,
      "step": 258900
    },
    {
      "epoch": 4.545550308116365,
      "grad_norm": 0.055800170586550076,
      "learning_rate": 6.172952379269717e-06,
      "loss": 0.8305,
      "step": 258910
    },
    {
      "epoch": 4.545725872996366,
      "grad_norm": 0.05238397160071007,
      "learning_rate": 6.1697517233408224e-06,
      "loss": 0.8341,
      "step": 258920
    },
    {
      "epoch": 4.545901437876367,
      "grad_norm": 0.26125304684616146,
      "learning_rate": 6.166552268929323e-06,
      "loss": 0.8406,
      "step": 258930
    },
    {
      "epoch": 4.546077002756369,
      "grad_norm": 0.03939678976118692,
      "learning_rate": 6.163354016075785e-06,
      "loss": 0.8416,
      "step": 258940
    },
    {
      "epoch": 4.54625256763637,
      "grad_norm": 0.046159556229806484,
      "learning_rate": 6.160156964820746e-06,
      "loss": 0.8416,
      "step": 258950
    },
    {
      "epoch": 4.546428132516372,
      "grad_norm": 0.05117020138329338,
      "learning_rate": 6.156961115204663e-06,
      "loss": 0.8406,
      "step": 258960
    },
    {
      "epoch": 4.546603697396373,
      "grad_norm": 0.0476919549280704,
      "learning_rate": 6.153766467268082e-06,
      "loss": 0.8322,
      "step": 258970
    },
    {
      "epoch": 4.546779262276374,
      "grad_norm": 0.0551141991437599,
      "learning_rate": 6.150573021051452e-06,
      "loss": 0.8436,
      "step": 258980
    },
    {
      "epoch": 4.546954827156376,
      "grad_norm": 0.05273284877743412,
      "learning_rate": 6.147380776595242e-06,
      "loss": 0.8416,
      "step": 258990
    },
    {
      "epoch": 4.547130392036377,
      "grad_norm": 0.05250306184166529,
      "learning_rate": 6.144189733939889e-06,
      "loss": 0.8396,
      "step": 259000
    },
    {
      "epoch": 4.547305956916379,
      "grad_norm": 0.050003757141442666,
      "learning_rate": 6.14099989312583e-06,
      "loss": 0.8336,
      "step": 259010
    },
    {
      "epoch": 4.54748152179638,
      "grad_norm": 0.04818139439068836,
      "learning_rate": 6.137811254193501e-06,
      "loss": 0.839,
      "step": 259020
    },
    {
      "epoch": 4.547657086676381,
      "grad_norm": 0.05489535911456474,
      "learning_rate": 6.13462381718325e-06,
      "loss": 0.8309,
      "step": 259030
    },
    {
      "epoch": 4.547832651556383,
      "grad_norm": 0.05508860550218829,
      "learning_rate": 6.1314375821355266e-06,
      "loss": 0.8407,
      "step": 259040
    },
    {
      "epoch": 4.548008216436384,
      "grad_norm": 0.045802800394279364,
      "learning_rate": 6.128252549090655e-06,
      "loss": 0.8348,
      "step": 259050
    },
    {
      "epoch": 4.548183781316386,
      "grad_norm": 0.05259928073489056,
      "learning_rate": 6.125068718089008e-06,
      "loss": 0.8351,
      "step": 259060
    },
    {
      "epoch": 4.548359346196387,
      "grad_norm": 0.048405345211736846,
      "learning_rate": 6.121886089170934e-06,
      "loss": 0.84,
      "step": 259070
    },
    {
      "epoch": 4.548534911076389,
      "grad_norm": 0.04746045558550393,
      "learning_rate": 6.118704662376738e-06,
      "loss": 0.8383,
      "step": 259080
    },
    {
      "epoch": 4.54871047595639,
      "grad_norm": 0.05448188592320247,
      "learning_rate": 6.115524437746746e-06,
      "loss": 0.8368,
      "step": 259090
    },
    {
      "epoch": 4.548886040836392,
      "grad_norm": 0.04895468540030248,
      "learning_rate": 6.112345415321229e-06,
      "loss": 0.8343,
      "step": 259100
    },
    {
      "epoch": 4.549061605716393,
      "grad_norm": 0.04899103559052429,
      "learning_rate": 6.1091675951404814e-06,
      "loss": 0.8334,
      "step": 259110
    },
    {
      "epoch": 4.549237170596394,
      "grad_norm": 0.04968612384310643,
      "learning_rate": 6.105990977244788e-06,
      "loss": 0.8354,
      "step": 259120
    },
    {
      "epoch": 4.549412735476396,
      "grad_norm": 0.06716369289948639,
      "learning_rate": 6.102815561674341e-06,
      "loss": 0.8325,
      "step": 259130
    },
    {
      "epoch": 4.549588300356397,
      "grad_norm": 0.05579920746798574,
      "learning_rate": 6.099641348469435e-06,
      "loss": 0.8476,
      "step": 259140
    },
    {
      "epoch": 4.5497638652363985,
      "grad_norm": 0.0482599611702587,
      "learning_rate": 6.096468337670243e-06,
      "loss": 0.8375,
      "step": 259150
    },
    {
      "epoch": 4.5499394301164,
      "grad_norm": 0.05677830108563702,
      "learning_rate": 6.093296529317005e-06,
      "loss": 0.8429,
      "step": 259160
    },
    {
      "epoch": 4.550114994996401,
      "grad_norm": 0.04991339550410038,
      "learning_rate": 6.090125923449869e-06,
      "loss": 0.8366,
      "step": 259170
    },
    {
      "epoch": 4.5502905598764025,
      "grad_norm": 0.1234072788418804,
      "learning_rate": 6.086956520109033e-06,
      "loss": 0.8394,
      "step": 259180
    },
    {
      "epoch": 4.5504661247564036,
      "grad_norm": 0.05287114628757094,
      "learning_rate": 6.083788319334667e-06,
      "loss": 0.8328,
      "step": 259190
    },
    {
      "epoch": 4.5506416896364055,
      "grad_norm": 0.056840768136038645,
      "learning_rate": 6.080621321166858e-06,
      "loss": 0.8343,
      "step": 259200
    },
    {
      "epoch": 4.5508172545164065,
      "grad_norm": 0.043763598292643334,
      "learning_rate": 6.0774555256458e-06,
      "loss": 0.8396,
      "step": 259210
    },
    {
      "epoch": 4.5509928193964075,
      "grad_norm": 0.05245497926377255,
      "learning_rate": 6.074290932811557e-06,
      "loss": 0.8457,
      "step": 259220
    },
    {
      "epoch": 4.5511683842764095,
      "grad_norm": 0.050417519441191605,
      "learning_rate": 6.0711275427042346e-06,
      "loss": 0.8319,
      "step": 259230
    },
    {
      "epoch": 4.5513439491564105,
      "grad_norm": 0.07190467630787786,
      "learning_rate": 6.0679653553639395e-06,
      "loss": 0.8388,
      "step": 259240
    },
    {
      "epoch": 4.551519514036412,
      "grad_norm": 0.05276945564900768,
      "learning_rate": 6.064804370830703e-06,
      "loss": 0.8395,
      "step": 259250
    },
    {
      "epoch": 4.5516950789164135,
      "grad_norm": 0.0494674696048589,
      "learning_rate": 6.061644589144598e-06,
      "loss": 0.8343,
      "step": 259260
    },
    {
      "epoch": 4.5518706437964145,
      "grad_norm": 0.08310166028530845,
      "learning_rate": 6.058486010345645e-06,
      "loss": 0.8381,
      "step": 259270
    },
    {
      "epoch": 4.552046208676416,
      "grad_norm": 0.038095251486930146,
      "learning_rate": 6.0553286344738605e-06,
      "loss": 0.8434,
      "step": 259280
    },
    {
      "epoch": 4.552221773556417,
      "grad_norm": 0.05838608885369294,
      "learning_rate": 6.052172461569277e-06,
      "loss": 0.8437,
      "step": 259290
    },
    {
      "epoch": 4.552397338436419,
      "grad_norm": 0.04474900962449523,
      "learning_rate": 6.0490174916718335e-06,
      "loss": 0.8274,
      "step": 259300
    },
    {
      "epoch": 4.55257290331642,
      "grad_norm": 0.046745786412668563,
      "learning_rate": 6.045863724821575e-06,
      "loss": 0.8327,
      "step": 259310
    },
    {
      "epoch": 4.552748468196422,
      "grad_norm": 0.04869502906108674,
      "learning_rate": 6.042711161058396e-06,
      "loss": 0.8373,
      "step": 259320
    },
    {
      "epoch": 4.552924033076423,
      "grad_norm": 0.05234288238250763,
      "learning_rate": 6.0395598004222845e-06,
      "loss": 0.8384,
      "step": 259330
    },
    {
      "epoch": 4.553099597956425,
      "grad_norm": 0.05150865756307625,
      "learning_rate": 6.036409642953128e-06,
      "loss": 0.8458,
      "step": 259340
    },
    {
      "epoch": 4.553275162836426,
      "grad_norm": 0.055847420899608514,
      "learning_rate": 6.033260688690866e-06,
      "loss": 0.8378,
      "step": 259350
    },
    {
      "epoch": 4.553450727716427,
      "grad_norm": 0.05732461644864,
      "learning_rate": 6.0301129376754096e-06,
      "loss": 0.8342,
      "step": 259360
    },
    {
      "epoch": 4.553626292596429,
      "grad_norm": 0.04440734318248186,
      "learning_rate": 6.0269663899465915e-06,
      "loss": 0.8359,
      "step": 259370
    },
    {
      "epoch": 4.55380185747643,
      "grad_norm": 0.08849749081337754,
      "learning_rate": 6.0238210455443295e-06,
      "loss": 0.8394,
      "step": 259380
    },
    {
      "epoch": 4.553977422356432,
      "grad_norm": 0.05424606060154002,
      "learning_rate": 6.0206769045084585e-06,
      "loss": 0.8446,
      "step": 259390
    },
    {
      "epoch": 4.554152987236433,
      "grad_norm": 0.07015791844447881,
      "learning_rate": 6.017533966878808e-06,
      "loss": 0.8447,
      "step": 259400
    },
    {
      "epoch": 4.554328552116434,
      "grad_norm": 0.057619771915141976,
      "learning_rate": 6.014392232695212e-06,
      "loss": 0.8494,
      "step": 259410
    },
    {
      "epoch": 4.554504116996436,
      "grad_norm": 0.0474802928253836,
      "learning_rate": 6.011251701997468e-06,
      "loss": 0.8391,
      "step": 259420
    },
    {
      "epoch": 4.554679681876437,
      "grad_norm": 0.06028191374746356,
      "learning_rate": 6.008112374825377e-06,
      "loss": 0.8414,
      "step": 259430
    },
    {
      "epoch": 4.554855246756439,
      "grad_norm": 0.042899421382947826,
      "learning_rate": 6.004974251218693e-06,
      "loss": 0.8382,
      "step": 259440
    },
    {
      "epoch": 4.55503081163644,
      "grad_norm": 0.05489864772767699,
      "learning_rate": 6.001837331217194e-06,
      "loss": 0.8383,
      "step": 259450
    },
    {
      "epoch": 4.555206376516441,
      "grad_norm": 0.06755220005296364,
      "learning_rate": 5.998701614860635e-06,
      "loss": 0.832,
      "step": 259460
    },
    {
      "epoch": 4.555381941396443,
      "grad_norm": 0.08235307663685043,
      "learning_rate": 5.995567102188704e-06,
      "loss": 0.8373,
      "step": 259470
    },
    {
      "epoch": 4.555557506276444,
      "grad_norm": 0.05668591964738623,
      "learning_rate": 5.992433793241179e-06,
      "loss": 0.8315,
      "step": 259480
    },
    {
      "epoch": 4.555733071156446,
      "grad_norm": 0.05253155810244448,
      "learning_rate": 5.9893016880577056e-06,
      "loss": 0.8418,
      "step": 259490
    },
    {
      "epoch": 4.555908636036447,
      "grad_norm": 0.037676356200062676,
      "learning_rate": 5.986170786678007e-06,
      "loss": 0.8376,
      "step": 259500
    },
    {
      "epoch": 4.556084200916449,
      "grad_norm": 0.0474889695283454,
      "learning_rate": 5.983041089141715e-06,
      "loss": 0.8381,
      "step": 259510
    },
    {
      "epoch": 4.55625976579645,
      "grad_norm": 0.05350610133714206,
      "learning_rate": 5.9799125954884985e-06,
      "loss": 0.8469,
      "step": 259520
    },
    {
      "epoch": 4.556435330676451,
      "grad_norm": 0.050498990256105505,
      "learning_rate": 5.9767853057580255e-06,
      "loss": 0.8386,
      "step": 259530
    },
    {
      "epoch": 4.556610895556453,
      "grad_norm": 0.05579717583665096,
      "learning_rate": 5.973659219989863e-06,
      "loss": 0.8455,
      "step": 259540
    },
    {
      "epoch": 4.556786460436454,
      "grad_norm": 0.047855928435461666,
      "learning_rate": 5.970534338223667e-06,
      "loss": 0.8421,
      "step": 259550
    },
    {
      "epoch": 4.556962025316456,
      "grad_norm": 0.06152928240615446,
      "learning_rate": 5.9674106604990176e-06,
      "loss": 0.8351,
      "step": 259560
    },
    {
      "epoch": 4.557137590196457,
      "grad_norm": 0.05118113007659675,
      "learning_rate": 5.964288186855472e-06,
      "loss": 0.8396,
      "step": 259570
    },
    {
      "epoch": 4.557313155076459,
      "grad_norm": 0.06862592946548947,
      "learning_rate": 5.961166917332632e-06,
      "loss": 0.8337,
      "step": 259580
    },
    {
      "epoch": 4.55748871995646,
      "grad_norm": 0.07110519223543639,
      "learning_rate": 5.9580468519699994e-06,
      "loss": 0.8423,
      "step": 259590
    },
    {
      "epoch": 4.557664284836461,
      "grad_norm": 0.059993035534576786,
      "learning_rate": 5.954927990807142e-06,
      "loss": 0.8451,
      "step": 259600
    },
    {
      "epoch": 4.557839849716463,
      "grad_norm": 0.057351776356578614,
      "learning_rate": 5.951810333883564e-06,
      "loss": 0.8359,
      "step": 259610
    },
    {
      "epoch": 4.558015414596464,
      "grad_norm": 0.0915054062290721,
      "learning_rate": 5.9486938812387435e-06,
      "loss": 0.8355,
      "step": 259620
    },
    {
      "epoch": 4.558190979476466,
      "grad_norm": 0.04652606697655266,
      "learning_rate": 5.945578632912218e-06,
      "loss": 0.8433,
      "step": 259630
    },
    {
      "epoch": 4.558366544356467,
      "grad_norm": 0.06113396298067097,
      "learning_rate": 5.94246458894339e-06,
      "loss": 0.8371,
      "step": 259640
    },
    {
      "epoch": 4.558542109236468,
      "grad_norm": 0.05043048114466396,
      "learning_rate": 5.9393517493717946e-06,
      "loss": 0.8369,
      "step": 259650
    },
    {
      "epoch": 4.55871767411647,
      "grad_norm": 0.07003155924434468,
      "learning_rate": 5.936240114236803e-06,
      "loss": 0.8416,
      "step": 259660
    },
    {
      "epoch": 4.558893238996471,
      "grad_norm": 0.05524450149841554,
      "learning_rate": 5.933129683577883e-06,
      "loss": 0.8419,
      "step": 259670
    },
    {
      "epoch": 4.559068803876473,
      "grad_norm": 0.04731454740649935,
      "learning_rate": 5.9300204574344194e-06,
      "loss": 0.8399,
      "step": 259680
    },
    {
      "epoch": 4.559244368756474,
      "grad_norm": 0.05855033725146765,
      "learning_rate": 5.926912435845824e-06,
      "loss": 0.8423,
      "step": 259690
    },
    {
      "epoch": 4.559419933636475,
      "grad_norm": 0.12059557496529623,
      "learning_rate": 5.923805618851467e-06,
      "loss": 0.8378,
      "step": 259700
    },
    {
      "epoch": 4.559595498516477,
      "grad_norm": 0.04667376808141787,
      "learning_rate": 5.920700006490699e-06,
      "loss": 0.8427,
      "step": 259710
    },
    {
      "epoch": 4.559771063396478,
      "grad_norm": 0.056482859829440374,
      "learning_rate": 5.917595598802911e-06,
      "loss": 0.8411,
      "step": 259720
    },
    {
      "epoch": 4.55994662827648,
      "grad_norm": 0.05813660239062548,
      "learning_rate": 5.9144923958273855e-06,
      "loss": 0.8395,
      "step": 259730
    },
    {
      "epoch": 4.560122193156481,
      "grad_norm": 0.04655029611685512,
      "learning_rate": 5.911390397603472e-06,
      "loss": 0.8474,
      "step": 259740
    },
    {
      "epoch": 4.560297758036483,
      "grad_norm": 0.04267522967129286,
      "learning_rate": 5.9082896041704745e-06,
      "loss": 0.8335,
      "step": 259750
    },
    {
      "epoch": 4.560473322916484,
      "grad_norm": 0.049215249556880065,
      "learning_rate": 5.905190015567664e-06,
      "loss": 0.8409,
      "step": 259760
    },
    {
      "epoch": 4.560648887796485,
      "grad_norm": 0.07741266561894908,
      "learning_rate": 5.9020916318343355e-06,
      "loss": 0.834,
      "step": 259770
    },
    {
      "epoch": 4.560824452676487,
      "grad_norm": 0.05372163317538071,
      "learning_rate": 5.898994453009715e-06,
      "loss": 0.843,
      "step": 259780
    },
    {
      "epoch": 4.561000017556488,
      "grad_norm": 0.06836553403045616,
      "learning_rate": 5.8958984791330535e-06,
      "loss": 0.838,
      "step": 259790
    },
    {
      "epoch": 4.56117558243649,
      "grad_norm": 0.060770418414859986,
      "learning_rate": 5.892803710243612e-06,
      "loss": 0.8425,
      "step": 259800
    },
    {
      "epoch": 4.561351147316491,
      "grad_norm": 0.05486604393426086,
      "learning_rate": 5.889710146380539e-06,
      "loss": 0.8376,
      "step": 259810
    },
    {
      "epoch": 4.561526712196493,
      "grad_norm": 0.04571282076716153,
      "learning_rate": 5.886617787583096e-06,
      "loss": 0.8361,
      "step": 259820
    },
    {
      "epoch": 4.561702277076494,
      "grad_norm": 0.05835412689362552,
      "learning_rate": 5.883526633890423e-06,
      "loss": 0.8385,
      "step": 259830
    },
    {
      "epoch": 4.561877841956495,
      "grad_norm": 0.06142850726714548,
      "learning_rate": 5.880436685341704e-06,
      "loss": 0.8348,
      "step": 259840
    },
    {
      "epoch": 4.562053406836497,
      "grad_norm": 0.056516976099498475,
      "learning_rate": 5.877347941976067e-06,
      "loss": 0.8379,
      "step": 259850
    },
    {
      "epoch": 4.562228971716498,
      "grad_norm": 0.05821057668122707,
      "learning_rate": 5.874260403832674e-06,
      "loss": 0.8419,
      "step": 259860
    },
    {
      "epoch": 4.5624045365965,
      "grad_norm": 0.06136540846989266,
      "learning_rate": 5.871174070950633e-06,
      "loss": 0.8385,
      "step": 259870
    },
    {
      "epoch": 4.562580101476501,
      "grad_norm": 0.05896038880468855,
      "learning_rate": 5.868088943369026e-06,
      "loss": 0.84,
      "step": 259880
    },
    {
      "epoch": 4.562755666356502,
      "grad_norm": 0.048431324203224886,
      "learning_rate": 5.865005021126995e-06,
      "loss": 0.8376,
      "step": 259890
    },
    {
      "epoch": 4.562931231236504,
      "grad_norm": 0.04587101381573632,
      "learning_rate": 5.861922304263558e-06,
      "loss": 0.8444,
      "step": 259900
    },
    {
      "epoch": 4.563106796116505,
      "grad_norm": 0.04329191342743193,
      "learning_rate": 5.85884079281781e-06,
      "loss": 0.8367,
      "step": 259910
    },
    {
      "epoch": 4.5632823609965065,
      "grad_norm": 0.04840083729593486,
      "learning_rate": 5.855760486828792e-06,
      "loss": 0.839,
      "step": 259920
    },
    {
      "epoch": 4.563457925876508,
      "grad_norm": 0.04945505197100781,
      "learning_rate": 5.852681386335513e-06,
      "loss": 0.8379,
      "step": 259930
    },
    {
      "epoch": 4.563633490756509,
      "grad_norm": 0.05303354974528548,
      "learning_rate": 5.8496034913770024e-06,
      "loss": 0.8455,
      "step": 259940
    },
    {
      "epoch": 4.5638090556365105,
      "grad_norm": 0.05203008426170722,
      "learning_rate": 5.846526801992234e-06,
      "loss": 0.8397,
      "step": 259950
    },
    {
      "epoch": 4.5639846205165115,
      "grad_norm": 0.05037883400398477,
      "learning_rate": 5.8434513182202165e-06,
      "loss": 0.8426,
      "step": 259960
    },
    {
      "epoch": 4.5641601853965135,
      "grad_norm": 0.05174587287591693,
      "learning_rate": 5.840377040099924e-06,
      "loss": 0.8425,
      "step": 259970
    },
    {
      "epoch": 4.5643357502765145,
      "grad_norm": 0.06573207660199426,
      "learning_rate": 5.837303967670255e-06,
      "loss": 0.8351,
      "step": 259980
    },
    {
      "epoch": 4.564511315156516,
      "grad_norm": 0.056205678887750014,
      "learning_rate": 5.834232100970218e-06,
      "loss": 0.837,
      "step": 259990
    },
    {
      "epoch": 4.5646868800365175,
      "grad_norm": 0.04510443221945392,
      "learning_rate": 5.831161440038676e-06,
      "loss": 0.8456,
      "step": 260000
    },
    {
      "epoch": 4.564862444916519,
      "grad_norm": 0.04111026328811391,
      "learning_rate": 5.828091984914572e-06,
      "loss": 0.8406,
      "step": 260010
    },
    {
      "epoch": 4.56503800979652,
      "grad_norm": 0.05264323750378233,
      "learning_rate": 5.825023735636782e-06,
      "loss": 0.8396,
      "step": 260020
    },
    {
      "epoch": 4.5652135746765214,
      "grad_norm": 0.051180000244255335,
      "learning_rate": 5.821956692244171e-06,
      "loss": 0.8393,
      "step": 260030
    },
    {
      "epoch": 4.565389139556523,
      "grad_norm": 0.05839756918855802,
      "learning_rate": 5.818890854775626e-06,
      "loss": 0.8413,
      "step": 260040
    },
    {
      "epoch": 4.565564704436524,
      "grad_norm": 0.05204214139320574,
      "learning_rate": 5.8158262232699445e-06,
      "loss": 0.8315,
      "step": 260050
    },
    {
      "epoch": 4.565740269316526,
      "grad_norm": 0.048009328800942726,
      "learning_rate": 5.812762797766015e-06,
      "loss": 0.8372,
      "step": 260060
    },
    {
      "epoch": 4.565915834196527,
      "grad_norm": 0.0435609533119045,
      "learning_rate": 5.809700578302614e-06,
      "loss": 0.8344,
      "step": 260070
    },
    {
      "epoch": 4.566091399076528,
      "grad_norm": 0.05508855716753935,
      "learning_rate": 5.80663956491854e-06,
      "loss": 0.8363,
      "step": 260080
    },
    {
      "epoch": 4.56626696395653,
      "grad_norm": 0.048789713119665276,
      "learning_rate": 5.803579757652604e-06,
      "loss": 0.8357,
      "step": 260090
    },
    {
      "epoch": 4.566442528836531,
      "grad_norm": 0.04557847417552077,
      "learning_rate": 5.8005211565435485e-06,
      "loss": 0.8369,
      "step": 260100
    },
    {
      "epoch": 4.566618093716533,
      "grad_norm": 0.045716955668551724,
      "learning_rate": 5.797463761630141e-06,
      "loss": 0.8404,
      "step": 260110
    },
    {
      "epoch": 4.566793658596534,
      "grad_norm": 0.055495404730600856,
      "learning_rate": 5.794407572951103e-06,
      "loss": 0.8302,
      "step": 260120
    },
    {
      "epoch": 4.566969223476535,
      "grad_norm": 0.05975511515104652,
      "learning_rate": 5.791352590545178e-06,
      "loss": 0.8437,
      "step": 260130
    },
    {
      "epoch": 4.567144788356537,
      "grad_norm": 0.05086446898446266,
      "learning_rate": 5.7882988144510555e-06,
      "loss": 0.8368,
      "step": 260140
    },
    {
      "epoch": 4.567320353236538,
      "grad_norm": 0.06079318770575196,
      "learning_rate": 5.785246244707425e-06,
      "loss": 0.8404,
      "step": 260150
    },
    {
      "epoch": 4.56749591811654,
      "grad_norm": 0.04696315597790492,
      "learning_rate": 5.7821948813529856e-06,
      "loss": 0.8373,
      "step": 260160
    },
    {
      "epoch": 4.567671482996541,
      "grad_norm": 0.044833973031519855,
      "learning_rate": 5.779144724426383e-06,
      "loss": 0.8374,
      "step": 260170
    },
    {
      "epoch": 4.567847047876542,
      "grad_norm": 0.046225774442482714,
      "learning_rate": 5.776095773966273e-06,
      "loss": 0.8439,
      "step": 260180
    },
    {
      "epoch": 4.568022612756544,
      "grad_norm": 0.051976248093320625,
      "learning_rate": 5.773048030011268e-06,
      "loss": 0.8407,
      "step": 260190
    },
    {
      "epoch": 4.568198177636545,
      "grad_norm": 0.0471567032357954,
      "learning_rate": 5.7700014925999904e-06,
      "loss": 0.8381,
      "step": 260200
    },
    {
      "epoch": 4.568373742516547,
      "grad_norm": 0.06639301444856582,
      "learning_rate": 5.766956161771064e-06,
      "loss": 0.8382,
      "step": 260210
    },
    {
      "epoch": 4.568549307396548,
      "grad_norm": 0.05123168655164977,
      "learning_rate": 5.763912037563024e-06,
      "loss": 0.8401,
      "step": 260220
    },
    {
      "epoch": 4.56872487227655,
      "grad_norm": 0.052113740574395856,
      "learning_rate": 5.7608691200145046e-06,
      "loss": 0.8382,
      "step": 260230
    },
    {
      "epoch": 4.568900437156551,
      "grad_norm": 0.055454847619952466,
      "learning_rate": 5.7578274091640076e-06,
      "loss": 0.8379,
      "step": 260240
    },
    {
      "epoch": 4.569076002036553,
      "grad_norm": 0.05422100983372784,
      "learning_rate": 5.75478690505009e-06,
      "loss": 0.8373,
      "step": 260250
    },
    {
      "epoch": 4.569251566916554,
      "grad_norm": 0.04416213261994837,
      "learning_rate": 5.751747607711287e-06,
      "loss": 0.8354,
      "step": 260260
    },
    {
      "epoch": 4.569427131796555,
      "grad_norm": 0.05320371268926052,
      "learning_rate": 5.74870951718608e-06,
      "loss": 0.8378,
      "step": 260270
    },
    {
      "epoch": 4.569602696676557,
      "grad_norm": 0.05281328323303454,
      "learning_rate": 5.745672633512991e-06,
      "loss": 0.8368,
      "step": 260280
    },
    {
      "epoch": 4.569778261556558,
      "grad_norm": 0.07644337514668471,
      "learning_rate": 5.74263695673048e-06,
      "loss": 0.8389,
      "step": 260290
    },
    {
      "epoch": 4.56995382643656,
      "grad_norm": 0.049935541121361254,
      "learning_rate": 5.739602486877016e-06,
      "loss": 0.8391,
      "step": 260300
    },
    {
      "epoch": 4.570129391316561,
      "grad_norm": 0.04775725200573514,
      "learning_rate": 5.736569223991045e-06,
      "loss": 0.8362,
      "step": 260310
    },
    {
      "epoch": 4.570304956196562,
      "grad_norm": 0.06587688618628366,
      "learning_rate": 5.733537168110984e-06,
      "loss": 0.8386,
      "step": 260320
    },
    {
      "epoch": 4.570480521076564,
      "grad_norm": 0.0639672920310937,
      "learning_rate": 5.730506319275288e-06,
      "loss": 0.8429,
      "step": 260330
    },
    {
      "epoch": 4.570656085956565,
      "grad_norm": 0.0675525179006928,
      "learning_rate": 5.727476677522329e-06,
      "loss": 0.8416,
      "step": 260340
    },
    {
      "epoch": 4.570831650836567,
      "grad_norm": 0.05391727357958606,
      "learning_rate": 5.724448242890499e-06,
      "loss": 0.8403,
      "step": 260350
    },
    {
      "epoch": 4.571007215716568,
      "grad_norm": 0.047049543248466545,
      "learning_rate": 5.721421015418169e-06,
      "loss": 0.8334,
      "step": 260360
    },
    {
      "epoch": 4.571182780596569,
      "grad_norm": 0.05198565436853326,
      "learning_rate": 5.718394995143686e-06,
      "loss": 0.8423,
      "step": 260370
    },
    {
      "epoch": 4.571358345476571,
      "grad_norm": 0.05053279719364805,
      "learning_rate": 5.715370182105411e-06,
      "loss": 0.8402,
      "step": 260380
    },
    {
      "epoch": 4.571533910356572,
      "grad_norm": 0.04574154490394495,
      "learning_rate": 5.712346576341635e-06,
      "loss": 0.8431,
      "step": 260390
    },
    {
      "epoch": 4.571709475236574,
      "grad_norm": 0.06027303008679638,
      "learning_rate": 5.709324177890719e-06,
      "loss": 0.8312,
      "step": 260400
    },
    {
      "epoch": 4.571885040116575,
      "grad_norm": 0.049681323681907044,
      "learning_rate": 5.706302986790912e-06,
      "loss": 0.8376,
      "step": 260410
    },
    {
      "epoch": 4.572060604996577,
      "grad_norm": 0.0538793099131825,
      "learning_rate": 5.703283003080508e-06,
      "loss": 0.8346,
      "step": 260420
    },
    {
      "epoch": 4.572236169876578,
      "grad_norm": 0.046053285740096095,
      "learning_rate": 5.700264226797789e-06,
      "loss": 0.8335,
      "step": 260430
    },
    {
      "epoch": 4.572411734756579,
      "grad_norm": 0.05574316191098782,
      "learning_rate": 5.697246657980971e-06,
      "loss": 0.8456,
      "step": 260440
    },
    {
      "epoch": 4.572587299636581,
      "grad_norm": 0.04901277015552072,
      "learning_rate": 5.694230296668314e-06,
      "loss": 0.8422,
      "step": 260450
    },
    {
      "epoch": 4.572762864516582,
      "grad_norm": 0.04849111246306439,
      "learning_rate": 5.691215142898015e-06,
      "loss": 0.8447,
      "step": 260460
    },
    {
      "epoch": 4.572938429396584,
      "grad_norm": 0.07516373297964565,
      "learning_rate": 5.688201196708276e-06,
      "loss": 0.8356,
      "step": 260470
    },
    {
      "epoch": 4.573113994276585,
      "grad_norm": 0.053558684518552314,
      "learning_rate": 5.6851884581373155e-06,
      "loss": 0.8409,
      "step": 260480
    },
    {
      "epoch": 4.573289559156587,
      "grad_norm": 0.05783665783449839,
      "learning_rate": 5.682176927223261e-06,
      "loss": 0.8416,
      "step": 260490
    },
    {
      "epoch": 4.573465124036588,
      "grad_norm": 0.04518310450602686,
      "learning_rate": 5.679166604004307e-06,
      "loss": 0.8453,
      "step": 260500
    },
    {
      "epoch": 4.573640688916589,
      "grad_norm": 0.052777001085918186,
      "learning_rate": 5.676157488518584e-06,
      "loss": 0.8318,
      "step": 260510
    },
    {
      "epoch": 4.573816253796591,
      "grad_norm": 0.0524946401886517,
      "learning_rate": 5.673149580804218e-06,
      "loss": 0.8374,
      "step": 260520
    },
    {
      "epoch": 4.573991818676592,
      "grad_norm": 0.045016386408987134,
      "learning_rate": 5.670142880899295e-06,
      "loss": 0.8414,
      "step": 260530
    },
    {
      "epoch": 4.574167383556594,
      "grad_norm": 0.05009107213653581,
      "learning_rate": 5.6671373888419414e-06,
      "loss": 0.8375,
      "step": 260540
    },
    {
      "epoch": 4.574342948436595,
      "grad_norm": 0.04803440327640358,
      "learning_rate": 5.664133104670245e-06,
      "loss": 0.836,
      "step": 260550
    },
    {
      "epoch": 4.574518513316596,
      "grad_norm": 0.06052962120610911,
      "learning_rate": 5.661130028422221e-06,
      "loss": 0.8418,
      "step": 260560
    },
    {
      "epoch": 4.574694078196598,
      "grad_norm": 0.046496818669047445,
      "learning_rate": 5.658128160135968e-06,
      "loss": 0.8366,
      "step": 260570
    },
    {
      "epoch": 4.574869643076599,
      "grad_norm": 0.04926835832103773,
      "learning_rate": 5.655127499849503e-06,
      "loss": 0.8389,
      "step": 260580
    },
    {
      "epoch": 4.575045207956601,
      "grad_norm": 0.047998383148685604,
      "learning_rate": 5.652128047600845e-06,
      "loss": 0.8405,
      "step": 260590
    },
    {
      "epoch": 4.575220772836602,
      "grad_norm": 0.062431703154047176,
      "learning_rate": 5.649129803428012e-06,
      "loss": 0.8441,
      "step": 260600
    },
    {
      "epoch": 4.575396337716603,
      "grad_norm": 0.041797479572235015,
      "learning_rate": 5.6461327673689565e-06,
      "loss": 0.8363,
      "step": 260610
    },
    {
      "epoch": 4.575571902596605,
      "grad_norm": 0.04492683427460048,
      "learning_rate": 5.643136939461687e-06,
      "loss": 0.8368,
      "step": 260620
    },
    {
      "epoch": 4.575747467476606,
      "grad_norm": 0.05544142129049827,
      "learning_rate": 5.640142319744145e-06,
      "loss": 0.8371,
      "step": 260630
    },
    {
      "epoch": 4.575923032356608,
      "grad_norm": 0.053793075674301694,
      "learning_rate": 5.637148908254272e-06,
      "loss": 0.8354,
      "step": 260640
    },
    {
      "epoch": 4.576098597236609,
      "grad_norm": 0.061829541878660174,
      "learning_rate": 5.634156705029999e-06,
      "loss": 0.8398,
      "step": 260650
    },
    {
      "epoch": 4.5762741621166105,
      "grad_norm": 0.05746092598315417,
      "learning_rate": 5.631165710109222e-06,
      "loss": 0.8347,
      "step": 260660
    },
    {
      "epoch": 4.576449726996612,
      "grad_norm": 0.05027338441307556,
      "learning_rate": 5.628175923529885e-06,
      "loss": 0.8326,
      "step": 260670
    },
    {
      "epoch": 4.5766252918766135,
      "grad_norm": 0.07218786187556964,
      "learning_rate": 5.625187345329817e-06,
      "loss": 0.8415,
      "step": 260680
    },
    {
      "epoch": 4.5768008567566145,
      "grad_norm": 0.07088710036086782,
      "learning_rate": 5.622199975546908e-06,
      "loss": 0.8446,
      "step": 260690
    },
    {
      "epoch": 4.5769764216366156,
      "grad_norm": 0.0451096606310002,
      "learning_rate": 5.6192138142189985e-06,
      "loss": 0.837,
      "step": 260700
    },
    {
      "epoch": 4.5771519865166175,
      "grad_norm": 0.04867928677439403,
      "learning_rate": 5.616228861383932e-06,
      "loss": 0.8349,
      "step": 260710
    },
    {
      "epoch": 4.5773275513966185,
      "grad_norm": 0.05602714622498141,
      "learning_rate": 5.61324511707954e-06,
      "loss": 0.8317,
      "step": 260720
    },
    {
      "epoch": 4.57750311627662,
      "grad_norm": 0.05197886395653192,
      "learning_rate": 5.610262581343588e-06,
      "loss": 0.8393,
      "step": 260730
    },
    {
      "epoch": 4.5776786811566215,
      "grad_norm": 0.044981952624113294,
      "learning_rate": 5.6072812542139085e-06,
      "loss": 0.8357,
      "step": 260740
    },
    {
      "epoch": 4.5778542460366225,
      "grad_norm": 0.044301617574425536,
      "learning_rate": 5.6043011357282555e-06,
      "loss": 0.8467,
      "step": 260750
    },
    {
      "epoch": 4.578029810916624,
      "grad_norm": 0.04803220795092703,
      "learning_rate": 5.6013222259243845e-06,
      "loss": 0.8335,
      "step": 260760
    },
    {
      "epoch": 4.5782053757966255,
      "grad_norm": 0.04939921471546582,
      "learning_rate": 5.5983445248400496e-06,
      "loss": 0.8405,
      "step": 260770
    },
    {
      "epoch": 4.578380940676627,
      "grad_norm": 0.05216337330491002,
      "learning_rate": 5.595368032512974e-06,
      "loss": 0.8414,
      "step": 260780
    },
    {
      "epoch": 4.578556505556628,
      "grad_norm": 0.04855345668985025,
      "learning_rate": 5.592392748980877e-06,
      "loss": 0.8372,
      "step": 260790
    },
    {
      "epoch": 4.578732070436629,
      "grad_norm": 0.046222495519975065,
      "learning_rate": 5.5894186742814275e-06,
      "loss": 0.8332,
      "step": 260800
    },
    {
      "epoch": 4.578907635316631,
      "grad_norm": 0.05336316054928661,
      "learning_rate": 5.5864458084523365e-06,
      "loss": 0.8403,
      "step": 260810
    },
    {
      "epoch": 4.579083200196632,
      "grad_norm": 0.04995009682370866,
      "learning_rate": 5.58347415153128e-06,
      "loss": 0.8401,
      "step": 260820
    },
    {
      "epoch": 4.579258765076634,
      "grad_norm": 0.051384326189187195,
      "learning_rate": 5.580503703555862e-06,
      "loss": 0.8396,
      "step": 260830
    },
    {
      "epoch": 4.579434329956635,
      "grad_norm": 0.05116121393796459,
      "learning_rate": 5.577534464563767e-06,
      "loss": 0.8394,
      "step": 260840
    },
    {
      "epoch": 4.579609894836636,
      "grad_norm": 0.05859434319731726,
      "learning_rate": 5.5745664345926e-06,
      "loss": 0.8416,
      "step": 260850
    },
    {
      "epoch": 4.579785459716638,
      "grad_norm": 0.043095294415940705,
      "learning_rate": 5.57159961367996e-06,
      "loss": 0.8417,
      "step": 260860
    },
    {
      "epoch": 4.579961024596639,
      "grad_norm": 0.05969538537577336,
      "learning_rate": 5.568634001863448e-06,
      "loss": 0.8371,
      "step": 260870
    },
    {
      "epoch": 4.580136589476641,
      "grad_norm": 0.06885761616278054,
      "learning_rate": 5.565669599180622e-06,
      "loss": 0.8415,
      "step": 260880
    },
    {
      "epoch": 4.580312154356642,
      "grad_norm": 0.046109495707382875,
      "learning_rate": 5.562706405669059e-06,
      "loss": 0.8405,
      "step": 260890
    },
    {
      "epoch": 4.580487719236644,
      "grad_norm": 0.04746156634874245,
      "learning_rate": 5.5597444213662846e-06,
      "loss": 0.8348,
      "step": 260900
    },
    {
      "epoch": 4.580663284116645,
      "grad_norm": 0.06640127780695726,
      "learning_rate": 5.556783646309855e-06,
      "loss": 0.8409,
      "step": 260910
    },
    {
      "epoch": 4.580838848996647,
      "grad_norm": 0.051953993820343214,
      "learning_rate": 5.553824080537261e-06,
      "loss": 0.8401,
      "step": 260920
    },
    {
      "epoch": 4.581014413876648,
      "grad_norm": 0.041957730093034225,
      "learning_rate": 5.550865724086006e-06,
      "loss": 0.8447,
      "step": 260930
    },
    {
      "epoch": 4.581189978756649,
      "grad_norm": 0.051330990796286904,
      "learning_rate": 5.5479085769935904e-06,
      "loss": 0.8351,
      "step": 260940
    },
    {
      "epoch": 4.581365543636651,
      "grad_norm": 0.05066504071073231,
      "learning_rate": 5.544952639297452e-06,
      "loss": 0.8401,
      "step": 260950
    },
    {
      "epoch": 4.581541108516652,
      "grad_norm": 0.04572979650038626,
      "learning_rate": 5.541997911035058e-06,
      "loss": 0.8425,
      "step": 260960
    },
    {
      "epoch": 4.581716673396654,
      "grad_norm": 0.04925079666008366,
      "learning_rate": 5.5390443922438455e-06,
      "loss": 0.8406,
      "step": 260970
    },
    {
      "epoch": 4.581892238276655,
      "grad_norm": 0.0546560375813379,
      "learning_rate": 5.536092082961239e-06,
      "loss": 0.8364,
      "step": 260980
    },
    {
      "epoch": 4.582067803156656,
      "grad_norm": 0.06390965600966374,
      "learning_rate": 5.533140983224641e-06,
      "loss": 0.833,
      "step": 260990
    },
    {
      "epoch": 4.582243368036658,
      "grad_norm": 0.04877339924278053,
      "learning_rate": 5.530191093071435e-06,
      "loss": 0.8412,
      "step": 261000
    },
    {
      "epoch": 4.582418932916659,
      "grad_norm": 0.04356293732649791,
      "learning_rate": 5.527242412539011e-06,
      "loss": 0.8369,
      "step": 261010
    },
    {
      "epoch": 4.582594497796661,
      "grad_norm": 0.11020853702765955,
      "learning_rate": 5.5242949416647285e-06,
      "loss": 0.8301,
      "step": 261020
    },
    {
      "epoch": 4.582770062676662,
      "grad_norm": 0.04891444650190832,
      "learning_rate": 5.5213486804859245e-06,
      "loss": 0.831,
      "step": 261030
    },
    {
      "epoch": 4.582945627556663,
      "grad_norm": 0.053811021422815235,
      "learning_rate": 5.518403629039926e-06,
      "loss": 0.8429,
      "step": 261040
    },
    {
      "epoch": 4.583121192436665,
      "grad_norm": 0.06212583897341815,
      "learning_rate": 5.515459787364047e-06,
      "loss": 0.8322,
      "step": 261050
    },
    {
      "epoch": 4.583296757316666,
      "grad_norm": 0.08135216745540368,
      "learning_rate": 5.512517155495615e-06,
      "loss": 0.833,
      "step": 261060
    },
    {
      "epoch": 4.583472322196668,
      "grad_norm": 0.0582254058197225,
      "learning_rate": 5.5095757334718555e-06,
      "loss": 0.8371,
      "step": 261070
    },
    {
      "epoch": 4.583647887076669,
      "grad_norm": 0.04525898633776517,
      "learning_rate": 5.506635521330096e-06,
      "loss": 0.8374,
      "step": 261080
    },
    {
      "epoch": 4.583823451956671,
      "grad_norm": 0.045161143593001885,
      "learning_rate": 5.503696519107564e-06,
      "loss": 0.8386,
      "step": 261090
    },
    {
      "epoch": 4.583999016836672,
      "grad_norm": 0.054478239876847384,
      "learning_rate": 5.500758726841484e-06,
      "loss": 0.8355,
      "step": 261100
    },
    {
      "epoch": 4.584174581716673,
      "grad_norm": 0.03987485641158511,
      "learning_rate": 5.4978221445691206e-06,
      "loss": 0.8404,
      "step": 261110
    },
    {
      "epoch": 4.584350146596675,
      "grad_norm": 0.06884366267151301,
      "learning_rate": 5.494886772327632e-06,
      "loss": 0.8368,
      "step": 261120
    },
    {
      "epoch": 4.584525711476676,
      "grad_norm": 0.08975631076551997,
      "learning_rate": 5.4919526101542455e-06,
      "loss": 0.8258,
      "step": 261130
    },
    {
      "epoch": 4.584701276356678,
      "grad_norm": 0.06578671920753462,
      "learning_rate": 5.489019658086112e-06,
      "loss": 0.8357,
      "step": 261140
    },
    {
      "epoch": 4.584876841236679,
      "grad_norm": 0.042872253333685155,
      "learning_rate": 5.486087916160405e-06,
      "loss": 0.8475,
      "step": 261150
    },
    {
      "epoch": 4.585052406116681,
      "grad_norm": 0.06270528070503455,
      "learning_rate": 5.483157384414283e-06,
      "loss": 0.8435,
      "step": 261160
    },
    {
      "epoch": 4.585227970996682,
      "grad_norm": 0.06330228248123161,
      "learning_rate": 5.480228062884833e-06,
      "loss": 0.8376,
      "step": 261170
    },
    {
      "epoch": 4.585403535876683,
      "grad_norm": 0.06205345793009414,
      "learning_rate": 5.4772999516092255e-06,
      "loss": 0.8417,
      "step": 261180
    },
    {
      "epoch": 4.585579100756685,
      "grad_norm": 0.0612597448306351,
      "learning_rate": 5.474373050624523e-06,
      "loss": 0.8383,
      "step": 261190
    },
    {
      "epoch": 4.585754665636686,
      "grad_norm": 0.04542710277092864,
      "learning_rate": 5.471447359967844e-06,
      "loss": 0.8426,
      "step": 261200
    },
    {
      "epoch": 4.585930230516688,
      "grad_norm": 0.09089939142432851,
      "learning_rate": 5.468522879676207e-06,
      "loss": 0.8297,
      "step": 261210
    },
    {
      "epoch": 4.586105795396689,
      "grad_norm": 0.04688301939907682,
      "learning_rate": 5.465599609786706e-06,
      "loss": 0.8377,
      "step": 261220
    },
    {
      "epoch": 4.58628136027669,
      "grad_norm": 0.06191464704347124,
      "learning_rate": 5.46267755033637e-06,
      "loss": 0.8427,
      "step": 261230
    },
    {
      "epoch": 4.586456925156692,
      "grad_norm": 0.05199584538448237,
      "learning_rate": 5.4597567013621965e-06,
      "loss": 0.8351,
      "step": 261240
    },
    {
      "epoch": 4.586632490036693,
      "grad_norm": 0.04372234658373586,
      "learning_rate": 5.456837062901237e-06,
      "loss": 0.84,
      "step": 261250
    },
    {
      "epoch": 4.586808054916695,
      "grad_norm": 0.04507521758211286,
      "learning_rate": 5.453918634990455e-06,
      "loss": 0.8387,
      "step": 261260
    },
    {
      "epoch": 4.586983619796696,
      "grad_norm": 0.05736018074111804,
      "learning_rate": 5.4510014176668355e-06,
      "loss": 0.8408,
      "step": 261270
    },
    {
      "epoch": 4.587159184676697,
      "grad_norm": 0.046175838597318705,
      "learning_rate": 5.448085410967353e-06,
      "loss": 0.843,
      "step": 261280
    },
    {
      "epoch": 4.587334749556699,
      "grad_norm": 0.04768990774115036,
      "learning_rate": 5.445170614928926e-06,
      "loss": 0.8346,
      "step": 261290
    },
    {
      "epoch": 4.5875103144367,
      "grad_norm": 0.05004968271774163,
      "learning_rate": 5.442257029588519e-06,
      "loss": 0.8421,
      "step": 261300
    },
    {
      "epoch": 4.587685879316702,
      "grad_norm": 0.055895407734581626,
      "learning_rate": 5.439344654983008e-06,
      "loss": 0.8455,
      "step": 261310
    },
    {
      "epoch": 4.587861444196703,
      "grad_norm": 0.04474108403478276,
      "learning_rate": 5.436433491149322e-06,
      "loss": 0.839,
      "step": 261320
    },
    {
      "epoch": 4.588037009076705,
      "grad_norm": 0.05295484747438008,
      "learning_rate": 5.43352353812435e-06,
      "loss": 0.8446,
      "step": 261330
    },
    {
      "epoch": 4.588212573956706,
      "grad_norm": 0.06042197190010765,
      "learning_rate": 5.430614795944919e-06,
      "loss": 0.8444,
      "step": 261340
    },
    {
      "epoch": 4.588388138836708,
      "grad_norm": 0.05476079585650681,
      "learning_rate": 5.427707264647943e-06,
      "loss": 0.838,
      "step": 261350
    },
    {
      "epoch": 4.588563703716709,
      "grad_norm": 0.052470474905806475,
      "learning_rate": 5.424800944270216e-06,
      "loss": 0.8389,
      "step": 261360
    },
    {
      "epoch": 4.58873926859671,
      "grad_norm": 0.05022561099222604,
      "learning_rate": 5.4218958348485945e-06,
      "loss": 0.8433,
      "step": 261370
    },
    {
      "epoch": 4.588914833476712,
      "grad_norm": 0.051891665132704656,
      "learning_rate": 5.4189919364198536e-06,
      "loss": 0.8436,
      "step": 261380
    },
    {
      "epoch": 4.589090398356713,
      "grad_norm": 0.06352308847681294,
      "learning_rate": 5.416089249020803e-06,
      "loss": 0.8432,
      "step": 261390
    },
    {
      "epoch": 4.5892659632367145,
      "grad_norm": 0.05905881554031981,
      "learning_rate": 5.413187772688242e-06,
      "loss": 0.84,
      "step": 261400
    },
    {
      "epoch": 4.589441528116716,
      "grad_norm": 0.04836234182212396,
      "learning_rate": 5.41028750745888e-06,
      "loss": 0.8372,
      "step": 261410
    },
    {
      "epoch": 4.589617092996717,
      "grad_norm": 0.07379331248016238,
      "learning_rate": 5.407388453369518e-06,
      "loss": 0.8341,
      "step": 261420
    },
    {
      "epoch": 4.5897926578767185,
      "grad_norm": 0.049465258169569884,
      "learning_rate": 5.4044906104568524e-06,
      "loss": 0.837,
      "step": 261430
    },
    {
      "epoch": 4.58996822275672,
      "grad_norm": 0.04990745171611921,
      "learning_rate": 5.401593978757618e-06,
      "loss": 0.8421,
      "step": 261440
    },
    {
      "epoch": 4.5901437876367215,
      "grad_norm": 0.052064588421338195,
      "learning_rate": 5.398698558308513e-06,
      "loss": 0.8352,
      "step": 261450
    },
    {
      "epoch": 4.5903193525167225,
      "grad_norm": 0.042675080524525295,
      "learning_rate": 5.395804349146205e-06,
      "loss": 0.836,
      "step": 261460
    },
    {
      "epoch": 4.5904949173967236,
      "grad_norm": 0.05880777043553095,
      "learning_rate": 5.392911351307394e-06,
      "loss": 0.8367,
      "step": 261470
    },
    {
      "epoch": 4.5906704822767255,
      "grad_norm": 0.058244423224309304,
      "learning_rate": 5.390019564828714e-06,
      "loss": 0.8388,
      "step": 261480
    },
    {
      "epoch": 4.5908460471567265,
      "grad_norm": 0.043041186786967535,
      "learning_rate": 5.3871289897468074e-06,
      "loss": 0.83,
      "step": 261490
    },
    {
      "epoch": 4.591021612036728,
      "grad_norm": 0.05435934218712979,
      "learning_rate": 5.38423962609831e-06,
      "loss": 0.8509,
      "step": 261500
    },
    {
      "epoch": 4.5911971769167295,
      "grad_norm": 0.06367198489612162,
      "learning_rate": 5.381351473919788e-06,
      "loss": 0.833,
      "step": 261510
    },
    {
      "epoch": 4.5913727417967305,
      "grad_norm": 0.4894884722619946,
      "learning_rate": 5.378464533247911e-06,
      "loss": 0.842,
      "step": 261520
    },
    {
      "epoch": 4.591548306676732,
      "grad_norm": 0.056647543700587506,
      "learning_rate": 5.375578804119188e-06,
      "loss": 0.8438,
      "step": 261530
    },
    {
      "epoch": 4.5917238715567334,
      "grad_norm": 0.059134882452687006,
      "learning_rate": 5.372694286570223e-06,
      "loss": 0.8368,
      "step": 261540
    },
    {
      "epoch": 4.591899436436735,
      "grad_norm": 0.05202157988126208,
      "learning_rate": 5.369810980637537e-06,
      "loss": 0.8349,
      "step": 261550
    },
    {
      "epoch": 4.592075001316736,
      "grad_norm": 0.05063056872337115,
      "learning_rate": 5.366928886357678e-06,
      "loss": 0.8389,
      "step": 261560
    },
    {
      "epoch": 4.592250566196738,
      "grad_norm": 0.06274773169934673,
      "learning_rate": 5.364048003767157e-06,
      "loss": 0.8359,
      "step": 261570
    },
    {
      "epoch": 4.592426131076739,
      "grad_norm": 0.04851988354739308,
      "learning_rate": 5.3611683329024665e-06,
      "loss": 0.8279,
      "step": 261580
    },
    {
      "epoch": 4.592601695956741,
      "grad_norm": 0.0629674767099638,
      "learning_rate": 5.358289873800118e-06,
      "loss": 0.8344,
      "step": 261590
    },
    {
      "epoch": 4.592777260836742,
      "grad_norm": 0.05504774831232978,
      "learning_rate": 5.355412626496559e-06,
      "loss": 0.8458,
      "step": 261600
    },
    {
      "epoch": 4.592952825716743,
      "grad_norm": 0.05086075851796276,
      "learning_rate": 5.352536591028247e-06,
      "loss": 0.8429,
      "step": 261610
    },
    {
      "epoch": 4.593128390596745,
      "grad_norm": 0.05318375489393171,
      "learning_rate": 5.349661767431651e-06,
      "loss": 0.8394,
      "step": 261620
    },
    {
      "epoch": 4.593303955476746,
      "grad_norm": 0.05109199823646804,
      "learning_rate": 5.346788155743152e-06,
      "loss": 0.8339,
      "step": 261630
    },
    {
      "epoch": 4.593479520356748,
      "grad_norm": 0.05060357423228389,
      "learning_rate": 5.343915755999186e-06,
      "loss": 0.839,
      "step": 261640
    },
    {
      "epoch": 4.593655085236749,
      "grad_norm": 0.04726535667380729,
      "learning_rate": 5.3410445682361325e-06,
      "loss": 0.8429,
      "step": 261650
    },
    {
      "epoch": 4.59383065011675,
      "grad_norm": 0.06144764670731672,
      "learning_rate": 5.338174592490375e-06,
      "loss": 0.8401,
      "step": 261660
    },
    {
      "epoch": 4.594006214996752,
      "grad_norm": 0.04120858648609949,
      "learning_rate": 5.335305828798292e-06,
      "loss": 0.8326,
      "step": 261670
    },
    {
      "epoch": 4.594181779876753,
      "grad_norm": 0.04722283323615609,
      "learning_rate": 5.3324382771962e-06,
      "loss": 0.8362,
      "step": 261680
    },
    {
      "epoch": 4.594357344756755,
      "grad_norm": 0.05064025788207252,
      "learning_rate": 5.3295719377204675e-06,
      "loss": 0.8479,
      "step": 261690
    },
    {
      "epoch": 4.594532909636756,
      "grad_norm": 0.04931411284425824,
      "learning_rate": 5.326706810407377e-06,
      "loss": 0.8482,
      "step": 261700
    },
    {
      "epoch": 4.594708474516757,
      "grad_norm": 0.05258786462744108,
      "learning_rate": 5.3238428952932665e-06,
      "loss": 0.8393,
      "step": 261710
    },
    {
      "epoch": 4.594884039396759,
      "grad_norm": 0.052785567101311354,
      "learning_rate": 5.320980192414384e-06,
      "loss": 0.8471,
      "step": 261720
    },
    {
      "epoch": 4.59505960427676,
      "grad_norm": 0.05487056716997818,
      "learning_rate": 5.3181187018070334e-06,
      "loss": 0.8397,
      "step": 261730
    },
    {
      "epoch": 4.595235169156762,
      "grad_norm": 0.05973123982192329,
      "learning_rate": 5.3152584235074534e-06,
      "loss": 0.8327,
      "step": 261740
    },
    {
      "epoch": 4.595410734036763,
      "grad_norm": 0.04647689758099906,
      "learning_rate": 5.31239935755187e-06,
      "loss": 0.8366,
      "step": 261750
    },
    {
      "epoch": 4.595586298916765,
      "grad_norm": 0.04055038081119253,
      "learning_rate": 5.309541503976545e-06,
      "loss": 0.8355,
      "step": 261760
    },
    {
      "epoch": 4.595761863796766,
      "grad_norm": 0.07272633034976074,
      "learning_rate": 5.306684862817659e-06,
      "loss": 0.8456,
      "step": 261770
    },
    {
      "epoch": 4.595937428676767,
      "grad_norm": 0.05343729401010844,
      "learning_rate": 5.303829434111419e-06,
      "loss": 0.8267,
      "step": 261780
    },
    {
      "epoch": 4.596112993556769,
      "grad_norm": 0.05439407485408468,
      "learning_rate": 5.300975217894008e-06,
      "loss": 0.8382,
      "step": 261790
    },
    {
      "epoch": 4.59628855843677,
      "grad_norm": 0.06074120099565308,
      "learning_rate": 5.298122214201574e-06,
      "loss": 0.8331,
      "step": 261800
    },
    {
      "epoch": 4.596464123316772,
      "grad_norm": 0.04942280604853766,
      "learning_rate": 5.29527042307028e-06,
      "loss": 0.8272,
      "step": 261810
    },
    {
      "epoch": 4.596639688196773,
      "grad_norm": 0.07027951276899684,
      "learning_rate": 5.292419844536232e-06,
      "loss": 0.8349,
      "step": 261820
    },
    {
      "epoch": 4.596815253076775,
      "grad_norm": 0.04930440915048122,
      "learning_rate": 5.289570478635579e-06,
      "loss": 0.8389,
      "step": 261830
    },
    {
      "epoch": 4.596990817956776,
      "grad_norm": 0.07960294656758801,
      "learning_rate": 5.286722325404417e-06,
      "loss": 0.8456,
      "step": 261840
    },
    {
      "epoch": 4.597166382836777,
      "grad_norm": 0.04235309073935361,
      "learning_rate": 5.2838753848788085e-06,
      "loss": 0.8384,
      "step": 261850
    },
    {
      "epoch": 4.597341947716779,
      "grad_norm": 0.06930937031068539,
      "learning_rate": 5.281029657094859e-06,
      "loss": 0.8425,
      "step": 261860
    },
    {
      "epoch": 4.59751751259678,
      "grad_norm": 0.05348690020540727,
      "learning_rate": 5.278185142088597e-06,
      "loss": 0.8365,
      "step": 261870
    },
    {
      "epoch": 4.597693077476782,
      "grad_norm": 0.048901616032612984,
      "learning_rate": 5.2753418398960964e-06,
      "loss": 0.8374,
      "step": 261880
    },
    {
      "epoch": 4.597868642356783,
      "grad_norm": 0.05132998055420026,
      "learning_rate": 5.272499750553343e-06,
      "loss": 0.8371,
      "step": 261890
    },
    {
      "epoch": 4.598044207236784,
      "grad_norm": 0.046157037657295626,
      "learning_rate": 5.269658874096355e-06,
      "loss": 0.835,
      "step": 261900
    },
    {
      "epoch": 4.598219772116786,
      "grad_norm": 0.048640766486540525,
      "learning_rate": 5.26681921056116e-06,
      "loss": 0.8303,
      "step": 261910
    },
    {
      "epoch": 4.598395336996787,
      "grad_norm": 0.04594977283853093,
      "learning_rate": 5.263980759983679e-06,
      "loss": 0.8396,
      "step": 261920
    },
    {
      "epoch": 4.598570901876789,
      "grad_norm": 0.0605704046153225,
      "learning_rate": 5.261143522399941e-06,
      "loss": 0.8404,
      "step": 261930
    },
    {
      "epoch": 4.59874646675679,
      "grad_norm": 0.05164178234943895,
      "learning_rate": 5.258307497845852e-06,
      "loss": 0.8407,
      "step": 261940
    },
    {
      "epoch": 4.598922031636791,
      "grad_norm": 0.042371515768627695,
      "learning_rate": 5.255472686357345e-06,
      "loss": 0.83,
      "step": 261950
    },
    {
      "epoch": 4.599097596516793,
      "grad_norm": 0.0562500618434656,
      "learning_rate": 5.252639087970371e-06,
      "loss": 0.8377,
      "step": 261960
    },
    {
      "epoch": 4.599273161396794,
      "grad_norm": 0.05226137731652457,
      "learning_rate": 5.249806702720795e-06,
      "loss": 0.8395,
      "step": 261970
    },
    {
      "epoch": 4.599448726276796,
      "grad_norm": 0.06784334474701646,
      "learning_rate": 5.246975530644524e-06,
      "loss": 0.8381,
      "step": 261980
    },
    {
      "epoch": 4.599624291156797,
      "grad_norm": 0.0515398444979551,
      "learning_rate": 5.2441455717774125e-06,
      "loss": 0.8371,
      "step": 261990
    },
    {
      "epoch": 4.599799856036799,
      "grad_norm": 0.05665140546975422,
      "learning_rate": 5.241316826155336e-06,
      "loss": 0.8438,
      "step": 262000
    },
    {
      "epoch": 4.5999754209168,
      "grad_norm": 0.044614239420091795,
      "learning_rate": 5.238489293814136e-06,
      "loss": 0.8374,
      "step": 262010
    },
    {
      "epoch": 4.600150985796802,
      "grad_norm": 0.05502169815144073,
      "learning_rate": 5.2356629747896095e-06,
      "loss": 0.8391,
      "step": 262020
    },
    {
      "epoch": 4.600326550676803,
      "grad_norm": 0.06209861017228469,
      "learning_rate": 5.232837869117602e-06,
      "loss": 0.8396,
      "step": 262030
    },
    {
      "epoch": 4.600502115556804,
      "grad_norm": 0.04312918294487879,
      "learning_rate": 5.2300139768338885e-06,
      "loss": 0.8406,
      "step": 262040
    },
    {
      "epoch": 4.600677680436806,
      "grad_norm": 0.0697830520980734,
      "learning_rate": 5.227191297974255e-06,
      "loss": 0.8326,
      "step": 262050
    },
    {
      "epoch": 4.600853245316807,
      "grad_norm": 0.047280217910018324,
      "learning_rate": 5.224369832574459e-06,
      "loss": 0.8397,
      "step": 262060
    },
    {
      "epoch": 4.601028810196809,
      "grad_norm": 0.04476938515267581,
      "learning_rate": 5.221549580670252e-06,
      "loss": 0.8377,
      "step": 262070
    },
    {
      "epoch": 4.60120437507681,
      "grad_norm": 0.05178397305072568,
      "learning_rate": 5.218730542297379e-06,
      "loss": 0.8362,
      "step": 262080
    },
    {
      "epoch": 4.601379939956811,
      "grad_norm": 0.05969399529703824,
      "learning_rate": 5.215912717491515e-06,
      "loss": 0.8351,
      "step": 262090
    },
    {
      "epoch": 4.601555504836813,
      "grad_norm": 0.05500071598341896,
      "learning_rate": 5.213096106288418e-06,
      "loss": 0.8342,
      "step": 262100
    },
    {
      "epoch": 4.601731069716814,
      "grad_norm": 0.057477159935514825,
      "learning_rate": 5.210280708723753e-06,
      "loss": 0.8363,
      "step": 262110
    },
    {
      "epoch": 4.601906634596816,
      "grad_norm": 0.04953224786202924,
      "learning_rate": 5.207466524833174e-06,
      "loss": 0.832,
      "step": 262120
    },
    {
      "epoch": 4.602082199476817,
      "grad_norm": 0.05847043068187543,
      "learning_rate": 5.204653554652371e-06,
      "loss": 0.8412,
      "step": 262130
    },
    {
      "epoch": 4.602257764356818,
      "grad_norm": 0.044493148635570176,
      "learning_rate": 5.201841798216956e-06,
      "loss": 0.8374,
      "step": 262140
    },
    {
      "epoch": 4.60243332923682,
      "grad_norm": 0.05181689950082289,
      "learning_rate": 5.1990312555625714e-06,
      "loss": 0.8389,
      "step": 262150
    },
    {
      "epoch": 4.602608894116821,
      "grad_norm": 0.0580517350789721,
      "learning_rate": 5.19622192672482e-06,
      "loss": 0.839,
      "step": 262160
    },
    {
      "epoch": 4.6027844589968225,
      "grad_norm": 0.04856664243933699,
      "learning_rate": 5.193413811739291e-06,
      "loss": 0.8399,
      "step": 262170
    },
    {
      "epoch": 4.602960023876824,
      "grad_norm": 0.05158850295190827,
      "learning_rate": 5.190606910641583e-06,
      "loss": 0.8309,
      "step": 262180
    },
    {
      "epoch": 4.603135588756825,
      "grad_norm": 0.0388385725233393,
      "learning_rate": 5.187801223467221e-06,
      "loss": 0.841,
      "step": 262190
    },
    {
      "epoch": 4.6033111536368265,
      "grad_norm": 0.055583810929912165,
      "learning_rate": 5.184996750251806e-06,
      "loss": 0.8348,
      "step": 262200
    },
    {
      "epoch": 4.603486718516828,
      "grad_norm": 0.0504311099503771,
      "learning_rate": 5.182193491030827e-06,
      "loss": 0.8383,
      "step": 262210
    },
    {
      "epoch": 4.6036622833968295,
      "grad_norm": 0.05084659149034944,
      "learning_rate": 5.179391445839843e-06,
      "loss": 0.8425,
      "step": 262220
    },
    {
      "epoch": 4.6038378482768305,
      "grad_norm": 0.04494099700107037,
      "learning_rate": 5.1765906147143086e-06,
      "loss": 0.842,
      "step": 262230
    },
    {
      "epoch": 4.604013413156832,
      "grad_norm": 0.04519878421919736,
      "learning_rate": 5.17379099768975e-06,
      "loss": 0.8389,
      "step": 262240
    },
    {
      "epoch": 4.6041889780368335,
      "grad_norm": 0.06774205097646795,
      "learning_rate": 5.170992594801623e-06,
      "loss": 0.8482,
      "step": 262250
    },
    {
      "epoch": 4.604364542916835,
      "grad_norm": 0.06497335966751934,
      "learning_rate": 5.168195406085364e-06,
      "loss": 0.8363,
      "step": 262260
    },
    {
      "epoch": 4.604540107796836,
      "grad_norm": 0.05619169930963423,
      "learning_rate": 5.165399431576465e-06,
      "loss": 0.8422,
      "step": 262270
    },
    {
      "epoch": 4.6047156726768375,
      "grad_norm": 0.04990488690517049,
      "learning_rate": 5.162604671310304e-06,
      "loss": 0.8458,
      "step": 262280
    },
    {
      "epoch": 4.604891237556839,
      "grad_norm": 0.05024322971854726,
      "learning_rate": 5.159811125322319e-06,
      "loss": 0.844,
      "step": 262290
    },
    {
      "epoch": 4.60506680243684,
      "grad_norm": 0.042978243655637766,
      "learning_rate": 5.1570187936478995e-06,
      "loss": 0.8385,
      "step": 262300
    },
    {
      "epoch": 4.605242367316842,
      "grad_norm": 0.07319919437733072,
      "learning_rate": 5.154227676322406e-06,
      "loss": 0.8371,
      "step": 262310
    },
    {
      "epoch": 4.605417932196843,
      "grad_norm": 0.059097121966824914,
      "learning_rate": 5.15143777338123e-06,
      "loss": 0.839,
      "step": 262320
    },
    {
      "epoch": 4.605593497076844,
      "grad_norm": 0.05311824750491535,
      "learning_rate": 5.148649084859696e-06,
      "loss": 0.8392,
      "step": 262330
    },
    {
      "epoch": 4.605769061956846,
      "grad_norm": 0.04885734624442374,
      "learning_rate": 5.145861610793154e-06,
      "loss": 0.8446,
      "step": 262340
    },
    {
      "epoch": 4.605944626836847,
      "grad_norm": 0.053978751727948165,
      "learning_rate": 5.143075351216916e-06,
      "loss": 0.8401,
      "step": 262350
    },
    {
      "epoch": 4.606120191716849,
      "grad_norm": 0.04899067526599813,
      "learning_rate": 5.140290306166266e-06,
      "loss": 0.8341,
      "step": 262360
    },
    {
      "epoch": 4.60629575659685,
      "grad_norm": 0.044814505905747794,
      "learning_rate": 5.137506475676539e-06,
      "loss": 0.8349,
      "step": 262370
    },
    {
      "epoch": 4.606471321476851,
      "grad_norm": 0.0567145014894231,
      "learning_rate": 5.134723859782952e-06,
      "loss": 0.8373,
      "step": 262380
    },
    {
      "epoch": 4.606646886356853,
      "grad_norm": 0.06982290778014882,
      "learning_rate": 5.131942458520808e-06,
      "loss": 0.8357,
      "step": 262390
    },
    {
      "epoch": 4.606822451236854,
      "grad_norm": 0.04626039203348567,
      "learning_rate": 5.129162271925312e-06,
      "loss": 0.838,
      "step": 262400
    },
    {
      "epoch": 4.606998016116856,
      "grad_norm": 0.04781626454207249,
      "learning_rate": 5.126383300031703e-06,
      "loss": 0.8374,
      "step": 262410
    },
    {
      "epoch": 4.607173580996857,
      "grad_norm": 0.0910695915867181,
      "learning_rate": 5.123605542875197e-06,
      "loss": 0.8343,
      "step": 262420
    },
    {
      "epoch": 4.607349145876859,
      "grad_norm": 0.05202094837408082,
      "learning_rate": 5.120829000490974e-06,
      "loss": 0.8414,
      "step": 262430
    },
    {
      "epoch": 4.60752471075686,
      "grad_norm": 0.05495286294745212,
      "learning_rate": 5.1180536729142304e-06,
      "loss": 0.8404,
      "step": 262440
    },
    {
      "epoch": 4.607700275636861,
      "grad_norm": 0.05312328069778979,
      "learning_rate": 5.115279560180125e-06,
      "loss": 0.8383,
      "step": 262450
    },
    {
      "epoch": 4.607875840516863,
      "grad_norm": 0.051559916420477137,
      "learning_rate": 5.112506662323811e-06,
      "loss": 0.847,
      "step": 262460
    },
    {
      "epoch": 4.608051405396864,
      "grad_norm": 0.052100321838481733,
      "learning_rate": 5.109734979380391e-06,
      "loss": 0.8386,
      "step": 262470
    },
    {
      "epoch": 4.608226970276866,
      "grad_norm": 0.07307196918693339,
      "learning_rate": 5.106964511385016e-06,
      "loss": 0.8411,
      "step": 262480
    },
    {
      "epoch": 4.608402535156867,
      "grad_norm": 0.04932808666309443,
      "learning_rate": 5.1041952583727925e-06,
      "loss": 0.8405,
      "step": 262490
    },
    {
      "epoch": 4.608578100036869,
      "grad_norm": 0.043303937756778946,
      "learning_rate": 5.101427220378782e-06,
      "loss": 0.8345,
      "step": 262500
    },
    {
      "epoch": 4.60875366491687,
      "grad_norm": 0.05384699597086268,
      "learning_rate": 5.098660397438068e-06,
      "loss": 0.8428,
      "step": 262510
    },
    {
      "epoch": 4.608929229796871,
      "grad_norm": 0.07842587886635578,
      "learning_rate": 5.095894789585714e-06,
      "loss": 0.8469,
      "step": 262520
    },
    {
      "epoch": 4.609104794676873,
      "grad_norm": 0.04178168786068841,
      "learning_rate": 5.093130396856726e-06,
      "loss": 0.8415,
      "step": 262530
    },
    {
      "epoch": 4.609280359556874,
      "grad_norm": 0.05524887552383218,
      "learning_rate": 5.090367219286186e-06,
      "loss": 0.8356,
      "step": 262540
    },
    {
      "epoch": 4.609455924436876,
      "grad_norm": 0.059085358925174705,
      "learning_rate": 5.08760525690906e-06,
      "loss": 0.8363,
      "step": 262550
    },
    {
      "epoch": 4.609631489316877,
      "grad_norm": 0.05098154862053491,
      "learning_rate": 5.084844509760366e-06,
      "loss": 0.8407,
      "step": 262560
    },
    {
      "epoch": 4.609807054196878,
      "grad_norm": 0.046775108907413,
      "learning_rate": 5.082084977875065e-06,
      "loss": 0.8369,
      "step": 262570
    },
    {
      "epoch": 4.60998261907688,
      "grad_norm": 0.051679098540568486,
      "learning_rate": 5.079326661288132e-06,
      "loss": 0.8405,
      "step": 262580
    },
    {
      "epoch": 4.610158183956881,
      "grad_norm": 0.051346225495821574,
      "learning_rate": 5.076569560034529e-06,
      "loss": 0.8356,
      "step": 262590
    },
    {
      "epoch": 4.610333748836883,
      "grad_norm": 0.05518616137037667,
      "learning_rate": 5.073813674149145e-06,
      "loss": 0.838,
      "step": 262600
    },
    {
      "epoch": 4.610509313716884,
      "grad_norm": 0.05915210079156219,
      "learning_rate": 5.071059003666939e-06,
      "loss": 0.8396,
      "step": 262610
    },
    {
      "epoch": 4.610684878596885,
      "grad_norm": 0.0420833242457466,
      "learning_rate": 5.0683055486228e-06,
      "loss": 0.8415,
      "step": 262620
    },
    {
      "epoch": 4.610860443476887,
      "grad_norm": 0.04372711856160739,
      "learning_rate": 5.065553309051625e-06,
      "loss": 0.843,
      "step": 262630
    },
    {
      "epoch": 4.611036008356888,
      "grad_norm": 0.05601461042901163,
      "learning_rate": 5.062802284988266e-06,
      "loss": 0.8416,
      "step": 262640
    },
    {
      "epoch": 4.61121157323689,
      "grad_norm": 0.05920866317005157,
      "learning_rate": 5.060052476467588e-06,
      "loss": 0.8424,
      "step": 262650
    },
    {
      "epoch": 4.611387138116891,
      "grad_norm": 0.05775269603359298,
      "learning_rate": 5.057303883524442e-06,
      "loss": 0.848,
      "step": 262660
    },
    {
      "epoch": 4.611562702996893,
      "grad_norm": 0.04803946052987673,
      "learning_rate": 5.0545565061936295e-06,
      "loss": 0.84,
      "step": 262670
    },
    {
      "epoch": 4.611738267876894,
      "grad_norm": 0.07518814730763504,
      "learning_rate": 5.051810344509979e-06,
      "loss": 0.8385,
      "step": 262680
    },
    {
      "epoch": 4.611913832756895,
      "grad_norm": 0.04255337104053123,
      "learning_rate": 5.049065398508278e-06,
      "loss": 0.8382,
      "step": 262690
    },
    {
      "epoch": 4.612089397636897,
      "grad_norm": 0.05312353697990771,
      "learning_rate": 5.046321668223293e-06,
      "loss": 0.8344,
      "step": 262700
    },
    {
      "epoch": 4.612264962516898,
      "grad_norm": 0.057074562687448795,
      "learning_rate": 5.0435791536898325e-06,
      "loss": 0.847,
      "step": 262710
    },
    {
      "epoch": 4.6124405273969,
      "grad_norm": 0.053746920406146315,
      "learning_rate": 5.0408378549425964e-06,
      "loss": 0.8358,
      "step": 262720
    },
    {
      "epoch": 4.612616092276901,
      "grad_norm": 0.055592490494945246,
      "learning_rate": 5.038097772016348e-06,
      "loss": 0.8438,
      "step": 262730
    },
    {
      "epoch": 4.612791657156903,
      "grad_norm": 0.05097585929062847,
      "learning_rate": 5.035358904945778e-06,
      "loss": 0.8288,
      "step": 262740
    },
    {
      "epoch": 4.612967222036904,
      "grad_norm": 0.04808025830276169,
      "learning_rate": 5.032621253765595e-06,
      "loss": 0.843,
      "step": 262750
    },
    {
      "epoch": 4.613142786916905,
      "grad_norm": 0.05258813890666286,
      "learning_rate": 5.02988481851052e-06,
      "loss": 0.835,
      "step": 262760
    },
    {
      "epoch": 4.613318351796907,
      "grad_norm": 0.0461579203531802,
      "learning_rate": 5.0271495992151656e-06,
      "loss": 0.8372,
      "step": 262770
    },
    {
      "epoch": 4.613493916676908,
      "grad_norm": 0.061252624265144724,
      "learning_rate": 5.02441559591424e-06,
      "loss": 0.8457,
      "step": 262780
    },
    {
      "epoch": 4.61366948155691,
      "grad_norm": 0.04343356785488192,
      "learning_rate": 5.021682808642357e-06,
      "loss": 0.8429,
      "step": 262790
    },
    {
      "epoch": 4.613845046436911,
      "grad_norm": 0.05250631926327239,
      "learning_rate": 5.018951237434148e-06,
      "loss": 0.8383,
      "step": 262800
    },
    {
      "epoch": 4.614020611316912,
      "grad_norm": 0.044954623872610074,
      "learning_rate": 5.016220882324216e-06,
      "loss": 0.8349,
      "step": 262810
    },
    {
      "epoch": 4.614196176196914,
      "grad_norm": 0.05493103409816574,
      "learning_rate": 5.013491743347168e-06,
      "loss": 0.8312,
      "step": 262820
    },
    {
      "epoch": 4.614371741076915,
      "grad_norm": 0.08134513376886711,
      "learning_rate": 5.0107638205375745e-06,
      "loss": 0.8346,
      "step": 262830
    },
    {
      "epoch": 4.614547305956917,
      "grad_norm": 0.04996722540237236,
      "learning_rate": 5.008037113929992e-06,
      "loss": 0.8315,
      "step": 262840
    },
    {
      "epoch": 4.614722870836918,
      "grad_norm": 0.0462794787400445,
      "learning_rate": 5.005311623558975e-06,
      "loss": 0.8439,
      "step": 262850
    },
    {
      "epoch": 4.614898435716919,
      "grad_norm": 0.051777807886414076,
      "learning_rate": 5.002587349459082e-06,
      "loss": 0.8332,
      "step": 262860
    },
    {
      "epoch": 4.615074000596921,
      "grad_norm": 0.05614153719100398,
      "learning_rate": 4.999864291664768e-06,
      "loss": 0.831,
      "step": 262870
    },
    {
      "epoch": 4.615249565476922,
      "grad_norm": 0.056770439184236084,
      "learning_rate": 4.997142450210602e-06,
      "loss": 0.8395,
      "step": 262880
    },
    {
      "epoch": 4.615425130356924,
      "grad_norm": 0.04367263970691313,
      "learning_rate": 4.994421825131019e-06,
      "loss": 0.8359,
      "step": 262890
    },
    {
      "epoch": 4.615600695236925,
      "grad_norm": 0.05575071262856129,
      "learning_rate": 4.991702416460531e-06,
      "loss": 0.8418,
      "step": 262900
    },
    {
      "epoch": 4.6157762601169265,
      "grad_norm": 0.07096368751848811,
      "learning_rate": 4.988984224233554e-06,
      "loss": 0.8361,
      "step": 262910
    },
    {
      "epoch": 4.615951824996928,
      "grad_norm": 0.06272715786380999,
      "learning_rate": 4.986267248484552e-06,
      "loss": 0.838,
      "step": 262920
    },
    {
      "epoch": 4.6161273898769295,
      "grad_norm": 0.04853821974672734,
      "learning_rate": 4.9835514892479635e-06,
      "loss": 0.8331,
      "step": 262930
    },
    {
      "epoch": 4.6163029547569305,
      "grad_norm": 0.057771270333987594,
      "learning_rate": 4.980836946558146e-06,
      "loss": 0.8359,
      "step": 262940
    },
    {
      "epoch": 4.616478519636932,
      "grad_norm": 0.05910407432109971,
      "learning_rate": 4.978123620449547e-06,
      "loss": 0.8316,
      "step": 262950
    },
    {
      "epoch": 4.6166540845169335,
      "grad_norm": 0.04522703855043616,
      "learning_rate": 4.975411510956523e-06,
      "loss": 0.8358,
      "step": 262960
    },
    {
      "epoch": 4.6168296493969345,
      "grad_norm": 0.051396536764259156,
      "learning_rate": 4.972700618113447e-06,
      "loss": 0.8482,
      "step": 262970
    },
    {
      "epoch": 4.617005214276936,
      "grad_norm": 0.09120769729766955,
      "learning_rate": 4.969990941954641e-06,
      "loss": 0.8369,
      "step": 262980
    },
    {
      "epoch": 4.6171807791569375,
      "grad_norm": 0.06346136899703875,
      "learning_rate": 4.9672824825144536e-06,
      "loss": 0.8343,
      "step": 262990
    },
    {
      "epoch": 4.6173563440369385,
      "grad_norm": 0.046718119683393844,
      "learning_rate": 4.964575239827222e-06,
      "loss": 0.8359,
      "step": 263000
    },
    {
      "epoch": 4.61753190891694,
      "grad_norm": 0.05667750561201669,
      "learning_rate": 4.961869213927205e-06,
      "loss": 0.8435,
      "step": 263010
    },
    {
      "epoch": 4.6177074737969415,
      "grad_norm": 0.05167457695056829,
      "learning_rate": 4.959164404848718e-06,
      "loss": 0.8398,
      "step": 263020
    },
    {
      "epoch": 4.617883038676943,
      "grad_norm": 0.05122828874051314,
      "learning_rate": 4.9564608126260315e-06,
      "loss": 0.8385,
      "step": 263030
    },
    {
      "epoch": 4.618058603556944,
      "grad_norm": 0.05694113128788712,
      "learning_rate": 4.953758437293382e-06,
      "loss": 0.8369,
      "step": 263040
    },
    {
      "epoch": 4.6182341684369455,
      "grad_norm": 0.07268963361741519,
      "learning_rate": 4.951057278885029e-06,
      "loss": 0.8343,
      "step": 263050
    },
    {
      "epoch": 4.618409733316947,
      "grad_norm": 0.05257455845356825,
      "learning_rate": 4.948357337435181e-06,
      "loss": 0.8342,
      "step": 263060
    },
    {
      "epoch": 4.618585298196948,
      "grad_norm": 0.05482131991244792,
      "learning_rate": 4.945658612978059e-06,
      "loss": 0.8376,
      "step": 263070
    },
    {
      "epoch": 4.61876086307695,
      "grad_norm": 0.05934558141284393,
      "learning_rate": 4.942961105547848e-06,
      "loss": 0.8301,
      "step": 263080
    },
    {
      "epoch": 4.618936427956951,
      "grad_norm": 0.0619729367028487,
      "learning_rate": 4.940264815178732e-06,
      "loss": 0.8452,
      "step": 263090
    },
    {
      "epoch": 4.619111992836953,
      "grad_norm": 0.04987904305540367,
      "learning_rate": 4.937569741904871e-06,
      "loss": 0.8445,
      "step": 263100
    },
    {
      "epoch": 4.619287557716954,
      "grad_norm": 0.06266614463261098,
      "learning_rate": 4.934875885760402e-06,
      "loss": 0.8299,
      "step": 263110
    },
    {
      "epoch": 4.619463122596955,
      "grad_norm": 0.05294933357515187,
      "learning_rate": 4.932183246779477e-06,
      "loss": 0.8421,
      "step": 263120
    },
    {
      "epoch": 4.619638687476957,
      "grad_norm": 0.045749846321355425,
      "learning_rate": 4.9294918249962005e-06,
      "loss": 0.835,
      "step": 263130
    },
    {
      "epoch": 4.619814252356958,
      "grad_norm": 0.04147609003910252,
      "learning_rate": 4.926801620444678e-06,
      "loss": 0.8292,
      "step": 263140
    },
    {
      "epoch": 4.61998981723696,
      "grad_norm": 0.05442572435130432,
      "learning_rate": 4.9241126331589826e-06,
      "loss": 0.8385,
      "step": 263150
    },
    {
      "epoch": 4.620165382116961,
      "grad_norm": 0.05270573245391319,
      "learning_rate": 4.921424863173187e-06,
      "loss": 0.8385,
      "step": 263160
    },
    {
      "epoch": 4.620340946996963,
      "grad_norm": 0.055721524913208635,
      "learning_rate": 4.918738310521376e-06,
      "loss": 0.837,
      "step": 263170
    },
    {
      "epoch": 4.620516511876964,
      "grad_norm": 0.04628580346369833,
      "learning_rate": 4.916052975237544e-06,
      "loss": 0.8387,
      "step": 263180
    },
    {
      "epoch": 4.620692076756965,
      "grad_norm": 0.048811936017096685,
      "learning_rate": 4.9133688573557405e-06,
      "loss": 0.8372,
      "step": 263190
    },
    {
      "epoch": 4.620867641636967,
      "grad_norm": 0.062153748121667074,
      "learning_rate": 4.910685956909976e-06,
      "loss": 0.8389,
      "step": 263200
    },
    {
      "epoch": 4.621043206516968,
      "grad_norm": 0.05042802192285471,
      "learning_rate": 4.90800427393421e-06,
      "loss": 0.8409,
      "step": 263210
    },
    {
      "epoch": 4.62121877139697,
      "grad_norm": 0.06057469196257636,
      "learning_rate": 4.905323808462474e-06,
      "loss": 0.8301,
      "step": 263220
    },
    {
      "epoch": 4.621394336276971,
      "grad_norm": 0.06739354189975966,
      "learning_rate": 4.902644560528684e-06,
      "loss": 0.8443,
      "step": 263230
    },
    {
      "epoch": 4.621569901156972,
      "grad_norm": 0.0591609275848092,
      "learning_rate": 4.899966530166827e-06,
      "loss": 0.832,
      "step": 263240
    },
    {
      "epoch": 4.621745466036974,
      "grad_norm": 0.05256360618515577,
      "learning_rate": 4.8972897174107885e-06,
      "loss": 0.8424,
      "step": 263250
    },
    {
      "epoch": 4.621921030916975,
      "grad_norm": 0.05703287172557032,
      "learning_rate": 4.894614122294509e-06,
      "loss": 0.8377,
      "step": 263260
    },
    {
      "epoch": 4.622096595796977,
      "grad_norm": 0.05289163300333757,
      "learning_rate": 4.8919397448519075e-06,
      "loss": 0.8425,
      "step": 263270
    },
    {
      "epoch": 4.622272160676978,
      "grad_norm": 0.05701270022284537,
      "learning_rate": 4.8892665851168255e-06,
      "loss": 0.8425,
      "step": 263280
    },
    {
      "epoch": 4.622447725556979,
      "grad_norm": 0.050830293602182976,
      "learning_rate": 4.8865946431231705e-06,
      "loss": 0.8323,
      "step": 263290
    },
    {
      "epoch": 4.622623290436981,
      "grad_norm": 0.058492664003798836,
      "learning_rate": 4.883923918904763e-06,
      "loss": 0.8387,
      "step": 263300
    },
    {
      "epoch": 4.622798855316982,
      "grad_norm": 0.04345870283496826,
      "learning_rate": 4.881254412495491e-06,
      "loss": 0.8329,
      "step": 263310
    },
    {
      "epoch": 4.622974420196984,
      "grad_norm": 0.0446948152263672,
      "learning_rate": 4.8785861239291154e-06,
      "loss": 0.8294,
      "step": 263320
    },
    {
      "epoch": 4.623149985076985,
      "grad_norm": 0.055409064261018555,
      "learning_rate": 4.875919053239493e-06,
      "loss": 0.8407,
      "step": 263330
    },
    {
      "epoch": 4.623325549956987,
      "grad_norm": 0.052357296106880453,
      "learning_rate": 4.873253200460396e-06,
      "loss": 0.8402,
      "step": 263340
    },
    {
      "epoch": 4.623501114836988,
      "grad_norm": 0.038652277936487685,
      "learning_rate": 4.8705885656256045e-06,
      "loss": 0.8412,
      "step": 263350
    },
    {
      "epoch": 4.623676679716989,
      "grad_norm": 0.04766647111306548,
      "learning_rate": 4.867925148768869e-06,
      "loss": 0.8391,
      "step": 263360
    },
    {
      "epoch": 4.623852244596991,
      "grad_norm": 0.047121225934717646,
      "learning_rate": 4.865262949923957e-06,
      "loss": 0.8339,
      "step": 263370
    },
    {
      "epoch": 4.624027809476992,
      "grad_norm": 0.0491039341734633,
      "learning_rate": 4.862601969124565e-06,
      "loss": 0.8324,
      "step": 263380
    },
    {
      "epoch": 4.624203374356994,
      "grad_norm": 0.040658664287921524,
      "learning_rate": 4.859942206404461e-06,
      "loss": 0.8428,
      "step": 263390
    },
    {
      "epoch": 4.624378939236995,
      "grad_norm": 0.05381771804776356,
      "learning_rate": 4.857283661797298e-06,
      "loss": 0.8424,
      "step": 263400
    },
    {
      "epoch": 4.624554504116997,
      "grad_norm": 0.05856686683448303,
      "learning_rate": 4.854626335336786e-06,
      "loss": 0.8375,
      "step": 263410
    },
    {
      "epoch": 4.624730068996998,
      "grad_norm": 0.06411387200453864,
      "learning_rate": 4.851970227056581e-06,
      "loss": 0.8355,
      "step": 263420
    },
    {
      "epoch": 4.624905633876999,
      "grad_norm": 0.04882954916766021,
      "learning_rate": 4.849315336990327e-06,
      "loss": 0.8472,
      "step": 263430
    },
    {
      "epoch": 4.625081198757001,
      "grad_norm": 0.044941394462790524,
      "learning_rate": 4.8466616651717e-06,
      "loss": 0.8379,
      "step": 263440
    },
    {
      "epoch": 4.625256763637002,
      "grad_norm": 0.07158229051450449,
      "learning_rate": 4.8440092116342695e-06,
      "loss": 0.8367,
      "step": 263450
    },
    {
      "epoch": 4.625432328517004,
      "grad_norm": 0.05687229797293317,
      "learning_rate": 4.841357976411687e-06,
      "loss": 0.8375,
      "step": 263460
    },
    {
      "epoch": 4.625607893397005,
      "grad_norm": 0.042359086321344816,
      "learning_rate": 4.8387079595375226e-06,
      "loss": 0.8356,
      "step": 263470
    },
    {
      "epoch": 4.625783458277006,
      "grad_norm": 0.05813180155967756,
      "learning_rate": 4.836059161045364e-06,
      "loss": 0.8398,
      "step": 263480
    },
    {
      "epoch": 4.625959023157008,
      "grad_norm": 0.05782432008685561,
      "learning_rate": 4.833411580968758e-06,
      "loss": 0.8373,
      "step": 263490
    },
    {
      "epoch": 4.626134588037009,
      "grad_norm": 0.052200365765224785,
      "learning_rate": 4.830765219341258e-06,
      "loss": 0.8312,
      "step": 263500
    },
    {
      "epoch": 4.626310152917011,
      "grad_norm": 0.05733073218346035,
      "learning_rate": 4.8281200761964015e-06,
      "loss": 0.839,
      "step": 263510
    },
    {
      "epoch": 4.626485717797012,
      "grad_norm": 0.04978269654501517,
      "learning_rate": 4.825476151567687e-06,
      "loss": 0.8388,
      "step": 263520
    },
    {
      "epoch": 4.626661282677013,
      "grad_norm": 0.05313757520639269,
      "learning_rate": 4.822833445488628e-06,
      "loss": 0.8431,
      "step": 263530
    },
    {
      "epoch": 4.626836847557015,
      "grad_norm": 0.05137794073631298,
      "learning_rate": 4.8201919579927035e-06,
      "loss": 0.8395,
      "step": 263540
    },
    {
      "epoch": 4.627012412437016,
      "grad_norm": 0.06477121149850469,
      "learning_rate": 4.817551689113359e-06,
      "loss": 0.8395,
      "step": 263550
    },
    {
      "epoch": 4.627187977317018,
      "grad_norm": 0.05407319035452907,
      "learning_rate": 4.814912638884086e-06,
      "loss": 0.8311,
      "step": 263560
    },
    {
      "epoch": 4.627363542197019,
      "grad_norm": 0.053140472471945935,
      "learning_rate": 4.812274807338297e-06,
      "loss": 0.8336,
      "step": 263570
    },
    {
      "epoch": 4.627539107077021,
      "grad_norm": 0.05042258001666748,
      "learning_rate": 4.809638194509438e-06,
      "loss": 0.8338,
      "step": 263580
    },
    {
      "epoch": 4.627714671957022,
      "grad_norm": 0.06368165461628876,
      "learning_rate": 4.80700280043088e-06,
      "loss": 0.8418,
      "step": 263590
    },
    {
      "epoch": 4.627890236837024,
      "grad_norm": 0.0860924895162878,
      "learning_rate": 4.804368625136022e-06,
      "loss": 0.8377,
      "step": 263600
    },
    {
      "epoch": 4.628065801717025,
      "grad_norm": 0.06020127438513643,
      "learning_rate": 4.8017356686582805e-06,
      "loss": 0.8431,
      "step": 263610
    },
    {
      "epoch": 4.628241366597026,
      "grad_norm": 0.055716901130259934,
      "learning_rate": 4.799103931030946e-06,
      "loss": 0.8349,
      "step": 263620
    },
    {
      "epoch": 4.628416931477028,
      "grad_norm": 0.06036734573620436,
      "learning_rate": 4.796473412287432e-06,
      "loss": 0.8321,
      "step": 263630
    },
    {
      "epoch": 4.628592496357029,
      "grad_norm": 0.05067152888298102,
      "learning_rate": 4.793844112461021e-06,
      "loss": 0.8382,
      "step": 263640
    },
    {
      "epoch": 4.6287680612370306,
      "grad_norm": 0.0579744521204318,
      "learning_rate": 4.791216031585059e-06,
      "loss": 0.8417,
      "step": 263650
    },
    {
      "epoch": 4.628943626117032,
      "grad_norm": 0.05116158733335425,
      "learning_rate": 4.788589169692818e-06,
      "loss": 0.8339,
      "step": 263660
    },
    {
      "epoch": 4.629119190997033,
      "grad_norm": 0.04185320358728536,
      "learning_rate": 4.785963526817588e-06,
      "loss": 0.8429,
      "step": 263670
    },
    {
      "epoch": 4.6292947558770345,
      "grad_norm": 0.05687132624824908,
      "learning_rate": 4.783339102992643e-06,
      "loss": 0.836,
      "step": 263680
    },
    {
      "epoch": 4.629470320757036,
      "grad_norm": 0.046485638068536374,
      "learning_rate": 4.780715898251238e-06,
      "loss": 0.835,
      "step": 263690
    },
    {
      "epoch": 4.6296458856370375,
      "grad_norm": 0.04669829816552097,
      "learning_rate": 4.778093912626591e-06,
      "loss": 0.8381,
      "step": 263700
    },
    {
      "epoch": 4.6298214505170385,
      "grad_norm": 0.043830703864413,
      "learning_rate": 4.775473146151939e-06,
      "loss": 0.8412,
      "step": 263710
    },
    {
      "epoch": 4.62999701539704,
      "grad_norm": 0.059805785652753206,
      "learning_rate": 4.7728535988604755e-06,
      "loss": 0.8366,
      "step": 263720
    },
    {
      "epoch": 4.6301725802770415,
      "grad_norm": 0.04611958108518331,
      "learning_rate": 4.7702352707854154e-06,
      "loss": 0.8325,
      "step": 263730
    },
    {
      "epoch": 4.6303481451570425,
      "grad_norm": 0.046361568295242586,
      "learning_rate": 4.767618161959909e-06,
      "loss": 0.8395,
      "step": 263740
    },
    {
      "epoch": 4.630523710037044,
      "grad_norm": 0.049824043041016175,
      "learning_rate": 4.765002272417128e-06,
      "loss": 0.8316,
      "step": 263750
    },
    {
      "epoch": 4.6306992749170455,
      "grad_norm": 0.04576663540817222,
      "learning_rate": 4.76238760219021e-06,
      "loss": 0.8424,
      "step": 263760
    },
    {
      "epoch": 4.6308748397970465,
      "grad_norm": 0.05367256957331146,
      "learning_rate": 4.759774151312271e-06,
      "loss": 0.8372,
      "step": 263770
    },
    {
      "epoch": 4.631050404677048,
      "grad_norm": 0.046570342556588346,
      "learning_rate": 4.757161919816462e-06,
      "loss": 0.8432,
      "step": 263780
    },
    {
      "epoch": 4.6312259695570495,
      "grad_norm": 0.047326827383342096,
      "learning_rate": 4.754550907735832e-06,
      "loss": 0.8363,
      "step": 263790
    },
    {
      "epoch": 4.631401534437051,
      "grad_norm": 0.06889011066153468,
      "learning_rate": 4.751941115103499e-06,
      "loss": 0.836,
      "step": 263800
    },
    {
      "epoch": 4.631577099317052,
      "grad_norm": 0.05235640554535179,
      "learning_rate": 4.749332541952513e-06,
      "loss": 0.8459,
      "step": 263810
    },
    {
      "epoch": 4.631752664197054,
      "grad_norm": 0.05378356383435131,
      "learning_rate": 4.746725188315947e-06,
      "loss": 0.8287,
      "step": 263820
    },
    {
      "epoch": 4.631928229077055,
      "grad_norm": 0.04226826069514739,
      "learning_rate": 4.744119054226806e-06,
      "loss": 0.8414,
      "step": 263830
    },
    {
      "epoch": 4.632103793957057,
      "grad_norm": 0.054813398870379025,
      "learning_rate": 4.7415141397181204e-06,
      "loss": 0.8332,
      "step": 263840
    },
    {
      "epoch": 4.632279358837058,
      "grad_norm": 0.05163006131865313,
      "learning_rate": 4.738910444822906e-06,
      "loss": 0.8442,
      "step": 263850
    },
    {
      "epoch": 4.632454923717059,
      "grad_norm": 0.04514845080522203,
      "learning_rate": 4.736307969574138e-06,
      "loss": 0.835,
      "step": 263860
    },
    {
      "epoch": 4.632630488597061,
      "grad_norm": 0.06057420187381858,
      "learning_rate": 4.7337067140048e-06,
      "loss": 0.8374,
      "step": 263870
    },
    {
      "epoch": 4.632806053477062,
      "grad_norm": 0.05646773680468028,
      "learning_rate": 4.7311066781478566e-06,
      "loss": 0.8397,
      "step": 263880
    },
    {
      "epoch": 4.632981618357064,
      "grad_norm": 0.04375916414288567,
      "learning_rate": 4.728507862036223e-06,
      "loss": 0.8345,
      "step": 263890
    },
    {
      "epoch": 4.633157183237065,
      "grad_norm": 0.06931433681885321,
      "learning_rate": 4.725910265702866e-06,
      "loss": 0.8357,
      "step": 263900
    },
    {
      "epoch": 4.633332748117066,
      "grad_norm": 0.06106629883021386,
      "learning_rate": 4.723313889180655e-06,
      "loss": 0.8404,
      "step": 263910
    },
    {
      "epoch": 4.633508312997068,
      "grad_norm": 0.05407512216471056,
      "learning_rate": 4.720718732502525e-06,
      "loss": 0.8352,
      "step": 263920
    },
    {
      "epoch": 4.633683877877069,
      "grad_norm": 0.04802963544510195,
      "learning_rate": 4.7181247957013365e-06,
      "loss": 0.8424,
      "step": 263930
    },
    {
      "epoch": 4.633859442757071,
      "grad_norm": 0.043298928569042286,
      "learning_rate": 4.7155320788099555e-06,
      "loss": 0.839,
      "step": 263940
    },
    {
      "epoch": 4.634035007637072,
      "grad_norm": 0.05369850193889572,
      "learning_rate": 4.712940581861243e-06,
      "loss": 0.8464,
      "step": 263950
    },
    {
      "epoch": 4.634210572517073,
      "grad_norm": 0.07332100773749704,
      "learning_rate": 4.71035030488801e-06,
      "loss": 0.8403,
      "step": 263960
    },
    {
      "epoch": 4.634386137397075,
      "grad_norm": 0.0539247649336384,
      "learning_rate": 4.707761247923108e-06,
      "loss": 0.8411,
      "step": 263970
    },
    {
      "epoch": 4.634561702277076,
      "grad_norm": 0.05588946147346,
      "learning_rate": 4.705173410999324e-06,
      "loss": 0.8415,
      "step": 263980
    },
    {
      "epoch": 4.634737267157078,
      "grad_norm": 0.047453320384465474,
      "learning_rate": 4.702586794149445e-06,
      "loss": 0.8365,
      "step": 263990
    },
    {
      "epoch": 4.634912832037079,
      "grad_norm": 0.054461226212843175,
      "learning_rate": 4.700001397406246e-06,
      "loss": 0.8311,
      "step": 264000
    },
    {
      "epoch": 4.635088396917081,
      "grad_norm": 0.05527298465300057,
      "learning_rate": 4.697417220802483e-06,
      "loss": 0.8399,
      "step": 264010
    },
    {
      "epoch": 4.635263961797082,
      "grad_norm": 0.043396981858347956,
      "learning_rate": 4.694834264370908e-06,
      "loss": 0.8374,
      "step": 264020
    },
    {
      "epoch": 4.635439526677083,
      "grad_norm": 0.053911617901286644,
      "learning_rate": 4.692252528144231e-06,
      "loss": 0.842,
      "step": 264030
    },
    {
      "epoch": 4.635615091557085,
      "grad_norm": 0.06413575983799963,
      "learning_rate": 4.689672012155172e-06,
      "loss": 0.832,
      "step": 264040
    },
    {
      "epoch": 4.635790656437086,
      "grad_norm": 0.05061685479676133,
      "learning_rate": 4.687092716436442e-06,
      "loss": 0.8315,
      "step": 264050
    },
    {
      "epoch": 4.635966221317088,
      "grad_norm": 0.05443206285599027,
      "learning_rate": 4.684514641020675e-06,
      "loss": 0.8404,
      "step": 264060
    },
    {
      "epoch": 4.636141786197089,
      "grad_norm": 0.04888814628604209,
      "learning_rate": 4.681937785940601e-06,
      "loss": 0.8357,
      "step": 264070
    },
    {
      "epoch": 4.636317351077091,
      "grad_norm": 0.04854432052662412,
      "learning_rate": 4.679362151228808e-06,
      "loss": 0.844,
      "step": 264080
    },
    {
      "epoch": 4.636492915957092,
      "grad_norm": 0.05699046634576645,
      "learning_rate": 4.676787736917976e-06,
      "loss": 0.8401,
      "step": 264090
    },
    {
      "epoch": 4.636668480837093,
      "grad_norm": 0.0445616594382379,
      "learning_rate": 4.674214543040691e-06,
      "loss": 0.8347,
      "step": 264100
    },
    {
      "epoch": 4.636844045717095,
      "grad_norm": 0.056954416934572163,
      "learning_rate": 4.671642569629566e-06,
      "loss": 0.8408,
      "step": 264110
    },
    {
      "epoch": 4.637019610597096,
      "grad_norm": 0.052457087878382234,
      "learning_rate": 4.66907181671721e-06,
      "loss": 0.8368,
      "step": 264120
    },
    {
      "epoch": 4.637195175477098,
      "grad_norm": 0.05382374542750904,
      "learning_rate": 4.666502284336148e-06,
      "loss": 0.8437,
      "step": 264130
    },
    {
      "epoch": 4.637370740357099,
      "grad_norm": 0.059237666047893975,
      "learning_rate": 4.663933972518979e-06,
      "loss": 0.8375,
      "step": 264140
    },
    {
      "epoch": 4.6375463052371,
      "grad_norm": 0.06504153986373828,
      "learning_rate": 4.661366881298225e-06,
      "loss": 0.8362,
      "step": 264150
    },
    {
      "epoch": 4.637721870117102,
      "grad_norm": 0.06004505161552498,
      "learning_rate": 4.658801010706433e-06,
      "loss": 0.8416,
      "step": 264160
    },
    {
      "epoch": 4.637897434997103,
      "grad_norm": 0.06547172258951575,
      "learning_rate": 4.65623636077608e-06,
      "loss": 0.8342,
      "step": 264170
    },
    {
      "epoch": 4.638072999877105,
      "grad_norm": 0.04947597606839504,
      "learning_rate": 4.653672931539669e-06,
      "loss": 0.8434,
      "step": 264180
    },
    {
      "epoch": 4.638248564757106,
      "grad_norm": 0.05051210533303105,
      "learning_rate": 4.65111072302971e-06,
      "loss": 0.8353,
      "step": 264190
    },
    {
      "epoch": 4.638424129637107,
      "grad_norm": 0.0529432312542431,
      "learning_rate": 4.648549735278628e-06,
      "loss": 0.8388,
      "step": 264200
    },
    {
      "epoch": 4.638599694517109,
      "grad_norm": 0.05661174988420741,
      "learning_rate": 4.645989968318881e-06,
      "loss": 0.8497,
      "step": 264210
    },
    {
      "epoch": 4.63877525939711,
      "grad_norm": 0.05163364901701673,
      "learning_rate": 4.6434314221829245e-06,
      "loss": 0.8357,
      "step": 264220
    },
    {
      "epoch": 4.638950824277112,
      "grad_norm": 0.06851313790590353,
      "learning_rate": 4.640874096903138e-06,
      "loss": 0.8367,
      "step": 264230
    },
    {
      "epoch": 4.639126389157113,
      "grad_norm": 0.06453176525859473,
      "learning_rate": 4.638317992511959e-06,
      "loss": 0.8372,
      "step": 264240
    },
    {
      "epoch": 4.639301954037115,
      "grad_norm": 0.05374193585135334,
      "learning_rate": 4.635763109041755e-06,
      "loss": 0.8278,
      "step": 264250
    },
    {
      "epoch": 4.639477518917116,
      "grad_norm": 0.06136486703120454,
      "learning_rate": 4.633209446524907e-06,
      "loss": 0.8377,
      "step": 264260
    },
    {
      "epoch": 4.639653083797118,
      "grad_norm": 0.05491593206210102,
      "learning_rate": 4.630657004993749e-06,
      "loss": 0.8326,
      "step": 264270
    },
    {
      "epoch": 4.639828648677119,
      "grad_norm": 0.04946396503472428,
      "learning_rate": 4.6281057844806425e-06,
      "loss": 0.8447,
      "step": 264280
    },
    {
      "epoch": 4.64000421355712,
      "grad_norm": 0.03698858190281376,
      "learning_rate": 4.62555578501791e-06,
      "loss": 0.846,
      "step": 264290
    },
    {
      "epoch": 4.640179778437122,
      "grad_norm": 0.04761279721875644,
      "learning_rate": 4.6230070066378444e-06,
      "loss": 0.8386,
      "step": 264300
    },
    {
      "epoch": 4.640355343317123,
      "grad_norm": 0.05439919460960522,
      "learning_rate": 4.62045944937276e-06,
      "loss": 0.8316,
      "step": 264310
    },
    {
      "epoch": 4.640530908197125,
      "grad_norm": 0.06010205509793542,
      "learning_rate": 4.617913113254927e-06,
      "loss": 0.8342,
      "step": 264320
    },
    {
      "epoch": 4.640706473077126,
      "grad_norm": 0.04912736924635315,
      "learning_rate": 4.615367998316604e-06,
      "loss": 0.8366,
      "step": 264330
    },
    {
      "epoch": 4.640882037957127,
      "grad_norm": 0.04855515342329472,
      "learning_rate": 4.61282410459004e-06,
      "loss": 0.84,
      "step": 264340
    },
    {
      "epoch": 4.641057602837129,
      "grad_norm": 0.05140429306159192,
      "learning_rate": 4.610281432107462e-06,
      "loss": 0.8377,
      "step": 264350
    },
    {
      "epoch": 4.64123316771713,
      "grad_norm": 0.044904946826181825,
      "learning_rate": 4.607739980901105e-06,
      "loss": 0.8419,
      "step": 264360
    },
    {
      "epoch": 4.641408732597132,
      "grad_norm": 0.05248743625410116,
      "learning_rate": 4.605199751003142e-06,
      "loss": 0.8394,
      "step": 264370
    },
    {
      "epoch": 4.641584297477133,
      "grad_norm": 0.06750938117804353,
      "learning_rate": 4.602660742445776e-06,
      "loss": 0.8302,
      "step": 264380
    },
    {
      "epoch": 4.641759862357134,
      "grad_norm": 0.055817008573486586,
      "learning_rate": 4.60012295526118e-06,
      "loss": 0.8419,
      "step": 264390
    },
    {
      "epoch": 4.641935427237136,
      "grad_norm": 0.05833335479797113,
      "learning_rate": 4.597586389481478e-06,
      "loss": 0.8394,
      "step": 264400
    },
    {
      "epoch": 4.642110992117137,
      "grad_norm": 0.0639116823553392,
      "learning_rate": 4.595051045138857e-06,
      "loss": 0.8377,
      "step": 264410
    },
    {
      "epoch": 4.6422865569971385,
      "grad_norm": 0.05495472563710806,
      "learning_rate": 4.5925169222654074e-06,
      "loss": 0.8427,
      "step": 264420
    },
    {
      "epoch": 4.64246212187714,
      "grad_norm": 0.049673957355314574,
      "learning_rate": 4.589984020893247e-06,
      "loss": 0.8396,
      "step": 264430
    },
    {
      "epoch": 4.642637686757141,
      "grad_norm": 0.0528974969585916,
      "learning_rate": 4.5874523410544705e-06,
      "loss": 0.8362,
      "step": 264440
    },
    {
      "epoch": 4.6428132516371425,
      "grad_norm": 0.04728634281305369,
      "learning_rate": 4.584921882781139e-06,
      "loss": 0.8351,
      "step": 264450
    },
    {
      "epoch": 4.642988816517144,
      "grad_norm": 0.059593540551551784,
      "learning_rate": 4.5823926461053355e-06,
      "loss": 0.8416,
      "step": 264460
    },
    {
      "epoch": 4.6431643813971455,
      "grad_norm": 0.06066644470220249,
      "learning_rate": 4.579864631059077e-06,
      "loss": 0.8442,
      "step": 264470
    },
    {
      "epoch": 4.6433399462771465,
      "grad_norm": 0.04789284813042157,
      "learning_rate": 4.5773378376744256e-06,
      "loss": 0.8356,
      "step": 264480
    },
    {
      "epoch": 4.6435155111571484,
      "grad_norm": 0.05366846749261967,
      "learning_rate": 4.574812265983387e-06,
      "loss": 0.8422,
      "step": 264490
    },
    {
      "epoch": 4.6436910760371495,
      "grad_norm": 0.0564483639828134,
      "learning_rate": 4.572287916017958e-06,
      "loss": 0.8394,
      "step": 264500
    },
    {
      "epoch": 4.643866640917151,
      "grad_norm": 0.06266667345718442,
      "learning_rate": 4.569764787810111e-06,
      "loss": 0.8405,
      "step": 264510
    },
    {
      "epoch": 4.644042205797152,
      "grad_norm": 0.049883230122205226,
      "learning_rate": 4.56724288139183e-06,
      "loss": 0.8328,
      "step": 264520
    },
    {
      "epoch": 4.6442177706771535,
      "grad_norm": 0.04784044354695512,
      "learning_rate": 4.564722196795067e-06,
      "loss": 0.8322,
      "step": 264530
    },
    {
      "epoch": 4.644393335557155,
      "grad_norm": 0.053882594353663125,
      "learning_rate": 4.56220273405174e-06,
      "loss": 0.8251,
      "step": 264540
    },
    {
      "epoch": 4.644568900437156,
      "grad_norm": 0.058783904975322356,
      "learning_rate": 4.559684493193803e-06,
      "loss": 0.8381,
      "step": 264550
    },
    {
      "epoch": 4.644744465317158,
      "grad_norm": 0.04566626570343365,
      "learning_rate": 4.557167474253148e-06,
      "loss": 0.8389,
      "step": 264560
    },
    {
      "epoch": 4.644920030197159,
      "grad_norm": 0.059072916742805036,
      "learning_rate": 4.554651677261642e-06,
      "loss": 0.8363,
      "step": 264570
    },
    {
      "epoch": 4.64509559507716,
      "grad_norm": 0.042835601116283976,
      "learning_rate": 4.552137102251202e-06,
      "loss": 0.846,
      "step": 264580
    },
    {
      "epoch": 4.645271159957162,
      "grad_norm": 0.049494970491922335,
      "learning_rate": 4.549623749253659e-06,
      "loss": 0.8361,
      "step": 264590
    },
    {
      "epoch": 4.645446724837163,
      "grad_norm": 0.05260355469612841,
      "learning_rate": 4.547111618300887e-06,
      "loss": 0.8354,
      "step": 264600
    },
    {
      "epoch": 4.645622289717165,
      "grad_norm": 0.05540646271552572,
      "learning_rate": 4.544600709424673e-06,
      "loss": 0.8388,
      "step": 264610
    },
    {
      "epoch": 4.645797854597166,
      "grad_norm": 0.04919611734093772,
      "learning_rate": 4.542091022656871e-06,
      "loss": 0.8343,
      "step": 264620
    },
    {
      "epoch": 4.645973419477167,
      "grad_norm": 0.0635725149924104,
      "learning_rate": 4.5395825580292525e-06,
      "loss": 0.8436,
      "step": 264630
    },
    {
      "epoch": 4.646148984357169,
      "grad_norm": 0.05461728408908432,
      "learning_rate": 4.537075315573598e-06,
      "loss": 0.8447,
      "step": 264640
    },
    {
      "epoch": 4.64632454923717,
      "grad_norm": 0.054644460819380465,
      "learning_rate": 4.5345692953217114e-06,
      "loss": 0.8401,
      "step": 264650
    },
    {
      "epoch": 4.646500114117172,
      "grad_norm": 0.04248622998908917,
      "learning_rate": 4.532064497305307e-06,
      "loss": 0.8415,
      "step": 264660
    },
    {
      "epoch": 4.646675678997173,
      "grad_norm": 0.05538328940121989,
      "learning_rate": 4.5295609215561345e-06,
      "loss": 0.8339,
      "step": 264670
    },
    {
      "epoch": 4.646851243877175,
      "grad_norm": 0.057060999244015906,
      "learning_rate": 4.527058568105915e-06,
      "loss": 0.8382,
      "step": 264680
    },
    {
      "epoch": 4.647026808757176,
      "grad_norm": 0.04465735672643092,
      "learning_rate": 4.524557436986339e-06,
      "loss": 0.8366,
      "step": 264690
    },
    {
      "epoch": 4.647202373637177,
      "grad_norm": 0.04946075945717429,
      "learning_rate": 4.522057528229136e-06,
      "loss": 0.8383,
      "step": 264700
    },
    {
      "epoch": 4.647377938517179,
      "grad_norm": 0.05385920999558381,
      "learning_rate": 4.5195588418659285e-06,
      "loss": 0.841,
      "step": 264710
    },
    {
      "epoch": 4.64755350339718,
      "grad_norm": 0.07435038655326356,
      "learning_rate": 4.517061377928414e-06,
      "loss": 0.8323,
      "step": 264720
    },
    {
      "epoch": 4.647729068277182,
      "grad_norm": 0.050856604315457674,
      "learning_rate": 4.514565136448226e-06,
      "loss": 0.8373,
      "step": 264730
    },
    {
      "epoch": 4.647904633157183,
      "grad_norm": 0.04655821582361148,
      "learning_rate": 4.5120701174569636e-06,
      "loss": 0.8424,
      "step": 264740
    },
    {
      "epoch": 4.648080198037185,
      "grad_norm": 0.048585196260082425,
      "learning_rate": 4.509576320986295e-06,
      "loss": 0.8368,
      "step": 264750
    },
    {
      "epoch": 4.648255762917186,
      "grad_norm": 0.06004243154013982,
      "learning_rate": 4.507083747067771e-06,
      "loss": 0.848,
      "step": 264760
    },
    {
      "epoch": 4.648431327797187,
      "grad_norm": 0.051834450867647715,
      "learning_rate": 4.504592395733007e-06,
      "loss": 0.837,
      "step": 264770
    },
    {
      "epoch": 4.648606892677189,
      "grad_norm": 0.050074053879178335,
      "learning_rate": 4.502102267013534e-06,
      "loss": 0.8463,
      "step": 264780
    },
    {
      "epoch": 4.64878245755719,
      "grad_norm": 0.04440352469306151,
      "learning_rate": 4.499613360940922e-06,
      "loss": 0.8318,
      "step": 264790
    },
    {
      "epoch": 4.648958022437192,
      "grad_norm": 0.045345144798561145,
      "learning_rate": 4.4971256775467e-06,
      "loss": 0.8357,
      "step": 264800
    },
    {
      "epoch": 4.649133587317193,
      "grad_norm": 0.06895106016344742,
      "learning_rate": 4.494639216862383e-06,
      "loss": 0.8388,
      "step": 264810
    },
    {
      "epoch": 4.649309152197194,
      "grad_norm": 0.07512498293195086,
      "learning_rate": 4.492153978919504e-06,
      "loss": 0.8402,
      "step": 264820
    },
    {
      "epoch": 4.649484717077196,
      "grad_norm": 0.057100210185968565,
      "learning_rate": 4.48966996374951e-06,
      "loss": 0.8456,
      "step": 264830
    },
    {
      "epoch": 4.649660281957197,
      "grad_norm": 0.05350265157751752,
      "learning_rate": 4.487187171383911e-06,
      "loss": 0.8426,
      "step": 264840
    },
    {
      "epoch": 4.649835846837199,
      "grad_norm": 0.05349509418493416,
      "learning_rate": 4.484705601854131e-06,
      "loss": 0.8361,
      "step": 264850
    },
    {
      "epoch": 4.6500114117172,
      "grad_norm": 0.07651858417423978,
      "learning_rate": 4.482225255191639e-06,
      "loss": 0.848,
      "step": 264860
    },
    {
      "epoch": 4.650186976597201,
      "grad_norm": 0.05679377477585151,
      "learning_rate": 4.479746131427848e-06,
      "loss": 0.8346,
      "step": 264870
    },
    {
      "epoch": 4.650362541477203,
      "grad_norm": 0.05149358211906644,
      "learning_rate": 4.4772682305941695e-06,
      "loss": 0.8382,
      "step": 264880
    },
    {
      "epoch": 4.650538106357204,
      "grad_norm": 0.05693837541040947,
      "learning_rate": 4.474791552722006e-06,
      "loss": 0.8415,
      "step": 264890
    },
    {
      "epoch": 4.650713671237206,
      "grad_norm": 0.047587807615302256,
      "learning_rate": 4.472316097842726e-06,
      "loss": 0.8339,
      "step": 264900
    },
    {
      "epoch": 4.650889236117207,
      "grad_norm": 0.044833395714368104,
      "learning_rate": 4.469841865987699e-06,
      "loss": 0.8389,
      "step": 264910
    },
    {
      "epoch": 4.651064800997209,
      "grad_norm": 0.04571273076395456,
      "learning_rate": 4.467368857188284e-06,
      "loss": 0.8369,
      "step": 264920
    },
    {
      "epoch": 4.65124036587721,
      "grad_norm": 0.0466372407697736,
      "learning_rate": 4.464897071475793e-06,
      "loss": 0.8383,
      "step": 264930
    },
    {
      "epoch": 4.651415930757212,
      "grad_norm": 0.05677768397981546,
      "learning_rate": 4.462426508881573e-06,
      "loss": 0.8392,
      "step": 264940
    },
    {
      "epoch": 4.651591495637213,
      "grad_norm": 0.05421100304244635,
      "learning_rate": 4.459957169436907e-06,
      "loss": 0.8381,
      "step": 264950
    },
    {
      "epoch": 4.651767060517214,
      "grad_norm": 0.05841488381149386,
      "learning_rate": 4.4574890531730735e-06,
      "loss": 0.8475,
      "step": 264960
    },
    {
      "epoch": 4.651942625397216,
      "grad_norm": 0.05214521062383346,
      "learning_rate": 4.455022160121377e-06,
      "loss": 0.8457,
      "step": 264970
    },
    {
      "epoch": 4.652118190277217,
      "grad_norm": 0.05029615289520486,
      "learning_rate": 4.452556490313021e-06,
      "loss": 0.8403,
      "step": 264980
    },
    {
      "epoch": 4.652293755157219,
      "grad_norm": 0.04781227447441461,
      "learning_rate": 4.4500920437793084e-06,
      "loss": 0.8416,
      "step": 264990
    },
    {
      "epoch": 4.65246932003722,
      "grad_norm": 0.054825189004599914,
      "learning_rate": 4.447628820551411e-06,
      "loss": 0.8417,
      "step": 265000
    },
    {
      "epoch": 4.652644884917221,
      "grad_norm": 0.06403652573769732,
      "learning_rate": 4.445166820660576e-06,
      "loss": 0.8304,
      "step": 265010
    },
    {
      "epoch": 4.652820449797223,
      "grad_norm": 0.04984875271368665,
      "learning_rate": 4.442706044137974e-06,
      "loss": 0.8416,
      "step": 265020
    },
    {
      "epoch": 4.652996014677224,
      "grad_norm": 0.06695252630449153,
      "learning_rate": 4.440246491014789e-06,
      "loss": 0.8352,
      "step": 265030
    },
    {
      "epoch": 4.653171579557226,
      "grad_norm": 0.04842746678489963,
      "learning_rate": 4.437788161322201e-06,
      "loss": 0.8326,
      "step": 265040
    },
    {
      "epoch": 4.653347144437227,
      "grad_norm": 0.06196139396821081,
      "learning_rate": 4.4353310550913265e-06,
      "loss": 0.8316,
      "step": 265050
    },
    {
      "epoch": 4.653522709317228,
      "grad_norm": 0.04622376174012648,
      "learning_rate": 4.432875172353328e-06,
      "loss": 0.8482,
      "step": 265060
    },
    {
      "epoch": 4.65369827419723,
      "grad_norm": 0.041362679145409954,
      "learning_rate": 4.430420513139307e-06,
      "loss": 0.84,
      "step": 265070
    },
    {
      "epoch": 4.653873839077231,
      "grad_norm": 0.04227888058076195,
      "learning_rate": 4.427967077480348e-06,
      "loss": 0.8417,
      "step": 265080
    },
    {
      "epoch": 4.654049403957233,
      "grad_norm": 0.0533132633056673,
      "learning_rate": 4.425514865407569e-06,
      "loss": 0.8447,
      "step": 265090
    },
    {
      "epoch": 4.654224968837234,
      "grad_norm": 0.048646408254322544,
      "learning_rate": 4.4230638769520175e-06,
      "loss": 0.8381,
      "step": 265100
    },
    {
      "epoch": 4.654400533717235,
      "grad_norm": 0.046631669255269376,
      "learning_rate": 4.4206141121447664e-06,
      "loss": 0.8413,
      "step": 265110
    },
    {
      "epoch": 4.654576098597237,
      "grad_norm": 0.04617452873503002,
      "learning_rate": 4.418165571016833e-06,
      "loss": 0.8377,
      "step": 265120
    },
    {
      "epoch": 4.654751663477238,
      "grad_norm": 0.04915997425296721,
      "learning_rate": 4.415718253599257e-06,
      "loss": 0.8384,
      "step": 265130
    },
    {
      "epoch": 4.65492722835724,
      "grad_norm": 0.08156091240670191,
      "learning_rate": 4.413272159923045e-06,
      "loss": 0.8369,
      "step": 265140
    },
    {
      "epoch": 4.655102793237241,
      "grad_norm": 0.049336936273594006,
      "learning_rate": 4.410827290019169e-06,
      "loss": 0.8359,
      "step": 265150
    },
    {
      "epoch": 4.6552783581172426,
      "grad_norm": 0.04871990914190845,
      "learning_rate": 4.408383643918646e-06,
      "loss": 0.8503,
      "step": 265160
    },
    {
      "epoch": 4.655453922997244,
      "grad_norm": 0.0632245658464954,
      "learning_rate": 4.405941221652396e-06,
      "loss": 0.8413,
      "step": 265170
    },
    {
      "epoch": 4.6556294878772455,
      "grad_norm": 0.05314838219795762,
      "learning_rate": 4.40350002325139e-06,
      "loss": 0.8464,
      "step": 265180
    },
    {
      "epoch": 4.6558050527572465,
      "grad_norm": 0.05022588351351373,
      "learning_rate": 4.401060048746549e-06,
      "loss": 0.84,
      "step": 265190
    },
    {
      "epoch": 4.655980617637248,
      "grad_norm": 0.1277967790041647,
      "learning_rate": 4.398621298168788e-06,
      "loss": 0.8326,
      "step": 265200
    },
    {
      "epoch": 4.6561561825172495,
      "grad_norm": 0.044322686984723435,
      "learning_rate": 4.396183771549016e-06,
      "loss": 0.8385,
      "step": 265210
    },
    {
      "epoch": 4.6563317473972505,
      "grad_norm": 0.055315082382266456,
      "learning_rate": 4.393747468918108e-06,
      "loss": 0.8433,
      "step": 265220
    },
    {
      "epoch": 4.6565073122772525,
      "grad_norm": 0.04749595310728598,
      "learning_rate": 4.3913123903069275e-06,
      "loss": 0.8407,
      "step": 265230
    },
    {
      "epoch": 4.6566828771572535,
      "grad_norm": 0.05168849083423946,
      "learning_rate": 4.388878535746346e-06,
      "loss": 0.8404,
      "step": 265240
    },
    {
      "epoch": 4.6568584420372545,
      "grad_norm": 0.04885998566893061,
      "learning_rate": 4.3864459052671745e-06,
      "loss": 0.8378,
      "step": 265250
    },
    {
      "epoch": 4.657034006917256,
      "grad_norm": 0.04426724812734429,
      "learning_rate": 4.384014498900254e-06,
      "loss": 0.8432,
      "step": 265260
    },
    {
      "epoch": 4.6572095717972575,
      "grad_norm": 0.057599586432929215,
      "learning_rate": 4.381584316676391e-06,
      "loss": 0.8326,
      "step": 265270
    },
    {
      "epoch": 4.657385136677259,
      "grad_norm": 0.056141594220397474,
      "learning_rate": 4.379155358626375e-06,
      "loss": 0.8386,
      "step": 265280
    },
    {
      "epoch": 4.65756070155726,
      "grad_norm": 0.06787767810702919,
      "learning_rate": 4.376727624780969e-06,
      "loss": 0.8393,
      "step": 265290
    },
    {
      "epoch": 4.6577362664372615,
      "grad_norm": 0.049002360483161624,
      "learning_rate": 4.374301115170935e-06,
      "loss": 0.8359,
      "step": 265300
    },
    {
      "epoch": 4.657911831317263,
      "grad_norm": 0.04480068238181881,
      "learning_rate": 4.37187582982704e-06,
      "loss": 0.8408,
      "step": 265310
    },
    {
      "epoch": 4.658087396197264,
      "grad_norm": 0.05206722726255003,
      "learning_rate": 4.369451768779983e-06,
      "loss": 0.8353,
      "step": 265320
    },
    {
      "epoch": 4.658262961077266,
      "grad_norm": 0.04854578803874495,
      "learning_rate": 4.3670289320605055e-06,
      "loss": 0.8418,
      "step": 265330
    },
    {
      "epoch": 4.658438525957267,
      "grad_norm": 0.052874573708117906,
      "learning_rate": 4.364607319699284e-06,
      "loss": 0.8383,
      "step": 265340
    },
    {
      "epoch": 4.658614090837269,
      "grad_norm": 0.05361078392526641,
      "learning_rate": 4.362186931727015e-06,
      "loss": 0.8427,
      "step": 265350
    },
    {
      "epoch": 4.65878965571727,
      "grad_norm": 0.052272235565682335,
      "learning_rate": 4.359767768174344e-06,
      "loss": 0.8403,
      "step": 265360
    },
    {
      "epoch": 4.658965220597271,
      "grad_norm": 0.05220970042361028,
      "learning_rate": 4.357349829071948e-06,
      "loss": 0.8422,
      "step": 265370
    },
    {
      "epoch": 4.659140785477273,
      "grad_norm": 0.05952320470460217,
      "learning_rate": 4.354933114450448e-06,
      "loss": 0.8378,
      "step": 265380
    },
    {
      "epoch": 4.659316350357274,
      "grad_norm": 0.06152824052229581,
      "learning_rate": 4.352517624340465e-06,
      "loss": 0.8322,
      "step": 265390
    },
    {
      "epoch": 4.659491915237276,
      "grad_norm": 0.053377477000211115,
      "learning_rate": 4.350103358772609e-06,
      "loss": 0.8386,
      "step": 265400
    },
    {
      "epoch": 4.659667480117277,
      "grad_norm": 0.04744644368136896,
      "learning_rate": 4.347690317777468e-06,
      "loss": 0.8412,
      "step": 265410
    },
    {
      "epoch": 4.659843044997279,
      "grad_norm": 0.06262292955444114,
      "learning_rate": 4.345278501385601e-06,
      "loss": 0.8366,
      "step": 265420
    },
    {
      "epoch": 4.66001860987728,
      "grad_norm": 0.06551791081072775,
      "learning_rate": 4.342867909627593e-06,
      "loss": 0.8377,
      "step": 265430
    },
    {
      "epoch": 4.660194174757281,
      "grad_norm": 0.0449105417253754,
      "learning_rate": 4.340458542533958e-06,
      "loss": 0.8395,
      "step": 265440
    },
    {
      "epoch": 4.660369739637283,
      "grad_norm": 0.05742282971527073,
      "learning_rate": 4.3380504001352615e-06,
      "loss": 0.8426,
      "step": 265450
    },
    {
      "epoch": 4.660545304517284,
      "grad_norm": 0.05553692246552626,
      "learning_rate": 4.33564348246197e-06,
      "loss": 0.8385,
      "step": 265460
    },
    {
      "epoch": 4.660720869397286,
      "grad_norm": 0.05604189290464433,
      "learning_rate": 4.3332377895446075e-06,
      "loss": 0.8378,
      "step": 265470
    },
    {
      "epoch": 4.660896434277287,
      "grad_norm": 0.07381937687519025,
      "learning_rate": 4.330833321413664e-06,
      "loss": 0.8374,
      "step": 265480
    },
    {
      "epoch": 4.661071999157288,
      "grad_norm": 0.05609552083028568,
      "learning_rate": 4.328430078099562e-06,
      "loss": 0.8417,
      "step": 265490
    },
    {
      "epoch": 4.66124756403729,
      "grad_norm": 0.04966165038782332,
      "learning_rate": 4.326028059632803e-06,
      "loss": 0.8403,
      "step": 265500
    },
    {
      "epoch": 4.661423128917291,
      "grad_norm": 0.05405034629802183,
      "learning_rate": 4.323627266043776e-06,
      "loss": 0.8396,
      "step": 265510
    },
    {
      "epoch": 4.661598693797293,
      "grad_norm": 0.06058425518726347,
      "learning_rate": 4.3212276973629394e-06,
      "loss": 0.8402,
      "step": 265520
    },
    {
      "epoch": 4.661774258677294,
      "grad_norm": 0.051116526998914696,
      "learning_rate": 4.318829353620663e-06,
      "loss": 0.836,
      "step": 265530
    },
    {
      "epoch": 4.661949823557295,
      "grad_norm": 0.054169478382998903,
      "learning_rate": 4.316432234847344e-06,
      "loss": 0.8337,
      "step": 265540
    },
    {
      "epoch": 4.662125388437297,
      "grad_norm": 0.04677932456739652,
      "learning_rate": 4.3140363410733676e-06,
      "loss": 0.8418,
      "step": 265550
    },
    {
      "epoch": 4.662300953317298,
      "grad_norm": 0.06596296038891632,
      "learning_rate": 4.3116416723290764e-06,
      "loss": 0.8398,
      "step": 265560
    },
    {
      "epoch": 4.6624765181973,
      "grad_norm": 0.04603541965423012,
      "learning_rate": 4.3092482286448074e-06,
      "loss": 0.8346,
      "step": 265570
    },
    {
      "epoch": 4.662652083077301,
      "grad_norm": 0.052126860655194555,
      "learning_rate": 4.306856010050908e-06,
      "loss": 0.8273,
      "step": 265580
    },
    {
      "epoch": 4.662827647957303,
      "grad_norm": 0.04510801114984144,
      "learning_rate": 4.304465016577659e-06,
      "loss": 0.8334,
      "step": 265590
    },
    {
      "epoch": 4.663003212837304,
      "grad_norm": 0.0444047452734566,
      "learning_rate": 4.3020752482553845e-06,
      "loss": 0.8294,
      "step": 265600
    },
    {
      "epoch": 4.663178777717305,
      "grad_norm": 0.050222164786892474,
      "learning_rate": 4.2996867051143436e-06,
      "loss": 0.8336,
      "step": 265610
    },
    {
      "epoch": 4.663354342597307,
      "grad_norm": 0.044587034969284636,
      "learning_rate": 4.2972993871848056e-06,
      "loss": 0.846,
      "step": 265620
    },
    {
      "epoch": 4.663529907477308,
      "grad_norm": 0.05160506683138448,
      "learning_rate": 4.294913294497019e-06,
      "loss": 0.8526,
      "step": 265630
    },
    {
      "epoch": 4.66370547235731,
      "grad_norm": 0.06845267640501167,
      "learning_rate": 4.292528427081208e-06,
      "loss": 0.8398,
      "step": 265640
    },
    {
      "epoch": 4.663881037237311,
      "grad_norm": 0.047741714995929586,
      "learning_rate": 4.290144784967601e-06,
      "loss": 0.842,
      "step": 265650
    },
    {
      "epoch": 4.664056602117313,
      "grad_norm": 0.04899951087333491,
      "learning_rate": 4.287762368186378e-06,
      "loss": 0.8339,
      "step": 265660
    },
    {
      "epoch": 4.664232166997314,
      "grad_norm": 0.041889489722192905,
      "learning_rate": 4.285381176767767e-06,
      "loss": 0.8425,
      "step": 265670
    },
    {
      "epoch": 4.664407731877315,
      "grad_norm": 0.04432999181290407,
      "learning_rate": 4.283001210741891e-06,
      "loss": 0.8446,
      "step": 265680
    },
    {
      "epoch": 4.664583296757317,
      "grad_norm": 0.05317577189194004,
      "learning_rate": 4.280622470138946e-06,
      "loss": 0.8383,
      "step": 265690
    },
    {
      "epoch": 4.664758861637318,
      "grad_norm": 0.053850042860729315,
      "learning_rate": 4.278244954989036e-06,
      "loss": 0.841,
      "step": 265700
    },
    {
      "epoch": 4.66493442651732,
      "grad_norm": 0.06305143996621833,
      "learning_rate": 4.2758686653223e-06,
      "loss": 0.8447,
      "step": 265710
    },
    {
      "epoch": 4.665109991397321,
      "grad_norm": 0.048097893412389355,
      "learning_rate": 4.273493601168852e-06,
      "loss": 0.8435,
      "step": 265720
    },
    {
      "epoch": 4.665285556277322,
      "grad_norm": 0.05893582780113315,
      "learning_rate": 4.271119762558763e-06,
      "loss": 0.8409,
      "step": 265730
    },
    {
      "epoch": 4.665461121157324,
      "grad_norm": 0.04894072827444121,
      "learning_rate": 4.2687471495221316e-06,
      "loss": 0.8393,
      "step": 265740
    },
    {
      "epoch": 4.665636686037325,
      "grad_norm": 0.06146187261879387,
      "learning_rate": 4.266375762089025e-06,
      "loss": 0.8372,
      "step": 265750
    },
    {
      "epoch": 4.665812250917327,
      "grad_norm": 0.04980956587009297,
      "learning_rate": 4.264005600289452e-06,
      "loss": 0.8363,
      "step": 265760
    },
    {
      "epoch": 4.665987815797328,
      "grad_norm": 0.053836010697510775,
      "learning_rate": 4.261636664153483e-06,
      "loss": 0.8392,
      "step": 265770
    },
    {
      "epoch": 4.666163380677329,
      "grad_norm": 0.08011089157167062,
      "learning_rate": 4.259268953711114e-06,
      "loss": 0.8284,
      "step": 265780
    },
    {
      "epoch": 4.666338945557331,
      "grad_norm": 0.04697104487817675,
      "learning_rate": 4.256902468992362e-06,
      "loss": 0.8375,
      "step": 265790
    },
    {
      "epoch": 4.666514510437332,
      "grad_norm": 0.044195051891115056,
      "learning_rate": 4.254537210027177e-06,
      "loss": 0.8336,
      "step": 265800
    },
    {
      "epoch": 4.666690075317334,
      "grad_norm": 0.03950652300914237,
      "learning_rate": 4.252173176845555e-06,
      "loss": 0.8451,
      "step": 265810
    },
    {
      "epoch": 4.666865640197335,
      "grad_norm": 0.045853023133502695,
      "learning_rate": 4.249810369477435e-06,
      "loss": 0.8433,
      "step": 265820
    },
    {
      "epoch": 4.667041205077337,
      "grad_norm": 0.051341046144287944,
      "learning_rate": 4.247448787952758e-06,
      "loss": 0.8372,
      "step": 265830
    },
    {
      "epoch": 4.667216769957338,
      "grad_norm": 0.0522277858571077,
      "learning_rate": 4.2450884323014546e-06,
      "loss": 0.8321,
      "step": 265840
    },
    {
      "epoch": 4.66739233483734,
      "grad_norm": 0.04512023593226506,
      "learning_rate": 4.2427293025534274e-06,
      "loss": 0.8375,
      "step": 265850
    },
    {
      "epoch": 4.667567899717341,
      "grad_norm": 0.0487335105420395,
      "learning_rate": 4.240371398738555e-06,
      "loss": 0.8332,
      "step": 265860
    },
    {
      "epoch": 4.667743464597342,
      "grad_norm": 0.059233626308728075,
      "learning_rate": 4.238014720886731e-06,
      "loss": 0.838,
      "step": 265870
    },
    {
      "epoch": 4.667919029477344,
      "grad_norm": 0.056216130355998796,
      "learning_rate": 4.235659269027786e-06,
      "loss": 0.8393,
      "step": 265880
    },
    {
      "epoch": 4.668094594357345,
      "grad_norm": 0.04915215729364625,
      "learning_rate": 4.233305043191606e-06,
      "loss": 0.8371,
      "step": 265890
    },
    {
      "epoch": 4.668270159237347,
      "grad_norm": 0.06936758371237513,
      "learning_rate": 4.2309520434079756e-06,
      "loss": 0.8353,
      "step": 265900
    },
    {
      "epoch": 4.668445724117348,
      "grad_norm": 0.06407505506255586,
      "learning_rate": 4.228600269706738e-06,
      "loss": 0.8408,
      "step": 265910
    },
    {
      "epoch": 4.668621288997349,
      "grad_norm": 0.05346090063868166,
      "learning_rate": 4.226249722117688e-06,
      "loss": 0.8373,
      "step": 265920
    },
    {
      "epoch": 4.6687968538773506,
      "grad_norm": 0.046841462324272644,
      "learning_rate": 4.22390040067058e-06,
      "loss": 0.8334,
      "step": 265930
    },
    {
      "epoch": 4.668972418757352,
      "grad_norm": 0.05673824642806295,
      "learning_rate": 4.2215523053952235e-06,
      "loss": 0.8455,
      "step": 265940
    },
    {
      "epoch": 4.6691479836373535,
      "grad_norm": 0.04967566094309692,
      "learning_rate": 4.2192054363213355e-06,
      "loss": 0.841,
      "step": 265950
    },
    {
      "epoch": 4.6693235485173545,
      "grad_norm": 0.05466150611273102,
      "learning_rate": 4.216859793478672e-06,
      "loss": 0.8433,
      "step": 265960
    },
    {
      "epoch": 4.669499113397356,
      "grad_norm": 0.05272064215812018,
      "learning_rate": 4.214515376896931e-06,
      "loss": 0.838,
      "step": 265970
    },
    {
      "epoch": 4.6696746782773575,
      "grad_norm": 0.05598441520952931,
      "learning_rate": 4.212172186605831e-06,
      "loss": 0.8317,
      "step": 265980
    },
    {
      "epoch": 4.6698502431573585,
      "grad_norm": 0.048135629300452856,
      "learning_rate": 4.209830222635072e-06,
      "loss": 0.8331,
      "step": 265990
    },
    {
      "epoch": 4.6700258080373604,
      "grad_norm": 0.05140265360119182,
      "learning_rate": 4.207489485014285e-06,
      "loss": 0.8413,
      "step": 266000
    },
    {
      "epoch": 4.6702013729173615,
      "grad_norm": 0.054729050623292474,
      "learning_rate": 4.205149973773191e-06,
      "loss": 0.8342,
      "step": 266010
    },
    {
      "epoch": 4.670376937797363,
      "grad_norm": 0.07540270959167492,
      "learning_rate": 4.202811688941379e-06,
      "loss": 0.8349,
      "step": 266020
    },
    {
      "epoch": 4.670552502677364,
      "grad_norm": 0.05988521721382386,
      "learning_rate": 4.200474630548501e-06,
      "loss": 0.8372,
      "step": 266030
    },
    {
      "epoch": 4.6707280675573655,
      "grad_norm": 0.05155999015250164,
      "learning_rate": 4.198138798624147e-06,
      "loss": 0.833,
      "step": 266040
    },
    {
      "epoch": 4.670903632437367,
      "grad_norm": 0.0647357911864209,
      "learning_rate": 4.195804193197939e-06,
      "loss": 0.841,
      "step": 266050
    },
    {
      "epoch": 4.671079197317368,
      "grad_norm": 0.04900386200470412,
      "learning_rate": 4.193470814299452e-06,
      "loss": 0.8365,
      "step": 266060
    },
    {
      "epoch": 4.67125476219737,
      "grad_norm": 0.048800972141669126,
      "learning_rate": 4.191138661958221e-06,
      "loss": 0.8424,
      "step": 266070
    },
    {
      "epoch": 4.671430327077371,
      "grad_norm": 0.047505544580597386,
      "learning_rate": 4.188807736203825e-06,
      "loss": 0.8469,
      "step": 266080
    },
    {
      "epoch": 4.671605891957373,
      "grad_norm": 0.05379836658351921,
      "learning_rate": 4.186478037065804e-06,
      "loss": 0.8281,
      "step": 266090
    },
    {
      "epoch": 4.671781456837374,
      "grad_norm": 0.055934955762778085,
      "learning_rate": 4.1841495645736295e-06,
      "loss": 0.844,
      "step": 266100
    },
    {
      "epoch": 4.671957021717375,
      "grad_norm": 0.04861937661394627,
      "learning_rate": 4.181822318756866e-06,
      "loss": 0.8365,
      "step": 266110
    },
    {
      "epoch": 4.672132586597377,
      "grad_norm": 0.046090690684590624,
      "learning_rate": 4.179496299644947e-06,
      "loss": 0.8352,
      "step": 266120
    },
    {
      "epoch": 4.672308151477378,
      "grad_norm": 0.07460998179795268,
      "learning_rate": 4.177171507267386e-06,
      "loss": 0.8446,
      "step": 266130
    },
    {
      "epoch": 4.67248371635738,
      "grad_norm": 0.046808622748815074,
      "learning_rate": 4.174847941653607e-06,
      "loss": 0.8368,
      "step": 266140
    },
    {
      "epoch": 4.672659281237381,
      "grad_norm": 0.0472288063617694,
      "learning_rate": 4.172525602833051e-06,
      "loss": 0.8444,
      "step": 266150
    },
    {
      "epoch": 4.672834846117382,
      "grad_norm": 0.05839807196018264,
      "learning_rate": 4.170204490835168e-06,
      "loss": 0.8316,
      "step": 266160
    },
    {
      "epoch": 4.673010410997384,
      "grad_norm": 0.04926396714143295,
      "learning_rate": 4.167884605689336e-06,
      "loss": 0.8354,
      "step": 266170
    },
    {
      "epoch": 4.673185975877385,
      "grad_norm": 0.05727439023972962,
      "learning_rate": 4.165565947424979e-06,
      "loss": 0.8383,
      "step": 266180
    },
    {
      "epoch": 4.673361540757387,
      "grad_norm": 0.07023896886856934,
      "learning_rate": 4.163248516071464e-06,
      "loss": 0.8435,
      "step": 266190
    },
    {
      "epoch": 4.673537105637388,
      "grad_norm": 0.05239959339981659,
      "learning_rate": 4.160932311658128e-06,
      "loss": 0.8341,
      "step": 266200
    },
    {
      "epoch": 4.673712670517389,
      "grad_norm": 0.04574998346044508,
      "learning_rate": 4.15861733421435e-06,
      "loss": 0.8369,
      "step": 266210
    },
    {
      "epoch": 4.673888235397391,
      "grad_norm": 0.04870733492327791,
      "learning_rate": 4.1563035837694425e-06,
      "loss": 0.8438,
      "step": 266220
    },
    {
      "epoch": 4.674063800277392,
      "grad_norm": 0.04938349264048995,
      "learning_rate": 4.153991060352743e-06,
      "loss": 0.8342,
      "step": 266230
    },
    {
      "epoch": 4.674239365157394,
      "grad_norm": 0.060625654788288665,
      "learning_rate": 4.151679763993531e-06,
      "loss": 0.8374,
      "step": 266240
    },
    {
      "epoch": 4.674414930037395,
      "grad_norm": 0.04282955694759285,
      "learning_rate": 4.149369694721098e-06,
      "loss": 0.836,
      "step": 266250
    },
    {
      "epoch": 4.674590494917397,
      "grad_norm": 0.051519627990433435,
      "learning_rate": 4.147060852564714e-06,
      "loss": 0.8405,
      "step": 266260
    },
    {
      "epoch": 4.674766059797398,
      "grad_norm": 0.04838209057172984,
      "learning_rate": 4.144753237553616e-06,
      "loss": 0.8422,
      "step": 266270
    },
    {
      "epoch": 4.674941624677399,
      "grad_norm": 0.05229888144152544,
      "learning_rate": 4.142446849717071e-06,
      "loss": 0.8331,
      "step": 266280
    },
    {
      "epoch": 4.675117189557401,
      "grad_norm": 0.05653964265342499,
      "learning_rate": 4.140141689084286e-06,
      "loss": 0.8349,
      "step": 266290
    },
    {
      "epoch": 4.675292754437402,
      "grad_norm": 0.06113543505129948,
      "learning_rate": 4.137837755684474e-06,
      "loss": 0.8376,
      "step": 266300
    },
    {
      "epoch": 4.675468319317404,
      "grad_norm": 0.061519716239832746,
      "learning_rate": 4.1355350495468055e-06,
      "loss": 0.8399,
      "step": 266310
    },
    {
      "epoch": 4.675643884197405,
      "grad_norm": 0.047933442384507065,
      "learning_rate": 4.133233570700474e-06,
      "loss": 0.8345,
      "step": 266320
    },
    {
      "epoch": 4.675819449077407,
      "grad_norm": 0.06017502468377595,
      "learning_rate": 4.13093331917465e-06,
      "loss": 0.8438,
      "step": 266330
    },
    {
      "epoch": 4.675995013957408,
      "grad_norm": 0.043836420650104554,
      "learning_rate": 4.128634294998448e-06,
      "loss": 0.8337,
      "step": 266340
    },
    {
      "epoch": 4.676170578837409,
      "grad_norm": 0.0657668555058995,
      "learning_rate": 4.126336498201041e-06,
      "loss": 0.834,
      "step": 266350
    },
    {
      "epoch": 4.676346143717411,
      "grad_norm": 0.04775484234332777,
      "learning_rate": 4.12403992881151e-06,
      "loss": 0.8384,
      "step": 266360
    },
    {
      "epoch": 4.676521708597412,
      "grad_norm": 0.06941045791473106,
      "learning_rate": 4.121744586858938e-06,
      "loss": 0.8343,
      "step": 266370
    },
    {
      "epoch": 4.676697273477414,
      "grad_norm": 0.04926311554912067,
      "learning_rate": 4.11945047237244e-06,
      "loss": 0.8321,
      "step": 266380
    },
    {
      "epoch": 4.676872838357415,
      "grad_norm": 0.04557572705129414,
      "learning_rate": 4.1171575853810775e-06,
      "loss": 0.8348,
      "step": 266390
    },
    {
      "epoch": 4.677048403237416,
      "grad_norm": 0.055111793168453466,
      "learning_rate": 4.11486592591389e-06,
      "loss": 0.8413,
      "step": 266400
    },
    {
      "epoch": 4.677223968117418,
      "grad_norm": 0.04480991477702378,
      "learning_rate": 4.112575493999927e-06,
      "loss": 0.8374,
      "step": 266410
    },
    {
      "epoch": 4.677399532997419,
      "grad_norm": 0.046752001547907505,
      "learning_rate": 4.110286289668193e-06,
      "loss": 0.8324,
      "step": 266420
    },
    {
      "epoch": 4.677575097877421,
      "grad_norm": 0.0553335350173884,
      "learning_rate": 4.1079983129477055e-06,
      "loss": 0.8435,
      "step": 266430
    },
    {
      "epoch": 4.677750662757422,
      "grad_norm": 0.045082253124120106,
      "learning_rate": 4.105711563867438e-06,
      "loss": 0.8415,
      "step": 266440
    },
    {
      "epoch": 4.677926227637423,
      "grad_norm": 0.04584229687853325,
      "learning_rate": 4.1034260424563845e-06,
      "loss": 0.838,
      "step": 266450
    },
    {
      "epoch": 4.678101792517425,
      "grad_norm": 0.051821742830595915,
      "learning_rate": 4.101141748743486e-06,
      "loss": 0.8377,
      "step": 266460
    },
    {
      "epoch": 4.678277357397426,
      "grad_norm": 0.061684019186398376,
      "learning_rate": 4.098858682757703e-06,
      "loss": 0.8374,
      "step": 266470
    },
    {
      "epoch": 4.678452922277428,
      "grad_norm": 0.04881409866333755,
      "learning_rate": 4.096576844527942e-06,
      "loss": 0.8355,
      "step": 266480
    },
    {
      "epoch": 4.678628487157429,
      "grad_norm": 0.048732622439878255,
      "learning_rate": 4.094296234083122e-06,
      "loss": 0.8367,
      "step": 266490
    },
    {
      "epoch": 4.678804052037431,
      "grad_norm": 0.08698005184767361,
      "learning_rate": 4.092016851452162e-06,
      "loss": 0.8399,
      "step": 266500
    },
    {
      "epoch": 4.678979616917432,
      "grad_norm": 0.05843450876619993,
      "learning_rate": 4.089738696663889e-06,
      "loss": 0.837,
      "step": 266510
    },
    {
      "epoch": 4.679155181797434,
      "grad_norm": 0.05048610495455827,
      "learning_rate": 4.087461769747223e-06,
      "loss": 0.8336,
      "step": 266520
    },
    {
      "epoch": 4.679330746677435,
      "grad_norm": 0.049907547188900094,
      "learning_rate": 4.085186070730982e-06,
      "loss": 0.842,
      "step": 266530
    },
    {
      "epoch": 4.679506311557436,
      "grad_norm": 0.05440961668103802,
      "learning_rate": 4.082911599644007e-06,
      "loss": 0.8395,
      "step": 266540
    },
    {
      "epoch": 4.679681876437438,
      "grad_norm": 0.044076052624389205,
      "learning_rate": 4.080638356515118e-06,
      "loss": 0.8361,
      "step": 266550
    },
    {
      "epoch": 4.679857441317439,
      "grad_norm": 0.05704032950178155,
      "learning_rate": 4.078366341373111e-06,
      "loss": 0.833,
      "step": 266560
    },
    {
      "epoch": 4.680033006197441,
      "grad_norm": 0.04470858184419258,
      "learning_rate": 4.0760955542467835e-06,
      "loss": 0.8354,
      "step": 266570
    },
    {
      "epoch": 4.680208571077442,
      "grad_norm": 0.0496052959728219,
      "learning_rate": 4.073825995164898e-06,
      "loss": 0.8379,
      "step": 266580
    },
    {
      "epoch": 4.680384135957443,
      "grad_norm": 0.04528067536477982,
      "learning_rate": 4.07155766415621e-06,
      "loss": 0.8379,
      "step": 266590
    },
    {
      "epoch": 4.680559700837445,
      "grad_norm": 0.05805193151413004,
      "learning_rate": 4.06929056124947e-06,
      "loss": 0.8366,
      "step": 266600
    },
    {
      "epoch": 4.680735265717446,
      "grad_norm": 0.055131614241161725,
      "learning_rate": 4.067024686473378e-06,
      "loss": 0.833,
      "step": 266610
    },
    {
      "epoch": 4.680910830597448,
      "grad_norm": 0.0450563724495724,
      "learning_rate": 4.064760039856674e-06,
      "loss": 0.843,
      "step": 266620
    },
    {
      "epoch": 4.681086395477449,
      "grad_norm": 0.05135533776211319,
      "learning_rate": 4.062496621428036e-06,
      "loss": 0.8362,
      "step": 266630
    },
    {
      "epoch": 4.68126196035745,
      "grad_norm": 0.053458573842434645,
      "learning_rate": 4.060234431216151e-06,
      "loss": 0.8445,
      "step": 266640
    },
    {
      "epoch": 4.681437525237452,
      "grad_norm": 0.04683713077323124,
      "learning_rate": 4.0579734692496606e-06,
      "loss": 0.8366,
      "step": 266650
    },
    {
      "epoch": 4.681613090117453,
      "grad_norm": 0.05561733556361488,
      "learning_rate": 4.05571373555723e-06,
      "loss": 0.8422,
      "step": 266660
    },
    {
      "epoch": 4.681788654997455,
      "grad_norm": 0.04857280803490181,
      "learning_rate": 4.053455230167493e-06,
      "loss": 0.8387,
      "step": 266670
    },
    {
      "epoch": 4.681964219877456,
      "grad_norm": 0.05506541154648383,
      "learning_rate": 4.051197953109048e-06,
      "loss": 0.8362,
      "step": 266680
    },
    {
      "epoch": 4.682139784757457,
      "grad_norm": 0.04183681532906464,
      "learning_rate": 4.048941904410515e-06,
      "loss": 0.8315,
      "step": 266690
    },
    {
      "epoch": 4.6823153496374585,
      "grad_norm": 0.06236881496468625,
      "learning_rate": 4.046687084100472e-06,
      "loss": 0.8363,
      "step": 266700
    },
    {
      "epoch": 4.68249091451746,
      "grad_norm": 0.042018771729010786,
      "learning_rate": 4.0444334922074745e-06,
      "loss": 0.8388,
      "step": 266710
    },
    {
      "epoch": 4.6826664793974615,
      "grad_norm": 0.0543086992509147,
      "learning_rate": 4.042181128760077e-06,
      "loss": 0.8324,
      "step": 266720
    },
    {
      "epoch": 4.6828420442774625,
      "grad_norm": 0.05099226674041593,
      "learning_rate": 4.039929993786835e-06,
      "loss": 0.835,
      "step": 266730
    },
    {
      "epoch": 4.6830176091574645,
      "grad_norm": 0.05349323900765579,
      "learning_rate": 4.03768008731627e-06,
      "loss": 0.8298,
      "step": 266740
    },
    {
      "epoch": 4.6831931740374655,
      "grad_norm": 0.051679155860820135,
      "learning_rate": 4.035431409376872e-06,
      "loss": 0.8378,
      "step": 266750
    },
    {
      "epoch": 4.683368738917467,
      "grad_norm": 0.0626765744693269,
      "learning_rate": 4.033183959997141e-06,
      "loss": 0.8406,
      "step": 266760
    },
    {
      "epoch": 4.6835443037974684,
      "grad_norm": 0.056205238620012074,
      "learning_rate": 4.030937739205567e-06,
      "loss": 0.8455,
      "step": 266770
    },
    {
      "epoch": 4.6837198686774695,
      "grad_norm": 0.058809764338237244,
      "learning_rate": 4.028692747030572e-06,
      "loss": 0.8367,
      "step": 266780
    },
    {
      "epoch": 4.683895433557471,
      "grad_norm": 0.04762997079605067,
      "learning_rate": 4.026448983500645e-06,
      "loss": 0.8394,
      "step": 266790
    },
    {
      "epoch": 4.684070998437472,
      "grad_norm": 0.04856248317318233,
      "learning_rate": 4.024206448644189e-06,
      "loss": 0.8384,
      "step": 266800
    },
    {
      "epoch": 4.684246563317474,
      "grad_norm": 0.054537555573307694,
      "learning_rate": 4.0219651424896255e-06,
      "loss": 0.8291,
      "step": 266810
    },
    {
      "epoch": 4.684422128197475,
      "grad_norm": 0.04937819952416661,
      "learning_rate": 4.0197250650653355e-06,
      "loss": 0.8306,
      "step": 266820
    },
    {
      "epoch": 4.684597693077476,
      "grad_norm": 0.04638174291213993,
      "learning_rate": 4.017486216399729e-06,
      "loss": 0.8339,
      "step": 266830
    },
    {
      "epoch": 4.684773257957478,
      "grad_norm": 0.05248566699816836,
      "learning_rate": 4.015248596521155e-06,
      "loss": 0.8415,
      "step": 266840
    },
    {
      "epoch": 4.684948822837479,
      "grad_norm": 0.05535330679461186,
      "learning_rate": 4.013012205457957e-06,
      "loss": 0.8436,
      "step": 266850
    },
    {
      "epoch": 4.685124387717481,
      "grad_norm": 0.06396312846552213,
      "learning_rate": 4.010777043238495e-06,
      "loss": 0.8418,
      "step": 266860
    },
    {
      "epoch": 4.685299952597482,
      "grad_norm": 0.04511677646578989,
      "learning_rate": 4.008543109891081e-06,
      "loss": 0.8323,
      "step": 266870
    },
    {
      "epoch": 4.685475517477483,
      "grad_norm": 0.0530175021914293,
      "learning_rate": 4.006310405444005e-06,
      "loss": 0.8496,
      "step": 266880
    },
    {
      "epoch": 4.685651082357485,
      "grad_norm": 0.05125673461785178,
      "learning_rate": 4.004078929925562e-06,
      "loss": 0.8353,
      "step": 266890
    },
    {
      "epoch": 4.685826647237486,
      "grad_norm": 0.04794601697423195,
      "learning_rate": 4.0018486833640286e-06,
      "loss": 0.8414,
      "step": 266900
    },
    {
      "epoch": 4.686002212117488,
      "grad_norm": 0.047257772389121974,
      "learning_rate": 3.9996196657876766e-06,
      "loss": 0.8362,
      "step": 266910
    },
    {
      "epoch": 4.686177776997489,
      "grad_norm": 0.04236968302833078,
      "learning_rate": 3.99739187722472e-06,
      "loss": 0.8406,
      "step": 266920
    },
    {
      "epoch": 4.686353341877491,
      "grad_norm": 0.05771800886024948,
      "learning_rate": 3.995165317703394e-06,
      "loss": 0.8421,
      "step": 266930
    },
    {
      "epoch": 4.686528906757492,
      "grad_norm": 0.05368147210072871,
      "learning_rate": 3.992939987251924e-06,
      "loss": 0.8322,
      "step": 266940
    },
    {
      "epoch": 4.686704471637493,
      "grad_norm": 0.050833270593746,
      "learning_rate": 3.990715885898483e-06,
      "loss": 0.8343,
      "step": 266950
    },
    {
      "epoch": 4.686880036517495,
      "grad_norm": 0.057729895305004456,
      "learning_rate": 3.988493013671284e-06,
      "loss": 0.8457,
      "step": 266960
    },
    {
      "epoch": 4.687055601397496,
      "grad_norm": 0.04847196499325078,
      "learning_rate": 3.986271370598452e-06,
      "loss": 0.8326,
      "step": 266970
    },
    {
      "epoch": 4.687231166277498,
      "grad_norm": 0.06949748785995478,
      "learning_rate": 3.984050956708171e-06,
      "loss": 0.8437,
      "step": 266980
    },
    {
      "epoch": 4.687406731157499,
      "grad_norm": 0.04920759622331658,
      "learning_rate": 3.9818317720285435e-06,
      "loss": 0.8357,
      "step": 266990
    },
    {
      "epoch": 4.687582296037501,
      "grad_norm": 0.06976097525782018,
      "learning_rate": 3.979613816587708e-06,
      "loss": 0.8395,
      "step": 267000
    },
    {
      "epoch": 4.687757860917502,
      "grad_norm": 0.06775888860421951,
      "learning_rate": 3.9773970904137574e-06,
      "loss": 0.8378,
      "step": 267010
    },
    {
      "epoch": 4.687933425797503,
      "grad_norm": 0.04325731051919898,
      "learning_rate": 3.975181593534775e-06,
      "loss": 0.8416,
      "step": 267020
    },
    {
      "epoch": 4.688108990677505,
      "grad_norm": 0.05319126073454429,
      "learning_rate": 3.972967325978843e-06,
      "loss": 0.845,
      "step": 267030
    },
    {
      "epoch": 4.688284555557506,
      "grad_norm": 0.05127940772790607,
      "learning_rate": 3.970754287774021e-06,
      "loss": 0.8355,
      "step": 267040
    },
    {
      "epoch": 4.688460120437508,
      "grad_norm": 0.05352201072357986,
      "learning_rate": 3.968542478948317e-06,
      "loss": 0.8361,
      "step": 267050
    },
    {
      "epoch": 4.688635685317509,
      "grad_norm": 0.052326633673514304,
      "learning_rate": 3.966331899529789e-06,
      "loss": 0.8396,
      "step": 267060
    },
    {
      "epoch": 4.68881125019751,
      "grad_norm": 0.04742240410797343,
      "learning_rate": 3.964122549546423e-06,
      "loss": 0.8374,
      "step": 267070
    },
    {
      "epoch": 4.688986815077512,
      "grad_norm": 0.04721589555397845,
      "learning_rate": 3.9619144290262325e-06,
      "loss": 0.8357,
      "step": 267080
    },
    {
      "epoch": 4.689162379957513,
      "grad_norm": 0.05892084909712918,
      "learning_rate": 3.959707537997171e-06,
      "loss": 0.8334,
      "step": 267090
    },
    {
      "epoch": 4.689337944837515,
      "grad_norm": 0.057966807692451036,
      "learning_rate": 3.95750187648722e-06,
      "loss": 0.8348,
      "step": 267100
    },
    {
      "epoch": 4.689513509717516,
      "grad_norm": 0.05946028953595775,
      "learning_rate": 3.9552974445243215e-06,
      "loss": 0.8399,
      "step": 267110
    },
    {
      "epoch": 4.689689074597517,
      "grad_norm": 0.057031723448869316,
      "learning_rate": 3.95309424213638e-06,
      "loss": 0.8364,
      "step": 267120
    },
    {
      "epoch": 4.689864639477519,
      "grad_norm": 0.041991433074674234,
      "learning_rate": 3.950892269351358e-06,
      "loss": 0.8316,
      "step": 267130
    },
    {
      "epoch": 4.69004020435752,
      "grad_norm": 0.05276576293307931,
      "learning_rate": 3.94869152619712e-06,
      "loss": 0.8408,
      "step": 267140
    },
    {
      "epoch": 4.690215769237522,
      "grad_norm": 0.04741873694795131,
      "learning_rate": 3.946492012701579e-06,
      "loss": 0.8366,
      "step": 267150
    },
    {
      "epoch": 4.690391334117523,
      "grad_norm": 0.07640464230386283,
      "learning_rate": 3.944293728892557e-06,
      "loss": 0.8377,
      "step": 267160
    },
    {
      "epoch": 4.690566898997525,
      "grad_norm": 0.05308150167577109,
      "learning_rate": 3.942096674797951e-06,
      "loss": 0.8347,
      "step": 267170
    },
    {
      "epoch": 4.690742463877526,
      "grad_norm": 0.05741710016525087,
      "learning_rate": 3.939900850445578e-06,
      "loss": 0.8366,
      "step": 267180
    },
    {
      "epoch": 4.690918028757528,
      "grad_norm": 0.055490616802752656,
      "learning_rate": 3.937706255863255e-06,
      "loss": 0.8394,
      "step": 267190
    },
    {
      "epoch": 4.691093593637529,
      "grad_norm": 0.04912146513138575,
      "learning_rate": 3.935512891078805e-06,
      "loss": 0.8388,
      "step": 267200
    },
    {
      "epoch": 4.69126915851753,
      "grad_norm": 0.047236394393004484,
      "learning_rate": 3.933320756120009e-06,
      "loss": 0.8366,
      "step": 267210
    },
    {
      "epoch": 4.691444723397532,
      "grad_norm": 0.04895519765991052,
      "learning_rate": 3.931129851014645e-06,
      "loss": 0.8321,
      "step": 267220
    },
    {
      "epoch": 4.691620288277533,
      "grad_norm": 0.04788850832137085,
      "learning_rate": 3.928940175790454e-06,
      "loss": 0.8371,
      "step": 267230
    },
    {
      "epoch": 4.691795853157535,
      "grad_norm": 0.05651568763324595,
      "learning_rate": 3.926751730475199e-06,
      "loss": 0.8406,
      "step": 267240
    },
    {
      "epoch": 4.691971418037536,
      "grad_norm": 0.046580418260578474,
      "learning_rate": 3.924564515096612e-06,
      "loss": 0.8344,
      "step": 267250
    },
    {
      "epoch": 4.692146982917537,
      "grad_norm": 0.04745823837022958,
      "learning_rate": 3.922378529682391e-06,
      "loss": 0.8436,
      "step": 267260
    },
    {
      "epoch": 4.692322547797539,
      "grad_norm": 0.06464798656542116,
      "learning_rate": 3.920193774260244e-06,
      "loss": 0.8392,
      "step": 267270
    },
    {
      "epoch": 4.69249811267754,
      "grad_norm": 0.06433269654296371,
      "learning_rate": 3.9180102488578465e-06,
      "loss": 0.8384,
      "step": 267280
    },
    {
      "epoch": 4.692673677557542,
      "grad_norm": 0.056798808733829184,
      "learning_rate": 3.915827953502853e-06,
      "loss": 0.8389,
      "step": 267290
    },
    {
      "epoch": 4.692849242437543,
      "grad_norm": 0.05051202524819432,
      "learning_rate": 3.91364688822294e-06,
      "loss": 0.8348,
      "step": 267300
    },
    {
      "epoch": 4.693024807317544,
      "grad_norm": 0.04316900497294111,
      "learning_rate": 3.911467053045716e-06,
      "loss": 0.8327,
      "step": 267310
    },
    {
      "epoch": 4.693200372197546,
      "grad_norm": 0.061969204649330305,
      "learning_rate": 3.909288447998824e-06,
      "loss": 0.8289,
      "step": 267320
    },
    {
      "epoch": 4.693375937077547,
      "grad_norm": 0.04632290525239616,
      "learning_rate": 3.9071110731098535e-06,
      "loss": 0.8411,
      "step": 267330
    },
    {
      "epoch": 4.693551501957549,
      "grad_norm": 0.04789655239751201,
      "learning_rate": 3.90493492840638e-06,
      "loss": 0.8442,
      "step": 267340
    },
    {
      "epoch": 4.69372706683755,
      "grad_norm": 0.048759035357047516,
      "learning_rate": 3.902760013916003e-06,
      "loss": 0.8399,
      "step": 267350
    },
    {
      "epoch": 4.693902631717551,
      "grad_norm": 0.04267342851101631,
      "learning_rate": 3.900586329666245e-06,
      "loss": 0.8444,
      "step": 267360
    },
    {
      "epoch": 4.694078196597553,
      "grad_norm": 0.051475865038630456,
      "learning_rate": 3.898413875684693e-06,
      "loss": 0.8378,
      "step": 267370
    },
    {
      "epoch": 4.694253761477554,
      "grad_norm": 0.05013455026967032,
      "learning_rate": 3.896242651998836e-06,
      "loss": 0.8346,
      "step": 267380
    },
    {
      "epoch": 4.694429326357556,
      "grad_norm": 0.048565435057501834,
      "learning_rate": 3.8940726586361965e-06,
      "loss": 0.8383,
      "step": 267390
    },
    {
      "epoch": 4.694604891237557,
      "grad_norm": 0.05542765872236257,
      "learning_rate": 3.8919038956242635e-06,
      "loss": 0.8403,
      "step": 267400
    },
    {
      "epoch": 4.694780456117559,
      "grad_norm": 0.054865968758593245,
      "learning_rate": 3.889736362990514e-06,
      "loss": 0.8385,
      "step": 267410
    },
    {
      "epoch": 4.69495602099756,
      "grad_norm": 0.07115759516580686,
      "learning_rate": 3.887570060762428e-06,
      "loss": 0.8421,
      "step": 267420
    },
    {
      "epoch": 4.6951315858775615,
      "grad_norm": 0.052713372832493506,
      "learning_rate": 3.885404988967427e-06,
      "loss": 0.8339,
      "step": 267430
    },
    {
      "epoch": 4.6953071507575626,
      "grad_norm": 0.046071225661225304,
      "learning_rate": 3.883241147632967e-06,
      "loss": 0.8353,
      "step": 267440
    },
    {
      "epoch": 4.695482715637564,
      "grad_norm": 0.05047572564467077,
      "learning_rate": 3.88107853678645e-06,
      "loss": 0.8397,
      "step": 267450
    },
    {
      "epoch": 4.6956582805175655,
      "grad_norm": 0.04080515202503704,
      "learning_rate": 3.878917156455276e-06,
      "loss": 0.8353,
      "step": 267460
    },
    {
      "epoch": 4.6958338453975665,
      "grad_norm": 0.06101269994428801,
      "learning_rate": 3.876757006666847e-06,
      "loss": 0.8363,
      "step": 267470
    },
    {
      "epoch": 4.6960094102775685,
      "grad_norm": 0.05705948526749106,
      "learning_rate": 3.8745980874485095e-06,
      "loss": 0.8349,
      "step": 267480
    },
    {
      "epoch": 4.6961849751575695,
      "grad_norm": 0.06370512782669564,
      "learning_rate": 3.872440398827641e-06,
      "loss": 0.8265,
      "step": 267490
    },
    {
      "epoch": 4.6963605400375705,
      "grad_norm": 0.052396943057172976,
      "learning_rate": 3.870283940831557e-06,
      "loss": 0.8304,
      "step": 267500
    },
    {
      "epoch": 4.6965361049175725,
      "grad_norm": 0.053428428692608367,
      "learning_rate": 3.868128713487579e-06,
      "loss": 0.8399,
      "step": 267510
    },
    {
      "epoch": 4.6967116697975735,
      "grad_norm": 0.05189453285462499,
      "learning_rate": 3.8659747168230535e-06,
      "loss": 0.8426,
      "step": 267520
    },
    {
      "epoch": 4.696887234677575,
      "grad_norm": 0.05309220565313678,
      "learning_rate": 3.863821950865208e-06,
      "loss": 0.8394,
      "step": 267530
    },
    {
      "epoch": 4.697062799557576,
      "grad_norm": 0.05722076682317009,
      "learning_rate": 3.8616704156413875e-06,
      "loss": 0.832,
      "step": 267540
    },
    {
      "epoch": 4.6972383644375775,
      "grad_norm": 0.04729842804315651,
      "learning_rate": 3.859520111178816e-06,
      "loss": 0.8409,
      "step": 267550
    },
    {
      "epoch": 4.697413929317579,
      "grad_norm": 0.0473312583662443,
      "learning_rate": 3.85737103750472e-06,
      "loss": 0.8443,
      "step": 267560
    },
    {
      "epoch": 4.69758949419758,
      "grad_norm": 0.06483039769884656,
      "learning_rate": 3.855223194646368e-06,
      "loss": 0.8291,
      "step": 267570
    },
    {
      "epoch": 4.697765059077582,
      "grad_norm": 0.052502909753894905,
      "learning_rate": 3.853076582630942e-06,
      "loss": 0.8408,
      "step": 267580
    },
    {
      "epoch": 4.697940623957583,
      "grad_norm": 0.04368234106113046,
      "learning_rate": 3.8509312014856665e-06,
      "loss": 0.8475,
      "step": 267590
    },
    {
      "epoch": 4.698116188837585,
      "grad_norm": 0.05335674473652865,
      "learning_rate": 3.848787051237712e-06,
      "loss": 0.836,
      "step": 267600
    },
    {
      "epoch": 4.698291753717586,
      "grad_norm": 0.06385919307131646,
      "learning_rate": 3.846644131914239e-06,
      "loss": 0.8321,
      "step": 267610
    },
    {
      "epoch": 4.698467318597587,
      "grad_norm": 0.0583129328229909,
      "learning_rate": 3.844502443542404e-06,
      "loss": 0.8315,
      "step": 267620
    },
    {
      "epoch": 4.698642883477589,
      "grad_norm": 0.06440548850230808,
      "learning_rate": 3.842361986149336e-06,
      "loss": 0.8414,
      "step": 267630
    },
    {
      "epoch": 4.69881844835759,
      "grad_norm": 0.07329617597450065,
      "learning_rate": 3.84022275976217e-06,
      "loss": 0.8403,
      "step": 267640
    },
    {
      "epoch": 4.698994013237592,
      "grad_norm": 0.08767050665903607,
      "learning_rate": 3.838084764408001e-06,
      "loss": 0.8319,
      "step": 267650
    },
    {
      "epoch": 4.699169578117593,
      "grad_norm": 0.04160079141717355,
      "learning_rate": 3.835948000113922e-06,
      "loss": 0.8357,
      "step": 267660
    },
    {
      "epoch": 4.699345142997595,
      "grad_norm": 0.047167075279582686,
      "learning_rate": 3.833812466907005e-06,
      "loss": 0.8387,
      "step": 267670
    },
    {
      "epoch": 4.699520707877596,
      "grad_norm": 0.04555878634379115,
      "learning_rate": 3.831678164814298e-06,
      "loss": 0.837,
      "step": 267680
    },
    {
      "epoch": 4.699696272757597,
      "grad_norm": 0.04545577425876813,
      "learning_rate": 3.829545093862862e-06,
      "loss": 0.8448,
      "step": 267690
    },
    {
      "epoch": 4.699871837637599,
      "grad_norm": 0.052463981854190156,
      "learning_rate": 3.827413254079691e-06,
      "loss": 0.8374,
      "step": 267700
    },
    {
      "epoch": 4.7000474025176,
      "grad_norm": 0.055277261384303604,
      "learning_rate": 3.825282645491835e-06,
      "loss": 0.8368,
      "step": 267710
    },
    {
      "epoch": 4.700222967397602,
      "grad_norm": 0.06115689324969741,
      "learning_rate": 3.823153268126277e-06,
      "loss": 0.8439,
      "step": 267720
    },
    {
      "epoch": 4.700398532277603,
      "grad_norm": 0.07261499974606021,
      "learning_rate": 3.821025122009979e-06,
      "loss": 0.8315,
      "step": 267730
    },
    {
      "epoch": 4.700574097157604,
      "grad_norm": 0.05153263786849403,
      "learning_rate": 3.818898207169911e-06,
      "loss": 0.8346,
      "step": 267740
    },
    {
      "epoch": 4.700749662037606,
      "grad_norm": 0.058820816778932764,
      "learning_rate": 3.8167725236330385e-06,
      "loss": 0.8443,
      "step": 267750
    },
    {
      "epoch": 4.700925226917607,
      "grad_norm": 0.058356241007583104,
      "learning_rate": 3.814648071426288e-06,
      "loss": 0.8304,
      "step": 267760
    },
    {
      "epoch": 4.701100791797609,
      "grad_norm": 0.0522849698428773,
      "learning_rate": 3.8125248505765542e-06,
      "loss": 0.8451,
      "step": 267770
    },
    {
      "epoch": 4.70127635667761,
      "grad_norm": 0.04415246472113507,
      "learning_rate": 3.8104028611107666e-06,
      "loss": 0.8363,
      "step": 267780
    },
    {
      "epoch": 4.701451921557611,
      "grad_norm": 0.046035285705998984,
      "learning_rate": 3.808282103055808e-06,
      "loss": 0.8374,
      "step": 267790
    },
    {
      "epoch": 4.701627486437613,
      "grad_norm": 0.050791778836912556,
      "learning_rate": 3.806162576438532e-06,
      "loss": 0.842,
      "step": 267800
    },
    {
      "epoch": 4.701803051317614,
      "grad_norm": 0.04498143166447269,
      "learning_rate": 3.804044281285801e-06,
      "loss": 0.8353,
      "step": 267810
    },
    {
      "epoch": 4.701978616197616,
      "grad_norm": 0.041056842886299634,
      "learning_rate": 3.8019272176244652e-06,
      "loss": 0.8353,
      "step": 267820
    },
    {
      "epoch": 4.702154181077617,
      "grad_norm": 0.05375852760739268,
      "learning_rate": 3.7998113854813334e-06,
      "loss": 0.8384,
      "step": 267830
    },
    {
      "epoch": 4.702329745957619,
      "grad_norm": 0.049126539923791235,
      "learning_rate": 3.7976967848832227e-06,
      "loss": 0.839,
      "step": 267840
    },
    {
      "epoch": 4.70250531083762,
      "grad_norm": 0.06392072175500237,
      "learning_rate": 3.7955834158569193e-06,
      "loss": 0.8395,
      "step": 267850
    },
    {
      "epoch": 4.702680875717622,
      "grad_norm": 0.050641115059739866,
      "learning_rate": 3.79347127842921e-06,
      "loss": 0.8488,
      "step": 267860
    },
    {
      "epoch": 4.702856440597623,
      "grad_norm": 0.05321547667054561,
      "learning_rate": 3.7913603726268358e-06,
      "loss": 0.8399,
      "step": 267870
    },
    {
      "epoch": 4.703032005477624,
      "grad_norm": 0.054352298472155454,
      "learning_rate": 3.7892506984765695e-06,
      "loss": 0.8403,
      "step": 267880
    },
    {
      "epoch": 4.703207570357626,
      "grad_norm": 0.049804500206362935,
      "learning_rate": 3.7871422560051327e-06,
      "loss": 0.8413,
      "step": 267890
    },
    {
      "epoch": 4.703383135237627,
      "grad_norm": 0.05921528545916879,
      "learning_rate": 3.7850350452392226e-06,
      "loss": 0.8379,
      "step": 267900
    },
    {
      "epoch": 4.703558700117629,
      "grad_norm": 0.04869878141737356,
      "learning_rate": 3.782929066205548e-06,
      "loss": 0.8451,
      "step": 267910
    },
    {
      "epoch": 4.70373426499763,
      "grad_norm": 0.06433041933637439,
      "learning_rate": 3.780824318930795e-06,
      "loss": 0.8335,
      "step": 267920
    },
    {
      "epoch": 4.703909829877631,
      "grad_norm": 0.045888515720987684,
      "learning_rate": 3.7787208034416393e-06,
      "loss": 0.838,
      "step": 267930
    },
    {
      "epoch": 4.704085394757633,
      "grad_norm": 0.05230375865738276,
      "learning_rate": 3.776618519764713e-06,
      "loss": 0.8414,
      "step": 267940
    },
    {
      "epoch": 4.704260959637634,
      "grad_norm": 0.041479614712258354,
      "learning_rate": 3.7745174679266582e-06,
      "loss": 0.8369,
      "step": 267950
    },
    {
      "epoch": 4.704436524517636,
      "grad_norm": 0.056199213238029996,
      "learning_rate": 3.7724176479541075e-06,
      "loss": 0.8324,
      "step": 267960
    },
    {
      "epoch": 4.704612089397637,
      "grad_norm": 0.04402875058024408,
      "learning_rate": 3.770319059873659e-06,
      "loss": 0.8422,
      "step": 267970
    },
    {
      "epoch": 4.704787654277638,
      "grad_norm": 0.04948152076759525,
      "learning_rate": 3.768221703711901e-06,
      "loss": 0.8419,
      "step": 267980
    },
    {
      "epoch": 4.70496321915764,
      "grad_norm": 0.050749469838027264,
      "learning_rate": 3.7661255794954096e-06,
      "loss": 0.8358,
      "step": 267990
    },
    {
      "epoch": 4.705138784037641,
      "grad_norm": 0.04818992754758732,
      "learning_rate": 3.7640306872507496e-06,
      "loss": 0.8378,
      "step": 268000
    },
    {
      "epoch": 4.705314348917643,
      "grad_norm": 0.04858907686783499,
      "learning_rate": 3.7619370270044453e-06,
      "loss": 0.838,
      "step": 268010
    },
    {
      "epoch": 4.705489913797644,
      "grad_norm": 0.05026133811011277,
      "learning_rate": 3.759844598783027e-06,
      "loss": 0.8388,
      "step": 268020
    },
    {
      "epoch": 4.705665478677645,
      "grad_norm": 0.04944871286473442,
      "learning_rate": 3.7577534026130295e-06,
      "loss": 0.8403,
      "step": 268030
    },
    {
      "epoch": 4.705841043557647,
      "grad_norm": 0.05132842869688889,
      "learning_rate": 3.755663438520919e-06,
      "loss": 0.84,
      "step": 268040
    },
    {
      "epoch": 4.706016608437648,
      "grad_norm": 0.05836427851759966,
      "learning_rate": 3.7535747065332054e-06,
      "loss": 0.8382,
      "step": 268050
    },
    {
      "epoch": 4.70619217331765,
      "grad_norm": 0.05641268315225926,
      "learning_rate": 3.7514872066763337e-06,
      "loss": 0.8402,
      "step": 268060
    },
    {
      "epoch": 4.706367738197651,
      "grad_norm": 0.04504060241912464,
      "learning_rate": 3.74940093897675e-06,
      "loss": 0.8403,
      "step": 268070
    },
    {
      "epoch": 4.706543303077653,
      "grad_norm": 0.04362603642609433,
      "learning_rate": 3.7473159034608976e-06,
      "loss": 0.8339,
      "step": 268080
    },
    {
      "epoch": 4.706718867957654,
      "grad_norm": 0.046312421192856806,
      "learning_rate": 3.745232100155179e-06,
      "loss": 0.8507,
      "step": 268090
    },
    {
      "epoch": 4.706894432837656,
      "grad_norm": 0.05050376132222978,
      "learning_rate": 3.743149529086027e-06,
      "loss": 0.8392,
      "step": 268100
    },
    {
      "epoch": 4.707069997717657,
      "grad_norm": 0.040868088687944115,
      "learning_rate": 3.741068190279809e-06,
      "loss": 0.8335,
      "step": 268110
    },
    {
      "epoch": 4.707245562597658,
      "grad_norm": 0.048492916702466186,
      "learning_rate": 3.738988083762883e-06,
      "loss": 0.8429,
      "step": 268120
    },
    {
      "epoch": 4.70742112747766,
      "grad_norm": 0.04243669346194222,
      "learning_rate": 3.736909209561628e-06,
      "loss": 0.834,
      "step": 268130
    },
    {
      "epoch": 4.707596692357661,
      "grad_norm": 0.0447825782779746,
      "learning_rate": 3.734831567702367e-06,
      "loss": 0.8365,
      "step": 268140
    },
    {
      "epoch": 4.707772257237663,
      "grad_norm": 0.06077382723983736,
      "learning_rate": 3.7327551582114358e-06,
      "loss": 0.8468,
      "step": 268150
    },
    {
      "epoch": 4.707947822117664,
      "grad_norm": 0.05574266363101695,
      "learning_rate": 3.7306799811151357e-06,
      "loss": 0.832,
      "step": 268160
    },
    {
      "epoch": 4.708123386997665,
      "grad_norm": 0.05018387083351964,
      "learning_rate": 3.728606036439768e-06,
      "loss": 0.8408,
      "step": 268170
    },
    {
      "epoch": 4.708298951877667,
      "grad_norm": 0.047827086955033175,
      "learning_rate": 3.726533324211592e-06,
      "loss": 0.8391,
      "step": 268180
    },
    {
      "epoch": 4.708474516757668,
      "grad_norm": 0.07035560723130747,
      "learning_rate": 3.7244618444568885e-06,
      "loss": 0.8409,
      "step": 268190
    },
    {
      "epoch": 4.7086500816376695,
      "grad_norm": 0.052023283170296905,
      "learning_rate": 3.7223915972019018e-06,
      "loss": 0.8345,
      "step": 268200
    },
    {
      "epoch": 4.7088256465176705,
      "grad_norm": 0.09031249538352809,
      "learning_rate": 3.7203225824728377e-06,
      "loss": 0.8368,
      "step": 268210
    },
    {
      "epoch": 4.709001211397672,
      "grad_norm": 0.0486240387696323,
      "learning_rate": 3.718254800295952e-06,
      "loss": 0.8381,
      "step": 268220
    },
    {
      "epoch": 4.7091767762776735,
      "grad_norm": 0.04055463146321824,
      "learning_rate": 3.7161882506974165e-06,
      "loss": 0.8437,
      "step": 268230
    },
    {
      "epoch": 4.7093523411576745,
      "grad_norm": 0.05013024350979586,
      "learning_rate": 3.7141229337034107e-06,
      "loss": 0.8454,
      "step": 268240
    },
    {
      "epoch": 4.7095279060376765,
      "grad_norm": 0.06472358484529285,
      "learning_rate": 3.7120588493401165e-06,
      "loss": 0.8402,
      "step": 268250
    },
    {
      "epoch": 4.7097034709176775,
      "grad_norm": 0.055675903210994,
      "learning_rate": 3.709995997633682e-06,
      "loss": 0.8346,
      "step": 268260
    },
    {
      "epoch": 4.709879035797679,
      "grad_norm": 0.05816234749183398,
      "learning_rate": 3.7079343786102547e-06,
      "loss": 0.841,
      "step": 268270
    },
    {
      "epoch": 4.7100546006776804,
      "grad_norm": 0.0552588742969346,
      "learning_rate": 3.705873992295929e-06,
      "loss": 0.8398,
      "step": 268280
    },
    {
      "epoch": 4.7102301655576815,
      "grad_norm": 0.050290155211311506,
      "learning_rate": 3.70381483871683e-06,
      "loss": 0.8299,
      "step": 268290
    },
    {
      "epoch": 4.710405730437683,
      "grad_norm": 0.0443909423870249,
      "learning_rate": 3.7017569178990507e-06,
      "loss": 0.8352,
      "step": 268300
    },
    {
      "epoch": 4.710581295317684,
      "grad_norm": 0.05046305017856473,
      "learning_rate": 3.6997002298686524e-06,
      "loss": 0.8397,
      "step": 268310
    },
    {
      "epoch": 4.710756860197686,
      "grad_norm": 0.051336890736669305,
      "learning_rate": 3.697644774651704e-06,
      "loss": 0.8401,
      "step": 268320
    },
    {
      "epoch": 4.710932425077687,
      "grad_norm": 0.05422910338105319,
      "learning_rate": 3.695590552274235e-06,
      "loss": 0.8356,
      "step": 268330
    },
    {
      "epoch": 4.711107989957689,
      "grad_norm": 0.05579917912327898,
      "learning_rate": 3.693537562762293e-06,
      "loss": 0.8364,
      "step": 268340
    },
    {
      "epoch": 4.71128355483769,
      "grad_norm": 0.06522176636176132,
      "learning_rate": 3.6914858061418743e-06,
      "loss": 0.8436,
      "step": 268350
    },
    {
      "epoch": 4.711459119717691,
      "grad_norm": 0.06130806855629285,
      "learning_rate": 3.6894352824389702e-06,
      "loss": 0.8365,
      "step": 268360
    },
    {
      "epoch": 4.711634684597693,
      "grad_norm": 0.04565619025526522,
      "learning_rate": 3.6873859916795887e-06,
      "loss": 0.8462,
      "step": 268370
    },
    {
      "epoch": 4.711810249477694,
      "grad_norm": 0.059040990567443004,
      "learning_rate": 3.6853379338896556e-06,
      "loss": 0.8446,
      "step": 268380
    },
    {
      "epoch": 4.711985814357696,
      "grad_norm": 0.05403848221015146,
      "learning_rate": 3.6832911090951563e-06,
      "loss": 0.8363,
      "step": 268390
    },
    {
      "epoch": 4.712161379237697,
      "grad_norm": 0.05689480997752656,
      "learning_rate": 3.681245517322017e-06,
      "loss": 0.8302,
      "step": 268400
    },
    {
      "epoch": 4.712336944117698,
      "grad_norm": 0.0403867055852201,
      "learning_rate": 3.679201158596134e-06,
      "loss": 0.8342,
      "step": 268410
    },
    {
      "epoch": 4.7125125089977,
      "grad_norm": 0.04902252652945765,
      "learning_rate": 3.6771580329434257e-06,
      "loss": 0.8438,
      "step": 268420
    },
    {
      "epoch": 4.712688073877701,
      "grad_norm": 0.05897045813171033,
      "learning_rate": 3.675116140389774e-06,
      "loss": 0.8338,
      "step": 268430
    },
    {
      "epoch": 4.712863638757703,
      "grad_norm": 0.06588643412872122,
      "learning_rate": 3.6730754809610648e-06,
      "loss": 0.8318,
      "step": 268440
    },
    {
      "epoch": 4.713039203637704,
      "grad_norm": 0.05542095460322553,
      "learning_rate": 3.671036054683127e-06,
      "loss": 0.838,
      "step": 268450
    },
    {
      "epoch": 4.713214768517705,
      "grad_norm": 0.04997675066370838,
      "learning_rate": 3.6689978615818245e-06,
      "loss": 0.8299,
      "step": 268460
    },
    {
      "epoch": 4.713390333397707,
      "grad_norm": 0.11109924894923964,
      "learning_rate": 3.6669609016829726e-06,
      "loss": 0.8388,
      "step": 268470
    },
    {
      "epoch": 4.713565898277708,
      "grad_norm": 0.0386646131723094,
      "learning_rate": 3.6649251750123694e-06,
      "loss": 0.8471,
      "step": 268480
    },
    {
      "epoch": 4.71374146315771,
      "grad_norm": 0.052773827256902765,
      "learning_rate": 3.662890681595833e-06,
      "loss": 0.8356,
      "step": 268490
    },
    {
      "epoch": 4.713917028037711,
      "grad_norm": 0.05065577891515663,
      "learning_rate": 3.6608574214591164e-06,
      "loss": 0.844,
      "step": 268500
    },
    {
      "epoch": 4.714092592917713,
      "grad_norm": 0.05207206632162683,
      "learning_rate": 3.6588253946279932e-06,
      "loss": 0.8367,
      "step": 268510
    },
    {
      "epoch": 4.714268157797714,
      "grad_norm": 0.045798426723981676,
      "learning_rate": 3.6567946011282163e-06,
      "loss": 0.8418,
      "step": 268520
    },
    {
      "epoch": 4.714443722677715,
      "grad_norm": 0.06520524129564617,
      "learning_rate": 3.654765040985493e-06,
      "loss": 0.836,
      "step": 268530
    },
    {
      "epoch": 4.714619287557717,
      "grad_norm": 0.08390155965076258,
      "learning_rate": 3.6527367142255667e-06,
      "loss": 0.8404,
      "step": 268540
    },
    {
      "epoch": 4.714794852437718,
      "grad_norm": 0.06386740217687596,
      "learning_rate": 3.6507096208741e-06,
      "loss": 0.8328,
      "step": 268550
    },
    {
      "epoch": 4.71497041731772,
      "grad_norm": 0.04562902804427964,
      "learning_rate": 3.648683760956824e-06,
      "loss": 0.8415,
      "step": 268560
    },
    {
      "epoch": 4.715145982197721,
      "grad_norm": 0.05449404725548073,
      "learning_rate": 3.6466591344993818e-06,
      "loss": 0.8431,
      "step": 268570
    },
    {
      "epoch": 4.715321547077723,
      "grad_norm": 0.0546248765180241,
      "learning_rate": 3.644635741527415e-06,
      "loss": 0.8344,
      "step": 268580
    },
    {
      "epoch": 4.715497111957724,
      "grad_norm": 0.04856117467364646,
      "learning_rate": 3.6426135820665666e-06,
      "loss": 0.8405,
      "step": 268590
    },
    {
      "epoch": 4.715672676837725,
      "grad_norm": 0.04323841681414147,
      "learning_rate": 3.640592656142468e-06,
      "loss": 0.8492,
      "step": 268600
    },
    {
      "epoch": 4.715848241717727,
      "grad_norm": 0.05092809809426667,
      "learning_rate": 3.6385729637807285e-06,
      "loss": 0.8377,
      "step": 268610
    },
    {
      "epoch": 4.716023806597728,
      "grad_norm": 0.053593703250541164,
      "learning_rate": 3.6365545050069254e-06,
      "loss": 0.837,
      "step": 268620
    },
    {
      "epoch": 4.71619937147773,
      "grad_norm": 0.08103235782942524,
      "learning_rate": 3.6345372798466225e-06,
      "loss": 0.844,
      "step": 268630
    },
    {
      "epoch": 4.716374936357731,
      "grad_norm": 0.056122499315986375,
      "learning_rate": 3.6325212883254094e-06,
      "loss": 0.8345,
      "step": 268640
    },
    {
      "epoch": 4.716550501237732,
      "grad_norm": 0.04874897235925137,
      "learning_rate": 3.6305065304688065e-06,
      "loss": 0.8415,
      "step": 268650
    },
    {
      "epoch": 4.716726066117734,
      "grad_norm": 0.04432560248580024,
      "learning_rate": 3.628493006302347e-06,
      "loss": 0.8417,
      "step": 268660
    },
    {
      "epoch": 4.716901630997735,
      "grad_norm": 0.043925526951845095,
      "learning_rate": 3.62648071585153e-06,
      "loss": 0.8446,
      "step": 268670
    },
    {
      "epoch": 4.717077195877737,
      "grad_norm": 0.04960844672784375,
      "learning_rate": 3.624469659141888e-06,
      "loss": 0.8431,
      "step": 268680
    },
    {
      "epoch": 4.717252760757738,
      "grad_norm": 0.05412862470503659,
      "learning_rate": 3.622459836198855e-06,
      "loss": 0.8379,
      "step": 268690
    },
    {
      "epoch": 4.717428325637739,
      "grad_norm": 0.06133956123760838,
      "learning_rate": 3.6204512470479304e-06,
      "loss": 0.8298,
      "step": 268700
    },
    {
      "epoch": 4.717603890517741,
      "grad_norm": 0.04721072868421729,
      "learning_rate": 3.6184438917145473e-06,
      "loss": 0.8401,
      "step": 268710
    },
    {
      "epoch": 4.717779455397742,
      "grad_norm": 0.059714094076333595,
      "learning_rate": 3.616437770224141e-06,
      "loss": 0.8376,
      "step": 268720
    },
    {
      "epoch": 4.717955020277744,
      "grad_norm": 0.05361939334742361,
      "learning_rate": 3.6144328826021326e-06,
      "loss": 0.8395,
      "step": 268730
    },
    {
      "epoch": 4.718130585157745,
      "grad_norm": 0.052166751594149914,
      "learning_rate": 3.6124292288739344e-06,
      "loss": 0.8389,
      "step": 268740
    },
    {
      "epoch": 4.718306150037747,
      "grad_norm": 0.04545535772040222,
      "learning_rate": 3.6104268090649034e-06,
      "loss": 0.838,
      "step": 268750
    },
    {
      "epoch": 4.718481714917748,
      "grad_norm": 0.05317437653373611,
      "learning_rate": 3.6084256232004397e-06,
      "loss": 0.836,
      "step": 268760
    },
    {
      "epoch": 4.71865727979775,
      "grad_norm": 0.05298638717813309,
      "learning_rate": 3.6064256713058894e-06,
      "loss": 0.8388,
      "step": 268770
    },
    {
      "epoch": 4.718832844677751,
      "grad_norm": 0.05005030641820667,
      "learning_rate": 3.604426953406587e-06,
      "loss": 0.8418,
      "step": 268780
    },
    {
      "epoch": 4.719008409557752,
      "grad_norm": 0.0559413773249924,
      "learning_rate": 3.6024294695278668e-06,
      "loss": 0.8369,
      "step": 268790
    },
    {
      "epoch": 4.719183974437754,
      "grad_norm": 0.05333492398650087,
      "learning_rate": 3.600433219695021e-06,
      "loss": 0.8427,
      "step": 268800
    },
    {
      "epoch": 4.719359539317755,
      "grad_norm": 0.04687568835939078,
      "learning_rate": 3.5984382039333714e-06,
      "loss": 0.8376,
      "step": 268810
    },
    {
      "epoch": 4.719535104197757,
      "grad_norm": 0.04449628826822418,
      "learning_rate": 3.5964444222681554e-06,
      "loss": 0.8343,
      "step": 268820
    },
    {
      "epoch": 4.719710669077758,
      "grad_norm": 0.05099617604266089,
      "learning_rate": 3.5944518747246636e-06,
      "loss": 0.8381,
      "step": 268830
    },
    {
      "epoch": 4.719886233957759,
      "grad_norm": 0.05887111457347818,
      "learning_rate": 3.5924605613281413e-06,
      "loss": 0.8393,
      "step": 268840
    },
    {
      "epoch": 4.720061798837761,
      "grad_norm": 0.04685161243813024,
      "learning_rate": 3.590470482103804e-06,
      "loss": 0.8365,
      "step": 268850
    },
    {
      "epoch": 4.720237363717762,
      "grad_norm": 0.04557980856678719,
      "learning_rate": 3.5884816370768765e-06,
      "loss": 0.8371,
      "step": 268860
    },
    {
      "epoch": 4.720412928597764,
      "grad_norm": 0.04302987046707537,
      "learning_rate": 3.5864940262725502e-06,
      "loss": 0.8386,
      "step": 268870
    },
    {
      "epoch": 4.720588493477765,
      "grad_norm": 0.05388780879320173,
      "learning_rate": 3.5845076497160277e-06,
      "loss": 0.8406,
      "step": 268880
    },
    {
      "epoch": 4.720764058357766,
      "grad_norm": 0.04636115846780166,
      "learning_rate": 3.5825225074324455e-06,
      "loss": 0.8424,
      "step": 268890
    },
    {
      "epoch": 4.720939623237768,
      "grad_norm": 0.05473438435222995,
      "learning_rate": 3.580538599446985e-06,
      "loss": 0.836,
      "step": 268900
    },
    {
      "epoch": 4.721115188117769,
      "grad_norm": 0.08695770468107591,
      "learning_rate": 3.5785559257847727e-06,
      "loss": 0.8318,
      "step": 268910
    },
    {
      "epoch": 4.721290752997771,
      "grad_norm": 0.06932711466030748,
      "learning_rate": 3.5765744864709225e-06,
      "loss": 0.8444,
      "step": 268920
    },
    {
      "epoch": 4.721466317877772,
      "grad_norm": 0.050924218267517606,
      "learning_rate": 3.574594281530538e-06,
      "loss": 0.8369,
      "step": 268930
    },
    {
      "epoch": 4.7216418827577735,
      "grad_norm": 0.1083503914414677,
      "learning_rate": 3.572615310988725e-06,
      "loss": 0.838,
      "step": 268940
    },
    {
      "epoch": 4.721817447637775,
      "grad_norm": 0.05088264901697227,
      "learning_rate": 3.5706375748705537e-06,
      "loss": 0.8388,
      "step": 268950
    },
    {
      "epoch": 4.721993012517776,
      "grad_norm": 0.05911842546522891,
      "learning_rate": 3.5686610732010606e-06,
      "loss": 0.8363,
      "step": 268960
    },
    {
      "epoch": 4.7221685773977775,
      "grad_norm": 0.04365734526738008,
      "learning_rate": 3.566685806005307e-06,
      "loss": 0.8373,
      "step": 268970
    },
    {
      "epoch": 4.7223441422777785,
      "grad_norm": 0.05024924303431537,
      "learning_rate": 3.564711773308331e-06,
      "loss": 0.8359,
      "step": 268980
    },
    {
      "epoch": 4.7225197071577805,
      "grad_norm": 0.04309548834692126,
      "learning_rate": 3.5627389751351147e-06,
      "loss": 0.8441,
      "step": 268990
    },
    {
      "epoch": 4.7226952720377815,
      "grad_norm": 0.05026617082062767,
      "learning_rate": 3.5607674115106752e-06,
      "loss": 0.8395,
      "step": 269000
    },
    {
      "epoch": 4.722870836917783,
      "grad_norm": 0.06282115733973297,
      "learning_rate": 3.5587970824599833e-06,
      "loss": 0.8354,
      "step": 269010
    },
    {
      "epoch": 4.7230464017977845,
      "grad_norm": 0.04561406244055906,
      "learning_rate": 3.5568279880080225e-06,
      "loss": 0.8343,
      "step": 269020
    },
    {
      "epoch": 4.7232219666777855,
      "grad_norm": 0.06219275744358239,
      "learning_rate": 3.5548601281797114e-06,
      "loss": 0.8415,
      "step": 269030
    },
    {
      "epoch": 4.723397531557787,
      "grad_norm": 0.046964502985725735,
      "learning_rate": 3.5528935029999976e-06,
      "loss": 0.8345,
      "step": 269040
    },
    {
      "epoch": 4.723573096437788,
      "grad_norm": 0.06292337348664616,
      "learning_rate": 3.55092811249381e-06,
      "loss": 0.8474,
      "step": 269050
    },
    {
      "epoch": 4.72374866131779,
      "grad_norm": 0.05810839397972513,
      "learning_rate": 3.5489639566860225e-06,
      "loss": 0.8388,
      "step": 269060
    },
    {
      "epoch": 4.723924226197791,
      "grad_norm": 0.042595063789370474,
      "learning_rate": 3.5470010356015515e-06,
      "loss": 0.8351,
      "step": 269070
    },
    {
      "epoch": 4.724099791077792,
      "grad_norm": 0.04447680112629358,
      "learning_rate": 3.545039349265248e-06,
      "loss": 0.8425,
      "step": 269080
    },
    {
      "epoch": 4.724275355957794,
      "grad_norm": 0.047674377811637685,
      "learning_rate": 3.5430788977019743e-06,
      "loss": 0.8357,
      "step": 269090
    },
    {
      "epoch": 4.724450920837795,
      "grad_norm": 0.05404903839945475,
      "learning_rate": 3.5411196809365708e-06,
      "loss": 0.846,
      "step": 269100
    },
    {
      "epoch": 4.724626485717797,
      "grad_norm": 0.06950413256772534,
      "learning_rate": 3.539161698993855e-06,
      "loss": 0.8354,
      "step": 269110
    },
    {
      "epoch": 4.724802050597798,
      "grad_norm": 0.04889943578408305,
      "learning_rate": 3.5372049518986457e-06,
      "loss": 0.8279,
      "step": 269120
    },
    {
      "epoch": 4.724977615477799,
      "grad_norm": 0.0480542436559821,
      "learning_rate": 3.535249439675717e-06,
      "loss": 0.842,
      "step": 269130
    },
    {
      "epoch": 4.725153180357801,
      "grad_norm": 0.040292265126041206,
      "learning_rate": 3.5332951623498646e-06,
      "loss": 0.8496,
      "step": 269140
    },
    {
      "epoch": 4.725328745237802,
      "grad_norm": 0.04880649381763071,
      "learning_rate": 3.5313421199458406e-06,
      "loss": 0.8433,
      "step": 269150
    },
    {
      "epoch": 4.725504310117804,
      "grad_norm": 0.05156150889033575,
      "learning_rate": 3.529390312488387e-06,
      "loss": 0.8395,
      "step": 269160
    },
    {
      "epoch": 4.725679874997805,
      "grad_norm": 0.047700728537964254,
      "learning_rate": 3.5274397400022455e-06,
      "loss": 0.8435,
      "step": 269170
    },
    {
      "epoch": 4.725855439877807,
      "grad_norm": 0.06715096942247298,
      "learning_rate": 3.525490402512111e-06,
      "loss": 0.8513,
      "step": 269180
    },
    {
      "epoch": 4.726031004757808,
      "grad_norm": 0.04104618342561526,
      "learning_rate": 3.5235423000427162e-06,
      "loss": 0.8362,
      "step": 269190
    },
    {
      "epoch": 4.726206569637809,
      "grad_norm": 0.054945337014982935,
      "learning_rate": 3.521595432618702e-06,
      "loss": 0.8407,
      "step": 269200
    },
    {
      "epoch": 4.726382134517811,
      "grad_norm": 0.04773776804990481,
      "learning_rate": 3.519649800264766e-06,
      "loss": 0.8348,
      "step": 269210
    },
    {
      "epoch": 4.726557699397812,
      "grad_norm": 0.04407269312451477,
      "learning_rate": 3.5177054030055506e-06,
      "loss": 0.8325,
      "step": 269220
    },
    {
      "epoch": 4.726733264277814,
      "grad_norm": 0.05597698368461701,
      "learning_rate": 3.5157622408656652e-06,
      "loss": 0.8393,
      "step": 269230
    },
    {
      "epoch": 4.726908829157815,
      "grad_norm": 0.05744742235048056,
      "learning_rate": 3.513820313869784e-06,
      "loss": 0.8371,
      "step": 269240
    },
    {
      "epoch": 4.727084394037817,
      "grad_norm": 0.04913781021358177,
      "learning_rate": 3.5118796220424742e-06,
      "loss": 0.8357,
      "step": 269250
    },
    {
      "epoch": 4.727259958917818,
      "grad_norm": 0.05038304050284769,
      "learning_rate": 3.5099401654083325e-06,
      "loss": 0.8428,
      "step": 269260
    },
    {
      "epoch": 4.727435523797819,
      "grad_norm": 0.06456451208723778,
      "learning_rate": 3.5080019439919362e-06,
      "loss": 0.8364,
      "step": 269270
    },
    {
      "epoch": 4.727611088677821,
      "grad_norm": 0.07261181225181651,
      "learning_rate": 3.5060649578178285e-06,
      "loss": 0.8376,
      "step": 269280
    },
    {
      "epoch": 4.727786653557822,
      "grad_norm": 0.04507683695476735,
      "learning_rate": 3.5041292069105747e-06,
      "loss": 0.8448,
      "step": 269290
    },
    {
      "epoch": 4.727962218437824,
      "grad_norm": 0.06541665504492263,
      "learning_rate": 3.5021946912946845e-06,
      "loss": 0.8266,
      "step": 269300
    },
    {
      "epoch": 4.728137783317825,
      "grad_norm": 0.058600724569097304,
      "learning_rate": 3.5002614109946693e-06,
      "loss": 0.8407,
      "step": 269310
    },
    {
      "epoch": 4.728313348197826,
      "grad_norm": 0.05292174167642983,
      "learning_rate": 3.4983293660350393e-06,
      "loss": 0.8403,
      "step": 269320
    },
    {
      "epoch": 4.728488913077828,
      "grad_norm": 0.06268306109018133,
      "learning_rate": 3.4963985564402487e-06,
      "loss": 0.8393,
      "step": 269330
    },
    {
      "epoch": 4.728664477957829,
      "grad_norm": 0.06134896965182258,
      "learning_rate": 3.4944689822347777e-06,
      "loss": 0.8356,
      "step": 269340
    },
    {
      "epoch": 4.728840042837831,
      "grad_norm": 0.054885534902796684,
      "learning_rate": 3.4925406434430674e-06,
      "loss": 0.8334,
      "step": 269350
    },
    {
      "epoch": 4.729015607717832,
      "grad_norm": 0.048584653554051574,
      "learning_rate": 3.490613540089564e-06,
      "loss": 0.8393,
      "step": 269360
    },
    {
      "epoch": 4.729191172597833,
      "grad_norm": 0.057768663868710254,
      "learning_rate": 3.4886876721986685e-06,
      "loss": 0.8431,
      "step": 269370
    },
    {
      "epoch": 4.729366737477835,
      "grad_norm": 0.046642804022550874,
      "learning_rate": 3.4867630397947914e-06,
      "loss": 0.8345,
      "step": 269380
    },
    {
      "epoch": 4.729542302357836,
      "grad_norm": 0.04052729191247539,
      "learning_rate": 3.4848396429023227e-06,
      "loss": 0.8465,
      "step": 269390
    },
    {
      "epoch": 4.729717867237838,
      "grad_norm": 0.0619121356996399,
      "learning_rate": 3.4829174815456077e-06,
      "loss": 0.8394,
      "step": 269400
    },
    {
      "epoch": 4.729893432117839,
      "grad_norm": 0.05610915667103477,
      "learning_rate": 3.480996555749025e-06,
      "loss": 0.8308,
      "step": 269410
    },
    {
      "epoch": 4.730068996997841,
      "grad_norm": 0.04593302920879009,
      "learning_rate": 3.479076865536908e-06,
      "loss": 0.8317,
      "step": 269420
    },
    {
      "epoch": 4.730244561877842,
      "grad_norm": 0.05633501317087239,
      "learning_rate": 3.477158410933582e-06,
      "loss": 0.8427,
      "step": 269430
    },
    {
      "epoch": 4.730420126757844,
      "grad_norm": 0.04704188402219465,
      "learning_rate": 3.4752411919633367e-06,
      "loss": 0.8376,
      "step": 269440
    },
    {
      "epoch": 4.730595691637845,
      "grad_norm": 0.058105486766959424,
      "learning_rate": 3.473325208650474e-06,
      "loss": 0.8365,
      "step": 269450
    },
    {
      "epoch": 4.730771256517846,
      "grad_norm": 0.05554448229249022,
      "learning_rate": 3.4714104610192853e-06,
      "loss": 0.8374,
      "step": 269460
    },
    {
      "epoch": 4.730946821397848,
      "grad_norm": 0.04350490422520714,
      "learning_rate": 3.4694969490940146e-06,
      "loss": 0.8451,
      "step": 269470
    },
    {
      "epoch": 4.731122386277849,
      "grad_norm": 0.09033786576875727,
      "learning_rate": 3.4675846728989e-06,
      "loss": 0.8303,
      "step": 269480
    },
    {
      "epoch": 4.731297951157851,
      "grad_norm": 0.06429272534682902,
      "learning_rate": 3.465673632458187e-06,
      "loss": 0.837,
      "step": 269490
    },
    {
      "epoch": 4.731473516037852,
      "grad_norm": 0.043894884781348266,
      "learning_rate": 3.463763827796078e-06,
      "loss": 0.8372,
      "step": 269500
    },
    {
      "epoch": 4.731649080917853,
      "grad_norm": 0.07330127344480698,
      "learning_rate": 3.461855258936776e-06,
      "loss": 0.8436,
      "step": 269510
    },
    {
      "epoch": 4.731824645797855,
      "grad_norm": 0.05948061335630108,
      "learning_rate": 3.4599479259044624e-06,
      "loss": 0.8411,
      "step": 269520
    },
    {
      "epoch": 4.732000210677856,
      "grad_norm": 0.04035450662529167,
      "learning_rate": 3.458041828723306e-06,
      "loss": 0.8416,
      "step": 269530
    },
    {
      "epoch": 4.732175775557858,
      "grad_norm": 0.05287651085333652,
      "learning_rate": 3.4561369674174434e-06,
      "loss": 0.8337,
      "step": 269540
    },
    {
      "epoch": 4.732351340437859,
      "grad_norm": 0.04213063850500086,
      "learning_rate": 3.4542333420110236e-06,
      "loss": 0.8375,
      "step": 269550
    },
    {
      "epoch": 4.73252690531786,
      "grad_norm": 0.046697410556363186,
      "learning_rate": 3.4523309525281704e-06,
      "loss": 0.8366,
      "step": 269560
    },
    {
      "epoch": 4.732702470197862,
      "grad_norm": 0.04926070673824322,
      "learning_rate": 3.4504297989929555e-06,
      "loss": 0.846,
      "step": 269570
    },
    {
      "epoch": 4.732878035077863,
      "grad_norm": 0.05368254963441094,
      "learning_rate": 3.448529881429516e-06,
      "loss": 0.8359,
      "step": 269580
    },
    {
      "epoch": 4.733053599957865,
      "grad_norm": 0.048665063336206835,
      "learning_rate": 3.446631199861888e-06,
      "loss": 0.8429,
      "step": 269590
    },
    {
      "epoch": 4.733229164837866,
      "grad_norm": 0.0529307168854671,
      "learning_rate": 3.4447337543141443e-06,
      "loss": 0.8397,
      "step": 269600
    },
    {
      "epoch": 4.733404729717867,
      "grad_norm": 0.05058404717534522,
      "learning_rate": 3.442837544810311e-06,
      "loss": 0.838,
      "step": 269610
    },
    {
      "epoch": 4.733580294597869,
      "grad_norm": 0.07772687331494625,
      "learning_rate": 3.4409425713744253e-06,
      "loss": 0.8333,
      "step": 269620
    },
    {
      "epoch": 4.73375585947787,
      "grad_norm": 0.04234404413806608,
      "learning_rate": 3.4390488340304927e-06,
      "loss": 0.8356,
      "step": 269630
    },
    {
      "epoch": 4.733931424357872,
      "grad_norm": 0.06404034287910282,
      "learning_rate": 3.437156332802518e-06,
      "loss": 0.8341,
      "step": 269640
    },
    {
      "epoch": 4.734106989237873,
      "grad_norm": 0.05839491460251181,
      "learning_rate": 3.4352650677144613e-06,
      "loss": 0.8445,
      "step": 269650
    },
    {
      "epoch": 4.734282554117875,
      "grad_norm": 0.03892606185574389,
      "learning_rate": 3.4333750387903054e-06,
      "loss": 0.8408,
      "step": 269660
    },
    {
      "epoch": 4.734458118997876,
      "grad_norm": 0.04713257478855645,
      "learning_rate": 3.431486246053968e-06,
      "loss": 0.8456,
      "step": 269670
    },
    {
      "epoch": 4.7346336838778775,
      "grad_norm": 0.06585172825441309,
      "learning_rate": 3.42959868952941e-06,
      "loss": 0.8351,
      "step": 269680
    },
    {
      "epoch": 4.734809248757879,
      "grad_norm": 0.05936409023889388,
      "learning_rate": 3.4277123692405363e-06,
      "loss": 0.832,
      "step": 269690
    },
    {
      "epoch": 4.73498481363788,
      "grad_norm": 0.052184318264816294,
      "learning_rate": 3.4258272852112425e-06,
      "loss": 0.842,
      "step": 269700
    },
    {
      "epoch": 4.7351603785178815,
      "grad_norm": 0.054481570397606836,
      "learning_rate": 3.4239434374654233e-06,
      "loss": 0.8312,
      "step": 269710
    },
    {
      "epoch": 4.7353359433978826,
      "grad_norm": 0.047576752968868645,
      "learning_rate": 3.4220608260269295e-06,
      "loss": 0.8357,
      "step": 269720
    },
    {
      "epoch": 4.7355115082778845,
      "grad_norm": 0.048991907386912864,
      "learning_rate": 3.420179450919645e-06,
      "loss": 0.8317,
      "step": 269730
    },
    {
      "epoch": 4.7356870731578855,
      "grad_norm": 0.04564526755012522,
      "learning_rate": 3.418299312167366e-06,
      "loss": 0.836,
      "step": 269740
    },
    {
      "epoch": 4.7358626380378865,
      "grad_norm": 0.056696439591082615,
      "learning_rate": 3.4164204097939543e-06,
      "loss": 0.8333,
      "step": 269750
    },
    {
      "epoch": 4.7360382029178885,
      "grad_norm": 0.054347137495963614,
      "learning_rate": 3.4145427438231943e-06,
      "loss": 0.8455,
      "step": 269760
    },
    {
      "epoch": 4.7362137677978895,
      "grad_norm": 0.08086928723077988,
      "learning_rate": 3.412666314278873e-06,
      "loss": 0.847,
      "step": 269770
    },
    {
      "epoch": 4.736389332677891,
      "grad_norm": 0.06403669676698799,
      "learning_rate": 3.4107911211847725e-06,
      "loss": 0.8423,
      "step": 269780
    },
    {
      "epoch": 4.7365648975578925,
      "grad_norm": 0.05864112859358978,
      "learning_rate": 3.408917164564658e-06,
      "loss": 0.8362,
      "step": 269790
    },
    {
      "epoch": 4.7367404624378935,
      "grad_norm": 0.04743468806774644,
      "learning_rate": 3.4070444444422584e-06,
      "loss": 0.8359,
      "step": 269800
    },
    {
      "epoch": 4.736916027317895,
      "grad_norm": 0.04452380574709382,
      "learning_rate": 3.4051729608413152e-06,
      "loss": 0.84,
      "step": 269810
    },
    {
      "epoch": 4.737091592197896,
      "grad_norm": 0.04695557089998945,
      "learning_rate": 3.403302713785536e-06,
      "loss": 0.844,
      "step": 269820
    },
    {
      "epoch": 4.737267157077898,
      "grad_norm": 0.05285343501918645,
      "learning_rate": 3.401433703298607e-06,
      "loss": 0.84,
      "step": 269830
    },
    {
      "epoch": 4.737442721957899,
      "grad_norm": 0.06115927283171212,
      "learning_rate": 3.3995659294042253e-06,
      "loss": 0.8371,
      "step": 269840
    },
    {
      "epoch": 4.737618286837901,
      "grad_norm": 0.05387899016999972,
      "learning_rate": 3.397699392126045e-06,
      "loss": 0.8403,
      "step": 269850
    },
    {
      "epoch": 4.737793851717902,
      "grad_norm": 0.04682105079652921,
      "learning_rate": 3.395834091487717e-06,
      "loss": 0.84,
      "step": 269860
    },
    {
      "epoch": 4.737969416597903,
      "grad_norm": 0.05180520754531565,
      "learning_rate": 3.3939700275128855e-06,
      "loss": 0.8368,
      "step": 269870
    },
    {
      "epoch": 4.738144981477905,
      "grad_norm": 0.04817191731868438,
      "learning_rate": 3.3921072002251477e-06,
      "loss": 0.8351,
      "step": 269880
    },
    {
      "epoch": 4.738320546357906,
      "grad_norm": 0.04868056540424003,
      "learning_rate": 3.3902456096481224e-06,
      "loss": 0.8404,
      "step": 269890
    },
    {
      "epoch": 4.738496111237908,
      "grad_norm": 0.05329617074551716,
      "learning_rate": 3.3883852558054e-06,
      "loss": 0.8318,
      "step": 269900
    },
    {
      "epoch": 4.738671676117909,
      "grad_norm": 0.05024651740091856,
      "learning_rate": 3.3865261387205314e-06,
      "loss": 0.8403,
      "step": 269910
    },
    {
      "epoch": 4.738847240997911,
      "grad_norm": 0.04628069948887886,
      "learning_rate": 3.3846682584170945e-06,
      "loss": 0.8419,
      "step": 269920
    },
    {
      "epoch": 4.739022805877912,
      "grad_norm": 0.04754536325785593,
      "learning_rate": 3.3828116149186212e-06,
      "loss": 0.8348,
      "step": 269930
    },
    {
      "epoch": 4.739198370757913,
      "grad_norm": 0.0504123447451637,
      "learning_rate": 3.3809562082486214e-06,
      "loss": 0.8305,
      "step": 269940
    },
    {
      "epoch": 4.739373935637915,
      "grad_norm": 0.04336624477960971,
      "learning_rate": 3.3791020384306163e-06,
      "loss": 0.8468,
      "step": 269950
    },
    {
      "epoch": 4.739549500517916,
      "grad_norm": 0.05272605369849274,
      "learning_rate": 3.3772491054881046e-06,
      "loss": 0.8318,
      "step": 269960
    },
    {
      "epoch": 4.739725065397918,
      "grad_norm": 0.04818259982933336,
      "learning_rate": 3.3753974094445544e-06,
      "loss": 0.8343,
      "step": 269970
    },
    {
      "epoch": 4.739900630277919,
      "grad_norm": 0.05307661307458954,
      "learning_rate": 3.37354695032343e-06,
      "loss": 0.8356,
      "step": 269980
    },
    {
      "epoch": 4.74007619515792,
      "grad_norm": 0.04797990388724096,
      "learning_rate": 3.3716977281481656e-06,
      "loss": 0.8389,
      "step": 269990
    },
    {
      "epoch": 4.740251760037922,
      "grad_norm": 0.045578064110531884,
      "learning_rate": 3.369849742942216e-06,
      "loss": 0.8442,
      "step": 270000
    },
    {
      "epoch": 4.740427324917923,
      "grad_norm": 0.06106701546868061,
      "learning_rate": 3.3680029947289708e-06,
      "loss": 0.843,
      "step": 270010
    },
    {
      "epoch": 4.740602889797925,
      "grad_norm": 0.051875651410074226,
      "learning_rate": 3.3661574835318415e-06,
      "loss": 0.8402,
      "step": 270020
    },
    {
      "epoch": 4.740778454677926,
      "grad_norm": 0.06168508521636349,
      "learning_rate": 3.364313209374205e-06,
      "loss": 0.836,
      "step": 270030
    },
    {
      "epoch": 4.740954019557927,
      "grad_norm": 0.05106223989197634,
      "learning_rate": 3.3624701722794305e-06,
      "loss": 0.8415,
      "step": 270040
    },
    {
      "epoch": 4.741129584437929,
      "grad_norm": 0.05129241486571749,
      "learning_rate": 3.3606283722708727e-06,
      "loss": 0.8363,
      "step": 270050
    },
    {
      "epoch": 4.74130514931793,
      "grad_norm": 0.06002905708372772,
      "learning_rate": 3.3587878093718674e-06,
      "loss": 0.8403,
      "step": 270060
    },
    {
      "epoch": 4.741480714197932,
      "grad_norm": 0.046288988501109934,
      "learning_rate": 3.3569484836057486e-06,
      "loss": 0.8425,
      "step": 270070
    },
    {
      "epoch": 4.741656279077933,
      "grad_norm": 0.09949551666996594,
      "learning_rate": 3.3551103949957724e-06,
      "loss": 0.8456,
      "step": 270080
    },
    {
      "epoch": 4.741831843957935,
      "grad_norm": 0.04433764247175743,
      "learning_rate": 3.3532735435652967e-06,
      "loss": 0.843,
      "step": 270090
    },
    {
      "epoch": 4.742007408837936,
      "grad_norm": 0.0512682538494922,
      "learning_rate": 3.351437929337545e-06,
      "loss": 0.8448,
      "step": 270100
    },
    {
      "epoch": 4.742182973717938,
      "grad_norm": 0.06002899844307116,
      "learning_rate": 3.3496035523357857e-06,
      "loss": 0.8384,
      "step": 270110
    },
    {
      "epoch": 4.742358538597939,
      "grad_norm": 0.04659568886838669,
      "learning_rate": 3.3477704125832552e-06,
      "loss": 0.8382,
      "step": 270120
    },
    {
      "epoch": 4.74253410347794,
      "grad_norm": 0.05258074437133813,
      "learning_rate": 3.3459385101031984e-06,
      "loss": 0.8387,
      "step": 270130
    },
    {
      "epoch": 4.742709668357942,
      "grad_norm": 0.05973101442031826,
      "learning_rate": 3.3441078449188185e-06,
      "loss": 0.8498,
      "step": 270140
    },
    {
      "epoch": 4.742885233237943,
      "grad_norm": 0.060622078124196156,
      "learning_rate": 3.342278417053308e-06,
      "loss": 0.8446,
      "step": 270150
    },
    {
      "epoch": 4.743060798117945,
      "grad_norm": 0.04523119719923329,
      "learning_rate": 3.3404502265298455e-06,
      "loss": 0.8376,
      "step": 270160
    },
    {
      "epoch": 4.743236362997946,
      "grad_norm": 0.041703477329352076,
      "learning_rate": 3.3386232733715905e-06,
      "loss": 0.8437,
      "step": 270170
    },
    {
      "epoch": 4.743411927877947,
      "grad_norm": 0.04936135184178244,
      "learning_rate": 3.3367975576017008e-06,
      "loss": 0.8307,
      "step": 270180
    },
    {
      "epoch": 4.743587492757949,
      "grad_norm": 0.05337973641459756,
      "learning_rate": 3.334973079243292e-06,
      "loss": 0.8399,
      "step": 270190
    },
    {
      "epoch": 4.74376305763795,
      "grad_norm": 0.04650477352865164,
      "learning_rate": 3.333149838319499e-06,
      "loss": 0.8339,
      "step": 270200
    },
    {
      "epoch": 4.743938622517952,
      "grad_norm": 0.04681623076168162,
      "learning_rate": 3.3313278348534273e-06,
      "loss": 0.8378,
      "step": 270210
    },
    {
      "epoch": 4.744114187397953,
      "grad_norm": 0.058882833674960416,
      "learning_rate": 3.3295070688681354e-06,
      "loss": 0.8408,
      "step": 270220
    },
    {
      "epoch": 4.744289752277954,
      "grad_norm": 0.054821604788888,
      "learning_rate": 3.327687540386705e-06,
      "loss": 0.839,
      "step": 270230
    },
    {
      "epoch": 4.744465317157956,
      "grad_norm": 0.06155126147987241,
      "learning_rate": 3.3258692494322065e-06,
      "loss": 0.8444,
      "step": 270240
    },
    {
      "epoch": 4.744640882037957,
      "grad_norm": 0.048239057229242883,
      "learning_rate": 3.3240521960276436e-06,
      "loss": 0.839,
      "step": 270250
    },
    {
      "epoch": 4.744816446917959,
      "grad_norm": 0.0555163726989012,
      "learning_rate": 3.3222363801960775e-06,
      "loss": 0.8379,
      "step": 270260
    },
    {
      "epoch": 4.74499201179796,
      "grad_norm": 0.05537660310329743,
      "learning_rate": 3.32042180196049e-06,
      "loss": 0.8376,
      "step": 270270
    },
    {
      "epoch": 4.745167576677961,
      "grad_norm": 0.057296672000888074,
      "learning_rate": 3.318608461343875e-06,
      "loss": 0.8381,
      "step": 270280
    },
    {
      "epoch": 4.745343141557963,
      "grad_norm": 0.043314535226188305,
      "learning_rate": 3.3167963583692148e-06,
      "loss": 0.8376,
      "step": 270290
    },
    {
      "epoch": 4.745518706437964,
      "grad_norm": 0.048847542111684156,
      "learning_rate": 3.31498549305947e-06,
      "loss": 0.8379,
      "step": 270300
    },
    {
      "epoch": 4.745694271317966,
      "grad_norm": 0.05141517850584879,
      "learning_rate": 3.313175865437579e-06,
      "loss": 0.8316,
      "step": 270310
    },
    {
      "epoch": 4.745869836197967,
      "grad_norm": 0.10218970380104432,
      "learning_rate": 3.31136747552646e-06,
      "loss": 0.849,
      "step": 270320
    },
    {
      "epoch": 4.746045401077969,
      "grad_norm": 0.07058957721385145,
      "learning_rate": 3.3095603233490494e-06,
      "loss": 0.8354,
      "step": 270330
    },
    {
      "epoch": 4.74622096595797,
      "grad_norm": 0.05203770393631048,
      "learning_rate": 3.3077544089282326e-06,
      "loss": 0.8416,
      "step": 270340
    },
    {
      "epoch": 4.746396530837972,
      "grad_norm": 0.05857776532731572,
      "learning_rate": 3.3059497322868815e-06,
      "loss": 0.835,
      "step": 270350
    },
    {
      "epoch": 4.746572095717973,
      "grad_norm": 0.04939546912427993,
      "learning_rate": 3.304146293447869e-06,
      "loss": 0.8361,
      "step": 270360
    },
    {
      "epoch": 4.746747660597974,
      "grad_norm": 0.06209039872546478,
      "learning_rate": 3.302344092434046e-06,
      "loss": 0.8405,
      "step": 270370
    },
    {
      "epoch": 4.746923225477976,
      "grad_norm": 0.051016435403898705,
      "learning_rate": 3.300543129268252e-06,
      "loss": 0.8364,
      "step": 270380
    },
    {
      "epoch": 4.747098790357977,
      "grad_norm": 0.06416328594709414,
      "learning_rate": 3.2987434039733047e-06,
      "loss": 0.8313,
      "step": 270390
    },
    {
      "epoch": 4.747274355237979,
      "grad_norm": 0.056149340235371284,
      "learning_rate": 3.2969449165719895e-06,
      "loss": 0.8416,
      "step": 270400
    },
    {
      "epoch": 4.74744992011798,
      "grad_norm": 0.0434353308858254,
      "learning_rate": 3.2951476670871235e-06,
      "loss": 0.8382,
      "step": 270410
    },
    {
      "epoch": 4.747625484997981,
      "grad_norm": 0.04866245250521003,
      "learning_rate": 3.2933516555414478e-06,
      "loss": 0.84,
      "step": 270420
    },
    {
      "epoch": 4.747801049877983,
      "grad_norm": 0.0571301710683115,
      "learning_rate": 3.291556881957747e-06,
      "loss": 0.8392,
      "step": 270430
    },
    {
      "epoch": 4.747976614757984,
      "grad_norm": 0.06537099344083497,
      "learning_rate": 3.289763346358739e-06,
      "loss": 0.8463,
      "step": 270440
    },
    {
      "epoch": 4.7481521796379855,
      "grad_norm": 0.05600803178631314,
      "learning_rate": 3.2879710487671557e-06,
      "loss": 0.8434,
      "step": 270450
    },
    {
      "epoch": 4.748327744517987,
      "grad_norm": 0.041087127703461165,
      "learning_rate": 3.286179989205692e-06,
      "loss": 0.8392,
      "step": 270460
    },
    {
      "epoch": 4.748503309397988,
      "grad_norm": 0.050434258680089224,
      "learning_rate": 3.2843901676970686e-06,
      "loss": 0.8373,
      "step": 270470
    },
    {
      "epoch": 4.7486788742779895,
      "grad_norm": 0.03716427768718725,
      "learning_rate": 3.2826015842639473e-06,
      "loss": 0.8468,
      "step": 270480
    },
    {
      "epoch": 4.7488544391579905,
      "grad_norm": 0.07201850458231546,
      "learning_rate": 3.2808142389289933e-06,
      "loss": 0.8387,
      "step": 270490
    },
    {
      "epoch": 4.7490300040379925,
      "grad_norm": 0.04294022736899864,
      "learning_rate": 3.279028131714848e-06,
      "loss": 0.8344,
      "step": 270500
    },
    {
      "epoch": 4.7492055689179935,
      "grad_norm": 0.0499524347791299,
      "learning_rate": 3.2772432626441535e-06,
      "loss": 0.834,
      "step": 270510
    },
    {
      "epoch": 4.749381133797995,
      "grad_norm": 0.05138282244571196,
      "learning_rate": 3.275459631739497e-06,
      "loss": 0.8383,
      "step": 270520
    },
    {
      "epoch": 4.7495566986779965,
      "grad_norm": 0.062100899140127654,
      "learning_rate": 3.273677239023509e-06,
      "loss": 0.8285,
      "step": 270530
    },
    {
      "epoch": 4.7497322635579975,
      "grad_norm": 0.058864676898113456,
      "learning_rate": 3.2718960845187553e-06,
      "loss": 0.8366,
      "step": 270540
    },
    {
      "epoch": 4.749907828437999,
      "grad_norm": 0.04894084416522389,
      "learning_rate": 3.2701161682478097e-06,
      "loss": 0.8429,
      "step": 270550
    },
    {
      "epoch": 4.7500833933180004,
      "grad_norm": 0.049139879353998425,
      "learning_rate": 3.268337490233217e-06,
      "loss": 0.8337,
      "step": 270560
    },
    {
      "epoch": 4.750258958198002,
      "grad_norm": 0.05708602351318422,
      "learning_rate": 3.2665600504975204e-06,
      "loss": 0.8325,
      "step": 270570
    },
    {
      "epoch": 4.750434523078003,
      "grad_norm": 0.04543547329410919,
      "learning_rate": 3.2647838490632397e-06,
      "loss": 0.8431,
      "step": 270580
    },
    {
      "epoch": 4.750610087958005,
      "grad_norm": 0.050039816193705074,
      "learning_rate": 3.2630088859528742e-06,
      "loss": 0.8359,
      "step": 270590
    },
    {
      "epoch": 4.750785652838006,
      "grad_norm": 0.053921826445360414,
      "learning_rate": 3.2612351611889233e-06,
      "loss": 0.8302,
      "step": 270600
    },
    {
      "epoch": 4.750961217718007,
      "grad_norm": 0.05228139559028975,
      "learning_rate": 3.2594626747938636e-06,
      "loss": 0.8448,
      "step": 270610
    },
    {
      "epoch": 4.751136782598009,
      "grad_norm": 0.048721128323644794,
      "learning_rate": 3.2576914267901286e-06,
      "loss": 0.8364,
      "step": 270620
    },
    {
      "epoch": 4.75131234747801,
      "grad_norm": 0.060704679337216925,
      "learning_rate": 3.255921417200185e-06,
      "loss": 0.8399,
      "step": 270630
    },
    {
      "epoch": 4.751487912358012,
      "grad_norm": 0.0446050423632679,
      "learning_rate": 3.254152646046442e-06,
      "loss": 0.8276,
      "step": 270640
    },
    {
      "epoch": 4.751663477238013,
      "grad_norm": 0.04647659806227319,
      "learning_rate": 3.2523851133513222e-06,
      "loss": 0.8432,
      "step": 270650
    },
    {
      "epoch": 4.751839042118014,
      "grad_norm": 0.053158569749119075,
      "learning_rate": 3.250618819137216e-06,
      "loss": 0.8543,
      "step": 270660
    },
    {
      "epoch": 4.752014606998016,
      "grad_norm": 0.04771647345033033,
      "learning_rate": 3.2488537634265e-06,
      "loss": 0.8337,
      "step": 270670
    },
    {
      "epoch": 4.752190171878017,
      "grad_norm": 0.0497789915833952,
      "learning_rate": 3.2470899462415424e-06,
      "loss": 0.8342,
      "step": 270680
    },
    {
      "epoch": 4.752365736758019,
      "grad_norm": 0.06969055934262718,
      "learning_rate": 3.245327367604688e-06,
      "loss": 0.831,
      "step": 270690
    },
    {
      "epoch": 4.75254130163802,
      "grad_norm": 0.04546738628059773,
      "learning_rate": 3.243566027538271e-06,
      "loss": 0.8374,
      "step": 270700
    },
    {
      "epoch": 4.752716866518021,
      "grad_norm": 0.06954260522572539,
      "learning_rate": 3.2418059260646025e-06,
      "loss": 0.8289,
      "step": 270710
    },
    {
      "epoch": 4.752892431398023,
      "grad_norm": 0.04883983841209657,
      "learning_rate": 3.2400470632059963e-06,
      "loss": 0.8374,
      "step": 270720
    },
    {
      "epoch": 4.753067996278024,
      "grad_norm": 0.05021226494281023,
      "learning_rate": 3.2382894389847304e-06,
      "loss": 0.8341,
      "step": 270730
    },
    {
      "epoch": 4.753243561158026,
      "grad_norm": 0.045622006661777126,
      "learning_rate": 3.2365330534230622e-06,
      "loss": 0.8354,
      "step": 270740
    },
    {
      "epoch": 4.753419126038027,
      "grad_norm": 0.05056057530822757,
      "learning_rate": 3.2347779065432717e-06,
      "loss": 0.8356,
      "step": 270750
    },
    {
      "epoch": 4.753594690918029,
      "grad_norm": 0.04871653113835665,
      "learning_rate": 3.233023998367571e-06,
      "loss": 0.8363,
      "step": 270760
    },
    {
      "epoch": 4.75377025579803,
      "grad_norm": 0.05897388101177949,
      "learning_rate": 3.2312713289181965e-06,
      "loss": 0.8485,
      "step": 270770
    },
    {
      "epoch": 4.753945820678032,
      "grad_norm": 0.050169921289439555,
      "learning_rate": 3.22951989821736e-06,
      "loss": 0.8419,
      "step": 270780
    },
    {
      "epoch": 4.754121385558033,
      "grad_norm": 0.055169919075575495,
      "learning_rate": 3.2277697062872313e-06,
      "loss": 0.8389,
      "step": 270790
    },
    {
      "epoch": 4.754296950438034,
      "grad_norm": 0.053570005521496974,
      "learning_rate": 3.2260207531500027e-06,
      "loss": 0.8458,
      "step": 270800
    },
    {
      "epoch": 4.754472515318036,
      "grad_norm": 0.04711782157778146,
      "learning_rate": 3.22427303882782e-06,
      "loss": 0.8433,
      "step": 270810
    },
    {
      "epoch": 4.754648080198037,
      "grad_norm": 0.049641366656153915,
      "learning_rate": 3.2225265633428535e-06,
      "loss": 0.8344,
      "step": 270820
    },
    {
      "epoch": 4.754823645078039,
      "grad_norm": 0.041519545558750835,
      "learning_rate": 3.220781326717206e-06,
      "loss": 0.8459,
      "step": 270830
    },
    {
      "epoch": 4.75499920995804,
      "grad_norm": 0.04739768527858636,
      "learning_rate": 3.2190373289730027e-06,
      "loss": 0.8444,
      "step": 270840
    },
    {
      "epoch": 4.755174774838041,
      "grad_norm": 0.05375754441511358,
      "learning_rate": 3.2172945701323363e-06,
      "loss": 0.8374,
      "step": 270850
    },
    {
      "epoch": 4.755350339718043,
      "grad_norm": 0.05376255607129604,
      "learning_rate": 3.215553050217277e-06,
      "loss": 0.8429,
      "step": 270860
    },
    {
      "epoch": 4.755525904598044,
      "grad_norm": 0.051695366041319524,
      "learning_rate": 3.213812769249906e-06,
      "loss": 0.8287,
      "step": 270870
    },
    {
      "epoch": 4.755701469478046,
      "grad_norm": 0.051838615157157146,
      "learning_rate": 3.2120737272522725e-06,
      "loss": 0.8298,
      "step": 270880
    },
    {
      "epoch": 4.755877034358047,
      "grad_norm": 0.04604301694625906,
      "learning_rate": 3.210335924246402e-06,
      "loss": 0.8423,
      "step": 270890
    },
    {
      "epoch": 4.756052599238048,
      "grad_norm": 0.04896197045001434,
      "learning_rate": 3.2085993602543202e-06,
      "loss": 0.8425,
      "step": 270900
    },
    {
      "epoch": 4.75622816411805,
      "grad_norm": 0.060778376759268206,
      "learning_rate": 3.2068640352980224e-06,
      "loss": 0.8366,
      "step": 270910
    },
    {
      "epoch": 4.756403728998051,
      "grad_norm": 0.06530358639316576,
      "learning_rate": 3.2051299493995003e-06,
      "loss": 0.8422,
      "step": 270920
    },
    {
      "epoch": 4.756579293878053,
      "grad_norm": 0.046333850275674986,
      "learning_rate": 3.2033971025807156e-06,
      "loss": 0.8331,
      "step": 270930
    },
    {
      "epoch": 4.756754858758054,
      "grad_norm": 0.05125117632559562,
      "learning_rate": 3.20166549486365e-06,
      "loss": 0.8428,
      "step": 270940
    },
    {
      "epoch": 4.756930423638055,
      "grad_norm": 0.0479131506170267,
      "learning_rate": 3.199935126270219e-06,
      "loss": 0.8362,
      "step": 270950
    },
    {
      "epoch": 4.757105988518057,
      "grad_norm": 0.06272170175082509,
      "learning_rate": 3.1982059968223524e-06,
      "loss": 0.8422,
      "step": 270960
    },
    {
      "epoch": 4.757281553398058,
      "grad_norm": 0.04934470200118367,
      "learning_rate": 3.1964781065419543e-06,
      "loss": 0.8396,
      "step": 270970
    },
    {
      "epoch": 4.75745711827806,
      "grad_norm": 0.04699288048044237,
      "learning_rate": 3.1947514554509185e-06,
      "loss": 0.8407,
      "step": 270980
    },
    {
      "epoch": 4.757632683158061,
      "grad_norm": 0.04750256753439146,
      "learning_rate": 3.1930260435711415e-06,
      "loss": 0.8405,
      "step": 270990
    },
    {
      "epoch": 4.757808248038063,
      "grad_norm": 0.0495094851224902,
      "learning_rate": 3.191301870924461e-06,
      "loss": 0.8382,
      "step": 271000
    },
    {
      "epoch": 4.757983812918064,
      "grad_norm": 0.05066369249900216,
      "learning_rate": 3.189578937532728e-06,
      "loss": 0.8442,
      "step": 271010
    },
    {
      "epoch": 4.758159377798066,
      "grad_norm": 0.042314152541360735,
      "learning_rate": 3.187857243417783e-06,
      "loss": 0.8382,
      "step": 271020
    },
    {
      "epoch": 4.758334942678067,
      "grad_norm": 0.049957018262588866,
      "learning_rate": 3.1861367886014196e-06,
      "loss": 0.8397,
      "step": 271030
    },
    {
      "epoch": 4.758510507558068,
      "grad_norm": 0.061857376265605474,
      "learning_rate": 3.184417573105457e-06,
      "loss": 0.8413,
      "step": 271040
    },
    {
      "epoch": 4.75868607243807,
      "grad_norm": 0.04350201672718212,
      "learning_rate": 3.182699596951669e-06,
      "loss": 0.8429,
      "step": 271050
    },
    {
      "epoch": 4.758861637318071,
      "grad_norm": 0.04828922331798451,
      "learning_rate": 3.1809828601618267e-06,
      "loss": 0.8274,
      "step": 271060
    },
    {
      "epoch": 4.759037202198073,
      "grad_norm": 0.04849686441288214,
      "learning_rate": 3.179267362757674e-06,
      "loss": 0.8424,
      "step": 271070
    },
    {
      "epoch": 4.759212767078074,
      "grad_norm": 0.06507825180000111,
      "learning_rate": 3.17755310476095e-06,
      "loss": 0.8375,
      "step": 271080
    },
    {
      "epoch": 4.759388331958075,
      "grad_norm": 0.05611490141302464,
      "learning_rate": 3.1758400861933836e-06,
      "loss": 0.8368,
      "step": 271090
    },
    {
      "epoch": 4.759563896838077,
      "grad_norm": 0.09122036002679286,
      "learning_rate": 3.1741283070766613e-06,
      "loss": 0.8342,
      "step": 271100
    },
    {
      "epoch": 4.759739461718078,
      "grad_norm": 0.08158851241790381,
      "learning_rate": 3.172417767432491e-06,
      "loss": 0.8379,
      "step": 271110
    },
    {
      "epoch": 4.75991502659808,
      "grad_norm": 0.05057759435715119,
      "learning_rate": 3.1707084672825358e-06,
      "loss": 0.844,
      "step": 271120
    },
    {
      "epoch": 4.760090591478081,
      "grad_norm": 0.04216180640753766,
      "learning_rate": 3.169000406648448e-06,
      "loss": 0.8396,
      "step": 271130
    },
    {
      "epoch": 4.760266156358082,
      "grad_norm": 0.04799174505604849,
      "learning_rate": 3.1672935855518814e-06,
      "loss": 0.839,
      "step": 271140
    },
    {
      "epoch": 4.760441721238084,
      "grad_norm": 0.044978553785657016,
      "learning_rate": 3.1655880040144444e-06,
      "loss": 0.843,
      "step": 271150
    },
    {
      "epoch": 4.760617286118085,
      "grad_norm": 0.07036420030013818,
      "learning_rate": 3.1638836620577665e-06,
      "loss": 0.8301,
      "step": 271160
    },
    {
      "epoch": 4.760792850998087,
      "grad_norm": 0.051508520919681014,
      "learning_rate": 3.162180559703425e-06,
      "loss": 0.8394,
      "step": 271170
    },
    {
      "epoch": 4.760968415878088,
      "grad_norm": 0.04616043876704121,
      "learning_rate": 3.1604786969730057e-06,
      "loss": 0.8351,
      "step": 271180
    },
    {
      "epoch": 4.7611439807580895,
      "grad_norm": 0.051002071924451275,
      "learning_rate": 3.158778073888086e-06,
      "loss": 0.8369,
      "step": 271190
    },
    {
      "epoch": 4.761319545638091,
      "grad_norm": 0.06320399686469767,
      "learning_rate": 3.157078690470183e-06,
      "loss": 0.8337,
      "step": 271200
    },
    {
      "epoch": 4.761495110518092,
      "grad_norm": 0.04910370676636998,
      "learning_rate": 3.1553805467408437e-06,
      "loss": 0.8376,
      "step": 271210
    },
    {
      "epoch": 4.7616706753980935,
      "grad_norm": 0.04402193894850012,
      "learning_rate": 3.1536836427215973e-06,
      "loss": 0.8375,
      "step": 271220
    },
    {
      "epoch": 4.7618462402780946,
      "grad_norm": 0.04511918416141858,
      "learning_rate": 3.1519879784339323e-06,
      "loss": 0.839,
      "step": 271230
    },
    {
      "epoch": 4.7620218051580965,
      "grad_norm": 0.04707698243762427,
      "learning_rate": 3.1502935538993147e-06,
      "loss": 0.8371,
      "step": 271240
    },
    {
      "epoch": 4.7621973700380975,
      "grad_norm": 0.061667066849666226,
      "learning_rate": 3.148600369139243e-06,
      "loss": 0.8393,
      "step": 271250
    },
    {
      "epoch": 4.762372934918099,
      "grad_norm": 0.06631474812330244,
      "learning_rate": 3.146908424175161e-06,
      "loss": 0.8332,
      "step": 271260
    },
    {
      "epoch": 4.7625484997981005,
      "grad_norm": 0.04894601317988778,
      "learning_rate": 3.1452177190284904e-06,
      "loss": 0.8308,
      "step": 271270
    },
    {
      "epoch": 4.7627240646781015,
      "grad_norm": 0.05075335331824453,
      "learning_rate": 3.1435282537206763e-06,
      "loss": 0.8431,
      "step": 271280
    },
    {
      "epoch": 4.762899629558103,
      "grad_norm": 0.04925546984764888,
      "learning_rate": 3.1418400282731174e-06,
      "loss": 0.8387,
      "step": 271290
    },
    {
      "epoch": 4.7630751944381045,
      "grad_norm": 0.06719340091583753,
      "learning_rate": 3.1401530427071918e-06,
      "loss": 0.8433,
      "step": 271300
    },
    {
      "epoch": 4.763250759318106,
      "grad_norm": 0.053641034017115494,
      "learning_rate": 3.1384672970442773e-06,
      "loss": 0.8364,
      "step": 271310
    },
    {
      "epoch": 4.763426324198107,
      "grad_norm": 0.05575639559834005,
      "learning_rate": 3.136782791305741e-06,
      "loss": 0.8356,
      "step": 271320
    },
    {
      "epoch": 4.763601889078108,
      "grad_norm": 0.056755631433561035,
      "learning_rate": 3.1350995255129273e-06,
      "loss": 0.8378,
      "step": 271330
    },
    {
      "epoch": 4.76377745395811,
      "grad_norm": 0.04556838681533919,
      "learning_rate": 3.1334174996871495e-06,
      "loss": 0.8414,
      "step": 271340
    },
    {
      "epoch": 4.763953018838111,
      "grad_norm": 0.07207157919685801,
      "learning_rate": 3.131736713849729e-06,
      "loss": 0.838,
      "step": 271350
    },
    {
      "epoch": 4.764128583718113,
      "grad_norm": 0.04286800940632896,
      "learning_rate": 3.130057168021979e-06,
      "loss": 0.8375,
      "step": 271360
    },
    {
      "epoch": 4.764304148598114,
      "grad_norm": 0.05331369675319611,
      "learning_rate": 3.1283788622251445e-06,
      "loss": 0.8423,
      "step": 271370
    },
    {
      "epoch": 4.764479713478115,
      "grad_norm": 0.05582712171674589,
      "learning_rate": 3.126701796480505e-06,
      "loss": 0.8375,
      "step": 271380
    },
    {
      "epoch": 4.764655278358117,
      "grad_norm": 0.04673141408576182,
      "learning_rate": 3.1250259708093066e-06,
      "loss": 0.8345,
      "step": 271390
    },
    {
      "epoch": 4.764830843238118,
      "grad_norm": 0.04614707878326132,
      "learning_rate": 3.1233513852328063e-06,
      "loss": 0.8419,
      "step": 271400
    },
    {
      "epoch": 4.76500640811812,
      "grad_norm": 0.05822677403639449,
      "learning_rate": 3.121678039772184e-06,
      "loss": 0.8355,
      "step": 271410
    },
    {
      "epoch": 4.765181972998121,
      "grad_norm": 0.04334761099375932,
      "learning_rate": 3.120005934448664e-06,
      "loss": 0.8409,
      "step": 271420
    },
    {
      "epoch": 4.765357537878123,
      "grad_norm": 0.0508257119019743,
      "learning_rate": 3.1183350692834358e-06,
      "loss": 0.8355,
      "step": 271430
    },
    {
      "epoch": 4.765533102758124,
      "grad_norm": 0.05079439966410048,
      "learning_rate": 3.116665444297649e-06,
      "loss": 0.8433,
      "step": 271440
    },
    {
      "epoch": 4.765708667638125,
      "grad_norm": 0.046292704470039955,
      "learning_rate": 3.114997059512482e-06,
      "loss": 0.8388,
      "step": 271450
    },
    {
      "epoch": 4.765884232518127,
      "grad_norm": 0.06148577121011536,
      "learning_rate": 3.11332991494906e-06,
      "loss": 0.8324,
      "step": 271460
    },
    {
      "epoch": 4.766059797398128,
      "grad_norm": 0.06236468299087744,
      "learning_rate": 3.1116640106284975e-06,
      "loss": 0.8433,
      "step": 271470
    },
    {
      "epoch": 4.76623536227813,
      "grad_norm": 0.048007557788704386,
      "learning_rate": 3.10999934657192e-06,
      "loss": 0.8299,
      "step": 271480
    },
    {
      "epoch": 4.766410927158131,
      "grad_norm": 0.05553435295644221,
      "learning_rate": 3.1083359228004076e-06,
      "loss": 0.8353,
      "step": 271490
    },
    {
      "epoch": 4.766586492038133,
      "grad_norm": 0.047181886976405615,
      "learning_rate": 3.106673739335053e-06,
      "loss": 0.8483,
      "step": 271500
    },
    {
      "epoch": 4.766762056918134,
      "grad_norm": 0.048184010531405606,
      "learning_rate": 3.105012796196883e-06,
      "loss": 0.839,
      "step": 271510
    },
    {
      "epoch": 4.766937621798135,
      "grad_norm": 0.06323833432354246,
      "learning_rate": 3.1033530934069775e-06,
      "loss": 0.8321,
      "step": 271520
    },
    {
      "epoch": 4.767113186678137,
      "grad_norm": 0.12148247422670377,
      "learning_rate": 3.1016946309863416e-06,
      "loss": 0.8343,
      "step": 271530
    },
    {
      "epoch": 4.767288751558138,
      "grad_norm": 0.054820959925492914,
      "learning_rate": 3.100037408956001e-06,
      "loss": 0.8372,
      "step": 271540
    },
    {
      "epoch": 4.76746431643814,
      "grad_norm": 0.07325980196775131,
      "learning_rate": 3.098381427336949e-06,
      "loss": 0.8422,
      "step": 271550
    },
    {
      "epoch": 4.767639881318141,
      "grad_norm": 0.0677934668803157,
      "learning_rate": 3.096726686150157e-06,
      "loss": 0.8359,
      "step": 271560
    },
    {
      "epoch": 4.767815446198142,
      "grad_norm": 0.058458925191270286,
      "learning_rate": 3.095073185416618e-06,
      "loss": 0.8332,
      "step": 271570
    },
    {
      "epoch": 4.767991011078144,
      "grad_norm": 0.052360144381423813,
      "learning_rate": 3.0934209251572596e-06,
      "loss": 0.8292,
      "step": 271580
    },
    {
      "epoch": 4.768166575958145,
      "grad_norm": 0.05595834606681241,
      "learning_rate": 3.0917699053930197e-06,
      "loss": 0.8312,
      "step": 271590
    },
    {
      "epoch": 4.768342140838147,
      "grad_norm": 0.0517464564305327,
      "learning_rate": 3.0901201261448145e-06,
      "loss": 0.8254,
      "step": 271600
    },
    {
      "epoch": 4.768517705718148,
      "grad_norm": 0.048215310524418506,
      "learning_rate": 3.08847158743355e-06,
      "loss": 0.8309,
      "step": 271610
    },
    {
      "epoch": 4.768693270598149,
      "grad_norm": 0.04695962487215543,
      "learning_rate": 3.0868242892801305e-06,
      "loss": 0.838,
      "step": 271620
    },
    {
      "epoch": 4.768868835478151,
      "grad_norm": 0.04685090833291896,
      "learning_rate": 3.085178231705407e-06,
      "loss": 0.8405,
      "step": 271630
    },
    {
      "epoch": 4.769044400358152,
      "grad_norm": 0.046947519785874275,
      "learning_rate": 3.0835334147302297e-06,
      "loss": 0.8405,
      "step": 271640
    },
    {
      "epoch": 4.769219965238154,
      "grad_norm": 0.04993960599291738,
      "learning_rate": 3.0818898383754595e-06,
      "loss": 0.8387,
      "step": 271650
    },
    {
      "epoch": 4.769395530118155,
      "grad_norm": 0.05442684057221733,
      "learning_rate": 3.080247502661903e-06,
      "loss": 0.8379,
      "step": 271660
    },
    {
      "epoch": 4.769571094998157,
      "grad_norm": 0.05475273258714771,
      "learning_rate": 3.078606407610389e-06,
      "loss": 0.8352,
      "step": 271670
    },
    {
      "epoch": 4.769746659878158,
      "grad_norm": 0.05188274088985188,
      "learning_rate": 3.0769665532416903e-06,
      "loss": 0.8355,
      "step": 271680
    },
    {
      "epoch": 4.76992222475816,
      "grad_norm": 0.04801000464249479,
      "learning_rate": 3.0753279395765807e-06,
      "loss": 0.8414,
      "step": 271690
    },
    {
      "epoch": 4.770097789638161,
      "grad_norm": 0.05114090982654578,
      "learning_rate": 3.0736905666358547e-06,
      "loss": 0.8374,
      "step": 271700
    },
    {
      "epoch": 4.770273354518162,
      "grad_norm": 0.052966068858296154,
      "learning_rate": 3.0720544344402207e-06,
      "loss": 0.8423,
      "step": 271710
    },
    {
      "epoch": 4.770448919398164,
      "grad_norm": 0.052374509632988106,
      "learning_rate": 3.0704195430104296e-06,
      "loss": 0.8351,
      "step": 271720
    },
    {
      "epoch": 4.770624484278165,
      "grad_norm": 0.049224668807189585,
      "learning_rate": 3.0687858923672004e-06,
      "loss": 0.8471,
      "step": 271730
    },
    {
      "epoch": 4.770800049158167,
      "grad_norm": 0.04893832427260954,
      "learning_rate": 3.067153482531218e-06,
      "loss": 0.8364,
      "step": 271740
    },
    {
      "epoch": 4.770975614038168,
      "grad_norm": 0.05726319810669407,
      "learning_rate": 3.065522313523157e-06,
      "loss": 0.8362,
      "step": 271750
    },
    {
      "epoch": 4.771151178918169,
      "grad_norm": 0.04456854875832515,
      "learning_rate": 3.0638923853637136e-06,
      "loss": 0.8378,
      "step": 271760
    },
    {
      "epoch": 4.771326743798171,
      "grad_norm": 0.09064007270316263,
      "learning_rate": 3.062263698073519e-06,
      "loss": 0.8369,
      "step": 271770
    },
    {
      "epoch": 4.771502308678172,
      "grad_norm": 0.05019757281402396,
      "learning_rate": 3.060636251673203e-06,
      "loss": 0.8368,
      "step": 271780
    },
    {
      "epoch": 4.771677873558174,
      "grad_norm": 0.04640220454695489,
      "learning_rate": 3.0590100461834073e-06,
      "loss": 0.8437,
      "step": 271790
    },
    {
      "epoch": 4.771853438438175,
      "grad_norm": 0.055834404937313266,
      "learning_rate": 3.0573850816247187e-06,
      "loss": 0.8429,
      "step": 271800
    },
    {
      "epoch": 4.772029003318176,
      "grad_norm": 0.04865562738890265,
      "learning_rate": 3.055761358017734e-06,
      "loss": 0.8365,
      "step": 271810
    },
    {
      "epoch": 4.772204568198178,
      "grad_norm": 0.050287794727624124,
      "learning_rate": 3.0541388753830188e-06,
      "loss": 0.8373,
      "step": 271820
    },
    {
      "epoch": 4.772380133078179,
      "grad_norm": 0.04745252734045799,
      "learning_rate": 3.052517633741148e-06,
      "loss": 0.8308,
      "step": 271830
    },
    {
      "epoch": 4.772555697958181,
      "grad_norm": 0.052719454337417225,
      "learning_rate": 3.050897633112642e-06,
      "loss": 0.8454,
      "step": 271840
    },
    {
      "epoch": 4.772731262838182,
      "grad_norm": 0.058145197703818235,
      "learning_rate": 3.0492788735180327e-06,
      "loss": 0.8357,
      "step": 271850
    },
    {
      "epoch": 4.772906827718184,
      "grad_norm": 0.0496581722149249,
      "learning_rate": 3.0476613549778295e-06,
      "loss": 0.8378,
      "step": 271860
    },
    {
      "epoch": 4.773082392598185,
      "grad_norm": 0.0475475905884917,
      "learning_rate": 3.0460450775125423e-06,
      "loss": 0.8409,
      "step": 271870
    },
    {
      "epoch": 4.773257957478186,
      "grad_norm": 0.0526072384018392,
      "learning_rate": 3.044430041142626e-06,
      "loss": 0.8365,
      "step": 271880
    },
    {
      "epoch": 4.773433522358188,
      "grad_norm": 0.041351976475230846,
      "learning_rate": 3.0428162458885565e-06,
      "loss": 0.8387,
      "step": 271890
    },
    {
      "epoch": 4.773609087238189,
      "grad_norm": 0.05796112984743137,
      "learning_rate": 3.0412036917707777e-06,
      "loss": 0.8377,
      "step": 271900
    },
    {
      "epoch": 4.773784652118191,
      "grad_norm": 0.05967469914285259,
      "learning_rate": 3.0395923788097217e-06,
      "loss": 0.8424,
      "step": 271910
    },
    {
      "epoch": 4.773960216998192,
      "grad_norm": 0.06128836117993918,
      "learning_rate": 3.037982307025812e-06,
      "loss": 0.8479,
      "step": 271920
    },
    {
      "epoch": 4.7741357818781935,
      "grad_norm": 0.041968077835822556,
      "learning_rate": 3.036373476439424e-06,
      "loss": 0.8382,
      "step": 271930
    },
    {
      "epoch": 4.774311346758195,
      "grad_norm": 0.06613798957388019,
      "learning_rate": 3.034765887070981e-06,
      "loss": 0.8295,
      "step": 271940
    },
    {
      "epoch": 4.774486911638196,
      "grad_norm": 0.051132717444606804,
      "learning_rate": 3.033159538940805e-06,
      "loss": 0.8437,
      "step": 271950
    },
    {
      "epoch": 4.7746624765181975,
      "grad_norm": 0.051986385549468765,
      "learning_rate": 3.031554432069296e-06,
      "loss": 0.8432,
      "step": 271960
    },
    {
      "epoch": 4.774838041398199,
      "grad_norm": 0.05131312205958178,
      "learning_rate": 3.029950566476754e-06,
      "loss": 0.8415,
      "step": 271970
    },
    {
      "epoch": 4.7750136062782005,
      "grad_norm": 0.04692997165847548,
      "learning_rate": 3.0283479421835253e-06,
      "loss": 0.8399,
      "step": 271980
    },
    {
      "epoch": 4.7751891711582015,
      "grad_norm": 0.04753910924717359,
      "learning_rate": 3.026746559209887e-06,
      "loss": 0.8352,
      "step": 271990
    },
    {
      "epoch": 4.7753647360382026,
      "grad_norm": 0.052590182153495056,
      "learning_rate": 3.0251464175761516e-06,
      "loss": 0.8387,
      "step": 272000
    },
    {
      "epoch": 4.7755403009182045,
      "grad_norm": 0.05079497141634782,
      "learning_rate": 3.023547517302598e-06,
      "loss": 0.8329,
      "step": 272010
    },
    {
      "epoch": 4.7757158657982055,
      "grad_norm": 0.04471549530958251,
      "learning_rate": 3.021949858409472e-06,
      "loss": 0.8377,
      "step": 272020
    },
    {
      "epoch": 4.775891430678207,
      "grad_norm": 0.07403224223379759,
      "learning_rate": 3.020353440917008e-06,
      "loss": 0.8392,
      "step": 272030
    },
    {
      "epoch": 4.7760669955582085,
      "grad_norm": 0.04523049307403165,
      "learning_rate": 3.0187582648454634e-06,
      "loss": 0.8457,
      "step": 272040
    },
    {
      "epoch": 4.7762425604382095,
      "grad_norm": 0.046249311421430225,
      "learning_rate": 3.0171643302150063e-06,
      "loss": 0.8421,
      "step": 272050
    },
    {
      "epoch": 4.776418125318211,
      "grad_norm": 0.05177348603866279,
      "learning_rate": 3.015571637045872e-06,
      "loss": 0.8315,
      "step": 272060
    },
    {
      "epoch": 4.7765936901982124,
      "grad_norm": 0.05429786167154706,
      "learning_rate": 3.0139801853582186e-06,
      "loss": 0.84,
      "step": 272070
    },
    {
      "epoch": 4.776769255078214,
      "grad_norm": 0.0472539858562014,
      "learning_rate": 3.012389975172215e-06,
      "loss": 0.8285,
      "step": 272080
    },
    {
      "epoch": 4.776944819958215,
      "grad_norm": 0.06726479266482321,
      "learning_rate": 3.0108010065080073e-06,
      "loss": 0.8401,
      "step": 272090
    },
    {
      "epoch": 4.777120384838217,
      "grad_norm": 0.045408363016968385,
      "learning_rate": 3.0092132793857315e-06,
      "loss": 0.8341,
      "step": 272100
    },
    {
      "epoch": 4.777295949718218,
      "grad_norm": 0.053299034049128405,
      "learning_rate": 3.0076267938255025e-06,
      "loss": 0.8341,
      "step": 272110
    },
    {
      "epoch": 4.777471514598219,
      "grad_norm": 0.057201910320460354,
      "learning_rate": 3.006041549847422e-06,
      "loss": 0.8308,
      "step": 272120
    },
    {
      "epoch": 4.777647079478221,
      "grad_norm": 0.06131261662966852,
      "learning_rate": 3.004457547471583e-06,
      "loss": 0.8405,
      "step": 272130
    },
    {
      "epoch": 4.777822644358222,
      "grad_norm": 0.06165089014902228,
      "learning_rate": 3.002874786718044e-06,
      "loss": 0.8403,
      "step": 272140
    },
    {
      "epoch": 4.777998209238224,
      "grad_norm": 0.053992010245312215,
      "learning_rate": 3.0012932676068527e-06,
      "loss": 0.8507,
      "step": 272150
    },
    {
      "epoch": 4.778173774118225,
      "grad_norm": 0.05397051905317473,
      "learning_rate": 2.9997129901580684e-06,
      "loss": 0.8432,
      "step": 272160
    },
    {
      "epoch": 4.778349338998227,
      "grad_norm": 0.049771610929222714,
      "learning_rate": 2.998133954391695e-06,
      "loss": 0.8388,
      "step": 272170
    },
    {
      "epoch": 4.778524903878228,
      "grad_norm": 0.04725557022636839,
      "learning_rate": 2.9965561603277586e-06,
      "loss": 0.8293,
      "step": 272180
    },
    {
      "epoch": 4.778700468758229,
      "grad_norm": 0.07401234985355087,
      "learning_rate": 2.99497960798623e-06,
      "loss": 0.8392,
      "step": 272190
    },
    {
      "epoch": 4.778876033638231,
      "grad_norm": 0.05036470722457178,
      "learning_rate": 2.9934042973870803e-06,
      "loss": 0.8417,
      "step": 272200
    },
    {
      "epoch": 4.779051598518232,
      "grad_norm": 0.051520491426241904,
      "learning_rate": 2.9918302285503028e-06,
      "loss": 0.8364,
      "step": 272210
    },
    {
      "epoch": 4.779227163398234,
      "grad_norm": 0.0763385776409845,
      "learning_rate": 2.9902574014958018e-06,
      "loss": 0.8468,
      "step": 272220
    },
    {
      "epoch": 4.779402728278235,
      "grad_norm": 0.05415339441461384,
      "learning_rate": 2.988685816243527e-06,
      "loss": 0.836,
      "step": 272230
    },
    {
      "epoch": 4.779578293158236,
      "grad_norm": 0.04355447847057992,
      "learning_rate": 2.987115472813394e-06,
      "loss": 0.8433,
      "step": 272240
    },
    {
      "epoch": 4.779753858038238,
      "grad_norm": 0.07974319836812874,
      "learning_rate": 2.985546371225287e-06,
      "loss": 0.8392,
      "step": 272250
    },
    {
      "epoch": 4.779929422918239,
      "grad_norm": 0.05745297876893198,
      "learning_rate": 2.9839785114990987e-06,
      "loss": 0.8396,
      "step": 272260
    },
    {
      "epoch": 4.780104987798241,
      "grad_norm": 0.06123438202070261,
      "learning_rate": 2.98241189365468e-06,
      "loss": 0.84,
      "step": 272270
    },
    {
      "epoch": 4.780280552678242,
      "grad_norm": 0.05252246026806511,
      "learning_rate": 2.9808465177118918e-06,
      "loss": 0.8373,
      "step": 272280
    },
    {
      "epoch": 4.780456117558243,
      "grad_norm": 0.0584847499596752,
      "learning_rate": 2.9792823836905514e-06,
      "loss": 0.8356,
      "step": 272290
    },
    {
      "epoch": 4.780631682438245,
      "grad_norm": 0.04092336029168831,
      "learning_rate": 2.9777194916104977e-06,
      "loss": 0.8463,
      "step": 272300
    },
    {
      "epoch": 4.780807247318246,
      "grad_norm": 0.06668370317051449,
      "learning_rate": 2.9761578414915373e-06,
      "loss": 0.8441,
      "step": 272310
    },
    {
      "epoch": 4.780982812198248,
      "grad_norm": 0.05887218172963556,
      "learning_rate": 2.9745974333534204e-06,
      "loss": 0.844,
      "step": 272320
    },
    {
      "epoch": 4.781158377078249,
      "grad_norm": 0.04963314592918964,
      "learning_rate": 2.9730382672159546e-06,
      "loss": 0.8231,
      "step": 272330
    },
    {
      "epoch": 4.781333941958251,
      "grad_norm": 0.05513565984105903,
      "learning_rate": 2.9714803430988676e-06,
      "loss": 0.8371,
      "step": 272340
    },
    {
      "epoch": 4.781509506838252,
      "grad_norm": 0.052073189648283925,
      "learning_rate": 2.969923661021934e-06,
      "loss": 0.8345,
      "step": 272350
    },
    {
      "epoch": 4.781685071718254,
      "grad_norm": 0.04956061466172039,
      "learning_rate": 2.968368221004838e-06,
      "loss": 0.8342,
      "step": 272360
    },
    {
      "epoch": 4.781860636598255,
      "grad_norm": 0.07003331534839052,
      "learning_rate": 2.966814023067309e-06,
      "loss": 0.8317,
      "step": 272370
    },
    {
      "epoch": 4.782036201478256,
      "grad_norm": 0.0566851936567614,
      "learning_rate": 2.9652610672290323e-06,
      "loss": 0.8428,
      "step": 272380
    },
    {
      "epoch": 4.782211766358258,
      "grad_norm": 0.04462201459975026,
      "learning_rate": 2.963709353509683e-06,
      "loss": 0.8326,
      "step": 272390
    },
    {
      "epoch": 4.782387331238259,
      "grad_norm": 0.053242754235159014,
      "learning_rate": 2.9621588819289343e-06,
      "loss": 0.8333,
      "step": 272400
    },
    {
      "epoch": 4.782562896118261,
      "grad_norm": 0.05542314436797711,
      "learning_rate": 2.9606096525064062e-06,
      "loss": 0.8267,
      "step": 272410
    },
    {
      "epoch": 4.782738460998262,
      "grad_norm": 0.05365925173725804,
      "learning_rate": 2.9590616652617615e-06,
      "loss": 0.8395,
      "step": 272420
    },
    {
      "epoch": 4.782914025878263,
      "grad_norm": 0.045062111203402845,
      "learning_rate": 2.9575149202145865e-06,
      "loss": 0.8357,
      "step": 272430
    },
    {
      "epoch": 4.783089590758265,
      "grad_norm": 0.04901015648438551,
      "learning_rate": 2.955969417384479e-06,
      "loss": 0.8411,
      "step": 272440
    },
    {
      "epoch": 4.783265155638266,
      "grad_norm": 0.05290357746798014,
      "learning_rate": 2.9544251567910464e-06,
      "loss": 0.8424,
      "step": 272450
    },
    {
      "epoch": 4.783440720518268,
      "grad_norm": 0.05779306273445232,
      "learning_rate": 2.952882138453821e-06,
      "loss": 0.8321,
      "step": 272460
    },
    {
      "epoch": 4.783616285398269,
      "grad_norm": 0.04210137667575115,
      "learning_rate": 2.9513403623923784e-06,
      "loss": 0.8466,
      "step": 272470
    },
    {
      "epoch": 4.78379185027827,
      "grad_norm": 0.04554060099178381,
      "learning_rate": 2.949799828626249e-06,
      "loss": 0.8415,
      "step": 272480
    },
    {
      "epoch": 4.783967415158272,
      "grad_norm": 0.06775259358198148,
      "learning_rate": 2.9482605371749425e-06,
      "loss": 0.8409,
      "step": 272490
    },
    {
      "epoch": 4.784142980038273,
      "grad_norm": 0.04898149312976878,
      "learning_rate": 2.9467224880579577e-06,
      "loss": 0.8415,
      "step": 272500
    },
    {
      "epoch": 4.784318544918275,
      "grad_norm": 0.07156119321163482,
      "learning_rate": 2.9451856812947923e-06,
      "loss": 0.8327,
      "step": 272510
    },
    {
      "epoch": 4.784494109798276,
      "grad_norm": 0.043847431509277106,
      "learning_rate": 2.9436501169049346e-06,
      "loss": 0.8326,
      "step": 272520
    },
    {
      "epoch": 4.784669674678277,
      "grad_norm": 0.05718004176203886,
      "learning_rate": 2.9421157949078055e-06,
      "loss": 0.8328,
      "step": 272530
    },
    {
      "epoch": 4.784845239558279,
      "grad_norm": 0.049987945184977775,
      "learning_rate": 2.9405827153228592e-06,
      "loss": 0.8364,
      "step": 272540
    },
    {
      "epoch": 4.78502080443828,
      "grad_norm": 0.04540508540671505,
      "learning_rate": 2.93905087816954e-06,
      "loss": 0.8363,
      "step": 272550
    },
    {
      "epoch": 4.785196369318282,
      "grad_norm": 0.050466865093385334,
      "learning_rate": 2.9375202834672245e-06,
      "loss": 0.8341,
      "step": 272560
    },
    {
      "epoch": 4.785371934198283,
      "grad_norm": 0.051836461132813504,
      "learning_rate": 2.9359909312353233e-06,
      "loss": 0.8404,
      "step": 272570
    },
    {
      "epoch": 4.785547499078285,
      "grad_norm": 0.05065113237817642,
      "learning_rate": 2.9344628214932147e-06,
      "loss": 0.8342,
      "step": 272580
    },
    {
      "epoch": 4.785723063958286,
      "grad_norm": 0.05849737797384587,
      "learning_rate": 2.9329359542602535e-06,
      "loss": 0.8395,
      "step": 272590
    },
    {
      "epoch": 4.785898628838288,
      "grad_norm": 0.061141963950995665,
      "learning_rate": 2.9314103295557846e-06,
      "loss": 0.8441,
      "step": 272600
    },
    {
      "epoch": 4.786074193718289,
      "grad_norm": 0.06200351088845982,
      "learning_rate": 2.9298859473991516e-06,
      "loss": 0.8324,
      "step": 272610
    },
    {
      "epoch": 4.78624975859829,
      "grad_norm": 0.1171198222851616,
      "learning_rate": 2.9283628078096564e-06,
      "loss": 0.8392,
      "step": 272620
    },
    {
      "epoch": 4.786425323478292,
      "grad_norm": 0.06627124053538233,
      "learning_rate": 2.926840910806588e-06,
      "loss": 0.8248,
      "step": 272630
    },
    {
      "epoch": 4.786600888358293,
      "grad_norm": 0.04256122894686353,
      "learning_rate": 2.9253202564092575e-06,
      "loss": 0.8391,
      "step": 272640
    },
    {
      "epoch": 4.786776453238295,
      "grad_norm": 0.05852519608021237,
      "learning_rate": 2.923800844636911e-06,
      "loss": 0.8399,
      "step": 272650
    },
    {
      "epoch": 4.786952018118296,
      "grad_norm": 0.05322086756714874,
      "learning_rate": 2.9222826755087944e-06,
      "loss": 0.8361,
      "step": 272660
    },
    {
      "epoch": 4.787127582998297,
      "grad_norm": 0.06075103182594135,
      "learning_rate": 2.9207657490441527e-06,
      "loss": 0.8475,
      "step": 272670
    },
    {
      "epoch": 4.787303147878299,
      "grad_norm": 0.060439836141617534,
      "learning_rate": 2.9192500652622094e-06,
      "loss": 0.8451,
      "step": 272680
    },
    {
      "epoch": 4.7874787127583,
      "grad_norm": 0.049366551753258235,
      "learning_rate": 2.9177356241821782e-06,
      "loss": 0.8374,
      "step": 272690
    },
    {
      "epoch": 4.7876542776383015,
      "grad_norm": 0.06717103203827511,
      "learning_rate": 2.9162224258232153e-06,
      "loss": 0.8367,
      "step": 272700
    },
    {
      "epoch": 4.787829842518303,
      "grad_norm": 0.07974625305893539,
      "learning_rate": 2.9147104702045236e-06,
      "loss": 0.8357,
      "step": 272710
    },
    {
      "epoch": 4.788005407398304,
      "grad_norm": 0.05187647392746153,
      "learning_rate": 2.9131997573452487e-06,
      "loss": 0.8373,
      "step": 272720
    },
    {
      "epoch": 4.7881809722783055,
      "grad_norm": 0.05031327692220071,
      "learning_rate": 2.9116902872645276e-06,
      "loss": 0.8353,
      "step": 272730
    },
    {
      "epoch": 4.788356537158307,
      "grad_norm": 0.05486571348650635,
      "learning_rate": 2.9101820599814837e-06,
      "loss": 0.8359,
      "step": 272740
    },
    {
      "epoch": 4.7885321020383085,
      "grad_norm": 0.051190094493191375,
      "learning_rate": 2.9086750755152316e-06,
      "loss": 0.8379,
      "step": 272750
    },
    {
      "epoch": 4.7887076669183095,
      "grad_norm": 0.05264359497721495,
      "learning_rate": 2.9071693338848736e-06,
      "loss": 0.842,
      "step": 272760
    },
    {
      "epoch": 4.788883231798311,
      "grad_norm": 0.06463692518768416,
      "learning_rate": 2.90566483510948e-06,
      "loss": 0.8287,
      "step": 272770
    },
    {
      "epoch": 4.7890587966783125,
      "grad_norm": 0.05270197616874153,
      "learning_rate": 2.904161579208108e-06,
      "loss": 0.8406,
      "step": 272780
    },
    {
      "epoch": 4.7892343615583135,
      "grad_norm": 0.0517538054160639,
      "learning_rate": 2.9026595661998073e-06,
      "loss": 0.8347,
      "step": 272790
    },
    {
      "epoch": 4.789409926438315,
      "grad_norm": 0.05271477471074964,
      "learning_rate": 2.9011587961036136e-06,
      "loss": 0.8311,
      "step": 272800
    },
    {
      "epoch": 4.7895854913183165,
      "grad_norm": 0.044253022947663305,
      "learning_rate": 2.8996592689385307e-06,
      "loss": 0.8375,
      "step": 272810
    },
    {
      "epoch": 4.789761056198318,
      "grad_norm": 0.04801087163220455,
      "learning_rate": 2.898160984723585e-06,
      "loss": 0.8316,
      "step": 272820
    },
    {
      "epoch": 4.789936621078319,
      "grad_norm": 0.07191219902452366,
      "learning_rate": 2.8966639434777252e-06,
      "loss": 0.8372,
      "step": 272830
    },
    {
      "epoch": 4.790112185958321,
      "grad_norm": 0.049254055095491356,
      "learning_rate": 2.8951681452199323e-06,
      "loss": 0.8352,
      "step": 272840
    },
    {
      "epoch": 4.790287750838322,
      "grad_norm": 0.05624239069998303,
      "learning_rate": 2.893673589969167e-06,
      "loss": 0.8339,
      "step": 272850
    },
    {
      "epoch": 4.790463315718323,
      "grad_norm": 0.07001655948731995,
      "learning_rate": 2.892180277744367e-06,
      "loss": 0.8452,
      "step": 272860
    },
    {
      "epoch": 4.790638880598325,
      "grad_norm": 0.05335236854064375,
      "learning_rate": 2.890688208564426e-06,
      "loss": 0.8447,
      "step": 272870
    },
    {
      "epoch": 4.790814445478326,
      "grad_norm": 0.05924205618755279,
      "learning_rate": 2.8891973824482814e-06,
      "loss": 0.8403,
      "step": 272880
    },
    {
      "epoch": 4.790990010358328,
      "grad_norm": 0.049782566080600096,
      "learning_rate": 2.887707799414806e-06,
      "loss": 0.8343,
      "step": 272890
    },
    {
      "epoch": 4.791165575238329,
      "grad_norm": 0.05061983145407054,
      "learning_rate": 2.886219459482882e-06,
      "loss": 0.8424,
      "step": 272900
    },
    {
      "epoch": 4.79134114011833,
      "grad_norm": 0.05001516970047681,
      "learning_rate": 2.884732362671349e-06,
      "loss": 0.8413,
      "step": 272910
    },
    {
      "epoch": 4.791516704998332,
      "grad_norm": 0.05902628642560134,
      "learning_rate": 2.8832465089990672e-06,
      "loss": 0.8342,
      "step": 272920
    },
    {
      "epoch": 4.791692269878333,
      "grad_norm": 0.05972825235462269,
      "learning_rate": 2.8817618984848545e-06,
      "loss": 0.8318,
      "step": 272930
    },
    {
      "epoch": 4.791867834758335,
      "grad_norm": 0.05783646018683862,
      "learning_rate": 2.880278531147527e-06,
      "loss": 0.8418,
      "step": 272940
    },
    {
      "epoch": 4.792043399638336,
      "grad_norm": 0.07222545560472893,
      "learning_rate": 2.8787964070058683e-06,
      "loss": 0.8352,
      "step": 272950
    },
    {
      "epoch": 4.792218964518337,
      "grad_norm": 0.04938546120214106,
      "learning_rate": 2.8773155260786747e-06,
      "loss": 0.8352,
      "step": 272960
    },
    {
      "epoch": 4.792394529398339,
      "grad_norm": 0.047309763079297476,
      "learning_rate": 2.875835888384674e-06,
      "loss": 0.8287,
      "step": 272970
    },
    {
      "epoch": 4.79257009427834,
      "grad_norm": 0.044903013533756836,
      "learning_rate": 2.874357493942662e-06,
      "loss": 0.8305,
      "step": 272980
    },
    {
      "epoch": 4.792745659158342,
      "grad_norm": 0.05223074968104964,
      "learning_rate": 2.872880342771345e-06,
      "loss": 0.8391,
      "step": 272990
    },
    {
      "epoch": 4.792921224038343,
      "grad_norm": 0.07007765930840969,
      "learning_rate": 2.8714044348894307e-06,
      "loss": 0.8385,
      "step": 273000
    },
    {
      "epoch": 4.793096788918345,
      "grad_norm": 0.06715124321390767,
      "learning_rate": 2.869929770315637e-06,
      "loss": 0.8346,
      "step": 273010
    },
    {
      "epoch": 4.793272353798346,
      "grad_norm": 0.04679157263483986,
      "learning_rate": 2.8684563490686267e-06,
      "loss": 0.8346,
      "step": 273020
    },
    {
      "epoch": 4.793447918678348,
      "grad_norm": 0.054952421320724934,
      "learning_rate": 2.866984171167096e-06,
      "loss": 0.8418,
      "step": 273030
    },
    {
      "epoch": 4.793623483558349,
      "grad_norm": 0.04355778966459859,
      "learning_rate": 2.8655132366296746e-06,
      "loss": 0.8399,
      "step": 273040
    },
    {
      "epoch": 4.79379904843835,
      "grad_norm": 0.06669245547686664,
      "learning_rate": 2.8640435454750045e-06,
      "loss": 0.8348,
      "step": 273050
    },
    {
      "epoch": 4.793974613318352,
      "grad_norm": 0.060256361709516745,
      "learning_rate": 2.862575097721726e-06,
      "loss": 0.8416,
      "step": 273060
    },
    {
      "epoch": 4.794150178198353,
      "grad_norm": 0.05779187902935643,
      "learning_rate": 2.8611078933884143e-06,
      "loss": 0.8403,
      "step": 273070
    },
    {
      "epoch": 4.794325743078355,
      "grad_norm": 0.05551534473036465,
      "learning_rate": 2.859641932493678e-06,
      "loss": 0.8343,
      "step": 273080
    },
    {
      "epoch": 4.794501307958356,
      "grad_norm": 0.05677549013349524,
      "learning_rate": 2.858177215056081e-06,
      "loss": 0.8387,
      "step": 273090
    },
    {
      "epoch": 4.794676872838357,
      "grad_norm": 0.05283057358707089,
      "learning_rate": 2.8567137410941984e-06,
      "loss": 0.8264,
      "step": 273100
    },
    {
      "epoch": 4.794852437718359,
      "grad_norm": 0.052024897547416744,
      "learning_rate": 2.855251510626561e-06,
      "loss": 0.8432,
      "step": 273110
    },
    {
      "epoch": 4.79502800259836,
      "grad_norm": 0.04724292497892216,
      "learning_rate": 2.8537905236716896e-06,
      "loss": 0.8397,
      "step": 273120
    },
    {
      "epoch": 4.795203567478362,
      "grad_norm": 0.053889806385812776,
      "learning_rate": 2.852330780248104e-06,
      "loss": 0.8396,
      "step": 273130
    },
    {
      "epoch": 4.795379132358363,
      "grad_norm": 0.06108352254654817,
      "learning_rate": 2.850872280374292e-06,
      "loss": 0.8366,
      "step": 273140
    },
    {
      "epoch": 4.795554697238364,
      "grad_norm": 0.06627764960451696,
      "learning_rate": 2.8494150240687508e-06,
      "loss": 0.84,
      "step": 273150
    },
    {
      "epoch": 4.795730262118366,
      "grad_norm": 0.04443483520565985,
      "learning_rate": 2.8479590113499247e-06,
      "loss": 0.8394,
      "step": 273160
    },
    {
      "epoch": 4.795905826998367,
      "grad_norm": 0.05072424258055007,
      "learning_rate": 2.846504242236267e-06,
      "loss": 0.8423,
      "step": 273170
    },
    {
      "epoch": 4.796081391878369,
      "grad_norm": 0.05704110196943755,
      "learning_rate": 2.8450507167462106e-06,
      "loss": 0.8305,
      "step": 273180
    },
    {
      "epoch": 4.79625695675837,
      "grad_norm": 0.05943635413399807,
      "learning_rate": 2.8435984348981766e-06,
      "loss": 0.8461,
      "step": 273190
    },
    {
      "epoch": 4.796432521638371,
      "grad_norm": 0.0564606158356772,
      "learning_rate": 2.8421473967105643e-06,
      "loss": 0.8348,
      "step": 273200
    },
    {
      "epoch": 4.796608086518373,
      "grad_norm": 0.046856436700104344,
      "learning_rate": 2.8406976022017507e-06,
      "loss": 0.8423,
      "step": 273210
    },
    {
      "epoch": 4.796783651398374,
      "grad_norm": 0.05476818821094388,
      "learning_rate": 2.8392490513901137e-06,
      "loss": 0.8376,
      "step": 273220
    },
    {
      "epoch": 4.796959216278376,
      "grad_norm": 0.05715358407742426,
      "learning_rate": 2.837801744294008e-06,
      "loss": 0.8324,
      "step": 273230
    },
    {
      "epoch": 4.797134781158377,
      "grad_norm": 0.0467898203480553,
      "learning_rate": 2.8363556809317564e-06,
      "loss": 0.8395,
      "step": 273240
    },
    {
      "epoch": 4.797310346038379,
      "grad_norm": 0.07099244577662506,
      "learning_rate": 2.8349108613217028e-06,
      "loss": 0.8407,
      "step": 273250
    },
    {
      "epoch": 4.79748591091838,
      "grad_norm": 0.04450691197176799,
      "learning_rate": 2.8334672854821264e-06,
      "loss": 0.8412,
      "step": 273260
    },
    {
      "epoch": 4.797661475798382,
      "grad_norm": 0.04596383473520709,
      "learning_rate": 2.832024953431348e-06,
      "loss": 0.8383,
      "step": 273270
    },
    {
      "epoch": 4.797837040678383,
      "grad_norm": 0.053478884993333774,
      "learning_rate": 2.8305838651876257e-06,
      "loss": 0.8442,
      "step": 273280
    },
    {
      "epoch": 4.798012605558384,
      "grad_norm": 0.0502824777620097,
      "learning_rate": 2.8291440207692145e-06,
      "loss": 0.8382,
      "step": 273290
    },
    {
      "epoch": 4.798188170438386,
      "grad_norm": 0.05523903914424387,
      "learning_rate": 2.8277054201943613e-06,
      "loss": 0.8372,
      "step": 273300
    },
    {
      "epoch": 4.798363735318387,
      "grad_norm": 0.0459653547727196,
      "learning_rate": 2.8262680634812988e-06,
      "loss": 0.8345,
      "step": 273310
    },
    {
      "epoch": 4.798539300198389,
      "grad_norm": 0.045190771556223266,
      "learning_rate": 2.8248319506482408e-06,
      "loss": 0.8381,
      "step": 273320
    },
    {
      "epoch": 4.79871486507839,
      "grad_norm": 0.04103432343079966,
      "learning_rate": 2.8233970817133663e-06,
      "loss": 0.8405,
      "step": 273330
    },
    {
      "epoch": 4.798890429958391,
      "grad_norm": 0.048575224040649144,
      "learning_rate": 2.821963456694876e-06,
      "loss": 0.846,
      "step": 273340
    },
    {
      "epoch": 4.799065994838393,
      "grad_norm": 0.045335192671927135,
      "learning_rate": 2.820531075610918e-06,
      "loss": 0.8433,
      "step": 273350
    },
    {
      "epoch": 4.799241559718394,
      "grad_norm": 0.05522109388490602,
      "learning_rate": 2.8190999384796373e-06,
      "loss": 0.8383,
      "step": 273360
    },
    {
      "epoch": 4.799417124598396,
      "grad_norm": 0.046278290612909986,
      "learning_rate": 2.8176700453191923e-06,
      "loss": 0.8353,
      "step": 273370
    },
    {
      "epoch": 4.799592689478397,
      "grad_norm": 0.06304691455283677,
      "learning_rate": 2.816241396147685e-06,
      "loss": 0.8314,
      "step": 273380
    },
    {
      "epoch": 4.799768254358398,
      "grad_norm": 0.04340434022187186,
      "learning_rate": 2.814813990983207e-06,
      "loss": 0.8338,
      "step": 273390
    },
    {
      "epoch": 4.7999438192384,
      "grad_norm": 0.05750832249920061,
      "learning_rate": 2.813387829843861e-06,
      "loss": 0.8387,
      "step": 273400
    },
    {
      "epoch": 4.800119384118401,
      "grad_norm": 0.04734387336845292,
      "learning_rate": 2.8119629127477053e-06,
      "loss": 0.8386,
      "step": 273410
    },
    {
      "epoch": 4.800294948998403,
      "grad_norm": 0.053375768160957314,
      "learning_rate": 2.810539239712788e-06,
      "loss": 0.8441,
      "step": 273420
    },
    {
      "epoch": 4.800470513878404,
      "grad_norm": 0.0501417982139343,
      "learning_rate": 2.8091168107571667e-06,
      "loss": 0.8409,
      "step": 273430
    },
    {
      "epoch": 4.8006460787584055,
      "grad_norm": 0.06007955756468875,
      "learning_rate": 2.8076956258988573e-06,
      "loss": 0.8414,
      "step": 273440
    },
    {
      "epoch": 4.800821643638407,
      "grad_norm": 0.05219339548589624,
      "learning_rate": 2.8062756851558514e-06,
      "loss": 0.834,
      "step": 273450
    },
    {
      "epoch": 4.800997208518408,
      "grad_norm": 0.055381367315034166,
      "learning_rate": 2.8048569885461527e-06,
      "loss": 0.8424,
      "step": 273460
    },
    {
      "epoch": 4.8011727733984095,
      "grad_norm": 0.05499448717127039,
      "learning_rate": 2.803439536087743e-06,
      "loss": 0.8364,
      "step": 273470
    },
    {
      "epoch": 4.801348338278411,
      "grad_norm": 0.043156935884811884,
      "learning_rate": 2.8020233277985713e-06,
      "loss": 0.8391,
      "step": 273480
    },
    {
      "epoch": 4.8015239031584125,
      "grad_norm": 0.0627095948238732,
      "learning_rate": 2.800608363696585e-06,
      "loss": 0.8436,
      "step": 273490
    },
    {
      "epoch": 4.8016994680384135,
      "grad_norm": 0.05600234448731083,
      "learning_rate": 2.799194643799712e-06,
      "loss": 0.8397,
      "step": 273500
    },
    {
      "epoch": 4.801875032918415,
      "grad_norm": 0.05669472623657756,
      "learning_rate": 2.797782168125857e-06,
      "loss": 0.8417,
      "step": 273510
    },
    {
      "epoch": 4.8020505977984165,
      "grad_norm": 0.05494440984824445,
      "learning_rate": 2.7963709366929242e-06,
      "loss": 0.8419,
      "step": 273520
    },
    {
      "epoch": 4.8022261626784175,
      "grad_norm": 0.05350391029861425,
      "learning_rate": 2.7949609495187857e-06,
      "loss": 0.8299,
      "step": 273530
    },
    {
      "epoch": 4.802401727558419,
      "grad_norm": 0.05701916678257653,
      "learning_rate": 2.7935522066213235e-06,
      "loss": 0.8332,
      "step": 273540
    },
    {
      "epoch": 4.8025772924384205,
      "grad_norm": 0.05270732642708108,
      "learning_rate": 2.7921447080183657e-06,
      "loss": 0.8406,
      "step": 273550
    },
    {
      "epoch": 4.802752857318422,
      "grad_norm": 0.07089415587909113,
      "learning_rate": 2.790738453727762e-06,
      "loss": 0.8454,
      "step": 273560
    },
    {
      "epoch": 4.802928422198423,
      "grad_norm": 0.046545926065815826,
      "learning_rate": 2.7893334437673293e-06,
      "loss": 0.8377,
      "step": 273570
    },
    {
      "epoch": 4.8031039870784245,
      "grad_norm": 0.05158493306681433,
      "learning_rate": 2.7879296781548625e-06,
      "loss": 0.836,
      "step": 273580
    },
    {
      "epoch": 4.803279551958426,
      "grad_norm": 0.04936613695602706,
      "learning_rate": 2.7865271569081334e-06,
      "loss": 0.8404,
      "step": 273590
    },
    {
      "epoch": 4.803455116838427,
      "grad_norm": 0.055032483199221854,
      "learning_rate": 2.7851258800449386e-06,
      "loss": 0.8362,
      "step": 273600
    },
    {
      "epoch": 4.803630681718429,
      "grad_norm": 0.08719845235547871,
      "learning_rate": 2.7837258475830267e-06,
      "loss": 0.8475,
      "step": 273610
    },
    {
      "epoch": 4.80380624659843,
      "grad_norm": 0.051376394459901005,
      "learning_rate": 2.7823270595401284e-06,
      "loss": 0.8386,
      "step": 273620
    },
    {
      "epoch": 4.803981811478431,
      "grad_norm": 0.04802552553335705,
      "learning_rate": 2.780929515933972e-06,
      "loss": 0.8387,
      "step": 273630
    },
    {
      "epoch": 4.804157376358433,
      "grad_norm": 0.055168456047945916,
      "learning_rate": 2.7795332167822523e-06,
      "loss": 0.8409,
      "step": 273640
    },
    {
      "epoch": 4.804332941238434,
      "grad_norm": 0.07140633419542976,
      "learning_rate": 2.778138162102677e-06,
      "loss": 0.8383,
      "step": 273650
    },
    {
      "epoch": 4.804508506118436,
      "grad_norm": 0.05048735364421116,
      "learning_rate": 2.7767443519129208e-06,
      "loss": 0.8417,
      "step": 273660
    },
    {
      "epoch": 4.804684070998437,
      "grad_norm": 0.053057210301874005,
      "learning_rate": 2.775351786230634e-06,
      "loss": 0.8367,
      "step": 273670
    },
    {
      "epoch": 4.804859635878439,
      "grad_norm": 0.051384205576900176,
      "learning_rate": 2.773960465073458e-06,
      "loss": 0.8332,
      "step": 273680
    },
    {
      "epoch": 4.80503520075844,
      "grad_norm": 0.057730887155330333,
      "learning_rate": 2.772570388459034e-06,
      "loss": 0.8468,
      "step": 273690
    },
    {
      "epoch": 4.805210765638442,
      "grad_norm": 0.04979515040868421,
      "learning_rate": 2.771181556404959e-06,
      "loss": 0.8376,
      "step": 273700
    },
    {
      "epoch": 4.805386330518443,
      "grad_norm": 0.05505416143428089,
      "learning_rate": 2.7697939689288515e-06,
      "loss": 0.838,
      "step": 273710
    },
    {
      "epoch": 4.805561895398444,
      "grad_norm": 0.05296292114786627,
      "learning_rate": 2.7684076260482758e-06,
      "loss": 0.8312,
      "step": 273720
    },
    {
      "epoch": 4.805737460278446,
      "grad_norm": 0.07356385281199877,
      "learning_rate": 2.7670225277808064e-06,
      "loss": 0.8338,
      "step": 273730
    },
    {
      "epoch": 4.805913025158447,
      "grad_norm": 0.049450834222695764,
      "learning_rate": 2.7656386741439856e-06,
      "loss": 0.8347,
      "step": 273740
    },
    {
      "epoch": 4.806088590038449,
      "grad_norm": 0.04835167639125916,
      "learning_rate": 2.7642560651553443e-06,
      "loss": 0.8387,
      "step": 273750
    },
    {
      "epoch": 4.80626415491845,
      "grad_norm": 0.05388142273795797,
      "learning_rate": 2.762874700832403e-06,
      "loss": 0.8482,
      "step": 273760
    },
    {
      "epoch": 4.806439719798451,
      "grad_norm": 0.0550001498786457,
      "learning_rate": 2.7614945811926694e-06,
      "loss": 0.8276,
      "step": 273770
    },
    {
      "epoch": 4.806615284678453,
      "grad_norm": 0.04871263777974383,
      "learning_rate": 2.760115706253632e-06,
      "loss": 0.8323,
      "step": 273780
    },
    {
      "epoch": 4.806790849558454,
      "grad_norm": 0.05140904201542726,
      "learning_rate": 2.758738076032755e-06,
      "loss": 0.8326,
      "step": 273790
    },
    {
      "epoch": 4.806966414438456,
      "grad_norm": 0.054478179861003624,
      "learning_rate": 2.7573616905474922e-06,
      "loss": 0.8419,
      "step": 273800
    },
    {
      "epoch": 4.807141979318457,
      "grad_norm": 0.05531657125836398,
      "learning_rate": 2.755986549815287e-06,
      "loss": 0.8243,
      "step": 273810
    },
    {
      "epoch": 4.807317544198458,
      "grad_norm": 0.041214739299007645,
      "learning_rate": 2.7546126538535503e-06,
      "loss": 0.8388,
      "step": 273820
    },
    {
      "epoch": 4.80749310907846,
      "grad_norm": 0.0490472170678095,
      "learning_rate": 2.753240002679713e-06,
      "loss": 0.8375,
      "step": 273830
    },
    {
      "epoch": 4.807668673958461,
      "grad_norm": 0.043477221850859823,
      "learning_rate": 2.7518685963111523e-06,
      "loss": 0.8449,
      "step": 273840
    },
    {
      "epoch": 4.807844238838463,
      "grad_norm": 0.054395727477970925,
      "learning_rate": 2.7504984347652343e-06,
      "loss": 0.8414,
      "step": 273850
    },
    {
      "epoch": 4.808019803718464,
      "grad_norm": 0.05885581000972204,
      "learning_rate": 2.749129518059337e-06,
      "loss": 0.8373,
      "step": 273860
    },
    {
      "epoch": 4.808195368598465,
      "grad_norm": 0.045735304436029606,
      "learning_rate": 2.7477618462108037e-06,
      "loss": 0.8425,
      "step": 273870
    },
    {
      "epoch": 4.808370933478467,
      "grad_norm": 0.05941723123880172,
      "learning_rate": 2.7463954192369563e-06,
      "loss": 0.8427,
      "step": 273880
    },
    {
      "epoch": 4.808546498358468,
      "grad_norm": 0.05490845292081517,
      "learning_rate": 2.7450302371551073e-06,
      "loss": 0.8355,
      "step": 273890
    },
    {
      "epoch": 4.80872206323847,
      "grad_norm": 0.04579878491329432,
      "learning_rate": 2.743666299982555e-06,
      "loss": 0.8385,
      "step": 273900
    },
    {
      "epoch": 4.808897628118471,
      "grad_norm": 0.0472595791606158,
      "learning_rate": 2.7423036077365795e-06,
      "loss": 0.8437,
      "step": 273910
    },
    {
      "epoch": 4.809073192998473,
      "grad_norm": 0.054958833484623035,
      "learning_rate": 2.740942160434457e-06,
      "loss": 0.8388,
      "step": 273920
    },
    {
      "epoch": 4.809248757878474,
      "grad_norm": 0.04287784919351478,
      "learning_rate": 2.7395819580934125e-06,
      "loss": 0.8351,
      "step": 273930
    },
    {
      "epoch": 4.809424322758476,
      "grad_norm": 0.07154161510818999,
      "learning_rate": 2.738223000730712e-06,
      "loss": 0.8415,
      "step": 273940
    },
    {
      "epoch": 4.809599887638477,
      "grad_norm": 0.048537101118523496,
      "learning_rate": 2.736865288363546e-06,
      "loss": 0.8375,
      "step": 273950
    },
    {
      "epoch": 4.809775452518478,
      "grad_norm": 0.06304159633431661,
      "learning_rate": 2.7355088210091382e-06,
      "loss": 0.8379,
      "step": 273960
    },
    {
      "epoch": 4.80995101739848,
      "grad_norm": 0.04757719785257009,
      "learning_rate": 2.7341535986846563e-06,
      "loss": 0.841,
      "step": 273970
    },
    {
      "epoch": 4.810126582278481,
      "grad_norm": 0.05250908224084286,
      "learning_rate": 2.7327996214072913e-06,
      "loss": 0.8294,
      "step": 273980
    },
    {
      "epoch": 4.810302147158483,
      "grad_norm": 0.05744159728192114,
      "learning_rate": 2.7314468891941886e-06,
      "loss": 0.8353,
      "step": 273990
    },
    {
      "epoch": 4.810477712038484,
      "grad_norm": 0.049875583689256785,
      "learning_rate": 2.730095402062473e-06,
      "loss": 0.8314,
      "step": 274000
    },
    {
      "epoch": 4.810653276918485,
      "grad_norm": 0.04432889845353877,
      "learning_rate": 2.7287451600293016e-06,
      "loss": 0.8426,
      "step": 274010
    },
    {
      "epoch": 4.810828841798487,
      "grad_norm": 0.04807263168156763,
      "learning_rate": 2.7273961631117436e-06,
      "loss": 0.8302,
      "step": 274020
    },
    {
      "epoch": 4.811004406678488,
      "grad_norm": 0.052133741825497446,
      "learning_rate": 2.726048411326913e-06,
      "loss": 0.8373,
      "step": 274030
    },
    {
      "epoch": 4.81117997155849,
      "grad_norm": 0.047510591391694154,
      "learning_rate": 2.7247019046918896e-06,
      "loss": 0.8291,
      "step": 274040
    },
    {
      "epoch": 4.811355536438491,
      "grad_norm": 0.0476299502665492,
      "learning_rate": 2.723356643223732e-06,
      "loss": 0.8408,
      "step": 274050
    },
    {
      "epoch": 4.811531101318492,
      "grad_norm": 0.06816853263031739,
      "learning_rate": 2.7220126269394754e-06,
      "loss": 0.8386,
      "step": 274060
    },
    {
      "epoch": 4.811706666198494,
      "grad_norm": 0.062460173064042133,
      "learning_rate": 2.720669855856147e-06,
      "loss": 0.8335,
      "step": 274070
    },
    {
      "epoch": 4.811882231078495,
      "grad_norm": 0.05588080699337944,
      "learning_rate": 2.719328329990782e-06,
      "loss": 0.8357,
      "step": 274080
    },
    {
      "epoch": 4.812057795958497,
      "grad_norm": 0.05027591745926784,
      "learning_rate": 2.7179880493603513e-06,
      "loss": 0.8394,
      "step": 274090
    },
    {
      "epoch": 4.812233360838498,
      "grad_norm": 0.05294402029794779,
      "learning_rate": 2.7166490139818473e-06,
      "loss": 0.8466,
      "step": 274100
    },
    {
      "epoch": 4.8124089257185,
      "grad_norm": 0.05857836935313177,
      "learning_rate": 2.7153112238722294e-06,
      "loss": 0.8349,
      "step": 274110
    },
    {
      "epoch": 4.812584490598501,
      "grad_norm": 0.052159958928901254,
      "learning_rate": 2.7139746790484676e-06,
      "loss": 0.8361,
      "step": 274120
    },
    {
      "epoch": 4.812760055478502,
      "grad_norm": 0.04611464364343843,
      "learning_rate": 2.712639379527478e-06,
      "loss": 0.8368,
      "step": 274130
    },
    {
      "epoch": 4.812935620358504,
      "grad_norm": 0.0505817095671278,
      "learning_rate": 2.7113053253261864e-06,
      "loss": 0.8349,
      "step": 274140
    },
    {
      "epoch": 4.813111185238505,
      "grad_norm": 0.047330964811583956,
      "learning_rate": 2.709972516461498e-06,
      "loss": 0.8411,
      "step": 274150
    },
    {
      "epoch": 4.813286750118507,
      "grad_norm": 0.07991379307677449,
      "learning_rate": 2.708640952950284e-06,
      "loss": 0.8422,
      "step": 274160
    },
    {
      "epoch": 4.813462314998508,
      "grad_norm": 0.046234165487679425,
      "learning_rate": 2.707310634809426e-06,
      "loss": 0.8371,
      "step": 274170
    },
    {
      "epoch": 4.8136378798785096,
      "grad_norm": 0.054470840165176816,
      "learning_rate": 2.7059815620557863e-06,
      "loss": 0.8347,
      "step": 274180
    },
    {
      "epoch": 4.813813444758511,
      "grad_norm": 0.06331153572832125,
      "learning_rate": 2.704653734706202e-06,
      "loss": 0.8409,
      "step": 274190
    },
    {
      "epoch": 4.813989009638512,
      "grad_norm": 0.0783254638578779,
      "learning_rate": 2.7033271527774795e-06,
      "loss": 0.8381,
      "step": 274200
    },
    {
      "epoch": 4.8141645745185135,
      "grad_norm": 0.04624760256893523,
      "learning_rate": 2.7020018162864567e-06,
      "loss": 0.8382,
      "step": 274210
    },
    {
      "epoch": 4.814340139398515,
      "grad_norm": 0.06850580339394562,
      "learning_rate": 2.7006777252499063e-06,
      "loss": 0.8329,
      "step": 274220
    },
    {
      "epoch": 4.8145157042785165,
      "grad_norm": 0.04177568116558813,
      "learning_rate": 2.6993548796846015e-06,
      "loss": 0.8376,
      "step": 274230
    },
    {
      "epoch": 4.8146912691585175,
      "grad_norm": 0.06267301148721093,
      "learning_rate": 2.698033279607303e-06,
      "loss": 0.8322,
      "step": 274240
    },
    {
      "epoch": 4.814866834038519,
      "grad_norm": 0.0716090966351115,
      "learning_rate": 2.696712925034773e-06,
      "loss": 0.836,
      "step": 274250
    },
    {
      "epoch": 4.8150423989185205,
      "grad_norm": 0.05441569158967458,
      "learning_rate": 2.695393815983729e-06,
      "loss": 0.8376,
      "step": 274260
    },
    {
      "epoch": 4.8152179637985215,
      "grad_norm": 0.0499427873689832,
      "learning_rate": 2.6940759524708887e-06,
      "loss": 0.8423,
      "step": 274270
    },
    {
      "epoch": 4.815393528678523,
      "grad_norm": 0.04635721532955182,
      "learning_rate": 2.6927593345129364e-06,
      "loss": 0.839,
      "step": 274280
    },
    {
      "epoch": 4.8155690935585245,
      "grad_norm": 0.04751693951685802,
      "learning_rate": 2.6914439621265787e-06,
      "loss": 0.839,
      "step": 274290
    },
    {
      "epoch": 4.8157446584385255,
      "grad_norm": 0.052849409514471264,
      "learning_rate": 2.6901298353284576e-06,
      "loss": 0.8381,
      "step": 274300
    },
    {
      "epoch": 4.815920223318527,
      "grad_norm": 0.052895367974738504,
      "learning_rate": 2.6888169541352223e-06,
      "loss": 0.8466,
      "step": 274310
    },
    {
      "epoch": 4.8160957881985285,
      "grad_norm": 0.05776267355681738,
      "learning_rate": 2.687505318563538e-06,
      "loss": 0.8406,
      "step": 274320
    },
    {
      "epoch": 4.81627135307853,
      "grad_norm": 0.05314365820847243,
      "learning_rate": 2.6861949286299893e-06,
      "loss": 0.8385,
      "step": 274330
    },
    {
      "epoch": 4.816446917958531,
      "grad_norm": 0.043585922067338356,
      "learning_rate": 2.6848857843511943e-06,
      "loss": 0.8457,
      "step": 274340
    },
    {
      "epoch": 4.816622482838533,
      "grad_norm": 0.04187818110033791,
      "learning_rate": 2.68357788574374e-06,
      "loss": 0.8365,
      "step": 274350
    },
    {
      "epoch": 4.816798047718534,
      "grad_norm": 0.0495627669318699,
      "learning_rate": 2.6822712328241994e-06,
      "loss": 0.84,
      "step": 274360
    },
    {
      "epoch": 4.816973612598535,
      "grad_norm": 0.06441702459249636,
      "learning_rate": 2.680965825609117e-06,
      "loss": 0.8451,
      "step": 274370
    },
    {
      "epoch": 4.817149177478537,
      "grad_norm": 0.04628159894711274,
      "learning_rate": 2.679661664115043e-06,
      "loss": 0.8424,
      "step": 274380
    },
    {
      "epoch": 4.817324742358538,
      "grad_norm": 0.043159504370785,
      "learning_rate": 2.678358748358509e-06,
      "loss": 0.8415,
      "step": 274390
    },
    {
      "epoch": 4.81750030723854,
      "grad_norm": 0.05234719638722515,
      "learning_rate": 2.677057078356002e-06,
      "loss": 0.8377,
      "step": 274400
    },
    {
      "epoch": 4.817675872118541,
      "grad_norm": 0.05365199459993951,
      "learning_rate": 2.6757566541240204e-06,
      "loss": 0.8362,
      "step": 274410
    },
    {
      "epoch": 4.817851436998543,
      "grad_norm": 0.05292267691780359,
      "learning_rate": 2.6744574756790604e-06,
      "loss": 0.8363,
      "step": 274420
    },
    {
      "epoch": 4.818027001878544,
      "grad_norm": 0.045726360428158454,
      "learning_rate": 2.6731595430375553e-06,
      "loss": 0.8338,
      "step": 274430
    },
    {
      "epoch": 4.818202566758545,
      "grad_norm": 0.054804219515055604,
      "learning_rate": 2.67186285621597e-06,
      "loss": 0.8334,
      "step": 274440
    },
    {
      "epoch": 4.818378131638547,
      "grad_norm": 0.048784839968412604,
      "learning_rate": 2.6705674152307237e-06,
      "loss": 0.8306,
      "step": 274450
    },
    {
      "epoch": 4.818553696518548,
      "grad_norm": 0.06639449827794473,
      "learning_rate": 2.6692732200982387e-06,
      "loss": 0.8426,
      "step": 274460
    },
    {
      "epoch": 4.81872926139855,
      "grad_norm": 0.060787795893001394,
      "learning_rate": 2.667980270834903e-06,
      "loss": 0.8347,
      "step": 274470
    },
    {
      "epoch": 4.818904826278551,
      "grad_norm": 0.04789555597792899,
      "learning_rate": 2.666688567457104e-06,
      "loss": 0.8435,
      "step": 274480
    },
    {
      "epoch": 4.819080391158552,
      "grad_norm": 0.05370176370490783,
      "learning_rate": 2.665398109981207e-06,
      "loss": 0.8417,
      "step": 274490
    },
    {
      "epoch": 4.819255956038554,
      "grad_norm": 0.04170856075861181,
      "learning_rate": 2.664108898423557e-06,
      "loss": 0.8363,
      "step": 274500
    },
    {
      "epoch": 4.819431520918555,
      "grad_norm": 0.039758572141325076,
      "learning_rate": 2.6628209328004977e-06,
      "loss": 0.8384,
      "step": 274510
    },
    {
      "epoch": 4.819607085798557,
      "grad_norm": 0.05041261471786215,
      "learning_rate": 2.6615342131283392e-06,
      "loss": 0.8334,
      "step": 274520
    },
    {
      "epoch": 4.819782650678558,
      "grad_norm": 0.07568819229042462,
      "learning_rate": 2.6602487394233935e-06,
      "loss": 0.8351,
      "step": 274530
    },
    {
      "epoch": 4.819958215558559,
      "grad_norm": 0.049969717678388856,
      "learning_rate": 2.6589645117019383e-06,
      "loss": 0.837,
      "step": 274540
    },
    {
      "epoch": 4.820133780438561,
      "grad_norm": 0.05764947460909601,
      "learning_rate": 2.6576815299802515e-06,
      "loss": 0.8356,
      "step": 274550
    },
    {
      "epoch": 4.820309345318562,
      "grad_norm": 0.04926990174381782,
      "learning_rate": 2.656399794274589e-06,
      "loss": 0.8412,
      "step": 274560
    },
    {
      "epoch": 4.820484910198564,
      "grad_norm": 0.05927757439200278,
      "learning_rate": 2.6551193046011845e-06,
      "loss": 0.8309,
      "step": 274570
    },
    {
      "epoch": 4.820660475078565,
      "grad_norm": 0.05294733621179258,
      "learning_rate": 2.6538400609762728e-06,
      "loss": 0.8342,
      "step": 274580
    },
    {
      "epoch": 4.820836039958567,
      "grad_norm": 0.053094900878462956,
      "learning_rate": 2.6525620634160545e-06,
      "loss": 0.8387,
      "step": 274590
    },
    {
      "epoch": 4.821011604838568,
      "grad_norm": 0.05272757671609848,
      "learning_rate": 2.6512853119367084e-06,
      "loss": 0.8392,
      "step": 274600
    },
    {
      "epoch": 4.82118716971857,
      "grad_norm": 0.11747388538621623,
      "learning_rate": 2.650009806554436e-06,
      "loss": 0.8363,
      "step": 274610
    },
    {
      "epoch": 4.821362734598571,
      "grad_norm": 0.05708374536696195,
      "learning_rate": 2.6487355472853942e-06,
      "loss": 0.8385,
      "step": 274620
    },
    {
      "epoch": 4.821538299478572,
      "grad_norm": 0.05367379767065688,
      "learning_rate": 2.647462534145718e-06,
      "loss": 0.8421,
      "step": 274630
    },
    {
      "epoch": 4.821713864358574,
      "grad_norm": 0.0467326387823937,
      "learning_rate": 2.646190767151543e-06,
      "loss": 0.838,
      "step": 274640
    },
    {
      "epoch": 4.821889429238575,
      "grad_norm": 0.05397787101301422,
      "learning_rate": 2.6449202463189704e-06,
      "loss": 0.8313,
      "step": 274650
    },
    {
      "epoch": 4.822064994118577,
      "grad_norm": 0.04691355722044592,
      "learning_rate": 2.643650971664125e-06,
      "loss": 0.8366,
      "step": 274660
    },
    {
      "epoch": 4.822240558998578,
      "grad_norm": 0.06322988722110144,
      "learning_rate": 2.6423829432030537e-06,
      "loss": 0.8376,
      "step": 274670
    },
    {
      "epoch": 4.822416123878579,
      "grad_norm": 0.04833355451809514,
      "learning_rate": 2.641116160951848e-06,
      "loss": 0.8404,
      "step": 274680
    },
    {
      "epoch": 4.822591688758581,
      "grad_norm": 0.042032972414826654,
      "learning_rate": 2.6398506249265545e-06,
      "loss": 0.8411,
      "step": 274690
    },
    {
      "epoch": 4.822767253638582,
      "grad_norm": 0.043564952241920744,
      "learning_rate": 2.6385863351431982e-06,
      "loss": 0.8389,
      "step": 274700
    },
    {
      "epoch": 4.822942818518584,
      "grad_norm": 0.05367997056934279,
      "learning_rate": 2.6373232916178055e-06,
      "loss": 0.838,
      "step": 274710
    },
    {
      "epoch": 4.823118383398585,
      "grad_norm": 0.05210193165190063,
      "learning_rate": 2.636061494366379e-06,
      "loss": 0.8443,
      "step": 274720
    },
    {
      "epoch": 4.823293948278586,
      "grad_norm": 0.04762947654934266,
      "learning_rate": 2.63480094340491e-06,
      "loss": 0.84,
      "step": 274730
    },
    {
      "epoch": 4.823469513158588,
      "grad_norm": 0.04655483271292889,
      "learning_rate": 2.633541638749359e-06,
      "loss": 0.8397,
      "step": 274740
    },
    {
      "epoch": 4.823645078038589,
      "grad_norm": 0.07853533613181898,
      "learning_rate": 2.6322835804156843e-06,
      "loss": 0.8446,
      "step": 274750
    },
    {
      "epoch": 4.823820642918591,
      "grad_norm": 0.0635948624330069,
      "learning_rate": 2.631026768419835e-06,
      "loss": 0.8356,
      "step": 274760
    },
    {
      "epoch": 4.823996207798592,
      "grad_norm": 0.06114650491965545,
      "learning_rate": 2.6297712027777255e-06,
      "loss": 0.8323,
      "step": 274770
    },
    {
      "epoch": 4.824171772678594,
      "grad_norm": 0.04838962814988243,
      "learning_rate": 2.62851688350526e-06,
      "loss": 0.8378,
      "step": 274780
    },
    {
      "epoch": 4.824347337558595,
      "grad_norm": 0.06101627614310327,
      "learning_rate": 2.6272638106183553e-06,
      "loss": 0.8431,
      "step": 274790
    },
    {
      "epoch": 4.824522902438596,
      "grad_norm": 0.05008757658798925,
      "learning_rate": 2.6260119841328594e-06,
      "loss": 0.8308,
      "step": 274800
    },
    {
      "epoch": 4.824698467318598,
      "grad_norm": 0.0476590880729688,
      "learning_rate": 2.6247614040646435e-06,
      "loss": 0.8407,
      "step": 274810
    },
    {
      "epoch": 4.824874032198599,
      "grad_norm": 0.052235951633122606,
      "learning_rate": 2.6235120704295574e-06,
      "loss": 0.8395,
      "step": 274820
    },
    {
      "epoch": 4.825049597078601,
      "grad_norm": 0.04793029521055945,
      "learning_rate": 2.6222639832434396e-06,
      "loss": 0.8458,
      "step": 274830
    },
    {
      "epoch": 4.825225161958602,
      "grad_norm": 0.05576796028477506,
      "learning_rate": 2.621017142522073e-06,
      "loss": 0.8432,
      "step": 274840
    },
    {
      "epoch": 4.825400726838604,
      "grad_norm": 0.049864772133782584,
      "learning_rate": 2.6197715482812747e-06,
      "loss": 0.8336,
      "step": 274850
    },
    {
      "epoch": 4.825576291718605,
      "grad_norm": 0.05422453430335835,
      "learning_rate": 2.618527200536838e-06,
      "loss": 0.8409,
      "step": 274860
    },
    {
      "epoch": 4.825751856598606,
      "grad_norm": 0.09981768138556454,
      "learning_rate": 2.6172840993045138e-06,
      "loss": 0.8316,
      "step": 274870
    },
    {
      "epoch": 4.825927421478608,
      "grad_norm": 0.0575394607189031,
      "learning_rate": 2.616042244600042e-06,
      "loss": 0.8374,
      "step": 274880
    },
    {
      "epoch": 4.826102986358609,
      "grad_norm": 0.05270341199762077,
      "learning_rate": 2.6148016364391827e-06,
      "loss": 0.8401,
      "step": 274890
    },
    {
      "epoch": 4.826278551238611,
      "grad_norm": 0.05862363979357398,
      "learning_rate": 2.613562274837633e-06,
      "loss": 0.8355,
      "step": 274900
    },
    {
      "epoch": 4.826454116118612,
      "grad_norm": 0.05954218983085429,
      "learning_rate": 2.6123241598111086e-06,
      "loss": 0.8364,
      "step": 274910
    },
    {
      "epoch": 4.826629680998613,
      "grad_norm": 0.05281234623293971,
      "learning_rate": 2.6110872913752958e-06,
      "loss": 0.8418,
      "step": 274920
    },
    {
      "epoch": 4.826805245878615,
      "grad_norm": 0.056388847867907446,
      "learning_rate": 2.6098516695458667e-06,
      "loss": 0.8409,
      "step": 274930
    },
    {
      "epoch": 4.826980810758616,
      "grad_norm": 0.0486185356400351,
      "learning_rate": 2.6086172943384622e-06,
      "loss": 0.8349,
      "step": 274940
    },
    {
      "epoch": 4.8271563756386175,
      "grad_norm": 0.08346628289157866,
      "learning_rate": 2.607384165768745e-06,
      "loss": 0.8353,
      "step": 274950
    },
    {
      "epoch": 4.827331940518619,
      "grad_norm": 0.04779139073439817,
      "learning_rate": 2.606152283852322e-06,
      "loss": 0.8384,
      "step": 274960
    },
    {
      "epoch": 4.82750750539862,
      "grad_norm": 0.048852519393326965,
      "learning_rate": 2.6049216486048014e-06,
      "loss": 0.8377,
      "step": 274970
    },
    {
      "epoch": 4.8276830702786215,
      "grad_norm": 0.050257859554560774,
      "learning_rate": 2.6036922600417905e-06,
      "loss": 0.8372,
      "step": 274980
    },
    {
      "epoch": 4.827858635158623,
      "grad_norm": 0.055336681752509755,
      "learning_rate": 2.602464118178853e-06,
      "loss": 0.834,
      "step": 274990
    },
    {
      "epoch": 4.8280342000386245,
      "grad_norm": 0.05434678065255687,
      "learning_rate": 2.6012372230315518e-06,
      "loss": 0.8503,
      "step": 275000
    },
    {
      "epoch": 4.8282097649186255,
      "grad_norm": 0.052684484263078535,
      "learning_rate": 2.600011574615429e-06,
      "loss": 0.8405,
      "step": 275010
    },
    {
      "epoch": 4.8283853297986274,
      "grad_norm": 0.06521905691636634,
      "learning_rate": 2.5987871729460147e-06,
      "loss": 0.8384,
      "step": 275020
    },
    {
      "epoch": 4.8285608946786285,
      "grad_norm": 0.051640154956327164,
      "learning_rate": 2.5975640180388403e-06,
      "loss": 0.8392,
      "step": 275030
    },
    {
      "epoch": 4.8287364595586295,
      "grad_norm": 0.05885807236413016,
      "learning_rate": 2.5963421099093807e-06,
      "loss": 0.8363,
      "step": 275040
    },
    {
      "epoch": 4.828912024438631,
      "grad_norm": 0.05173920681681754,
      "learning_rate": 2.595121448573123e-06,
      "loss": 0.8408,
      "step": 275050
    },
    {
      "epoch": 4.8290875893186325,
      "grad_norm": 0.05597530773778599,
      "learning_rate": 2.5939020340455316e-06,
      "loss": 0.8411,
      "step": 275060
    },
    {
      "epoch": 4.829263154198634,
      "grad_norm": 0.05262749342030055,
      "learning_rate": 2.592683866342071e-06,
      "loss": 0.8348,
      "step": 275070
    },
    {
      "epoch": 4.829438719078635,
      "grad_norm": 0.054545607354975614,
      "learning_rate": 2.5914669454781623e-06,
      "loss": 0.8344,
      "step": 275080
    },
    {
      "epoch": 4.829614283958637,
      "grad_norm": 0.04718014786378896,
      "learning_rate": 2.5902512714692257e-06,
      "loss": 0.8359,
      "step": 275090
    },
    {
      "epoch": 4.829789848838638,
      "grad_norm": 0.05157151816232482,
      "learning_rate": 2.589036844330671e-06,
      "loss": 0.8404,
      "step": 275100
    },
    {
      "epoch": 4.829965413718639,
      "grad_norm": 0.050692196710622056,
      "learning_rate": 2.587823664077875e-06,
      "loss": 0.8367,
      "step": 275110
    },
    {
      "epoch": 4.830140978598641,
      "grad_norm": 0.05372257961884488,
      "learning_rate": 2.586611730726214e-06,
      "loss": 0.8305,
      "step": 275120
    },
    {
      "epoch": 4.830316543478642,
      "grad_norm": 0.048736657223681625,
      "learning_rate": 2.5854010442910437e-06,
      "loss": 0.8381,
      "step": 275130
    },
    {
      "epoch": 4.830492108358644,
      "grad_norm": 0.051002250305920815,
      "learning_rate": 2.584191604787707e-06,
      "loss": 0.8339,
      "step": 275140
    },
    {
      "epoch": 4.830667673238645,
      "grad_norm": 0.04422490906868055,
      "learning_rate": 2.5829834122315145e-06,
      "loss": 0.8325,
      "step": 275150
    },
    {
      "epoch": 4.830843238118646,
      "grad_norm": 0.04397252465971796,
      "learning_rate": 2.5817764666377887e-06,
      "loss": 0.8395,
      "step": 275160
    },
    {
      "epoch": 4.831018802998648,
      "grad_norm": 0.04628613705195507,
      "learning_rate": 2.5805707680218176e-06,
      "loss": 0.8349,
      "step": 275170
    },
    {
      "epoch": 4.831194367878649,
      "grad_norm": 0.049161076470542485,
      "learning_rate": 2.5793663163988793e-06,
      "loss": 0.837,
      "step": 275180
    },
    {
      "epoch": 4.831369932758651,
      "grad_norm": 0.05100449603308824,
      "learning_rate": 2.57816311178423e-06,
      "loss": 0.8404,
      "step": 275190
    },
    {
      "epoch": 4.831545497638652,
      "grad_norm": 0.06363703358697169,
      "learning_rate": 2.5769611541931137e-06,
      "loss": 0.8417,
      "step": 275200
    },
    {
      "epoch": 4.831721062518653,
      "grad_norm": 0.06872160612271697,
      "learning_rate": 2.575760443640765e-06,
      "loss": 0.8445,
      "step": 275210
    },
    {
      "epoch": 4.831896627398655,
      "grad_norm": 0.05203915890964668,
      "learning_rate": 2.5745609801423953e-06,
      "loss": 0.8417,
      "step": 275220
    },
    {
      "epoch": 4.832072192278656,
      "grad_norm": 0.05593835073480344,
      "learning_rate": 2.5733627637131945e-06,
      "loss": 0.8337,
      "step": 275230
    },
    {
      "epoch": 4.832247757158658,
      "grad_norm": 0.047085507357207824,
      "learning_rate": 2.5721657943683635e-06,
      "loss": 0.836,
      "step": 275240
    },
    {
      "epoch": 4.832423322038659,
      "grad_norm": 0.09524147808357304,
      "learning_rate": 2.570970072123048e-06,
      "loss": 0.8422,
      "step": 275250
    },
    {
      "epoch": 4.832598886918661,
      "grad_norm": 0.051942729066597956,
      "learning_rate": 2.569775596992394e-06,
      "loss": 0.8378,
      "step": 275260
    },
    {
      "epoch": 4.832774451798662,
      "grad_norm": 0.05412395162369996,
      "learning_rate": 2.5685823689915696e-06,
      "loss": 0.8395,
      "step": 275270
    },
    {
      "epoch": 4.832950016678664,
      "grad_norm": 0.05948476887440438,
      "learning_rate": 2.5673903881356545e-06,
      "loss": 0.8372,
      "step": 275280
    },
    {
      "epoch": 4.833125581558665,
      "grad_norm": 0.049804838660985826,
      "learning_rate": 2.5661996544397726e-06,
      "loss": 0.8388,
      "step": 275290
    },
    {
      "epoch": 4.833301146438666,
      "grad_norm": 0.05177429703471617,
      "learning_rate": 2.5650101679190042e-06,
      "loss": 0.841,
      "step": 275300
    },
    {
      "epoch": 4.833476711318668,
      "grad_norm": 0.04593440643607699,
      "learning_rate": 2.563821928588428e-06,
      "loss": 0.84,
      "step": 275310
    },
    {
      "epoch": 4.833652276198669,
      "grad_norm": 0.046546517254690646,
      "learning_rate": 2.562634936463093e-06,
      "loss": 0.8412,
      "step": 275320
    },
    {
      "epoch": 4.833827841078671,
      "grad_norm": 0.053945689611284466,
      "learning_rate": 2.5614491915580333e-06,
      "loss": 0.8361,
      "step": 275330
    },
    {
      "epoch": 4.834003405958672,
      "grad_norm": 0.04283580434020563,
      "learning_rate": 2.560264693888275e-06,
      "loss": 0.8395,
      "step": 275340
    },
    {
      "epoch": 4.834178970838673,
      "grad_norm": 0.05618254999070159,
      "learning_rate": 2.559081443468843e-06,
      "loss": 0.8389,
      "step": 275350
    },
    {
      "epoch": 4.834354535718675,
      "grad_norm": 0.06366411461407076,
      "learning_rate": 2.5578994403146954e-06,
      "loss": 0.8289,
      "step": 275360
    },
    {
      "epoch": 4.834530100598676,
      "grad_norm": 0.042819735730527664,
      "learning_rate": 2.5567186844408472e-06,
      "loss": 0.8399,
      "step": 275370
    },
    {
      "epoch": 4.834705665478678,
      "grad_norm": 0.03780578032069957,
      "learning_rate": 2.5555391758622243e-06,
      "loss": 0.8345,
      "step": 275380
    },
    {
      "epoch": 4.834881230358679,
      "grad_norm": 0.05182234473742897,
      "learning_rate": 2.5543609145937963e-06,
      "loss": 0.8348,
      "step": 275390
    },
    {
      "epoch": 4.83505679523868,
      "grad_norm": 0.10085801540654526,
      "learning_rate": 2.5531839006504785e-06,
      "loss": 0.8377,
      "step": 275400
    },
    {
      "epoch": 4.835232360118682,
      "grad_norm": 0.06209971587856139,
      "learning_rate": 2.5520081340471966e-06,
      "loss": 0.8398,
      "step": 275410
    },
    {
      "epoch": 4.835407924998683,
      "grad_norm": 0.05762555703290678,
      "learning_rate": 2.5508336147988338e-06,
      "loss": 0.8396,
      "step": 275420
    },
    {
      "epoch": 4.835583489878685,
      "grad_norm": 0.05130963281937535,
      "learning_rate": 2.549660342920271e-06,
      "loss": 0.8459,
      "step": 275430
    },
    {
      "epoch": 4.835759054758686,
      "grad_norm": 0.04909627370199849,
      "learning_rate": 2.5484883184263795e-06,
      "loss": 0.8368,
      "step": 275440
    },
    {
      "epoch": 4.835934619638687,
      "grad_norm": 0.053860178333573566,
      "learning_rate": 2.54731754133202e-06,
      "loss": 0.8356,
      "step": 275450
    },
    {
      "epoch": 4.836110184518689,
      "grad_norm": 0.04605035020565666,
      "learning_rate": 2.5461480116520085e-06,
      "loss": 0.8335,
      "step": 275460
    },
    {
      "epoch": 4.83628574939869,
      "grad_norm": 0.05883822193550608,
      "learning_rate": 2.5449797294011716e-06,
      "loss": 0.8397,
      "step": 275470
    },
    {
      "epoch": 4.836461314278692,
      "grad_norm": 0.05412926537854916,
      "learning_rate": 2.543812694594304e-06,
      "loss": 0.8415,
      "step": 275480
    },
    {
      "epoch": 4.836636879158693,
      "grad_norm": 0.04851425914999154,
      "learning_rate": 2.542646907246211e-06,
      "loss": 0.8478,
      "step": 275490
    },
    {
      "epoch": 4.836812444038695,
      "grad_norm": 0.04416477076971267,
      "learning_rate": 2.541482367371642e-06,
      "loss": 0.8451,
      "step": 275500
    },
    {
      "epoch": 4.836988008918696,
      "grad_norm": 0.04532828759384817,
      "learning_rate": 2.5403190749853585e-06,
      "loss": 0.8311,
      "step": 275510
    },
    {
      "epoch": 4.837163573798698,
      "grad_norm": 0.08048066112268504,
      "learning_rate": 2.539157030102111e-06,
      "loss": 0.8455,
      "step": 275520
    },
    {
      "epoch": 4.837339138678699,
      "grad_norm": 0.053524413827905466,
      "learning_rate": 2.5379962327366057e-06,
      "loss": 0.8289,
      "step": 275530
    },
    {
      "epoch": 4.8375147035587,
      "grad_norm": 0.04396071282071184,
      "learning_rate": 2.536836682903571e-06,
      "loss": 0.8291,
      "step": 275540
    },
    {
      "epoch": 4.837690268438702,
      "grad_norm": 0.06706345461420463,
      "learning_rate": 2.5356783806176688e-06,
      "loss": 0.8358,
      "step": 275550
    },
    {
      "epoch": 4.837865833318703,
      "grad_norm": 0.05665783822882606,
      "learning_rate": 2.534521325893605e-06,
      "loss": 0.8353,
      "step": 275560
    },
    {
      "epoch": 4.838041398198705,
      "grad_norm": 0.04556393566022263,
      "learning_rate": 2.5333655187460207e-06,
      "loss": 0.8456,
      "step": 275570
    },
    {
      "epoch": 4.838216963078706,
      "grad_norm": 0.05369280877628928,
      "learning_rate": 2.532210959189567e-06,
      "loss": 0.8404,
      "step": 275580
    },
    {
      "epoch": 4.838392527958707,
      "grad_norm": 0.044311803142480566,
      "learning_rate": 2.5310576472388726e-06,
      "loss": 0.8455,
      "step": 275590
    },
    {
      "epoch": 4.838568092838709,
      "grad_norm": 0.049509042018977246,
      "learning_rate": 2.5299055829085456e-06,
      "loss": 0.8465,
      "step": 275600
    },
    {
      "epoch": 4.83874365771871,
      "grad_norm": 0.04594018448852915,
      "learning_rate": 2.5287547662131925e-06,
      "loss": 0.8277,
      "step": 275610
    },
    {
      "epoch": 4.838919222598712,
      "grad_norm": 0.08187710357293193,
      "learning_rate": 2.5276051971673884e-06,
      "loss": 0.8315,
      "step": 275620
    },
    {
      "epoch": 4.839094787478713,
      "grad_norm": 0.04193027949336782,
      "learning_rate": 2.5264568757856966e-06,
      "loss": 0.8406,
      "step": 275630
    },
    {
      "epoch": 4.839270352358714,
      "grad_norm": 0.0473975199940962,
      "learning_rate": 2.5253098020826577e-06,
      "loss": 0.8347,
      "step": 275640
    },
    {
      "epoch": 4.839445917238716,
      "grad_norm": 0.04849877965430718,
      "learning_rate": 2.524163976072835e-06,
      "loss": 0.8357,
      "step": 275650
    },
    {
      "epoch": 4.839621482118717,
      "grad_norm": 0.06155250209652752,
      "learning_rate": 2.5230193977707163e-06,
      "loss": 0.8452,
      "step": 275660
    },
    {
      "epoch": 4.839797046998719,
      "grad_norm": 0.06199273650825144,
      "learning_rate": 2.5218760671908205e-06,
      "loss": 0.8327,
      "step": 275670
    },
    {
      "epoch": 4.83997261187872,
      "grad_norm": 0.057364288380320434,
      "learning_rate": 2.5207339843476218e-06,
      "loss": 0.8329,
      "step": 275680
    },
    {
      "epoch": 4.8401481767587216,
      "grad_norm": 0.14788286082051325,
      "learning_rate": 2.519593149255608e-06,
      "loss": 0.8383,
      "step": 275690
    },
    {
      "epoch": 4.840323741638723,
      "grad_norm": 0.05700793108994467,
      "learning_rate": 2.5184535619292097e-06,
      "loss": 0.8309,
      "step": 275700
    },
    {
      "epoch": 4.840499306518724,
      "grad_norm": 0.06555397919989249,
      "learning_rate": 2.5173152223828915e-06,
      "loss": 0.83,
      "step": 275710
    },
    {
      "epoch": 4.8406748713987255,
      "grad_norm": 0.07986507132894739,
      "learning_rate": 2.5161781306310525e-06,
      "loss": 0.844,
      "step": 275720
    },
    {
      "epoch": 4.840850436278727,
      "grad_norm": 0.05686283488626891,
      "learning_rate": 2.5150422866881126e-06,
      "loss": 0.8413,
      "step": 275730
    },
    {
      "epoch": 4.8410260011587285,
      "grad_norm": 0.05130869893717305,
      "learning_rate": 2.5139076905684594e-06,
      "loss": 0.8439,
      "step": 275740
    },
    {
      "epoch": 4.8412015660387295,
      "grad_norm": 0.07498377017035839,
      "learning_rate": 2.5127743422864803e-06,
      "loss": 0.839,
      "step": 275750
    },
    {
      "epoch": 4.8413771309187315,
      "grad_norm": 0.04660732178528223,
      "learning_rate": 2.5116422418565185e-06,
      "loss": 0.8362,
      "step": 275760
    },
    {
      "epoch": 4.8415526957987325,
      "grad_norm": 0.0458257915781942,
      "learning_rate": 2.510511389292919e-06,
      "loss": 0.8319,
      "step": 275770
    },
    {
      "epoch": 4.8417282606787335,
      "grad_norm": 0.04802905838148354,
      "learning_rate": 2.509381784610034e-06,
      "loss": 0.8324,
      "step": 275780
    },
    {
      "epoch": 4.841903825558735,
      "grad_norm": 0.051097637100227564,
      "learning_rate": 2.5082534278221432e-06,
      "loss": 0.8402,
      "step": 275790
    },
    {
      "epoch": 4.8420793904387365,
      "grad_norm": 0.048735907257290434,
      "learning_rate": 2.5071263189435562e-06,
      "loss": 0.8405,
      "step": 275800
    },
    {
      "epoch": 4.842254955318738,
      "grad_norm": 0.0505762838295391,
      "learning_rate": 2.5060004579885612e-06,
      "loss": 0.8317,
      "step": 275810
    },
    {
      "epoch": 4.842430520198739,
      "grad_norm": 0.055370647702656865,
      "learning_rate": 2.5048758449714146e-06,
      "loss": 0.8401,
      "step": 275820
    },
    {
      "epoch": 4.8426060850787405,
      "grad_norm": 0.044215602476360014,
      "learning_rate": 2.503752479906361e-06,
      "loss": 0.8407,
      "step": 275830
    },
    {
      "epoch": 4.842781649958742,
      "grad_norm": 0.04804937722503847,
      "learning_rate": 2.5026303628076444e-06,
      "loss": 0.8462,
      "step": 275840
    },
    {
      "epoch": 4.842957214838743,
      "grad_norm": 0.06136555097975872,
      "learning_rate": 2.5015094936894875e-06,
      "loss": 0.8439,
      "step": 275850
    },
    {
      "epoch": 4.843132779718745,
      "grad_norm": 0.07742245251607874,
      "learning_rate": 2.500389872566069e-06,
      "loss": 0.8375,
      "step": 275860
    },
    {
      "epoch": 4.843308344598746,
      "grad_norm": 0.05057122235690557,
      "learning_rate": 2.499271499451601e-06,
      "loss": 0.8366,
      "step": 275870
    },
    {
      "epoch": 4.843483909478747,
      "grad_norm": 0.04322844832150202,
      "learning_rate": 2.498154374360229e-06,
      "loss": 0.8274,
      "step": 275880
    },
    {
      "epoch": 4.843659474358749,
      "grad_norm": 0.052340456405622154,
      "learning_rate": 2.497038497306131e-06,
      "loss": 0.839,
      "step": 275890
    },
    {
      "epoch": 4.84383503923875,
      "grad_norm": 0.049263236388273256,
      "learning_rate": 2.495923868303421e-06,
      "loss": 0.8381,
      "step": 275900
    },
    {
      "epoch": 4.844010604118752,
      "grad_norm": 0.060230123523162926,
      "learning_rate": 2.4948104873662323e-06,
      "loss": 0.8306,
      "step": 275910
    },
    {
      "epoch": 4.844186168998753,
      "grad_norm": 0.058678017553847876,
      "learning_rate": 2.4936983545086896e-06,
      "loss": 0.8342,
      "step": 275920
    },
    {
      "epoch": 4.844361733878755,
      "grad_norm": 0.051773741856335785,
      "learning_rate": 2.49258746974485e-06,
      "loss": 0.8412,
      "step": 275930
    },
    {
      "epoch": 4.844537298758756,
      "grad_norm": 0.04648072228877954,
      "learning_rate": 2.491477833088816e-06,
      "loss": 0.8382,
      "step": 275940
    },
    {
      "epoch": 4.844712863638758,
      "grad_norm": 0.04901295136462109,
      "learning_rate": 2.490369444554633e-06,
      "loss": 0.8374,
      "step": 275950
    },
    {
      "epoch": 4.844888428518759,
      "grad_norm": 0.04329450378663491,
      "learning_rate": 2.4892623041563492e-06,
      "loss": 0.8298,
      "step": 275960
    },
    {
      "epoch": 4.84506399339876,
      "grad_norm": 0.05486242725083093,
      "learning_rate": 2.488156411907989e-06,
      "loss": 0.8337,
      "step": 275970
    },
    {
      "epoch": 4.845239558278762,
      "grad_norm": 0.058391220082592336,
      "learning_rate": 2.4870517678235764e-06,
      "loss": 0.8374,
      "step": 275980
    },
    {
      "epoch": 4.845415123158763,
      "grad_norm": 0.05159378883692771,
      "learning_rate": 2.4859483719170936e-06,
      "loss": 0.8426,
      "step": 275990
    },
    {
      "epoch": 4.845590688038765,
      "grad_norm": 0.05160197297247187,
      "learning_rate": 2.48484622420252e-06,
      "loss": 0.8404,
      "step": 276000
    },
    {
      "epoch": 4.845766252918766,
      "grad_norm": 0.06867158476020538,
      "learning_rate": 2.483745324693826e-06,
      "loss": 0.8373,
      "step": 276010
    },
    {
      "epoch": 4.845941817798767,
      "grad_norm": 0.06819101443023562,
      "learning_rate": 2.48264567340496e-06,
      "loss": 0.832,
      "step": 276020
    },
    {
      "epoch": 4.846117382678769,
      "grad_norm": 0.05318365951033614,
      "learning_rate": 2.481547270349858e-06,
      "loss": 0.8399,
      "step": 276030
    },
    {
      "epoch": 4.84629294755877,
      "grad_norm": 0.05632614185900767,
      "learning_rate": 2.4804501155424244e-06,
      "loss": 0.8324,
      "step": 276040
    },
    {
      "epoch": 4.846468512438772,
      "grad_norm": 0.04447132396113028,
      "learning_rate": 2.479354208996585e-06,
      "loss": 0.8397,
      "step": 276050
    },
    {
      "epoch": 4.846644077318773,
      "grad_norm": 0.05314569981349564,
      "learning_rate": 2.4782595507261995e-06,
      "loss": 0.8371,
      "step": 276060
    },
    {
      "epoch": 4.846819642198774,
      "grad_norm": 0.055679117474786245,
      "learning_rate": 2.477166140745139e-06,
      "loss": 0.8381,
      "step": 276070
    },
    {
      "epoch": 4.846995207078776,
      "grad_norm": 0.049484615961554505,
      "learning_rate": 2.4760739790672735e-06,
      "loss": 0.8379,
      "step": 276080
    },
    {
      "epoch": 4.847170771958777,
      "grad_norm": 0.07838242435434019,
      "learning_rate": 2.4749830657064317e-06,
      "loss": 0.8357,
      "step": 276090
    },
    {
      "epoch": 4.847346336838779,
      "grad_norm": 0.0532957241837975,
      "learning_rate": 2.473893400676439e-06,
      "loss": 0.8445,
      "step": 276100
    },
    {
      "epoch": 4.84752190171878,
      "grad_norm": 0.05089799263539613,
      "learning_rate": 2.4728049839911016e-06,
      "loss": 0.8379,
      "step": 276110
    },
    {
      "epoch": 4.847697466598781,
      "grad_norm": 0.05699995316333513,
      "learning_rate": 2.4717178156642012e-06,
      "loss": 0.8418,
      "step": 276120
    },
    {
      "epoch": 4.847873031478783,
      "grad_norm": 0.05349635477107371,
      "learning_rate": 2.4706318957095217e-06,
      "loss": 0.8333,
      "step": 276130
    },
    {
      "epoch": 4.848048596358784,
      "grad_norm": 0.04723675702953651,
      "learning_rate": 2.469547224140813e-06,
      "loss": 0.8364,
      "step": 276140
    },
    {
      "epoch": 4.848224161238786,
      "grad_norm": 0.055297224968775295,
      "learning_rate": 2.4684638009718353e-06,
      "loss": 0.8362,
      "step": 276150
    },
    {
      "epoch": 4.848399726118787,
      "grad_norm": 0.048493980133704095,
      "learning_rate": 2.4673816262162957e-06,
      "loss": 0.8377,
      "step": 276160
    },
    {
      "epoch": 4.848575290998789,
      "grad_norm": 0.04394980395573552,
      "learning_rate": 2.466300699887922e-06,
      "loss": 0.8433,
      "step": 276170
    },
    {
      "epoch": 4.84875085587879,
      "grad_norm": 0.042493998597317166,
      "learning_rate": 2.465221022000397e-06,
      "loss": 0.846,
      "step": 276180
    },
    {
      "epoch": 4.848926420758792,
      "grad_norm": 0.05819088024197446,
      "learning_rate": 2.464142592567407e-06,
      "loss": 0.8449,
      "step": 276190
    },
    {
      "epoch": 4.849101985638793,
      "grad_norm": 0.06545653627919384,
      "learning_rate": 2.463065411602612e-06,
      "loss": 0.8377,
      "step": 276200
    },
    {
      "epoch": 4.849277550518794,
      "grad_norm": 0.05491103641251533,
      "learning_rate": 2.461989479119664e-06,
      "loss": 0.8463,
      "step": 276210
    },
    {
      "epoch": 4.849453115398796,
      "grad_norm": 0.04656565088005356,
      "learning_rate": 2.4609147951322032e-06,
      "loss": 0.8364,
      "step": 276220
    },
    {
      "epoch": 4.849628680278797,
      "grad_norm": 0.04328329642344152,
      "learning_rate": 2.4598413596538255e-06,
      "loss": 0.845,
      "step": 276230
    },
    {
      "epoch": 4.849804245158799,
      "grad_norm": 0.050196473362790116,
      "learning_rate": 2.4587691726981483e-06,
      "loss": 0.8357,
      "step": 276240
    },
    {
      "epoch": 4.8499798100388,
      "grad_norm": 0.056717227431924006,
      "learning_rate": 2.4576982342787475e-06,
      "loss": 0.8281,
      "step": 276250
    },
    {
      "epoch": 4.850155374918801,
      "grad_norm": 0.058811752138621144,
      "learning_rate": 2.456628544409207e-06,
      "loss": 0.8309,
      "step": 276260
    },
    {
      "epoch": 4.850330939798803,
      "grad_norm": 0.0647470095609219,
      "learning_rate": 2.455560103103057e-06,
      "loss": 0.8344,
      "step": 276270
    },
    {
      "epoch": 4.850506504678804,
      "grad_norm": 0.06661353454187778,
      "learning_rate": 2.4544929103738387e-06,
      "loss": 0.8321,
      "step": 276280
    },
    {
      "epoch": 4.850682069558806,
      "grad_norm": 0.04373796452696412,
      "learning_rate": 2.4534269662351044e-06,
      "loss": 0.8323,
      "step": 276290
    },
    {
      "epoch": 4.850857634438807,
      "grad_norm": 0.06027810260059136,
      "learning_rate": 2.4523622707003185e-06,
      "loss": 0.8371,
      "step": 276300
    },
    {
      "epoch": 4.851033199318808,
      "grad_norm": 0.05859440555758353,
      "learning_rate": 2.451298823782989e-06,
      "loss": 0.8405,
      "step": 276310
    },
    {
      "epoch": 4.85120876419881,
      "grad_norm": 0.04331950437295324,
      "learning_rate": 2.450236625496602e-06,
      "loss": 0.8401,
      "step": 276320
    },
    {
      "epoch": 4.851384329078811,
      "grad_norm": 0.06369793618981957,
      "learning_rate": 2.4491756758546e-06,
      "loss": 0.8406,
      "step": 276330
    },
    {
      "epoch": 4.851559893958813,
      "grad_norm": 0.06843181043045812,
      "learning_rate": 2.448115974870425e-06,
      "loss": 0.8423,
      "step": 276340
    },
    {
      "epoch": 4.851735458838814,
      "grad_norm": 0.04344347243344211,
      "learning_rate": 2.4470575225575087e-06,
      "loss": 0.8414,
      "step": 276350
    },
    {
      "epoch": 4.851911023718816,
      "grad_norm": 0.0523883445168756,
      "learning_rate": 2.44600031892926e-06,
      "loss": 0.8377,
      "step": 276360
    },
    {
      "epoch": 4.852086588598817,
      "grad_norm": 0.052692592425111685,
      "learning_rate": 2.4449443639990772e-06,
      "loss": 0.8381,
      "step": 276370
    },
    {
      "epoch": 4.852262153478818,
      "grad_norm": 0.06527803053020698,
      "learning_rate": 2.4438896577803364e-06,
      "loss": 0.827,
      "step": 276380
    },
    {
      "epoch": 4.85243771835882,
      "grad_norm": 0.04824221271245805,
      "learning_rate": 2.4428362002864036e-06,
      "loss": 0.8393,
      "step": 276390
    },
    {
      "epoch": 4.852613283238821,
      "grad_norm": 0.06520037753436923,
      "learning_rate": 2.441783991530632e-06,
      "loss": 0.8438,
      "step": 276400
    },
    {
      "epoch": 4.852788848118823,
      "grad_norm": 0.04809592940544788,
      "learning_rate": 2.4407330315263334e-06,
      "loss": 0.8358,
      "step": 276410
    },
    {
      "epoch": 4.852964412998824,
      "grad_norm": 0.05387894258989316,
      "learning_rate": 2.439683320286839e-06,
      "loss": 0.8406,
      "step": 276420
    },
    {
      "epoch": 4.853139977878826,
      "grad_norm": 0.05940606072828876,
      "learning_rate": 2.4386348578254484e-06,
      "loss": 0.8429,
      "step": 276430
    },
    {
      "epoch": 4.853315542758827,
      "grad_norm": 0.05348443066623755,
      "learning_rate": 2.4375876441554397e-06,
      "loss": 0.8425,
      "step": 276440
    },
    {
      "epoch": 4.853491107638828,
      "grad_norm": 0.045909406474687366,
      "learning_rate": 2.436541679290089e-06,
      "loss": 0.8366,
      "step": 276450
    },
    {
      "epoch": 4.8536666725188296,
      "grad_norm": 0.043549743611623844,
      "learning_rate": 2.4354969632426414e-06,
      "loss": 0.8375,
      "step": 276460
    },
    {
      "epoch": 4.853842237398831,
      "grad_norm": 0.05512808942021891,
      "learning_rate": 2.4344534960263415e-06,
      "loss": 0.8495,
      "step": 276470
    },
    {
      "epoch": 4.8540178022788325,
      "grad_norm": 0.06407925986476999,
      "learning_rate": 2.4334112776544007e-06,
      "loss": 0.8399,
      "step": 276480
    },
    {
      "epoch": 4.8541933671588335,
      "grad_norm": 0.048968263130823385,
      "learning_rate": 2.43237030814003e-06,
      "loss": 0.8391,
      "step": 276490
    },
    {
      "epoch": 4.854368932038835,
      "grad_norm": 0.06542310584154047,
      "learning_rate": 2.4313305874964304e-06,
      "loss": 0.8378,
      "step": 276500
    },
    {
      "epoch": 4.8545444969188365,
      "grad_norm": 0.04884759788808477,
      "learning_rate": 2.430292115736747e-06,
      "loss": 0.8395,
      "step": 276510
    },
    {
      "epoch": 4.8547200617988375,
      "grad_norm": 0.05447029383285702,
      "learning_rate": 2.4292548928741584e-06,
      "loss": 0.8415,
      "step": 276520
    },
    {
      "epoch": 4.8548956266788394,
      "grad_norm": 0.05073164949252843,
      "learning_rate": 2.4282189189217985e-06,
      "loss": 0.8411,
      "step": 276530
    },
    {
      "epoch": 4.8550711915588405,
      "grad_norm": 0.05319537321030178,
      "learning_rate": 2.4271841938928028e-06,
      "loss": 0.8376,
      "step": 276540
    },
    {
      "epoch": 4.8552467564388415,
      "grad_norm": 0.05098083977395918,
      "learning_rate": 2.426150717800272e-06,
      "loss": 0.8371,
      "step": 276550
    },
    {
      "epoch": 4.855422321318843,
      "grad_norm": 0.049546691505777715,
      "learning_rate": 2.4251184906572974e-06,
      "loss": 0.8391,
      "step": 276560
    },
    {
      "epoch": 4.8555978861988445,
      "grad_norm": 0.04314729154365082,
      "learning_rate": 2.4240875124769687e-06,
      "loss": 0.8439,
      "step": 276570
    },
    {
      "epoch": 4.855773451078846,
      "grad_norm": 0.05368713802787919,
      "learning_rate": 2.4230577832723444e-06,
      "loss": 0.8334,
      "step": 276580
    },
    {
      "epoch": 4.855949015958847,
      "grad_norm": 0.054508763806073175,
      "learning_rate": 2.42202930305647e-06,
      "loss": 0.8364,
      "step": 276590
    },
    {
      "epoch": 4.856124580838849,
      "grad_norm": 0.05892810592262698,
      "learning_rate": 2.421002071842371e-06,
      "loss": 0.8326,
      "step": 276600
    },
    {
      "epoch": 4.85630014571885,
      "grad_norm": 0.056607946749304826,
      "learning_rate": 2.419976089643071e-06,
      "loss": 0.8336,
      "step": 276610
    },
    {
      "epoch": 4.856475710598852,
      "grad_norm": 0.05416060055113515,
      "learning_rate": 2.4189513564715735e-06,
      "loss": 0.8464,
      "step": 276620
    },
    {
      "epoch": 4.856651275478853,
      "grad_norm": 0.05159538645805233,
      "learning_rate": 2.4179278723408482e-06,
      "loss": 0.8494,
      "step": 276630
    },
    {
      "epoch": 4.856826840358854,
      "grad_norm": 0.04883583262373563,
      "learning_rate": 2.416905637263875e-06,
      "loss": 0.8307,
      "step": 276640
    },
    {
      "epoch": 4.857002405238856,
      "grad_norm": 0.06954892570127008,
      "learning_rate": 2.415884651253591e-06,
      "loss": 0.8387,
      "step": 276650
    },
    {
      "epoch": 4.857177970118857,
      "grad_norm": 0.06553701484895307,
      "learning_rate": 2.4148649143229545e-06,
      "loss": 0.8376,
      "step": 276660
    },
    {
      "epoch": 4.857353534998859,
      "grad_norm": 0.07778962064569517,
      "learning_rate": 2.4138464264848693e-06,
      "loss": 0.8422,
      "step": 276670
    },
    {
      "epoch": 4.85752909987886,
      "grad_norm": 0.057663401924703164,
      "learning_rate": 2.4128291877522395e-06,
      "loss": 0.8396,
      "step": 276680
    },
    {
      "epoch": 4.857704664758861,
      "grad_norm": 0.04161193910815434,
      "learning_rate": 2.4118131981379573e-06,
      "loss": 0.8371,
      "step": 276690
    },
    {
      "epoch": 4.857880229638863,
      "grad_norm": 0.0557763765595428,
      "learning_rate": 2.4107984576549044e-06,
      "loss": 0.8411,
      "step": 276700
    },
    {
      "epoch": 4.858055794518864,
      "grad_norm": 0.04779868596290374,
      "learning_rate": 2.409784966315919e-06,
      "loss": 0.8338,
      "step": 276710
    },
    {
      "epoch": 4.858231359398866,
      "grad_norm": 0.04418221175500474,
      "learning_rate": 2.4087727241338603e-06,
      "loss": 0.8322,
      "step": 276720
    },
    {
      "epoch": 4.858406924278867,
      "grad_norm": 0.058312347057849406,
      "learning_rate": 2.4077617311215444e-06,
      "loss": 0.836,
      "step": 276730
    },
    {
      "epoch": 4.858582489158868,
      "grad_norm": 0.04641820138388282,
      "learning_rate": 2.4067519872917867e-06,
      "loss": 0.8516,
      "step": 276740
    },
    {
      "epoch": 4.85875805403887,
      "grad_norm": 0.03865854099883858,
      "learning_rate": 2.4057434926573703e-06,
      "loss": 0.8435,
      "step": 276750
    },
    {
      "epoch": 4.858933618918871,
      "grad_norm": 0.05442460674651128,
      "learning_rate": 2.4047362472310777e-06,
      "loss": 0.8325,
      "step": 276760
    },
    {
      "epoch": 4.859109183798873,
      "grad_norm": 0.05994117320073124,
      "learning_rate": 2.403730251025681e-06,
      "loss": 0.8488,
      "step": 276770
    },
    {
      "epoch": 4.859284748678874,
      "grad_norm": 0.04911111664374036,
      "learning_rate": 2.402725504053908e-06,
      "loss": 0.8315,
      "step": 276780
    },
    {
      "epoch": 4.859460313558875,
      "grad_norm": 0.04960633629942295,
      "learning_rate": 2.401722006328497e-06,
      "loss": 0.8359,
      "step": 276790
    },
    {
      "epoch": 4.859635878438877,
      "grad_norm": 0.04760099398077638,
      "learning_rate": 2.4007197578621763e-06,
      "loss": 0.8355,
      "step": 276800
    },
    {
      "epoch": 4.859811443318878,
      "grad_norm": 0.0649158384603694,
      "learning_rate": 2.3997187586676294e-06,
      "loss": 0.8306,
      "step": 276810
    },
    {
      "epoch": 4.85998700819888,
      "grad_norm": 0.06024094945675965,
      "learning_rate": 2.398719008757541e-06,
      "loss": 0.8415,
      "step": 276820
    },
    {
      "epoch": 4.860162573078881,
      "grad_norm": 0.04982201705123357,
      "learning_rate": 2.397720508144572e-06,
      "loss": 0.8381,
      "step": 276830
    },
    {
      "epoch": 4.860338137958883,
      "grad_norm": 0.061975843147084354,
      "learning_rate": 2.3967232568413955e-06,
      "loss": 0.8438,
      "step": 276840
    },
    {
      "epoch": 4.860513702838884,
      "grad_norm": 0.0662055213992838,
      "learning_rate": 2.3957272548606297e-06,
      "loss": 0.8358,
      "step": 276850
    },
    {
      "epoch": 4.860689267718886,
      "grad_norm": 0.05001104044130969,
      "learning_rate": 2.394732502214892e-06,
      "loss": 0.833,
      "step": 276860
    },
    {
      "epoch": 4.860864832598887,
      "grad_norm": 0.05260662935296916,
      "learning_rate": 2.39373899891679e-06,
      "loss": 0.8372,
      "step": 276870
    },
    {
      "epoch": 4.861040397478888,
      "grad_norm": 0.053319419348396374,
      "learning_rate": 2.392746744978919e-06,
      "loss": 0.8479,
      "step": 276880
    },
    {
      "epoch": 4.86121596235889,
      "grad_norm": 0.06306468142262597,
      "learning_rate": 2.3917557404138316e-06,
      "loss": 0.8381,
      "step": 276890
    },
    {
      "epoch": 4.861391527238891,
      "grad_norm": 0.054155415269286154,
      "learning_rate": 2.390765985234112e-06,
      "loss": 0.8365,
      "step": 276900
    },
    {
      "epoch": 4.861567092118893,
      "grad_norm": 0.04555726628157994,
      "learning_rate": 2.38977747945228e-06,
      "loss": 0.8373,
      "step": 276910
    },
    {
      "epoch": 4.861742656998894,
      "grad_norm": 0.058150079038495416,
      "learning_rate": 2.3887902230808647e-06,
      "loss": 0.837,
      "step": 276920
    },
    {
      "epoch": 4.861918221878895,
      "grad_norm": 0.04850078884688074,
      "learning_rate": 2.3878042161323746e-06,
      "loss": 0.8325,
      "step": 276930
    },
    {
      "epoch": 4.862093786758897,
      "grad_norm": 0.04689012869985865,
      "learning_rate": 2.3868194586193067e-06,
      "loss": 0.8344,
      "step": 276940
    },
    {
      "epoch": 4.862269351638898,
      "grad_norm": 0.04887574303728333,
      "learning_rate": 2.385835950554136e-06,
      "loss": 0.8415,
      "step": 276950
    },
    {
      "epoch": 4.8624449165189,
      "grad_norm": 0.04966807185123525,
      "learning_rate": 2.3848536919493152e-06,
      "loss": 0.8276,
      "step": 276960
    },
    {
      "epoch": 4.862620481398901,
      "grad_norm": 0.05444379217321977,
      "learning_rate": 2.383872682817309e-06,
      "loss": 0.8428,
      "step": 276970
    },
    {
      "epoch": 4.862796046278902,
      "grad_norm": 0.06021040823578459,
      "learning_rate": 2.382892923170537e-06,
      "loss": 0.839,
      "step": 276980
    },
    {
      "epoch": 4.862971611158904,
      "grad_norm": 0.05416367209313438,
      "learning_rate": 2.381914413021397e-06,
      "loss": 0.8402,
      "step": 276990
    },
    {
      "epoch": 4.863147176038905,
      "grad_norm": 0.043157651613715986,
      "learning_rate": 2.3809371523823096e-06,
      "loss": 0.8306,
      "step": 277000
    },
    {
      "epoch": 4.863322740918907,
      "grad_norm": 0.0511080186433846,
      "learning_rate": 2.3799611412656508e-06,
      "loss": 0.8352,
      "step": 277010
    },
    {
      "epoch": 4.863498305798908,
      "grad_norm": 0.050029273164310425,
      "learning_rate": 2.378986379683786e-06,
      "loss": 0.8412,
      "step": 277020
    },
    {
      "epoch": 4.86367387067891,
      "grad_norm": 0.05285014175964713,
      "learning_rate": 2.3780128676490573e-06,
      "loss": 0.8438,
      "step": 277030
    },
    {
      "epoch": 4.863849435558911,
      "grad_norm": 0.045815316761178476,
      "learning_rate": 2.3770406051738197e-06,
      "loss": 0.8333,
      "step": 277040
    },
    {
      "epoch": 4.864025000438912,
      "grad_norm": 0.056458357979612246,
      "learning_rate": 2.3760695922703727e-06,
      "loss": 0.835,
      "step": 277050
    },
    {
      "epoch": 4.864200565318914,
      "grad_norm": 0.05373996469512881,
      "learning_rate": 2.375099828951014e-06,
      "loss": 0.8388,
      "step": 277060
    },
    {
      "epoch": 4.864376130198915,
      "grad_norm": 0.052202125085458864,
      "learning_rate": 2.374131315228055e-06,
      "loss": 0.8431,
      "step": 277070
    },
    {
      "epoch": 4.864551695078917,
      "grad_norm": 0.06248961676193048,
      "learning_rate": 2.3731640511137507e-06,
      "loss": 0.8316,
      "step": 277080
    },
    {
      "epoch": 4.864727259958918,
      "grad_norm": 0.06417479145541581,
      "learning_rate": 2.3721980366203563e-06,
      "loss": 0.8343,
      "step": 277090
    },
    {
      "epoch": 4.86490282483892,
      "grad_norm": 0.05279870527726918,
      "learning_rate": 2.3712332717601154e-06,
      "loss": 0.8334,
      "step": 277100
    },
    {
      "epoch": 4.865078389718921,
      "grad_norm": 0.05974434683991479,
      "learning_rate": 2.370269756545251e-06,
      "loss": 0.8389,
      "step": 277110
    },
    {
      "epoch": 4.865253954598922,
      "grad_norm": 0.04615149423052299,
      "learning_rate": 2.369307490987974e-06,
      "loss": 0.8435,
      "step": 277120
    },
    {
      "epoch": 4.865429519478924,
      "grad_norm": 0.049617499921966346,
      "learning_rate": 2.3683464751004627e-06,
      "loss": 0.8342,
      "step": 277130
    },
    {
      "epoch": 4.865605084358925,
      "grad_norm": 0.0760057223311987,
      "learning_rate": 2.367386708894917e-06,
      "loss": 0.836,
      "step": 277140
    },
    {
      "epoch": 4.865780649238927,
      "grad_norm": 0.04984481165671735,
      "learning_rate": 2.366428192383483e-06,
      "loss": 0.8417,
      "step": 277150
    },
    {
      "epoch": 4.865956214118928,
      "grad_norm": 0.040523243208213815,
      "learning_rate": 2.3654709255783053e-06,
      "loss": 0.8451,
      "step": 277160
    },
    {
      "epoch": 4.866131778998929,
      "grad_norm": 0.04736242142151796,
      "learning_rate": 2.364514908491507e-06,
      "loss": 0.8349,
      "step": 277170
    },
    {
      "epoch": 4.866307343878931,
      "grad_norm": 0.04034271552790983,
      "learning_rate": 2.363560141135223e-06,
      "loss": 0.8356,
      "step": 277180
    },
    {
      "epoch": 4.866482908758932,
      "grad_norm": 0.052032715564242805,
      "learning_rate": 2.3626066235215216e-06,
      "loss": 0.8378,
      "step": 277190
    },
    {
      "epoch": 4.866658473638934,
      "grad_norm": 0.05369074275971313,
      "learning_rate": 2.3616543556625037e-06,
      "loss": 0.8416,
      "step": 277200
    },
    {
      "epoch": 4.866834038518935,
      "grad_norm": 0.04855708869916052,
      "learning_rate": 2.360703337570238e-06,
      "loss": 0.8371,
      "step": 277210
    },
    {
      "epoch": 4.867009603398936,
      "grad_norm": 0.04900107812927804,
      "learning_rate": 2.3597535692567487e-06,
      "loss": 0.8455,
      "step": 277220
    },
    {
      "epoch": 4.8671851682789375,
      "grad_norm": 0.06743829085559142,
      "learning_rate": 2.3588050507340935e-06,
      "loss": 0.84,
      "step": 277230
    },
    {
      "epoch": 4.867360733158939,
      "grad_norm": 0.04454336247087602,
      "learning_rate": 2.357857782014285e-06,
      "loss": 0.8295,
      "step": 277240
    },
    {
      "epoch": 4.8675362980389405,
      "grad_norm": 0.045380465725132275,
      "learning_rate": 2.3569117631093154e-06,
      "loss": 0.8413,
      "step": 277250
    },
    {
      "epoch": 4.8677118629189415,
      "grad_norm": 0.04949551295321744,
      "learning_rate": 2.355966994031187e-06,
      "loss": 0.8429,
      "step": 277260
    },
    {
      "epoch": 4.8678874277989435,
      "grad_norm": 0.047766221560478966,
      "learning_rate": 2.3550234747918578e-06,
      "loss": 0.8416,
      "step": 277270
    },
    {
      "epoch": 4.8680629926789445,
      "grad_norm": 0.0452426218785097,
      "learning_rate": 2.3540812054032863e-06,
      "loss": 0.835,
      "step": 277280
    },
    {
      "epoch": 4.8682385575589455,
      "grad_norm": 0.042602584699881214,
      "learning_rate": 2.3531401858774206e-06,
      "loss": 0.8444,
      "step": 277290
    },
    {
      "epoch": 4.8684141224389474,
      "grad_norm": 0.04368750948059472,
      "learning_rate": 2.3522004162261637e-06,
      "loss": 0.8325,
      "step": 277300
    },
    {
      "epoch": 4.8685896873189485,
      "grad_norm": 0.05845885531055001,
      "learning_rate": 2.3512618964614298e-06,
      "loss": 0.8348,
      "step": 277310
    },
    {
      "epoch": 4.86876525219895,
      "grad_norm": 0.04792462454968859,
      "learning_rate": 2.3503246265951227e-06,
      "loss": 0.8375,
      "step": 277320
    },
    {
      "epoch": 4.868940817078951,
      "grad_norm": 0.052606405786951436,
      "learning_rate": 2.349388606639102e-06,
      "loss": 0.8417,
      "step": 277330
    },
    {
      "epoch": 4.869116381958953,
      "grad_norm": 0.048216316734338495,
      "learning_rate": 2.348453836605238e-06,
      "loss": 0.8291,
      "step": 277340
    },
    {
      "epoch": 4.869291946838954,
      "grad_norm": 0.06070337181364155,
      "learning_rate": 2.3475203165053792e-06,
      "loss": 0.8361,
      "step": 277350
    },
    {
      "epoch": 4.869467511718955,
      "grad_norm": 0.06353469194366601,
      "learning_rate": 2.3465880463513304e-06,
      "loss": 0.8403,
      "step": 277360
    },
    {
      "epoch": 4.869643076598957,
      "grad_norm": 0.056728655461388784,
      "learning_rate": 2.3456570261549177e-06,
      "loss": 0.8324,
      "step": 277370
    },
    {
      "epoch": 4.869818641478958,
      "grad_norm": 0.04684084792774294,
      "learning_rate": 2.3447272559279462e-06,
      "loss": 0.839,
      "step": 277380
    },
    {
      "epoch": 4.86999420635896,
      "grad_norm": 0.05376680040491101,
      "learning_rate": 2.3437987356821877e-06,
      "loss": 0.8422,
      "step": 277390
    },
    {
      "epoch": 4.870169771238961,
      "grad_norm": 0.05499766185130584,
      "learning_rate": 2.3428714654294014e-06,
      "loss": 0.8354,
      "step": 277400
    },
    {
      "epoch": 4.870345336118962,
      "grad_norm": 0.060098283066076286,
      "learning_rate": 2.3419454451813387e-06,
      "loss": 0.8336,
      "step": 277410
    },
    {
      "epoch": 4.870520900998964,
      "grad_norm": 0.04528530433418624,
      "learning_rate": 2.341020674949748e-06,
      "loss": 0.842,
      "step": 277420
    },
    {
      "epoch": 4.870696465878965,
      "grad_norm": 0.04518100583161324,
      "learning_rate": 2.340097154746313e-06,
      "loss": 0.8308,
      "step": 277430
    },
    {
      "epoch": 4.870872030758967,
      "grad_norm": 0.06289203051735735,
      "learning_rate": 2.339174884582773e-06,
      "loss": 0.8405,
      "step": 277440
    },
    {
      "epoch": 4.871047595638968,
      "grad_norm": 0.056291950914469535,
      "learning_rate": 2.338253864470789e-06,
      "loss": 0.8417,
      "step": 277450
    },
    {
      "epoch": 4.871223160518969,
      "grad_norm": 0.043895817237339485,
      "learning_rate": 2.3373340944220336e-06,
      "loss": 0.8363,
      "step": 277460
    },
    {
      "epoch": 4.871398725398971,
      "grad_norm": 0.04954573930877574,
      "learning_rate": 2.3364155744481695e-06,
      "loss": 0.8372,
      "step": 277470
    },
    {
      "epoch": 4.871574290278972,
      "grad_norm": 0.06686163789769836,
      "learning_rate": 2.3354983045608244e-06,
      "loss": 0.8471,
      "step": 277480
    },
    {
      "epoch": 4.871749855158974,
      "grad_norm": 0.05785954781628937,
      "learning_rate": 2.3345822847716273e-06,
      "loss": 0.8401,
      "step": 277490
    },
    {
      "epoch": 4.871925420038975,
      "grad_norm": 0.06631210507120402,
      "learning_rate": 2.333667515092185e-06,
      "loss": 0.831,
      "step": 277500
    },
    {
      "epoch": 4.872100984918977,
      "grad_norm": 0.04417037874497076,
      "learning_rate": 2.3327539955340712e-06,
      "loss": 0.8348,
      "step": 277510
    },
    {
      "epoch": 4.872276549798978,
      "grad_norm": 0.05113445704410173,
      "learning_rate": 2.3318417261088922e-06,
      "loss": 0.8405,
      "step": 277520
    },
    {
      "epoch": 4.87245211467898,
      "grad_norm": 0.04881209013074881,
      "learning_rate": 2.3309307068281786e-06,
      "loss": 0.8352,
      "step": 277530
    },
    {
      "epoch": 4.872627679558981,
      "grad_norm": 0.05819976046334759,
      "learning_rate": 2.3300209377034814e-06,
      "loss": 0.8274,
      "step": 277540
    },
    {
      "epoch": 4.872803244438982,
      "grad_norm": 0.0570634228875145,
      "learning_rate": 2.3291124187463307e-06,
      "loss": 0.8372,
      "step": 277550
    },
    {
      "epoch": 4.872978809318984,
      "grad_norm": 0.05175751448350101,
      "learning_rate": 2.3282051499682344e-06,
      "loss": 0.8368,
      "step": 277560
    },
    {
      "epoch": 4.873154374198985,
      "grad_norm": 0.05402693145926012,
      "learning_rate": 2.3272991313806893e-06,
      "loss": 0.8353,
      "step": 277570
    },
    {
      "epoch": 4.873329939078987,
      "grad_norm": 0.059394982057102745,
      "learning_rate": 2.326394362995181e-06,
      "loss": 0.8454,
      "step": 277580
    },
    {
      "epoch": 4.873505503958988,
      "grad_norm": 0.043150555593406865,
      "learning_rate": 2.3254908448231625e-06,
      "loss": 0.8386,
      "step": 277590
    },
    {
      "epoch": 4.873681068838989,
      "grad_norm": 0.054015204574329755,
      "learning_rate": 2.3245885768760864e-06,
      "loss": 0.8336,
      "step": 277600
    },
    {
      "epoch": 4.873856633718991,
      "grad_norm": 0.04318843061781092,
      "learning_rate": 2.3236875591653834e-06,
      "loss": 0.8249,
      "step": 277610
    },
    {
      "epoch": 4.874032198598992,
      "grad_norm": 0.058468849722661635,
      "learning_rate": 2.3227877917024744e-06,
      "loss": 0.8428,
      "step": 277620
    },
    {
      "epoch": 4.874207763478994,
      "grad_norm": 0.0663196693819406,
      "learning_rate": 2.3218892744987453e-06,
      "loss": 0.8384,
      "step": 277630
    },
    {
      "epoch": 4.874383328358995,
      "grad_norm": 0.04552678346590824,
      "learning_rate": 2.320992007565594e-06,
      "loss": 0.8404,
      "step": 277640
    },
    {
      "epoch": 4.874558893238996,
      "grad_norm": 0.04788939158724716,
      "learning_rate": 2.320095990914397e-06,
      "loss": 0.8316,
      "step": 277650
    },
    {
      "epoch": 4.874734458118998,
      "grad_norm": 0.08029834139414921,
      "learning_rate": 2.3192012245564858e-06,
      "loss": 0.8391,
      "step": 277660
    },
    {
      "epoch": 4.874910022998999,
      "grad_norm": 0.0489498986627386,
      "learning_rate": 2.3183077085032035e-06,
      "loss": 0.8396,
      "step": 277670
    },
    {
      "epoch": 4.875085587879001,
      "grad_norm": 0.05406835086921593,
      "learning_rate": 2.3174154427658816e-06,
      "loss": 0.8328,
      "step": 277680
    },
    {
      "epoch": 4.875261152759002,
      "grad_norm": 0.049601886440406244,
      "learning_rate": 2.31652442735582e-06,
      "loss": 0.8414,
      "step": 277690
    },
    {
      "epoch": 4.875436717639004,
      "grad_norm": 0.0588710816333404,
      "learning_rate": 2.315634662284306e-06,
      "loss": 0.8343,
      "step": 277700
    },
    {
      "epoch": 4.875612282519005,
      "grad_norm": 0.05438950186616897,
      "learning_rate": 2.3147461475626055e-06,
      "loss": 0.8367,
      "step": 277710
    },
    {
      "epoch": 4.875787847399006,
      "grad_norm": 0.0483107067094726,
      "learning_rate": 2.3138588832019954e-06,
      "loss": 0.8356,
      "step": 277720
    },
    {
      "epoch": 4.875963412279008,
      "grad_norm": 0.0620369504572558,
      "learning_rate": 2.3129728692136982e-06,
      "loss": 0.8379,
      "step": 277730
    },
    {
      "epoch": 4.876138977159009,
      "grad_norm": 0.061341783916448143,
      "learning_rate": 2.312088105608947e-06,
      "loss": 0.8396,
      "step": 277740
    },
    {
      "epoch": 4.876314542039011,
      "grad_norm": 0.0554070363706951,
      "learning_rate": 2.3112045923989522e-06,
      "loss": 0.832,
      "step": 277750
    },
    {
      "epoch": 4.876490106919012,
      "grad_norm": 0.05895349014674556,
      "learning_rate": 2.3103223295949146e-06,
      "loss": 0.8434,
      "step": 277760
    },
    {
      "epoch": 4.876665671799014,
      "grad_norm": 0.0571989668218448,
      "learning_rate": 2.309441317208011e-06,
      "loss": 0.8372,
      "step": 277770
    },
    {
      "epoch": 4.876841236679015,
      "grad_norm": 0.06320315504766369,
      "learning_rate": 2.3085615552493883e-06,
      "loss": 0.8396,
      "step": 277780
    },
    {
      "epoch": 4.877016801559016,
      "grad_norm": 0.057099118963570825,
      "learning_rate": 2.307683043730212e-06,
      "loss": 0.8409,
      "step": 277790
    },
    {
      "epoch": 4.877192366439018,
      "grad_norm": 0.08663020941453886,
      "learning_rate": 2.3068057826616057e-06,
      "loss": 0.8387,
      "step": 277800
    },
    {
      "epoch": 4.877367931319019,
      "grad_norm": 0.04515207942027647,
      "learning_rate": 2.305929772054682e-06,
      "loss": 0.8394,
      "step": 277810
    },
    {
      "epoch": 4.877543496199021,
      "grad_norm": 0.05229526692854949,
      "learning_rate": 2.305055011920541e-06,
      "loss": 0.8417,
      "step": 277820
    },
    {
      "epoch": 4.877719061079022,
      "grad_norm": 0.05088786043337373,
      "learning_rate": 2.3041815022702746e-06,
      "loss": 0.834,
      "step": 277830
    },
    {
      "epoch": 4.877894625959023,
      "grad_norm": 0.052798702070406524,
      "learning_rate": 2.303309243114938e-06,
      "loss": 0.8394,
      "step": 277840
    },
    {
      "epoch": 4.878070190839025,
      "grad_norm": 0.04268913493797968,
      "learning_rate": 2.302438234465589e-06,
      "loss": 0.842,
      "step": 277850
    },
    {
      "epoch": 4.878245755719026,
      "grad_norm": 0.04385788511983237,
      "learning_rate": 2.301568476333263e-06,
      "loss": 0.8315,
      "step": 277860
    },
    {
      "epoch": 4.878421320599028,
      "grad_norm": 0.049231253225883904,
      "learning_rate": 2.3006999687289842e-06,
      "loss": 0.8418,
      "step": 277870
    },
    {
      "epoch": 4.878596885479029,
      "grad_norm": 0.05242012726370273,
      "learning_rate": 2.299832711663744e-06,
      "loss": 0.837,
      "step": 277880
    },
    {
      "epoch": 4.87877245035903,
      "grad_norm": 0.17656638631343696,
      "learning_rate": 2.2989667051485553e-06,
      "loss": 0.8383,
      "step": 277890
    },
    {
      "epoch": 4.878948015239032,
      "grad_norm": 0.043898826029743675,
      "learning_rate": 2.298101949194354e-06,
      "loss": 0.8334,
      "step": 277900
    },
    {
      "epoch": 4.879123580119033,
      "grad_norm": 0.04843320078922225,
      "learning_rate": 2.297238443812132e-06,
      "loss": 0.8471,
      "step": 277910
    },
    {
      "epoch": 4.879299144999035,
      "grad_norm": 0.04344077113600318,
      "learning_rate": 2.296376189012814e-06,
      "loss": 0.8436,
      "step": 277920
    },
    {
      "epoch": 4.879474709879036,
      "grad_norm": 0.052921927643427784,
      "learning_rate": 2.2955151848073146e-06,
      "loss": 0.8303,
      "step": 277930
    },
    {
      "epoch": 4.879650274759038,
      "grad_norm": 0.04492382557362727,
      "learning_rate": 2.294655431206559e-06,
      "loss": 0.8482,
      "step": 277940
    },
    {
      "epoch": 4.879825839639039,
      "grad_norm": 0.05489876132898841,
      "learning_rate": 2.2937969282214392e-06,
      "loss": 0.8389,
      "step": 277950
    },
    {
      "epoch": 4.88000140451904,
      "grad_norm": 0.07227579802417675,
      "learning_rate": 2.2929396758628374e-06,
      "loss": 0.8351,
      "step": 277960
    },
    {
      "epoch": 4.8801769693990416,
      "grad_norm": 0.0633064561175574,
      "learning_rate": 2.29208367414159e-06,
      "loss": 0.842,
      "step": 277970
    },
    {
      "epoch": 4.880352534279043,
      "grad_norm": 0.054815895606689244,
      "learning_rate": 2.291228923068567e-06,
      "loss": 0.8355,
      "step": 277980
    },
    {
      "epoch": 4.8805280991590445,
      "grad_norm": 0.06498866483678091,
      "learning_rate": 2.2903754226545957e-06,
      "loss": 0.8357,
      "step": 277990
    },
    {
      "epoch": 4.8807036640390455,
      "grad_norm": 0.055785782429241364,
      "learning_rate": 2.2895231729104804e-06,
      "loss": 0.8397,
      "step": 278000
    },
    {
      "epoch": 4.8808792289190475,
      "grad_norm": 0.04100923565122837,
      "learning_rate": 2.288672173847025e-06,
      "loss": 0.84,
      "step": 278010
    },
    {
      "epoch": 4.8810547937990485,
      "grad_norm": 0.05900372105573216,
      "learning_rate": 2.2878224254750117e-06,
      "loss": 0.8418,
      "step": 278020
    },
    {
      "epoch": 4.8812303586790495,
      "grad_norm": 0.09698479942102946,
      "learning_rate": 2.2869739278052122e-06,
      "loss": 0.8322,
      "step": 278030
    },
    {
      "epoch": 4.8814059235590515,
      "grad_norm": 0.040062980763133214,
      "learning_rate": 2.2861266808483656e-06,
      "loss": 0.8499,
      "step": 278040
    },
    {
      "epoch": 4.8815814884390525,
      "grad_norm": 0.04289182117816468,
      "learning_rate": 2.2852806846152087e-06,
      "loss": 0.8358,
      "step": 278050
    },
    {
      "epoch": 4.881757053319054,
      "grad_norm": 0.04731217538521671,
      "learning_rate": 2.2844359391164706e-06,
      "loss": 0.8471,
      "step": 278060
    },
    {
      "epoch": 4.881932618199055,
      "grad_norm": 0.044729003322457925,
      "learning_rate": 2.283592444362844e-06,
      "loss": 0.8376,
      "step": 278070
    },
    {
      "epoch": 4.8821081830790565,
      "grad_norm": 0.06041796705566438,
      "learning_rate": 2.282750200365025e-06,
      "loss": 0.8501,
      "step": 278080
    },
    {
      "epoch": 4.882283747959058,
      "grad_norm": 0.05366805393258608,
      "learning_rate": 2.2819092071336747e-06,
      "loss": 0.8335,
      "step": 278090
    },
    {
      "epoch": 4.882459312839059,
      "grad_norm": 0.05063609869302914,
      "learning_rate": 2.2810694646794537e-06,
      "loss": 0.8377,
      "step": 278100
    },
    {
      "epoch": 4.882634877719061,
      "grad_norm": 0.04158450008760223,
      "learning_rate": 2.280230973013002e-06,
      "loss": 0.8449,
      "step": 278110
    },
    {
      "epoch": 4.882810442599062,
      "grad_norm": 0.04754579419919649,
      "learning_rate": 2.279393732144949e-06,
      "loss": 0.8359,
      "step": 278120
    },
    {
      "epoch": 4.882986007479063,
      "grad_norm": 0.055576526504273645,
      "learning_rate": 2.2785577420858886e-06,
      "loss": 0.8345,
      "step": 278130
    },
    {
      "epoch": 4.883161572359065,
      "grad_norm": 0.04848045112989425,
      "learning_rate": 2.277723002846428e-06,
      "loss": 0.8482,
      "step": 278140
    },
    {
      "epoch": 4.883337137239066,
      "grad_norm": 0.047667219301417825,
      "learning_rate": 2.27688951443713e-06,
      "loss": 0.8351,
      "step": 278150
    },
    {
      "epoch": 4.883512702119068,
      "grad_norm": 0.0482325798303706,
      "learning_rate": 2.2760572768685685e-06,
      "loss": 0.8289,
      "step": 278160
    },
    {
      "epoch": 4.883688266999069,
      "grad_norm": 0.06902302981277714,
      "learning_rate": 2.2752262901512827e-06,
      "loss": 0.8427,
      "step": 278170
    },
    {
      "epoch": 4.883863831879071,
      "grad_norm": 0.04864813090620879,
      "learning_rate": 2.2743965542957926e-06,
      "loss": 0.8385,
      "step": 278180
    },
    {
      "epoch": 4.884039396759072,
      "grad_norm": 0.049049771387816105,
      "learning_rate": 2.2735680693126274e-06,
      "loss": 0.8337,
      "step": 278190
    },
    {
      "epoch": 4.884214961639074,
      "grad_norm": 0.06349549147189137,
      "learning_rate": 2.2727408352122722e-06,
      "loss": 0.8404,
      "step": 278200
    },
    {
      "epoch": 4.884390526519075,
      "grad_norm": 0.047563218056591396,
      "learning_rate": 2.2719148520052134e-06,
      "loss": 0.8395,
      "step": 278210
    },
    {
      "epoch": 4.884566091399076,
      "grad_norm": 0.047378611869736714,
      "learning_rate": 2.2710901197019138e-06,
      "loss": 0.8392,
      "step": 278220
    },
    {
      "epoch": 4.884741656279078,
      "grad_norm": 0.052731741955891684,
      "learning_rate": 2.270266638312826e-06,
      "loss": 0.8385,
      "step": 278230
    },
    {
      "epoch": 4.884917221159079,
      "grad_norm": 0.05264190349071234,
      "learning_rate": 2.269444407848371e-06,
      "loss": 0.8425,
      "step": 278240
    },
    {
      "epoch": 4.885092786039081,
      "grad_norm": 0.058138778176642224,
      "learning_rate": 2.2686234283189884e-06,
      "loss": 0.8425,
      "step": 278250
    },
    {
      "epoch": 4.885268350919082,
      "grad_norm": 0.044505415544785765,
      "learning_rate": 2.267803699735067e-06,
      "loss": 0.8388,
      "step": 278260
    },
    {
      "epoch": 4.885443915799083,
      "grad_norm": 0.05650075823860792,
      "learning_rate": 2.266985222106992e-06,
      "loss": 0.8374,
      "step": 278270
    },
    {
      "epoch": 4.885619480679085,
      "grad_norm": 0.06539298645253816,
      "learning_rate": 2.2661679954451286e-06,
      "loss": 0.8345,
      "step": 278280
    },
    {
      "epoch": 4.885795045559086,
      "grad_norm": 0.054282053828018983,
      "learning_rate": 2.265352019759853e-06,
      "loss": 0.8369,
      "step": 278290
    },
    {
      "epoch": 4.885970610439088,
      "grad_norm": 0.056573522906115475,
      "learning_rate": 2.264537295061485e-06,
      "loss": 0.843,
      "step": 278300
    },
    {
      "epoch": 4.886146175319089,
      "grad_norm": 0.06748099569611775,
      "learning_rate": 2.263723821360346e-06,
      "loss": 0.8414,
      "step": 278310
    },
    {
      "epoch": 4.88632174019909,
      "grad_norm": 0.04400455933173876,
      "learning_rate": 2.2629115986667567e-06,
      "loss": 0.8343,
      "step": 278320
    },
    {
      "epoch": 4.886497305079092,
      "grad_norm": 0.054460474371262564,
      "learning_rate": 2.2621006269909937e-06,
      "loss": 0.8379,
      "step": 278330
    },
    {
      "epoch": 4.886672869959093,
      "grad_norm": 0.04205758264410936,
      "learning_rate": 2.2612909063433445e-06,
      "loss": 0.8373,
      "step": 278340
    },
    {
      "epoch": 4.886848434839095,
      "grad_norm": 0.055217750565734434,
      "learning_rate": 2.260482436734064e-06,
      "loss": 0.8409,
      "step": 278350
    },
    {
      "epoch": 4.887023999719096,
      "grad_norm": 0.06307567997969087,
      "learning_rate": 2.2596752181733858e-06,
      "loss": 0.8369,
      "step": 278360
    },
    {
      "epoch": 4.887199564599097,
      "grad_norm": 0.07310672063884452,
      "learning_rate": 2.258869250671553e-06,
      "loss": 0.8259,
      "step": 278370
    },
    {
      "epoch": 4.887375129479099,
      "grad_norm": 0.06162925353423119,
      "learning_rate": 2.258064534238765e-06,
      "loss": 0.8334,
      "step": 278380
    },
    {
      "epoch": 4.8875506943591,
      "grad_norm": 0.05127352227780282,
      "learning_rate": 2.257261068885234e-06,
      "loss": 0.8472,
      "step": 278390
    },
    {
      "epoch": 4.887726259239102,
      "grad_norm": 0.04409399464831387,
      "learning_rate": 2.2564588546211144e-06,
      "loss": 0.8351,
      "step": 278400
    },
    {
      "epoch": 4.887901824119103,
      "grad_norm": 0.050159388711853374,
      "learning_rate": 2.2556578914565957e-06,
      "loss": 0.8359,
      "step": 278410
    },
    {
      "epoch": 4.888077388999105,
      "grad_norm": 0.045226167319418295,
      "learning_rate": 2.2548581794018123e-06,
      "loss": 0.8391,
      "step": 278420
    },
    {
      "epoch": 4.888252953879106,
      "grad_norm": 0.050429314250120115,
      "learning_rate": 2.2540597184669086e-06,
      "loss": 0.8391,
      "step": 278430
    },
    {
      "epoch": 4.888428518759108,
      "grad_norm": 0.052557606607808734,
      "learning_rate": 2.2532625086619745e-06,
      "loss": 0.8409,
      "step": 278440
    },
    {
      "epoch": 4.888604083639109,
      "grad_norm": 0.04599643086267006,
      "learning_rate": 2.2524665499971434e-06,
      "loss": 0.8411,
      "step": 278450
    },
    {
      "epoch": 4.88877964851911,
      "grad_norm": 0.09521453958415674,
      "learning_rate": 2.2516718424824845e-06,
      "loss": 0.8363,
      "step": 278460
    },
    {
      "epoch": 4.888955213399112,
      "grad_norm": 0.05384470711025122,
      "learning_rate": 2.2508783861280647e-06,
      "loss": 0.8361,
      "step": 278470
    },
    {
      "epoch": 4.889130778279113,
      "grad_norm": 0.048855296033197514,
      "learning_rate": 2.250086180943941e-06,
      "loss": 0.8429,
      "step": 278480
    },
    {
      "epoch": 4.889306343159115,
      "grad_norm": 0.05327099847422694,
      "learning_rate": 2.249295226940149e-06,
      "loss": 0.8372,
      "step": 278490
    },
    {
      "epoch": 4.889481908039116,
      "grad_norm": 0.04683294212189078,
      "learning_rate": 2.2485055241267228e-06,
      "loss": 0.8334,
      "step": 278500
    },
    {
      "epoch": 4.889657472919117,
      "grad_norm": 0.0618024755718891,
      "learning_rate": 2.2477170725136435e-06,
      "loss": 0.8405,
      "step": 278510
    },
    {
      "epoch": 4.889833037799119,
      "grad_norm": 0.06839605901016345,
      "learning_rate": 2.2469298721109236e-06,
      "loss": 0.8412,
      "step": 278520
    },
    {
      "epoch": 4.89000860267912,
      "grad_norm": 0.07712691406268805,
      "learning_rate": 2.2461439229285208e-06,
      "loss": 0.8393,
      "step": 278530
    },
    {
      "epoch": 4.890184167559122,
      "grad_norm": 0.06665950976515594,
      "learning_rate": 2.245359224976404e-06,
      "loss": 0.8299,
      "step": 278540
    },
    {
      "epoch": 4.890359732439123,
      "grad_norm": 0.07399322024581181,
      "learning_rate": 2.244575778264521e-06,
      "loss": 0.8347,
      "step": 278550
    },
    {
      "epoch": 4.890535297319124,
      "grad_norm": 0.03979243302458181,
      "learning_rate": 2.243793582802775e-06,
      "loss": 0.8326,
      "step": 278560
    },
    {
      "epoch": 4.890710862199126,
      "grad_norm": 0.052175661813315004,
      "learning_rate": 2.2430126386011006e-06,
      "loss": 0.833,
      "step": 278570
    },
    {
      "epoch": 4.890886427079127,
      "grad_norm": 0.05261326169584823,
      "learning_rate": 2.2422329456693804e-06,
      "loss": 0.842,
      "step": 278580
    },
    {
      "epoch": 4.891061991959129,
      "grad_norm": 0.04878331209543456,
      "learning_rate": 2.241454504017495e-06,
      "loss": 0.8314,
      "step": 278590
    },
    {
      "epoch": 4.89123755683913,
      "grad_norm": 0.04689565832430067,
      "learning_rate": 2.240677313655303e-06,
      "loss": 0.8478,
      "step": 278600
    },
    {
      "epoch": 4.891413121719132,
      "grad_norm": 0.05221460430315399,
      "learning_rate": 2.2399013745926643e-06,
      "loss": 0.8341,
      "step": 278610
    },
    {
      "epoch": 4.891588686599133,
      "grad_norm": 0.05145104945388073,
      "learning_rate": 2.239126686839394e-06,
      "loss": 0.8343,
      "step": 278620
    },
    {
      "epoch": 4.891764251479134,
      "grad_norm": 0.051053261118615474,
      "learning_rate": 2.238353250405328e-06,
      "loss": 0.8352,
      "step": 278630
    },
    {
      "epoch": 4.891939816359136,
      "grad_norm": 0.04880858847466602,
      "learning_rate": 2.2375810653002395e-06,
      "loss": 0.8385,
      "step": 278640
    },
    {
      "epoch": 4.892115381239137,
      "grad_norm": 0.05543594798725162,
      "learning_rate": 2.236810131533942e-06,
      "loss": 0.846,
      "step": 278650
    },
    {
      "epoch": 4.892290946119139,
      "grad_norm": 0.05192552768549944,
      "learning_rate": 2.2360404491161743e-06,
      "loss": 0.833,
      "step": 278660
    },
    {
      "epoch": 4.89246651099914,
      "grad_norm": 0.057205350420153366,
      "learning_rate": 2.235272018056708e-06,
      "loss": 0.8299,
      "step": 278670
    },
    {
      "epoch": 4.892642075879142,
      "grad_norm": 0.04963350882636005,
      "learning_rate": 2.2345048383652808e-06,
      "loss": 0.8445,
      "step": 278680
    },
    {
      "epoch": 4.892817640759143,
      "grad_norm": 0.05326250889020277,
      "learning_rate": 2.2337389100515977e-06,
      "loss": 0.839,
      "step": 278690
    },
    {
      "epoch": 4.892993205639144,
      "grad_norm": 0.05072235350829191,
      "learning_rate": 2.2329742331253643e-06,
      "loss": 0.8342,
      "step": 278700
    },
    {
      "epoch": 4.893168770519146,
      "grad_norm": 0.06487674690735258,
      "learning_rate": 2.232210807596286e-06,
      "loss": 0.8417,
      "step": 278710
    },
    {
      "epoch": 4.893344335399147,
      "grad_norm": 0.05521036590664638,
      "learning_rate": 2.231448633474024e-06,
      "loss": 0.8334,
      "step": 278720
    },
    {
      "epoch": 4.8935199002791485,
      "grad_norm": 0.055803420315471926,
      "learning_rate": 2.230687710768239e-06,
      "loss": 0.8356,
      "step": 278730
    },
    {
      "epoch": 4.8936954651591495,
      "grad_norm": 0.056248982327419414,
      "learning_rate": 2.2299280394885603e-06,
      "loss": 0.8479,
      "step": 278740
    },
    {
      "epoch": 4.893871030039151,
      "grad_norm": 0.059976376334121805,
      "learning_rate": 2.229169619644637e-06,
      "loss": 0.8387,
      "step": 278750
    },
    {
      "epoch": 4.8940465949191525,
      "grad_norm": 0.06209168189454056,
      "learning_rate": 2.2284124512460547e-06,
      "loss": 0.8382,
      "step": 278760
    },
    {
      "epoch": 4.8942221597991535,
      "grad_norm": 0.06878826917141462,
      "learning_rate": 2.2276565343024187e-06,
      "loss": 0.8415,
      "step": 278770
    },
    {
      "epoch": 4.8943977246791555,
      "grad_norm": 0.04649203747074873,
      "learning_rate": 2.2269018688232913e-06,
      "loss": 0.8378,
      "step": 278780
    },
    {
      "epoch": 4.8945732895591565,
      "grad_norm": 0.055556209128093594,
      "learning_rate": 2.226148454818258e-06,
      "loss": 0.8379,
      "step": 278790
    },
    {
      "epoch": 4.8947488544391575,
      "grad_norm": 0.05496131926505662,
      "learning_rate": 2.225396292296858e-06,
      "loss": 0.8367,
      "step": 278800
    },
    {
      "epoch": 4.8949244193191594,
      "grad_norm": 0.05801774918334893,
      "learning_rate": 2.2246453812686103e-06,
      "loss": 0.8405,
      "step": 278810
    },
    {
      "epoch": 4.8950999841991605,
      "grad_norm": 0.0472910479809456,
      "learning_rate": 2.223895721743044e-06,
      "loss": 0.8295,
      "step": 278820
    },
    {
      "epoch": 4.895275549079162,
      "grad_norm": 0.05122560454556435,
      "learning_rate": 2.2231473137296447e-06,
      "loss": 0.8389,
      "step": 278830
    },
    {
      "epoch": 4.895451113959163,
      "grad_norm": 0.056013324072003,
      "learning_rate": 2.2224001572378974e-06,
      "loss": 0.8304,
      "step": 278840
    },
    {
      "epoch": 4.895626678839165,
      "grad_norm": 0.05012957623410529,
      "learning_rate": 2.2216542522772664e-06,
      "loss": 0.8434,
      "step": 278850
    },
    {
      "epoch": 4.895802243719166,
      "grad_norm": 0.06925939268954806,
      "learning_rate": 2.220909598857214e-06,
      "loss": 0.8391,
      "step": 278860
    },
    {
      "epoch": 4.895977808599168,
      "grad_norm": 0.055425291918549446,
      "learning_rate": 2.2201661969871607e-06,
      "loss": 0.8315,
      "step": 278870
    },
    {
      "epoch": 4.896153373479169,
      "grad_norm": 0.046764257468795256,
      "learning_rate": 2.2194240466765473e-06,
      "loss": 0.8364,
      "step": 278880
    },
    {
      "epoch": 4.89632893835917,
      "grad_norm": 0.061332902399378136,
      "learning_rate": 2.2186831479347496e-06,
      "loss": 0.8358,
      "step": 278890
    },
    {
      "epoch": 4.896504503239172,
      "grad_norm": 0.04812721115076412,
      "learning_rate": 2.217943500771176e-06,
      "loss": 0.8343,
      "step": 278900
    },
    {
      "epoch": 4.896680068119173,
      "grad_norm": 0.049910574815562644,
      "learning_rate": 2.2172051051951796e-06,
      "loss": 0.8371,
      "step": 278910
    },
    {
      "epoch": 4.896855632999175,
      "grad_norm": 0.05550576573079736,
      "learning_rate": 2.2164679612161366e-06,
      "loss": 0.8402,
      "step": 278920
    },
    {
      "epoch": 4.897031197879176,
      "grad_norm": 0.04787282432936886,
      "learning_rate": 2.2157320688433667e-06,
      "loss": 0.8457,
      "step": 278930
    },
    {
      "epoch": 4.897206762759177,
      "grad_norm": 0.05563499437827437,
      "learning_rate": 2.2149974280862135e-06,
      "loss": 0.8418,
      "step": 278940
    },
    {
      "epoch": 4.897382327639179,
      "grad_norm": 0.050404840686993475,
      "learning_rate": 2.214264038953964e-06,
      "loss": 0.8409,
      "step": 278950
    },
    {
      "epoch": 4.89755789251918,
      "grad_norm": 0.048198400712412263,
      "learning_rate": 2.2135319014559274e-06,
      "loss": 0.8399,
      "step": 278960
    },
    {
      "epoch": 4.897733457399182,
      "grad_norm": 0.05052357082306422,
      "learning_rate": 2.21280101560137e-06,
      "loss": 0.8343,
      "step": 278970
    },
    {
      "epoch": 4.897909022279183,
      "grad_norm": 0.057731122713568624,
      "learning_rate": 2.2120713813995677e-06,
      "loss": 0.8375,
      "step": 278980
    },
    {
      "epoch": 4.898084587159184,
      "grad_norm": 0.05081446340645816,
      "learning_rate": 2.211342998859743e-06,
      "loss": 0.842,
      "step": 278990
    },
    {
      "epoch": 4.898260152039186,
      "grad_norm": 0.05587891997816039,
      "learning_rate": 2.210615867991139e-06,
      "loss": 0.8335,
      "step": 279000
    },
    {
      "epoch": 4.898435716919187,
      "grad_norm": 0.06035027903754979,
      "learning_rate": 2.2098899888029655e-06,
      "loss": 0.8417,
      "step": 279010
    },
    {
      "epoch": 4.898611281799189,
      "grad_norm": 0.046754470594056664,
      "learning_rate": 2.2091653613044126e-06,
      "loss": 0.833,
      "step": 279020
    },
    {
      "epoch": 4.89878684667919,
      "grad_norm": 0.04032333011065203,
      "learning_rate": 2.208441985504668e-06,
      "loss": 0.8371,
      "step": 279030
    },
    {
      "epoch": 4.898962411559191,
      "grad_norm": 0.05167981130926976,
      "learning_rate": 2.207719861412899e-06,
      "loss": 0.8376,
      "step": 279040
    },
    {
      "epoch": 4.899137976439193,
      "grad_norm": 0.06430372013647201,
      "learning_rate": 2.2069989890382607e-06,
      "loss": 0.8385,
      "step": 279050
    },
    {
      "epoch": 4.899313541319194,
      "grad_norm": 0.05034317800361945,
      "learning_rate": 2.206279368389876e-06,
      "loss": 0.8318,
      "step": 279060
    },
    {
      "epoch": 4.899489106199196,
      "grad_norm": 0.04825736862765043,
      "learning_rate": 2.205560999476857e-06,
      "loss": 0.8343,
      "step": 279070
    },
    {
      "epoch": 4.899664671079197,
      "grad_norm": 0.07276220921868239,
      "learning_rate": 2.2048438823083257e-06,
      "loss": 0.8405,
      "step": 279080
    },
    {
      "epoch": 4.899840235959199,
      "grad_norm": 0.053154753193462245,
      "learning_rate": 2.20412801689335e-06,
      "loss": 0.8333,
      "step": 279090
    },
    {
      "epoch": 4.9000158008392,
      "grad_norm": 0.04331622805782756,
      "learning_rate": 2.2034134032409986e-06,
      "loss": 0.835,
      "step": 279100
    },
    {
      "epoch": 4.900191365719202,
      "grad_norm": 0.0444677905484624,
      "learning_rate": 2.202700041360349e-06,
      "loss": 0.8309,
      "step": 279110
    },
    {
      "epoch": 4.900366930599203,
      "grad_norm": 0.05015715321949416,
      "learning_rate": 2.2019879312604146e-06,
      "loss": 0.8347,
      "step": 279120
    },
    {
      "epoch": 4.900542495479204,
      "grad_norm": 0.043173930789633067,
      "learning_rate": 2.20127707295023e-06,
      "loss": 0.8319,
      "step": 279130
    },
    {
      "epoch": 4.900718060359206,
      "grad_norm": 0.06468821655756614,
      "learning_rate": 2.200567466438797e-06,
      "loss": 0.8279,
      "step": 279140
    },
    {
      "epoch": 4.900893625239207,
      "grad_norm": 0.04231210687396608,
      "learning_rate": 2.1998591117351178e-06,
      "loss": 0.8406,
      "step": 279150
    },
    {
      "epoch": 4.901069190119209,
      "grad_norm": 0.07191216745453653,
      "learning_rate": 2.1991520088481494e-06,
      "loss": 0.8399,
      "step": 279160
    },
    {
      "epoch": 4.90124475499921,
      "grad_norm": 0.04993714945850589,
      "learning_rate": 2.198446157786861e-06,
      "loss": 0.835,
      "step": 279170
    },
    {
      "epoch": 4.901420319879211,
      "grad_norm": 0.05761185434038643,
      "learning_rate": 2.1977415585602004e-06,
      "loss": 0.84,
      "step": 279180
    },
    {
      "epoch": 4.901595884759213,
      "grad_norm": 0.041888489082501035,
      "learning_rate": 2.19703821117708e-06,
      "loss": 0.8478,
      "step": 279190
    },
    {
      "epoch": 4.901771449639214,
      "grad_norm": 0.04805956843946023,
      "learning_rate": 2.1963361156464253e-06,
      "loss": 0.837,
      "step": 279200
    },
    {
      "epoch": 4.901947014519216,
      "grad_norm": 0.08199819254167685,
      "learning_rate": 2.1956352719771282e-06,
      "loss": 0.8457,
      "step": 279210
    },
    {
      "epoch": 4.902122579399217,
      "grad_norm": 0.058827150423677414,
      "learning_rate": 2.1949356801780804e-06,
      "loss": 0.8383,
      "step": 279220
    },
    {
      "epoch": 4.902298144279218,
      "grad_norm": 0.04485202712486317,
      "learning_rate": 2.1942373402581198e-06,
      "loss": 0.8375,
      "step": 279230
    },
    {
      "epoch": 4.90247370915922,
      "grad_norm": 0.04649297824535971,
      "learning_rate": 2.1935402522261037e-06,
      "loss": 0.8376,
      "step": 279240
    },
    {
      "epoch": 4.902649274039221,
      "grad_norm": 0.05114129542404288,
      "learning_rate": 2.192844416090881e-06,
      "loss": 0.8353,
      "step": 279250
    },
    {
      "epoch": 4.902824838919223,
      "grad_norm": 0.046469863676526744,
      "learning_rate": 2.1921498318612557e-06,
      "loss": 0.8423,
      "step": 279260
    },
    {
      "epoch": 4.903000403799224,
      "grad_norm": 0.05144290962780012,
      "learning_rate": 2.1914564995460207e-06,
      "loss": 0.8348,
      "step": 279270
    },
    {
      "epoch": 4.903175968679226,
      "grad_norm": 0.047777590885201775,
      "learning_rate": 2.190764419153968e-06,
      "loss": 0.8348,
      "step": 279280
    },
    {
      "epoch": 4.903351533559227,
      "grad_norm": 0.07157630274946687,
      "learning_rate": 2.1900735906938705e-06,
      "loss": 0.8464,
      "step": 279290
    },
    {
      "epoch": 4.903527098439228,
      "grad_norm": 0.04617307671764577,
      "learning_rate": 2.1893840141744752e-06,
      "loss": 0.843,
      "step": 279300
    },
    {
      "epoch": 4.90370266331923,
      "grad_norm": 0.05980516433723175,
      "learning_rate": 2.1886956896045213e-06,
      "loss": 0.8365,
      "step": 279310
    },
    {
      "epoch": 4.903878228199231,
      "grad_norm": 0.07316911806676787,
      "learning_rate": 2.1880086169927357e-06,
      "loss": 0.8306,
      "step": 279320
    },
    {
      "epoch": 4.904053793079233,
      "grad_norm": 0.04729600073509166,
      "learning_rate": 2.1873227963478124e-06,
      "loss": 0.8367,
      "step": 279330
    },
    {
      "epoch": 4.904229357959234,
      "grad_norm": 0.043056974917395026,
      "learning_rate": 2.1866382276784453e-06,
      "loss": 0.8365,
      "step": 279340
    },
    {
      "epoch": 4.904404922839236,
      "grad_norm": 0.05043370629262698,
      "learning_rate": 2.185954910993317e-06,
      "loss": 0.8404,
      "step": 279350
    },
    {
      "epoch": 4.904580487719237,
      "grad_norm": 0.052100155729152824,
      "learning_rate": 2.1852728463010783e-06,
      "loss": 0.8307,
      "step": 279360
    },
    {
      "epoch": 4.904756052599238,
      "grad_norm": 0.05175030584511767,
      "learning_rate": 2.184592033610357e-06,
      "loss": 0.8378,
      "step": 279370
    },
    {
      "epoch": 4.90493161747924,
      "grad_norm": 0.05335181932513785,
      "learning_rate": 2.1839124729298036e-06,
      "loss": 0.8333,
      "step": 279380
    },
    {
      "epoch": 4.905107182359241,
      "grad_norm": 0.08539987228708797,
      "learning_rate": 2.1832341642680126e-06,
      "loss": 0.8329,
      "step": 279390
    },
    {
      "epoch": 4.905282747239243,
      "grad_norm": 0.05093353269388306,
      "learning_rate": 2.1825571076335904e-06,
      "loss": 0.8331,
      "step": 279400
    },
    {
      "epoch": 4.905458312119244,
      "grad_norm": 0.04104946421152567,
      "learning_rate": 2.1818813030350988e-06,
      "loss": 0.8422,
      "step": 279410
    },
    {
      "epoch": 4.905633876999245,
      "grad_norm": 0.07324502464453338,
      "learning_rate": 2.181206750481112e-06,
      "loss": 0.8352,
      "step": 279420
    },
    {
      "epoch": 4.905809441879247,
      "grad_norm": 0.056873450190901134,
      "learning_rate": 2.18053344998018e-06,
      "loss": 0.8361,
      "step": 279430
    },
    {
      "epoch": 4.905985006759248,
      "grad_norm": 0.04838378960420139,
      "learning_rate": 2.1798614015408214e-06,
      "loss": 0.8349,
      "step": 279440
    },
    {
      "epoch": 4.90616057163925,
      "grad_norm": 0.04010266085159644,
      "learning_rate": 2.1791906051715546e-06,
      "loss": 0.8385,
      "step": 279450
    },
    {
      "epoch": 4.906336136519251,
      "grad_norm": 0.06160821926695372,
      "learning_rate": 2.1785210608808865e-06,
      "loss": 0.8392,
      "step": 279460
    },
    {
      "epoch": 4.906511701399252,
      "grad_norm": 0.04326563547929671,
      "learning_rate": 2.1778527686773025e-06,
      "loss": 0.8348,
      "step": 279470
    },
    {
      "epoch": 4.906687266279254,
      "grad_norm": 0.0644465843623066,
      "learning_rate": 2.1771857285692544e-06,
      "loss": 0.8468,
      "step": 279480
    },
    {
      "epoch": 4.906862831159255,
      "grad_norm": 0.05352062705063244,
      "learning_rate": 2.176519940565206e-06,
      "loss": 0.8334,
      "step": 279490
    },
    {
      "epoch": 4.9070383960392565,
      "grad_norm": 0.047710068235627824,
      "learning_rate": 2.175855404673587e-06,
      "loss": 0.839,
      "step": 279500
    },
    {
      "epoch": 4.9072139609192575,
      "grad_norm": 0.06324675606614232,
      "learning_rate": 2.175192120902818e-06,
      "loss": 0.8343,
      "step": 279510
    },
    {
      "epoch": 4.9073895257992595,
      "grad_norm": 0.04956125167354765,
      "learning_rate": 2.1745300892613057e-06,
      "loss": 0.8352,
      "step": 279520
    },
    {
      "epoch": 4.9075650906792605,
      "grad_norm": 0.037112469927390895,
      "learning_rate": 2.1738693097574368e-06,
      "loss": 0.8366,
      "step": 279530
    },
    {
      "epoch": 4.907740655559262,
      "grad_norm": 0.050669385771234345,
      "learning_rate": 2.173209782399587e-06,
      "loss": 0.8409,
      "step": 279540
    },
    {
      "epoch": 4.9079162204392635,
      "grad_norm": 0.04574227321383287,
      "learning_rate": 2.172551507196109e-06,
      "loss": 0.8363,
      "step": 279550
    },
    {
      "epoch": 4.9080917853192645,
      "grad_norm": 0.04551084246897065,
      "learning_rate": 2.171894484155335e-06,
      "loss": 0.8434,
      "step": 279560
    },
    {
      "epoch": 4.908267350199266,
      "grad_norm": 0.04361574789713374,
      "learning_rate": 2.171238713285606e-06,
      "loss": 0.8369,
      "step": 279570
    },
    {
      "epoch": 4.908442915079267,
      "grad_norm": 0.04724029641923724,
      "learning_rate": 2.1705841945952224e-06,
      "loss": 0.8461,
      "step": 279580
    },
    {
      "epoch": 4.908618479959269,
      "grad_norm": 0.052440321673310926,
      "learning_rate": 2.169930928092481e-06,
      "loss": 0.8391,
      "step": 279590
    },
    {
      "epoch": 4.90879404483927,
      "grad_norm": 0.04581415199248438,
      "learning_rate": 2.169278913785647e-06,
      "loss": 0.8354,
      "step": 279600
    },
    {
      "epoch": 4.908969609719271,
      "grad_norm": 0.05226010322180794,
      "learning_rate": 2.1686281516829983e-06,
      "loss": 0.8463,
      "step": 279610
    },
    {
      "epoch": 4.909145174599273,
      "grad_norm": 0.044276340760401156,
      "learning_rate": 2.1679786417927654e-06,
      "loss": 0.8398,
      "step": 279620
    },
    {
      "epoch": 4.909320739479274,
      "grad_norm": 0.05288930929219624,
      "learning_rate": 2.167330384123193e-06,
      "loss": 0.8384,
      "step": 279630
    },
    {
      "epoch": 4.909496304359276,
      "grad_norm": 0.05755266905539326,
      "learning_rate": 2.1666833786824803e-06,
      "loss": 0.8417,
      "step": 279640
    },
    {
      "epoch": 4.909671869239277,
      "grad_norm": 0.06568486436433703,
      "learning_rate": 2.1660376254788266e-06,
      "loss": 0.8322,
      "step": 279650
    },
    {
      "epoch": 4.909847434119278,
      "grad_norm": 0.05035782646462016,
      "learning_rate": 2.1653931245204205e-06,
      "loss": 0.8402,
      "step": 279660
    },
    {
      "epoch": 4.91002299899928,
      "grad_norm": 0.06210543654294787,
      "learning_rate": 2.1647498758154284e-06,
      "loss": 0.8357,
      "step": 279670
    },
    {
      "epoch": 4.910198563879281,
      "grad_norm": 0.04704650552818354,
      "learning_rate": 2.1641078793719945e-06,
      "loss": 0.841,
      "step": 279680
    },
    {
      "epoch": 4.910374128759283,
      "grad_norm": 0.04916405314266536,
      "learning_rate": 2.1634671351982527e-06,
      "loss": 0.8412,
      "step": 279690
    },
    {
      "epoch": 4.910549693639284,
      "grad_norm": 0.0527733013129654,
      "learning_rate": 2.1628276433023254e-06,
      "loss": 0.8419,
      "step": 279700
    },
    {
      "epoch": 4.910725258519285,
      "grad_norm": 0.049605644519071325,
      "learning_rate": 2.1621894036923127e-06,
      "loss": 0.8402,
      "step": 279710
    },
    {
      "epoch": 4.910900823399287,
      "grad_norm": 0.050821098745100934,
      "learning_rate": 2.161552416376305e-06,
      "loss": 0.8354,
      "step": 279720
    },
    {
      "epoch": 4.911076388279288,
      "grad_norm": 0.06046703143729991,
      "learning_rate": 2.1609166813623694e-06,
      "loss": 0.8378,
      "step": 279730
    },
    {
      "epoch": 4.91125195315929,
      "grad_norm": 0.05044453773583008,
      "learning_rate": 2.1602821986585617e-06,
      "loss": 0.839,
      "step": 279740
    },
    {
      "epoch": 4.911427518039291,
      "grad_norm": 0.05514556390601098,
      "learning_rate": 2.1596489682729286e-06,
      "loss": 0.8358,
      "step": 279750
    },
    {
      "epoch": 4.911603082919293,
      "grad_norm": 0.057981331300718766,
      "learning_rate": 2.1590169902134824e-06,
      "loss": 0.8416,
      "step": 279760
    },
    {
      "epoch": 4.911778647799294,
      "grad_norm": 0.05426131070894731,
      "learning_rate": 2.1583862644882244e-06,
      "loss": 0.8426,
      "step": 279770
    },
    {
      "epoch": 4.911954212679296,
      "grad_norm": 0.05641558529129608,
      "learning_rate": 2.157756791105167e-06,
      "loss": 0.8375,
      "step": 279780
    },
    {
      "epoch": 4.912129777559297,
      "grad_norm": 0.05642186222656356,
      "learning_rate": 2.157128570072269e-06,
      "loss": 0.8468,
      "step": 279790
    },
    {
      "epoch": 4.912305342439298,
      "grad_norm": 0.03954130662683614,
      "learning_rate": 2.1565016013974975e-06,
      "loss": 0.8389,
      "step": 279800
    },
    {
      "epoch": 4.9124809073193,
      "grad_norm": 0.059825246012372255,
      "learning_rate": 2.155875885088789e-06,
      "loss": 0.8367,
      "step": 279810
    },
    {
      "epoch": 4.912656472199301,
      "grad_norm": 0.06253953049105121,
      "learning_rate": 2.1552514211540793e-06,
      "loss": 0.829,
      "step": 279820
    },
    {
      "epoch": 4.912832037079303,
      "grad_norm": 0.0551245140919602,
      "learning_rate": 2.1546282096012704e-06,
      "loss": 0.8307,
      "step": 279830
    },
    {
      "epoch": 4.913007601959304,
      "grad_norm": 0.0534558208617877,
      "learning_rate": 2.1540062504382765e-06,
      "loss": 0.8365,
      "step": 279840
    },
    {
      "epoch": 4.913183166839305,
      "grad_norm": 0.04717544386298845,
      "learning_rate": 2.153385543672967e-06,
      "loss": 0.8435,
      "step": 279850
    },
    {
      "epoch": 4.913358731719307,
      "grad_norm": 0.04772322205221167,
      "learning_rate": 2.152766089313201e-06,
      "loss": 0.8333,
      "step": 279860
    },
    {
      "epoch": 4.913534296599308,
      "grad_norm": 0.05597869377708241,
      "learning_rate": 2.1521478873668367e-06,
      "loss": 0.8381,
      "step": 279870
    },
    {
      "epoch": 4.91370986147931,
      "grad_norm": 0.05213046536879911,
      "learning_rate": 2.1515309378416894e-06,
      "loss": 0.8317,
      "step": 279880
    },
    {
      "epoch": 4.913885426359311,
      "grad_norm": 0.05700192271910947,
      "learning_rate": 2.1509152407456065e-06,
      "loss": 0.8355,
      "step": 279890
    },
    {
      "epoch": 4.914060991239312,
      "grad_norm": 0.04657504380023435,
      "learning_rate": 2.15030079608637e-06,
      "loss": 0.8401,
      "step": 279900
    },
    {
      "epoch": 4.914236556119314,
      "grad_norm": 0.05659475025374277,
      "learning_rate": 2.1496876038717616e-06,
      "loss": 0.8376,
      "step": 279910
    },
    {
      "epoch": 4.914412120999315,
      "grad_norm": 0.11344765372029422,
      "learning_rate": 2.1490756641095516e-06,
      "loss": 0.8453,
      "step": 279920
    },
    {
      "epoch": 4.914587685879317,
      "grad_norm": 0.055910393210492064,
      "learning_rate": 2.1484649768075e-06,
      "loss": 0.8407,
      "step": 279930
    },
    {
      "epoch": 4.914763250759318,
      "grad_norm": 0.05581482747389268,
      "learning_rate": 2.1478555419733452e-06,
      "loss": 0.8459,
      "step": 279940
    },
    {
      "epoch": 4.91493881563932,
      "grad_norm": 0.047898834586171536,
      "learning_rate": 2.1472473596148137e-06,
      "loss": 0.836,
      "step": 279950
    },
    {
      "epoch": 4.915114380519321,
      "grad_norm": 0.051149828668926034,
      "learning_rate": 2.146640429739588e-06,
      "loss": 0.8442,
      "step": 279960
    },
    {
      "epoch": 4.915289945399322,
      "grad_norm": 0.05359341235530319,
      "learning_rate": 2.1460347523553847e-06,
      "loss": 0.8296,
      "step": 279970
    },
    {
      "epoch": 4.915465510279324,
      "grad_norm": 0.05467313358930575,
      "learning_rate": 2.1454303274698634e-06,
      "loss": 0.8395,
      "step": 279980
    },
    {
      "epoch": 4.915641075159325,
      "grad_norm": 0.04736530029630476,
      "learning_rate": 2.1448271550906754e-06,
      "loss": 0.8312,
      "step": 279990
    },
    {
      "epoch": 4.915816640039327,
      "grad_norm": 0.04552368774162066,
      "learning_rate": 2.1442252352254805e-06,
      "loss": 0.8492,
      "step": 280000
    }
  ],
  "logging_steps": 10,
  "max_steps": 284795,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 5000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.422163750513541e+17,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}