{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 1313,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.007616146230007616,
      "grad_norm": 9.639493086797174,
      "learning_rate": 8.333333333333333e-07,
      "loss": 1.2528,
      "step": 10
    },
    {
      "epoch": 0.015232292460015232,
      "grad_norm": 8.49004769576833,
      "learning_rate": 2.5e-06,
      "loss": 0.5082,
      "step": 20
    },
    {
      "epoch": 0.02284843869002285,
      "grad_norm": 6.555158164896051,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.3399,
      "step": 30
    },
    {
      "epoch": 0.030464584920030464,
      "grad_norm": 8.564554290087065,
      "learning_rate": 4.9805144193296965e-06,
      "loss": 0.3035,
      "step": 40
    },
    {
      "epoch": 0.03808073115003808,
      "grad_norm": 6.702661496981953,
      "learning_rate": 4.9454403741231495e-06,
      "loss": 0.3163,
      "step": 50
    },
    {
      "epoch": 0.0456968773800457,
      "grad_norm": 11.658563206953636,
      "learning_rate": 4.906469212782542e-06,
      "loss": 0.3014,
      "step": 60
    },
    {
      "epoch": 0.053313023610053314,
      "grad_norm": 9.653812130588781,
      "learning_rate": 4.867498051441934e-06,
      "loss": 0.28,
      "step": 70
    },
    {
      "epoch": 0.06092916984006093,
      "grad_norm": 6.566810094632355,
      "learning_rate": 4.832424006235386e-06,
      "loss": 0.2801,
      "step": 80
    },
    {
      "epoch": 0.06854531607006854,
      "grad_norm": 9.977494581893676,
      "learning_rate": 4.793452844894778e-06,
      "loss": 0.299,
      "step": 90
    },
    {
      "epoch": 0.07616146230007616,
      "grad_norm": 3.7598332928698057,
      "learning_rate": 4.75448168355417e-06,
      "loss": 0.2936,
      "step": 100
    },
    {
      "epoch": 0.08377760853008377,
      "grad_norm": 3.5678745009701713,
      "learning_rate": 4.715510522213562e-06,
      "loss": 0.3044,
      "step": 110
    },
    {
      "epoch": 0.0913937547600914,
      "grad_norm": 13.805243688595516,
      "learning_rate": 4.676539360872954e-06,
      "loss": 0.2683,
      "step": 120
    },
    {
      "epoch": 0.09900990099009901,
      "grad_norm": 5.852174401983805,
      "learning_rate": 4.6375681995323465e-06,
      "loss": 0.2444,
      "step": 130
    },
    {
      "epoch": 0.10662604722010663,
      "grad_norm": 7.855194077383853,
      "learning_rate": 4.5985970381917386e-06,
      "loss": 0.3041,
      "step": 140
    },
    {
      "epoch": 0.11424219345011424,
      "grad_norm": 8.043209218403092,
      "learning_rate": 4.559625876851131e-06,
      "loss": 0.2442,
      "step": 150
    },
    {
      "epoch": 0.12185833968012186,
      "grad_norm": 25.144531941751673,
      "learning_rate": 4.520654715510523e-06,
      "loss": 0.272,
      "step": 160
    },
    {
      "epoch": 0.12947448591012947,
      "grad_norm": 6.509174092287003,
      "learning_rate": 4.481683554169914e-06,
      "loss": 0.2623,
      "step": 170
    },
    {
      "epoch": 0.1370906321401371,
      "grad_norm": 9.500128771255806,
      "learning_rate": 4.442712392829306e-06,
      "loss": 0.2665,
      "step": 180
    },
    {
      "epoch": 0.1447067783701447,
      "grad_norm": 3.607016552182795,
      "learning_rate": 4.403741231488698e-06,
      "loss": 0.2802,
      "step": 190
    },
    {
      "epoch": 0.15232292460015232,
      "grad_norm": 4.42691540524899,
      "learning_rate": 4.36477007014809e-06,
      "loss": 0.2363,
      "step": 200
    },
    {
      "epoch": 0.15993907083015993,
      "grad_norm": 3.3159003473586197,
      "learning_rate": 4.3257989088074824e-06,
      "loss": 0.2326,
      "step": 210
    },
    {
      "epoch": 0.16755521706016754,
      "grad_norm": 4.689165992829481,
      "learning_rate": 4.286827747466875e-06,
      "loss": 0.2433,
      "step": 220
    },
    {
      "epoch": 0.17517136329017516,
      "grad_norm": 9.737741759699885,
      "learning_rate": 4.2478565861262675e-06,
      "loss": 0.2467,
      "step": 230
    },
    {
      "epoch": 0.1827875095201828,
      "grad_norm": 2.836341979209339,
      "learning_rate": 4.20888542478566e-06,
      "loss": 0.2391,
      "step": 240
    },
    {
      "epoch": 0.19040365575019041,
      "grad_norm": 1.5359361442871327,
      "learning_rate": 4.169914263445051e-06,
      "loss": 0.2367,
      "step": 250
    },
    {
      "epoch": 0.19801980198019803,
      "grad_norm": 4.250993517900826,
      "learning_rate": 4.130943102104443e-06,
      "loss": 0.2719,
      "step": 260
    },
    {
      "epoch": 0.20563594821020564,
      "grad_norm": 2.537220060582072,
      "learning_rate": 4.091971940763835e-06,
      "loss": 0.2269,
      "step": 270
    },
    {
      "epoch": 0.21325209444021326,
      "grad_norm": 1.3331572997631018,
      "learning_rate": 4.053000779423227e-06,
      "loss": 0.2316,
      "step": 280
    },
    {
      "epoch": 0.22086824067022087,
      "grad_norm": 3.608424270155421,
      "learning_rate": 4.014029618082619e-06,
      "loss": 0.248,
      "step": 290
    },
    {
      "epoch": 0.2284843869002285,
      "grad_norm": 4.4991481026863225,
      "learning_rate": 3.975058456742011e-06,
      "loss": 0.2397,
      "step": 300
    },
    {
      "epoch": 0.2361005331302361,
      "grad_norm": 3.8860134497329843,
      "learning_rate": 3.9360872954014035e-06,
      "loss": 0.2615,
      "step": 310
    },
    {
      "epoch": 0.24371667936024372,
      "grad_norm": 4.04363997587026,
      "learning_rate": 3.897116134060796e-06,
      "loss": 0.2428,
      "step": 320
    },
    {
      "epoch": 0.25133282559025133,
      "grad_norm": 1.8755213132629207,
      "learning_rate": 3.858144972720188e-06,
      "loss": 0.2192,
      "step": 330
    },
    {
      "epoch": 0.25894897182025894,
      "grad_norm": 6.113234551860946,
      "learning_rate": 3.81917381137958e-06,
      "loss": 0.2393,
      "step": 340
    },
    {
      "epoch": 0.26656511805026656,
      "grad_norm": 9.925025613514833,
      "learning_rate": 3.7802026500389715e-06,
      "loss": 0.2229,
      "step": 350
    },
    {
      "epoch": 0.2741812642802742,
      "grad_norm": 5.214638205967677,
      "learning_rate": 3.7412314886983636e-06,
      "loss": 0.2224,
      "step": 360
    },
    {
      "epoch": 0.2817974105102818,
      "grad_norm": 6.087030860156098,
      "learning_rate": 3.7022603273577557e-06,
      "loss": 0.2173,
      "step": 370
    },
    {
      "epoch": 0.2894135567402894,
      "grad_norm": 6.180290616323521,
      "learning_rate": 3.6632891660171478e-06,
      "loss": 0.2144,
      "step": 380
    },
    {
      "epoch": 0.297029702970297,
      "grad_norm": 2.402073266844856,
      "learning_rate": 3.62431800467654e-06,
      "loss": 0.2118,
      "step": 390
    },
    {
      "epoch": 0.30464584920030463,
      "grad_norm": 6.174997984528584,
      "learning_rate": 3.5853468433359316e-06,
      "loss": 0.2615,
      "step": 400
    },
    {
      "epoch": 0.31226199543031224,
      "grad_norm": 8.879077685914272,
      "learning_rate": 3.5463756819953237e-06,
      "loss": 0.2529,
      "step": 410
    },
    {
      "epoch": 0.31987814166031986,
      "grad_norm": 5.468782572651813,
      "learning_rate": 3.5074045206547158e-06,
      "loss": 0.2669,
      "step": 420
    },
    {
      "epoch": 0.3274942878903275,
      "grad_norm": 7.406738215292078,
      "learning_rate": 3.468433359314108e-06,
      "loss": 0.2632,
      "step": 430
    },
    {
      "epoch": 0.3351104341203351,
      "grad_norm": 3.70629740090698,
      "learning_rate": 3.4294621979735e-06,
      "loss": 0.2412,
      "step": 440
    },
    {
      "epoch": 0.3427265803503427,
      "grad_norm": 3.3764650248375676,
      "learning_rate": 3.390491036632892e-06,
      "loss": 0.2461,
      "step": 450
    },
    {
      "epoch": 0.3503427265803503,
      "grad_norm": 4.286293923557235,
      "learning_rate": 3.3515198752922838e-06,
      "loss": 0.2367,
      "step": 460
    },
    {
      "epoch": 0.357958872810358,
      "grad_norm": 3.2261118617888767,
      "learning_rate": 3.312548713951676e-06,
      "loss": 0.2508,
      "step": 470
    },
    {
      "epoch": 0.3655750190403656,
      "grad_norm": 2.000066076371398,
      "learning_rate": 3.273577552611068e-06,
      "loss": 0.2355,
      "step": 480
    },
    {
      "epoch": 0.3731911652703732,
      "grad_norm": 1.6980552649530136,
      "learning_rate": 3.23460639127046e-06,
      "loss": 0.2329,
      "step": 490
    },
    {
      "epoch": 0.38080731150038083,
      "grad_norm": 3.5865596510799524,
      "learning_rate": 3.195635229929852e-06,
      "loss": 0.2463,
      "step": 500
    },
    {
      "epoch": 0.38842345773038844,
      "grad_norm": 2.0737159974836428,
      "learning_rate": 3.1566640685892443e-06,
      "loss": 0.2296,
      "step": 510
    },
    {
      "epoch": 0.39603960396039606,
      "grad_norm": 12.615894511310687,
      "learning_rate": 3.117692907248636e-06,
      "loss": 0.2479,
      "step": 520
    },
    {
      "epoch": 0.4036557501904037,
      "grad_norm": 2.941177874523108,
      "learning_rate": 3.078721745908028e-06,
      "loss": 0.2382,
      "step": 530
    },
    {
      "epoch": 0.4112718964204113,
      "grad_norm": 3.4235048536712824,
      "learning_rate": 3.03975058456742e-06,
      "loss": 0.2655,
      "step": 540
    },
    {
      "epoch": 0.4188880426504189,
      "grad_norm": 2.280585583419532,
      "learning_rate": 3.0007794232268123e-06,
      "loss": 0.226,
      "step": 550
    },
    {
      "epoch": 0.4265041888804265,
      "grad_norm": 4.192992543465362,
      "learning_rate": 2.9618082618862044e-06,
      "loss": 0.237,
      "step": 560
    },
    {
      "epoch": 0.43412033511043413,
      "grad_norm": 4.376833355588179,
      "learning_rate": 2.9228371005455965e-06,
      "loss": 0.2354,
      "step": 570
    },
    {
      "epoch": 0.44173648134044174,
      "grad_norm": 5.5764858101422465,
      "learning_rate": 2.883865939204988e-06,
      "loss": 0.2377,
      "step": 580
    },
    {
      "epoch": 0.44935262757044936,
      "grad_norm": 11.079112280738615,
      "learning_rate": 2.8448947778643803e-06,
      "loss": 0.2302,
      "step": 590
    },
    {
      "epoch": 0.456968773800457,
      "grad_norm": 4.653154577810809,
      "learning_rate": 2.8059236165237724e-06,
      "loss": 0.2271,
      "step": 600
    },
    {
      "epoch": 0.4645849200304646,
      "grad_norm": 5.627141358500255,
      "learning_rate": 2.7669524551831645e-06,
      "loss": 0.214,
      "step": 610
    },
    {
      "epoch": 0.4722010662604722,
      "grad_norm": 3.0799884402462583,
      "learning_rate": 2.7279812938425566e-06,
      "loss": 0.2401,
      "step": 620
    },
    {
      "epoch": 0.4798172124904798,
      "grad_norm": 2.9094630692232566,
      "learning_rate": 2.6890101325019487e-06,
      "loss": 0.2156,
      "step": 630
    },
    {
      "epoch": 0.48743335872048743,
      "grad_norm": 4.7368426822797955,
      "learning_rate": 2.6500389711613408e-06,
      "loss": 0.2398,
      "step": 640
    },
    {
      "epoch": 0.49504950495049505,
      "grad_norm": 2.694907521185521,
      "learning_rate": 2.6110678098207325e-06,
      "loss": 0.2005,
      "step": 650
    },
    {
      "epoch": 0.5026656511805027,
      "grad_norm": 4.336379900966037,
      "learning_rate": 2.572096648480125e-06,
      "loss": 0.1999,
      "step": 660
    },
    {
      "epoch": 0.5102817974105103,
      "grad_norm": 5.533701246102975,
      "learning_rate": 2.533125487139517e-06,
      "loss": 0.2066,
      "step": 670
    },
    {
      "epoch": 0.5178979436405179,
      "grad_norm": 4.264409295509442,
      "learning_rate": 2.494154325798909e-06,
      "loss": 0.2127,
      "step": 680
    },
    {
      "epoch": 0.5255140898705255,
      "grad_norm": 6.757877305010082,
      "learning_rate": 2.455183164458301e-06,
      "loss": 0.2107,
      "step": 690
    },
    {
      "epoch": 0.5331302361005331,
      "grad_norm": 21.156572431608588,
      "learning_rate": 2.416212003117693e-06,
      "loss": 0.2004,
      "step": 700
    },
    {
      "epoch": 0.5407463823305407,
      "grad_norm": 7.9735056443944154,
      "learning_rate": 2.377240841777085e-06,
      "loss": 0.2365,
      "step": 710
    },
    {
      "epoch": 0.5483625285605483,
      "grad_norm": 5.553215674492296,
      "learning_rate": 2.338269680436477e-06,
      "loss": 0.1814,
      "step": 720
    },
    {
      "epoch": 0.555978674790556,
      "grad_norm": 94.69988106143563,
      "learning_rate": 2.3031956352299302e-06,
      "loss": 0.1908,
      "step": 730
    },
    {
      "epoch": 0.5635948210205636,
      "grad_norm": 1.649907078405919,
      "learning_rate": 2.264224473889322e-06,
      "loss": 0.1883,
      "step": 740
    },
    {
      "epoch": 0.5712109672505712,
      "grad_norm": 2.6805491138942052,
      "learning_rate": 2.225253312548714e-06,
      "loss": 0.2365,
      "step": 750
    },
    {
      "epoch": 0.5788271134805788,
      "grad_norm": 7.23861556333092,
      "learning_rate": 2.186282151208106e-06,
      "loss": 0.1761,
      "step": 760
    },
    {
      "epoch": 0.5864432597105864,
      "grad_norm": 3.2894178397808846,
      "learning_rate": 2.1473109898674982e-06,
      "loss": 0.2062,
      "step": 770
    },
    {
      "epoch": 0.594059405940594,
      "grad_norm": 10.630820097778582,
      "learning_rate": 2.1083398285268903e-06,
      "loss": 0.221,
      "step": 780
    },
    {
      "epoch": 0.6016755521706016,
      "grad_norm": 4.304468571093565,
      "learning_rate": 2.0693686671862824e-06,
      "loss": 0.1962,
      "step": 790
    },
    {
      "epoch": 0.6092916984006093,
      "grad_norm": 25.23668397466602,
      "learning_rate": 2.030397505845674e-06,
      "loss": 0.2064,
      "step": 800
    },
    {
      "epoch": 0.6169078446306169,
      "grad_norm": 7.801217919207177,
      "learning_rate": 1.9914263445050662e-06,
      "loss": 0.1903,
      "step": 810
    },
    {
      "epoch": 0.6245239908606245,
      "grad_norm": 2.751715853008115,
      "learning_rate": 1.9524551831644583e-06,
      "loss": 0.201,
      "step": 820
    },
    {
      "epoch": 0.6321401370906321,
      "grad_norm": 5.690806891610993,
      "learning_rate": 1.9134840218238504e-06,
      "loss": 0.1945,
      "step": 830
    },
    {
      "epoch": 0.6397562833206397,
      "grad_norm": 14.562381559438334,
      "learning_rate": 1.8745128604832425e-06,
      "loss": 0.1825,
      "step": 840
    },
    {
      "epoch": 0.6473724295506473,
      "grad_norm": 3.134405388535424,
      "learning_rate": 1.8355416991426344e-06,
      "loss": 0.2219,
      "step": 850
    },
    {
      "epoch": 0.654988575780655,
      "grad_norm": 4.669745434608825,
      "learning_rate": 1.7965705378020267e-06,
      "loss": 0.2153,
      "step": 860
    },
    {
      "epoch": 0.6626047220106626,
      "grad_norm": 7.086116154905358,
      "learning_rate": 1.7575993764614188e-06,
      "loss": 0.2205,
      "step": 870
    },
    {
      "epoch": 0.6702208682406702,
      "grad_norm": 3.572081402217666,
      "learning_rate": 1.718628215120811e-06,
      "loss": 0.1877,
      "step": 880
    },
    {
      "epoch": 0.6778370144706778,
      "grad_norm": 2.925715103242468,
      "learning_rate": 1.6796570537802028e-06,
      "loss": 0.2187,
      "step": 890
    },
    {
      "epoch": 0.6854531607006854,
      "grad_norm": 9.493856956533788,
      "learning_rate": 1.640685892439595e-06,
      "loss": 0.2125,
      "step": 900
    },
    {
      "epoch": 0.693069306930693,
      "grad_norm": 27.037214056603638,
      "learning_rate": 1.601714731098987e-06,
      "loss": 0.2317,
      "step": 910
    },
    {
      "epoch": 0.7006854531607006,
      "grad_norm": 4.472164841493445,
      "learning_rate": 1.562743569758379e-06,
      "loss": 0.2044,
      "step": 920
    },
    {
      "epoch": 0.7083015993907082,
      "grad_norm": 16.172978825850585,
      "learning_rate": 1.523772408417771e-06,
      "loss": 0.1894,
      "step": 930
    },
    {
      "epoch": 0.715917745620716,
      "grad_norm": 4.881279517002309,
      "learning_rate": 1.4848012470771631e-06,
      "loss": 0.2003,
      "step": 940
    },
    {
      "epoch": 0.7235338918507236,
      "grad_norm": 3.2541774506080903,
      "learning_rate": 1.445830085736555e-06,
      "loss": 0.19,
      "step": 950
    },
    {
      "epoch": 0.7311500380807312,
      "grad_norm": 4.81369927464371,
      "learning_rate": 1.4068589243959471e-06,
      "loss": 0.2147,
      "step": 960
    },
    {
      "epoch": 0.7387661843107388,
      "grad_norm": 4.714739168817072,
      "learning_rate": 1.3678877630553392e-06,
      "loss": 0.216,
      "step": 970
    },
    {
      "epoch": 0.7463823305407464,
      "grad_norm": 17.223534266258675,
      "learning_rate": 1.3289166017147311e-06,
      "loss": 0.1985,
      "step": 980
    },
    {
      "epoch": 0.753998476770754,
      "grad_norm": 8.493294082063118,
      "learning_rate": 1.2899454403741232e-06,
      "loss": 0.1977,
      "step": 990
    },
    {
      "epoch": 0.7616146230007617,
      "grad_norm": 9.836008899951883,
      "learning_rate": 1.2509742790335153e-06,
      "loss": 0.2164,
      "step": 1000
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 6.722203243193697,
      "learning_rate": 1.2120031176929072e-06,
      "loss": 0.1702,
      "step": 1010
    },
    {
      "epoch": 0.7768469154607769,
      "grad_norm": 6.285912185515779,
      "learning_rate": 1.1730319563522993e-06,
      "loss": 0.1839,
      "step": 1020
    },
    {
      "epoch": 0.7844630616907845,
      "grad_norm": 4.843126546344412,
      "learning_rate": 1.1340607950116914e-06,
      "loss": 0.1821,
      "step": 1030
    },
    {
      "epoch": 0.7920792079207921,
      "grad_norm": 4.402783083650455,
      "learning_rate": 1.0950896336710835e-06,
      "loss": 0.2193,
      "step": 1040
    },
    {
      "epoch": 0.7996953541507997,
      "grad_norm": 3.2568451258156808,
      "learning_rate": 1.0561184723304756e-06,
      "loss": 0.2043,
      "step": 1050
    },
    {
      "epoch": 0.8073115003808073,
      "grad_norm": 161.43472406681008,
      "learning_rate": 1.0171473109898675e-06,
      "loss": 0.1897,
      "step": 1060
    },
    {
      "epoch": 0.814927646610815,
      "grad_norm": 5.379668281357475,
      "learning_rate": 9.781761496492596e-07,
      "loss": 0.2093,
      "step": 1070
    },
    {
      "epoch": 0.8225437928408226,
      "grad_norm": 13.223854900370439,
      "learning_rate": 9.392049883086516e-07,
      "loss": 0.2536,
      "step": 1080
    },
    {
      "epoch": 0.8301599390708302,
      "grad_norm": 2.5590636509915097,
      "learning_rate": 9.002338269680437e-07,
      "loss": 0.1686,
      "step": 1090
    },
    {
      "epoch": 0.8377760853008378,
      "grad_norm": 2.4022953941248506,
      "learning_rate": 8.612626656274357e-07,
      "loss": 0.1858,
      "step": 1100
    },
    {
      "epoch": 0.8453922315308454,
      "grad_norm": 5.610490607350316,
      "learning_rate": 8.222915042868277e-07,
      "loss": 0.2067,
      "step": 1110
    },
    {
      "epoch": 0.853008377760853,
      "grad_norm": 2.6949684335315385,
      "learning_rate": 7.833203429462198e-07,
      "loss": 0.2081,
      "step": 1120
    },
    {
      "epoch": 0.8606245239908606,
      "grad_norm": 2.363658845139482,
      "learning_rate": 7.443491816056118e-07,
      "loss": 0.2167,
      "step": 1130
    },
    {
      "epoch": 0.8682406702208683,
      "grad_norm": 54.850194962785416,
      "learning_rate": 7.05378020265004e-07,
      "loss": 0.1661,
      "step": 1140
    },
    {
      "epoch": 0.8758568164508759,
      "grad_norm": 2.4702676793543312,
      "learning_rate": 6.66406858924396e-07,
      "loss": 0.1837,
      "step": 1150
    },
    {
      "epoch": 0.8834729626808835,
      "grad_norm": 1.291212127209575,
      "learning_rate": 6.27435697583788e-07,
      "loss": 0.1639,
      "step": 1160
    },
    {
      "epoch": 0.8910891089108911,
      "grad_norm": 4.989079218052827,
      "learning_rate": 5.884645362431801e-07,
      "loss": 0.1941,
      "step": 1170
    },
    {
      "epoch": 0.8987052551408987,
      "grad_norm": 3.934798653103373,
      "learning_rate": 5.494933749025721e-07,
      "loss": 0.2115,
      "step": 1180
    },
    {
      "epoch": 0.9063214013709063,
      "grad_norm": 8.76267158776298,
      "learning_rate": 5.105222135619641e-07,
      "loss": 0.1888,
      "step": 1190
    },
    {
      "epoch": 0.913937547600914,
      "grad_norm": 37.34043564266434,
      "learning_rate": 4.715510522213562e-07,
      "loss": 0.1932,
      "step": 1200
    },
    {
      "epoch": 0.9215536938309216,
      "grad_norm": 5.936087774353652,
      "learning_rate": 4.3257989088074824e-07,
      "loss": 0.201,
      "step": 1210
    },
    {
      "epoch": 0.9291698400609292,
      "grad_norm": 3.2120948066407005,
      "learning_rate": 3.9360872954014035e-07,
      "loss": 0.2004,
      "step": 1220
    },
    {
      "epoch": 0.9367859862909368,
      "grad_norm": 2.2714590166034525,
      "learning_rate": 3.546375681995324e-07,
      "loss": 0.1936,
      "step": 1230
    },
    {
      "epoch": 0.9444021325209444,
      "grad_norm": 3.1017617351534406,
      "learning_rate": 3.1566640685892445e-07,
      "loss": 0.1925,
      "step": 1240
    },
    {
      "epoch": 0.952018278750952,
      "grad_norm": 3.814780236337704,
      "learning_rate": 2.7669524551831645e-07,
      "loss": 0.2221,
      "step": 1250
    },
    {
      "epoch": 0.9596344249809596,
      "grad_norm": 7.0716534670331,
      "learning_rate": 2.3772408417770852e-07,
      "loss": 0.1985,
      "step": 1260
    },
    {
      "epoch": 0.9672505712109672,
      "grad_norm": 48.227925344783095,
      "learning_rate": 1.9875292283710057e-07,
      "loss": 0.1864,
      "step": 1270
    },
    {
      "epoch": 0.9748667174409749,
      "grad_norm": 4.310515199128544,
      "learning_rate": 1.597817614964926e-07,
      "loss": 0.2306,
      "step": 1280
    },
    {
      "epoch": 0.9824828636709825,
      "grad_norm": 5.071672951133308,
      "learning_rate": 1.2081060015588465e-07,
      "loss": 0.1761,
      "step": 1290
    },
    {
      "epoch": 0.9900990099009901,
      "grad_norm": 3.32925400548507,
      "learning_rate": 8.18394388152767e-08,
      "loss": 0.1799,
      "step": 1300
    },
    {
      "epoch": 0.9977151561309977,
      "grad_norm": 3.1779605362401053,
      "learning_rate": 4.286827747466875e-08,
      "loss": 0.1876,
      "step": 1310
    }
  ],
  "logging_steps": 10,
  "max_steps": 1313,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 300,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1244660420050944.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}