{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 2698,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.007414100692755034,
      "grad_norm": 32.06580537112801,
      "learning_rate": 7.037037037037038e-07,
      "loss": 3.3934,
      "step": 20
    },
    {
      "epoch": 0.014828201385510068,
      "grad_norm": 3.3326163294817954,
      "learning_rate": 1.4444444444444445e-06,
      "loss": 1.2307,
      "step": 40
    },
    {
      "epoch": 0.0222423020782651,
      "grad_norm": 4.290780419763869,
      "learning_rate": 2.1851851851851853e-06,
      "loss": 0.4061,
      "step": 60
    },
    {
      "epoch": 0.029656402771020135,
      "grad_norm": 2.6943584691235576,
      "learning_rate": 2.9259259259259257e-06,
      "loss": 0.3405,
      "step": 80
    },
    {
      "epoch": 0.03707050346377517,
      "grad_norm": 2.266412502759934,
      "learning_rate": 3.6666666666666666e-06,
      "loss": 0.3194,
      "step": 100
    },
    {
      "epoch": 0.0444846041565302,
      "grad_norm": 2.0806674330352632,
      "learning_rate": 4.407407407407408e-06,
      "loss": 0.3032,
      "step": 120
    },
    {
      "epoch": 0.051898704849285236,
      "grad_norm": 1.8472539326338435,
      "learning_rate": 5.148148148148149e-06,
      "loss": 0.309,
      "step": 140
    },
    {
      "epoch": 0.05931280554204027,
      "grad_norm": 1.7376245233341043,
      "learning_rate": 5.88888888888889e-06,
      "loss": 0.2961,
      "step": 160
    },
    {
      "epoch": 0.0667269062347953,
      "grad_norm": 1.516283408516978,
      "learning_rate": 6.62962962962963e-06,
      "loss": 0.2912,
      "step": 180
    },
    {
      "epoch": 0.07414100692755034,
      "grad_norm": 1.2314417372784106,
      "learning_rate": 7.370370370370371e-06,
      "loss": 0.2912,
      "step": 200
    },
    {
      "epoch": 0.08155510762030536,
      "grad_norm": 1.4472634085041824,
      "learning_rate": 8.111111111111112e-06,
      "loss": 0.3035,
      "step": 220
    },
    {
      "epoch": 0.0889692083130604,
      "grad_norm": 1.2659145795134694,
      "learning_rate": 8.851851851851853e-06,
      "loss": 0.2862,
      "step": 240
    },
    {
      "epoch": 0.09638330900581543,
      "grad_norm": 1.2890767825393588,
      "learning_rate": 9.592592592592593e-06,
      "loss": 0.293,
      "step": 260
    },
    {
      "epoch": 0.10379740969857047,
      "grad_norm": 3.448944854479974,
      "learning_rate": 9.999660982200767e-06,
      "loss": 0.3,
      "step": 280
    },
    {
      "epoch": 0.1112115103913255,
      "grad_norm": 1.2378181628212996,
      "learning_rate": 9.996480447662715e-06,
      "loss": 0.29,
      "step": 300
    },
    {
      "epoch": 0.11862561108408054,
      "grad_norm": 1.3427597598729493,
      "learning_rate": 9.989954095012863e-06,
      "loss": 0.2937,
      "step": 320
    },
    {
      "epoch": 0.12603971177683557,
      "grad_norm": 0.975575976420179,
      "learning_rate": 9.980086294525275e-06,
      "loss": 0.2921,
      "step": 340
    },
    {
      "epoch": 0.1334538124695906,
      "grad_norm": 0.9658470351220226,
      "learning_rate": 9.966883654024384e-06,
      "loss": 0.285,
      "step": 360
    },
    {
      "epoch": 0.14086791316234565,
      "grad_norm": 0.907938463759864,
      "learning_rate": 9.950355014460171e-06,
      "loss": 0.2803,
      "step": 380
    },
    {
      "epoch": 0.14828201385510068,
      "grad_norm": 1.0837571176440133,
      "learning_rate": 9.93051144398795e-06,
      "loss": 0.2774,
      "step": 400
    },
    {
      "epoch": 0.1556961145478557,
      "grad_norm": 1.0159088629816762,
      "learning_rate": 9.907366230556745e-06,
      "loss": 0.2796,
      "step": 420
    },
    {
      "epoch": 0.16311021524061073,
      "grad_norm": 0.8347916457344657,
      "learning_rate": 9.880934873011216e-06,
      "loss": 0.2797,
      "step": 440
    },
    {
      "epoch": 0.17052431593336578,
      "grad_norm": 1.1330585238504898,
      "learning_rate": 9.851235070713086e-06,
      "loss": 0.281,
      "step": 460
    },
    {
      "epoch": 0.1779384166261208,
      "grad_norm": 0.9622555723755646,
      "learning_rate": 9.818286711689036e-06,
      "loss": 0.2795,
      "step": 480
    },
    {
      "epoch": 0.18535251731887584,
      "grad_norm": 0.99111564845969,
      "learning_rate": 9.78211185931298e-06,
      "loss": 0.2687,
      "step": 500
    },
    {
      "epoch": 0.19276661801163086,
      "grad_norm": 0.9679547178989175,
      "learning_rate": 9.742734737531662e-06,
      "loss": 0.2707,
      "step": 520
    },
    {
      "epoch": 0.2001807187043859,
      "grad_norm": 1.4672952066714184,
      "learning_rate": 9.700181714643453e-06,
      "loss": 0.2657,
      "step": 540
    },
    {
      "epoch": 0.20759481939714095,
      "grad_norm": 0.9454408078270417,
      "learning_rate": 9.654481285641214e-06,
      "loss": 0.2834,
      "step": 560
    },
    {
      "epoch": 0.21500892008989597,
      "grad_norm": 2.185790922947196,
      "learning_rate": 9.605664053131058e-06,
      "loss": 0.263,
      "step": 580
    },
    {
      "epoch": 0.222423020782651,
      "grad_norm": 0.8118181314852884,
      "learning_rate": 9.553762706839763e-06,
      "loss": 0.2706,
      "step": 600
    },
    {
      "epoch": 0.22983712147540603,
      "grad_norm": 0.9392583618462029,
      "learning_rate": 9.498812001724594e-06,
      "loss": 0.2614,
      "step": 620
    },
    {
      "epoch": 0.23725122216816108,
      "grad_norm": 0.6948312735577339,
      "learning_rate": 9.440848734700175e-06,
      "loss": 0.2621,
      "step": 640
    },
    {
      "epoch": 0.2446653228609161,
      "grad_norm": 0.8256442560540643,
      "learning_rate": 9.379911719997975e-06,
      "loss": 0.2617,
      "step": 660
    },
    {
      "epoch": 0.25207942355367113,
      "grad_norm": 0.7805987945354576,
      "learning_rate": 9.316041763174946e-06,
      "loss": 0.2643,
      "step": 680
    },
    {
      "epoch": 0.2594935242464262,
      "grad_norm": 0.8124960825173237,
      "learning_rate": 9.249281633788692e-06,
      "loss": 0.2624,
      "step": 700
    },
    {
      "epoch": 0.2669076249391812,
      "grad_norm": 0.7823012396587974,
      "learning_rate": 9.179676036757488e-06,
      "loss": 0.2659,
      "step": 720
    },
    {
      "epoch": 0.27432172563193624,
      "grad_norm": 0.7555442715531621,
      "learning_rate": 9.107271582424284e-06,
      "loss": 0.2644,
      "step": 740
    },
    {
      "epoch": 0.2817358263246913,
      "grad_norm": 1.1171738106079774,
      "learning_rate": 9.032116755344804e-06,
      "loss": 0.2512,
      "step": 760
    },
    {
      "epoch": 0.2891499270174463,
      "grad_norm": 0.7798366005214705,
      "learning_rate": 8.954261881820584e-06,
      "loss": 0.2647,
      "step": 780
    },
    {
      "epoch": 0.29656402771020135,
      "grad_norm": 0.8245286080721071,
      "learning_rate": 8.873759096198724e-06,
      "loss": 0.2578,
      "step": 800
    },
    {
      "epoch": 0.30397812840295635,
      "grad_norm": 0.648298467437291,
      "learning_rate": 8.790662305960899e-06,
      "loss": 0.2556,
      "step": 820
    },
    {
      "epoch": 0.3113922290957114,
      "grad_norm": 0.7821424784357656,
      "learning_rate": 8.705027155625034e-06,
      "loss": 0.2465,
      "step": 840
    },
    {
      "epoch": 0.31880632978846646,
      "grad_norm": 0.7859052019683617,
      "learning_rate": 8.616910989483773e-06,
      "loss": 0.2624,
      "step": 860
    },
    {
      "epoch": 0.32622043048122146,
      "grad_norm": 0.75339069893752,
      "learning_rate": 8.526372813204749e-06,
      "loss": 0.2511,
      "step": 880
    },
    {
      "epoch": 0.3336345311739765,
      "grad_norm": 1.4580418599743148,
      "learning_rate": 8.433473254318313e-06,
      "loss": 0.2571,
      "step": 900
    },
    {
      "epoch": 0.34104863186673157,
      "grad_norm": 0.8481013751282837,
      "learning_rate": 8.33827452161923e-06,
      "loss": 0.2512,
      "step": 920
    },
    {
      "epoch": 0.34846273255948657,
      "grad_norm": 0.8089440081418998,
      "learning_rate": 8.240840363509482e-06,
      "loss": 0.2549,
      "step": 940
    },
    {
      "epoch": 0.3558768332522416,
      "grad_norm": 0.7352455790133732,
      "learning_rate": 8.141236025310118e-06,
      "loss": 0.2594,
      "step": 960
    },
    {
      "epoch": 0.3632909339449966,
      "grad_norm": 1.0831597837201794,
      "learning_rate": 8.039528205570695e-06,
      "loss": 0.2444,
      "step": 980
    },
    {
      "epoch": 0.3707050346377517,
      "grad_norm": 0.7585612413284839,
      "learning_rate": 7.935785011405605e-06,
      "loss": 0.2478,
      "step": 1000
    },
    {
      "epoch": 0.37811913533050673,
      "grad_norm": 0.727649147663723,
      "learning_rate": 7.83007591288716e-06,
      "loss": 0.2523,
      "step": 1020
    },
    {
      "epoch": 0.3855332360232617,
      "grad_norm": 0.7787866088201918,
      "learning_rate": 7.722471696526007e-06,
      "loss": 0.2589,
      "step": 1040
    },
    {
      "epoch": 0.3929473367160168,
      "grad_norm": 0.7765136850082168,
      "learning_rate": 7.613044417870006e-06,
      "loss": 0.2491,
      "step": 1060
    },
    {
      "epoch": 0.4003614374087718,
      "grad_norm": 0.690468569324881,
      "learning_rate": 7.501867353253301e-06,
      "loss": 0.2462,
      "step": 1080
    },
    {
      "epoch": 0.40777553810152684,
      "grad_norm": 0.7182183611659085,
      "learning_rate": 7.389014950727938e-06,
      "loss": 0.2582,
      "step": 1100
    },
    {
      "epoch": 0.4151896387942819,
      "grad_norm": 0.6634951338167964,
      "learning_rate": 7.274562780210837e-06,
      "loss": 0.2399,
      "step": 1120
    },
    {
      "epoch": 0.4226037394870369,
      "grad_norm": 0.7930541091163981,
      "learning_rate": 7.158587482879542e-06,
      "loss": 0.2419,
      "step": 1140
    },
    {
      "epoch": 0.43001784017979194,
      "grad_norm": 0.7591718992979861,
      "learning_rate": 7.041166719850585e-06,
      "loss": 0.2361,
      "step": 1160
    },
    {
      "epoch": 0.437431940872547,
      "grad_norm": 0.657480629189675,
      "learning_rate": 6.922379120174911e-06,
      "loss": 0.2426,
      "step": 1180
    },
    {
      "epoch": 0.444846041565302,
      "grad_norm": 0.856188180561323,
      "learning_rate": 6.802304228185097e-06,
      "loss": 0.2499,
      "step": 1200
    },
    {
      "epoch": 0.45226014225805705,
      "grad_norm": 0.7103672994416902,
      "learning_rate": 6.6810224502296924e-06,
      "loss": 0.2431,
      "step": 1220
    },
    {
      "epoch": 0.45967424295081205,
      "grad_norm": 0.7655598824506112,
      "learning_rate": 6.55861500083031e-06,
      "loss": 0.2475,
      "step": 1240
    },
    {
      "epoch": 0.4670883436435671,
      "grad_norm": 0.7285887684926099,
      "learning_rate": 6.435163848297549e-06,
      "loss": 0.2434,
      "step": 1260
    },
    {
      "epoch": 0.47450244433632216,
      "grad_norm": 0.7404441533804746,
      "learning_rate": 6.310751659842136e-06,
      "loss": 0.2446,
      "step": 1280
    },
    {
      "epoch": 0.48191654502907716,
      "grad_norm": 0.7192394640726304,
      "learning_rate": 6.185461746218072e-06,
      "loss": 0.2402,
      "step": 1300
    },
    {
      "epoch": 0.4893306457218322,
      "grad_norm": 0.6374798064639219,
      "learning_rate": 6.05937800593485e-06,
      "loss": 0.24,
      "step": 1320
    },
    {
      "epoch": 0.49674474641458727,
      "grad_norm": 0.7031390506652865,
      "learning_rate": 5.932584869076049e-06,
      "loss": 0.2493,
      "step": 1340
    },
    {
      "epoch": 0.5041588471073423,
      "grad_norm": 0.7586530636558019,
      "learning_rate": 5.8051672407620295e-06,
      "loss": 0.2351,
      "step": 1360
    },
    {
      "epoch": 0.5115729478000973,
      "grad_norm": 0.7281460672456646,
      "learning_rate": 5.6772104442944674e-06,
      "loss": 0.2434,
      "step": 1380
    },
    {
      "epoch": 0.5189870484928524,
      "grad_norm": 0.7585358565709085,
      "learning_rate": 5.548800164020905e-06,
      "loss": 0.2427,
      "step": 1400
    },
    {
      "epoch": 0.5264011491856074,
      "grad_norm": 0.6534120478200554,
      "learning_rate": 5.420022387957501e-06,
      "loss": 0.2408,
      "step": 1420
    },
    {
      "epoch": 0.5338152498783624,
      "grad_norm": 1.0252998400955873,
      "learning_rate": 5.290963350208455e-06,
      "loss": 0.2353,
      "step": 1440
    },
    {
      "epoch": 0.5412293505711174,
      "grad_norm": 0.7331074205501853,
      "learning_rate": 5.161709473220628e-06,
      "loss": 0.237,
      "step": 1460
    },
    {
      "epoch": 0.5486434512638725,
      "grad_norm": 0.6449563139414835,
      "learning_rate": 5.032347309912058e-06,
      "loss": 0.2418,
      "step": 1480
    },
    {
      "epoch": 0.5560575519566275,
      "grad_norm": 0.6315702237924259,
      "learning_rate": 4.902963485713087e-06,
      "loss": 0.2381,
      "step": 1500
    },
    {
      "epoch": 0.5634716526493826,
      "grad_norm": 0.6051123125654648,
      "learning_rate": 4.773644640558944e-06,
      "loss": 0.2388,
      "step": 1520
    },
    {
      "epoch": 0.5708857533421375,
      "grad_norm": 0.6580102590702694,
      "learning_rate": 4.644477370872627e-06,
      "loss": 0.2388,
      "step": 1540
    },
    {
      "epoch": 0.5782998540348926,
      "grad_norm": 0.7237643321863204,
      "learning_rate": 4.515548171576886e-06,
      "loss": 0.2363,
      "step": 1560
    },
    {
      "epoch": 0.5857139547276476,
      "grad_norm": 0.6923389676943642,
      "learning_rate": 4.386943378174214e-06,
      "loss": 0.236,
      "step": 1580
    },
    {
      "epoch": 0.5931280554204027,
      "grad_norm": 0.6934233196397561,
      "learning_rate": 4.258749108933571e-06,
      "loss": 0.2369,
      "step": 1600
    },
    {
      "epoch": 0.6005421561131578,
      "grad_norm": 0.6182921314601578,
      "learning_rate": 4.131051207222571e-06,
      "loss": 0.2387,
      "step": 1620
    },
    {
      "epoch": 0.6079562568059127,
      "grad_norm": 0.7468100667443839,
      "learning_rate": 4.003935184023768e-06,
      "loss": 0.2479,
      "step": 1640
    },
    {
      "epoch": 0.6153703574986678,
      "grad_norm": 0.623631126507463,
      "learning_rate": 3.877486160673502e-06,
      "loss": 0.235,
      "step": 1660
    },
    {
      "epoch": 0.6227844581914228,
      "grad_norm": 0.7107052234543987,
      "learning_rate": 3.7517888118616853e-06,
      "loss": 0.2346,
      "step": 1680
    },
    {
      "epoch": 0.6301985588841779,
      "grad_norm": 0.8029055529060507,
      "learning_rate": 3.626927308930658e-06,
      "loss": 0.2457,
      "step": 1700
    },
    {
      "epoch": 0.6376126595769329,
      "grad_norm": 0.6874508607391684,
      "learning_rate": 3.5029852635111172e-06,
      "loss": 0.2308,
      "step": 1720
    },
    {
      "epoch": 0.6450267602696879,
      "grad_norm": 0.6908040428007661,
      "learning_rate": 3.380045671532839e-06,
      "loss": 0.2329,
      "step": 1740
    },
    {
      "epoch": 0.6524408609624429,
      "grad_norm": 0.5547457031999729,
      "learning_rate": 3.2581908576477127e-06,
      "loss": 0.2334,
      "step": 1760
    },
    {
      "epoch": 0.659854961655198,
      "grad_norm": 0.6349433771354605,
      "learning_rate": 3.1375024201022563e-06,
      "loss": 0.2349,
      "step": 1780
    },
    {
      "epoch": 0.667269062347953,
      "grad_norm": 0.6010959441423943,
      "learning_rate": 3.0180611760965877e-06,
      "loss": 0.2297,
      "step": 1800
    },
    {
      "epoch": 0.6746831630407081,
      "grad_norm": 0.699521154565116,
      "learning_rate": 2.8999471076663967e-06,
      "loss": 0.2367,
      "step": 1820
    },
    {
      "epoch": 0.6820972637334631,
      "grad_norm": 0.6187654276233007,
      "learning_rate": 2.783239308124178e-06,
      "loss": 0.231,
      "step": 1840
    },
    {
      "epoch": 0.6895113644262181,
      "grad_norm": 0.6044970045544199,
      "learning_rate": 2.6680159290955777e-06,
      "loss": 0.2285,
      "step": 1860
    },
    {
      "epoch": 0.6969254651189731,
      "grad_norm": 0.7284350015662899,
      "learning_rate": 2.5543541281863334e-06,
      "loss": 0.2365,
      "step": 1880
    },
    {
      "epoch": 0.7043395658117282,
      "grad_norm": 0.7220795849243428,
      "learning_rate": 2.4423300173148407e-06,
      "loss": 0.2293,
      "step": 1900
    },
    {
      "epoch": 0.7117536665044832,
      "grad_norm": 0.6987651618762949,
      "learning_rate": 2.332018611744952e-06,
      "loss": 0.2317,
      "step": 1920
    },
    {
      "epoch": 0.7191677671972383,
      "grad_norm": 0.6900248322430578,
      "learning_rate": 2.22349377985313e-06,
      "loss": 0.2212,
      "step": 1940
    },
    {
      "epoch": 0.7265818678899932,
      "grad_norm": 0.6517311748768397,
      "learning_rate": 2.1168281936635964e-06,
      "loss": 0.2265,
      "step": 1960
    },
    {
      "epoch": 0.7339959685827483,
      "grad_norm": 0.6933857965660299,
      "learning_rate": 2.012093280184607e-06,
      "loss": 0.2192,
      "step": 1980
    },
    {
      "epoch": 0.7414100692755033,
      "grad_norm": 0.6690150882563629,
      "learning_rate": 1.909359173578424e-06,
      "loss": 0.2311,
      "step": 2000
    },
    {
      "epoch": 0.7488241699682584,
      "grad_norm": 0.6880821782327176,
      "learning_rate": 1.8086946681970274e-06,
      "loss": 0.223,
      "step": 2020
    },
    {
      "epoch": 0.7562382706610135,
      "grad_norm": 0.7671933520011451,
      "learning_rate": 1.710167172515011e-06,
      "loss": 0.2266,
      "step": 2040
    },
    {
      "epoch": 0.7636523713537684,
      "grad_norm": 0.7073742558100695,
      "learning_rate": 1.6138426639905163e-06,
      "loss": 0.2262,
      "step": 2060
    },
    {
      "epoch": 0.7710664720465235,
      "grad_norm": 0.6602266104494168,
      "learning_rate": 1.5197856448844e-06,
      "loss": 0.2256,
      "step": 2080
    },
    {
      "epoch": 0.7784805727392785,
      "grad_norm": 0.6504469232479442,
      "learning_rate": 1.4280590990672827e-06,
      "loss": 0.2259,
      "step": 2100
    },
    {
      "epoch": 0.7858946734320336,
      "grad_norm": 0.6423895377225716,
      "learning_rate": 1.3387244498433389e-06,
      "loss": 0.2286,
      "step": 2120
    },
    {
      "epoch": 0.7933087741247886,
      "grad_norm": 0.6849903734668334,
      "learning_rate": 1.2518415188191008e-06,
      "loss": 0.2236,
      "step": 2140
    },
    {
      "epoch": 0.8007228748175436,
      "grad_norm": 0.6536439344710893,
      "learning_rate": 1.1674684858448164e-06,
      "loss": 0.225,
      "step": 2160
    },
    {
      "epoch": 0.8081369755102986,
      "grad_norm": 0.8214938667433946,
      "learning_rate": 1.085661850055193e-06,
      "loss": 0.2276,
      "step": 2180
    },
    {
      "epoch": 0.8155510762030537,
      "grad_norm": 0.7409010792932056,
      "learning_rate": 1.006476392035598e-06,
      "loss": 0.2188,
      "step": 2200
    },
    {
      "epoch": 0.8229651768958087,
      "grad_norm": 0.6696568442942546,
      "learning_rate": 9.299651371390589e-07,
      "loss": 0.2172,
      "step": 2220
    },
    {
      "epoch": 0.8303792775885638,
      "grad_norm": 0.6913664887131121,
      "learning_rate": 8.561793199786339e-07,
      "loss": 0.2243,
      "step": 2240
    },
    {
      "epoch": 0.8377933782813188,
      "grad_norm": 0.6651230207577203,
      "learning_rate": 7.8516835011892e-07,
      "loss": 0.2271,
      "step": 2260
    },
    {
      "epoch": 0.8452074789740738,
      "grad_norm": 0.6718428566405614,
      "learning_rate": 7.169797789896876e-07,
      "loss": 0.2237,
      "step": 2280
    },
    {
      "epoch": 0.8526215796668288,
      "grad_norm": 0.7501096157117769,
      "learning_rate": 6.516592680437717e-07,
      "loss": 0.2236,
      "step": 2300
    },
    {
      "epoch": 0.8600356803595839,
      "grad_norm": 0.625025697061966,
      "learning_rate": 5.892505581805669e-07,
      "loss": 0.2216,
      "step": 2320
    },
    {
      "epoch": 0.8674497810523389,
      "grad_norm": 0.6238881264357159,
      "learning_rate": 5.297954404555944e-07,
      "loss": 0.2232,
      "step": 2340
    },
    {
      "epoch": 0.874863881745094,
      "grad_norm": 0.5852724716452957,
      "learning_rate": 4.7333372809574616e-07,
      "loss": 0.2279,
      "step": 2360
    },
    {
      "epoch": 0.8822779824378489,
      "grad_norm": 0.6481022641381101,
      "learning_rate": 4.199032298389549e-07,
      "loss": 0.2232,
      "step": 2380
    },
    {
      "epoch": 0.889692083130604,
      "grad_norm": 0.7060913539739083,
      "learning_rate": 3.6953972461614364e-07,
      "loss": 0.2217,
      "step": 2400
    },
    {
      "epoch": 0.897106183823359,
      "grad_norm": 0.6677018339279599,
      "learning_rate": 3.2227693759239785e-07,
      "loss": 0.2161,
      "step": 2420
    },
    {
      "epoch": 0.9045202845161141,
      "grad_norm": 0.6581848099265243,
      "learning_rate": 2.781465175834186e-07,
      "loss": 0.2228,
      "step": 2440
    },
    {
      "epoch": 0.9119343852088692,
      "grad_norm": 0.7113437251505971,
      "learning_rate": 2.3717801586236544e-07,
      "loss": 0.2144,
      "step": 2460
    },
    {
      "epoch": 0.9193484859016241,
      "grad_norm": 0.6220302232780794,
      "learning_rate": 1.9939886637129145e-07,
      "loss": 0.2173,
      "step": 2480
    },
    {
      "epoch": 0.9267625865943792,
      "grad_norm": 0.6691128505437876,
      "learning_rate": 1.648343673504166e-07,
      "loss": 0.2184,
      "step": 2500
    },
    {
      "epoch": 0.9341766872871342,
      "grad_norm": 0.6487345437674443,
      "learning_rate": 1.3350766439753914e-07,
      "loss": 0.2193,
      "step": 2520
    },
    {
      "epoch": 0.9415907879798893,
      "grad_norm": 0.7736974881210988,
      "learning_rate": 1.0543973496893401e-07,
      "loss": 0.2232,
      "step": 2540
    },
    {
      "epoch": 0.9490048886726443,
      "grad_norm": 0.7116280365195675,
      "learning_rate": 8.064937433211518e-08,
      "loss": 0.22,
      "step": 2560
    },
    {
      "epoch": 0.9564189893653993,
      "grad_norm": 0.6630183579839392,
      "learning_rate": 5.91531829798675e-08,
      "loss": 0.2249,
      "step": 2580
    },
    {
      "epoch": 0.9638330900581543,
      "grad_norm": 0.6777889817351815,
      "learning_rate": 4.0965555513974297e-08,
      "loss": 0.2203,
      "step": 2600
    },
    {
      "epoch": 0.9712471907509094,
      "grad_norm": 0.6925637296642155,
      "learning_rate": 2.6098671006087873e-08,
      "loss": 0.2224,
      "step": 2620
    },
    {
      "epoch": 0.9786612914436644,
      "grad_norm": 0.736161972188353,
      "learning_rate": 1.4562484842199354e-08,
      "loss": 0.2204,
      "step": 2640
    },
    {
      "epoch": 0.9860753921364195,
      "grad_norm": 0.6366143976132524,
      "learning_rate": 6.3647220561624535e-09,
      "loss": 0.222,
      "step": 2660
    },
    {
      "epoch": 0.9934894928291745,
      "grad_norm": 0.6355076829280015,
      "learning_rate": 1.5108721567419493e-09,
      "loss": 0.2176,
      "step": 2680
    },
    {
      "epoch": 1.0,
      "step": 2698,
      "total_flos": 1877022766792704.0,
      "train_loss": 0.2793045342631301,
      "train_runtime": 58447.463,
      "train_samples_per_second": 5.908,
      "train_steps_per_second": 0.046
    }
  ],
  "logging_steps": 20,
  "max_steps": 2698,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1877022766792704.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}