{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 19.995409685563462,
  "eval_steps": 500,
  "global_step": 21780,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.09180628873077806,
      "grad_norm": 0.3286115229129791,
      "learning_rate": 0.0003,
      "loss": 1.8271,
      "step": 100
    },
    {
      "epoch": 0.18361257746155613,
      "grad_norm": 0.401665598154068,
      "learning_rate": 0.0003,
      "loss": 1.7773,
      "step": 200
    },
    {
      "epoch": 0.2754188661923342,
      "grad_norm": 0.27676922082901,
      "learning_rate": 0.0003,
      "loss": 1.7687,
      "step": 300
    },
    {
      "epoch": 0.36722515492311225,
      "grad_norm": 0.3495130240917206,
      "learning_rate": 0.0003,
      "loss": 1.7648,
      "step": 400
    },
    {
      "epoch": 0.4590314436538903,
      "grad_norm": 0.3620341718196869,
      "learning_rate": 0.0003,
      "loss": 1.7494,
      "step": 500
    },
    {
      "epoch": 0.5508377323846684,
      "grad_norm": 0.3991380035877228,
      "learning_rate": 0.0003,
      "loss": 1.7729,
      "step": 600
    },
    {
      "epoch": 0.6426440211154464,
      "grad_norm": 0.4441532492637634,
      "learning_rate": 0.0003,
      "loss": 1.7407,
      "step": 700
    },
    {
      "epoch": 0.7344503098462245,
      "grad_norm": 0.38743725419044495,
      "learning_rate": 0.0003,
      "loss": 1.7537,
      "step": 800
    },
    {
      "epoch": 0.8262565985770025,
      "grad_norm": 0.42725738883018494,
      "learning_rate": 0.0003,
      "loss": 1.7377,
      "step": 900
    },
    {
      "epoch": 0.9180628873077806,
      "grad_norm": 0.43618279695510864,
      "learning_rate": 0.0003,
      "loss": 1.7567,
      "step": 1000
    },
    {
      "epoch": 0.999770484278173,
      "eval_accuracy": 0.5190476190476191,
      "eval_loss": 2.262606382369995,
      "eval_runtime": 5.8101,
      "eval_samples_per_second": 86.057,
      "eval_steps_per_second": 10.843,
      "step": 1089
    },
    {
      "epoch": 1.0098691760385587,
      "grad_norm": 0.4314129650592804,
      "learning_rate": 0.0003,
      "loss": 1.7239,
      "step": 1100
    },
    {
      "epoch": 1.1016754647693368,
      "grad_norm": 0.5108153820037842,
      "learning_rate": 0.0003,
      "loss": 1.5348,
      "step": 1200
    },
    {
      "epoch": 1.1934817535001148,
      "grad_norm": 0.4964566230773926,
      "learning_rate": 0.0003,
      "loss": 1.5236,
      "step": 1300
    },
    {
      "epoch": 1.2852880422308928,
      "grad_norm": 0.4912070035934448,
      "learning_rate": 0.0003,
      "loss": 1.5342,
      "step": 1400
    },
    {
      "epoch": 1.377094330961671,
      "grad_norm": 0.5223585367202759,
      "learning_rate": 0.0003,
      "loss": 1.5405,
      "step": 1500
    },
    {
      "epoch": 1.468900619692449,
      "grad_norm": 0.4836246371269226,
      "learning_rate": 0.0003,
      "loss": 1.546,
      "step": 1600
    },
    {
      "epoch": 1.560706908423227,
      "grad_norm": 0.6014620661735535,
      "learning_rate": 0.0003,
      "loss": 1.5504,
      "step": 1700
    },
    {
      "epoch": 1.652513197154005,
      "grad_norm": 0.47756990790367126,
      "learning_rate": 0.0003,
      "loss": 1.5461,
      "step": 1800
    },
    {
      "epoch": 1.744319485884783,
      "grad_norm": 0.5226470828056335,
      "learning_rate": 0.0003,
      "loss": 1.5502,
      "step": 1900
    },
    {
      "epoch": 1.836125774615561,
      "grad_norm": 0.5086384415626526,
      "learning_rate": 0.0003,
      "loss": 1.5474,
      "step": 2000
    },
    {
      "epoch": 1.9279320633463393,
      "grad_norm": 0.5471687912940979,
      "learning_rate": 0.0003,
      "loss": 1.5617,
      "step": 2100
    },
    {
      "epoch": 1.999540968556346,
      "eval_accuracy": 0.5246031746031746,
      "eval_loss": 2.2435803413391113,
      "eval_runtime": 5.3712,
      "eval_samples_per_second": 93.09,
      "eval_steps_per_second": 11.729,
      "step": 2178
    },
    {
      "epoch": 2.0197383520771175,
      "grad_norm": 0.5887908935546875,
      "learning_rate": 0.0003,
      "loss": 1.4881,
      "step": 2200
    },
    {
      "epoch": 2.1115446408078955,
      "grad_norm": 0.6036558151245117,
      "learning_rate": 0.0003,
      "loss": 1.267,
      "step": 2300
    },
    {
      "epoch": 2.2033509295386735,
      "grad_norm": 0.724539041519165,
      "learning_rate": 0.0003,
      "loss": 1.2882,
      "step": 2400
    },
    {
      "epoch": 2.2951572182694515,
      "grad_norm": 0.6767842173576355,
      "learning_rate": 0.0003,
      "loss": 1.2945,
      "step": 2500
    },
    {
      "epoch": 2.3869635070002295,
      "grad_norm": 0.5520135760307312,
      "learning_rate": 0.0003,
      "loss": 1.3038,
      "step": 2600
    },
    {
      "epoch": 2.4787697957310075,
      "grad_norm": 0.6388158798217773,
      "learning_rate": 0.0003,
      "loss": 1.3132,
      "step": 2700
    },
    {
      "epoch": 2.5705760844617855,
      "grad_norm": 0.6712493896484375,
      "learning_rate": 0.0003,
      "loss": 1.3205,
      "step": 2800
    },
    {
      "epoch": 2.6623823731925635,
      "grad_norm": 0.6082379221916199,
      "learning_rate": 0.0003,
      "loss": 1.3272,
      "step": 2900
    },
    {
      "epoch": 2.754188661923342,
      "grad_norm": 0.6824610233306885,
      "learning_rate": 0.0003,
      "loss": 1.3514,
      "step": 3000
    },
    {
      "epoch": 2.84599495065412,
      "grad_norm": 0.6347182989120483,
      "learning_rate": 0.0003,
      "loss": 1.3454,
      "step": 3100
    },
    {
      "epoch": 2.937801239384898,
      "grad_norm": 0.6638208627700806,
      "learning_rate": 0.0003,
      "loss": 1.343,
      "step": 3200
    },
    {
      "epoch": 2.9993114528345193,
      "eval_accuracy": 0.5236825396825396,
      "eval_loss": 2.3385279178619385,
      "eval_runtime": 5.417,
      "eval_samples_per_second": 92.301,
      "eval_steps_per_second": 11.63,
      "step": 3267
    },
    {
      "epoch": 3.029607528115676,
      "grad_norm": 0.8752885460853577,
      "learning_rate": 0.0003,
      "loss": 1.264,
      "step": 3300
    },
    {
      "epoch": 3.121413816846454,
      "grad_norm": 0.8712926506996155,
      "learning_rate": 0.0003,
      "loss": 1.0576,
      "step": 3400
    },
    {
      "epoch": 3.213220105577232,
      "grad_norm": 0.9298152327537537,
      "learning_rate": 0.0003,
      "loss": 1.0825,
      "step": 3500
    },
    {
      "epoch": 3.30502639430801,
      "grad_norm": 0.9302284717559814,
      "learning_rate": 0.0003,
      "loss": 1.0984,
      "step": 3600
    },
    {
      "epoch": 3.396832683038788,
      "grad_norm": 0.9003168940544128,
      "learning_rate": 0.0003,
      "loss": 1.112,
      "step": 3700
    },
    {
      "epoch": 3.488638971769566,
      "grad_norm": 0.8025063276290894,
      "learning_rate": 0.0003,
      "loss": 1.1348,
      "step": 3800
    },
    {
      "epoch": 3.580445260500344,
      "grad_norm": 0.9571109414100647,
      "learning_rate": 0.0003,
      "loss": 1.1117,
      "step": 3900
    },
    {
      "epoch": 3.672251549231122,
      "grad_norm": 0.8698576092720032,
      "learning_rate": 0.0003,
      "loss": 1.1431,
      "step": 4000
    },
    {
      "epoch": 3.7640578379619005,
      "grad_norm": 0.9028270244598389,
      "learning_rate": 0.0003,
      "loss": 1.1455,
      "step": 4100
    },
    {
      "epoch": 3.8558641266926785,
      "grad_norm": 0.851496696472168,
      "learning_rate": 0.0003,
      "loss": 1.1769,
      "step": 4200
    },
    {
      "epoch": 3.9476704154234565,
      "grad_norm": 0.8378119468688965,
      "learning_rate": 0.0003,
      "loss": 1.1682,
      "step": 4300
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.5215238095238095,
      "eval_loss": 2.499518394470215,
      "eval_runtime": 5.6608,
      "eval_samples_per_second": 88.327,
      "eval_steps_per_second": 11.129,
      "step": 4357
    },
    {
      "epoch": 4.039476704154235,
      "grad_norm": 0.7991979718208313,
      "learning_rate": 0.0003,
      "loss": 1.0485,
      "step": 4400
    },
    {
      "epoch": 4.131282992885013,
      "grad_norm": 0.7637009024620056,
      "learning_rate": 0.0003,
      "loss": 0.8982,
      "step": 4500
    },
    {
      "epoch": 4.223089281615791,
      "grad_norm": 0.8598212003707886,
      "learning_rate": 0.0003,
      "loss": 0.9252,
      "step": 4600
    },
    {
      "epoch": 4.314895570346569,
      "grad_norm": 0.8812803030014038,
      "learning_rate": 0.0003,
      "loss": 0.9372,
      "step": 4700
    },
    {
      "epoch": 4.406701859077347,
      "grad_norm": 0.8750812411308289,
      "learning_rate": 0.0003,
      "loss": 0.9422,
      "step": 4800
    },
    {
      "epoch": 4.498508147808125,
      "grad_norm": 1.0362316370010376,
      "learning_rate": 0.0003,
      "loss": 0.9617,
      "step": 4900
    },
    {
      "epoch": 4.590314436538903,
      "grad_norm": 1.1548477411270142,
      "learning_rate": 0.0003,
      "loss": 0.9721,
      "step": 5000
    },
    {
      "epoch": 4.682120725269681,
      "grad_norm": 1.0806621313095093,
      "learning_rate": 0.0003,
      "loss": 0.9884,
      "step": 5100
    },
    {
      "epoch": 4.773927014000459,
      "grad_norm": 0.8893236517906189,
      "learning_rate": 0.0003,
      "loss": 0.9934,
      "step": 5200
    },
    {
      "epoch": 4.865733302731237,
      "grad_norm": 0.9162046313285828,
      "learning_rate": 0.0003,
      "loss": 1.0058,
      "step": 5300
    },
    {
      "epoch": 4.957539591462015,
      "grad_norm": 0.8382129073143005,
      "learning_rate": 0.0003,
      "loss": 1.0141,
      "step": 5400
    },
    {
      "epoch": 4.999770484278173,
      "eval_accuracy": 0.5181587301587302,
      "eval_loss": 2.6397223472595215,
      "eval_runtime": 5.7176,
      "eval_samples_per_second": 87.45,
      "eval_steps_per_second": 11.019,
      "step": 5446
    },
    {
      "epoch": 5.049345880192793,
      "grad_norm": 0.8659687042236328,
      "learning_rate": 0.0003,
      "loss": 0.8863,
      "step": 5500
    },
    {
      "epoch": 5.141152168923571,
      "grad_norm": 0.9213656187057495,
      "learning_rate": 0.0003,
      "loss": 0.7697,
      "step": 5600
    },
    {
      "epoch": 5.232958457654349,
      "grad_norm": 1.1257935762405396,
      "learning_rate": 0.0003,
      "loss": 0.7966,
      "step": 5700
    },
    {
      "epoch": 5.324764746385127,
      "grad_norm": 0.9488617777824402,
      "learning_rate": 0.0003,
      "loss": 0.7945,
      "step": 5800
    },
    {
      "epoch": 5.416571035115905,
      "grad_norm": 1.0306527614593506,
      "learning_rate": 0.0003,
      "loss": 0.8215,
      "step": 5900
    },
    {
      "epoch": 5.508377323846684,
      "grad_norm": 1.0242127180099487,
      "learning_rate": 0.0003,
      "loss": 0.8387,
      "step": 6000
    },
    {
      "epoch": 5.600183612577462,
      "grad_norm": 1.0761654376983643,
      "learning_rate": 0.0003,
      "loss": 0.8484,
      "step": 6100
    },
    {
      "epoch": 5.69198990130824,
      "grad_norm": 1.1436479091644287,
      "learning_rate": 0.0003,
      "loss": 0.8564,
      "step": 6200
    },
    {
      "epoch": 5.783796190039018,
      "grad_norm": 0.9349524974822998,
      "learning_rate": 0.0003,
      "loss": 0.8736,
      "step": 6300
    },
    {
      "epoch": 5.875602478769796,
      "grad_norm": 0.9073782563209534,
      "learning_rate": 0.0003,
      "loss": 0.8956,
      "step": 6400
    },
    {
      "epoch": 5.967408767500574,
      "grad_norm": 1.085017442703247,
      "learning_rate": 0.0003,
      "loss": 0.9023,
      "step": 6500
    },
    {
      "epoch": 5.999540968556346,
      "eval_accuracy": 0.516984126984127,
      "eval_loss": 2.7929389476776123,
      "eval_runtime": 6.1646,
      "eval_samples_per_second": 81.108,
      "eval_steps_per_second": 10.22,
      "step": 6535
    },
    {
      "epoch": 6.059215056231352,
      "grad_norm": 0.881386935710907,
      "learning_rate": 0.0003,
      "loss": 0.7462,
      "step": 6600
    },
    {
      "epoch": 6.15102134496213,
      "grad_norm": 1.0695873498916626,
      "learning_rate": 0.0003,
      "loss": 0.6751,
      "step": 6700
    },
    {
      "epoch": 6.242827633692908,
      "grad_norm": 0.9458519816398621,
      "learning_rate": 0.0003,
      "loss": 0.7007,
      "step": 6800
    },
    {
      "epoch": 6.334633922423686,
      "grad_norm": 1.045708179473877,
      "learning_rate": 0.0003,
      "loss": 0.7105,
      "step": 6900
    },
    {
      "epoch": 6.426440211154464,
      "grad_norm": 0.9866082668304443,
      "learning_rate": 0.0003,
      "loss": 0.7335,
      "step": 7000
    },
    {
      "epoch": 6.518246499885242,
      "grad_norm": 0.9309425950050354,
      "learning_rate": 0.0003,
      "loss": 0.7354,
      "step": 7100
    },
    {
      "epoch": 6.61005278861602,
      "grad_norm": 0.921173632144928,
      "learning_rate": 0.0003,
      "loss": 0.7593,
      "step": 7200
    },
    {
      "epoch": 6.701859077346798,
      "grad_norm": 1.2775295972824097,
      "learning_rate": 0.0003,
      "loss": 0.7603,
      "step": 7300
    },
    {
      "epoch": 6.793665366077576,
      "grad_norm": 1.1141977310180664,
      "learning_rate": 0.0003,
      "loss": 0.7792,
      "step": 7400
    },
    {
      "epoch": 6.885471654808354,
      "grad_norm": 1.1323295831680298,
      "learning_rate": 0.0003,
      "loss": 0.7966,
      "step": 7500
    },
    {
      "epoch": 6.977277943539132,
      "grad_norm": 1.2780287265777588,
      "learning_rate": 0.0003,
      "loss": 0.8008,
      "step": 7600
    },
    {
      "epoch": 6.999311452834519,
      "eval_accuracy": 0.5162222222222222,
      "eval_loss": 2.8233399391174316,
      "eval_runtime": 5.472,
      "eval_samples_per_second": 91.374,
      "eval_steps_per_second": 11.513,
      "step": 7624
    },
    {
      "epoch": 7.06908423226991,
      "grad_norm": 1.0170750617980957,
      "learning_rate": 0.0003,
      "loss": 0.6451,
      "step": 7700
    },
    {
      "epoch": 7.160890521000688,
      "grad_norm": 0.8980923891067505,
      "learning_rate": 0.0003,
      "loss": 0.6104,
      "step": 7800
    },
    {
      "epoch": 7.252696809731467,
      "grad_norm": 1.0722894668579102,
      "learning_rate": 0.0003,
      "loss": 0.6217,
      "step": 7900
    },
    {
      "epoch": 7.344503098462245,
      "grad_norm": 1.0610452890396118,
      "learning_rate": 0.0003,
      "loss": 0.6405,
      "step": 8000
    },
    {
      "epoch": 7.436309387193023,
      "grad_norm": 0.9819157123565674,
      "learning_rate": 0.0003,
      "loss": 0.6599,
      "step": 8100
    },
    {
      "epoch": 7.528115675923801,
      "grad_norm": 1.0208162069320679,
      "learning_rate": 0.0003,
      "loss": 0.6663,
      "step": 8200
    },
    {
      "epoch": 7.619921964654579,
      "grad_norm": 1.1422516107559204,
      "learning_rate": 0.0003,
      "loss": 0.6826,
      "step": 8300
    },
    {
      "epoch": 7.711728253385357,
      "grad_norm": 1.179495930671692,
      "learning_rate": 0.0003,
      "loss": 0.7054,
      "step": 8400
    },
    {
      "epoch": 7.803534542116135,
      "grad_norm": 1.155306100845337,
      "learning_rate": 0.0003,
      "loss": 0.7126,
      "step": 8500
    },
    {
      "epoch": 7.895340830846913,
      "grad_norm": 1.1133177280426025,
      "learning_rate": 0.0003,
      "loss": 0.7246,
      "step": 8600
    },
    {
      "epoch": 7.987147119577691,
      "grad_norm": 1.1150102615356445,
      "learning_rate": 0.0003,
      "loss": 0.7377,
      "step": 8700
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.5180317460317461,
      "eval_loss": 2.883272886276245,
      "eval_runtime": 5.7501,
      "eval_samples_per_second": 86.955,
      "eval_steps_per_second": 10.956,
      "step": 8714
    },
    {
      "epoch": 8.07895340830847,
      "grad_norm": 0.9108123779296875,
      "learning_rate": 0.0003,
      "loss": 0.5637,
      "step": 8800
    },
    {
      "epoch": 8.170759697039248,
      "grad_norm": 1.1592975854873657,
      "learning_rate": 0.0003,
      "loss": 0.5671,
      "step": 8900
    },
    {
      "epoch": 8.262565985770026,
      "grad_norm": 1.1172692775726318,
      "learning_rate": 0.0003,
      "loss": 0.5708,
      "step": 9000
    },
    {
      "epoch": 8.354372274500804,
      "grad_norm": 1.0453649759292603,
      "learning_rate": 0.0003,
      "loss": 0.587,
      "step": 9100
    },
    {
      "epoch": 8.446178563231582,
      "grad_norm": 1.0275832414627075,
      "learning_rate": 0.0003,
      "loss": 0.6085,
      "step": 9200
    },
    {
      "epoch": 8.53798485196236,
      "grad_norm": 1.0241683721542358,
      "learning_rate": 0.0003,
      "loss": 0.6214,
      "step": 9300
    },
    {
      "epoch": 8.629791140693138,
      "grad_norm": 1.1279902458190918,
      "learning_rate": 0.0003,
      "loss": 0.6232,
      "step": 9400
    },
    {
      "epoch": 8.721597429423916,
      "grad_norm": 1.143254041671753,
      "learning_rate": 0.0003,
      "loss": 0.6446,
      "step": 9500
    },
    {
      "epoch": 8.813403718154694,
      "grad_norm": 1.138592004776001,
      "learning_rate": 0.0003,
      "loss": 0.6472,
      "step": 9600
    },
    {
      "epoch": 8.905210006885472,
      "grad_norm": 1.1288646459579468,
      "learning_rate": 0.0003,
      "loss": 0.6801,
      "step": 9700
    },
    {
      "epoch": 8.99701629561625,
      "grad_norm": 1.1128137111663818,
      "learning_rate": 0.0003,
      "loss": 0.6732,
      "step": 9800
    },
    {
      "epoch": 8.999770484278173,
      "eval_accuracy": 0.5165079365079365,
      "eval_loss": 2.9550092220306396,
      "eval_runtime": 5.5706,
      "eval_samples_per_second": 89.756,
      "eval_steps_per_second": 11.309,
      "step": 9803
    },
    {
      "epoch": 9.088822584347028,
      "grad_norm": 1.0105735063552856,
      "learning_rate": 0.0003,
      "loss": 0.5071,
      "step": 9900
    },
    {
      "epoch": 9.180628873077806,
      "grad_norm": 1.0888274908065796,
      "learning_rate": 0.0003,
      "loss": 0.5164,
      "step": 10000
    },
    {
      "epoch": 9.272435161808584,
      "grad_norm": 1.067857027053833,
      "learning_rate": 0.0003,
      "loss": 0.5311,
      "step": 10100
    },
    {
      "epoch": 9.364241450539362,
      "grad_norm": 1.0484533309936523,
      "learning_rate": 0.0003,
      "loss": 0.5501,
      "step": 10200
    },
    {
      "epoch": 9.45604773927014,
      "grad_norm": 1.0611076354980469,
      "learning_rate": 0.0003,
      "loss": 0.5635,
      "step": 10300
    },
    {
      "epoch": 9.547854028000918,
      "grad_norm": 1.1622737646102905,
      "learning_rate": 0.0003,
      "loss": 0.5713,
      "step": 10400
    },
    {
      "epoch": 9.639660316731696,
      "grad_norm": 1.272035837173462,
      "learning_rate": 0.0003,
      "loss": 0.5933,
      "step": 10500
    },
    {
      "epoch": 9.731466605462474,
      "grad_norm": 1.2226234674453735,
      "learning_rate": 0.0003,
      "loss": 0.5939,
      "step": 10600
    },
    {
      "epoch": 9.823272894193252,
      "grad_norm": 1.197426199913025,
      "learning_rate": 0.0003,
      "loss": 0.6289,
      "step": 10700
    },
    {
      "epoch": 9.91507918292403,
      "grad_norm": 1.2565838098526,
      "learning_rate": 0.0003,
      "loss": 0.6225,
      "step": 10800
    },
    {
      "epoch": 9.999540968556346,
      "eval_accuracy": 0.5165079365079365,
      "eval_loss": 2.9767260551452637,
      "eval_runtime": 5.7303,
      "eval_samples_per_second": 87.255,
      "eval_steps_per_second": 10.994,
      "step": 10892
    },
    {
      "epoch": 10.006885471654808,
      "grad_norm": 1.0739877223968506,
      "learning_rate": 0.0003,
      "loss": 0.6213,
      "step": 10900
    },
    {
      "epoch": 10.098691760385586,
      "grad_norm": 1.0812422037124634,
      "learning_rate": 0.0003,
      "loss": 0.4637,
      "step": 11000
    },
    {
      "epoch": 10.190498049116364,
      "grad_norm": 1.0868052244186401,
      "learning_rate": 0.0003,
      "loss": 0.485,
      "step": 11100
    },
    {
      "epoch": 10.282304337847142,
      "grad_norm": 1.1750460863113403,
      "learning_rate": 0.0003,
      "loss": 0.5093,
      "step": 11200
    },
    {
      "epoch": 10.37411062657792,
      "grad_norm": 1.2779016494750977,
      "learning_rate": 0.0003,
      "loss": 0.5184,
      "step": 11300
    },
    {
      "epoch": 10.465916915308698,
      "grad_norm": 1.2549773454666138,
      "learning_rate": 0.0003,
      "loss": 0.5342,
      "step": 11400
    },
    {
      "epoch": 10.557723204039476,
      "grad_norm": 1.008012294769287,
      "learning_rate": 0.0003,
      "loss": 0.5409,
      "step": 11500
    },
    {
      "epoch": 10.649529492770254,
      "grad_norm": 1.1311882734298706,
      "learning_rate": 0.0003,
      "loss": 0.5569,
      "step": 11600
    },
    {
      "epoch": 10.741335781501032,
      "grad_norm": 1.20146644115448,
      "learning_rate": 0.0003,
      "loss": 0.5644,
      "step": 11700
    },
    {
      "epoch": 10.83314207023181,
      "grad_norm": 1.1888364553451538,
      "learning_rate": 0.0003,
      "loss": 0.5914,
      "step": 11800
    },
    {
      "epoch": 10.924948358962588,
      "grad_norm": 1.1159058809280396,
      "learning_rate": 0.0003,
      "loss": 0.5858,
      "step": 11900
    },
    {
      "epoch": 10.999311452834519,
      "eval_accuracy": 0.5165396825396825,
      "eval_loss": 3.011744737625122,
      "eval_runtime": 5.5804,
      "eval_samples_per_second": 89.599,
      "eval_steps_per_second": 11.29,
      "step": 11981
    },
    {
      "epoch": 11.016754647693366,
      "grad_norm": 0.9989836812019348,
      "learning_rate": 0.0003,
      "loss": 0.5703,
      "step": 12000
    },
    {
      "epoch": 11.108560936424144,
      "grad_norm": 1.2236177921295166,
      "learning_rate": 0.0003,
      "loss": 0.4423,
      "step": 12100
    },
    {
      "epoch": 11.200367225154924,
      "grad_norm": 1.0680415630340576,
      "learning_rate": 0.0003,
      "loss": 0.4644,
      "step": 12200
    },
    {
      "epoch": 11.292173513885702,
      "grad_norm": 1.3059639930725098,
      "learning_rate": 0.0003,
      "loss": 0.4797,
      "step": 12300
    },
    {
      "epoch": 11.38397980261648,
      "grad_norm": 1.1068886518478394,
      "learning_rate": 0.0003,
      "loss": 0.4968,
      "step": 12400
    },
    {
      "epoch": 11.475786091347258,
      "grad_norm": 1.2058519124984741,
      "learning_rate": 0.0003,
      "loss": 0.5027,
      "step": 12500
    },
    {
      "epoch": 11.567592380078036,
      "grad_norm": 1.222372055053711,
      "learning_rate": 0.0003,
      "loss": 0.5157,
      "step": 12600
    },
    {
      "epoch": 11.659398668808814,
      "grad_norm": 1.2184628248214722,
      "learning_rate": 0.0003,
      "loss": 0.531,
      "step": 12700
    },
    {
      "epoch": 11.751204957539592,
      "grad_norm": 1.1435773372650146,
      "learning_rate": 0.0003,
      "loss": 0.5436,
      "step": 12800
    },
    {
      "epoch": 11.84301124627037,
      "grad_norm": 1.3586217164993286,
      "learning_rate": 0.0003,
      "loss": 0.5591,
      "step": 12900
    },
    {
      "epoch": 11.934817535001148,
      "grad_norm": 1.2372316122055054,
      "learning_rate": 0.0003,
      "loss": 0.5618,
      "step": 13000
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.517015873015873,
      "eval_loss": 3.031684160232544,
      "eval_runtime": 5.7443,
      "eval_samples_per_second": 87.043,
      "eval_steps_per_second": 10.967,
      "step": 13071
    },
    {
      "epoch": 12.026623823731926,
      "grad_norm": 1.0793490409851074,
      "learning_rate": 0.0003,
      "loss": 0.531,
      "step": 13100
    },
    {
      "epoch": 12.118430112462704,
      "grad_norm": 1.0374959707260132,
      "learning_rate": 0.0003,
      "loss": 0.4188,
      "step": 13200
    },
    {
      "epoch": 12.210236401193482,
      "grad_norm": 1.1079918146133423,
      "learning_rate": 0.0003,
      "loss": 0.4472,
      "step": 13300
    },
    {
      "epoch": 12.30204268992426,
      "grad_norm": 1.2988286018371582,
      "learning_rate": 0.0003,
      "loss": 0.4579,
      "step": 13400
    },
    {
      "epoch": 12.393848978655038,
      "grad_norm": 1.110042691230774,
      "learning_rate": 0.0003,
      "loss": 0.4741,
      "step": 13500
    },
    {
      "epoch": 12.485655267385816,
      "grad_norm": 1.2625929117202759,
      "learning_rate": 0.0003,
      "loss": 0.4879,
      "step": 13600
    },
    {
      "epoch": 12.577461556116594,
      "grad_norm": 1.1651537418365479,
      "learning_rate": 0.0003,
      "loss": 0.4988,
      "step": 13700
    },
    {
      "epoch": 12.669267844847372,
      "grad_norm": 1.2762588262557983,
      "learning_rate": 0.0003,
      "loss": 0.5068,
      "step": 13800
    },
    {
      "epoch": 12.76107413357815,
      "grad_norm": 1.3166848421096802,
      "learning_rate": 0.0003,
      "loss": 0.5279,
      "step": 13900
    },
    {
      "epoch": 12.852880422308928,
      "grad_norm": 1.3245456218719482,
      "learning_rate": 0.0003,
      "loss": 0.5316,
      "step": 14000
    },
    {
      "epoch": 12.944686711039706,
      "grad_norm": 1.2448067665100098,
      "learning_rate": 0.0003,
      "loss": 0.5464,
      "step": 14100
    },
    {
      "epoch": 12.999770484278173,
      "eval_accuracy": 0.5166984126984127,
      "eval_loss": 3.0685949325561523,
      "eval_runtime": 5.5092,
      "eval_samples_per_second": 90.758,
      "eval_steps_per_second": 11.435,
      "step": 14160
    },
    {
      "epoch": 13.036492999770484,
      "grad_norm": 1.0070549249649048,
      "learning_rate": 0.0003,
      "loss": 0.4909,
      "step": 14200
    },
    {
      "epoch": 13.128299288501262,
      "grad_norm": 1.078823447227478,
      "learning_rate": 0.0003,
      "loss": 0.4126,
      "step": 14300
    },
    {
      "epoch": 13.22010557723204,
      "grad_norm": 1.2008830308914185,
      "learning_rate": 0.0003,
      "loss": 0.4279,
      "step": 14400
    },
    {
      "epoch": 13.311911865962818,
      "grad_norm": 1.258834958076477,
      "learning_rate": 0.0003,
      "loss": 0.445,
      "step": 14500
    },
    {
      "epoch": 13.403718154693596,
      "grad_norm": 1.2432317733764648,
      "learning_rate": 0.0003,
      "loss": 0.4563,
      "step": 14600
    },
    {
      "epoch": 13.495524443424374,
      "grad_norm": 1.2906520366668701,
      "learning_rate": 0.0003,
      "loss": 0.47,
      "step": 14700
    },
    {
      "epoch": 13.587330732155152,
      "grad_norm": 1.2167085409164429,
      "learning_rate": 0.0003,
      "loss": 0.481,
      "step": 14800
    },
    {
      "epoch": 13.67913702088593,
      "grad_norm": 1.3750929832458496,
      "learning_rate": 0.0003,
      "loss": 0.4949,
      "step": 14900
    },
    {
      "epoch": 13.770943309616708,
      "grad_norm": 1.3509438037872314,
      "learning_rate": 0.0003,
      "loss": 0.5036,
      "step": 15000
    },
    {
      "epoch": 13.862749598347486,
      "grad_norm": 1.2090297937393188,
      "learning_rate": 0.0003,
      "loss": 0.5101,
      "step": 15100
    },
    {
      "epoch": 13.954555887078264,
      "grad_norm": 1.3497878313064575,
      "learning_rate": 0.0003,
      "loss": 0.5243,
      "step": 15200
    },
    {
      "epoch": 13.999540968556346,
      "eval_accuracy": 0.5148571428571429,
      "eval_loss": 3.0829057693481445,
      "eval_runtime": 5.7223,
      "eval_samples_per_second": 87.377,
      "eval_steps_per_second": 11.01,
      "step": 15249
    },
    {
      "epoch": 14.046362175809042,
      "grad_norm": 1.0588971376419067,
      "learning_rate": 0.0003,
      "loss": 0.458,
      "step": 15300
    },
    {
      "epoch": 14.13816846453982,
      "grad_norm": 1.2202554941177368,
      "learning_rate": 0.0003,
      "loss": 0.3978,
      "step": 15400
    },
    {
      "epoch": 14.229974753270598,
      "grad_norm": 1.2965370416641235,
      "learning_rate": 0.0003,
      "loss": 0.4225,
      "step": 15500
    },
    {
      "epoch": 14.321781042001376,
      "grad_norm": 1.2187902927398682,
      "learning_rate": 0.0003,
      "loss": 0.4312,
      "step": 15600
    },
    {
      "epoch": 14.413587330732156,
      "grad_norm": 1.2526473999023438,
      "learning_rate": 0.0003,
      "loss": 0.4479,
      "step": 15700
    },
    {
      "epoch": 14.505393619462934,
      "grad_norm": 1.1553575992584229,
      "learning_rate": 0.0003,
      "loss": 0.4589,
      "step": 15800
    },
    {
      "epoch": 14.597199908193712,
      "grad_norm": 1.2674176692962646,
      "learning_rate": 0.0003,
      "loss": 0.4654,
      "step": 15900
    },
    {
      "epoch": 14.68900619692449,
      "grad_norm": 1.2161575555801392,
      "learning_rate": 0.0003,
      "loss": 0.475,
      "step": 16000
    },
    {
      "epoch": 14.780812485655268,
      "grad_norm": 1.078759789466858,
      "learning_rate": 0.0003,
      "loss": 0.4919,
      "step": 16100
    },
    {
      "epoch": 14.872618774386046,
      "grad_norm": 1.3057931661605835,
      "learning_rate": 0.0003,
      "loss": 0.4965,
      "step": 16200
    },
    {
      "epoch": 14.964425063116824,
      "grad_norm": 1.212488055229187,
      "learning_rate": 0.0003,
      "loss": 0.5066,
      "step": 16300
    },
    {
      "epoch": 14.999311452834519,
      "eval_accuracy": 0.5127301587301587,
      "eval_loss": 3.0957658290863037,
      "eval_runtime": 6.0214,
      "eval_samples_per_second": 83.037,
      "eval_steps_per_second": 10.463,
      "step": 16338
    },
    {
      "epoch": 15.056231351847602,
      "grad_norm": 0.9950692653656006,
      "learning_rate": 0.0003,
      "loss": 0.4266,
      "step": 16400
    },
    {
      "epoch": 15.14803764057838,
      "grad_norm": 1.0924224853515625,
      "learning_rate": 0.0003,
      "loss": 0.3937,
      "step": 16500
    },
    {
      "epoch": 15.239843929309158,
      "grad_norm": 1.1368046998977661,
      "learning_rate": 0.0003,
      "loss": 0.4018,
      "step": 16600
    },
    {
      "epoch": 15.331650218039936,
      "grad_norm": 1.1744548082351685,
      "learning_rate": 0.0003,
      "loss": 0.4195,
      "step": 16700
    },
    {
      "epoch": 15.423456506770714,
      "grad_norm": 1.1728862524032593,
      "learning_rate": 0.0003,
      "loss": 0.4324,
      "step": 16800
    },
    {
      "epoch": 15.515262795501492,
      "grad_norm": 1.2837214469909668,
      "learning_rate": 0.0003,
      "loss": 0.4444,
      "step": 16900
    },
    {
      "epoch": 15.60706908423227,
      "grad_norm": 1.2437090873718262,
      "learning_rate": 0.0003,
      "loss": 0.4551,
      "step": 17000
    },
    {
      "epoch": 15.698875372963048,
      "grad_norm": 1.2437779903411865,
      "learning_rate": 0.0003,
      "loss": 0.4676,
      "step": 17100
    },
    {
      "epoch": 15.790681661693826,
      "grad_norm": 1.182637095451355,
      "learning_rate": 0.0003,
      "loss": 0.478,
      "step": 17200
    },
    {
      "epoch": 15.882487950424604,
      "grad_norm": 1.199127435684204,
      "learning_rate": 0.0003,
      "loss": 0.4881,
      "step": 17300
    },
    {
      "epoch": 15.974294239155382,
      "grad_norm": 1.251943826675415,
      "learning_rate": 0.0003,
      "loss": 0.4947,
      "step": 17400
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.5152698412698413,
      "eval_loss": 3.0920920372009277,
      "eval_runtime": 5.478,
      "eval_samples_per_second": 91.273,
      "eval_steps_per_second": 11.5,
      "step": 17428
    },
    {
      "epoch": 16.06610052788616,
      "grad_norm": 1.2216734886169434,
      "learning_rate": 0.0003,
      "loss": 0.4103,
      "step": 17500
    },
    {
      "epoch": 16.15790681661694,
      "grad_norm": 1.1105889081954956,
      "learning_rate": 0.0003,
      "loss": 0.3818,
      "step": 17600
    },
    {
      "epoch": 16.249713105347716,
      "grad_norm": 1.1674516201019287,
      "learning_rate": 0.0003,
      "loss": 0.3983,
      "step": 17700
    },
    {
      "epoch": 16.341519394078496,
      "grad_norm": 1.0942326784133911,
      "learning_rate": 0.0003,
      "loss": 0.4102,
      "step": 17800
    },
    {
      "epoch": 16.433325682809272,
      "grad_norm": 1.0696207284927368,
      "learning_rate": 0.0003,
      "loss": 0.4188,
      "step": 17900
    },
    {
      "epoch": 16.525131971540052,
      "grad_norm": 1.3964133262634277,
      "learning_rate": 0.0003,
      "loss": 0.434,
      "step": 18000
    },
    {
      "epoch": 16.616938260270828,
      "grad_norm": 1.2896054983139038,
      "learning_rate": 0.0003,
      "loss": 0.4465,
      "step": 18100
    },
    {
      "epoch": 16.708744549001608,
      "grad_norm": 1.2668951749801636,
      "learning_rate": 0.0003,
      "loss": 0.4544,
      "step": 18200
    },
    {
      "epoch": 16.800550837732384,
      "grad_norm": 1.2267131805419922,
      "learning_rate": 0.0003,
      "loss": 0.4647,
      "step": 18300
    },
    {
      "epoch": 16.892357126463164,
      "grad_norm": 1.5275602340698242,
      "learning_rate": 0.0003,
      "loss": 0.4756,
      "step": 18400
    },
    {
      "epoch": 16.98416341519394,
      "grad_norm": 1.2730522155761719,
      "learning_rate": 0.0003,
      "loss": 0.4841,
      "step": 18500
    },
    {
      "epoch": 16.999770484278173,
      "eval_accuracy": 0.5162222222222222,
      "eval_loss": 3.116957902908325,
      "eval_runtime": 5.4049,
      "eval_samples_per_second": 92.509,
      "eval_steps_per_second": 11.656,
      "step": 18517
    },
    {
      "epoch": 17.07596970392472,
      "grad_norm": 1.0755029916763306,
      "learning_rate": 0.0003,
      "loss": 0.382,
      "step": 18600
    },
    {
      "epoch": 17.167775992655496,
      "grad_norm": 1.0808229446411133,
      "learning_rate": 0.0003,
      "loss": 0.3769,
      "step": 18700
    },
    {
      "epoch": 17.259582281386276,
      "grad_norm": 1.1092764139175415,
      "learning_rate": 0.0003,
      "loss": 0.3884,
      "step": 18800
    },
    {
      "epoch": 17.351388570117052,
      "grad_norm": 1.169694423675537,
      "learning_rate": 0.0003,
      "loss": 0.4039,
      "step": 18900
    },
    {
      "epoch": 17.443194858847832,
      "grad_norm": 1.2209644317626953,
      "learning_rate": 0.0003,
      "loss": 0.419,
      "step": 19000
    },
    {
      "epoch": 17.53500114757861,
      "grad_norm": 1.3729647397994995,
      "learning_rate": 0.0003,
      "loss": 0.4292,
      "step": 19100
    },
    {
      "epoch": 17.626807436309388,
      "grad_norm": 1.1597981452941895,
      "learning_rate": 0.0003,
      "loss": 0.4326,
      "step": 19200
    },
    {
      "epoch": 17.718613725040164,
      "grad_norm": 1.608828067779541,
      "learning_rate": 0.0003,
      "loss": 0.4511,
      "step": 19300
    },
    {
      "epoch": 17.810420013770944,
      "grad_norm": 1.3395555019378662,
      "learning_rate": 0.0003,
      "loss": 0.4542,
      "step": 19400
    },
    {
      "epoch": 17.90222630250172,
      "grad_norm": 1.3384451866149902,
      "learning_rate": 0.0003,
      "loss": 0.4609,
      "step": 19500
    },
    {
      "epoch": 17.9940325912325,
      "grad_norm": 1.3556160926818848,
      "learning_rate": 0.0003,
      "loss": 0.4727,
      "step": 19600
    },
    {
      "epoch": 17.999540968556346,
      "eval_accuracy": 0.5172063492063492,
      "eval_loss": 3.137547492980957,
      "eval_runtime": 5.667,
      "eval_samples_per_second": 88.231,
      "eval_steps_per_second": 11.117,
      "step": 19606
    },
    {
      "epoch": 18.085838879963276,
      "grad_norm": 1.1512075662612915,
      "learning_rate": 0.0003,
      "loss": 0.3544,
      "step": 19700
    },
    {
      "epoch": 18.177645168694056,
      "grad_norm": 1.1157560348510742,
      "learning_rate": 0.0003,
      "loss": 0.3698,
      "step": 19800
    },
    {
      "epoch": 18.269451457424832,
      "grad_norm": 1.0202255249023438,
      "learning_rate": 0.0003,
      "loss": 0.3897,
      "step": 19900
    },
    {
      "epoch": 18.361257746155612,
      "grad_norm": 1.1312451362609863,
      "learning_rate": 0.0003,
      "loss": 0.3938,
      "step": 20000
    },
    {
      "epoch": 18.45306403488639,
      "grad_norm": 1.5055644512176514,
      "learning_rate": 0.0003,
      "loss": 0.4062,
      "step": 20100
    },
    {
      "epoch": 18.544870323617168,
      "grad_norm": 1.3119601011276245,
      "learning_rate": 0.0003,
      "loss": 0.4172,
      "step": 20200
    },
    {
      "epoch": 18.636676612347944,
      "grad_norm": 1.229153037071228,
      "learning_rate": 0.0003,
      "loss": 0.4294,
      "step": 20300
    },
    {
      "epoch": 18.728482901078724,
      "grad_norm": 1.1669813394546509,
      "learning_rate": 0.0003,
      "loss": 0.4396,
      "step": 20400
    },
    {
      "epoch": 18.8202891898095,
      "grad_norm": 1.4410171508789062,
      "learning_rate": 0.0003,
      "loss": 0.4459,
      "step": 20500
    },
    {
      "epoch": 18.91209547854028,
      "grad_norm": 1.4326797723770142,
      "learning_rate": 0.0003,
      "loss": 0.4634,
      "step": 20600
    },
    {
      "epoch": 18.99931145283452,
      "eval_accuracy": 0.514984126984127,
      "eval_loss": 3.132270097732544,
      "eval_runtime": 5.3564,
      "eval_samples_per_second": 93.346,
      "eval_steps_per_second": 11.762,
      "step": 20695
    },
    {
      "epoch": 19.003901767271056,
      "grad_norm": 1.1022387742996216,
      "learning_rate": 0.0003,
      "loss": 0.461,
      "step": 20700
    },
    {
      "epoch": 19.095708056001836,
      "grad_norm": 1.2258243560791016,
      "learning_rate": 0.0003,
      "loss": 0.3484,
      "step": 20800
    },
    {
      "epoch": 19.187514344732612,
      "grad_norm": 1.2331010103225708,
      "learning_rate": 0.0003,
      "loss": 0.3623,
      "step": 20900
    },
    {
      "epoch": 19.279320633463392,
      "grad_norm": 1.1330702304840088,
      "learning_rate": 0.0003,
      "loss": 0.3792,
      "step": 21000
    },
    {
      "epoch": 19.371126922194172,
      "grad_norm": 1.272649884223938,
      "learning_rate": 0.0003,
      "loss": 0.3929,
      "step": 21100
    },
    {
      "epoch": 19.462933210924948,
      "grad_norm": 1.1329351663589478,
      "learning_rate": 0.0003,
      "loss": 0.4027,
      "step": 21200
    },
    {
      "epoch": 19.554739499655728,
      "grad_norm": 1.2024905681610107,
      "learning_rate": 0.0003,
      "loss": 0.4081,
      "step": 21300
    },
    {
      "epoch": 19.646545788386504,
      "grad_norm": 1.2089053392410278,
      "learning_rate": 0.0003,
      "loss": 0.4156,
      "step": 21400
    },
    {
      "epoch": 19.738352077117284,
      "grad_norm": 1.2542715072631836,
      "learning_rate": 0.0003,
      "loss": 0.4377,
      "step": 21500
    },
    {
      "epoch": 19.83015836584806,
      "grad_norm": 1.5096327066421509,
      "learning_rate": 0.0003,
      "loss": 0.4446,
      "step": 21600
    },
    {
      "epoch": 19.92196465457884,
      "grad_norm": 1.2375317811965942,
      "learning_rate": 0.0003,
      "loss": 0.4468,
      "step": 21700
    },
    {
      "epoch": 19.995409685563462,
      "eval_accuracy": 0.5165079365079365,
      "eval_loss": 3.163097381591797,
      "eval_runtime": 5.9319,
      "eval_samples_per_second": 84.29,
      "eval_steps_per_second": 10.62,
      "step": 21780
    },
    {
      "epoch": 19.995409685563462,
      "step": 21780,
      "total_flos": 1.5026894410916823e+18,
      "train_loss": 0.737595841904317,
      "train_runtime": 47608.6465,
      "train_samples_per_second": 14.642,
      "train_steps_per_second": 0.457
    }
  ],
  "logging_steps": 100,
  "max_steps": 21780,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 500,
  "total_flos": 1.5026894410916823e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}