{
  "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": 0.229265034198761,
      "learning_rate": 0.0003,
      "loss": 1.6148,
      "step": 100
    },
    {
      "epoch": 0.18361257746155613,
      "grad_norm": 0.2753294110298157,
      "learning_rate": 0.0003,
      "loss": 1.5846,
      "step": 200
    },
    {
      "epoch": 0.2754188661923342,
      "grad_norm": 0.2562897801399231,
      "learning_rate": 0.0003,
      "loss": 1.5631,
      "step": 300
    },
    {
      "epoch": 0.36722515492311225,
      "grad_norm": 0.24621829390525818,
      "learning_rate": 0.0003,
      "loss": 1.5662,
      "step": 400
    },
    {
      "epoch": 0.4590314436538903,
      "grad_norm": 0.2583877742290497,
      "learning_rate": 0.0003,
      "loss": 1.561,
      "step": 500
    },
    {
      "epoch": 0.5508377323846684,
      "grad_norm": 0.275136262178421,
      "learning_rate": 0.0003,
      "loss": 1.5561,
      "step": 600
    },
    {
      "epoch": 0.6426440211154464,
      "grad_norm": 0.2927466928958893,
      "learning_rate": 0.0003,
      "loss": 1.5552,
      "step": 700
    },
    {
      "epoch": 0.7344503098462245,
      "grad_norm": 0.3323851525783539,
      "learning_rate": 0.0003,
      "loss": 1.5608,
      "step": 800
    },
    {
      "epoch": 0.8262565985770025,
      "grad_norm": 0.33202576637268066,
      "learning_rate": 0.0003,
      "loss": 1.5254,
      "step": 900
    },
    {
      "epoch": 0.9180628873077806,
      "grad_norm": 0.3486601412296295,
      "learning_rate": 0.0003,
      "loss": 1.5356,
      "step": 1000
    },
    {
      "epoch": 0.999770484278173,
      "eval_accuracy": 0.6863755458515284,
      "eval_loss": 1.3710707426071167,
      "eval_runtime": 8.9317,
      "eval_samples_per_second": 55.98,
      "eval_steps_per_second": 7.054,
      "step": 1089
    },
    {
      "epoch": 1.0098691760385587,
      "grad_norm": 0.34695035219192505,
      "learning_rate": 0.0003,
      "loss": 1.5044,
      "step": 1100
    },
    {
      "epoch": 1.1016754647693368,
      "grad_norm": 0.3605581820011139,
      "learning_rate": 0.0003,
      "loss": 1.3328,
      "step": 1200
    },
    {
      "epoch": 1.1934817535001148,
      "grad_norm": 0.37978020310401917,
      "learning_rate": 0.0003,
      "loss": 1.3246,
      "step": 1300
    },
    {
      "epoch": 1.2852880422308928,
      "grad_norm": 0.40200668573379517,
      "learning_rate": 0.0003,
      "loss": 1.3183,
      "step": 1400
    },
    {
      "epoch": 1.377094330961671,
      "grad_norm": 0.49451494216918945,
      "learning_rate": 0.0003,
      "loss": 1.3153,
      "step": 1500
    },
    {
      "epoch": 1.468900619692449,
      "grad_norm": 0.4396137297153473,
      "learning_rate": 0.0003,
      "loss": 1.3301,
      "step": 1600
    },
    {
      "epoch": 1.560706908423227,
      "grad_norm": 0.4403189718723297,
      "learning_rate": 0.0003,
      "loss": 1.2971,
      "step": 1700
    },
    {
      "epoch": 1.652513197154005,
      "grad_norm": 0.4628995656967163,
      "learning_rate": 0.0003,
      "loss": 1.3159,
      "step": 1800
    },
    {
      "epoch": 1.744319485884783,
      "grad_norm": 0.4826644957065582,
      "learning_rate": 0.0003,
      "loss": 1.3027,
      "step": 1900
    },
    {
      "epoch": 1.836125774615561,
      "grad_norm": 0.48213547468185425,
      "learning_rate": 0.0003,
      "loss": 1.2977,
      "step": 2000
    },
    {
      "epoch": 1.9279320633463393,
      "grad_norm": 0.5228666067123413,
      "learning_rate": 0.0003,
      "loss": 1.3102,
      "step": 2100
    },
    {
      "epoch": 1.999540968556346,
      "eval_accuracy": 0.7020116448326056,
      "eval_loss": 1.1752803325653076,
      "eval_runtime": 8.9457,
      "eval_samples_per_second": 55.892,
      "eval_steps_per_second": 7.042,
      "step": 2178
    },
    {
      "epoch": 2.0197383520771175,
      "grad_norm": 0.4973185360431671,
      "learning_rate": 0.0003,
      "loss": 1.2434,
      "step": 2200
    },
    {
      "epoch": 2.1115446408078955,
      "grad_norm": 0.5200226902961731,
      "learning_rate": 0.0003,
      "loss": 1.0191,
      "step": 2300
    },
    {
      "epoch": 2.2033509295386735,
      "grad_norm": 0.6249756217002869,
      "learning_rate": 0.0003,
      "loss": 1.0435,
      "step": 2400
    },
    {
      "epoch": 2.2951572182694515,
      "grad_norm": 0.5348188877105713,
      "learning_rate": 0.0003,
      "loss": 1.0434,
      "step": 2500
    },
    {
      "epoch": 2.3869635070002295,
      "grad_norm": 0.6106544733047485,
      "learning_rate": 0.0003,
      "loss": 1.0546,
      "step": 2600
    },
    {
      "epoch": 2.4787697957310075,
      "grad_norm": 0.5895304679870605,
      "learning_rate": 0.0003,
      "loss": 1.0437,
      "step": 2700
    },
    {
      "epoch": 2.5705760844617855,
      "grad_norm": 0.5809823274612427,
      "learning_rate": 0.0003,
      "loss": 1.0628,
      "step": 2800
    },
    {
      "epoch": 2.6623823731925635,
      "grad_norm": 0.6009641289710999,
      "learning_rate": 0.0003,
      "loss": 1.0478,
      "step": 2900
    },
    {
      "epoch": 2.754188661923342,
      "grad_norm": 0.679328978061676,
      "learning_rate": 0.0003,
      "loss": 1.0674,
      "step": 3000
    },
    {
      "epoch": 2.84599495065412,
      "grad_norm": 0.5471652746200562,
      "learning_rate": 0.0003,
      "loss": 1.0462,
      "step": 3100
    },
    {
      "epoch": 2.937801239384898,
      "grad_norm": 0.6065019965171814,
      "learning_rate": 0.0003,
      "loss": 1.0549,
      "step": 3200
    },
    {
      "epoch": 2.9993114528345193,
      "eval_accuracy": 0.716448326055313,
      "eval_loss": 1.0094784498214722,
      "eval_runtime": 8.9265,
      "eval_samples_per_second": 56.013,
      "eval_steps_per_second": 7.058,
      "step": 3267
    },
    {
      "epoch": 3.029607528115676,
      "grad_norm": 0.6480036973953247,
      "learning_rate": 0.0003,
      "loss": 0.9648,
      "step": 3300
    },
    {
      "epoch": 3.121413816846454,
      "grad_norm": 0.7049514651298523,
      "learning_rate": 0.0003,
      "loss": 0.7919,
      "step": 3400
    },
    {
      "epoch": 3.213220105577232,
      "grad_norm": 0.7723010182380676,
      "learning_rate": 0.0003,
      "loss": 0.804,
      "step": 3500
    },
    {
      "epoch": 3.30502639430801,
      "grad_norm": 0.7994470596313477,
      "learning_rate": 0.0003,
      "loss": 0.8323,
      "step": 3600
    },
    {
      "epoch": 3.396832683038788,
      "grad_norm": 0.7567194700241089,
      "learning_rate": 0.0003,
      "loss": 0.8261,
      "step": 3700
    },
    {
      "epoch": 3.488638971769566,
      "grad_norm": 0.7832695245742798,
      "learning_rate": 0.0003,
      "loss": 0.8337,
      "step": 3800
    },
    {
      "epoch": 3.580445260500344,
      "grad_norm": 0.7570985555648804,
      "learning_rate": 0.0003,
      "loss": 0.8334,
      "step": 3900
    },
    {
      "epoch": 3.672251549231122,
      "grad_norm": 0.7717698216438293,
      "learning_rate": 0.0003,
      "loss": 0.8455,
      "step": 4000
    },
    {
      "epoch": 3.7640578379619005,
      "grad_norm": 0.8522344827651978,
      "learning_rate": 0.0003,
      "loss": 0.8448,
      "step": 4100
    },
    {
      "epoch": 3.8558641266926785,
      "grad_norm": 0.8296842575073242,
      "learning_rate": 0.0003,
      "loss": 0.8534,
      "step": 4200
    },
    {
      "epoch": 3.9476704154234565,
      "grad_norm": 0.7863412499427795,
      "learning_rate": 0.0003,
      "loss": 0.8461,
      "step": 4300
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.7297409024745269,
      "eval_loss": 0.8722099661827087,
      "eval_runtime": 8.8823,
      "eval_samples_per_second": 56.292,
      "eval_steps_per_second": 7.093,
      "step": 4357
    },
    {
      "epoch": 4.039476704154235,
      "grad_norm": 0.7983632683753967,
      "learning_rate": 0.0003,
      "loss": 0.7547,
      "step": 4400
    },
    {
      "epoch": 4.131282992885013,
      "grad_norm": 0.699720561504364,
      "learning_rate": 0.0003,
      "loss": 0.618,
      "step": 4500
    },
    {
      "epoch": 4.223089281615791,
      "grad_norm": 0.8726249933242798,
      "learning_rate": 0.0003,
      "loss": 0.6373,
      "step": 4600
    },
    {
      "epoch": 4.314895570346569,
      "grad_norm": 0.8412553071975708,
      "learning_rate": 0.0003,
      "loss": 0.6582,
      "step": 4700
    },
    {
      "epoch": 4.406701859077347,
      "grad_norm": 0.86082923412323,
      "learning_rate": 0.0003,
      "loss": 0.6612,
      "step": 4800
    },
    {
      "epoch": 4.498508147808125,
      "grad_norm": 0.8449448943138123,
      "learning_rate": 0.0003,
      "loss": 0.6704,
      "step": 4900
    },
    {
      "epoch": 4.590314436538903,
      "grad_norm": 0.9901677370071411,
      "learning_rate": 0.0003,
      "loss": 0.683,
      "step": 5000
    },
    {
      "epoch": 4.682120725269681,
      "grad_norm": 0.884429395198822,
      "learning_rate": 0.0003,
      "loss": 0.6903,
      "step": 5100
    },
    {
      "epoch": 4.773927014000459,
      "grad_norm": 0.8438558578491211,
      "learning_rate": 0.0003,
      "loss": 0.693,
      "step": 5200
    },
    {
      "epoch": 4.865733302731237,
      "grad_norm": 0.8200265169143677,
      "learning_rate": 0.0003,
      "loss": 0.6986,
      "step": 5300
    },
    {
      "epoch": 4.957539591462015,
      "grad_norm": 0.9410732984542847,
      "learning_rate": 0.0003,
      "loss": 0.701,
      "step": 5400
    },
    {
      "epoch": 4.999770484278173,
      "eval_accuracy": 0.7405647743813683,
      "eval_loss": 0.7641452550888062,
      "eval_runtime": 8.8775,
      "eval_samples_per_second": 56.322,
      "eval_steps_per_second": 7.097,
      "step": 5446
    },
    {
      "epoch": 5.049345880192793,
      "grad_norm": 0.8364540934562683,
      "learning_rate": 0.0003,
      "loss": 0.5945,
      "step": 5500
    },
    {
      "epoch": 5.141152168923571,
      "grad_norm": 0.8985306620597839,
      "learning_rate": 0.0003,
      "loss": 0.5072,
      "step": 5600
    },
    {
      "epoch": 5.232958457654349,
      "grad_norm": 0.9036381244659424,
      "learning_rate": 0.0003,
      "loss": 0.5187,
      "step": 5700
    },
    {
      "epoch": 5.324764746385127,
      "grad_norm": 0.9836907386779785,
      "learning_rate": 0.0003,
      "loss": 0.5468,
      "step": 5800
    },
    {
      "epoch": 5.416571035115905,
      "grad_norm": 0.8932607769966125,
      "learning_rate": 0.0003,
      "loss": 0.546,
      "step": 5900
    },
    {
      "epoch": 5.508377323846684,
      "grad_norm": 0.9328716397285461,
      "learning_rate": 0.0003,
      "loss": 0.5636,
      "step": 6000
    },
    {
      "epoch": 5.600183612577462,
      "grad_norm": 1.0993613004684448,
      "learning_rate": 0.0003,
      "loss": 0.567,
      "step": 6100
    },
    {
      "epoch": 5.69198990130824,
      "grad_norm": 0.9644467830657959,
      "learning_rate": 0.0003,
      "loss": 0.5853,
      "step": 6200
    },
    {
      "epoch": 5.783796190039018,
      "grad_norm": 1.0126985311508179,
      "learning_rate": 0.0003,
      "loss": 0.5883,
      "step": 6300
    },
    {
      "epoch": 5.875602478769796,
      "grad_norm": 0.9119430184364319,
      "learning_rate": 0.0003,
      "loss": 0.5853,
      "step": 6400
    },
    {
      "epoch": 5.967408767500574,
      "grad_norm": 0.88072669506073,
      "learning_rate": 0.0003,
      "loss": 0.5977,
      "step": 6500
    },
    {
      "epoch": 5.999540968556346,
      "eval_accuracy": 0.7490451237263465,
      "eval_loss": 0.6796724796295166,
      "eval_runtime": 8.9191,
      "eval_samples_per_second": 56.059,
      "eval_steps_per_second": 7.063,
      "step": 6535
    },
    {
      "epoch": 6.059215056231352,
      "grad_norm": 0.9065115451812744,
      "learning_rate": 0.0003,
      "loss": 0.4864,
      "step": 6600
    },
    {
      "epoch": 6.15102134496213,
      "grad_norm": 0.8660620450973511,
      "learning_rate": 0.0003,
      "loss": 0.4337,
      "step": 6700
    },
    {
      "epoch": 6.242827633692908,
      "grad_norm": 0.9163237810134888,
      "learning_rate": 0.0003,
      "loss": 0.4496,
      "step": 6800
    },
    {
      "epoch": 6.334633922423686,
      "grad_norm": 0.8863268494606018,
      "learning_rate": 0.0003,
      "loss": 0.4626,
      "step": 6900
    },
    {
      "epoch": 6.426440211154464,
      "grad_norm": 0.9574721455574036,
      "learning_rate": 0.0003,
      "loss": 0.4722,
      "step": 7000
    },
    {
      "epoch": 6.518246499885242,
      "grad_norm": 1.096871256828308,
      "learning_rate": 0.0003,
      "loss": 0.4887,
      "step": 7100
    },
    {
      "epoch": 6.61005278861602,
      "grad_norm": 1.0900388956069946,
      "learning_rate": 0.0003,
      "loss": 0.4958,
      "step": 7200
    },
    {
      "epoch": 6.701859077346798,
      "grad_norm": 0.9678546190261841,
      "learning_rate": 0.0003,
      "loss": 0.4991,
      "step": 7300
    },
    {
      "epoch": 6.793665366077576,
      "grad_norm": 1.0782684087753296,
      "learning_rate": 0.0003,
      "loss": 0.5068,
      "step": 7400
    },
    {
      "epoch": 6.885471654808354,
      "grad_norm": 1.0213241577148438,
      "learning_rate": 0.0003,
      "loss": 0.512,
      "step": 7500
    },
    {
      "epoch": 6.977277943539132,
      "grad_norm": 1.049910545349121,
      "learning_rate": 0.0003,
      "loss": 0.5238,
      "step": 7600
    },
    {
      "epoch": 6.999311452834519,
      "eval_accuracy": 0.7558719068413392,
      "eval_loss": 0.620939314365387,
      "eval_runtime": 9.0834,
      "eval_samples_per_second": 55.045,
      "eval_steps_per_second": 6.936,
      "step": 7624
    },
    {
      "epoch": 7.06908423226991,
      "grad_norm": 0.9470316767692566,
      "learning_rate": 0.0003,
      "loss": 0.4131,
      "step": 7700
    },
    {
      "epoch": 7.160890521000688,
      "grad_norm": 0.8299586772918701,
      "learning_rate": 0.0003,
      "loss": 0.39,
      "step": 7800
    },
    {
      "epoch": 7.252696809731467,
      "grad_norm": 0.9369521141052246,
      "learning_rate": 0.0003,
      "loss": 0.3953,
      "step": 7900
    },
    {
      "epoch": 7.344503098462245,
      "grad_norm": 1.0843796730041504,
      "learning_rate": 0.0003,
      "loss": 0.4144,
      "step": 8000
    },
    {
      "epoch": 7.436309387193023,
      "grad_norm": 0.9615817666053772,
      "learning_rate": 0.0003,
      "loss": 0.4266,
      "step": 8100
    },
    {
      "epoch": 7.528115675923801,
      "grad_norm": 0.9017691612243652,
      "learning_rate": 0.0003,
      "loss": 0.4411,
      "step": 8200
    },
    {
      "epoch": 7.619921964654579,
      "grad_norm": 1.0230755805969238,
      "learning_rate": 0.0003,
      "loss": 0.4392,
      "step": 8300
    },
    {
      "epoch": 7.711728253385357,
      "grad_norm": 1.0185121297836304,
      "learning_rate": 0.0003,
      "loss": 0.4489,
      "step": 8400
    },
    {
      "epoch": 7.803534542116135,
      "grad_norm": 1.003201961517334,
      "learning_rate": 0.0003,
      "loss": 0.4475,
      "step": 8500
    },
    {
      "epoch": 7.895340830846913,
      "grad_norm": 0.8930547833442688,
      "learning_rate": 0.0003,
      "loss": 0.4569,
      "step": 8600
    },
    {
      "epoch": 7.987147119577691,
      "grad_norm": 1.1096137762069702,
      "learning_rate": 0.0003,
      "loss": 0.4742,
      "step": 8700
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.7599941775836973,
      "eval_loss": 0.5836606025695801,
      "eval_runtime": 8.8891,
      "eval_samples_per_second": 56.249,
      "eval_steps_per_second": 7.087,
      "step": 8714
    },
    {
      "epoch": 8.07895340830847,
      "grad_norm": 0.9343357086181641,
      "learning_rate": 0.0003,
      "loss": 0.3623,
      "step": 8800
    },
    {
      "epoch": 8.170759697039248,
      "grad_norm": 1.0631523132324219,
      "learning_rate": 0.0003,
      "loss": 0.3456,
      "step": 8900
    },
    {
      "epoch": 8.262565985770026,
      "grad_norm": 0.9419318437576294,
      "learning_rate": 0.0003,
      "loss": 0.3648,
      "step": 9000
    },
    {
      "epoch": 8.354372274500804,
      "grad_norm": 0.8649080395698547,
      "learning_rate": 0.0003,
      "loss": 0.3758,
      "step": 9100
    },
    {
      "epoch": 8.446178563231582,
      "grad_norm": 0.9168979525566101,
      "learning_rate": 0.0003,
      "loss": 0.3843,
      "step": 9200
    },
    {
      "epoch": 8.53798485196236,
      "grad_norm": 0.9176032543182373,
      "learning_rate": 0.0003,
      "loss": 0.3942,
      "step": 9300
    },
    {
      "epoch": 8.629791140693138,
      "grad_norm": 0.9991121292114258,
      "learning_rate": 0.0003,
      "loss": 0.4084,
      "step": 9400
    },
    {
      "epoch": 8.721597429423916,
      "grad_norm": 1.0064568519592285,
      "learning_rate": 0.0003,
      "loss": 0.4115,
      "step": 9500
    },
    {
      "epoch": 8.813403718154694,
      "grad_norm": 1.0827409029006958,
      "learning_rate": 0.0003,
      "loss": 0.4254,
      "step": 9600
    },
    {
      "epoch": 8.905210006885472,
      "grad_norm": 1.1357035636901855,
      "learning_rate": 0.0003,
      "loss": 0.425,
      "step": 9700
    },
    {
      "epoch": 8.99701629561625,
      "grad_norm": 1.0116100311279297,
      "learning_rate": 0.0003,
      "loss": 0.438,
      "step": 9800
    },
    {
      "epoch": 8.999770484278173,
      "eval_accuracy": 0.7638136826783115,
      "eval_loss": 0.5532092452049255,
      "eval_runtime": 8.896,
      "eval_samples_per_second": 56.205,
      "eval_steps_per_second": 7.082,
      "step": 9803
    },
    {
      "epoch": 9.088822584347028,
      "grad_norm": 0.943336546421051,
      "learning_rate": 0.0003,
      "loss": 0.3196,
      "step": 9900
    },
    {
      "epoch": 9.180628873077806,
      "grad_norm": 0.966768205165863,
      "learning_rate": 0.0003,
      "loss": 0.3288,
      "step": 10000
    },
    {
      "epoch": 9.272435161808584,
      "grad_norm": 1.0522419214248657,
      "learning_rate": 0.0003,
      "loss": 0.3416,
      "step": 10100
    },
    {
      "epoch": 9.364241450539362,
      "grad_norm": 0.9877403378486633,
      "learning_rate": 0.0003,
      "loss": 0.3522,
      "step": 10200
    },
    {
      "epoch": 9.45604773927014,
      "grad_norm": 1.1037845611572266,
      "learning_rate": 0.0003,
      "loss": 0.3595,
      "step": 10300
    },
    {
      "epoch": 9.547854028000918,
      "grad_norm": 1.1503223180770874,
      "learning_rate": 0.0003,
      "loss": 0.3734,
      "step": 10400
    },
    {
      "epoch": 9.639660316731696,
      "grad_norm": 1.0321091413497925,
      "learning_rate": 0.0003,
      "loss": 0.3808,
      "step": 10500
    },
    {
      "epoch": 9.731466605462474,
      "grad_norm": 0.9578890204429626,
      "learning_rate": 0.0003,
      "loss": 0.3878,
      "step": 10600
    },
    {
      "epoch": 9.823272894193252,
      "grad_norm": 1.0930492877960205,
      "learning_rate": 0.0003,
      "loss": 0.3943,
      "step": 10700
    },
    {
      "epoch": 9.91507918292403,
      "grad_norm": 1.2598655223846436,
      "learning_rate": 0.0003,
      "loss": 0.402,
      "step": 10800
    },
    {
      "epoch": 9.999540968556346,
      "eval_accuracy": 0.7663522561863173,
      "eval_loss": 0.5331353545188904,
      "eval_runtime": 8.8958,
      "eval_samples_per_second": 56.206,
      "eval_steps_per_second": 7.082,
      "step": 10892
    },
    {
      "epoch": 10.006885471654808,
      "grad_norm": 0.9872549176216125,
      "learning_rate": 0.0003,
      "loss": 0.4016,
      "step": 10900
    },
    {
      "epoch": 10.098691760385586,
      "grad_norm": 0.9196072220802307,
      "learning_rate": 0.0003,
      "loss": 0.2985,
      "step": 11000
    },
    {
      "epoch": 10.190498049116364,
      "grad_norm": 0.9028146266937256,
      "learning_rate": 0.0003,
      "loss": 0.311,
      "step": 11100
    },
    {
      "epoch": 10.282304337847142,
      "grad_norm": 0.9019381999969482,
      "learning_rate": 0.0003,
      "loss": 0.3199,
      "step": 11200
    },
    {
      "epoch": 10.37411062657792,
      "grad_norm": 1.0821175575256348,
      "learning_rate": 0.0003,
      "loss": 0.3366,
      "step": 11300
    },
    {
      "epoch": 10.465916915308698,
      "grad_norm": 0.9856448769569397,
      "learning_rate": 0.0003,
      "loss": 0.3391,
      "step": 11400
    },
    {
      "epoch": 10.557723204039476,
      "grad_norm": 1.0605835914611816,
      "learning_rate": 0.0003,
      "loss": 0.3512,
      "step": 11500
    },
    {
      "epoch": 10.649529492770254,
      "grad_norm": 0.9575846791267395,
      "learning_rate": 0.0003,
      "loss": 0.3546,
      "step": 11600
    },
    {
      "epoch": 10.741335781501032,
      "grad_norm": 1.0335845947265625,
      "learning_rate": 0.0003,
      "loss": 0.3692,
      "step": 11700
    },
    {
      "epoch": 10.83314207023181,
      "grad_norm": 1.131901502609253,
      "learning_rate": 0.0003,
      "loss": 0.3747,
      "step": 11800
    },
    {
      "epoch": 10.924948358962588,
      "grad_norm": 0.9360109567642212,
      "learning_rate": 0.0003,
      "loss": 0.383,
      "step": 11900
    },
    {
      "epoch": 10.999311452834519,
      "eval_accuracy": 0.7685181950509461,
      "eval_loss": 0.5156339406967163,
      "eval_runtime": 8.925,
      "eval_samples_per_second": 56.022,
      "eval_steps_per_second": 7.059,
      "step": 11981
    },
    {
      "epoch": 11.016754647693366,
      "grad_norm": 0.8987607359886169,
      "learning_rate": 0.0003,
      "loss": 0.3677,
      "step": 12000
    },
    {
      "epoch": 11.108560936424144,
      "grad_norm": 0.882723867893219,
      "learning_rate": 0.0003,
      "loss": 0.2865,
      "step": 12100
    },
    {
      "epoch": 11.200367225154924,
      "grad_norm": 0.9493284821510315,
      "learning_rate": 0.0003,
      "loss": 0.2998,
      "step": 12200
    },
    {
      "epoch": 11.292173513885702,
      "grad_norm": 0.9848419427871704,
      "learning_rate": 0.0003,
      "loss": 0.3062,
      "step": 12300
    },
    {
      "epoch": 11.38397980261648,
      "grad_norm": 0.9860394597053528,
      "learning_rate": 0.0003,
      "loss": 0.3197,
      "step": 12400
    },
    {
      "epoch": 11.475786091347258,
      "grad_norm": 1.246336579322815,
      "learning_rate": 0.0003,
      "loss": 0.3318,
      "step": 12500
    },
    {
      "epoch": 11.567592380078036,
      "grad_norm": 1.0225517749786377,
      "learning_rate": 0.0003,
      "loss": 0.3333,
      "step": 12600
    },
    {
      "epoch": 11.659398668808814,
      "grad_norm": 1.043228268623352,
      "learning_rate": 0.0003,
      "loss": 0.3496,
      "step": 12700
    },
    {
      "epoch": 11.751204957539592,
      "grad_norm": 1.0331432819366455,
      "learning_rate": 0.0003,
      "loss": 0.3561,
      "step": 12800
    },
    {
      "epoch": 11.84301124627037,
      "grad_norm": 1.009142279624939,
      "learning_rate": 0.0003,
      "loss": 0.3591,
      "step": 12900
    },
    {
      "epoch": 11.934817535001148,
      "grad_norm": 0.9662990570068359,
      "learning_rate": 0.0003,
      "loss": 0.3627,
      "step": 13000
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.7702008733624454,
      "eval_loss": 0.5070006251335144,
      "eval_runtime": 8.8961,
      "eval_samples_per_second": 56.204,
      "eval_steps_per_second": 7.082,
      "step": 13071
    },
    {
      "epoch": 12.026623823731926,
      "grad_norm": 0.8747033476829529,
      "learning_rate": 0.0003,
      "loss": 0.3406,
      "step": 13100
    },
    {
      "epoch": 12.118430112462704,
      "grad_norm": 1.044632911682129,
      "learning_rate": 0.0003,
      "loss": 0.2725,
      "step": 13200
    },
    {
      "epoch": 12.210236401193482,
      "grad_norm": 1.118489384651184,
      "learning_rate": 0.0003,
      "loss": 0.2867,
      "step": 13300
    },
    {
      "epoch": 12.30204268992426,
      "grad_norm": 1.1297811269760132,
      "learning_rate": 0.0003,
      "loss": 0.3031,
      "step": 13400
    },
    {
      "epoch": 12.393848978655038,
      "grad_norm": 1.005750298500061,
      "learning_rate": 0.0003,
      "loss": 0.308,
      "step": 13500
    },
    {
      "epoch": 12.485655267385816,
      "grad_norm": 1.1464134454727173,
      "learning_rate": 0.0003,
      "loss": 0.3173,
      "step": 13600
    },
    {
      "epoch": 12.577461556116594,
      "grad_norm": 1.0724166631698608,
      "learning_rate": 0.0003,
      "loss": 0.3263,
      "step": 13700
    },
    {
      "epoch": 12.669267844847372,
      "grad_norm": 0.9200640320777893,
      "learning_rate": 0.0003,
      "loss": 0.3348,
      "step": 13800
    },
    {
      "epoch": 12.76107413357815,
      "grad_norm": 1.012441635131836,
      "learning_rate": 0.0003,
      "loss": 0.34,
      "step": 13900
    },
    {
      "epoch": 12.852880422308928,
      "grad_norm": 1.1097270250320435,
      "learning_rate": 0.0003,
      "loss": 0.3515,
      "step": 14000
    },
    {
      "epoch": 12.944686711039706,
      "grad_norm": 1.0985119342803955,
      "learning_rate": 0.0003,
      "loss": 0.3521,
      "step": 14100
    },
    {
      "epoch": 12.999770484278173,
      "eval_accuracy": 0.7713799126637555,
      "eval_loss": 0.4984254539012909,
      "eval_runtime": 7.8448,
      "eval_samples_per_second": 63.737,
      "eval_steps_per_second": 8.031,
      "step": 14160
    },
    {
      "epoch": 13.036492999770484,
      "grad_norm": 0.8010920882225037,
      "learning_rate": 0.0003,
      "loss": 0.3178,
      "step": 14200
    },
    {
      "epoch": 13.128299288501262,
      "grad_norm": 0.7816724181175232,
      "learning_rate": 0.0003,
      "loss": 0.2667,
      "step": 14300
    },
    {
      "epoch": 13.22010557723204,
      "grad_norm": 0.9443672895431519,
      "learning_rate": 0.0003,
      "loss": 0.2779,
      "step": 14400
    },
    {
      "epoch": 13.311911865962818,
      "grad_norm": 0.9471251964569092,
      "learning_rate": 0.0003,
      "loss": 0.2889,
      "step": 14500
    },
    {
      "epoch": 13.403718154693596,
      "grad_norm": 1.0766810178756714,
      "learning_rate": 0.0003,
      "loss": 0.301,
      "step": 14600
    },
    {
      "epoch": 13.495524443424374,
      "grad_norm": 1.0341782569885254,
      "learning_rate": 0.0003,
      "loss": 0.3079,
      "step": 14700
    },
    {
      "epoch": 13.587330732155152,
      "grad_norm": 0.9588432312011719,
      "learning_rate": 0.0003,
      "loss": 0.3184,
      "step": 14800
    },
    {
      "epoch": 13.67913702088593,
      "grad_norm": 1.0522781610488892,
      "learning_rate": 0.0003,
      "loss": 0.3228,
      "step": 14900
    },
    {
      "epoch": 13.770943309616708,
      "grad_norm": 1.0346490144729614,
      "learning_rate": 0.0003,
      "loss": 0.33,
      "step": 15000
    },
    {
      "epoch": 13.862749598347486,
      "grad_norm": 1.129906177520752,
      "learning_rate": 0.0003,
      "loss": 0.3378,
      "step": 15100
    },
    {
      "epoch": 13.954555887078264,
      "grad_norm": 0.9843702912330627,
      "learning_rate": 0.0003,
      "loss": 0.344,
      "step": 15200
    },
    {
      "epoch": 13.999540968556346,
      "eval_accuracy": 0.7721863173216885,
      "eval_loss": 0.4925004243850708,
      "eval_runtime": 8.9168,
      "eval_samples_per_second": 56.074,
      "eval_steps_per_second": 7.065,
      "step": 15249
    },
    {
      "epoch": 14.046362175809042,
      "grad_norm": 0.9825355410575867,
      "learning_rate": 0.0003,
      "loss": 0.2998,
      "step": 15300
    },
    {
      "epoch": 14.13816846453982,
      "grad_norm": 1.0309946537017822,
      "learning_rate": 0.0003,
      "loss": 0.2621,
      "step": 15400
    },
    {
      "epoch": 14.229974753270598,
      "grad_norm": 0.8775806427001953,
      "learning_rate": 0.0003,
      "loss": 0.2744,
      "step": 15500
    },
    {
      "epoch": 14.321781042001376,
      "grad_norm": 0.9635316133499146,
      "learning_rate": 0.0003,
      "loss": 0.2828,
      "step": 15600
    },
    {
      "epoch": 14.413587330732156,
      "grad_norm": 1.0143553018569946,
      "learning_rate": 0.0003,
      "loss": 0.29,
      "step": 15700
    },
    {
      "epoch": 14.505393619462934,
      "grad_norm": 1.0207178592681885,
      "learning_rate": 0.0003,
      "loss": 0.3022,
      "step": 15800
    },
    {
      "epoch": 14.597199908193712,
      "grad_norm": 1.07411527633667,
      "learning_rate": 0.0003,
      "loss": 0.3113,
      "step": 15900
    },
    {
      "epoch": 14.68900619692449,
      "grad_norm": 1.0961089134216309,
      "learning_rate": 0.0003,
      "loss": 0.3153,
      "step": 16000
    },
    {
      "epoch": 14.780812485655268,
      "grad_norm": 0.9983770847320557,
      "learning_rate": 0.0003,
      "loss": 0.323,
      "step": 16100
    },
    {
      "epoch": 14.872618774386046,
      "grad_norm": 1.0680583715438843,
      "learning_rate": 0.0003,
      "loss": 0.3275,
      "step": 16200
    },
    {
      "epoch": 14.964425063116824,
      "grad_norm": 1.0475704669952393,
      "learning_rate": 0.0003,
      "loss": 0.3341,
      "step": 16300
    },
    {
      "epoch": 14.999311452834519,
      "eval_accuracy": 0.7735604075691412,
      "eval_loss": 0.4847230315208435,
      "eval_runtime": 9.1634,
      "eval_samples_per_second": 54.565,
      "eval_steps_per_second": 6.875,
      "step": 16338
    },
    {
      "epoch": 15.056231351847602,
      "grad_norm": 0.7305497527122498,
      "learning_rate": 0.0003,
      "loss": 0.2829,
      "step": 16400
    },
    {
      "epoch": 15.14803764057838,
      "grad_norm": 0.936439037322998,
      "learning_rate": 0.0003,
      "loss": 0.2585,
      "step": 16500
    },
    {
      "epoch": 15.239843929309158,
      "grad_norm": 0.9285767078399658,
      "learning_rate": 0.0003,
      "loss": 0.262,
      "step": 16600
    },
    {
      "epoch": 15.331650218039936,
      "grad_norm": 1.0587061643600464,
      "learning_rate": 0.0003,
      "loss": 0.2795,
      "step": 16700
    },
    {
      "epoch": 15.423456506770714,
      "grad_norm": 1.0315121412277222,
      "learning_rate": 0.0003,
      "loss": 0.2858,
      "step": 16800
    },
    {
      "epoch": 15.515262795501492,
      "grad_norm": 1.007581114768982,
      "learning_rate": 0.0003,
      "loss": 0.2941,
      "step": 16900
    },
    {
      "epoch": 15.60706908423227,
      "grad_norm": 1.123679280281067,
      "learning_rate": 0.0003,
      "loss": 0.2998,
      "step": 17000
    },
    {
      "epoch": 15.698875372963048,
      "grad_norm": 1.245398759841919,
      "learning_rate": 0.0003,
      "loss": 0.3086,
      "step": 17100
    },
    {
      "epoch": 15.790681661693826,
      "grad_norm": 0.9799935817718506,
      "learning_rate": 0.0003,
      "loss": 0.3133,
      "step": 17200
    },
    {
      "epoch": 15.882487950424604,
      "grad_norm": 1.1230483055114746,
      "learning_rate": 0.0003,
      "loss": 0.3251,
      "step": 17300
    },
    {
      "epoch": 15.974294239155382,
      "grad_norm": 1.2086437940597534,
      "learning_rate": 0.0003,
      "loss": 0.3275,
      "step": 17400
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.7748180494905386,
      "eval_loss": 0.4808007478713989,
      "eval_runtime": 8.8958,
      "eval_samples_per_second": 56.206,
      "eval_steps_per_second": 7.082,
      "step": 17428
    },
    {
      "epoch": 16.06610052788616,
      "grad_norm": 0.8386597633361816,
      "learning_rate": 0.0003,
      "loss": 0.2679,
      "step": 17500
    },
    {
      "epoch": 16.15790681661694,
      "grad_norm": 0.865949273109436,
      "learning_rate": 0.0003,
      "loss": 0.2542,
      "step": 17600
    },
    {
      "epoch": 16.249713105347716,
      "grad_norm": 1.052322506904602,
      "learning_rate": 0.0003,
      "loss": 0.2646,
      "step": 17700
    },
    {
      "epoch": 16.341519394078496,
      "grad_norm": 0.8787432312965393,
      "learning_rate": 0.0003,
      "loss": 0.2671,
      "step": 17800
    },
    {
      "epoch": 16.433325682809272,
      "grad_norm": 0.9127426147460938,
      "learning_rate": 0.0003,
      "loss": 0.2772,
      "step": 17900
    },
    {
      "epoch": 16.525131971540052,
      "grad_norm": 0.9554562568664551,
      "learning_rate": 0.0003,
      "loss": 0.286,
      "step": 18000
    },
    {
      "epoch": 16.616938260270828,
      "grad_norm": 0.9382279515266418,
      "learning_rate": 0.0003,
      "loss": 0.2972,
      "step": 18100
    },
    {
      "epoch": 16.708744549001608,
      "grad_norm": 0.928916335105896,
      "learning_rate": 0.0003,
      "loss": 0.3031,
      "step": 18200
    },
    {
      "epoch": 16.800550837732384,
      "grad_norm": 0.9512932300567627,
      "learning_rate": 0.0003,
      "loss": 0.3058,
      "step": 18300
    },
    {
      "epoch": 16.892357126463164,
      "grad_norm": 1.2069571018218994,
      "learning_rate": 0.0003,
      "loss": 0.3168,
      "step": 18400
    },
    {
      "epoch": 16.98416341519394,
      "grad_norm": 1.1233407258987427,
      "learning_rate": 0.0003,
      "loss": 0.3223,
      "step": 18500
    },
    {
      "epoch": 16.999770484278173,
      "eval_accuracy": 0.7750887918486172,
      "eval_loss": 0.4775759279727936,
      "eval_runtime": 8.8897,
      "eval_samples_per_second": 56.245,
      "eval_steps_per_second": 7.087,
      "step": 18517
    },
    {
      "epoch": 17.07596970392472,
      "grad_norm": 0.9757447838783264,
      "learning_rate": 0.0003,
      "loss": 0.2534,
      "step": 18600
    },
    {
      "epoch": 17.167775992655496,
      "grad_norm": 0.7774830460548401,
      "learning_rate": 0.0003,
      "loss": 0.2487,
      "step": 18700
    },
    {
      "epoch": 17.259582281386276,
      "grad_norm": 0.9797073602676392,
      "learning_rate": 0.0003,
      "loss": 0.2564,
      "step": 18800
    },
    {
      "epoch": 17.351388570117052,
      "grad_norm": 0.9221961498260498,
      "learning_rate": 0.0003,
      "loss": 0.2675,
      "step": 18900
    },
    {
      "epoch": 17.443194858847832,
      "grad_norm": 0.9980019927024841,
      "learning_rate": 0.0003,
      "loss": 0.2759,
      "step": 19000
    },
    {
      "epoch": 17.53500114757861,
      "grad_norm": 1.1230412721633911,
      "learning_rate": 0.0003,
      "loss": 0.2884,
      "step": 19100
    },
    {
      "epoch": 17.626807436309388,
      "grad_norm": 1.0681424140930176,
      "learning_rate": 0.0003,
      "loss": 0.2894,
      "step": 19200
    },
    {
      "epoch": 17.718613725040164,
      "grad_norm": 1.0071791410446167,
      "learning_rate": 0.0003,
      "loss": 0.294,
      "step": 19300
    },
    {
      "epoch": 17.810420013770944,
      "grad_norm": 1.0229710340499878,
      "learning_rate": 0.0003,
      "loss": 0.305,
      "step": 19400
    },
    {
      "epoch": 17.90222630250172,
      "grad_norm": 1.1899082660675049,
      "learning_rate": 0.0003,
      "loss": 0.3114,
      "step": 19500
    },
    {
      "epoch": 17.9940325912325,
      "grad_norm": 1.1344189643859863,
      "learning_rate": 0.0003,
      "loss": 0.3155,
      "step": 19600
    },
    {
      "epoch": 17.999540968556346,
      "eval_accuracy": 0.7757671033478893,
      "eval_loss": 0.4804111421108246,
      "eval_runtime": 8.8913,
      "eval_samples_per_second": 56.234,
      "eval_steps_per_second": 7.086,
      "step": 19606
    },
    {
      "epoch": 18.085838879963276,
      "grad_norm": 0.9845724105834961,
      "learning_rate": 0.0003,
      "loss": 0.2418,
      "step": 19700
    },
    {
      "epoch": 18.177645168694056,
      "grad_norm": 0.879173994064331,
      "learning_rate": 0.0003,
      "loss": 0.2453,
      "step": 19800
    },
    {
      "epoch": 18.269451457424832,
      "grad_norm": 0.8902878761291504,
      "learning_rate": 0.0003,
      "loss": 0.2545,
      "step": 19900
    },
    {
      "epoch": 18.361257746155612,
      "grad_norm": 1.160914421081543,
      "learning_rate": 0.0003,
      "loss": 0.2648,
      "step": 20000
    },
    {
      "epoch": 18.45306403488639,
      "grad_norm": 0.9650329947471619,
      "learning_rate": 0.0003,
      "loss": 0.271,
      "step": 20100
    },
    {
      "epoch": 18.544870323617168,
      "grad_norm": 1.0495669841766357,
      "learning_rate": 0.0003,
      "loss": 0.2782,
      "step": 20200
    },
    {
      "epoch": 18.636676612347944,
      "grad_norm": 0.9071341156959534,
      "learning_rate": 0.0003,
      "loss": 0.2886,
      "step": 20300
    },
    {
      "epoch": 18.728482901078724,
      "grad_norm": 1.0429463386535645,
      "learning_rate": 0.0003,
      "loss": 0.2934,
      "step": 20400
    },
    {
      "epoch": 18.8202891898095,
      "grad_norm": 1.096994400024414,
      "learning_rate": 0.0003,
      "loss": 0.2992,
      "step": 20500
    },
    {
      "epoch": 18.91209547854028,
      "grad_norm": 1.0189508199691772,
      "learning_rate": 0.0003,
      "loss": 0.3033,
      "step": 20600
    },
    {
      "epoch": 18.99931145283452,
      "eval_accuracy": 0.7760727802037846,
      "eval_loss": 0.4787416458129883,
      "eval_runtime": 9.278,
      "eval_samples_per_second": 53.891,
      "eval_steps_per_second": 6.79,
      "step": 20695
    },
    {
      "epoch": 19.003901767271056,
      "grad_norm": 0.8995434045791626,
      "learning_rate": 0.0003,
      "loss": 0.3075,
      "step": 20700
    },
    {
      "epoch": 19.095708056001836,
      "grad_norm": 0.8744321465492249,
      "learning_rate": 0.0003,
      "loss": 0.2342,
      "step": 20800
    },
    {
      "epoch": 19.187514344732612,
      "grad_norm": 0.9471040964126587,
      "learning_rate": 0.0003,
      "loss": 0.2422,
      "step": 20900
    },
    {
      "epoch": 19.279320633463392,
      "grad_norm": 1.0133309364318848,
      "learning_rate": 0.0003,
      "loss": 0.2533,
      "step": 21000
    },
    {
      "epoch": 19.371126922194172,
      "grad_norm": 1.0121233463287354,
      "learning_rate": 0.0003,
      "loss": 0.2607,
      "step": 21100
    },
    {
      "epoch": 19.462933210924948,
      "grad_norm": 1.1256521940231323,
      "learning_rate": 0.0003,
      "loss": 0.2706,
      "step": 21200
    },
    {
      "epoch": 19.554739499655728,
      "grad_norm": 1.0843769311904907,
      "learning_rate": 0.0003,
      "loss": 0.2735,
      "step": 21300
    },
    {
      "epoch": 19.646545788386504,
      "grad_norm": 1.1019864082336426,
      "learning_rate": 0.0003,
      "loss": 0.286,
      "step": 21400
    },
    {
      "epoch": 19.738352077117284,
      "grad_norm": 1.135688066482544,
      "learning_rate": 0.0003,
      "loss": 0.2886,
      "step": 21500
    },
    {
      "epoch": 19.83015836584806,
      "grad_norm": 1.0445754528045654,
      "learning_rate": 0.0003,
      "loss": 0.2953,
      "step": 21600
    },
    {
      "epoch": 19.92196465457884,
      "grad_norm": 1.035718560218811,
      "learning_rate": 0.0003,
      "loss": 0.2989,
      "step": 21700
    },
    {
      "epoch": 19.995409685563462,
      "eval_accuracy": 0.7763580786026201,
      "eval_loss": 0.475557804107666,
      "eval_runtime": 9.1559,
      "eval_samples_per_second": 54.61,
      "eval_steps_per_second": 6.881,
      "step": 21780
    },
    {
      "epoch": 19.995409685563462,
      "step": 21780,
      "total_flos": 2.2953223726028554e+18,
      "train_loss": 0.5338389528903313,
      "train_runtime": 46593.736,
      "train_samples_per_second": 14.961,
      "train_steps_per_second": 0.467
    }
  ],
  "logging_steps": 100,
  "max_steps": 21780,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 500,
  "total_flos": 2.2953223726028554e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}