{
  "best_global_step": 59997,
  "best_metric": 0.8794049695261634,
  "best_model_checkpoint": "./doc_type_v1_primary_model_multilingual-e5-small/checkpoint-59997",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 59997,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.025001250062503127,
      "grad_norm": 6.811492443084717,
      "learning_rate": 4.958414587396037e-05,
      "loss": 1.2379,
      "step": 500
    },
    {
      "epoch": 0.050002500125006254,
      "grad_norm": 27.201915740966797,
      "learning_rate": 4.916745837291865e-05,
      "loss": 0.8651,
      "step": 1000
    },
    {
      "epoch": 0.07500375018750938,
      "grad_norm": 38.990150451660156,
      "learning_rate": 4.875077087187693e-05,
      "loss": 0.7379,
      "step": 1500
    },
    {
      "epoch": 0.10000500025001251,
      "grad_norm": 7.001402378082275,
      "learning_rate": 4.833408337083521e-05,
      "loss": 0.7292,
      "step": 2000
    },
    {
      "epoch": 0.12500625031251564,
      "grad_norm": 13.325078010559082,
      "learning_rate": 4.791739586979349e-05,
      "loss": 0.696,
      "step": 2500
    },
    {
      "epoch": 0.15000750037501875,
      "grad_norm": 8.727783203125,
      "learning_rate": 4.7500708368751775e-05,
      "loss": 0.711,
      "step": 3000
    },
    {
      "epoch": 0.17500875043752187,
      "grad_norm": 16.32622528076172,
      "learning_rate": 4.708402086771005e-05,
      "loss": 0.6598,
      "step": 3500
    },
    {
      "epoch": 0.20001000050002501,
      "grad_norm": 7.856052875518799,
      "learning_rate": 4.6667333366668335e-05,
      "loss": 0.6057,
      "step": 4000
    },
    {
      "epoch": 0.22501125056252813,
      "grad_norm": 7.5791335105896,
      "learning_rate": 4.625064586562662e-05,
      "loss": 0.585,
      "step": 4500
    },
    {
      "epoch": 0.2500125006250313,
      "grad_norm": 10.478697776794434,
      "learning_rate": 4.58339583645849e-05,
      "loss": 0.5894,
      "step": 5000
    },
    {
      "epoch": 0.27501375068753436,
      "grad_norm": 12.097478866577148,
      "learning_rate": 4.541727086354318e-05,
      "loss": 0.5759,
      "step": 5500
    },
    {
      "epoch": 0.3000150007500375,
      "grad_norm": 9.861705780029297,
      "learning_rate": 4.5000583362501456e-05,
      "loss": 0.5605,
      "step": 6000
    },
    {
      "epoch": 0.32501625081254065,
      "grad_norm": 23.986101150512695,
      "learning_rate": 4.458389586145974e-05,
      "loss": 0.5548,
      "step": 6500
    },
    {
      "epoch": 0.35001750087504374,
      "grad_norm": 28.77222442626953,
      "learning_rate": 4.416720836041802e-05,
      "loss": 0.5508,
      "step": 7000
    },
    {
      "epoch": 0.3750187509375469,
      "grad_norm": 5.619863510131836,
      "learning_rate": 4.3750520859376306e-05,
      "loss": 0.5182,
      "step": 7500
    },
    {
      "epoch": 0.40002000100005003,
      "grad_norm": 5.042713642120361,
      "learning_rate": 4.333383335833458e-05,
      "loss": 0.5597,
      "step": 8000
    },
    {
      "epoch": 0.4250212510625531,
      "grad_norm": 8.755766868591309,
      "learning_rate": 4.291714585729287e-05,
      "loss": 0.5342,
      "step": 8500
    },
    {
      "epoch": 0.45002250112505626,
      "grad_norm": 9.902745246887207,
      "learning_rate": 4.250045835625115e-05,
      "loss": 0.5154,
      "step": 9000
    },
    {
      "epoch": 0.47502375118755935,
      "grad_norm": 18.608970642089844,
      "learning_rate": 4.2083770855209434e-05,
      "loss": 0.5101,
      "step": 9500
    },
    {
      "epoch": 0.5000250012500626,
      "grad_norm": 7.529025077819824,
      "learning_rate": 4.166708335416771e-05,
      "loss": 0.5153,
      "step": 10000
    },
    {
      "epoch": 0.5250262513125656,
      "grad_norm": 4.368379592895508,
      "learning_rate": 4.125039585312599e-05,
      "loss": 0.4962,
      "step": 10500
    },
    {
      "epoch": 0.5500275013750687,
      "grad_norm": 5.460580348968506,
      "learning_rate": 4.083370835208427e-05,
      "loss": 0.5055,
      "step": 11000
    },
    {
      "epoch": 0.5750287514375719,
      "grad_norm": 17.047475814819336,
      "learning_rate": 4.0417020851042555e-05,
      "loss": 0.5289,
      "step": 11500
    },
    {
      "epoch": 0.600030001500075,
      "grad_norm": 8.671713829040527,
      "learning_rate": 4.000033335000083e-05,
      "loss": 0.5024,
      "step": 12000
    },
    {
      "epoch": 0.6250312515625781,
      "grad_norm": 13.457786560058594,
      "learning_rate": 3.9583645848959115e-05,
      "loss": 0.481,
      "step": 12500
    },
    {
      "epoch": 0.6500325016250813,
      "grad_norm": 1.2953449487686157,
      "learning_rate": 3.91669583479174e-05,
      "loss": 0.4843,
      "step": 13000
    },
    {
      "epoch": 0.6750337516875844,
      "grad_norm": 0.7997756004333496,
      "learning_rate": 3.875027084687568e-05,
      "loss": 0.4519,
      "step": 13500
    },
    {
      "epoch": 0.7000350017500875,
      "grad_norm": 5.482394218444824,
      "learning_rate": 3.833358334583396e-05,
      "loss": 0.4829,
      "step": 14000
    },
    {
      "epoch": 0.7250362518125907,
      "grad_norm": 13.774917602539062,
      "learning_rate": 3.791689584479224e-05,
      "loss": 0.4746,
      "step": 14500
    },
    {
      "epoch": 0.7500375018750938,
      "grad_norm": 7.448230743408203,
      "learning_rate": 3.750020834375052e-05,
      "loss": 0.5123,
      "step": 15000
    },
    {
      "epoch": 0.7750387519375969,
      "grad_norm": 11.042137145996094,
      "learning_rate": 3.70835208427088e-05,
      "loss": 0.5058,
      "step": 15500
    },
    {
      "epoch": 0.8000400020001001,
      "grad_norm": 9.733834266662598,
      "learning_rate": 3.666683334166709e-05,
      "loss": 0.453,
      "step": 16000
    },
    {
      "epoch": 0.8250412520626031,
      "grad_norm": 9.845958709716797,
      "learning_rate": 3.6250145840625363e-05,
      "loss": 0.4604,
      "step": 16500
    },
    {
      "epoch": 0.8500425021251062,
      "grad_norm": 24.350200653076172,
      "learning_rate": 3.583345833958365e-05,
      "loss": 0.4689,
      "step": 17000
    },
    {
      "epoch": 0.8750437521876093,
      "grad_norm": 25.33015251159668,
      "learning_rate": 3.541677083854193e-05,
      "loss": 0.4689,
      "step": 17500
    },
    {
      "epoch": 0.9000450022501125,
      "grad_norm": 3.9758975505828857,
      "learning_rate": 3.5000083337500214e-05,
      "loss": 0.4704,
      "step": 18000
    },
    {
      "epoch": 0.9250462523126156,
      "grad_norm": 5.771573066711426,
      "learning_rate": 3.458339583645849e-05,
      "loss": 0.4367,
      "step": 18500
    },
    {
      "epoch": 0.9500475023751187,
      "grad_norm": 5.9610819816589355,
      "learning_rate": 3.4166708335416775e-05,
      "loss": 0.451,
      "step": 19000
    },
    {
      "epoch": 0.9750487524376219,
      "grad_norm": 13.63049030303955,
      "learning_rate": 3.375002083437505e-05,
      "loss": 0.4538,
      "step": 19500
    },
    {
      "epoch": 1.0,
      "eval_f1": 0.8655922269409259,
      "eval_loss": 0.4387092590332031,
      "eval_runtime": 10.5773,
      "eval_samples_per_second": 1891.79,
      "eval_steps_per_second": 236.545,
      "step": 19999
    },
    {
      "epoch": 1.000050002500125,
      "grad_norm": 5.536550045013428,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.4367,
      "step": 20000
    },
    {
      "epoch": 1.025051252562628,
      "grad_norm": 16.51177406311035,
      "learning_rate": 3.291664583229161e-05,
      "loss": 0.3614,
      "step": 20500
    },
    {
      "epoch": 1.0500525026251313,
      "grad_norm": 20.4051570892334,
      "learning_rate": 3.2499958331249895e-05,
      "loss": 0.3757,
      "step": 21000
    },
    {
      "epoch": 1.0750537526876345,
      "grad_norm": 4.4692912101745605,
      "learning_rate": 3.208327083020818e-05,
      "loss": 0.3197,
      "step": 21500
    },
    {
      "epoch": 1.1000550027501375,
      "grad_norm": 9.222668647766113,
      "learning_rate": 3.166658332916646e-05,
      "loss": 0.3649,
      "step": 22000
    },
    {
      "epoch": 1.1250562528126407,
      "grad_norm": 20.703449249267578,
      "learning_rate": 3.124989582812474e-05,
      "loss": 0.3736,
      "step": 22500
    },
    {
      "epoch": 1.1500575028751436,
      "grad_norm": 10.590250015258789,
      "learning_rate": 3.083320832708302e-05,
      "loss": 0.3325,
      "step": 23000
    },
    {
      "epoch": 1.1750587529376468,
      "grad_norm": 39.34541320800781,
      "learning_rate": 3.0416520826041306e-05,
      "loss": 0.3472,
      "step": 23500
    },
    {
      "epoch": 1.20006000300015,
      "grad_norm": 9.01701545715332,
      "learning_rate": 2.9999833324999583e-05,
      "loss": 0.3513,
      "step": 24000
    },
    {
      "epoch": 1.2250612530626532,
      "grad_norm": 20.631200790405273,
      "learning_rate": 2.9583145823957863e-05,
      "loss": 0.3699,
      "step": 24500
    },
    {
      "epoch": 1.2500625031251562,
      "grad_norm": 0.09197826683521271,
      "learning_rate": 2.9166458322916147e-05,
      "loss": 0.3847,
      "step": 25000
    },
    {
      "epoch": 1.2750637531876594,
      "grad_norm": 5.867861747741699,
      "learning_rate": 2.8749770821874427e-05,
      "loss": 0.3252,
      "step": 25500
    },
    {
      "epoch": 1.3000650032501624,
      "grad_norm": 21.79376220703125,
      "learning_rate": 2.833308332083271e-05,
      "loss": 0.3573,
      "step": 26000
    },
    {
      "epoch": 1.3250662533126656,
      "grad_norm": 10.486989974975586,
      "learning_rate": 2.791639581979099e-05,
      "loss": 0.3704,
      "step": 26500
    },
    {
      "epoch": 1.3500675033751688,
      "grad_norm": 10.07336139678955,
      "learning_rate": 2.7499708318749275e-05,
      "loss": 0.3269,
      "step": 27000
    },
    {
      "epoch": 1.375068753437672,
      "grad_norm": 0.7917349338531494,
      "learning_rate": 2.7083020817707555e-05,
      "loss": 0.3637,
      "step": 27500
    },
    {
      "epoch": 1.400070003500175,
      "grad_norm": 4.454699993133545,
      "learning_rate": 2.666633331666584e-05,
      "loss": 0.3503,
      "step": 28000
    },
    {
      "epoch": 1.4250712535626782,
      "grad_norm": 4.157910346984863,
      "learning_rate": 2.6249645815624112e-05,
      "loss": 0.3503,
      "step": 28500
    },
    {
      "epoch": 1.4500725036251811,
      "grad_norm": 0.06325356662273407,
      "learning_rate": 2.5832958314582395e-05,
      "loss": 0.3246,
      "step": 29000
    },
    {
      "epoch": 1.4750737536876843,
      "grad_norm": 19.81574058532715,
      "learning_rate": 2.5416270813540676e-05,
      "loss": 0.3507,
      "step": 29500
    },
    {
      "epoch": 1.5000750037501875,
      "grad_norm": 0.12496486306190491,
      "learning_rate": 2.499958331249896e-05,
      "loss": 0.3274,
      "step": 30000
    },
    {
      "epoch": 1.5250762538126907,
      "grad_norm": 14.442770004272461,
      "learning_rate": 2.458289581145724e-05,
      "loss": 0.3926,
      "step": 30500
    },
    {
      "epoch": 1.5500775038751937,
      "grad_norm": 30.787338256835938,
      "learning_rate": 2.4166208310415523e-05,
      "loss": 0.3445,
      "step": 31000
    },
    {
      "epoch": 1.575078753937697,
      "grad_norm": 16.105411529541016,
      "learning_rate": 2.3749520809373803e-05,
      "loss": 0.3397,
      "step": 31500
    },
    {
      "epoch": 1.6000800040002,
      "grad_norm": 0.16195891797542572,
      "learning_rate": 2.3332833308332083e-05,
      "loss": 0.3337,
      "step": 32000
    },
    {
      "epoch": 1.625081254062703,
      "grad_norm": 5.337989807128906,
      "learning_rate": 2.2916145807290363e-05,
      "loss": 0.3398,
      "step": 32500
    },
    {
      "epoch": 1.6500825041252063,
      "grad_norm": 12.693595886230469,
      "learning_rate": 2.2499458306248647e-05,
      "loss": 0.3457,
      "step": 33000
    },
    {
      "epoch": 1.6750837541877095,
      "grad_norm": 0.6089347004890442,
      "learning_rate": 2.2082770805206927e-05,
      "loss": 0.3252,
      "step": 33500
    },
    {
      "epoch": 1.7000850042502125,
      "grad_norm": 21.343097686767578,
      "learning_rate": 2.166608330416521e-05,
      "loss": 0.3691,
      "step": 34000
    },
    {
      "epoch": 1.7250862543127157,
      "grad_norm": 7.946640968322754,
      "learning_rate": 2.124939580312349e-05,
      "loss": 0.3334,
      "step": 34500
    },
    {
      "epoch": 1.7500875043752186,
      "grad_norm": 7.454244613647461,
      "learning_rate": 2.083270830208177e-05,
      "loss": 0.3363,
      "step": 35000
    },
    {
      "epoch": 1.7750887544377218,
      "grad_norm": 0.1911257654428482,
      "learning_rate": 2.0416020801040055e-05,
      "loss": 0.3454,
      "step": 35500
    },
    {
      "epoch": 1.800090004500225,
      "grad_norm": 13.01964282989502,
      "learning_rate": 1.9999333299998335e-05,
      "loss": 0.3189,
      "step": 36000
    },
    {
      "epoch": 1.8250912545627282,
      "grad_norm": 0.19861529767513275,
      "learning_rate": 1.9582645798956615e-05,
      "loss": 0.3422,
      "step": 36500
    },
    {
      "epoch": 1.8500925046252312,
      "grad_norm": 19.57059097290039,
      "learning_rate": 1.9165958297914895e-05,
      "loss": 0.3355,
      "step": 37000
    },
    {
      "epoch": 1.8750937546877344,
      "grad_norm": 4.671505451202393,
      "learning_rate": 1.874927079687318e-05,
      "loss": 0.3195,
      "step": 37500
    },
    {
      "epoch": 1.9000950047502374,
      "grad_norm": 13.636198997497559,
      "learning_rate": 1.833258329583146e-05,
      "loss": 0.2937,
      "step": 38000
    },
    {
      "epoch": 1.9250962548127406,
      "grad_norm": 3.3196206092834473,
      "learning_rate": 1.7915895794789743e-05,
      "loss": 0.3382,
      "step": 38500
    },
    {
      "epoch": 1.9500975048752438,
      "grad_norm": 10.638530731201172,
      "learning_rate": 1.749920829374802e-05,
      "loss": 0.3509,
      "step": 39000
    },
    {
      "epoch": 1.975098754937747,
      "grad_norm": 25.718746185302734,
      "learning_rate": 1.7082520792706303e-05,
      "loss": 0.3244,
      "step": 39500
    },
    {
      "epoch": 2.0,
      "eval_f1": 0.8739136212588781,
      "eval_loss": 0.5150496959686279,
      "eval_runtime": 10.5221,
      "eval_samples_per_second": 1901.71,
      "eval_steps_per_second": 237.785,
      "step": 39998
    },
    {
      "epoch": 2.00010000500025,
      "grad_norm": 5.354879856109619,
      "learning_rate": 1.6665833291664583e-05,
      "loss": 0.3325,
      "step": 40000
    },
    {
      "epoch": 2.025101255062753,
      "grad_norm": 26.597732543945312,
      "learning_rate": 1.6249145790622867e-05,
      "loss": 0.2202,
      "step": 40500
    },
    {
      "epoch": 2.050102505125256,
      "grad_norm": 19.418195724487305,
      "learning_rate": 1.5832458289581144e-05,
      "loss": 0.2126,
      "step": 41000
    },
    {
      "epoch": 2.0751037551877594,
      "grad_norm": 30.896820068359375,
      "learning_rate": 1.5415770788539427e-05,
      "loss": 0.1978,
      "step": 41500
    },
    {
      "epoch": 2.1001050052502626,
      "grad_norm": 33.62570571899414,
      "learning_rate": 1.4999083287497709e-05,
      "loss": 0.2235,
      "step": 42000
    },
    {
      "epoch": 2.1251062553127658,
      "grad_norm": 34.06229782104492,
      "learning_rate": 1.4582395786455991e-05,
      "loss": 0.2285,
      "step": 42500
    },
    {
      "epoch": 2.150107505375269,
      "grad_norm": 7.419727325439453,
      "learning_rate": 1.4165708285414273e-05,
      "loss": 0.2114,
      "step": 43000
    },
    {
      "epoch": 2.1751087554377717,
      "grad_norm": 0.09190714359283447,
      "learning_rate": 1.3749020784372551e-05,
      "loss": 0.2401,
      "step": 43500
    },
    {
      "epoch": 2.200110005500275,
      "grad_norm": 3.367650032043457,
      "learning_rate": 1.3332333283330833e-05,
      "loss": 0.2316,
      "step": 44000
    },
    {
      "epoch": 2.225111255562778,
      "grad_norm": 1.1675509214401245,
      "learning_rate": 1.2915645782289115e-05,
      "loss": 0.2356,
      "step": 44500
    },
    {
      "epoch": 2.2501125056252813,
      "grad_norm": 9.24765682220459,
      "learning_rate": 1.2498958281247395e-05,
      "loss": 0.2265,
      "step": 45000
    },
    {
      "epoch": 2.2751137556877845,
      "grad_norm": 49.01567459106445,
      "learning_rate": 1.2082270780205677e-05,
      "loss": 0.2156,
      "step": 45500
    },
    {
      "epoch": 2.3001150057502873,
      "grad_norm": 1.4991744756698608,
      "learning_rate": 1.1665583279163959e-05,
      "loss": 0.1985,
      "step": 46000
    },
    {
      "epoch": 2.3251162558127905,
      "grad_norm": 23.972734451293945,
      "learning_rate": 1.124889577812224e-05,
      "loss": 0.2341,
      "step": 46500
    },
    {
      "epoch": 2.3501175058752937,
      "grad_norm": 0.17075876891613007,
      "learning_rate": 1.0832208277080521e-05,
      "loss": 0.2253,
      "step": 47000
    },
    {
      "epoch": 2.375118755937797,
      "grad_norm": 31.144304275512695,
      "learning_rate": 1.0415520776038801e-05,
      "loss": 0.2155,
      "step": 47500
    },
    {
      "epoch": 2.4001200060003,
      "grad_norm": 0.0451333224773407,
      "learning_rate": 9.998833274997083e-06,
      "loss": 0.1964,
      "step": 48000
    },
    {
      "epoch": 2.4251212560628033,
      "grad_norm": 0.07999496906995773,
      "learning_rate": 9.582145773955365e-06,
      "loss": 0.2406,
      "step": 48500
    },
    {
      "epoch": 2.4501225061253065,
      "grad_norm": 0.9701845049858093,
      "learning_rate": 9.165458272913647e-06,
      "loss": 0.2345,
      "step": 49000
    },
    {
      "epoch": 2.475123756187809,
      "grad_norm": 1.0465730428695679,
      "learning_rate": 8.748770771871927e-06,
      "loss": 0.2179,
      "step": 49500
    },
    {
      "epoch": 2.5001250062503124,
      "grad_norm": 0.01202826015651226,
      "learning_rate": 8.332083270830209e-06,
      "loss": 0.2076,
      "step": 50000
    },
    {
      "epoch": 2.5251262563128156,
      "grad_norm": 0.02468780241906643,
      "learning_rate": 7.915395769788491e-06,
      "loss": 0.2387,
      "step": 50500
    },
    {
      "epoch": 2.550127506375319,
      "grad_norm": 2.958998918533325,
      "learning_rate": 7.498708268746771e-06,
      "loss": 0.2114,
      "step": 51000
    },
    {
      "epoch": 2.575128756437822,
      "grad_norm": 0.1065281331539154,
      "learning_rate": 7.082020767705053e-06,
      "loss": 0.1916,
      "step": 51500
    },
    {
      "epoch": 2.6001300065003248,
      "grad_norm": 34.561431884765625,
      "learning_rate": 6.665333266663333e-06,
      "loss": 0.2074,
      "step": 52000
    },
    {
      "epoch": 2.625131256562828,
      "grad_norm": 0.007779615931212902,
      "learning_rate": 6.248645765621615e-06,
      "loss": 0.2133,
      "step": 52500
    },
    {
      "epoch": 2.650132506625331,
      "grad_norm": 0.13916213810443878,
      "learning_rate": 5.831958264579896e-06,
      "loss": 0.2301,
      "step": 53000
    },
    {
      "epoch": 2.6751337566878344,
      "grad_norm": 42.24327087402344,
      "learning_rate": 5.415270763538177e-06,
      "loss": 0.2216,
      "step": 53500
    },
    {
      "epoch": 2.7001350067503376,
      "grad_norm": 7.163196563720703,
      "learning_rate": 4.998583262496458e-06,
      "loss": 0.2313,
      "step": 54000
    },
    {
      "epoch": 2.7251362568128408,
      "grad_norm": 0.8102510571479797,
      "learning_rate": 4.581895761454739e-06,
      "loss": 0.1916,
      "step": 54500
    },
    {
      "epoch": 2.750137506875344,
      "grad_norm": 7.8397417068481445,
      "learning_rate": 4.16520826041302e-06,
      "loss": 0.2055,
      "step": 55000
    },
    {
      "epoch": 2.7751387569378467,
      "grad_norm": 1.0101325511932373,
      "learning_rate": 3.7485207593713018e-06,
      "loss": 0.2059,
      "step": 55500
    },
    {
      "epoch": 2.80014000700035,
      "grad_norm": 0.05691508203744888,
      "learning_rate": 3.3318332583295837e-06,
      "loss": 0.2021,
      "step": 56000
    },
    {
      "epoch": 2.825141257062853,
      "grad_norm": 0.08343327045440674,
      "learning_rate": 2.9151457572878643e-06,
      "loss": 0.2075,
      "step": 56500
    },
    {
      "epoch": 2.8501425071253563,
      "grad_norm": 0.06794146448373795,
      "learning_rate": 2.4984582562461457e-06,
      "loss": 0.1644,
      "step": 57000
    },
    {
      "epoch": 2.8751437571878595,
      "grad_norm": 0.023177076131105423,
      "learning_rate": 2.081770755204427e-06,
      "loss": 0.2023,
      "step": 57500
    },
    {
      "epoch": 2.9001450072503623,
      "grad_norm": 0.15367339551448822,
      "learning_rate": 1.6650832541627082e-06,
      "loss": 0.2175,
      "step": 58000
    },
    {
      "epoch": 2.9251462573128655,
      "grad_norm": 0.029408954083919525,
      "learning_rate": 1.2483957531209895e-06,
      "loss": 0.2073,
      "step": 58500
    },
    {
      "epoch": 2.9501475073753687,
      "grad_norm": 2.278526782989502,
      "learning_rate": 8.317082520792706e-07,
      "loss": 0.2154,
      "step": 59000
    },
    {
      "epoch": 2.975148757437872,
      "grad_norm": 15.807899475097656,
      "learning_rate": 4.150207510375519e-07,
      "loss": 0.2132,
      "step": 59500
    },
    {
      "epoch": 3.0,
      "eval_f1": 0.8794049695261634,
      "eval_loss": 0.6096033453941345,
      "eval_runtime": 10.6241,
      "eval_samples_per_second": 1883.457,
      "eval_steps_per_second": 235.503,
      "step": 59997
    },
    {
      "epoch": 3.0,
      "step": 59997,
      "total_flos": 7906263495201792.0,
      "train_loss": 0.3725671201518653,
      "train_runtime": 1624.1756,
      "train_samples_per_second": 295.512,
      "train_steps_per_second": 36.94
    }
  ],
  "logging_steps": 500,
  "max_steps": 59997,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7906263495201792.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}