{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 50.0,
  "global_step": 80650,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.31,
      "learning_rate": 6.195786864931847e-05,
      "loss": 7.3347,
      "step": 500
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.97582756158962e-05,
      "loss": 4.6943,
      "step": 1000
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.913204664153402e-05,
      "loss": 1.9617,
      "step": 1500
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.850581766717182e-05,
      "loss": 1.4872,
      "step": 2000
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.787958869280964e-05,
      "loss": 1.3048,
      "step": 2500
    },
    {
      "epoch": 1.86,
      "learning_rate": 9.725335971844745e-05,
      "loss": 1.1881,
      "step": 3000
    },
    {
      "epoch": 2.17,
      "learning_rate": 9.662713074408527e-05,
      "loss": 1.105,
      "step": 3500
    },
    {
      "epoch": 2.48,
      "learning_rate": 9.600090176972308e-05,
      "loss": 1.0428,
      "step": 4000
    },
    {
      "epoch": 2.79,
      "learning_rate": 9.53746727953609e-05,
      "loss": 0.9918,
      "step": 4500
    },
    {
      "epoch": 3.1,
      "learning_rate": 9.47484438209987e-05,
      "loss": 0.9489,
      "step": 5000
    },
    {
      "epoch": 3.41,
      "learning_rate": 9.412221484663653e-05,
      "loss": 0.9121,
      "step": 5500
    },
    {
      "epoch": 3.72,
      "learning_rate": 9.349598587227433e-05,
      "loss": 0.8837,
      "step": 6000
    },
    {
      "epoch": 4.03,
      "learning_rate": 9.286975689791215e-05,
      "loss": 0.8581,
      "step": 6500
    },
    {
      "epoch": 4.34,
      "learning_rate": 9.224352792354997e-05,
      "loss": 0.8315,
      "step": 7000
    },
    {
      "epoch": 4.65,
      "learning_rate": 9.161729894918779e-05,
      "loss": 0.8117,
      "step": 7500
    },
    {
      "epoch": 4.96,
      "learning_rate": 9.09910699748256e-05,
      "loss": 0.795,
      "step": 8000
    },
    {
      "epoch": 5.27,
      "learning_rate": 9.036484100046342e-05,
      "loss": 0.776,
      "step": 8500
    },
    {
      "epoch": 5.58,
      "learning_rate": 8.973861202610123e-05,
      "loss": 0.76,
      "step": 9000
    },
    {
      "epoch": 5.89,
      "learning_rate": 8.911238305173905e-05,
      "loss": 0.7464,
      "step": 9500
    },
    {
      "epoch": 6.2,
      "learning_rate": 8.848615407737685e-05,
      "loss": 0.732,
      "step": 10000
    },
    {
      "epoch": 6.51,
      "learning_rate": 8.785992510301467e-05,
      "loss": 0.7206,
      "step": 10500
    },
    {
      "epoch": 6.82,
      "learning_rate": 8.723369612865248e-05,
      "loss": 0.709,
      "step": 11000
    },
    {
      "epoch": 7.13,
      "learning_rate": 8.66074671542903e-05,
      "loss": 0.7003,
      "step": 11500
    },
    {
      "epoch": 7.44,
      "learning_rate": 8.598123817992811e-05,
      "loss": 0.6869,
      "step": 12000
    },
    {
      "epoch": 7.75,
      "learning_rate": 8.535500920556593e-05,
      "loss": 0.6808,
      "step": 12500
    },
    {
      "epoch": 8.06,
      "learning_rate": 8.472878023120375e-05,
      "loss": 0.671,
      "step": 13000
    },
    {
      "epoch": 8.37,
      "learning_rate": 8.410255125684155e-05,
      "loss": 0.6615,
      "step": 13500
    },
    {
      "epoch": 8.68,
      "learning_rate": 8.347632228247937e-05,
      "loss": 0.6552,
      "step": 14000
    },
    {
      "epoch": 8.99,
      "learning_rate": 8.285009330811718e-05,
      "loss": 0.6484,
      "step": 14500
    },
    {
      "epoch": 9.3,
      "learning_rate": 8.2223864333755e-05,
      "loss": 0.6395,
      "step": 15000
    },
    {
      "epoch": 9.61,
      "learning_rate": 8.159763535939281e-05,
      "loss": 0.6351,
      "step": 15500
    },
    {
      "epoch": 9.92,
      "learning_rate": 8.097140638503063e-05,
      "loss": 0.6297,
      "step": 16000
    },
    {
      "epoch": 10.23,
      "learning_rate": 8.034642986861716e-05,
      "loss": 0.6214,
      "step": 16500
    },
    {
      "epoch": 10.54,
      "learning_rate": 7.972020089425498e-05,
      "loss": 0.617,
      "step": 17000
    },
    {
      "epoch": 10.85,
      "learning_rate": 7.909397191989279e-05,
      "loss": 0.6122,
      "step": 17500
    },
    {
      "epoch": 11.16,
      "learning_rate": 7.846774294553061e-05,
      "loss": 0.6055,
      "step": 18000
    },
    {
      "epoch": 11.47,
      "learning_rate": 7.784276642911714e-05,
      "loss": 0.5992,
      "step": 18500
    },
    {
      "epoch": 11.78,
      "learning_rate": 7.721653745475495e-05,
      "loss": 0.5957,
      "step": 19000
    },
    {
      "epoch": 12.09,
      "learning_rate": 7.65915609383415e-05,
      "loss": 0.5919,
      "step": 19500
    },
    {
      "epoch": 12.4,
      "learning_rate": 7.596533196397932e-05,
      "loss": 0.5857,
      "step": 20000
    },
    {
      "epoch": 12.71,
      "learning_rate": 7.533910298961712e-05,
      "loss": 0.5825,
      "step": 20500
    },
    {
      "epoch": 13.02,
      "learning_rate": 7.471287401525494e-05,
      "loss": 0.5787,
      "step": 21000
    },
    {
      "epoch": 13.33,
      "learning_rate": 7.408664504089275e-05,
      "loss": 0.5721,
      "step": 21500
    },
    {
      "epoch": 13.64,
      "learning_rate": 7.346041606653057e-05,
      "loss": 0.5694,
      "step": 22000
    },
    {
      "epoch": 13.95,
      "learning_rate": 7.283418709216838e-05,
      "loss": 0.5672,
      "step": 22500
    },
    {
      "epoch": 14.26,
      "learning_rate": 7.22079581178062e-05,
      "loss": 0.5619,
      "step": 23000
    },
    {
      "epoch": 14.57,
      "learning_rate": 7.158298160139275e-05,
      "loss": 0.5586,
      "step": 23500
    },
    {
      "epoch": 14.88,
      "learning_rate": 7.095800508497928e-05,
      "loss": 0.5559,
      "step": 24000
    },
    {
      "epoch": 15.19,
      "learning_rate": 7.033177611061709e-05,
      "loss": 0.5519,
      "step": 24500
    },
    {
      "epoch": 15.5,
      "learning_rate": 6.97055471362549e-05,
      "loss": 0.5485,
      "step": 25000
    },
    {
      "epoch": 15.81,
      "learning_rate": 6.907931816189271e-05,
      "loss": 0.5459,
      "step": 25500
    },
    {
      "epoch": 16.12,
      "learning_rate": 6.845308918753053e-05,
      "loss": 0.5423,
      "step": 26000
    },
    {
      "epoch": 16.43,
      "learning_rate": 6.782686021316834e-05,
      "loss": 0.538,
      "step": 26500
    },
    {
      "epoch": 16.74,
      "learning_rate": 6.720063123880616e-05,
      "loss": 0.5366,
      "step": 27000
    },
    {
      "epoch": 17.05,
      "learning_rate": 6.657440226444397e-05,
      "loss": 0.5351,
      "step": 27500
    },
    {
      "epoch": 17.36,
      "learning_rate": 6.594817329008179e-05,
      "loss": 0.5299,
      "step": 28000
    },
    {
      "epoch": 17.67,
      "learning_rate": 6.532194431571961e-05,
      "loss": 0.5283,
      "step": 28500
    },
    {
      "epoch": 17.98,
      "learning_rate": 6.469571534135743e-05,
      "loss": 0.5259,
      "step": 29000
    },
    {
      "epoch": 18.29,
      "learning_rate": 6.406948636699523e-05,
      "loss": 0.5221,
      "step": 29500
    },
    {
      "epoch": 18.6,
      "learning_rate": 6.344325739263305e-05,
      "loss": 0.5197,
      "step": 30000
    },
    {
      "epoch": 18.91,
      "learning_rate": 6.281702841827086e-05,
      "loss": 0.5185,
      "step": 30500
    },
    {
      "epoch": 19.22,
      "learning_rate": 6.219079944390868e-05,
      "loss": 0.5138,
      "step": 31000
    },
    {
      "epoch": 19.53,
      "learning_rate": 6.156457046954649e-05,
      "loss": 0.5121,
      "step": 31500
    },
    {
      "epoch": 19.84,
      "learning_rate": 6.09383414951843e-05,
      "loss": 0.5112,
      "step": 32000
    },
    {
      "epoch": 20.15,
      "learning_rate": 6.0312112520822115e-05,
      "loss": 0.5072,
      "step": 32500
    },
    {
      "epoch": 20.46,
      "learning_rate": 5.968588354645993e-05,
      "loss": 0.5049,
      "step": 33000
    },
    {
      "epoch": 20.77,
      "learning_rate": 5.905965457209774e-05,
      "loss": 0.5043,
      "step": 33500
    },
    {
      "epoch": 21.08,
      "learning_rate": 5.8433425597735556e-05,
      "loss": 0.5013,
      "step": 34000
    },
    {
      "epoch": 21.39,
      "learning_rate": 5.780719662337337e-05,
      "loss": 0.4981,
      "step": 34500
    },
    {
      "epoch": 21.7,
      "learning_rate": 5.718096764901118e-05,
      "loss": 0.4975,
      "step": 35000
    },
    {
      "epoch": 22.01,
      "learning_rate": 5.6554738674648996e-05,
      "loss": 0.4959,
      "step": 35500
    },
    {
      "epoch": 22.32,
      "learning_rate": 5.5928509700286816e-05,
      "loss": 0.4921,
      "step": 36000
    },
    {
      "epoch": 22.63,
      "learning_rate": 5.530228072592463e-05,
      "loss": 0.4921,
      "step": 36500
    },
    {
      "epoch": 22.94,
      "learning_rate": 5.467605175156244e-05,
      "loss": 0.4886,
      "step": 37000
    },
    {
      "epoch": 23.25,
      "learning_rate": 5.404982277720026e-05,
      "loss": 0.4856,
      "step": 37500
    },
    {
      "epoch": 23.56,
      "learning_rate": 5.342359380283807e-05,
      "loss": 0.4847,
      "step": 38000
    },
    {
      "epoch": 23.87,
      "learning_rate": 5.2797364828475884e-05,
      "loss": 0.4839,
      "step": 38500
    },
    {
      "epoch": 24.18,
      "learning_rate": 5.21711358541137e-05,
      "loss": 0.4814,
      "step": 39000
    },
    {
      "epoch": 24.49,
      "learning_rate": 5.154490687975152e-05,
      "loss": 0.4796,
      "step": 39500
    },
    {
      "epoch": 24.8,
      "learning_rate": 5.091867790538933e-05,
      "loss": 0.4784,
      "step": 40000
    },
    {
      "epoch": 25.11,
      "learning_rate": 5.0292448931027144e-05,
      "loss": 0.4757,
      "step": 40500
    },
    {
      "epoch": 25.42,
      "learning_rate": 4.966621995666496e-05,
      "loss": 0.4744,
      "step": 41000
    },
    {
      "epoch": 25.73,
      "learning_rate": 4.903999098230277e-05,
      "loss": 0.4734,
      "step": 41500
    },
    {
      "epoch": 26.04,
      "learning_rate": 4.841376200794059e-05,
      "loss": 0.4715,
      "step": 42000
    },
    {
      "epoch": 26.35,
      "learning_rate": 4.7787533033578405e-05,
      "loss": 0.4682,
      "step": 42500
    },
    {
      "epoch": 26.66,
      "learning_rate": 4.716130405921622e-05,
      "loss": 0.4683,
      "step": 43000
    },
    {
      "epoch": 26.97,
      "learning_rate": 4.653507508485403e-05,
      "loss": 0.4671,
      "step": 43500
    },
    {
      "epoch": 27.28,
      "learning_rate": 4.5908846110491845e-05,
      "loss": 0.4637,
      "step": 44000
    },
    {
      "epoch": 27.59,
      "learning_rate": 4.528261713612966e-05,
      "loss": 0.4633,
      "step": 44500
    },
    {
      "epoch": 27.9,
      "learning_rate": 4.465638816176747e-05,
      "loss": 0.4623,
      "step": 45000
    },
    {
      "epoch": 28.21,
      "learning_rate": 4.4030159187405286e-05,
      "loss": 0.4595,
      "step": 45500
    },
    {
      "epoch": 28.52,
      "learning_rate": 4.34039302130431e-05,
      "loss": 0.4593,
      "step": 46000
    },
    {
      "epoch": 28.83,
      "learning_rate": 4.277770123868091e-05,
      "loss": 0.4581,
      "step": 46500
    },
    {
      "epoch": 29.14,
      "learning_rate": 4.2151472264318726e-05,
      "loss": 0.4558,
      "step": 47000
    },
    {
      "epoch": 29.45,
      "learning_rate": 4.152524328995654e-05,
      "loss": 0.4534,
      "step": 47500
    },
    {
      "epoch": 29.76,
      "learning_rate": 4.089901431559436e-05,
      "loss": 0.4538,
      "step": 48000
    },
    {
      "epoch": 30.07,
      "learning_rate": 4.0274037799180894e-05,
      "loss": 0.4518,
      "step": 48500
    },
    {
      "epoch": 30.38,
      "learning_rate": 3.9647808824818714e-05,
      "loss": 0.4499,
      "step": 49000
    },
    {
      "epoch": 30.69,
      "learning_rate": 3.902157985045653e-05,
      "loss": 0.4503,
      "step": 49500
    },
    {
      "epoch": 31.0,
      "learning_rate": 3.839535087609434e-05,
      "loss": 0.4482,
      "step": 50000
    },
    {
      "epoch": 31.31,
      "learning_rate": 3.7770374359680875e-05,
      "loss": 0.4454,
      "step": 50500
    },
    {
      "epoch": 31.62,
      "learning_rate": 3.714414538531869e-05,
      "loss": 0.445,
      "step": 51000
    },
    {
      "epoch": 31.93,
      "learning_rate": 3.65179164109565e-05,
      "loss": 0.4443,
      "step": 51500
    },
    {
      "epoch": 32.24,
      "learning_rate": 3.5891687436594315e-05,
      "loss": 0.443,
      "step": 52000
    },
    {
      "epoch": 32.55,
      "learning_rate": 3.5266710920180856e-05,
      "loss": 0.4416,
      "step": 52500
    },
    {
      "epoch": 32.86,
      "learning_rate": 3.464048194581867e-05,
      "loss": 0.4414,
      "step": 53000
    },
    {
      "epoch": 33.17,
      "learning_rate": 3.401425297145648e-05,
      "loss": 0.4391,
      "step": 53500
    },
    {
      "epoch": 33.48,
      "learning_rate": 3.3388023997094296e-05,
      "loss": 0.4369,
      "step": 54000
    },
    {
      "epoch": 33.79,
      "learning_rate": 3.2763047480680836e-05,
      "loss": 0.4368,
      "step": 54500
    },
    {
      "epoch": 34.1,
      "learning_rate": 3.213681850631865e-05,
      "loss": 0.4363,
      "step": 55000
    },
    {
      "epoch": 34.41,
      "learning_rate": 3.1510589531956463e-05,
      "loss": 0.434,
      "step": 55500
    },
    {
      "epoch": 34.72,
      "learning_rate": 3.0884360557594284e-05,
      "loss": 0.4337,
      "step": 56000
    },
    {
      "epoch": 35.03,
      "learning_rate": 3.0259384041180817e-05,
      "loss": 0.4324,
      "step": 56500
    },
    {
      "epoch": 35.34,
      "learning_rate": 2.9633155066818634e-05,
      "loss": 0.4311,
      "step": 57000
    },
    {
      "epoch": 35.65,
      "learning_rate": 2.9006926092456448e-05,
      "loss": 0.4298,
      "step": 57500
    },
    {
      "epoch": 35.96,
      "learning_rate": 2.838069711809426e-05,
      "loss": 0.4305,
      "step": 58000
    },
    {
      "epoch": 36.27,
      "learning_rate": 2.77557206016808e-05,
      "loss": 0.4273,
      "step": 58500
    },
    {
      "epoch": 36.58,
      "learning_rate": 2.7129491627318615e-05,
      "loss": 0.4274,
      "step": 59000
    },
    {
      "epoch": 36.89,
      "learning_rate": 2.650326265295643e-05,
      "loss": 0.4267,
      "step": 59500
    },
    {
      "epoch": 37.2,
      "learning_rate": 2.5877033678594242e-05,
      "loss": 0.4245,
      "step": 60000
    },
    {
      "epoch": 37.51,
      "learning_rate": 2.5250804704232056e-05,
      "loss": 0.4234,
      "step": 60500
    },
    {
      "epoch": 37.82,
      "learning_rate": 2.462457572986987e-05,
      "loss": 0.4237,
      "step": 61000
    },
    {
      "epoch": 38.13,
      "learning_rate": 2.3998346755507686e-05,
      "loss": 0.4217,
      "step": 61500
    },
    {
      "epoch": 38.44,
      "learning_rate": 2.33721177811455e-05,
      "loss": 0.42,
      "step": 62000
    },
    {
      "epoch": 38.75,
      "learning_rate": 2.274714126473204e-05,
      "loss": 0.4207,
      "step": 62500
    },
    {
      "epoch": 39.06,
      "learning_rate": 2.2120912290369853e-05,
      "loss": 0.4206,
      "step": 63000
    },
    {
      "epoch": 39.37,
      "learning_rate": 2.1494683316007667e-05,
      "loss": 0.4177,
      "step": 63500
    },
    {
      "epoch": 39.68,
      "learning_rate": 2.086845434164548e-05,
      "loss": 0.4183,
      "step": 64000
    },
    {
      "epoch": 39.99,
      "learning_rate": 2.0243477825232017e-05,
      "loss": 0.4163,
      "step": 64500
    },
    {
      "epoch": 40.3,
      "learning_rate": 1.961724885086983e-05,
      "loss": 0.4151,
      "step": 65000
    },
    {
      "epoch": 40.61,
      "learning_rate": 1.8991019876507644e-05,
      "loss": 0.4159,
      "step": 65500
    },
    {
      "epoch": 40.92,
      "learning_rate": 1.836479090214546e-05,
      "loss": 0.4146,
      "step": 66000
    },
    {
      "epoch": 41.23,
      "learning_rate": 1.7738561927783275e-05,
      "loss": 0.4134,
      "step": 66500
    },
    {
      "epoch": 41.54,
      "learning_rate": 1.7112332953421088e-05,
      "loss": 0.4124,
      "step": 67000
    },
    {
      "epoch": 41.85,
      "learning_rate": 1.6486103979058905e-05,
      "loss": 0.4119,
      "step": 67500
    },
    {
      "epoch": 42.16,
      "learning_rate": 1.585987500469672e-05,
      "loss": 0.4106,
      "step": 68000
    },
    {
      "epoch": 42.47,
      "learning_rate": 1.5233646030334534e-05,
      "loss": 0.4104,
      "step": 68500
    },
    {
      "epoch": 42.78,
      "learning_rate": 1.4607417055972347e-05,
      "loss": 0.4094,
      "step": 69000
    },
    {
      "epoch": 43.09,
      "learning_rate": 1.398118808161016e-05,
      "loss": 0.4091,
      "step": 69500
    },
    {
      "epoch": 43.4,
      "learning_rate": 1.3354959107247974e-05,
      "loss": 0.4082,
      "step": 70000
    },
    {
      "epoch": 43.71,
      "learning_rate": 1.2728730132885787e-05,
      "loss": 0.4074,
      "step": 70500
    },
    {
      "epoch": 44.02,
      "learning_rate": 1.2102501158523603e-05,
      "loss": 0.4067,
      "step": 71000
    },
    {
      "epoch": 44.33,
      "learning_rate": 1.1476272184161418e-05,
      "loss": 0.4062,
      "step": 71500
    },
    {
      "epoch": 44.64,
      "learning_rate": 1.0850043209799233e-05,
      "loss": 0.4051,
      "step": 72000
    },
    {
      "epoch": 44.95,
      "learning_rate": 1.0223814235437046e-05,
      "loss": 0.4048,
      "step": 72500
    },
    {
      "epoch": 45.26,
      "learning_rate": 9.59758526107486e-06,
      "loss": 0.4034,
      "step": 73000
    },
    {
      "epoch": 45.57,
      "learning_rate": 8.971356286712675e-06,
      "loss": 0.4036,
      "step": 73500
    },
    {
      "epoch": 45.88,
      "learning_rate": 8.345127312350489e-06,
      "loss": 0.4038,
      "step": 74000
    },
    {
      "epoch": 46.19,
      "learning_rate": 7.718898337988302e-06,
      "loss": 0.4021,
      "step": 74500
    },
    {
      "epoch": 46.5,
      "learning_rate": 7.092669363626117e-06,
      "loss": 0.4023,
      "step": 75000
    },
    {
      "epoch": 46.81,
      "learning_rate": 6.466440389263931e-06,
      "loss": 0.4011,
      "step": 75500
    },
    {
      "epoch": 47.12,
      "learning_rate": 5.840211414901745e-06,
      "loss": 0.4004,
      "step": 76000
    },
    {
      "epoch": 47.43,
      "learning_rate": 5.2139824405395585e-06,
      "loss": 0.3997,
      "step": 76500
    },
    {
      "epoch": 47.74,
      "learning_rate": 4.587753466177374e-06,
      "loss": 0.3999,
      "step": 77000
    },
    {
      "epoch": 48.05,
      "learning_rate": 3.961524491815188e-06,
      "loss": 0.3988,
      "step": 77500
    },
    {
      "epoch": 48.36,
      "learning_rate": 3.3352955174530015e-06,
      "loss": 0.3988,
      "step": 78000
    },
    {
      "epoch": 48.67,
      "learning_rate": 2.709066543090816e-06,
      "loss": 0.3978,
      "step": 78500
    },
    {
      "epoch": 48.98,
      "learning_rate": 2.08283756872863e-06,
      "loss": 0.3983,
      "step": 79000
    },
    {
      "epoch": 49.29,
      "learning_rate": 1.4566085943664442e-06,
      "loss": 0.398,
      "step": 79500
    },
    {
      "epoch": 49.6,
      "learning_rate": 8.303796200042584e-07,
      "loss": 0.3974,
      "step": 80000
    },
    {
      "epoch": 49.91,
      "learning_rate": 2.0415064564207257e-07,
      "loss": 0.3975,
      "step": 80500
    },
    {
      "epoch": 50.0,
      "step": 80650,
      "total_flos": 2.180340696717392e+19,
      "train_loss": 0.005664803918222426,
      "train_runtime": 6206.924,
      "train_samples_per_second": 19956.785,
      "train_steps_per_second": 12.994
    }
  ],
  "max_steps": 80650,
  "num_train_epochs": 50,
  "total_flos": 2.180340696717392e+19,
  "trial_name": null,
  "trial_params": null
}