{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9789902790843525,
  "global_step": 38000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "learning_rate": 4.934671265809554e-05,
      "loss": 2.5519,
      "step": 500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.8693425316191076e-05,
      "loss": 2.5343,
      "step": 1000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.804013797428661e-05,
      "loss": 2.4991,
      "step": 1500
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.738685063238215e-05,
      "loss": 2.475,
      "step": 2000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.6733563290477686e-05,
      "loss": 2.4791,
      "step": 2500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.608027594857322e-05,
      "loss": 2.4725,
      "step": 3000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.542698860666876e-05,
      "loss": 2.4255,
      "step": 3500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.4773701264764296e-05,
      "loss": 2.4452,
      "step": 4000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.412041392285983e-05,
      "loss": 2.4272,
      "step": 4500
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.346712658095537e-05,
      "loss": 2.435,
      "step": 5000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.2813839239050906e-05,
      "loss": 2.4164,
      "step": 5500
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.216055189714644e-05,
      "loss": 2.3954,
      "step": 6000
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.150726455524198e-05,
      "loss": 2.3677,
      "step": 6500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.0853977213337515e-05,
      "loss": 2.3643,
      "step": 7000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.020068987143305e-05,
      "loss": 2.3596,
      "step": 7500
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.954740252952859e-05,
      "loss": 2.3601,
      "step": 8000
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.8894115187624125e-05,
      "loss": 2.3391,
      "step": 8500
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.824082784571966e-05,
      "loss": 2.3418,
      "step": 9000
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.75875405038152e-05,
      "loss": 2.3256,
      "step": 9500
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.6934253161910735e-05,
      "loss": 2.3386,
      "step": 10000
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.628096582000627e-05,
      "loss": 2.3078,
      "step": 10500
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.562767847810181e-05,
      "loss": 2.3059,
      "step": 11000
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.497439113619735e-05,
      "loss": 2.3098,
      "step": 11500
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.432110379429289e-05,
      "loss": 2.3021,
      "step": 12000
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.366781645238842e-05,
      "loss": 2.2801,
      "step": 12500
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.3014529110483955e-05,
      "loss": 2.0588,
      "step": 13000
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.236124176857949e-05,
      "loss": 1.8245,
      "step": 13500
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.170795442667503e-05,
      "loss": 1.8411,
      "step": 14000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.1054667084770565e-05,
      "loss": 1.8267,
      "step": 14500
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0401379742866105e-05,
      "loss": 1.8376,
      "step": 15000
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9748092400961642e-05,
      "loss": 1.863,
      "step": 15500
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.909480505905718e-05,
      "loss": 1.8302,
      "step": 16000
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8441517717152715e-05,
      "loss": 1.8444,
      "step": 16500
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.7788230375248252e-05,
      "loss": 1.8417,
      "step": 17000
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.7134943033343785e-05,
      "loss": 1.8576,
      "step": 17500
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.648165569143932e-05,
      "loss": 1.8421,
      "step": 18000
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5828368349534858e-05,
      "loss": 1.8459,
      "step": 18500
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.51750810076304e-05,
      "loss": 1.8438,
      "step": 19000
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4521793665725935e-05,
      "loss": 1.8263,
      "step": 19500
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.386850632382147e-05,
      "loss": 1.8451,
      "step": 20000
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.3215218981917008e-05,
      "loss": 1.8439,
      "step": 20500
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.256193164001254e-05,
      "loss": 1.8529,
      "step": 21000
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.190864429810808e-05,
      "loss": 1.8184,
      "step": 21500
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.1255356956203618e-05,
      "loss": 1.8364,
      "step": 22000
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.0602069614299155e-05,
      "loss": 1.831,
      "step": 22500
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.994878227239469e-05,
      "loss": 1.8103,
      "step": 23000
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9295494930490228e-05,
      "loss": 1.8078,
      "step": 23500
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8642207588585765e-05,
      "loss": 1.8047,
      "step": 24000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.79889202466813e-05,
      "loss": 1.8155,
      "step": 24500
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7335632904776838e-05,
      "loss": 1.8136,
      "step": 25000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6682345562872375e-05,
      "loss": 1.8019,
      "step": 25500
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.602905822096791e-05,
      "loss": 1.3885,
      "step": 26000
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5375770879063448e-05,
      "loss": 1.3774,
      "step": 26500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4722483537158985e-05,
      "loss": 1.3693,
      "step": 27000
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.4069196195254521e-05,
      "loss": 1.3827,
      "step": 27500
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.341590885335006e-05,
      "loss": 1.3755,
      "step": 28000
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2762621511445595e-05,
      "loss": 1.3739,
      "step": 28500
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2109334169541131e-05,
      "loss": 1.3798,
      "step": 29000
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.1456046827636668e-05,
      "loss": 1.3705,
      "step": 29500
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.0802759485732206e-05,
      "loss": 1.3847,
      "step": 30000
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0149472143827741e-05,
      "loss": 1.371,
      "step": 30500
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.49618480192328e-06,
      "loss": 1.3661,
      "step": 31000
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.842897460018814e-06,
      "loss": 1.3863,
      "step": 31500
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.189610118114353e-06,
      "loss": 1.3819,
      "step": 32000
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.536322776209888e-06,
      "loss": 1.3762,
      "step": 32500
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.883035434305425e-06,
      "loss": 1.3733,
      "step": 33000
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.229748092400962e-06,
      "loss": 1.3591,
      "step": 33500
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.576460750496499e-06,
      "loss": 1.3756,
      "step": 34000
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.923173408592035e-06,
      "loss": 1.3447,
      "step": 34500
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.269886066687572e-06,
      "loss": 1.3656,
      "step": 35000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.616598724783109e-06,
      "loss": 1.3549,
      "step": 35500
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.963311382878645e-06,
      "loss": 1.3691,
      "step": 36000
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.310024040974182e-06,
      "loss": 1.3534,
      "step": 36500
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.6567366990697188e-06,
      "loss": 1.3391,
      "step": 37000
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.0034493571652555e-06,
      "loss": 1.3589,
      "step": 37500
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.5016201526079235e-07,
      "loss": 1.3393,
      "step": 38000
    }
  ],
  "max_steps": 38268,
  "num_train_epochs": 3,
  "total_flos": 1.948577863019397e+17,
  "trial_name": null,
  "trial_params": null
}