{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 375,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "learning_rate": 4.997807075247146e-05,
      "loss": 1.446,
      "step": 5
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.991232148123761e-05,
      "loss": 1.2288,
      "step": 10
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.980286753286195e-05,
      "loss": 1.2506,
      "step": 15
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.964990092676263e-05,
      "loss": 1.1189,
      "step": 20
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9453690018345144e-05,
      "loss": 1.104,
      "step": 25
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.9214579028215776e-05,
      "loss": 1.2144,
      "step": 30
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.893298743830168e-05,
      "loss": 1.0983,
      "step": 35
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.860940925593703e-05,
      "loss": 1.1491,
      "step": 40
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.8244412147206284e-05,
      "loss": 1.1163,
      "step": 45
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.783863644106502e-05,
      "loss": 1.0851,
      "step": 50
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.7392794005985326e-05,
      "loss": 1.1639,
      "step": 55
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.690766700109659e-05,
      "loss": 1.1499,
      "step": 60
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.638410650401267e-05,
      "loss": 1.0992,
      "step": 65
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.5823031017752485e-05,
      "loss": 1.1705,
      "step": 70
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.522542485937369e-05,
      "loss": 1.0919,
      "step": 75
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.4592336433146e-05,
      "loss": 1.1709,
      "step": 80
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.3924876391293915e-05,
      "loss": 1.2184,
      "step": 85
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.3224215685535294e-05,
      "loss": 1.1752,
      "step": 90
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.249158351283414e-05,
      "loss": 1.1208,
      "step": 95
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.172826515897146e-05,
      "loss": 1.0406,
      "step": 100
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.093559974371725e-05,
      "loss": 1.1003,
      "step": 105
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.011497787155938e-05,
      "loss": 1.0862,
      "step": 110
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.92678391921108e-05,
      "loss": 1.0911,
      "step": 115
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.8395669874474915e-05,
      "loss": 1.2021,
      "step": 120
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 1.1072,
      "step": 125
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.6582400877996546e-05,
      "loss": 0.7294,
      "step": 130
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.564448228912682e-05,
      "loss": 0.7575,
      "step": 135
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.4687889661302576e-05,
      "loss": 0.7256,
      "step": 140
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.3714301183045385e-05,
      "loss": 0.6674,
      "step": 145
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.272542485937369e-05,
      "loss": 0.6673,
      "step": 150
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.172299551538164e-05,
      "loss": 0.6968,
      "step": 155
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.0708771752766394e-05,
      "loss": 0.6963,
      "step": 160
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.9684532864643122e-05,
      "loss": 0.6005,
      "step": 165
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.8652075714060295e-05,
      "loss": 0.6712,
      "step": 170
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.761321158169134e-05,
      "loss": 0.6859,
      "step": 175
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.656976298823284e-05,
      "loss": 0.6128,
      "step": 180
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.5523560497083926e-05,
      "loss": 0.6395,
      "step": 185
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.447643950291608e-05,
      "loss": 0.7517,
      "step": 190
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.3430237011767167e-05,
      "loss": 0.6147,
      "step": 195
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.238678841830867e-05,
      "loss": 0.7306,
      "step": 200
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.1347924285939714e-05,
      "loss": 0.7235,
      "step": 205
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.031546713535688e-05,
      "loss": 0.595,
      "step": 210
    },
    {
      "epoch": 1.72,
      "learning_rate": 1.9291228247233605e-05,
      "loss": 0.6347,
      "step": 215
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.827700448461836e-05,
      "loss": 0.6938,
      "step": 220
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.7274575140626318e-05,
      "loss": 0.6408,
      "step": 225
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.6285698816954624e-05,
      "loss": 0.6296,
      "step": 230
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.5312110338697426e-05,
      "loss": 0.6688,
      "step": 235
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.4355517710873184e-05,
      "loss": 0.661,
      "step": 240
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.3417599122003464e-05,
      "loss": 0.6846,
      "step": 245
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.2500000000000006e-05,
      "loss": 0.6501,
      "step": 250
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.1604330125525079e-05,
      "loss": 0.5096,
      "step": 255
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.0732160807889211e-05,
      "loss": 0.4049,
      "step": 260
    },
    {
      "epoch": 2.12,
      "learning_rate": 9.88502212844063e-06,
      "loss": 0.4556,
      "step": 265
    },
    {
      "epoch": 2.16,
      "learning_rate": 9.064400256282757e-06,
      "loss": 0.4205,
      "step": 270
    },
    {
      "epoch": 2.2,
      "learning_rate": 8.271734841028553e-06,
      "loss": 0.3667,
      "step": 275
    },
    {
      "epoch": 2.24,
      "learning_rate": 7.508416487165862e-06,
      "loss": 0.3817,
      "step": 280
    },
    {
      "epoch": 2.28,
      "learning_rate": 6.775784314464717e-06,
      "loss": 0.4369,
      "step": 285
    },
    {
      "epoch": 2.32,
      "learning_rate": 6.075123608706093e-06,
      "loss": 0.3672,
      "step": 290
    },
    {
      "epoch": 2.36,
      "learning_rate": 5.4076635668540075e-06,
      "loss": 0.4072,
      "step": 295
    },
    {
      "epoch": 2.4,
      "learning_rate": 4.7745751406263165e-06,
      "loss": 0.416,
      "step": 300
    },
    {
      "epoch": 2.44,
      "learning_rate": 4.176968982247514e-06,
      "loss": 0.3504,
      "step": 305
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.6158934959873353e-06,
      "loss": 0.4019,
      "step": 310
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.092332998903416e-06,
      "loss": 0.4057,
      "step": 315
    },
    {
      "epoch": 2.56,
      "learning_rate": 2.6072059940146775e-06,
      "loss": 0.3894,
      "step": 320
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.1613635589349756e-06,
      "loss": 0.3795,
      "step": 325
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.7555878527937164e-06,
      "loss": 0.3934,
      "step": 330
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.3905907440629752e-06,
      "loss": 0.423,
      "step": 335
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.067012561698319e-06,
      "loss": 0.4121,
      "step": 340
    },
    {
      "epoch": 2.76,
      "learning_rate": 7.854209717842231e-07,
      "loss": 0.4026,
      "step": 345
    },
    {
      "epoch": 2.8,
      "learning_rate": 5.463099816548579e-07,
      "loss": 0.4056,
      "step": 350
    },
    {
      "epoch": 2.84,
      "learning_rate": 3.5009907323737825e-07,
      "loss": 0.3625,
      "step": 355
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.9713246713805588e-07,
      "loss": 0.4279,
      "step": 360
    },
    {
      "epoch": 2.92,
      "learning_rate": 8.767851876239074e-08,
      "loss": 0.3752,
      "step": 365
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.192924752854042e-08,
      "loss": 0.3992,
      "step": 370
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0,
      "loss": 0.3638,
      "step": 375
    },
    {
      "epoch": 3.0,
      "step": 375,
      "total_flos": 6.763019396186112e+16,
      "train_loss": 0.7424986867904663,
      "train_runtime": 1158.4304,
      "train_samples_per_second": 5.179,
      "train_steps_per_second": 0.324
    }
  ],
  "logging_steps": 5,
  "max_steps": 375,
  "num_train_epochs": 3,
  "save_steps": 100,
  "total_flos": 6.763019396186112e+16,
  "trial_name": null,
  "trial_params": null
}