{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.389776357827476,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.06,
      "learning_rate": 4.999493072462126e-05,
      "loss": 0.7286,
      "step": 10
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.9974340316719906e-05,
      "loss": 0.6461,
      "step": 20
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.993792498360407e-05,
      "loss": 0.5379,
      "step": 30
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.9885707799754036e-05,
      "loss": 0.5064,
      "step": 40
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.981772185245135e-05,
      "loss": 0.4653,
      "step": 50
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.9734010220813144e-05,
      "loss": 0.4585,
      "step": 60
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.963462594849515e-05,
      "loss": 0.4565,
      "step": 70
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.951963201008076e-05,
      "loss": 0.442,
      "step": 80
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.9389101271177355e-05,
      "loss": 0.4378,
      "step": 90
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.9243116442245124e-05,
      "loss": 0.4238,
      "step": 100
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.9081770026187914e-05,
      "loss": 0.4417,
      "step": 110
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.89051642597389e-05,
      "loss": 0.4266,
      "step": 120
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.8713411048678635e-05,
      "loss": 0.4272,
      "step": 130
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.850663189692619e-05,
      "loss": 0.4253,
      "step": 140
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.828495782954852e-05,
      "loss": 0.4334,
      "step": 150
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.804852930973679e-05,
      "loss": 0.4334,
      "step": 160
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.7797496149802256e-05,
      "loss": 0.4143,
      "step": 170
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.753201741624804e-05,
      "loss": 0.4245,
      "step": 180
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.7252261328977115e-05,
      "loss": 0.4214,
      "step": 190
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.6958405154700154e-05,
      "loss": 0.4308,
      "step": 200
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.665063509461097e-05,
      "loss": 0.4232,
      "step": 210
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.6329146166400614e-05,
      "loss": 0.4037,
      "step": 220
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.599414208068495e-05,
      "loss": 0.4025,
      "step": 230
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.5645835111923896e-05,
      "loss": 0.4105,
      "step": 240
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.528444596391433e-05,
      "loss": 0.4129,
      "step": 250
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.491020362994168e-05,
      "loss": 0.4112,
      "step": 260
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.452334524767894e-05,
      "loss": 0.4149,
      "step": 270
    },
    {
      "epoch": 1.79,
      "learning_rate": 4.4124115948925e-05,
      "loss": 0.4104,
      "step": 280
    },
    {
      "epoch": 1.85,
      "learning_rate": 4.371276870427753e-05,
      "loss": 0.3973,
      "step": 290
    },
    {
      "epoch": 1.92,
      "learning_rate": 4.3289564162838806e-05,
      "loss": 0.4288,
      "step": 300
    },
    {
      "epoch": 1.98,
      "learning_rate": 4.2854770487056114e-05,
      "loss": 0.4095,
      "step": 310
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.240866318280132e-05,
      "loss": 0.4163,
      "step": 320
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.195152492479727e-05,
      "loss": 0.3989,
      "step": 330
    },
    {
      "epoch": 2.17,
      "learning_rate": 4.148364537750172e-05,
      "loss": 0.4179,
      "step": 340
    },
    {
      "epoch": 2.24,
      "learning_rate": 4.1005321011562206e-05,
      "loss": 0.408,
      "step": 350
    },
    {
      "epoch": 2.3,
      "learning_rate": 4.051685491595817e-05,
      "loss": 0.4083,
      "step": 360
    },
    {
      "epoch": 2.36,
      "learning_rate": 4.001855660594948e-05,
      "loss": 0.4225,
      "step": 370
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.951074182695284e-05,
      "loss": 0.4152,
      "step": 380
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.8993732354470556e-05,
      "loss": 0.3981,
      "step": 390
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.8467855790198366e-05,
      "loss": 0.4082,
      "step": 400
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.793344535444142e-05,
      "loss": 0.3991,
      "step": 410
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.739083967497017e-05,
      "loss": 0.4041,
      "step": 420
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.684038257244973e-05,
      "loss": 0.3868,
      "step": 430
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.6282422842578845e-05,
      "loss": 0.4161,
      "step": 440
    },
    {
      "epoch": 2.88,
      "learning_rate": 3.5717314035076355e-05,
      "loss": 0.3967,
      "step": 450
    },
    {
      "epoch": 2.94,
      "learning_rate": 3.5145414229655435e-05,
      "loss": 0.3786,
      "step": 460
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.456708580912725e-05,
      "loss": 0.4084,
      "step": 470
    },
    {
      "epoch": 3.07,
      "learning_rate": 3.3982695229778045e-05,
      "loss": 0.4099,
      "step": 480
    },
    {
      "epoch": 3.13,
      "learning_rate": 3.339261278916512e-05,
      "loss": 0.4037,
      "step": 490
    },
    {
      "epoch": 3.19,
      "learning_rate": 3.2797212391478724e-05,
      "loss": 0.3938,
      "step": 500
    },
    {
      "epoch": 3.26,
      "learning_rate": 3.2196871310618655e-05,
      "loss": 0.4048,
      "step": 510
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.159196995113563e-05,
      "loss": 0.396,
      "step": 520
    },
    {
      "epoch": 3.39,
      "learning_rate": 3.098289160718895e-05,
      "loss": 0.407,
      "step": 530
    },
    {
      "epoch": 3.45,
      "learning_rate": 3.0370022219673173e-05,
      "loss": 0.3919,
      "step": 540
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.975375013166767e-05,
      "loss": 0.3961,
      "step": 550
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.9134465842364035e-05,
      "loss": 0.3882,
      "step": 560
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.851256175962732e-05,
      "loss": 0.3885,
      "step": 570
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.788843195134781e-05,
      "loss": 0.4067,
      "step": 580
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.726247189574095e-05,
      "loss": 0.3994,
      "step": 590
    },
    {
      "epoch": 3.83,
      "learning_rate": 2.663507823075358e-05,
      "loss": 0.3993,
      "step": 600
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.600664850273538e-05,
      "loss": 0.4045,
      "step": 610
    },
    {
      "epoch": 3.96,
      "learning_rate": 2.5377580914534647e-05,
      "loss": 0.4035,
      "step": 620
    },
    {
      "epoch": 4.03,
      "learning_rate": 2.4748274073178114e-05,
      "loss": 0.39,
      "step": 630
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.411912673729464e-05,
      "loss": 0.3876,
      "step": 640
    },
    {
      "epoch": 4.15,
      "learning_rate": 2.3490537564442847e-05,
      "loss": 0.3915,
      "step": 650
    },
    {
      "epoch": 4.22,
      "learning_rate": 2.2862904858502856e-05,
      "loss": 0.3896,
      "step": 660
    },
    {
      "epoch": 4.28,
      "learning_rate": 2.2236626317292007e-05,
      "loss": 0.408,
      "step": 670
    },
    {
      "epoch": 4.35,
      "learning_rate": 2.1612098780564714e-05,
      "loss": 0.3927,
      "step": 680
    },
    {
      "epoch": 4.41,
      "learning_rate": 2.098971797855599e-05,
      "loss": 0.3882,
      "step": 690
    },
    {
      "epoch": 4.47,
      "learning_rate": 2.0369878281228067e-05,
      "loss": 0.397,
      "step": 700
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.9752972448378814e-05,
      "loss": 0.3955,
      "step": 710
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.9139391380770594e-05,
      "loss": 0.4069,
      "step": 720
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.852952387243698e-05,
      "loss": 0.3826,
      "step": 730
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.7923756364324492e-05,
      "loss": 0.3868,
      "step": 740
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.7322472699425234e-05,
      "loss": 0.4035,
      "step": 750
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.672605387955583e-05,
      "loss": 0.3924,
      "step": 760
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.613487782393661e-05,
      "loss": 0.3951,
      "step": 770
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.5549319129724115e-05,
      "loss": 0.3904,
      "step": 780
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.4969748834648556e-05,
      "loss": 0.3931,
      "step": 790
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.4396534181906725e-05,
      "loss": 0.3961,
      "step": 800
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.3830038387459354e-05,
      "loss": 0.3893,
      "step": 810
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.327062040988024e-05,
      "loss": 0.3767,
      "step": 820
    },
    {
      "epoch": 5.3,
      "learning_rate": 1.2718634722903073e-05,
      "loss": 0.3855,
      "step": 830
    },
    {
      "epoch": 5.37,
      "learning_rate": 1.2174431090810112e-05,
      "loss": 0.3932,
      "step": 840
    },
    {
      "epoch": 5.43,
      "learning_rate": 1.1638354346804971e-05,
      "loss": 0.401,
      "step": 850
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.1110744174509952e-05,
      "loss": 0.4076,
      "step": 860
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.059193489272637e-05,
      "loss": 0.3749,
      "step": 870
    },
    {
      "epoch": 5.62,
      "learning_rate": 1.0082255243594399e-05,
      "loss": 0.3881,
      "step": 880
    },
    {
      "epoch": 5.69,
      "learning_rate": 9.582028184286423e-06,
      "loss": 0.3845,
      "step": 890
    },
    {
      "epoch": 5.75,
      "learning_rate": 9.091570682366086e-06,
      "loss": 0.4057,
      "step": 900
    },
    {
      "epoch": 5.81,
      "learning_rate": 8.611193514942634e-06,
      "loss": 0.3855,
      "step": 910
    },
    {
      "epoch": 5.88,
      "learning_rate": 8.141201071747784e-06,
      "loss": 0.3875,
      "step": 920
    },
    {
      "epoch": 5.94,
      "learning_rate": 7.681891162260015e-06,
      "loss": 0.3762,
      "step": 930
    },
    {
      "epoch": 6.01,
      "learning_rate": 7.2335548269983485e-06,
      "loss": 0.3959,
      "step": 940
    },
    {
      "epoch": 6.07,
      "learning_rate": 6.796476153105294e-06,
      "loss": 0.3851,
      "step": 950
    },
    {
      "epoch": 6.13,
      "learning_rate": 6.370932094335746e-06,
      "loss": 0.3876,
      "step": 960
    },
    {
      "epoch": 6.2,
      "learning_rate": 5.957192295566022e-06,
      "loss": 0.4071,
      "step": 970
    },
    {
      "epoch": 6.26,
      "learning_rate": 5.555518921934047e-06,
      "loss": 0.3933,
      "step": 980
    },
    {
      "epoch": 6.33,
      "learning_rate": 5.166166492719124e-06,
      "loss": 0.3877,
      "step": 990
    },
    {
      "epoch": 6.39,
      "learning_rate": 4.789381720066441e-06,
      "loss": 0.3762,
      "step": 1000
    }
  ],
  "max_steps": 1248,
  "num_train_epochs": 8,
  "total_flos": 1.0651837376642417e+18,
  "trial_name": null,
  "trial_params": null
}