{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.4246575342465753,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.7063,
      "step": 10
    },
    {
      "epoch": 0.07,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.6727,
      "step": 20
    },
    {
      "epoch": 0.1,
      "learning_rate": 3e-06,
      "loss": 0.6957,
      "step": 30
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.6775,
      "step": 40
    },
    {
      "epoch": 0.17,
      "learning_rate": 5e-06,
      "loss": 0.6918,
      "step": 50
    },
    {
      "epoch": 0.21,
      "learning_rate": 6e-06,
      "loss": 0.6939,
      "step": 60
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.000000000000001e-06,
      "loss": 0.6684,
      "step": 70
    },
    {
      "epoch": 0.27,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.6312,
      "step": 80
    },
    {
      "epoch": 0.31,
      "learning_rate": 9e-06,
      "loss": 0.5444,
      "step": 90
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 0.435,
      "step": 100
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.3208,
      "step": 110
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.2e-05,
      "loss": 0.3709,
      "step": 120
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.4004,
      "step": 130
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 0.3061,
      "step": 140
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.5e-05,
      "loss": 0.3556,
      "step": 150
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.2945,
      "step": 160
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 0.3984,
      "step": 170
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8e-05,
      "loss": 0.3539,
      "step": 180
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.9e-05,
      "loss": 0.4119,
      "step": 190
    },
    {
      "epoch": 0.68,
      "learning_rate": 2e-05,
      "loss": 0.3457,
      "step": 200
    },
    {
      "epoch": 0.72,
      "learning_rate": 2.1e-05,
      "loss": 0.3241,
      "step": 210
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 0.4169,
      "step": 220
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 0.3863,
      "step": 230
    },
    {
      "epoch": 0.82,
      "learning_rate": 2.4e-05,
      "loss": 0.2749,
      "step": 240
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.5e-05,
      "loss": 0.3399,
      "step": 250
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 0.2771,
      "step": 260
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 0.3155,
      "step": 270
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 0.3677,
      "step": 280
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.9e-05,
      "loss": 0.2678,
      "step": 290
    },
    {
      "epoch": 1.03,
      "learning_rate": 3e-05,
      "loss": 0.2909,
      "step": 300
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.1e-05,
      "loss": 0.2997,
      "step": 310
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.2799,
      "step": 320
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.3e-05,
      "loss": 0.2793,
      "step": 330
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 0.2785,
      "step": 340
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.5e-05,
      "loss": 0.292,
      "step": 350
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.6e-05,
      "loss": 0.2058,
      "step": 360
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.7e-05,
      "loss": 0.2439,
      "step": 370
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.8e-05,
      "loss": 0.4566,
      "step": 380
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.9000000000000006e-05,
      "loss": 0.5624,
      "step": 390
    },
    {
      "epoch": 1.37,
      "learning_rate": 4e-05,
      "loss": 0.2315,
      "step": 400
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.1e-05,
      "loss": 0.184,
      "step": 410
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.2e-05,
      "loss": 0.3491,
      "step": 420
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.3e-05,
      "loss": 0.3401,
      "step": 430
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.2944,
      "step": 440
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.5e-05,
      "loss": 0.244,
      "step": 450
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.3155,
      "step": 460
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.7e-05,
      "loss": 0.2387,
      "step": 470
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.8e-05,
      "loss": 0.2712,
      "step": 480
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.9e-05,
      "loss": 0.3125,
      "step": 490
    },
    {
      "epoch": 1.71,
      "learning_rate": 5e-05,
      "loss": 0.2915,
      "step": 500
    },
    {
      "epoch": 1.75,
      "learning_rate": 4.947916666666667e-05,
      "loss": 0.3305,
      "step": 510
    },
    {
      "epoch": 1.78,
      "learning_rate": 4.8958333333333335e-05,
      "loss": 0.2439,
      "step": 520
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.8437500000000005e-05,
      "loss": 0.2771,
      "step": 530
    },
    {
      "epoch": 1.85,
      "learning_rate": 4.791666666666667e-05,
      "loss": 0.357,
      "step": 540
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.739583333333333e-05,
      "loss": 0.2287,
      "step": 550
    },
    {
      "epoch": 1.92,
      "learning_rate": 4.6875e-05,
      "loss": 0.3991,
      "step": 560
    },
    {
      "epoch": 1.95,
      "learning_rate": 4.635416666666667e-05,
      "loss": 0.3384,
      "step": 570
    },
    {
      "epoch": 1.99,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 0.2882,
      "step": 580
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.5312500000000004e-05,
      "loss": 0.2081,
      "step": 590
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.4791666666666673e-05,
      "loss": 0.2584,
      "step": 600
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.4270833333333337e-05,
      "loss": 0.153,
      "step": 610
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.375e-05,
      "loss": 0.3285,
      "step": 620
    },
    {
      "epoch": 2.16,
      "learning_rate": 4.322916666666667e-05,
      "loss": 0.215,
      "step": 630
    },
    {
      "epoch": 2.19,
      "learning_rate": 4.270833333333333e-05,
      "loss": 0.0573,
      "step": 640
    },
    {
      "epoch": 2.23,
      "learning_rate": 4.21875e-05,
      "loss": 0.2465,
      "step": 650
    },
    {
      "epoch": 2.26,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.243,
      "step": 660
    },
    {
      "epoch": 2.29,
      "learning_rate": 4.1145833333333335e-05,
      "loss": 0.1553,
      "step": 670
    },
    {
      "epoch": 2.33,
      "learning_rate": 4.0625000000000005e-05,
      "loss": 0.2165,
      "step": 680
    },
    {
      "epoch": 2.36,
      "learning_rate": 4.010416666666667e-05,
      "loss": 0.2816,
      "step": 690
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.958333333333333e-05,
      "loss": 0.1815,
      "step": 700
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.90625e-05,
      "loss": 0.1879,
      "step": 710
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.854166666666667e-05,
      "loss": 0.1997,
      "step": 720
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.8020833333333334e-05,
      "loss": 0.2046,
      "step": 730
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.1886,
      "step": 740
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.697916666666667e-05,
      "loss": 0.2487,
      "step": 750
    },
    {
      "epoch": 2.6,
      "learning_rate": 3.6458333333333336e-05,
      "loss": 0.212,
      "step": 760
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.59375e-05,
      "loss": 0.171,
      "step": 770
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.541666666666667e-05,
      "loss": 0.2213,
      "step": 780
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.489583333333333e-05,
      "loss": 0.3446,
      "step": 790
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.4375e-05,
      "loss": 0.2946,
      "step": 800
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.385416666666667e-05,
      "loss": 0.1453,
      "step": 810
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.1054,
      "step": 820
    },
    {
      "epoch": 2.84,
      "learning_rate": 3.2812500000000005e-05,
      "loss": 0.1581,
      "step": 830
    },
    {
      "epoch": 2.88,
      "learning_rate": 3.229166666666667e-05,
      "loss": 0.2906,
      "step": 840
    },
    {
      "epoch": 2.91,
      "learning_rate": 3.177083333333333e-05,
      "loss": 0.194,
      "step": 850
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.125e-05,
      "loss": 0.2613,
      "step": 860
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.072916666666667e-05,
      "loss": 0.2135,
      "step": 870
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.0208333333333334e-05,
      "loss": 0.2054,
      "step": 880
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.96875e-05,
      "loss": 0.1041,
      "step": 890
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.916666666666667e-05,
      "loss": 0.1748,
      "step": 900
    },
    {
      "epoch": 3.12,
      "learning_rate": 2.8645833333333333e-05,
      "loss": 0.1735,
      "step": 910
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.8125000000000003e-05,
      "loss": 0.0759,
      "step": 920
    },
    {
      "epoch": 3.18,
      "learning_rate": 2.760416666666667e-05,
      "loss": 0.2027,
      "step": 930
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.7083333333333332e-05,
      "loss": 0.0628,
      "step": 940
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.6562500000000002e-05,
      "loss": 0.1241,
      "step": 950
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.604166666666667e-05,
      "loss": 0.1004,
      "step": 960
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.552083333333333e-05,
      "loss": 0.0394,
      "step": 970
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.5e-05,
      "loss": 0.1247,
      "step": 980
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.4479166666666668e-05,
      "loss": 0.1056,
      "step": 990
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.3958333333333334e-05,
      "loss": 0.1697,
      "step": 1000
    }
  ],
  "max_steps": 1460,
  "num_train_epochs": 5,
  "total_flos": 2119080976300032.0,
  "trial_name": null,
  "trial_params": null
}