{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 50.0,
  "global_step": 2100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.48,
      "learning_rate": 1.980952380952381e-05,
      "loss": 1.4241,
      "step": 20
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.961904761904762e-05,
      "loss": 1.3539,
      "step": 40
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.942857142857143e-05,
      "loss": 0.8083,
      "step": 60
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.923809523809524e-05,
      "loss": 0.7274,
      "step": 80
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.904761904761905e-05,
      "loss": 0.3911,
      "step": 100
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.885714285714286e-05,
      "loss": 0.2785,
      "step": 120
    },
    {
      "epoch": 3.33,
      "learning_rate": 1.866666666666667e-05,
      "loss": 0.1367,
      "step": 140
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.8476190476190478e-05,
      "loss": 0.1004,
      "step": 160
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.8285714285714288e-05,
      "loss": 0.0718,
      "step": 180
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.8095238095238097e-05,
      "loss": 0.0588,
      "step": 200
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.7904761904761907e-05,
      "loss": 0.0456,
      "step": 220
    },
    {
      "epoch": 5.71,
      "learning_rate": 1.7714285714285717e-05,
      "loss": 0.0378,
      "step": 240
    },
    {
      "epoch": 6.19,
      "learning_rate": 1.7523809523809526e-05,
      "loss": 0.0297,
      "step": 260
    },
    {
      "epoch": 6.67,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 0.0232,
      "step": 280
    },
    {
      "epoch": 7.14,
      "learning_rate": 1.7142857142857142e-05,
      "loss": 0.0231,
      "step": 300
    },
    {
      "epoch": 7.62,
      "learning_rate": 1.6952380952380955e-05,
      "loss": 0.016,
      "step": 320
    },
    {
      "epoch": 8.1,
      "learning_rate": 1.6761904761904764e-05,
      "loss": 0.016,
      "step": 340
    },
    {
      "epoch": 8.57,
      "learning_rate": 1.6571428571428574e-05,
      "loss": 0.0109,
      "step": 360
    },
    {
      "epoch": 9.05,
      "learning_rate": 1.6380952380952384e-05,
      "loss": 0.0114,
      "step": 380
    },
    {
      "epoch": 9.52,
      "learning_rate": 1.6190476190476193e-05,
      "loss": 0.0078,
      "step": 400
    },
    {
      "epoch": 10.0,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.0103,
      "step": 420
    },
    {
      "epoch": 10.48,
      "learning_rate": 1.580952380952381e-05,
      "loss": 0.0061,
      "step": 440
    },
    {
      "epoch": 10.95,
      "learning_rate": 1.5619047619047622e-05,
      "loss": 0.0085,
      "step": 460
    },
    {
      "epoch": 11.43,
      "learning_rate": 1.542857142857143e-05,
      "loss": 0.005,
      "step": 480
    },
    {
      "epoch": 11.9,
      "learning_rate": 1.523809523809524e-05,
      "loss": 0.0093,
      "step": 500
    },
    {
      "epoch": 12.38,
      "learning_rate": 1.5047619047619049e-05,
      "loss": 0.0066,
      "step": 520
    },
    {
      "epoch": 12.86,
      "learning_rate": 1.4857142857142858e-05,
      "loss": 0.0061,
      "step": 540
    },
    {
      "epoch": 13.33,
      "learning_rate": 1.4666666666666666e-05,
      "loss": 0.0048,
      "step": 560
    },
    {
      "epoch": 13.81,
      "learning_rate": 1.4476190476190478e-05,
      "loss": 0.0051,
      "step": 580
    },
    {
      "epoch": 14.29,
      "learning_rate": 1.4285714285714287e-05,
      "loss": 0.0036,
      "step": 600
    },
    {
      "epoch": 14.76,
      "learning_rate": 1.4095238095238097e-05,
      "loss": 0.0053,
      "step": 620
    },
    {
      "epoch": 15.24,
      "learning_rate": 1.3904761904761905e-05,
      "loss": 0.0052,
      "step": 640
    },
    {
      "epoch": 15.71,
      "learning_rate": 1.3714285714285716e-05,
      "loss": 0.0054,
      "step": 660
    },
    {
      "epoch": 16.19,
      "learning_rate": 1.3523809523809525e-05,
      "loss": 0.0049,
      "step": 680
    },
    {
      "epoch": 16.67,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.0055,
      "step": 700
    },
    {
      "epoch": 17.14,
      "learning_rate": 1.3142857142857145e-05,
      "loss": 0.0046,
      "step": 720
    },
    {
      "epoch": 17.62,
      "learning_rate": 1.2952380952380954e-05,
      "loss": 0.004,
      "step": 740
    },
    {
      "epoch": 18.1,
      "learning_rate": 1.2761904761904762e-05,
      "loss": 0.0047,
      "step": 760
    },
    {
      "epoch": 18.57,
      "learning_rate": 1.2571428571428572e-05,
      "loss": 0.0047,
      "step": 780
    },
    {
      "epoch": 19.05,
      "learning_rate": 1.2380952380952383e-05,
      "loss": 0.0047,
      "step": 800
    },
    {
      "epoch": 19.52,
      "learning_rate": 1.2190476190476192e-05,
      "loss": 0.0038,
      "step": 820
    },
    {
      "epoch": 20.0,
      "learning_rate": 1.2e-05,
      "loss": 0.0046,
      "step": 840
    },
    {
      "epoch": 20.48,
      "learning_rate": 1.180952380952381e-05,
      "loss": 0.0034,
      "step": 860
    },
    {
      "epoch": 20.95,
      "learning_rate": 1.1619047619047621e-05,
      "loss": 0.004,
      "step": 880
    },
    {
      "epoch": 21.43,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 0.0032,
      "step": 900
    },
    {
      "epoch": 21.9,
      "learning_rate": 1.1238095238095239e-05,
      "loss": 0.0039,
      "step": 920
    },
    {
      "epoch": 22.38,
      "learning_rate": 1.104761904761905e-05,
      "loss": 0.0033,
      "step": 940
    },
    {
      "epoch": 22.86,
      "learning_rate": 1.0857142857142858e-05,
      "loss": 0.0032,
      "step": 960
    },
    {
      "epoch": 23.33,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 0.0033,
      "step": 980
    },
    {
      "epoch": 23.81,
      "learning_rate": 1.0476190476190477e-05,
      "loss": 0.003,
      "step": 1000
    },
    {
      "epoch": 24.29,
      "learning_rate": 1.0285714285714285e-05,
      "loss": 0.003,
      "step": 1020
    },
    {
      "epoch": 24.76,
      "learning_rate": 1.0095238095238096e-05,
      "loss": 0.003,
      "step": 1040
    },
    {
      "epoch": 25.24,
      "learning_rate": 9.904761904761906e-06,
      "loss": 0.0034,
      "step": 1060
    },
    {
      "epoch": 25.71,
      "learning_rate": 9.714285714285715e-06,
      "loss": 0.0029,
      "step": 1080
    },
    {
      "epoch": 26.19,
      "learning_rate": 9.523809523809525e-06,
      "loss": 0.0031,
      "step": 1100
    },
    {
      "epoch": 26.67,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.0032,
      "step": 1120
    },
    {
      "epoch": 27.14,
      "learning_rate": 9.142857142857144e-06,
      "loss": 0.0032,
      "step": 1140
    },
    {
      "epoch": 27.62,
      "learning_rate": 8.952380952380953e-06,
      "loss": 0.003,
      "step": 1160
    },
    {
      "epoch": 28.1,
      "learning_rate": 8.761904761904763e-06,
      "loss": 0.0031,
      "step": 1180
    },
    {
      "epoch": 28.57,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.0026,
      "step": 1200
    },
    {
      "epoch": 29.05,
      "learning_rate": 8.380952380952382e-06,
      "loss": 0.0032,
      "step": 1220
    },
    {
      "epoch": 29.52,
      "learning_rate": 8.190476190476192e-06,
      "loss": 0.0026,
      "step": 1240
    },
    {
      "epoch": 30.0,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.004,
      "step": 1260
    },
    {
      "epoch": 30.48,
      "learning_rate": 7.809523809523811e-06,
      "loss": 0.0031,
      "step": 1280
    },
    {
      "epoch": 30.95,
      "learning_rate": 7.61904761904762e-06,
      "loss": 0.0034,
      "step": 1300
    },
    {
      "epoch": 31.43,
      "learning_rate": 7.428571428571429e-06,
      "loss": 0.0032,
      "step": 1320
    },
    {
      "epoch": 31.9,
      "learning_rate": 7.238095238095239e-06,
      "loss": 0.0031,
      "step": 1340
    },
    {
      "epoch": 32.38,
      "learning_rate": 7.047619047619048e-06,
      "loss": 0.0029,
      "step": 1360
    },
    {
      "epoch": 32.86,
      "learning_rate": 6.857142857142858e-06,
      "loss": 0.0031,
      "step": 1380
    },
    {
      "epoch": 33.33,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.0033,
      "step": 1400
    },
    {
      "epoch": 33.81,
      "learning_rate": 6.476190476190477e-06,
      "loss": 0.0031,
      "step": 1420
    },
    {
      "epoch": 34.29,
      "learning_rate": 6.285714285714286e-06,
      "loss": 0.0027,
      "step": 1440
    },
    {
      "epoch": 34.76,
      "learning_rate": 6.095238095238096e-06,
      "loss": 0.0028,
      "step": 1460
    },
    {
      "epoch": 35.24,
      "learning_rate": 5.904761904761905e-06,
      "loss": 0.0027,
      "step": 1480
    },
    {
      "epoch": 35.71,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.0029,
      "step": 1500
    },
    {
      "epoch": 36.19,
      "learning_rate": 5.523809523809525e-06,
      "loss": 0.0026,
      "step": 1520
    },
    {
      "epoch": 36.67,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.0024,
      "step": 1540
    },
    {
      "epoch": 37.14,
      "learning_rate": 5.142857142857142e-06,
      "loss": 0.0029,
      "step": 1560
    },
    {
      "epoch": 37.62,
      "learning_rate": 4.952380952380953e-06,
      "loss": 0.0023,
      "step": 1580
    },
    {
      "epoch": 38.1,
      "learning_rate": 4.761904761904762e-06,
      "loss": 0.0029,
      "step": 1600
    },
    {
      "epoch": 38.57,
      "learning_rate": 4.571428571428572e-06,
      "loss": 0.0023,
      "step": 1620
    },
    {
      "epoch": 39.05,
      "learning_rate": 4.3809523809523815e-06,
      "loss": 0.0027,
      "step": 1640
    },
    {
      "epoch": 39.52,
      "learning_rate": 4.190476190476191e-06,
      "loss": 0.0022,
      "step": 1660
    },
    {
      "epoch": 40.0,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0028,
      "step": 1680
    },
    {
      "epoch": 40.48,
      "learning_rate": 3.80952380952381e-06,
      "loss": 0.0024,
      "step": 1700
    },
    {
      "epoch": 40.95,
      "learning_rate": 3.6190476190476194e-06,
      "loss": 0.0025,
      "step": 1720
    },
    {
      "epoch": 41.43,
      "learning_rate": 3.428571428571429e-06,
      "loss": 0.0022,
      "step": 1740
    },
    {
      "epoch": 41.9,
      "learning_rate": 3.2380952380952385e-06,
      "loss": 0.0026,
      "step": 1760
    },
    {
      "epoch": 42.38,
      "learning_rate": 3.047619047619048e-06,
      "loss": 0.0021,
      "step": 1780
    },
    {
      "epoch": 42.86,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.0027,
      "step": 1800
    },
    {
      "epoch": 43.33,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.0023,
      "step": 1820
    },
    {
      "epoch": 43.81,
      "learning_rate": 2.4761904761904764e-06,
      "loss": 0.0025,
      "step": 1840
    },
    {
      "epoch": 44.29,
      "learning_rate": 2.285714285714286e-06,
      "loss": 0.0022,
      "step": 1860
    },
    {
      "epoch": 44.76,
      "learning_rate": 2.0952380952380955e-06,
      "loss": 0.0024,
      "step": 1880
    },
    {
      "epoch": 45.24,
      "learning_rate": 1.904761904761905e-06,
      "loss": 0.0023,
      "step": 1900
    },
    {
      "epoch": 45.71,
      "learning_rate": 1.7142857142857145e-06,
      "loss": 0.0023,
      "step": 1920
    },
    {
      "epoch": 46.19,
      "learning_rate": 1.523809523809524e-06,
      "loss": 0.0022,
      "step": 1940
    },
    {
      "epoch": 46.67,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.0024,
      "step": 1960
    },
    {
      "epoch": 47.14,
      "learning_rate": 1.142857142857143e-06,
      "loss": 0.002,
      "step": 1980
    },
    {
      "epoch": 47.62,
      "learning_rate": 9.523809523809525e-07,
      "loss": 0.0021,
      "step": 2000
    },
    {
      "epoch": 48.1,
      "learning_rate": 7.61904761904762e-07,
      "loss": 0.0023,
      "step": 2020
    },
    {
      "epoch": 48.57,
      "learning_rate": 5.714285714285715e-07,
      "loss": 0.0022,
      "step": 2040
    },
    {
      "epoch": 49.05,
      "learning_rate": 3.80952380952381e-07,
      "loss": 0.0021,
      "step": 2060
    },
    {
      "epoch": 49.52,
      "learning_rate": 1.904761904761905e-07,
      "loss": 0.0021,
      "step": 2080
    },
    {
      "epoch": 50.0,
      "learning_rate": 0.0,
      "loss": 0.0021,
      "step": 2100
    },
    {
      "epoch": 50.0,
      "step": 2100,
      "total_flos": 13733996789760.0,
      "train_loss": 0.0559101588953109,
      "train_runtime": 46787.7757,
      "train_samples_per_second": 0.357,
      "train_steps_per_second": 0.045
    }
  ],
  "max_steps": 2100,
  "num_train_epochs": 50,
  "total_flos": 13733996789760.0,
  "trial_name": null,
  "trial_params": null
}