{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.839587932303164,
  "eval_steps": 500,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.4e-07,
      "loss": 0.8278,
      "step": 25
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.400000000000001e-07,
      "loss": 0.7595,
      "step": 50
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.44e-06,
      "loss": 0.6724,
      "step": 75
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.94e-06,
      "loss": 0.6072,
      "step": 100
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.4400000000000004e-06,
      "loss": 0.5705,
      "step": 125
    },
    {
      "epoch": 0.06,
      "learning_rate": 2.9400000000000002e-06,
      "loss": 0.5734,
      "step": 150
    },
    {
      "epoch": 0.06,
      "learning_rate": 3.44e-06,
      "loss": 0.5258,
      "step": 175
    },
    {
      "epoch": 0.07,
      "learning_rate": 3.94e-06,
      "loss": 0.5347,
      "step": 200
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.440000000000001e-06,
      "loss": 0.4947,
      "step": 225
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.94e-06,
      "loss": 0.5224,
      "step": 250
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.4400000000000004e-06,
      "loss": 0.4735,
      "step": 275
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.94e-06,
      "loss": 0.5,
      "step": 300
    },
    {
      "epoch": 0.12,
      "learning_rate": 6.440000000000001e-06,
      "loss": 0.461,
      "step": 325
    },
    {
      "epoch": 0.13,
      "learning_rate": 6.9400000000000005e-06,
      "loss": 0.4331,
      "step": 350
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.440000000000001e-06,
      "loss": 0.4501,
      "step": 375
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.94e-06,
      "loss": 0.4465,
      "step": 400
    },
    {
      "epoch": 0.16,
      "learning_rate": 8.44e-06,
      "loss": 0.411,
      "step": 425
    },
    {
      "epoch": 0.17,
      "learning_rate": 8.94e-06,
      "loss": 0.4331,
      "step": 450
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.440000000000001e-06,
      "loss": 0.4032,
      "step": 475
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.940000000000001e-06,
      "loss": 0.4036,
      "step": 500
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.951111111111111e-06,
      "loss": 0.4213,
      "step": 525
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.895555555555557e-06,
      "loss": 0.3808,
      "step": 550
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.84e-06,
      "loss": 0.3869,
      "step": 575
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.784444444444445e-06,
      "loss": 0.388,
      "step": 600
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.72888888888889e-06,
      "loss": 0.3969,
      "step": 625
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.673333333333334e-06,
      "loss": 0.3631,
      "step": 650
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.617777777777778e-06,
      "loss": 0.3637,
      "step": 675
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.562222222222223e-06,
      "loss": 0.3426,
      "step": 700
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.506666666666667e-06,
      "loss": 0.3473,
      "step": 725
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.451111111111112e-06,
      "loss": 0.3602,
      "step": 750
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.395555555555556e-06,
      "loss": 0.3523,
      "step": 775
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.340000000000002e-06,
      "loss": 0.3331,
      "step": 800
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.284444444444444e-06,
      "loss": 0.3487,
      "step": 825
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.22888888888889e-06,
      "loss": 0.3493,
      "step": 850
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.173333333333334e-06,
      "loss": 0.3236,
      "step": 875
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.117777777777778e-06,
      "loss": 0.3117,
      "step": 900
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.062222222222224e-06,
      "loss": 0.3302,
      "step": 925
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.006666666666666e-06,
      "loss": 0.3204,
      "step": 950
    },
    {
      "epoch": 0.36,
      "learning_rate": 8.951111111111112e-06,
      "loss": 0.3328,
      "step": 975
    },
    {
      "epoch": 0.37,
      "learning_rate": 8.895555555555556e-06,
      "loss": 0.2967,
      "step": 1000
    },
    {
      "epoch": 0.38,
      "learning_rate": 8.84e-06,
      "loss": 0.325,
      "step": 1025
    },
    {
      "epoch": 0.39,
      "learning_rate": 8.784444444444446e-06,
      "loss": 0.3207,
      "step": 1050
    },
    {
      "epoch": 0.4,
      "learning_rate": 8.72888888888889e-06,
      "loss": 0.3248,
      "step": 1075
    },
    {
      "epoch": 0.4,
      "learning_rate": 8.673333333333334e-06,
      "loss": 0.2953,
      "step": 1100
    },
    {
      "epoch": 0.41,
      "learning_rate": 8.617777777777778e-06,
      "loss": 0.3112,
      "step": 1125
    },
    {
      "epoch": 0.42,
      "learning_rate": 8.562222222222224e-06,
      "loss": 0.2969,
      "step": 1150
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.506666666666668e-06,
      "loss": 0.3064,
      "step": 1175
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.451111111111112e-06,
      "loss": 0.3169,
      "step": 1200
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.395555555555557e-06,
      "loss": 0.3039,
      "step": 1225
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.34e-06,
      "loss": 0.2728,
      "step": 1250
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.284444444444446e-06,
      "loss": 0.3109,
      "step": 1275
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.22888888888889e-06,
      "loss": 0.2991,
      "step": 1300
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.173333333333334e-06,
      "loss": 0.2908,
      "step": 1325
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.11777777777778e-06,
      "loss": 0.2856,
      "step": 1350
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.062222222222222e-06,
      "loss": 0.281,
      "step": 1375
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.006666666666667e-06,
      "loss": 0.2924,
      "step": 1400
    },
    {
      "epoch": 0.52,
      "learning_rate": 7.951111111111111e-06,
      "loss": 0.2729,
      "step": 1425
    },
    {
      "epoch": 0.53,
      "learning_rate": 7.895555555555557e-06,
      "loss": 0.2803,
      "step": 1450
    },
    {
      "epoch": 0.54,
      "learning_rate": 7.840000000000001e-06,
      "loss": 0.3032,
      "step": 1475
    },
    {
      "epoch": 0.55,
      "learning_rate": 7.784444444444445e-06,
      "loss": 0.271,
      "step": 1500
    },
    {
      "epoch": 0.56,
      "learning_rate": 7.72888888888889e-06,
      "loss": 0.295,
      "step": 1525
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.673333333333333e-06,
      "loss": 0.2806,
      "step": 1550
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.617777777777778e-06,
      "loss": 0.2674,
      "step": 1575
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.562222222222223e-06,
      "loss": 0.2688,
      "step": 1600
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.506666666666668e-06,
      "loss": 0.2791,
      "step": 1625
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.451111111111111e-06,
      "loss": 0.2825,
      "step": 1650
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.395555555555556e-06,
      "loss": 0.2814,
      "step": 1675
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.340000000000001e-06,
      "loss": 0.279,
      "step": 1700
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.284444444444445e-06,
      "loss": 0.2626,
      "step": 1725
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.22888888888889e-06,
      "loss": 0.2858,
      "step": 1750
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.173333333333335e-06,
      "loss": 0.2508,
      "step": 1775
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.117777777777778e-06,
      "loss": 0.2866,
      "step": 1800
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.062222222222223e-06,
      "loss": 0.2485,
      "step": 1825
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.006666666666667e-06,
      "loss": 0.2533,
      "step": 1850
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.951111111111112e-06,
      "loss": 0.2731,
      "step": 1875
    },
    {
      "epoch": 0.7,
      "learning_rate": 6.8955555555555565e-06,
      "loss": 0.2509,
      "step": 1900
    },
    {
      "epoch": 0.71,
      "learning_rate": 6.8400000000000014e-06,
      "loss": 0.2411,
      "step": 1925
    },
    {
      "epoch": 0.72,
      "learning_rate": 6.784444444444445e-06,
      "loss": 0.2356,
      "step": 1950
    },
    {
      "epoch": 0.73,
      "learning_rate": 6.7288888888888895e-06,
      "loss": 0.247,
      "step": 1975
    },
    {
      "epoch": 0.74,
      "learning_rate": 6.6733333333333335e-06,
      "loss": 0.2614,
      "step": 2000
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.617777777777778e-06,
      "loss": 0.2502,
      "step": 2025
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.562222222222223e-06,
      "loss": 0.2521,
      "step": 2050
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.5066666666666665e-06,
      "loss": 0.2488,
      "step": 2075
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.451111111111111e-06,
      "loss": 0.2361,
      "step": 2100
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.395555555555556e-06,
      "loss": 0.2657,
      "step": 2125
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.34e-06,
      "loss": 0.2635,
      "step": 2150
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.284444444444445e-06,
      "loss": 0.2408,
      "step": 2175
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.22888888888889e-06,
      "loss": 0.2486,
      "step": 2200
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.173333333333333e-06,
      "loss": 0.2496,
      "step": 2225
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.117777777777778e-06,
      "loss": 0.2363,
      "step": 2250
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.062222222222223e-06,
      "loss": 0.2616,
      "step": 2275
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.006666666666667e-06,
      "loss": 0.2622,
      "step": 2300
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.951111111111112e-06,
      "loss": 0.2273,
      "step": 2325
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.895555555555557e-06,
      "loss": 0.2339,
      "step": 2350
    },
    {
      "epoch": 0.87,
      "learning_rate": 5.84e-06,
      "loss": 0.2378,
      "step": 2375
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.784444444444445e-06,
      "loss": 0.231,
      "step": 2400
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.72888888888889e-06,
      "loss": 0.2367,
      "step": 2425
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.673333333333334e-06,
      "loss": 0.2204,
      "step": 2450
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.617777777777779e-06,
      "loss": 0.2062,
      "step": 2475
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.562222222222222e-06,
      "loss": 0.2331,
      "step": 2500
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.506666666666667e-06,
      "loss": 0.2313,
      "step": 2525
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.451111111111112e-06,
      "loss": 0.2343,
      "step": 2550
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.3955555555555565e-06,
      "loss": 0.2215,
      "step": 2575
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.3400000000000005e-06,
      "loss": 0.2259,
      "step": 2600
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.2844444444444454e-06,
      "loss": 0.2524,
      "step": 2625
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.228888888888889e-06,
      "loss": 0.2182,
      "step": 2650
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.1733333333333335e-06,
      "loss": 0.2118,
      "step": 2675
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.117777777777778e-06,
      "loss": 0.2363,
      "step": 2700
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.062222222222222e-06,
      "loss": 0.2106,
      "step": 2725
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.006666666666667e-06,
      "loss": 0.1688,
      "step": 2750
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.951111111111111e-06,
      "loss": 0.174,
      "step": 2775
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.895555555555556e-06,
      "loss": 0.1585,
      "step": 2800
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.84e-06,
      "loss": 0.1751,
      "step": 2825
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.784444444444445e-06,
      "loss": 0.1575,
      "step": 2850
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.728888888888889e-06,
      "loss": 0.159,
      "step": 2875
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.673333333333333e-06,
      "loss": 0.151,
      "step": 2900
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.617777777777778e-06,
      "loss": 0.1836,
      "step": 2925
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.562222222222222e-06,
      "loss": 0.1625,
      "step": 2950
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.506666666666667e-06,
      "loss": 0.1672,
      "step": 2975
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.451111111111112e-06,
      "loss": 0.1918,
      "step": 3000
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.395555555555556e-06,
      "loss": 0.1535,
      "step": 3025
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.34e-06,
      "loss": 0.1558,
      "step": 3050
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.284444444444445e-06,
      "loss": 0.1703,
      "step": 3075
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.228888888888889e-06,
      "loss": 0.148,
      "step": 3100
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.173333333333334e-06,
      "loss": 0.1555,
      "step": 3125
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.117777777777779e-06,
      "loss": 0.1549,
      "step": 3150
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.062222222222223e-06,
      "loss": 0.1532,
      "step": 3175
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.006666666666667e-06,
      "loss": 0.1605,
      "step": 3200
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.951111111111112e-06,
      "loss": 0.1639,
      "step": 3225
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.895555555555556e-06,
      "loss": 0.1592,
      "step": 3250
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.8400000000000005e-06,
      "loss": 0.1484,
      "step": 3275
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.784444444444445e-06,
      "loss": 0.1592,
      "step": 3300
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.728888888888889e-06,
      "loss": 0.1706,
      "step": 3325
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.673333333333334e-06,
      "loss": 0.1754,
      "step": 3350
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.617777777777778e-06,
      "loss": 0.157,
      "step": 3375
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.5622222222222224e-06,
      "loss": 0.1624,
      "step": 3400
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.5066666666666673e-06,
      "loss": 0.1689,
      "step": 3425
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.4511111111111113e-06,
      "loss": 0.164,
      "step": 3450
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.3955555555555558e-06,
      "loss": 0.1553,
      "step": 3475
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.3400000000000006e-06,
      "loss": 0.1613,
      "step": 3500
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.2844444444444447e-06,
      "loss": 0.1479,
      "step": 3525
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.228888888888889e-06,
      "loss": 0.1619,
      "step": 3550
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.173333333333334e-06,
      "loss": 0.1642,
      "step": 3575
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.117777777777778e-06,
      "loss": 0.1411,
      "step": 3600
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.0622222222222225e-06,
      "loss": 0.1589,
      "step": 3625
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.0066666666666674e-06,
      "loss": 0.1657,
      "step": 3650
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.9511111111111114e-06,
      "loss": 0.1611,
      "step": 3675
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.895555555555556e-06,
      "loss": 0.1416,
      "step": 3700
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.84e-06,
      "loss": 0.1417,
      "step": 3725
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.784444444444445e-06,
      "loss": 0.1794,
      "step": 3750
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.7288888888888893e-06,
      "loss": 0.1545,
      "step": 3775
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6733333333333333e-06,
      "loss": 0.1603,
      "step": 3800
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.617777777777778e-06,
      "loss": 0.1513,
      "step": 3825
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.5622222222222226e-06,
      "loss": 0.1419,
      "step": 3850
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.5066666666666667e-06,
      "loss": 0.1448,
      "step": 3875
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.451111111111111e-06,
      "loss": 0.147,
      "step": 3900
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.3955555555555556e-06,
      "loss": 0.1653,
      "step": 3925
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.3400000000000005e-06,
      "loss": 0.1577,
      "step": 3950
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.2844444444444445e-06,
      "loss": 0.1285,
      "step": 3975
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.228888888888889e-06,
      "loss": 0.1602,
      "step": 4000
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.1733333333333334e-06,
      "loss": 0.1785,
      "step": 4025
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.117777777777778e-06,
      "loss": 0.1553,
      "step": 4050
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.0622222222222223e-06,
      "loss": 0.1546,
      "step": 4075
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.006666666666667e-06,
      "loss": 0.1477,
      "step": 4100
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.9511111111111113e-06,
      "loss": 0.1363,
      "step": 4125
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.8955555555555557e-06,
      "loss": 0.1619,
      "step": 4150
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.8400000000000002e-06,
      "loss": 0.1545,
      "step": 4175
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.7844444444444444e-06,
      "loss": 0.1451,
      "step": 4200
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.728888888888889e-06,
      "loss": 0.1493,
      "step": 4225
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.6733333333333335e-06,
      "loss": 0.1455,
      "step": 4250
    },
    {
      "epoch": 1.57,
      "learning_rate": 1.6177777777777778e-06,
      "loss": 0.1409,
      "step": 4275
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.5622222222222225e-06,
      "loss": 0.1587,
      "step": 4300
    },
    {
      "epoch": 1.59,
      "learning_rate": 1.506666666666667e-06,
      "loss": 0.1494,
      "step": 4325
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.4511111111111112e-06,
      "loss": 0.1539,
      "step": 4350
    },
    {
      "epoch": 1.61,
      "learning_rate": 1.3955555555555556e-06,
      "loss": 0.1512,
      "step": 4375
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.34e-06,
      "loss": 0.15,
      "step": 4400
    },
    {
      "epoch": 1.63,
      "learning_rate": 1.2844444444444445e-06,
      "loss": 0.1511,
      "step": 4425
    },
    {
      "epoch": 1.64,
      "learning_rate": 1.228888888888889e-06,
      "loss": 0.1517,
      "step": 4450
    },
    {
      "epoch": 1.65,
      "learning_rate": 1.1733333333333335e-06,
      "loss": 0.1485,
      "step": 4475
    },
    {
      "epoch": 1.66,
      "learning_rate": 1.117777777777778e-06,
      "loss": 0.1382,
      "step": 4500
    },
    {
      "epoch": 1.66,
      "learning_rate": 1.0622222222222222e-06,
      "loss": 0.1427,
      "step": 4525
    },
    {
      "epoch": 1.67,
      "learning_rate": 1.0066666666666668e-06,
      "loss": 0.1448,
      "step": 4550
    },
    {
      "epoch": 1.68,
      "learning_rate": 9.511111111111111e-07,
      "loss": 0.1417,
      "step": 4575
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.955555555555557e-07,
      "loss": 0.1364,
      "step": 4600
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.400000000000001e-07,
      "loss": 0.1421,
      "step": 4625
    },
    {
      "epoch": 1.71,
      "learning_rate": 7.844444444444445e-07,
      "loss": 0.1488,
      "step": 4650
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.28888888888889e-07,
      "loss": 0.1425,
      "step": 4675
    },
    {
      "epoch": 1.73,
      "learning_rate": 6.733333333333334e-07,
      "loss": 0.1487,
      "step": 4700
    },
    {
      "epoch": 1.74,
      "learning_rate": 6.177777777777778e-07,
      "loss": 0.1529,
      "step": 4725
    },
    {
      "epoch": 1.75,
      "learning_rate": 5.622222222222223e-07,
      "loss": 0.1401,
      "step": 4750
    },
    {
      "epoch": 1.76,
      "learning_rate": 5.066666666666667e-07,
      "loss": 0.1489,
      "step": 4775
    },
    {
      "epoch": 1.77,
      "learning_rate": 4.511111111111111e-07,
      "loss": 0.1457,
      "step": 4800
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.9555555555555557e-07,
      "loss": 0.15,
      "step": 4825
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.4000000000000003e-07,
      "loss": 0.1538,
      "step": 4850
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.844444444444445e-07,
      "loss": 0.149,
      "step": 4875
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.2888888888888892e-07,
      "loss": 0.1604,
      "step": 4900
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.7333333333333335e-07,
      "loss": 0.1402,
      "step": 4925
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.1777777777777778e-07,
      "loss": 0.1385,
      "step": 4950
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.222222222222223e-08,
      "loss": 0.1396,
      "step": 4975
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.666666666666667e-09,
      "loss": 0.1486,
      "step": 5000
    },
    {
      "epoch": 1.84,
      "step": 5000,
      "total_flos": 2.308423475699712e+19,
      "train_loss": 0.24893692359924316,
      "train_runtime": 5414.4985,
      "train_samples_per_second": 14.775,
      "train_steps_per_second": 0.923
    }
  ],
  "logging_steps": 25,
  "max_steps": 5000,
  "num_train_epochs": 2,
  "save_steps": 2000,
  "total_flos": 2.308423475699712e+19,
  "trial_name": null,
  "trial_params": null
}