{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0993816733561657,
  "global_step": 50000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4e-06,
      "loss": 10.285,
      "step": 100
    },
    {
      "epoch": 0.01,
      "learning_rate": 8e-06,
      "loss": 8.5629,
      "step": 200
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.2e-05,
      "loss": 7.7941,
      "step": 300
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.6e-05,
      "loss": 7.2214,
      "step": 400
    },
    {
      "epoch": 0.03,
      "learning_rate": 2e-05,
      "loss": 6.7445,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.4e-05,
      "loss": 6.5017,
      "step": 600
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.8e-05,
      "loss": 6.3973,
      "step": 700
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.2e-05,
      "loss": 6.3519,
      "step": 800
    },
    {
      "epoch": 0.06,
      "learning_rate": 3.6e-05,
      "loss": 6.341,
      "step": 900
    },
    {
      "epoch": 0.06,
      "learning_rate": 4e-05,
      "loss": 6.3197,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.4e-05,
      "loss": 6.3186,
      "step": 1100
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.8e-05,
      "loss": 6.3114,
      "step": 1200
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.2e-05,
      "loss": 6.3023,
      "step": 1300
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.6e-05,
      "loss": 6.2749,
      "step": 1400
    },
    {
      "epoch": 0.09,
      "learning_rate": 6e-05,
      "loss": 6.2705,
      "step": 1500
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.4e-05,
      "loss": 6.2538,
      "step": 1600
    },
    {
      "epoch": 0.11,
      "learning_rate": 6.800000000000001e-05,
      "loss": 6.2464,
      "step": 1700
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.2e-05,
      "loss": 6.2073,
      "step": 1800
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.6e-05,
      "loss": 6.101,
      "step": 1900
    },
    {
      "epoch": 0.12,
      "learning_rate": 8e-05,
      "loss": 5.9768,
      "step": 2000
    },
    {
      "epoch": 0.13,
      "learning_rate": 8.400000000000001e-05,
      "loss": 5.7986,
      "step": 2100
    },
    {
      "epoch": 0.14,
      "learning_rate": 8.8e-05,
      "loss": 5.6088,
      "step": 2200
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.2e-05,
      "loss": 5.4903,
      "step": 2300
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.6e-05,
      "loss": 5.3685,
      "step": 2400
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001,
      "loss": 5.282,
      "step": 2500
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.000104,
      "loss": 5.1831,
      "step": 2600
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.000108,
      "loss": 5.1007,
      "step": 2700
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.000112,
      "loss": 5.0023,
      "step": 2800
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00011600000000000001,
      "loss": 4.9239,
      "step": 2900
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00012,
      "loss": 4.8625,
      "step": 3000
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.000124,
      "loss": 4.7771,
      "step": 3100
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.000128,
      "loss": 4.7051,
      "step": 3200
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.000132,
      "loss": 4.6168,
      "step": 3300
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00013600000000000003,
      "loss": 4.5591,
      "step": 3400
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00014000000000000001,
      "loss": 4.5026,
      "step": 3500
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.000144,
      "loss": 4.4251,
      "step": 3600
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.000148,
      "loss": 4.3889,
      "step": 3700
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.000152,
      "loss": 4.3326,
      "step": 3800
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.000156,
      "loss": 4.288,
      "step": 3900
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00016,
      "loss": 4.2549,
      "step": 4000
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.000164,
      "loss": 4.19,
      "step": 4100
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016800000000000002,
      "loss": 4.1428,
      "step": 4200
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00017199999999999998,
      "loss": 4.1085,
      "step": 4300
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.000176,
      "loss": 4.078,
      "step": 4400
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00017999999999999998,
      "loss": 4.0206,
      "step": 4500
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.000184,
      "loss": 4.0012,
      "step": 4600
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00018800000000000002,
      "loss": 3.98,
      "step": 4700
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.000192,
      "loss": 3.9526,
      "step": 4800
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019600000000000002,
      "loss": 3.8986,
      "step": 4900
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 3.8794,
      "step": 5000
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.000204,
      "loss": 3.84,
      "step": 5100
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.000208,
      "loss": 3.8412,
      "step": 5200
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.000212,
      "loss": 3.8021,
      "step": 5300
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.000216,
      "loss": 3.7633,
      "step": 5400
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00022,
      "loss": 3.7475,
      "step": 5500
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.000224,
      "loss": 3.7366,
      "step": 5600
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.000228,
      "loss": 3.7149,
      "step": 5700
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00023200000000000003,
      "loss": 3.6964,
      "step": 5800
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.000236,
      "loss": 3.6868,
      "step": 5900
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00024,
      "loss": 3.6684,
      "step": 6000
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.000244,
      "loss": 3.6452,
      "step": 6100
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.000248,
      "loss": 3.6353,
      "step": 6200
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.000252,
      "loss": 3.6115,
      "step": 6300
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.000256,
      "loss": 3.6012,
      "step": 6400
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00026000000000000003,
      "loss": 3.5799,
      "step": 6500
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.000264,
      "loss": 3.5517,
      "step": 6600
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.000268,
      "loss": 3.5509,
      "step": 6700
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00027200000000000005,
      "loss": 3.539,
      "step": 6800
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00027600000000000004,
      "loss": 3.5227,
      "step": 6900
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00028000000000000003,
      "loss": 3.5066,
      "step": 7000
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00028399999999999996,
      "loss": 3.5071,
      "step": 7100
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.000288,
      "loss": 3.5012,
      "step": 7200
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.000292,
      "loss": 3.4881,
      "step": 7300
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.000296,
      "loss": 3.483,
      "step": 7400
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0003,
      "loss": 3.4781,
      "step": 7500
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.000304,
      "loss": 3.4433,
      "step": 7600
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.000308,
      "loss": 3.4457,
      "step": 7700
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.000312,
      "loss": 3.4491,
      "step": 7800
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.000316,
      "loss": 3.4211,
      "step": 7900
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00032,
      "loss": 3.416,
      "step": 8000
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.000324,
      "loss": 3.4149,
      "step": 8100
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.000328,
      "loss": 3.3943,
      "step": 8200
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00033200000000000005,
      "loss": 3.3854,
      "step": 8300
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00033600000000000004,
      "loss": 3.389,
      "step": 8400
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00034,
      "loss": 3.3672,
      "step": 8500
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00034399999999999996,
      "loss": 3.3656,
      "step": 8600
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.000348,
      "loss": 3.3347,
      "step": 8700
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.000352,
      "loss": 3.3448,
      "step": 8800
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.000356,
      "loss": 3.3228,
      "step": 8900
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00035999999999999997,
      "loss": 3.3337,
      "step": 9000
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.000364,
      "loss": 3.3183,
      "step": 9100
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.000368,
      "loss": 3.33,
      "step": 9200
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.000372,
      "loss": 3.3125,
      "step": 9300
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00037600000000000003,
      "loss": 3.2962,
      "step": 9400
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00038,
      "loss": 3.2982,
      "step": 9500
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.000384,
      "loss": 3.295,
      "step": 9600
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.000388,
      "loss": 3.2839,
      "step": 9700
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00039200000000000004,
      "loss": 3.2859,
      "step": 9800
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00039600000000000003,
      "loss": 3.2746,
      "step": 9900
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0004,
      "loss": 3.2595,
      "step": 10000
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.000404,
      "loss": 3.2802,
      "step": 10100
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.000408,
      "loss": 3.2474,
      "step": 10200
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.000412,
      "loss": 3.2535,
      "step": 10300
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.000416,
      "loss": 3.2276,
      "step": 10400
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00042,
      "loss": 3.2365,
      "step": 10500
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.000424,
      "loss": 3.2413,
      "step": 10600
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.000428,
      "loss": 3.2175,
      "step": 10700
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.000432,
      "loss": 3.2231,
      "step": 10800
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.000436,
      "loss": 3.2403,
      "step": 10900
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00044,
      "loss": 3.2022,
      "step": 11000
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.000444,
      "loss": 3.2017,
      "step": 11100
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.000448,
      "loss": 3.2083,
      "step": 11200
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00045200000000000004,
      "loss": 3.1954,
      "step": 11300
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.000456,
      "loss": 3.1911,
      "step": 11400
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00046,
      "loss": 3.1823,
      "step": 11500
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00046400000000000006,
      "loss": 3.1863,
      "step": 11600
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00046800000000000005,
      "loss": 3.1878,
      "step": 11700
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.000472,
      "loss": 3.1734,
      "step": 11800
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00047599999999999997,
      "loss": 3.1782,
      "step": 11900
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00048,
      "loss": 3.1711,
      "step": 12000
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.000484,
      "loss": 3.1747,
      "step": 12100
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.000488,
      "loss": 3.1527,
      "step": 12200
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.000492,
      "loss": 3.15,
      "step": 12300
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.000496,
      "loss": 3.1416,
      "step": 12400
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0005,
      "loss": 3.1455,
      "step": 12500
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.000504,
      "loss": 3.1441,
      "step": 12600
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.000508,
      "loss": 3.1462,
      "step": 12700
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.000512,
      "loss": 3.1531,
      "step": 12800
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0005160000000000001,
      "loss": 3.1394,
      "step": 12900
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0005200000000000001,
      "loss": 3.1304,
      "step": 13000
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.000524,
      "loss": 3.1292,
      "step": 13100
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.000528,
      "loss": 3.1334,
      "step": 13200
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.000532,
      "loss": 3.1129,
      "step": 13300
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.000536,
      "loss": 3.1171,
      "step": 13400
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00054,
      "loss": 3.1147,
      "step": 13500
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0005440000000000001,
      "loss": 3.1122,
      "step": 13600
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0005480000000000001,
      "loss": 3.101,
      "step": 13700
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0005520000000000001,
      "loss": 3.1152,
      "step": 13800
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0005560000000000001,
      "loss": 3.0982,
      "step": 13900
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0005600000000000001,
      "loss": 3.1003,
      "step": 14000
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0005639999999999999,
      "loss": 3.1038,
      "step": 14100
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0005679999999999999,
      "loss": 3.1052,
      "step": 14200
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0005719999999999999,
      "loss": 3.0987,
      "step": 14300
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.000576,
      "loss": 3.0798,
      "step": 14400
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00058,
      "loss": 3.0898,
      "step": 14500
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.000584,
      "loss": 3.0645,
      "step": 14600
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.000588,
      "loss": 3.085,
      "step": 14700
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.000592,
      "loss": 3.0773,
      "step": 14800
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.000596,
      "loss": 3.081,
      "step": 14900
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0006,
      "loss": 3.0728,
      "step": 15000
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.000604,
      "loss": 3.0774,
      "step": 15100
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.000608,
      "loss": 3.0598,
      "step": 15200
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.000612,
      "loss": 3.064,
      "step": 15300
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.000616,
      "loss": 3.0875,
      "step": 15400
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00062,
      "loss": 3.0779,
      "step": 15500
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.000624,
      "loss": 3.0626,
      "step": 15600
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.000628,
      "loss": 3.0632,
      "step": 15700
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.000632,
      "loss": 3.0534,
      "step": 15800
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0006360000000000001,
      "loss": 3.0666,
      "step": 15900
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00064,
      "loss": 3.0665,
      "step": 16000
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.000644,
      "loss": 3.0614,
      "step": 16100
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.000648,
      "loss": 3.0527,
      "step": 16200
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.000652,
      "loss": 3.0384,
      "step": 16300
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.000656,
      "loss": 3.0633,
      "step": 16400
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00066,
      "loss": 3.0308,
      "step": 16500
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0006640000000000001,
      "loss": 3.0476,
      "step": 16600
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0006680000000000001,
      "loss": 3.0514,
      "step": 16700
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0006720000000000001,
      "loss": 3.057,
      "step": 16800
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0006760000000000001,
      "loss": 3.0202,
      "step": 16900
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00068,
      "loss": 3.0328,
      "step": 17000
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.000684,
      "loss": 3.0422,
      "step": 17100
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0006879999999999999,
      "loss": 3.0236,
      "step": 17200
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.000692,
      "loss": 3.0257,
      "step": 17300
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.000696,
      "loss": 3.0346,
      "step": 17400
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0007,
      "loss": 3.0428,
      "step": 17500
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.000704,
      "loss": 3.0297,
      "step": 17600
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.000708,
      "loss": 3.035,
      "step": 17700
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.000712,
      "loss": 3.0265,
      "step": 17800
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.000716,
      "loss": 3.0069,
      "step": 17900
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0007199999999999999,
      "loss": 3.044,
      "step": 18000
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.000724,
      "loss": 3.0174,
      "step": 18100
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.000728,
      "loss": 3.0153,
      "step": 18200
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.000732,
      "loss": 3.0149,
      "step": 18300
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.000736,
      "loss": 3.0094,
      "step": 18400
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00074,
      "loss": 3.0139,
      "step": 18500
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.000744,
      "loss": 3.0115,
      "step": 18600
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.000748,
      "loss": 3.0049,
      "step": 18700
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0007520000000000001,
      "loss": 3.0011,
      "step": 18800
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.000756,
      "loss": 3.0098,
      "step": 18900
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00076,
      "loss": 2.9991,
      "step": 19000
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.000764,
      "loss": 3.0096,
      "step": 19100
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.000768,
      "loss": 3.0174,
      "step": 19200
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.000772,
      "loss": 3.0131,
      "step": 19300
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.000776,
      "loss": 3.003,
      "step": 19400
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0007800000000000001,
      "loss": 3.0081,
      "step": 19500
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0007840000000000001,
      "loss": 2.9973,
      "step": 19600
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0007880000000000001,
      "loss": 3.0083,
      "step": 19700
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0007920000000000001,
      "loss": 3.0031,
      "step": 19800
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.000796,
      "loss": 2.9853,
      "step": 19900
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0008,
      "loss": 2.9972,
      "step": 20000
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.000804,
      "loss": 2.9945,
      "step": 20100
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.000808,
      "loss": 2.9802,
      "step": 20200
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0008120000000000001,
      "loss": 2.9833,
      "step": 20300
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.000816,
      "loss": 2.9901,
      "step": 20400
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00082,
      "loss": 2.9983,
      "step": 20500
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.000824,
      "loss": 2.9861,
      "step": 20600
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.000828,
      "loss": 2.9787,
      "step": 20700
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.000832,
      "loss": 2.9893,
      "step": 20800
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0008359999999999999,
      "loss": 2.9919,
      "step": 20900
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00084,
      "loss": 2.9547,
      "step": 21000
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.000844,
      "loss": 2.9825,
      "step": 21100
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.000848,
      "loss": 2.9905,
      "step": 21200
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.000852,
      "loss": 2.982,
      "step": 21300
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.000856,
      "loss": 2.9726,
      "step": 21400
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00086,
      "loss": 2.9559,
      "step": 21500
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.000864,
      "loss": 2.9818,
      "step": 21600
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0008680000000000001,
      "loss": 2.978,
      "step": 21700
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.000872,
      "loss": 2.9851,
      "step": 21800
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.000876,
      "loss": 2.9761,
      "step": 21900
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00088,
      "loss": 2.9584,
      "step": 22000
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.000884,
      "loss": 2.9792,
      "step": 22100
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.000888,
      "loss": 2.9692,
      "step": 22200
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.000892,
      "loss": 2.9636,
      "step": 22300
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.000896,
      "loss": 2.9668,
      "step": 22400
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0009000000000000001,
      "loss": 2.9632,
      "step": 22500
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0009040000000000001,
      "loss": 2.9857,
      "step": 22600
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0009080000000000001,
      "loss": 2.9574,
      "step": 22700
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.000912,
      "loss": 2.9548,
      "step": 22800
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.000916,
      "loss": 2.9772,
      "step": 22900
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00092,
      "loss": 2.9661,
      "step": 23000
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.000924,
      "loss": 2.9418,
      "step": 23100
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0009280000000000001,
      "loss": 2.9544,
      "step": 23200
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0009320000000000001,
      "loss": 2.978,
      "step": 23300
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0009360000000000001,
      "loss": 2.9759,
      "step": 23400
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00094,
      "loss": 2.9494,
      "step": 23500
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.000944,
      "loss": 2.9474,
      "step": 23600
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.000948,
      "loss": 2.9466,
      "step": 23700
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0009519999999999999,
      "loss": 2.9378,
      "step": 23800
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0009559999999999999,
      "loss": 3.0109,
      "step": 23900
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00096,
      "loss": 2.969,
      "step": 24000
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.000964,
      "loss": 2.9366,
      "step": 24100
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.000968,
      "loss": 2.9381,
      "step": 24200
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.000972,
      "loss": 2.9382,
      "step": 24300
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.000976,
      "loss": 2.9519,
      "step": 24400
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00098,
      "loss": 2.9452,
      "step": 24500
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.000984,
      "loss": 2.942,
      "step": 24600
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.000988,
      "loss": 2.9315,
      "step": 24700
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.000992,
      "loss": 2.938,
      "step": 24800
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.000996,
      "loss": 2.9331,
      "step": 24900
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.001,
      "loss": 2.9488,
      "step": 25000
    },
    {
      "epoch": 1.55,
      "eval_accuracy": 0.49347509891775837,
      "eval_loss": 2.7666990756988525,
      "eval_runtime": 6163.6406,
      "eval_samples_per_second": 35.265,
      "eval_steps_per_second": 2.204,
      "step": 25000
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.000996,
      "loss": 2.9298,
      "step": 25100
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.000992,
      "loss": 2.93,
      "step": 25200
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.000988,
      "loss": 2.9365,
      "step": 25300
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.000984,
      "loss": 2.9522,
      "step": 25400
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00098,
      "loss": 2.9344,
      "step": 25500
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.000976,
      "loss": 2.9349,
      "step": 25600
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.000972,
      "loss": 2.922,
      "step": 25700
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.000968,
      "loss": 2.9094,
      "step": 25800
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.000964,
      "loss": 2.9079,
      "step": 25900
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00096,
      "loss": 2.9033,
      "step": 26000
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0009559999999999999,
      "loss": 2.8957,
      "step": 26100
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0009519999999999999,
      "loss": 2.9106,
      "step": 26200
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.000948,
      "loss": 2.9186,
      "step": 26300
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.000944,
      "loss": 2.8864,
      "step": 26400
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00094,
      "loss": 2.9045,
      "step": 26500
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0009360000000000001,
      "loss": 2.8999,
      "step": 26600
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0009320000000000001,
      "loss": 2.8869,
      "step": 26700
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0009280000000000001,
      "loss": 2.8616,
      "step": 26800
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.000924,
      "loss": 2.8826,
      "step": 26900
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.00092,
      "loss": 2.8858,
      "step": 27000
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.000916,
      "loss": 2.8868,
      "step": 27100
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.000912,
      "loss": 2.8586,
      "step": 27200
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0009080000000000001,
      "loss": 2.872,
      "step": 27300
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0009040000000000001,
      "loss": 2.8575,
      "step": 27400
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0009000000000000001,
      "loss": 2.8648,
      "step": 27500
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.000896,
      "loss": 2.8639,
      "step": 27600
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.000892,
      "loss": 2.8573,
      "step": 27700
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.000888,
      "loss": 2.849,
      "step": 27800
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.000884,
      "loss": 2.8566,
      "step": 27900
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.00088,
      "loss": 2.8574,
      "step": 28000
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.000876,
      "loss": 2.8519,
      "step": 28100
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.000872,
      "loss": 2.8346,
      "step": 28200
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0008680000000000001,
      "loss": 2.8502,
      "step": 28300
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.000864,
      "loss": 2.8269,
      "step": 28400
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.00086,
      "loss": 2.842,
      "step": 28500
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.000856,
      "loss": 2.8386,
      "step": 28600
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.000852,
      "loss": 2.8188,
      "step": 28700
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.000848,
      "loss": 2.8147,
      "step": 28800
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.000844,
      "loss": 2.82,
      "step": 28900
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00084,
      "loss": 2.8251,
      "step": 29000
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0008359999999999999,
      "loss": 2.8202,
      "step": 29100
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.000832,
      "loss": 2.808,
      "step": 29200
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.000828,
      "loss": 2.819,
      "step": 29300
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.000824,
      "loss": 2.8127,
      "step": 29400
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00082,
      "loss": 2.7945,
      "step": 29500
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.000816,
      "loss": 2.8057,
      "step": 29600
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0008120000000000001,
      "loss": 2.8077,
      "step": 29700
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.000808,
      "loss": 2.809,
      "step": 29800
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.000804,
      "loss": 2.8029,
      "step": 29900
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0008,
      "loss": 2.7945,
      "step": 30000
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.000796,
      "loss": 2.7991,
      "step": 30100
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0007920000000000001,
      "loss": 2.7869,
      "step": 30200
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0007880000000000001,
      "loss": 2.7931,
      "step": 30300
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0007840000000000001,
      "loss": 2.7927,
      "step": 30400
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0007800000000000001,
      "loss": 2.787,
      "step": 30500
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.000776,
      "loss": 2.7747,
      "step": 30600
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.000772,
      "loss": 2.7916,
      "step": 30700
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.000768,
      "loss": 2.7759,
      "step": 30800
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.000764,
      "loss": 2.7599,
      "step": 30900
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00076,
      "loss": 2.7578,
      "step": 31000
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.000756,
      "loss": 2.7688,
      "step": 31100
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0007520000000000001,
      "loss": 2.7586,
      "step": 31200
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.000748,
      "loss": 2.7641,
      "step": 31300
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.000744,
      "loss": 2.7688,
      "step": 31400
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00074,
      "loss": 2.761,
      "step": 31500
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.000736,
      "loss": 2.7543,
      "step": 31600
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.000732,
      "loss": 2.7434,
      "step": 31700
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.000728,
      "loss": 2.7458,
      "step": 31800
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.000724,
      "loss": 2.7583,
      "step": 31900
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0007199999999999999,
      "loss": 2.7356,
      "step": 32000
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.000716,
      "loss": 2.7437,
      "step": 32100
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.000712,
      "loss": 2.7392,
      "step": 32200
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.000708,
      "loss": 2.7443,
      "step": 32300
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.000704,
      "loss": 2.7234,
      "step": 32400
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0007,
      "loss": 2.7302,
      "step": 32500
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.000696,
      "loss": 2.72,
      "step": 32600
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.000692,
      "loss": 2.7183,
      "step": 32700
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0006879999999999999,
      "loss": 2.7129,
      "step": 32800
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.000684,
      "loss": 2.7253,
      "step": 32900
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00068,
      "loss": 2.7299,
      "step": 33000
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0006760000000000001,
      "loss": 2.7195,
      "step": 33100
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0006720000000000001,
      "loss": 2.7026,
      "step": 33200
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0006680000000000001,
      "loss": 2.7002,
      "step": 33300
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0006640000000000001,
      "loss": 2.7065,
      "step": 33400
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.00066,
      "loss": 2.7092,
      "step": 33500
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.000656,
      "loss": 2.7014,
      "step": 33600
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.000652,
      "loss": 2.7083,
      "step": 33700
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.000648,
      "loss": 2.6784,
      "step": 33800
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.000644,
      "loss": 2.6933,
      "step": 33900
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.00064,
      "loss": 2.7003,
      "step": 34000
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.0006360000000000001,
      "loss": 2.6866,
      "step": 34100
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.000632,
      "loss": 2.6984,
      "step": 34200
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.000628,
      "loss": 2.678,
      "step": 34300
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.000624,
      "loss": 2.698,
      "step": 34400
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00062,
      "loss": 2.6927,
      "step": 34500
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.000616,
      "loss": 2.6787,
      "step": 34600
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.000612,
      "loss": 2.6916,
      "step": 34700
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.000608,
      "loss": 2.6794,
      "step": 34800
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.000604,
      "loss": 2.6723,
      "step": 34900
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.0006,
      "loss": 2.6582,
      "step": 35000
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.000596,
      "loss": 2.6683,
      "step": 35100
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.000592,
      "loss": 2.6748,
      "step": 35200
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.000588,
      "loss": 2.6711,
      "step": 35300
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.000584,
      "loss": 2.6566,
      "step": 35400
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.00058,
      "loss": 2.6589,
      "step": 35500
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.000576,
      "loss": 2.6656,
      "step": 35600
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.0005719999999999999,
      "loss": 2.6423,
      "step": 35700
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0005679999999999999,
      "loss": 2.6561,
      "step": 35800
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0005639999999999999,
      "loss": 2.6579,
      "step": 35900
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0005600000000000001,
      "loss": 2.6596,
      "step": 36000
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0005560000000000001,
      "loss": 2.6419,
      "step": 36100
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0005520000000000001,
      "loss": 2.6477,
      "step": 36200
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.0005480000000000001,
      "loss": 2.6391,
      "step": 36300
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.0005440000000000001,
      "loss": 2.6337,
      "step": 36400
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.00054,
      "loss": 2.6417,
      "step": 36500
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.000536,
      "loss": 2.6318,
      "step": 36600
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.000532,
      "loss": 2.6409,
      "step": 36700
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.000528,
      "loss": 2.6469,
      "step": 36800
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.000524,
      "loss": 2.6394,
      "step": 36900
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.0005200000000000001,
      "loss": 2.6342,
      "step": 37000
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0005160000000000001,
      "loss": 2.6335,
      "step": 37100
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.000512,
      "loss": 2.6274,
      "step": 37200
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.000508,
      "loss": 2.6209,
      "step": 37300
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.000504,
      "loss": 2.6213,
      "step": 37400
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.0005,
      "loss": 2.623,
      "step": 37500
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.000496,
      "loss": 2.617,
      "step": 37600
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.000492,
      "loss": 2.6276,
      "step": 37700
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.000488,
      "loss": 2.627,
      "step": 37800
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.000484,
      "loss": 2.6128,
      "step": 37900
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.00048,
      "loss": 2.5874,
      "step": 38000
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.00047599999999999997,
      "loss": 2.6047,
      "step": 38100
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.000472,
      "loss": 2.6107,
      "step": 38200
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.00046800000000000005,
      "loss": 2.6277,
      "step": 38300
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00046400000000000006,
      "loss": 2.6046,
      "step": 38400
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.00046,
      "loss": 2.6021,
      "step": 38500
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.000456,
      "loss": 2.5964,
      "step": 38600
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.00045200000000000004,
      "loss": 2.5984,
      "step": 38700
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.000448,
      "loss": 2.6017,
      "step": 38800
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.000444,
      "loss": 2.5829,
      "step": 38900
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.00044,
      "loss": 2.6043,
      "step": 39000
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.000436,
      "loss": 2.5952,
      "step": 39100
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.000432,
      "loss": 2.5914,
      "step": 39200
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.000428,
      "loss": 2.5807,
      "step": 39300
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.000424,
      "loss": 2.5741,
      "step": 39400
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00042,
      "loss": 2.5896,
      "step": 39500
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.000416,
      "loss": 2.5787,
      "step": 39600
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.000412,
      "loss": 2.5855,
      "step": 39700
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.000408,
      "loss": 2.5841,
      "step": 39800
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.000404,
      "loss": 2.5796,
      "step": 39900
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0004,
      "loss": 2.5741,
      "step": 40000
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00039600000000000003,
      "loss": 2.5621,
      "step": 40100
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00039200000000000004,
      "loss": 2.5595,
      "step": 40200
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.000388,
      "loss": 2.5762,
      "step": 40300
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.000384,
      "loss": 2.557,
      "step": 40400
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.00038,
      "loss": 2.572,
      "step": 40500
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.00037600000000000003,
      "loss": 2.5602,
      "step": 40600
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.000372,
      "loss": 2.5448,
      "step": 40700
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.000368,
      "loss": 2.5611,
      "step": 40800
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.000364,
      "loss": 2.5426,
      "step": 40900
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.00035999999999999997,
      "loss": 2.5544,
      "step": 41000
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.000356,
      "loss": 2.5621,
      "step": 41100
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.000352,
      "loss": 2.5451,
      "step": 41200
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.000348,
      "loss": 2.5358,
      "step": 41300
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00034399999999999996,
      "loss": 2.5481,
      "step": 41400
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00034,
      "loss": 2.5407,
      "step": 41500
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00033600000000000004,
      "loss": 2.5451,
      "step": 41600
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00033200000000000005,
      "loss": 2.5317,
      "step": 41700
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.000328,
      "loss": 2.5421,
      "step": 41800
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.000324,
      "loss": 2.5286,
      "step": 41900
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00032,
      "loss": 2.5366,
      "step": 42000
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.000316,
      "loss": 2.5288,
      "step": 42100
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.000312,
      "loss": 2.5221,
      "step": 42200
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.000308,
      "loss": 2.5278,
      "step": 42300
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.000304,
      "loss": 2.545,
      "step": 42400
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.0003,
      "loss": 2.5246,
      "step": 42500
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.000296,
      "loss": 2.5333,
      "step": 42600
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.000292,
      "loss": 2.5258,
      "step": 42700
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.000288,
      "loss": 2.5334,
      "step": 42800
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00028399999999999996,
      "loss": 2.5101,
      "step": 42900
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.00028000000000000003,
      "loss": 2.5168,
      "step": 43000
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.00027600000000000004,
      "loss": 2.5184,
      "step": 43100
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00027200000000000005,
      "loss": 2.5236,
      "step": 43200
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.000268,
      "loss": 2.5015,
      "step": 43300
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.000264,
      "loss": 2.5106,
      "step": 43400
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.00026000000000000003,
      "loss": 2.5095,
      "step": 43500
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.000256,
      "loss": 2.5089,
      "step": 43600
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.000252,
      "loss": 2.5086,
      "step": 43700
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.000248,
      "loss": 2.4996,
      "step": 43800
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.000244,
      "loss": 2.5202,
      "step": 43900
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00024,
      "loss": 2.5012,
      "step": 44000
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.000236,
      "loss": 2.51,
      "step": 44100
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00023200000000000003,
      "loss": 2.5011,
      "step": 44200
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.000228,
      "loss": 2.4794,
      "step": 44300
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.000224,
      "loss": 2.4736,
      "step": 44400
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00022,
      "loss": 2.5016,
      "step": 44500
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.000216,
      "loss": 2.4874,
      "step": 44600
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.000212,
      "loss": 2.4932,
      "step": 44700
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.000208,
      "loss": 2.5032,
      "step": 44800
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.000204,
      "loss": 2.4956,
      "step": 44900
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.0002,
      "loss": 2.5014,
      "step": 45000
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.00019600000000000002,
      "loss": 2.4986,
      "step": 45100
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.000192,
      "loss": 2.4685,
      "step": 45200
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00018800000000000002,
      "loss": 2.4901,
      "step": 45300
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.000184,
      "loss": 2.4783,
      "step": 45400
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00017999999999999998,
      "loss": 2.4947,
      "step": 45500
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.000176,
      "loss": 2.4696,
      "step": 45600
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.00017199999999999998,
      "loss": 2.4728,
      "step": 45700
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00016800000000000002,
      "loss": 2.4791,
      "step": 45800
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.000164,
      "loss": 2.4764,
      "step": 45900
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00016,
      "loss": 2.4861,
      "step": 46000
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.000156,
      "loss": 2.4619,
      "step": 46100
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.000152,
      "loss": 2.4698,
      "step": 46200
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.000148,
      "loss": 2.489,
      "step": 46300
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.000144,
      "loss": 2.4607,
      "step": 46400
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00014000000000000001,
      "loss": 2.4727,
      "step": 46500
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.00013600000000000003,
      "loss": 2.4548,
      "step": 46600
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.000132,
      "loss": 2.4553,
      "step": 46700
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.000128,
      "loss": 2.465,
      "step": 46800
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.000124,
      "loss": 2.4601,
      "step": 46900
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00012,
      "loss": 2.4554,
      "step": 47000
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.00011600000000000001,
      "loss": 2.4575,
      "step": 47100
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.000112,
      "loss": 2.4515,
      "step": 47200
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.000108,
      "loss": 2.4613,
      "step": 47300
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.000104,
      "loss": 2.4583,
      "step": 47400
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0001,
      "loss": 2.4455,
      "step": 47500
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.6e-05,
      "loss": 2.4446,
      "step": 47600
    },
    {
      "epoch": 2.96,
      "learning_rate": 9.2e-05,
      "loss": 2.4488,
      "step": 47700
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.8e-05,
      "loss": 2.4454,
      "step": 47800
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.400000000000001e-05,
      "loss": 2.4445,
      "step": 47900
    },
    {
      "epoch": 2.98,
      "learning_rate": 8e-05,
      "loss": 2.4388,
      "step": 48000
    },
    {
      "epoch": 2.98,
      "learning_rate": 7.6e-05,
      "loss": 2.4506,
      "step": 48100
    },
    {
      "epoch": 2.99,
      "learning_rate": 7.2e-05,
      "loss": 2.4392,
      "step": 48200
    },
    {
      "epoch": 2.99,
      "learning_rate": 6.800000000000001e-05,
      "loss": 2.4408,
      "step": 48300
    },
    {
      "epoch": 3.0,
      "learning_rate": 6.4e-05,
      "loss": 2.4391,
      "step": 48400
    },
    {
      "epoch": 3.01,
      "learning_rate": 6e-05,
      "loss": 2.4281,
      "step": 48500
    },
    {
      "epoch": 3.01,
      "learning_rate": 5.6e-05,
      "loss": 2.4451,
      "step": 48600
    },
    {
      "epoch": 3.02,
      "learning_rate": 5.2e-05,
      "loss": 2.4277,
      "step": 48700
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.8e-05,
      "loss": 2.4388,
      "step": 48800
    },
    {
      "epoch": 3.03,
      "learning_rate": 4.4e-05,
      "loss": 2.4368,
      "step": 48900
    },
    {
      "epoch": 3.04,
      "learning_rate": 4e-05,
      "loss": 2.423,
      "step": 49000
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.6e-05,
      "loss": 2.4458,
      "step": 49100
    },
    {
      "epoch": 3.05,
      "learning_rate": 3.2e-05,
      "loss": 2.4218,
      "step": 49200
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.8e-05,
      "loss": 2.4226,
      "step": 49300
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.4e-05,
      "loss": 2.4287,
      "step": 49400
    },
    {
      "epoch": 3.07,
      "learning_rate": 2e-05,
      "loss": 2.4237,
      "step": 49500
    },
    {
      "epoch": 3.07,
      "learning_rate": 1.6e-05,
      "loss": 2.421,
      "step": 49600
    },
    {
      "epoch": 3.08,
      "learning_rate": 1.2e-05,
      "loss": 2.4214,
      "step": 49700
    },
    {
      "epoch": 3.09,
      "learning_rate": 8e-06,
      "loss": 2.413,
      "step": 49800
    },
    {
      "epoch": 3.09,
      "learning_rate": 4e-06,
      "loss": 2.4309,
      "step": 49900
    },
    {
      "epoch": 3.1,
      "learning_rate": 0.0,
      "loss": 2.4233,
      "step": 50000
    },
    {
      "epoch": 3.1,
      "eval_accuracy": 0.5611543501568519,
      "eval_loss": 2.2921805381774902,
      "eval_runtime": 6159.6175,
      "eval_samples_per_second": 35.288,
      "eval_steps_per_second": 2.206,
      "step": 50000
    },
    {
      "epoch": 3.1,
      "step": 50000,
      "total_flos": 7.554075721728e+17,
      "train_loss": 3.1285950134277343,
      "train_runtime": 151231.41,
      "train_samples_per_second": 84.639,
      "train_steps_per_second": 0.331
    }
  ],
  "max_steps": 50000,
  "num_train_epochs": 4,
  "total_flos": 7.554075721728e+17,
  "trial_name": null,
  "trial_params": null
}