{
  "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": 10.195,
      "step": 100
    },
    {
      "epoch": 0.01,
      "learning_rate": 8e-06,
      "loss": 8.4977,
      "step": 200
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.2e-05,
      "loss": 7.7115,
      "step": 300
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.6e-05,
      "loss": 7.0073,
      "step": 400
    },
    {
      "epoch": 0.03,
      "learning_rate": 2e-05,
      "loss": 6.386,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.4e-05,
      "loss": 5.9934,
      "step": 600
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.8e-05,
      "loss": 5.7449,
      "step": 700
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.2e-05,
      "loss": 5.5865,
      "step": 800
    },
    {
      "epoch": 0.06,
      "learning_rate": 3.6e-05,
      "loss": 5.4709,
      "step": 900
    },
    {
      "epoch": 0.06,
      "learning_rate": 4e-05,
      "loss": 5.3401,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.4e-05,
      "loss": 5.2552,
      "step": 1100
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.8e-05,
      "loss": 5.165,
      "step": 1200
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.2e-05,
      "loss": 5.0858,
      "step": 1300
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.6e-05,
      "loss": 4.9837,
      "step": 1400
    },
    {
      "epoch": 0.09,
      "learning_rate": 6e-05,
      "loss": 4.9136,
      "step": 1500
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.4e-05,
      "loss": 4.8214,
      "step": 1600
    },
    {
      "epoch": 0.11,
      "learning_rate": 6.800000000000001e-05,
      "loss": 4.7654,
      "step": 1700
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.2e-05,
      "loss": 4.693,
      "step": 1800
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.6e-05,
      "loss": 4.6122,
      "step": 1900
    },
    {
      "epoch": 0.12,
      "learning_rate": 8e-05,
      "loss": 4.5828,
      "step": 2000
    },
    {
      "epoch": 0.13,
      "learning_rate": 8.400000000000001e-05,
      "loss": 4.5066,
      "step": 2100
    },
    {
      "epoch": 0.14,
      "learning_rate": 8.8e-05,
      "loss": 4.4385,
      "step": 2200
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.2e-05,
      "loss": 4.4198,
      "step": 2300
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.6e-05,
      "loss": 4.3611,
      "step": 2400
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001,
      "loss": 4.3158,
      "step": 2500
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.000104,
      "loss": 4.2597,
      "step": 2600
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.000108,
      "loss": 4.2201,
      "step": 2700
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.000112,
      "loss": 4.1731,
      "step": 2800
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00011600000000000001,
      "loss": 4.1267,
      "step": 2900
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00012,
      "loss": 4.0979,
      "step": 3000
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.000124,
      "loss": 4.0423,
      "step": 3100
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.000128,
      "loss": 4.0117,
      "step": 3200
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.000132,
      "loss": 3.9583,
      "step": 3300
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00013600000000000003,
      "loss": 3.9295,
      "step": 3400
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00014000000000000001,
      "loss": 3.898,
      "step": 3500
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.000144,
      "loss": 3.8391,
      "step": 3600
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.000148,
      "loss": 3.8254,
      "step": 3700
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.000152,
      "loss": 3.784,
      "step": 3800
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.000156,
      "loss": 3.7579,
      "step": 3900
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00016,
      "loss": 3.7455,
      "step": 4000
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.000164,
      "loss": 3.6903,
      "step": 4100
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00016800000000000002,
      "loss": 3.6621,
      "step": 4200
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00017199999999999998,
      "loss": 3.6469,
      "step": 4300
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.000176,
      "loss": 3.6307,
      "step": 4400
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00017999999999999998,
      "loss": 3.5867,
      "step": 4500
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.000184,
      "loss": 3.5759,
      "step": 4600
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00018800000000000002,
      "loss": 3.5661,
      "step": 4700
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.000192,
      "loss": 3.5462,
      "step": 4800
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019600000000000002,
      "loss": 3.5015,
      "step": 4900
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002,
      "loss": 3.4948,
      "step": 5000
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.000204,
      "loss": 3.4657,
      "step": 5100
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.000208,
      "loss": 3.4767,
      "step": 5200
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.000212,
      "loss": 3.436,
      "step": 5300
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.000216,
      "loss": 3.4037,
      "step": 5400
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00022,
      "loss": 3.3969,
      "step": 5500
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.000224,
      "loss": 3.3928,
      "step": 5600
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.000228,
      "loss": 3.37,
      "step": 5700
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00023200000000000003,
      "loss": 3.3616,
      "step": 5800
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.000236,
      "loss": 3.3497,
      "step": 5900
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00024,
      "loss": 3.3297,
      "step": 6000
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.000244,
      "loss": 3.3118,
      "step": 6100
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.000248,
      "loss": 3.3046,
      "step": 6200
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.000252,
      "loss": 3.2869,
      "step": 6300
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.000256,
      "loss": 3.2808,
      "step": 6400
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00026000000000000003,
      "loss": 3.258,
      "step": 6500
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.000264,
      "loss": 3.2364,
      "step": 6600
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.000268,
      "loss": 3.2288,
      "step": 6700
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00027200000000000005,
      "loss": 3.2161,
      "step": 6800
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00027600000000000004,
      "loss": 3.2074,
      "step": 6900
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00028000000000000003,
      "loss": 3.197,
      "step": 7000
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00028399999999999996,
      "loss": 3.1904,
      "step": 7100
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.000288,
      "loss": 3.1833,
      "step": 7200
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.000292,
      "loss": 3.176,
      "step": 7300
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.000296,
      "loss": 3.1737,
      "step": 7400
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0003,
      "loss": 3.1608,
      "step": 7500
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.000304,
      "loss": 3.1385,
      "step": 7600
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.000308,
      "loss": 3.1352,
      "step": 7700
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.000312,
      "loss": 3.1349,
      "step": 7800
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.000316,
      "loss": 3.1161,
      "step": 7900
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00032,
      "loss": 3.1061,
      "step": 8000
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.000324,
      "loss": 3.1074,
      "step": 8100
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.000328,
      "loss": 3.0859,
      "step": 8200
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00033200000000000005,
      "loss": 3.0818,
      "step": 8300
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00033600000000000004,
      "loss": 3.0857,
      "step": 8400
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00034,
      "loss": 3.0687,
      "step": 8500
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00034399999999999996,
      "loss": 3.0619,
      "step": 8600
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.000348,
      "loss": 3.0348,
      "step": 8700
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.000352,
      "loss": 3.0381,
      "step": 8800
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.000356,
      "loss": 3.0196,
      "step": 8900
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00035999999999999997,
      "loss": 3.0301,
      "step": 9000
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.000364,
      "loss": 3.0148,
      "step": 9100
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.000368,
      "loss": 3.0309,
      "step": 9200
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.000372,
      "loss": 3.0174,
      "step": 9300
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00037600000000000003,
      "loss": 3.0017,
      "step": 9400
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00038,
      "loss": 2.9999,
      "step": 9500
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.000384,
      "loss": 2.9966,
      "step": 9600
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.000388,
      "loss": 2.9843,
      "step": 9700
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00039200000000000004,
      "loss": 2.9854,
      "step": 9800
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00039600000000000003,
      "loss": 2.9827,
      "step": 9900
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0004,
      "loss": 2.965,
      "step": 10000
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.000404,
      "loss": 2.9811,
      "step": 10100
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.000408,
      "loss": 2.9534,
      "step": 10200
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.000412,
      "loss": 2.9656,
      "step": 10300
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.000416,
      "loss": 2.9361,
      "step": 10400
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00042,
      "loss": 2.9461,
      "step": 10500
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.000424,
      "loss": 2.9473,
      "step": 10600
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.000428,
      "loss": 2.9238,
      "step": 10700
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.000432,
      "loss": 2.9323,
      "step": 10800
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.000436,
      "loss": 2.9481,
      "step": 10900
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00044,
      "loss": 2.9187,
      "step": 11000
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.000444,
      "loss": 2.9166,
      "step": 11100
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.000448,
      "loss": 2.9183,
      "step": 11200
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00045200000000000004,
      "loss": 2.9118,
      "step": 11300
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.000456,
      "loss": 2.9008,
      "step": 11400
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00046,
      "loss": 2.8965,
      "step": 11500
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00046400000000000006,
      "loss": 2.8985,
      "step": 11600
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00046800000000000005,
      "loss": 2.9029,
      "step": 11700
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.000472,
      "loss": 2.8906,
      "step": 11800
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00047599999999999997,
      "loss": 2.9007,
      "step": 11900
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00048,
      "loss": 2.8885,
      "step": 12000
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.000484,
      "loss": 2.8926,
      "step": 12100
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.000488,
      "loss": 2.8688,
      "step": 12200
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.000492,
      "loss": 2.8697,
      "step": 12300
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.000496,
      "loss": 2.8661,
      "step": 12400
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0005,
      "loss": 2.8604,
      "step": 12500
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.000504,
      "loss": 2.8673,
      "step": 12600
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.000508,
      "loss": 2.8675,
      "step": 12700
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.000512,
      "loss": 2.8707,
      "step": 12800
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0005160000000000001,
      "loss": 2.8666,
      "step": 12900
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0005200000000000001,
      "loss": 2.85,
      "step": 13000
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.000524,
      "loss": 2.8556,
      "step": 13100
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.000528,
      "loss": 2.8575,
      "step": 13200
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.000532,
      "loss": 2.8361,
      "step": 13300
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.000536,
      "loss": 2.8411,
      "step": 13400
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00054,
      "loss": 2.8409,
      "step": 13500
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0005440000000000001,
      "loss": 2.8363,
      "step": 13600
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0005480000000000001,
      "loss": 2.8257,
      "step": 13700
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0005520000000000001,
      "loss": 2.8399,
      "step": 13800
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0005560000000000001,
      "loss": 2.8253,
      "step": 13900
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0005600000000000001,
      "loss": 2.8277,
      "step": 14000
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0005639999999999999,
      "loss": 2.834,
      "step": 14100
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0005679999999999999,
      "loss": 2.831,
      "step": 14200
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0005719999999999999,
      "loss": 2.8291,
      "step": 14300
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.000576,
      "loss": 2.8085,
      "step": 14400
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00058,
      "loss": 2.8219,
      "step": 14500
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.000584,
      "loss": 2.7945,
      "step": 14600
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.000588,
      "loss": 2.8197,
      "step": 14700
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.000592,
      "loss": 2.8132,
      "step": 14800
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.000596,
      "loss": 2.8157,
      "step": 14900
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0006,
      "loss": 2.8101,
      "step": 15000
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.000604,
      "loss": 2.8123,
      "step": 15100
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.000608,
      "loss": 2.8017,
      "step": 15200
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.000612,
      "loss": 2.8047,
      "step": 15300
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.000616,
      "loss": 2.8224,
      "step": 15400
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00062,
      "loss": 2.8159,
      "step": 15500
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.000624,
      "loss": 2.8053,
      "step": 15600
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.000628,
      "loss": 2.8039,
      "step": 15700
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.000632,
      "loss": 2.7909,
      "step": 15800
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0006360000000000001,
      "loss": 2.8059,
      "step": 15900
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00064,
      "loss": 2.8008,
      "step": 16000
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.000644,
      "loss": 2.8071,
      "step": 16100
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.000648,
      "loss": 2.7936,
      "step": 16200
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.000652,
      "loss": 2.7817,
      "step": 16300
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.000656,
      "loss": 2.8028,
      "step": 16400
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00066,
      "loss": 2.7742,
      "step": 16500
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0006640000000000001,
      "loss": 2.7892,
      "step": 16600
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0006680000000000001,
      "loss": 2.7957,
      "step": 16700
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0006720000000000001,
      "loss": 2.7993,
      "step": 16800
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0006760000000000001,
      "loss": 2.7669,
      "step": 16900
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00068,
      "loss": 2.7788,
      "step": 17000
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.000684,
      "loss": 2.7812,
      "step": 17100
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0006879999999999999,
      "loss": 2.7659,
      "step": 17200
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.000692,
      "loss": 2.7729,
      "step": 17300
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.000696,
      "loss": 2.7875,
      "step": 17400
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0007,
      "loss": 2.7876,
      "step": 17500
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.000704,
      "loss": 2.7802,
      "step": 17600
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.000708,
      "loss": 2.7826,
      "step": 17700
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.000712,
      "loss": 2.7742,
      "step": 17800
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.000716,
      "loss": 2.7578,
      "step": 17900
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0007199999999999999,
      "loss": 2.7884,
      "step": 18000
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.000724,
      "loss": 2.7712,
      "step": 18100
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.000728,
      "loss": 2.7708,
      "step": 18200
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.000732,
      "loss": 2.771,
      "step": 18300
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.000736,
      "loss": 2.7691,
      "step": 18400
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00074,
      "loss": 2.7716,
      "step": 18500
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.000744,
      "loss": 2.7658,
      "step": 18600
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.000748,
      "loss": 2.7614,
      "step": 18700
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0007520000000000001,
      "loss": 2.7769,
      "step": 18800
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.000756,
      "loss": 2.7717,
      "step": 18900
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00076,
      "loss": 2.7608,
      "step": 19000
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.000764,
      "loss": 2.7644,
      "step": 19100
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.000768,
      "loss": 2.7791,
      "step": 19200
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.000772,
      "loss": 2.77,
      "step": 19300
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.000776,
      "loss": 2.7608,
      "step": 19400
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0007800000000000001,
      "loss": 2.7697,
      "step": 19500
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0007840000000000001,
      "loss": 2.7597,
      "step": 19600
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0007880000000000001,
      "loss": 2.7686,
      "step": 19700
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0007920000000000001,
      "loss": 2.7595,
      "step": 19800
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.000796,
      "loss": 2.745,
      "step": 19900
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0008,
      "loss": 2.7619,
      "step": 20000
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.000804,
      "loss": 2.7535,
      "step": 20100
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.000808,
      "loss": 2.7519,
      "step": 20200
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0008120000000000001,
      "loss": 2.7566,
      "step": 20300
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.000816,
      "loss": 2.7683,
      "step": 20400
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00082,
      "loss": 2.7643,
      "step": 20500
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.000824,
      "loss": 2.7591,
      "step": 20600
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.000828,
      "loss": 2.7513,
      "step": 20700
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.000832,
      "loss": 2.7585,
      "step": 20800
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0008359999999999999,
      "loss": 2.7608,
      "step": 20900
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00084,
      "loss": 2.7324,
      "step": 21000
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.000844,
      "loss": 2.7575,
      "step": 21100
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.000848,
      "loss": 2.7659,
      "step": 21200
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.000852,
      "loss": 2.7628,
      "step": 21300
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.000856,
      "loss": 2.7476,
      "step": 21400
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00086,
      "loss": 2.7348,
      "step": 21500
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.000864,
      "loss": 2.7582,
      "step": 21600
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0008680000000000001,
      "loss": 2.7569,
      "step": 21700
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.000872,
      "loss": 2.754,
      "step": 21800
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.000876,
      "loss": 2.7571,
      "step": 21900
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00088,
      "loss": 2.7427,
      "step": 22000
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.000884,
      "loss": 2.7657,
      "step": 22100
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.000888,
      "loss": 2.7527,
      "step": 22200
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.000892,
      "loss": 2.7499,
      "step": 22300
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.000896,
      "loss": 2.7459,
      "step": 22400
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0009000000000000001,
      "loss": 2.7472,
      "step": 22500
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0009040000000000001,
      "loss": 2.7761,
      "step": 22600
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0009080000000000001,
      "loss": 2.7421,
      "step": 22700
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.000912,
      "loss": 2.7459,
      "step": 22800
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.000916,
      "loss": 2.7594,
      "step": 22900
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00092,
      "loss": 2.7552,
      "step": 23000
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.000924,
      "loss": 2.7298,
      "step": 23100
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0009280000000000001,
      "loss": 2.7413,
      "step": 23200
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0009320000000000001,
      "loss": 2.7566,
      "step": 23300
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0009360000000000001,
      "loss": 2.7536,
      "step": 23400
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00094,
      "loss": 2.7431,
      "step": 23500
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.000944,
      "loss": 2.7381,
      "step": 23600
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.000948,
      "loss": 2.7428,
      "step": 23700
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0009519999999999999,
      "loss": 2.7352,
      "step": 23800
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0009559999999999999,
      "loss": 2.7462,
      "step": 23900
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00096,
      "loss": 2.7498,
      "step": 24000
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.000964,
      "loss": 2.7327,
      "step": 24100
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.000968,
      "loss": 2.7351,
      "step": 24200
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.000972,
      "loss": 2.7368,
      "step": 24300
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.000976,
      "loss": 2.7504,
      "step": 24400
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00098,
      "loss": 2.7456,
      "step": 24500
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.000984,
      "loss": 2.7398,
      "step": 24600
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.000988,
      "loss": 2.7282,
      "step": 24700
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.000992,
      "loss": 2.7393,
      "step": 24800
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.000996,
      "loss": 2.7343,
      "step": 24900
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.001,
      "loss": 2.7518,
      "step": 25000
    },
    {
      "epoch": 1.55,
      "eval_accuracy": 0.5213318080618469,
      "eval_loss": 2.5872762203216553,
      "eval_runtime": 6159.3531,
      "eval_samples_per_second": 35.29,
      "eval_steps_per_second": 2.206,
      "step": 25000
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.000996,
      "loss": 2.7304,
      "step": 25100
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.000992,
      "loss": 2.7407,
      "step": 25200
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.000988,
      "loss": 2.7432,
      "step": 25300
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.000984,
      "loss": 2.7605,
      "step": 25400
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00098,
      "loss": 2.7445,
      "step": 25500
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.000976,
      "loss": 2.7514,
      "step": 25600
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.000972,
      "loss": 2.7391,
      "step": 25700
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.000968,
      "loss": 2.7188,
      "step": 25800
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.000964,
      "loss": 2.7218,
      "step": 25900
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00096,
      "loss": 2.7134,
      "step": 26000
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0009559999999999999,
      "loss": 2.7103,
      "step": 26100
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0009519999999999999,
      "loss": 2.755,
      "step": 26200
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.000948,
      "loss": 2.7243,
      "step": 26300
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.000944,
      "loss": 2.6953,
      "step": 26400
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00094,
      "loss": 2.7182,
      "step": 26500
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0009360000000000001,
      "loss": 2.7154,
      "step": 26600
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0009320000000000001,
      "loss": 2.7025,
      "step": 26700
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0009280000000000001,
      "loss": 2.6803,
      "step": 26800
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.000924,
      "loss": 2.6936,
      "step": 26900
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.00092,
      "loss": 2.7001,
      "step": 27000
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.000916,
      "loss": 2.6993,
      "step": 27100
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.000912,
      "loss": 2.6761,
      "step": 27200
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0009080000000000001,
      "loss": 2.6881,
      "step": 27300
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0009040000000000001,
      "loss": 2.6784,
      "step": 27400
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0009000000000000001,
      "loss": 2.6896,
      "step": 27500
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.000896,
      "loss": 2.6848,
      "step": 27600
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.000892,
      "loss": 2.6726,
      "step": 27700
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.000888,
      "loss": 2.667,
      "step": 27800
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.000884,
      "loss": 2.6761,
      "step": 27900
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.00088,
      "loss": 2.6782,
      "step": 28000
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.000876,
      "loss": 2.6735,
      "step": 28100
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.000872,
      "loss": 2.6588,
      "step": 28200
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0008680000000000001,
      "loss": 2.668,
      "step": 28300
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.000864,
      "loss": 2.6486,
      "step": 28400
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.00086,
      "loss": 2.6698,
      "step": 28500
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.000856,
      "loss": 2.6624,
      "step": 28600
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.000852,
      "loss": 2.6426,
      "step": 28700
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.000848,
      "loss": 2.6427,
      "step": 28800
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.000844,
      "loss": 2.6428,
      "step": 28900
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00084,
      "loss": 2.6502,
      "step": 29000
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0008359999999999999,
      "loss": 2.6465,
      "step": 29100
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.000832,
      "loss": 2.6323,
      "step": 29200
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.000828,
      "loss": 2.6417,
      "step": 29300
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.000824,
      "loss": 2.6398,
      "step": 29400
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00082,
      "loss": 2.624,
      "step": 29500
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.000816,
      "loss": 2.6285,
      "step": 29600
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0008120000000000001,
      "loss": 2.6319,
      "step": 29700
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.000808,
      "loss": 2.6353,
      "step": 29800
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.000804,
      "loss": 2.6301,
      "step": 29900
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0008,
      "loss": 2.6278,
      "step": 30000
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.000796,
      "loss": 2.6239,
      "step": 30100
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0007920000000000001,
      "loss": 2.6101,
      "step": 30200
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0007880000000000001,
      "loss": 2.62,
      "step": 30300
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0007840000000000001,
      "loss": 2.6216,
      "step": 30400
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0007800000000000001,
      "loss": 2.6152,
      "step": 30500
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.000776,
      "loss": 2.6032,
      "step": 30600
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.000772,
      "loss": 2.6224,
      "step": 30700
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.000768,
      "loss": 2.6032,
      "step": 30800
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.000764,
      "loss": 2.5919,
      "step": 30900
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00076,
      "loss": 2.5904,
      "step": 31000
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.000756,
      "loss": 2.5956,
      "step": 31100
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0007520000000000001,
      "loss": 2.5912,
      "step": 31200
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.000748,
      "loss": 2.5968,
      "step": 31300
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.000744,
      "loss": 2.6009,
      "step": 31400
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00074,
      "loss": 2.5942,
      "step": 31500
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.000736,
      "loss": 2.588,
      "step": 31600
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.000732,
      "loss": 2.5748,
      "step": 31700
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.000728,
      "loss": 2.5746,
      "step": 31800
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.000724,
      "loss": 2.5879,
      "step": 31900
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0007199999999999999,
      "loss": 2.57,
      "step": 32000
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.000716,
      "loss": 2.5722,
      "step": 32100
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.000712,
      "loss": 2.5742,
      "step": 32200
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.000708,
      "loss": 2.5778,
      "step": 32300
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.000704,
      "loss": 2.5566,
      "step": 32400
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0007,
      "loss": 2.5584,
      "step": 32500
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.000696,
      "loss": 2.5534,
      "step": 32600
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.000692,
      "loss": 2.5508,
      "step": 32700
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0006879999999999999,
      "loss": 2.545,
      "step": 32800
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.000684,
      "loss": 2.5651,
      "step": 32900
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00068,
      "loss": 2.5641,
      "step": 33000
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0006760000000000001,
      "loss": 2.5541,
      "step": 33100
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0006720000000000001,
      "loss": 2.5422,
      "step": 33200
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0006680000000000001,
      "loss": 2.538,
      "step": 33300
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0006640000000000001,
      "loss": 2.5718,
      "step": 33400
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.00066,
      "loss": 2.5429,
      "step": 33500
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.000656,
      "loss": 2.5339,
      "step": 33600
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.000652,
      "loss": 2.5395,
      "step": 33700
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.000648,
      "loss": 2.5152,
      "step": 33800
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.000644,
      "loss": 2.5249,
      "step": 33900
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.00064,
      "loss": 2.5356,
      "step": 34000
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.0006360000000000001,
      "loss": 2.5249,
      "step": 34100
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.000632,
      "loss": 2.5293,
      "step": 34200
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.000628,
      "loss": 2.5128,
      "step": 34300
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.000624,
      "loss": 2.5339,
      "step": 34400
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00062,
      "loss": 2.5242,
      "step": 34500
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.000616,
      "loss": 2.519,
      "step": 34600
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.000612,
      "loss": 2.5265,
      "step": 34700
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.000608,
      "loss": 2.5162,
      "step": 34800
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.000604,
      "loss": 2.5088,
      "step": 34900
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.0006,
      "loss": 2.4965,
      "step": 35000
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.000596,
      "loss": 2.5009,
      "step": 35100
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.000592,
      "loss": 2.5142,
      "step": 35200
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.000588,
      "loss": 2.5097,
      "step": 35300
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.000584,
      "loss": 2.492,
      "step": 35400
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.00058,
      "loss": 2.4913,
      "step": 35500
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.000576,
      "loss": 2.5008,
      "step": 35600
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.0005719999999999999,
      "loss": 2.4828,
      "step": 35700
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0005679999999999999,
      "loss": 2.493,
      "step": 35800
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0005639999999999999,
      "loss": 2.4934,
      "step": 35900
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0005600000000000001,
      "loss": 2.4966,
      "step": 36000
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0005560000000000001,
      "loss": 2.4813,
      "step": 36100
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0005520000000000001,
      "loss": 2.4881,
      "step": 36200
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.0005480000000000001,
      "loss": 2.4766,
      "step": 36300
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.0005440000000000001,
      "loss": 2.4717,
      "step": 36400
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.00054,
      "loss": 2.4767,
      "step": 36500
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.000536,
      "loss": 2.4731,
      "step": 36600
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.000532,
      "loss": 2.4789,
      "step": 36700
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.000528,
      "loss": 2.4785,
      "step": 36800
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.000524,
      "loss": 2.4717,
      "step": 36900
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.0005200000000000001,
      "loss": 2.4715,
      "step": 37000
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0005160000000000001,
      "loss": 2.4764,
      "step": 37100
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.000512,
      "loss": 2.4633,
      "step": 37200
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.000508,
      "loss": 2.4619,
      "step": 37300
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.000504,
      "loss": 2.4614,
      "step": 37400
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.0005,
      "loss": 2.4626,
      "step": 37500
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.000496,
      "loss": 2.4524,
      "step": 37600
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.000492,
      "loss": 2.4671,
      "step": 37700
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.000488,
      "loss": 2.4685,
      "step": 37800
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.000484,
      "loss": 2.4533,
      "step": 37900
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.00048,
      "loss": 2.4287,
      "step": 38000
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.00047599999999999997,
      "loss": 2.4472,
      "step": 38100
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.000472,
      "loss": 2.4451,
      "step": 38200
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.00046800000000000005,
      "loss": 2.4631,
      "step": 38300
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00046400000000000006,
      "loss": 2.4404,
      "step": 38400
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.00046,
      "loss": 2.4413,
      "step": 38500
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.000456,
      "loss": 2.4345,
      "step": 38600
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.00045200000000000004,
      "loss": 2.4403,
      "step": 38700
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.000448,
      "loss": 2.4408,
      "step": 38800
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.000444,
      "loss": 2.4233,
      "step": 38900
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.00044,
      "loss": 2.4415,
      "step": 39000
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.000436,
      "loss": 2.4349,
      "step": 39100
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.000432,
      "loss": 2.4287,
      "step": 39200
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.000428,
      "loss": 2.4232,
      "step": 39300
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.000424,
      "loss": 2.416,
      "step": 39400
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00042,
      "loss": 2.4277,
      "step": 39500
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.000416,
      "loss": 2.4156,
      "step": 39600
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.000412,
      "loss": 2.4263,
      "step": 39700
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.000408,
      "loss": 2.4283,
      "step": 39800
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.000404,
      "loss": 2.4172,
      "step": 39900
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0004,
      "loss": 2.4129,
      "step": 40000
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00039600000000000003,
      "loss": 2.4034,
      "step": 40100
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00039200000000000004,
      "loss": 2.4019,
      "step": 40200
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.000388,
      "loss": 2.4149,
      "step": 40300
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.000384,
      "loss": 2.395,
      "step": 40400
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.00038,
      "loss": 2.4106,
      "step": 40500
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.00037600000000000003,
      "loss": 2.3981,
      "step": 40600
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.000372,
      "loss": 2.3877,
      "step": 40700
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.000368,
      "loss": 2.3989,
      "step": 40800
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.000364,
      "loss": 2.3811,
      "step": 40900
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.00035999999999999997,
      "loss": 2.3954,
      "step": 41000
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.000356,
      "loss": 2.4033,
      "step": 41100
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.000352,
      "loss": 2.3806,
      "step": 41200
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.000348,
      "loss": 2.3749,
      "step": 41300
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00034399999999999996,
      "loss": 2.3896,
      "step": 41400
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00034,
      "loss": 2.3799,
      "step": 41500
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00033600000000000004,
      "loss": 2.385,
      "step": 41600
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00033200000000000005,
      "loss": 2.3733,
      "step": 41700
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.000328,
      "loss": 2.381,
      "step": 41800
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.000324,
      "loss": 2.3716,
      "step": 41900
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00032,
      "loss": 2.3779,
      "step": 42000
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.000316,
      "loss": 2.3693,
      "step": 42100
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.000312,
      "loss": 2.3618,
      "step": 42200
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.000308,
      "loss": 2.3636,
      "step": 42300
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.000304,
      "loss": 2.3797,
      "step": 42400
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.0003,
      "loss": 2.3638,
      "step": 42500
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.000296,
      "loss": 2.3765,
      "step": 42600
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.000292,
      "loss": 2.3635,
      "step": 42700
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.000288,
      "loss": 2.3695,
      "step": 42800
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00028399999999999996,
      "loss": 2.35,
      "step": 42900
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.00028000000000000003,
      "loss": 2.3555,
      "step": 43000
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.00027600000000000004,
      "loss": 2.3592,
      "step": 43100
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00027200000000000005,
      "loss": 2.3623,
      "step": 43200
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.000268,
      "loss": 2.3422,
      "step": 43300
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.000264,
      "loss": 2.3495,
      "step": 43400
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.00026000000000000003,
      "loss": 2.3479,
      "step": 43500
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.000256,
      "loss": 2.3492,
      "step": 43600
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.000252,
      "loss": 2.3475,
      "step": 43700
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.000248,
      "loss": 2.3366,
      "step": 43800
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.000244,
      "loss": 2.3599,
      "step": 43900
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00024,
      "loss": 2.3434,
      "step": 44000
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.000236,
      "loss": 2.3507,
      "step": 44100
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00023200000000000003,
      "loss": 2.3421,
      "step": 44200
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.000228,
      "loss": 2.3198,
      "step": 44300
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.000224,
      "loss": 2.31,
      "step": 44400
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00022,
      "loss": 2.3369,
      "step": 44500
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.000216,
      "loss": 2.3275,
      "step": 44600
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.000212,
      "loss": 2.3318,
      "step": 44700
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.000208,
      "loss": 2.3457,
      "step": 44800
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.000204,
      "loss": 2.3336,
      "step": 44900
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.0002,
      "loss": 2.3411,
      "step": 45000
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.00019600000000000002,
      "loss": 2.338,
      "step": 45100
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.000192,
      "loss": 2.3059,
      "step": 45200
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00018800000000000002,
      "loss": 2.3253,
      "step": 45300
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.000184,
      "loss": 2.3181,
      "step": 45400
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00017999999999999998,
      "loss": 2.3329,
      "step": 45500
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.000176,
      "loss": 2.3128,
      "step": 45600
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.00017199999999999998,
      "loss": 2.3133,
      "step": 45700
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00016800000000000002,
      "loss": 2.3154,
      "step": 45800
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.000164,
      "loss": 2.3158,
      "step": 45900
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00016,
      "loss": 2.3276,
      "step": 46000
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.000156,
      "loss": 2.3017,
      "step": 46100
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.000152,
      "loss": 2.3043,
      "step": 46200
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.000148,
      "loss": 2.3292,
      "step": 46300
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.000144,
      "loss": 2.3008,
      "step": 46400
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00014000000000000001,
      "loss": 2.3083,
      "step": 46500
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.00013600000000000003,
      "loss": 2.297,
      "step": 46600
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.000132,
      "loss": 2.2969,
      "step": 46700
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.000128,
      "loss": 2.304,
      "step": 46800
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.000124,
      "loss": 2.2976,
      "step": 46900
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00012,
      "loss": 2.2944,
      "step": 47000
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.00011600000000000001,
      "loss": 2.2965,
      "step": 47100
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.000112,
      "loss": 2.2915,
      "step": 47200
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.000108,
      "loss": 2.2986,
      "step": 47300
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.000104,
      "loss": 2.2986,
      "step": 47400
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0001,
      "loss": 2.2876,
      "step": 47500
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.6e-05,
      "loss": 2.2889,
      "step": 47600
    },
    {
      "epoch": 2.96,
      "learning_rate": 9.2e-05,
      "loss": 2.2914,
      "step": 47700
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.8e-05,
      "loss": 2.2848,
      "step": 47800
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.400000000000001e-05,
      "loss": 2.2822,
      "step": 47900
    },
    {
      "epoch": 2.98,
      "learning_rate": 8e-05,
      "loss": 2.272,
      "step": 48000
    },
    {
      "epoch": 2.98,
      "learning_rate": 7.6e-05,
      "loss": 2.2862,
      "step": 48100
    },
    {
      "epoch": 2.99,
      "learning_rate": 7.2e-05,
      "loss": 2.2756,
      "step": 48200
    },
    {
      "epoch": 2.99,
      "learning_rate": 6.800000000000001e-05,
      "loss": 2.2796,
      "step": 48300
    },
    {
      "epoch": 3.0,
      "learning_rate": 6.4e-05,
      "loss": 2.2766,
      "step": 48400
    },
    {
      "epoch": 3.01,
      "learning_rate": 6e-05,
      "loss": 2.2662,
      "step": 48500
    },
    {
      "epoch": 3.01,
      "learning_rate": 5.6e-05,
      "loss": 2.2805,
      "step": 48600
    },
    {
      "epoch": 3.02,
      "learning_rate": 5.2e-05,
      "loss": 2.2668,
      "step": 48700
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.8e-05,
      "loss": 2.2776,
      "step": 48800
    },
    {
      "epoch": 3.03,
      "learning_rate": 4.4e-05,
      "loss": 2.2768,
      "step": 48900
    },
    {
      "epoch": 3.04,
      "learning_rate": 4e-05,
      "loss": 2.2644,
      "step": 49000
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.6e-05,
      "loss": 2.2822,
      "step": 49100
    },
    {
      "epoch": 3.05,
      "learning_rate": 3.2e-05,
      "loss": 2.2597,
      "step": 49200
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.8e-05,
      "loss": 2.2602,
      "step": 49300
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.4e-05,
      "loss": 2.2677,
      "step": 49400
    },
    {
      "epoch": 3.07,
      "learning_rate": 2e-05,
      "loss": 2.2615,
      "step": 49500
    },
    {
      "epoch": 3.07,
      "learning_rate": 1.6e-05,
      "loss": 2.26,
      "step": 49600
    },
    {
      "epoch": 3.08,
      "learning_rate": 1.2e-05,
      "loss": 2.2612,
      "step": 49700
    },
    {
      "epoch": 3.09,
      "learning_rate": 8e-06,
      "loss": 2.2487,
      "step": 49800
    },
    {
      "epoch": 3.09,
      "learning_rate": 4e-06,
      "loss": 2.267,
      "step": 49900
    },
    {
      "epoch": 3.1,
      "learning_rate": 0.0,
      "loss": 2.2587,
      "step": 50000
    },
    {
      "epoch": 3.1,
      "eval_accuracy": 0.5823700512250327,
      "eval_loss": 2.1487433910369873,
      "eval_runtime": 6151.3334,
      "eval_samples_per_second": 35.336,
      "eval_steps_per_second": 2.209,
      "step": 50000
    },
    {
      "epoch": 3.1,
      "step": 50000,
      "total_flos": 7.55408847241216e+17,
      "train_loss": 2.8628560260009768,
      "train_runtime": 151959.6932,
      "train_samples_per_second": 84.233,
      "train_steps_per_second": 0.329
    }
  ],
  "max_steps": 50000,
  "num_train_epochs": 4,
  "total_flos": 7.55408847241216e+17,
  "trial_name": null,
  "trial_params": null
}