{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 18.0,
  "global_step": 94050,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 5e-09,
      "loss": 10.4865,
      "step": 1
    },
    {
      "epoch": 0.1,
      "learning_rate": 2.5e-06,
      "loss": 9.4439,
      "step": 500
    },
    {
      "epoch": 0.19,
      "learning_rate": 5e-06,
      "loss": 7.6179,
      "step": 1000
    },
    {
      "epoch": 0.29,
      "learning_rate": 7.5e-06,
      "loss": 6.3619,
      "step": 1500
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 6.0809,
      "step": 2000
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.25e-05,
      "loss": 5.957,
      "step": 2500
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.5e-05,
      "loss": 5.8758,
      "step": 3000
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.75e-05,
      "loss": 5.8102,
      "step": 3500
    },
    {
      "epoch": 0.77,
      "learning_rate": 2e-05,
      "loss": 5.7625,
      "step": 4000
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.25e-05,
      "loss": 5.7195,
      "step": 4500
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.5e-05,
      "loss": 5.6801,
      "step": 5000
    },
    {
      "epoch": 1.05,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 5.6449,
      "step": 5500
    },
    {
      "epoch": 1.15,
      "learning_rate": 3e-05,
      "loss": 5.6134,
      "step": 6000
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 5.591,
      "step": 6500
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.5e-05,
      "loss": 5.5683,
      "step": 7000
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 5.5419,
      "step": 7500
    },
    {
      "epoch": 1.53,
      "learning_rate": 4e-05,
      "loss": 5.5231,
      "step": 8000
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.25e-05,
      "loss": 5.506,
      "step": 8500
    },
    {
      "epoch": 1.72,
      "learning_rate": 4.4995000000000005e-05,
      "loss": 5.4871,
      "step": 9000
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.7495e-05,
      "loss": 5.4763,
      "step": 9500
    },
    {
      "epoch": 1.91,
      "learning_rate": 4.9995000000000005e-05,
      "loss": 5.4615,
      "step": 10000
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.998800480769231e-05,
      "loss": 5.4468,
      "step": 10500
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.997600961538462e-05,
      "loss": 5.4341,
      "step": 11000
    },
    {
      "epoch": 2.2,
      "learning_rate": 4.996399038461539e-05,
      "loss": 5.4224,
      "step": 11500
    },
    {
      "epoch": 2.3,
      "learning_rate": 4.995197115384615e-05,
      "loss": 5.4099,
      "step": 12000
    },
    {
      "epoch": 2.39,
      "learning_rate": 4.993995192307693e-05,
      "loss": 5.3978,
      "step": 12500
    },
    {
      "epoch": 2.49,
      "learning_rate": 4.992795673076923e-05,
      "loss": 5.3897,
      "step": 13000
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.991593750000001e-05,
      "loss": 5.3836,
      "step": 13500
    },
    {
      "epoch": 2.68,
      "learning_rate": 4.990391826923077e-05,
      "loss": 5.3737,
      "step": 14000
    },
    {
      "epoch": 2.78,
      "learning_rate": 4.989189903846154e-05,
      "loss": 5.3668,
      "step": 14500
    },
    {
      "epoch": 2.87,
      "learning_rate": 4.987990384615385e-05,
      "loss": 5.3597,
      "step": 15000
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.986788461538462e-05,
      "loss": 5.3485,
      "step": 15500
    },
    {
      "epoch": 3.06,
      "learning_rate": 4.9855889423076926e-05,
      "loss": 5.3413,
      "step": 16000
    },
    {
      "epoch": 3.16,
      "learning_rate": 4.9843870192307694e-05,
      "loss": 5.338,
      "step": 16500
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.983185096153846e-05,
      "loss": 5.3304,
      "step": 17000
    },
    {
      "epoch": 3.35,
      "learning_rate": 4.981983173076924e-05,
      "loss": 5.3258,
      "step": 17500
    },
    {
      "epoch": 3.44,
      "learning_rate": 4.98078125e-05,
      "loss": 5.317,
      "step": 18000
    },
    {
      "epoch": 3.54,
      "learning_rate": 4.9795793269230774e-05,
      "loss": 5.3134,
      "step": 18500
    },
    {
      "epoch": 3.64,
      "learning_rate": 4.978377403846154e-05,
      "loss": 5.3097,
      "step": 19000
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.977175480769231e-05,
      "loss": 5.3019,
      "step": 19500
    },
    {
      "epoch": 3.83,
      "learning_rate": 4.9759759615384614e-05,
      "loss": 5.2985,
      "step": 20000
    },
    {
      "epoch": 3.92,
      "learning_rate": 4.974774038461539e-05,
      "loss": 5.2942,
      "step": 20500
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.973572115384616e-05,
      "loss": 5.2893,
      "step": 21000
    },
    {
      "epoch": 4.11,
      "learning_rate": 4.9723701923076925e-05,
      "loss": 5.2843,
      "step": 21500
    },
    {
      "epoch": 4.21,
      "learning_rate": 4.971170673076923e-05,
      "loss": 5.2784,
      "step": 22000
    },
    {
      "epoch": 4.31,
      "learning_rate": 4.9699687500000004e-05,
      "loss": 5.2732,
      "step": 22500
    },
    {
      "epoch": 4.4,
      "learning_rate": 4.968766826923077e-05,
      "loss": 5.2701,
      "step": 23000
    },
    {
      "epoch": 4.5,
      "learning_rate": 4.967564903846154e-05,
      "loss": 5.2677,
      "step": 23500
    },
    {
      "epoch": 4.59,
      "learning_rate": 4.9663653846153844e-05,
      "loss": 5.2644,
      "step": 24000
    },
    {
      "epoch": 4.69,
      "learning_rate": 4.965163461538462e-05,
      "loss": 5.2562,
      "step": 24500
    },
    {
      "epoch": 4.78,
      "learning_rate": 4.963963942307693e-05,
      "loss": 5.2557,
      "step": 25000
    },
    {
      "epoch": 4.88,
      "learning_rate": 4.96276201923077e-05,
      "loss": 5.2529,
      "step": 25500
    },
    {
      "epoch": 4.98,
      "learning_rate": 4.9615600961538466e-05,
      "loss": 5.2504,
      "step": 26000
    },
    {
      "epoch": 5.07,
      "learning_rate": 4.9603581730769234e-05,
      "loss": 5.1431,
      "step": 26500
    },
    {
      "epoch": 5.17,
      "learning_rate": 4.95915625e-05,
      "loss": 4.8144,
      "step": 27000
    },
    {
      "epoch": 5.26,
      "learning_rate": 4.957954326923077e-05,
      "loss": 4.5379,
      "step": 27500
    },
    {
      "epoch": 5.36,
      "learning_rate": 4.9567524038461545e-05,
      "loss": 4.282,
      "step": 28000
    },
    {
      "epoch": 5.45,
      "learning_rate": 4.955550480769231e-05,
      "loss": 3.7642,
      "step": 28500
    },
    {
      "epoch": 5.55,
      "learning_rate": 4.9543485576923075e-05,
      "loss": 3.0854,
      "step": 29000
    },
    {
      "epoch": 5.65,
      "learning_rate": 4.9531490384615385e-05,
      "loss": 2.6674,
      "step": 29500
    },
    {
      "epoch": 5.74,
      "learning_rate": 4.951947115384616e-05,
      "loss": 2.2254,
      "step": 30000
    },
    {
      "epoch": 5.84,
      "learning_rate": 4.950745192307693e-05,
      "loss": 1.9446,
      "step": 30500
    },
    {
      "epoch": 5.93,
      "learning_rate": 4.94954326923077e-05,
      "loss": 1.7693,
      "step": 31000
    },
    {
      "epoch": 6.03,
      "learning_rate": 4.94834375e-05,
      "loss": 1.6527,
      "step": 31500
    },
    {
      "epoch": 6.12,
      "learning_rate": 4.9471418269230775e-05,
      "loss": 1.5696,
      "step": 32000
    },
    {
      "epoch": 6.22,
      "learning_rate": 4.9459399038461544e-05,
      "loss": 1.5054,
      "step": 32500
    },
    {
      "epoch": 6.32,
      "learning_rate": 4.944737980769231e-05,
      "loss": 1.4447,
      "step": 33000
    },
    {
      "epoch": 6.41,
      "learning_rate": 4.9435384615384616e-05,
      "loss": 1.3901,
      "step": 33500
    },
    {
      "epoch": 6.51,
      "learning_rate": 4.942336538461539e-05,
      "loss": 1.3332,
      "step": 34000
    },
    {
      "epoch": 6.6,
      "learning_rate": 4.941134615384615e-05,
      "loss": 1.2692,
      "step": 34500
    },
    {
      "epoch": 6.7,
      "learning_rate": 4.939932692307693e-05,
      "loss": 1.217,
      "step": 35000
    },
    {
      "epoch": 6.79,
      "learning_rate": 4.938733173076923e-05,
      "loss": 1.178,
      "step": 35500
    },
    {
      "epoch": 6.89,
      "learning_rate": 4.9375312500000006e-05,
      "loss": 1.143,
      "step": 36000
    },
    {
      "epoch": 6.99,
      "learning_rate": 4.936329326923077e-05,
      "loss": 1.1109,
      "step": 36500
    },
    {
      "epoch": 7.08,
      "learning_rate": 4.935129807692308e-05,
      "loss": 1.0859,
      "step": 37000
    },
    {
      "epoch": 7.18,
      "learning_rate": 4.9339278846153846e-05,
      "loss": 1.0619,
      "step": 37500
    },
    {
      "epoch": 7.27,
      "learning_rate": 4.932725961538462e-05,
      "loss": 1.0387,
      "step": 38000
    },
    {
      "epoch": 7.37,
      "learning_rate": 4.931524038461538e-05,
      "loss": 1.0205,
      "step": 38500
    },
    {
      "epoch": 7.46,
      "learning_rate": 4.930322115384616e-05,
      "loss": 1.0017,
      "step": 39000
    },
    {
      "epoch": 7.56,
      "learning_rate": 4.929122596153846e-05,
      "loss": 0.9856,
      "step": 39500
    },
    {
      "epoch": 7.66,
      "learning_rate": 4.9279206730769236e-05,
      "loss": 0.9707,
      "step": 40000
    },
    {
      "epoch": 7.75,
      "learning_rate": 4.92671875e-05,
      "loss": 0.9574,
      "step": 40500
    },
    {
      "epoch": 7.85,
      "learning_rate": 4.925516826923077e-05,
      "loss": 0.9455,
      "step": 41000
    },
    {
      "epoch": 7.94,
      "learning_rate": 4.924314903846154e-05,
      "loss": 0.9323,
      "step": 41500
    },
    {
      "epoch": 8.04,
      "learning_rate": 4.923112980769231e-05,
      "loss": 0.9199,
      "step": 42000
    },
    {
      "epoch": 8.13,
      "learning_rate": 4.921913461538461e-05,
      "loss": 0.9113,
      "step": 42500
    },
    {
      "epoch": 8.23,
      "learning_rate": 4.920713942307692e-05,
      "loss": 0.9012,
      "step": 43000
    },
    {
      "epoch": 8.33,
      "learning_rate": 4.919512019230769e-05,
      "loss": 0.8939,
      "step": 43500
    },
    {
      "epoch": 8.42,
      "learning_rate": 4.9183100961538466e-05,
      "loss": 0.8851,
      "step": 44000
    },
    {
      "epoch": 8.52,
      "learning_rate": 4.917108173076923e-05,
      "loss": 0.8745,
      "step": 44500
    },
    {
      "epoch": 8.61,
      "learning_rate": 4.91590625e-05,
      "loss": 0.8651,
      "step": 45000
    },
    {
      "epoch": 8.71,
      "learning_rate": 4.914704326923077e-05,
      "loss": 0.8578,
      "step": 45500
    },
    {
      "epoch": 8.8,
      "learning_rate": 4.913502403846154e-05,
      "loss": 0.8519,
      "step": 46000
    },
    {
      "epoch": 8.9,
      "learning_rate": 4.912300480769231e-05,
      "loss": 0.8457,
      "step": 46500
    },
    {
      "epoch": 9.0,
      "learning_rate": 4.911098557692308e-05,
      "loss": 0.8389,
      "step": 47000
    },
    {
      "epoch": 9.09,
      "learning_rate": 4.9098990384615386e-05,
      "loss": 0.8305,
      "step": 47500
    },
    {
      "epoch": 9.19,
      "learning_rate": 4.9086971153846154e-05,
      "loss": 0.8233,
      "step": 48000
    },
    {
      "epoch": 9.28,
      "learning_rate": 4.907495192307692e-05,
      "loss": 0.8189,
      "step": 48500
    },
    {
      "epoch": 9.38,
      "learning_rate": 4.90629326923077e-05,
      "loss": 0.8129,
      "step": 49000
    },
    {
      "epoch": 9.47,
      "learning_rate": 4.905093750000001e-05,
      "loss": 0.8076,
      "step": 49500
    },
    {
      "epoch": 9.57,
      "learning_rate": 4.903894230769231e-05,
      "loss": 0.8019,
      "step": 50000
    },
    {
      "epoch": 9.67,
      "learning_rate": 4.902692307692308e-05,
      "loss": 0.7962,
      "step": 50500
    },
    {
      "epoch": 9.76,
      "learning_rate": 4.901490384615385e-05,
      "loss": 0.7904,
      "step": 51000
    },
    {
      "epoch": 9.86,
      "learning_rate": 4.900288461538462e-05,
      "loss": 0.7879,
      "step": 51500
    },
    {
      "epoch": 9.95,
      "learning_rate": 4.8990865384615384e-05,
      "loss": 0.7811,
      "step": 52000
    },
    {
      "epoch": 10.05,
      "learning_rate": 4.897884615384616e-05,
      "loss": 0.7781,
      "step": 52500
    },
    {
      "epoch": 10.14,
      "learning_rate": 4.896682692307693e-05,
      "loss": 0.7724,
      "step": 53000
    },
    {
      "epoch": 10.24,
      "learning_rate": 4.8954807692307695e-05,
      "loss": 0.7682,
      "step": 53500
    },
    {
      "epoch": 10.33,
      "learning_rate": 4.89428125e-05,
      "loss": 0.7637,
      "step": 54000
    },
    {
      "epoch": 10.43,
      "learning_rate": 4.893081730769231e-05,
      "loss": 0.7592,
      "step": 54500
    },
    {
      "epoch": 10.53,
      "learning_rate": 4.891879807692308e-05,
      "loss": 0.7541,
      "step": 55000
    },
    {
      "epoch": 10.62,
      "learning_rate": 4.890677884615385e-05,
      "loss": 0.75,
      "step": 55500
    },
    {
      "epoch": 10.72,
      "learning_rate": 4.8894759615384614e-05,
      "loss": 0.749,
      "step": 56000
    },
    {
      "epoch": 10.81,
      "learning_rate": 4.888274038461539e-05,
      "loss": 0.7434,
      "step": 56500
    },
    {
      "epoch": 10.91,
      "learning_rate": 4.887074519230769e-05,
      "loss": 0.7407,
      "step": 57000
    },
    {
      "epoch": 11.0,
      "learning_rate": 4.885872596153847e-05,
      "loss": 0.7381,
      "step": 57500
    },
    {
      "epoch": 11.1,
      "learning_rate": 4.884670673076923e-05,
      "loss": 0.7344,
      "step": 58000
    },
    {
      "epoch": 11.2,
      "learning_rate": 4.8834687500000004e-05,
      "loss": 0.7282,
      "step": 58500
    },
    {
      "epoch": 11.29,
      "learning_rate": 4.882269230769231e-05,
      "loss": 0.7272,
      "step": 59000
    },
    {
      "epoch": 11.39,
      "learning_rate": 4.881067307692308e-05,
      "loss": 0.7236,
      "step": 59500
    },
    {
      "epoch": 11.48,
      "learning_rate": 4.8798653846153845e-05,
      "loss": 0.7196,
      "step": 60000
    },
    {
      "epoch": 11.58,
      "learning_rate": 4.878663461538462e-05,
      "loss": 0.7164,
      "step": 60500
    },
    {
      "epoch": 11.67,
      "learning_rate": 4.877461538461539e-05,
      "loss": 0.7129,
      "step": 61000
    },
    {
      "epoch": 11.77,
      "learning_rate": 4.87626201923077e-05,
      "loss": 0.71,
      "step": 61500
    },
    {
      "epoch": 11.87,
      "learning_rate": 4.875060096153846e-05,
      "loss": 0.7088,
      "step": 62000
    },
    {
      "epoch": 11.96,
      "learning_rate": 4.8738581730769235e-05,
      "loss": 0.7057,
      "step": 62500
    },
    {
      "epoch": 12.06,
      "learning_rate": 4.87265625e-05,
      "loss": 0.7022,
      "step": 63000
    },
    {
      "epoch": 12.15,
      "learning_rate": 4.8714567307692313e-05,
      "loss": 0.6977,
      "step": 63500
    },
    {
      "epoch": 12.25,
      "learning_rate": 4.8702548076923075e-05,
      "loss": 0.6988,
      "step": 64000
    },
    {
      "epoch": 12.34,
      "learning_rate": 4.869052884615385e-05,
      "loss": 0.6943,
      "step": 64500
    },
    {
      "epoch": 12.44,
      "learning_rate": 4.867850961538462e-05,
      "loss": 0.6919,
      "step": 65000
    },
    {
      "epoch": 12.54,
      "learning_rate": 4.8666490384615386e-05,
      "loss": 0.6888,
      "step": 65500
    },
    {
      "epoch": 12.63,
      "learning_rate": 4.865449519230769e-05,
      "loss": 0.686,
      "step": 66000
    },
    {
      "epoch": 12.73,
      "learning_rate": 4.8642475961538465e-05,
      "loss": 0.6843,
      "step": 66500
    },
    {
      "epoch": 12.82,
      "learning_rate": 4.863045673076923e-05,
      "loss": 0.681,
      "step": 67000
    },
    {
      "epoch": 12.92,
      "learning_rate": 4.86184375e-05,
      "loss": 0.68,
      "step": 67500
    },
    {
      "epoch": 13.01,
      "learning_rate": 4.860644230769231e-05,
      "loss": 0.6775,
      "step": 68000
    },
    {
      "epoch": 13.11,
      "learning_rate": 4.859442307692308e-05,
      "loss": 0.6745,
      "step": 68500
    },
    {
      "epoch": 13.21,
      "learning_rate": 4.858240384615385e-05,
      "loss": 0.6726,
      "step": 69000
    },
    {
      "epoch": 13.3,
      "learning_rate": 4.8570384615384616e-05,
      "loss": 0.6716,
      "step": 69500
    },
    {
      "epoch": 13.4,
      "learning_rate": 4.855836538461539e-05,
      "loss": 0.6691,
      "step": 70000
    },
    {
      "epoch": 13.49,
      "learning_rate": 4.8546370192307695e-05,
      "loss": 0.6665,
      "step": 70500
    },
    {
      "epoch": 13.59,
      "learning_rate": 4.853435096153846e-05,
      "loss": 0.6625,
      "step": 71000
    },
    {
      "epoch": 13.68,
      "learning_rate": 4.852233173076923e-05,
      "loss": 0.6609,
      "step": 71500
    },
    {
      "epoch": 13.78,
      "learning_rate": 4.8510312500000006e-05,
      "loss": 0.66,
      "step": 72000
    },
    {
      "epoch": 13.88,
      "learning_rate": 4.8498293269230774e-05,
      "loss": 0.6566,
      "step": 72500
    },
    {
      "epoch": 13.97,
      "learning_rate": 4.848627403846154e-05,
      "loss": 0.6561,
      "step": 73000
    },
    {
      "epoch": 14.07,
      "learning_rate": 4.8474278846153847e-05,
      "loss": 0.6536,
      "step": 73500
    },
    {
      "epoch": 14.16,
      "learning_rate": 4.846225961538462e-05,
      "loss": 0.6509,
      "step": 74000
    },
    {
      "epoch": 14.26,
      "learning_rate": 4.845024038461539e-05,
      "loss": 0.6509,
      "step": 74500
    },
    {
      "epoch": 14.35,
      "learning_rate": 4.843822115384616e-05,
      "loss": 0.6471,
      "step": 75000
    },
    {
      "epoch": 14.45,
      "learning_rate": 4.8426201923076926e-05,
      "loss": 0.6459,
      "step": 75500
    },
    {
      "epoch": 14.55,
      "learning_rate": 4.8414206730769237e-05,
      "loss": 0.6452,
      "step": 76000
    },
    {
      "epoch": 14.64,
      "learning_rate": 4.840221153846154e-05,
      "loss": 0.6425,
      "step": 76500
    },
    {
      "epoch": 14.74,
      "learning_rate": 4.839019230769231e-05,
      "loss": 0.6406,
      "step": 77000
    },
    {
      "epoch": 14.83,
      "learning_rate": 4.837817307692308e-05,
      "loss": 0.6393,
      "step": 77500
    },
    {
      "epoch": 14.93,
      "learning_rate": 4.836615384615385e-05,
      "loss": 0.6362,
      "step": 78000
    },
    {
      "epoch": 15.02,
      "learning_rate": 4.835413461538461e-05,
      "loss": 0.6368,
      "step": 78500
    },
    {
      "epoch": 15.12,
      "learning_rate": 4.834211538461539e-05,
      "loss": 0.634,
      "step": 79000
    },
    {
      "epoch": 15.22,
      "learning_rate": 4.8330096153846156e-05,
      "loss": 0.6324,
      "step": 79500
    },
    {
      "epoch": 15.31,
      "learning_rate": 4.831810096153847e-05,
      "loss": 0.6287,
      "step": 80000
    },
    {
      "epoch": 15.41,
      "learning_rate": 4.830608173076923e-05,
      "loss": 0.6286,
      "step": 80500
    },
    {
      "epoch": 15.5,
      "learning_rate": 4.82940625e-05,
      "loss": 0.6267,
      "step": 81000
    },
    {
      "epoch": 15.6,
      "learning_rate": 4.828204326923077e-05,
      "loss": 0.6257,
      "step": 81500
    },
    {
      "epoch": 15.69,
      "learning_rate": 4.827002403846154e-05,
      "loss": 0.6245,
      "step": 82000
    },
    {
      "epoch": 15.79,
      "learning_rate": 4.825800480769231e-05,
      "loss": 0.6216,
      "step": 82500
    },
    {
      "epoch": 15.89,
      "learning_rate": 4.824598557692308e-05,
      "loss": 0.6208,
      "step": 83000
    },
    {
      "epoch": 15.98,
      "learning_rate": 4.8233966346153844e-05,
      "loss": 0.621,
      "step": 83500
    },
    {
      "epoch": 16.08,
      "learning_rate": 4.822194711538462e-05,
      "loss": 0.617,
      "step": 84000
    },
    {
      "epoch": 16.17,
      "learning_rate": 4.820995192307692e-05,
      "loss": 0.6158,
      "step": 84500
    },
    {
      "epoch": 16.27,
      "learning_rate": 4.819795673076923e-05,
      "loss": 0.617,
      "step": 85000
    },
    {
      "epoch": 16.36,
      "learning_rate": 4.81859375e-05,
      "loss": 0.6149,
      "step": 85500
    },
    {
      "epoch": 16.46,
      "learning_rate": 4.817391826923077e-05,
      "loss": 0.6128,
      "step": 86000
    },
    {
      "epoch": 16.56,
      "learning_rate": 4.816189903846154e-05,
      "loss": 0.6119,
      "step": 86500
    },
    {
      "epoch": 16.65,
      "learning_rate": 4.814987980769231e-05,
      "loss": 0.6104,
      "step": 87000
    },
    {
      "epoch": 16.75,
      "learning_rate": 4.8137860576923074e-05,
      "loss": 0.6082,
      "step": 87500
    },
    {
      "epoch": 16.84,
      "learning_rate": 4.812584134615385e-05,
      "loss": 0.6077,
      "step": 88000
    },
    {
      "epoch": 16.94,
      "learning_rate": 4.811382211538462e-05,
      "loss": 0.6066,
      "step": 88500
    },
    {
      "epoch": 17.03,
      "learning_rate": 4.810182692307693e-05,
      "loss": 0.6051,
      "step": 89000
    },
    {
      "epoch": 17.13,
      "learning_rate": 4.808980769230769e-05,
      "loss": 0.6035,
      "step": 89500
    },
    {
      "epoch": 17.22,
      "learning_rate": 4.8077788461538464e-05,
      "loss": 0.6032,
      "step": 90000
    },
    {
      "epoch": 17.32,
      "learning_rate": 4.806576923076923e-05,
      "loss": 0.6001,
      "step": 90500
    },
    {
      "epoch": 17.42,
      "learning_rate": 4.805375e-05,
      "loss": 0.598,
      "step": 91000
    },
    {
      "epoch": 17.51,
      "learning_rate": 4.8041754807692304e-05,
      "loss": 0.5996,
      "step": 91500
    },
    {
      "epoch": 17.61,
      "learning_rate": 4.802973557692308e-05,
      "loss": 0.5987,
      "step": 92000
    },
    {
      "epoch": 17.7,
      "learning_rate": 4.801771634615385e-05,
      "loss": 0.5973,
      "step": 92500
    },
    {
      "epoch": 17.8,
      "learning_rate": 4.800572115384616e-05,
      "loss": 0.5957,
      "step": 93000
    },
    {
      "epoch": 17.89,
      "learning_rate": 4.7993701923076926e-05,
      "loss": 0.5938,
      "step": 93500
    },
    {
      "epoch": 17.99,
      "learning_rate": 4.7981682692307694e-05,
      "loss": 0.593,
      "step": 94000
    }
  ],
  "max_steps": 2090000,
  "num_train_epochs": 400,
  "total_flos": 2.5346531711380357e+19,
  "trial_name": null,
  "trial_params": null
}