{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 73.36757153338225,
  "eval_steps": 500,
  "global_step": 100000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.73,
      "learning_rate": 4.000000000000001e-06,
      "loss": 7.1234,
      "step": 1000
    },
    {
      "epoch": 1.47,
      "learning_rate": 8.000000000000001e-06,
      "loss": 6.5465,
      "step": 2000
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.2e-05,
      "loss": 6.3804,
      "step": 3000
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 6.3328,
      "step": 4000
    },
    {
      "epoch": 3.67,
      "learning_rate": 2e-05,
      "loss": 6.3097,
      "step": 5000
    },
    {
      "epoch": 4.4,
      "learning_rate": 2.4e-05,
      "loss": 6.2978,
      "step": 6000
    },
    {
      "epoch": 5.14,
      "learning_rate": 2.8e-05,
      "loss": 6.2887,
      "step": 7000
    },
    {
      "epoch": 5.87,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 6.278,
      "step": 8000
    },
    {
      "epoch": 6.6,
      "learning_rate": 3.6e-05,
      "loss": 6.2731,
      "step": 9000
    },
    {
      "epoch": 7.34,
      "learning_rate": 4e-05,
      "loss": 6.2669,
      "step": 10000
    },
    {
      "epoch": 8.07,
      "learning_rate": 3.955555555555556e-05,
      "loss": 6.2622,
      "step": 11000
    },
    {
      "epoch": 8.8,
      "learning_rate": 3.9111111111111115e-05,
      "loss": 6.2592,
      "step": 12000
    },
    {
      "epoch": 9.54,
      "learning_rate": 3.866666666666667e-05,
      "loss": 6.2562,
      "step": 13000
    },
    {
      "epoch": 10.27,
      "learning_rate": 3.8222222222222226e-05,
      "loss": 6.2553,
      "step": 14000
    },
    {
      "epoch": 11.01,
      "learning_rate": 3.777777777777778e-05,
      "loss": 6.25,
      "step": 15000
    },
    {
      "epoch": 11.74,
      "learning_rate": 3.733333333333334e-05,
      "loss": 6.249,
      "step": 16000
    },
    {
      "epoch": 12.47,
      "learning_rate": 3.6888888888888896e-05,
      "loss": 6.1997,
      "step": 17000
    },
    {
      "epoch": 13.21,
      "learning_rate": 3.644444444444445e-05,
      "loss": 6.1146,
      "step": 18000
    },
    {
      "epoch": 13.94,
      "learning_rate": 3.6e-05,
      "loss": 6.0516,
      "step": 19000
    },
    {
      "epoch": 14.67,
      "learning_rate": 3.555555555555555e-05,
      "loss": 5.99,
      "step": 20000
    },
    {
      "epoch": 15.41,
      "learning_rate": 3.511111111111111e-05,
      "loss": 5.9323,
      "step": 21000
    },
    {
      "epoch": 16.14,
      "learning_rate": 3.466666666666667e-05,
      "loss": 5.8738,
      "step": 22000
    },
    {
      "epoch": 16.87,
      "learning_rate": 3.4222222222222224e-05,
      "loss": 5.8147,
      "step": 23000
    },
    {
      "epoch": 17.61,
      "learning_rate": 3.377777777777778e-05,
      "loss": 5.7298,
      "step": 24000
    },
    {
      "epoch": 18.34,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 5.5738,
      "step": 25000
    },
    {
      "epoch": 19.08,
      "learning_rate": 3.288888888888889e-05,
      "loss": 5.3626,
      "step": 26000
    },
    {
      "epoch": 19.81,
      "learning_rate": 3.2444444444444446e-05,
      "loss": 5.2659,
      "step": 27000
    },
    {
      "epoch": 20.54,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 5.2049,
      "step": 28000
    },
    {
      "epoch": 21.28,
      "learning_rate": 3.155555555555556e-05,
      "loss": 5.1549,
      "step": 29000
    },
    {
      "epoch": 22.01,
      "learning_rate": 3.111111111111112e-05,
      "loss": 5.1192,
      "step": 30000
    },
    {
      "epoch": 22.74,
      "learning_rate": 3.066666666666667e-05,
      "loss": 5.0853,
      "step": 31000
    },
    {
      "epoch": 23.48,
      "learning_rate": 3.0222222222222225e-05,
      "loss": 5.0582,
      "step": 32000
    },
    {
      "epoch": 24.21,
      "learning_rate": 2.977777777777778e-05,
      "loss": 5.0324,
      "step": 33000
    },
    {
      "epoch": 24.94,
      "learning_rate": 2.9333333333333333e-05,
      "loss": 5.0123,
      "step": 34000
    },
    {
      "epoch": 25.68,
      "learning_rate": 2.888888888888889e-05,
      "loss": 4.9906,
      "step": 35000
    },
    {
      "epoch": 26.41,
      "learning_rate": 2.8444444444444447e-05,
      "loss": 4.9727,
      "step": 36000
    },
    {
      "epoch": 27.15,
      "learning_rate": 2.8e-05,
      "loss": 4.9547,
      "step": 37000
    },
    {
      "epoch": 27.88,
      "learning_rate": 2.755555555555556e-05,
      "loss": 4.9421,
      "step": 38000
    },
    {
      "epoch": 28.61,
      "learning_rate": 2.7111111111111114e-05,
      "loss": 4.927,
      "step": 39000
    },
    {
      "epoch": 29.35,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 4.9147,
      "step": 40000
    },
    {
      "epoch": 30.08,
      "learning_rate": 2.6222222222222226e-05,
      "loss": 4.9032,
      "step": 41000
    },
    {
      "epoch": 30.81,
      "learning_rate": 2.577777777777778e-05,
      "loss": 4.8913,
      "step": 42000
    },
    {
      "epoch": 31.55,
      "learning_rate": 2.5333333333333334e-05,
      "loss": 4.8793,
      "step": 43000
    },
    {
      "epoch": 32.28,
      "learning_rate": 2.4888888888888893e-05,
      "loss": 4.8695,
      "step": 44000
    },
    {
      "epoch": 33.02,
      "learning_rate": 2.444444444444445e-05,
      "loss": 4.8613,
      "step": 45000
    },
    {
      "epoch": 33.75,
      "learning_rate": 2.4e-05,
      "loss": 4.8524,
      "step": 46000
    },
    {
      "epoch": 34.48,
      "learning_rate": 2.355555555555556e-05,
      "loss": 4.8416,
      "step": 47000
    },
    {
      "epoch": 35.22,
      "learning_rate": 2.3111111111111112e-05,
      "loss": 4.836,
      "step": 48000
    },
    {
      "epoch": 35.95,
      "learning_rate": 2.2666666666666668e-05,
      "loss": 4.8259,
      "step": 49000
    },
    {
      "epoch": 36.68,
      "learning_rate": 2.2222222222222227e-05,
      "loss": 4.8206,
      "step": 50000
    },
    {
      "epoch": 37.42,
      "learning_rate": 2.177777777777778e-05,
      "loss": 4.8121,
      "step": 51000
    },
    {
      "epoch": 38.15,
      "learning_rate": 2.1333333333333335e-05,
      "loss": 4.8051,
      "step": 52000
    },
    {
      "epoch": 38.88,
      "learning_rate": 2.088888888888889e-05,
      "loss": 4.7994,
      "step": 53000
    },
    {
      "epoch": 39.62,
      "learning_rate": 2.0444444444444446e-05,
      "loss": 4.7937,
      "step": 54000
    },
    {
      "epoch": 40.35,
      "learning_rate": 2e-05,
      "loss": 4.7877,
      "step": 55000
    },
    {
      "epoch": 41.09,
      "learning_rate": 1.9555555555555557e-05,
      "loss": 4.7825,
      "step": 56000
    },
    {
      "epoch": 41.82,
      "learning_rate": 1.9111111111111113e-05,
      "loss": 4.7752,
      "step": 57000
    },
    {
      "epoch": 42.55,
      "learning_rate": 1.866666666666667e-05,
      "loss": 4.7724,
      "step": 58000
    },
    {
      "epoch": 43.29,
      "learning_rate": 1.8222222222222224e-05,
      "loss": 4.7656,
      "step": 59000
    },
    {
      "epoch": 44.02,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 4.7604,
      "step": 60000
    },
    {
      "epoch": 44.75,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 4.7554,
      "step": 61000
    },
    {
      "epoch": 45.49,
      "learning_rate": 1.688888888888889e-05,
      "loss": 4.7521,
      "step": 62000
    },
    {
      "epoch": 46.22,
      "learning_rate": 1.6444444444444444e-05,
      "loss": 4.7477,
      "step": 63000
    },
    {
      "epoch": 46.96,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 4.7434,
      "step": 64000
    },
    {
      "epoch": 47.69,
      "learning_rate": 1.555555555555556e-05,
      "loss": 4.7393,
      "step": 65000
    },
    {
      "epoch": 48.42,
      "learning_rate": 1.5111111111111112e-05,
      "loss": 4.7363,
      "step": 66000
    },
    {
      "epoch": 49.16,
      "learning_rate": 1.4666666666666666e-05,
      "loss": 4.7306,
      "step": 67000
    },
    {
      "epoch": 49.89,
      "learning_rate": 1.4222222222222224e-05,
      "loss": 4.7291,
      "step": 68000
    },
    {
      "epoch": 50.62,
      "learning_rate": 1.377777777777778e-05,
      "loss": 4.7253,
      "step": 69000
    },
    {
      "epoch": 51.36,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 4.7215,
      "step": 70000
    },
    {
      "epoch": 52.09,
      "learning_rate": 1.288888888888889e-05,
      "loss": 4.7175,
      "step": 71000
    },
    {
      "epoch": 52.82,
      "learning_rate": 1.2444444444444446e-05,
      "loss": 4.7149,
      "step": 72000
    },
    {
      "epoch": 53.56,
      "learning_rate": 1.2e-05,
      "loss": 4.7141,
      "step": 73000
    },
    {
      "epoch": 54.29,
      "learning_rate": 1.1555555555555556e-05,
      "loss": 4.7086,
      "step": 74000
    },
    {
      "epoch": 55.03,
      "learning_rate": 1.1111111111111113e-05,
      "loss": 4.7089,
      "step": 75000
    },
    {
      "epoch": 55.76,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 4.7049,
      "step": 76000
    },
    {
      "epoch": 56.49,
      "learning_rate": 1.0222222222222223e-05,
      "loss": 4.7007,
      "step": 77000
    },
    {
      "epoch": 57.23,
      "learning_rate": 9.777777777777779e-06,
      "loss": 4.7008,
      "step": 78000
    },
    {
      "epoch": 57.96,
      "learning_rate": 9.333333333333334e-06,
      "loss": 4.6981,
      "step": 79000
    },
    {
      "epoch": 58.69,
      "learning_rate": 8.888888888888888e-06,
      "loss": 4.6962,
      "step": 80000
    },
    {
      "epoch": 59.43,
      "learning_rate": 8.444444444444446e-06,
      "loss": 4.6943,
      "step": 81000
    },
    {
      "epoch": 60.16,
      "learning_rate": 8.000000000000001e-06,
      "loss": 4.6907,
      "step": 82000
    },
    {
      "epoch": 60.9,
      "learning_rate": 7.555555555555556e-06,
      "loss": 4.6921,
      "step": 83000
    },
    {
      "epoch": 61.63,
      "learning_rate": 7.111111111111112e-06,
      "loss": 4.6866,
      "step": 84000
    },
    {
      "epoch": 62.36,
      "learning_rate": 6.666666666666667e-06,
      "loss": 4.6864,
      "step": 85000
    },
    {
      "epoch": 63.1,
      "learning_rate": 6.222222222222223e-06,
      "loss": 4.6841,
      "step": 86000
    },
    {
      "epoch": 63.83,
      "learning_rate": 5.777777777777778e-06,
      "loss": 4.6831,
      "step": 87000
    },
    {
      "epoch": 64.56,
      "learning_rate": 5.333333333333334e-06,
      "loss": 4.6829,
      "step": 88000
    },
    {
      "epoch": 65.3,
      "learning_rate": 4.888888888888889e-06,
      "loss": 4.6783,
      "step": 89000
    },
    {
      "epoch": 66.03,
      "learning_rate": 4.444444444444444e-06,
      "loss": 4.6789,
      "step": 90000
    },
    {
      "epoch": 66.76,
      "learning_rate": 4.000000000000001e-06,
      "loss": 4.6787,
      "step": 91000
    },
    {
      "epoch": 67.5,
      "learning_rate": 3.555555555555556e-06,
      "loss": 4.6763,
      "step": 92000
    },
    {
      "epoch": 68.23,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 4.6766,
      "step": 93000
    },
    {
      "epoch": 68.97,
      "learning_rate": 2.666666666666667e-06,
      "loss": 4.6766,
      "step": 94000
    },
    {
      "epoch": 69.7,
      "learning_rate": 2.222222222222222e-06,
      "loss": 4.6757,
      "step": 95000
    },
    {
      "epoch": 70.43,
      "learning_rate": 1.777777777777778e-06,
      "loss": 4.6728,
      "step": 96000
    },
    {
      "epoch": 71.17,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 4.6734,
      "step": 97000
    },
    {
      "epoch": 71.9,
      "learning_rate": 8.88888888888889e-07,
      "loss": 4.6732,
      "step": 98000
    },
    {
      "epoch": 72.63,
      "learning_rate": 4.444444444444445e-07,
      "loss": 4.6735,
      "step": 99000
    },
    {
      "epoch": 73.37,
      "learning_rate": 0.0,
      "loss": 4.67,
      "step": 100000
    },
    {
      "epoch": 73.37,
      "step": 100000,
      "total_flos": 2.0207652764961997e+19,
      "train_loss": 5.1539158251953126,
      "train_runtime": 83105.1599,
      "train_samples_per_second": 924.13,
      "train_steps_per_second": 1.203
    }
  ],
  "logging_steps": 1000,
  "max_steps": 100000,
  "num_train_epochs": 74,
  "save_steps": 1000,
  "total_flos": 2.0207652764961997e+19,
  "trial_name": null,
  "trial_params": null
}