{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.999970181119878,
  "eval_steps": 500,
  "global_step": 419195,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 500
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 1000
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 1500
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 2000
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 2500
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 3000
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 3500
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 4000
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 4500
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 5000
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 5500
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 6000
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 6500
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 7000
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 7500
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 8000
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 8500
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 9000
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 9500
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 10000
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 10500
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 11000
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 11500
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 12000
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 12500
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 13000
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 13500
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 14000
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 14500
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 15000
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 15500
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 16000
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 16500
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 17000
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 17500
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 18000
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 18500
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 19000
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 19500
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 20000
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 20500
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 21000
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 21500
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 22000
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 22500
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 23000
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 23500
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 24000
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 24500
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 25000
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 25500
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 26000
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 26500
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 27000
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 27500
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 28000
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 28500
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 29000
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 29500
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 30000
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 30500
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 31000
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 31500
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 32000
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 32500
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 33000
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 33500
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 34000
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 34500
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 35000
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 35500
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 36000
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 36500
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 37000
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 37500
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 38000
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 38500
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 39000
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 39500
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 40000
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 40500
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 41000
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 41500
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 42000
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 42500
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 43000
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 43500
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 44000
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 44500
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 45000
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 45500
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 46000
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 46500
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 47000
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 47500
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 48000
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 48500
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 49000
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 49500
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 50000
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 50500
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 51000
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 51500
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 52000
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 52500
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 53000
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 53500
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 54000
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 54500
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 55000
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 55500
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 56000
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 56500
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 57000
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 57500
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 58000
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 58500
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 59000
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 59500
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 60000
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 60500
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 61000
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 61500
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 62000
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 62500
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 63000
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 63500
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 64000
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 64500
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 65000
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 65500
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 66000
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 66500
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 67000
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 67500
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 68000
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 68500
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 69000
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 69500
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 70000
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 70500
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 71000
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 71500
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 72000
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 72500
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 73000
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 73500
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 74000
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 74500
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 75000
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 75500
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 76000
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 76500
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 77000
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 77500
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 78000
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 78500
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 79000
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 79500
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 80000
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 80500
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 81000
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 81500
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 82000
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 82500
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 83000
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 83500
    },
    {
      "epoch": 1.0,
      "eval_loss": NaN,
      "eval_runtime": 2514.1501,
      "eval_samples_per_second": 16.674,
      "eval_steps_per_second": 8.337,
      "step": 83839
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 84000
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 84500
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 85000
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 85500
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 86000
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 86500
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 87000
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 87500
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 88000
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 88500
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 89000
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 89500
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 90000
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 90500
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 91000
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 91500
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 92000
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 92500
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 93000
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 93500
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 94000
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 94500
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 95000
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 95500
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 96000
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 96500
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 97000
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 97500
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 98000
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 98500
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 99000
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 99500
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 100000
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 100500
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 101000
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 101500
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 102000
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 102500
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 103000
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 103500
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 104000
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 104500
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 105000
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 105500
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 106000
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 106500
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 107000
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 107500
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 108000
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 108500
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 109000
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 109500
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 110000
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 110500
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 111000
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 111500
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 112000
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 112500
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 113000
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 113500
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 114000
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 114500
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 115000
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 115500
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 116000
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 116500
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 117000
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 117500
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 118000
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 118500
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 119000
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 119500
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 120000
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 120500
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 121000
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 121500
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 122000
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 122500
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 123000
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 123500
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 124000
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 124500
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 125000
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 125500
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 126000
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 126500
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 127000
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 127500
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 128000
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 128500
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 129000
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 129500
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 130000
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 130500
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 131000
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 131500
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 132000
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 132500
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 133000
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 133500
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 134000
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 134500
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 135000
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 135500
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 136000
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 136500
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 137000
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 137500
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 138000
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 138500
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 139000
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 139500
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 140000
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 140500
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 141000
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 141500
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 142000
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 142500
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 143000
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 143500
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 144000
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 144500
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 145000
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 145500
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 146000
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 146500
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 147000
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 147500
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 148000
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 148500
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 149000
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 149500
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 150000
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 150500
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 151000
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 151500
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 152000
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 152500
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 153000
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 153500
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 154000
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 154500
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 155000
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 155500
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 156000
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 156500
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 157000
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 157500
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 158000
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 158500
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 159000
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 159500
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 160000
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 160500
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 161000
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 161500
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 162000
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 162500
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 163000
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 163500
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 164000
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 164500
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 165000
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 165500
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 166000
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 166500
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 167000
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 167500
    },
    {
      "epoch": 2.0,
      "eval_loss": NaN,
      "eval_runtime": 2358.0971,
      "eval_samples_per_second": 17.777,
      "eval_steps_per_second": 8.889,
      "step": 167679
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 168000
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 168500
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 169000
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 169500
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 170000
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 170500
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 171000
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 171500
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 172000
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 172500
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 173000
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 173500
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 174000
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 174500
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 175000
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 175500
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 176000
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 176500
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 177000
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 177500
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 178000
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 178500
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 179000
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 179500
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 180000
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 180500
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 181000
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 181500
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 182000
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 182500
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 183000
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 183500
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 184000
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 184500
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 185000
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 185500
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 186000
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 186500
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 187000
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 187500
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 188000
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 188500
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 189000
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 189500
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 190000
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 190500
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 191000
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 191500
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 192000
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 192500
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 193000
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 193500
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 194000
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 194500
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 195000
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 195500
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 196000
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 196500
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 197000
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 197500
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 198000
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 198500
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 199000
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 199500
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 200000
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 200500
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 201000
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 201500
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 202000
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 202500
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 203000
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 203500
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 204000
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 204500
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 205000
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 205500
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 206000
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 206500
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 207000
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 207500
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 208000
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 208500
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 209000
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 209500
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 210000
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 210500
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 211000
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 211500
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 212000
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 212500
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 213000
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 213500
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 214000
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 214500
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 215000
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 215500
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 216000
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 216500
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 217000
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 217500
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 218000
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 218500
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 219000
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 219500
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 220000
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 220500
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 221000
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 221500
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 222000
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 222500
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 223000
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 223500
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 224000
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 224500
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 225000
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 225500
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 226000
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 226500
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 227000
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 227500
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 228000
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 228500
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 229000
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 229500
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 230000
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 230500
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 231000
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 231500
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 232000
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 232500
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 233000
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 233500
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 234000
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 234500
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 235000
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 235500
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 236000
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 236500
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 237000
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 237500
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 238000
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 238500
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 239000
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 239500
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 240000
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 240500
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 241000
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 241500
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 242000
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 242500
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 243000
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 243500
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 244000
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 244500
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 245000
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 245500
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 246000
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 246500
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 247000
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 247500
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 248000
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 248500
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 249000
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 249500
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 250000
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 250500
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 251000
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 251500
    },
    {
      "epoch": 3.0,
      "eval_loss": NaN,
      "eval_runtime": 2358.2867,
      "eval_samples_per_second": 17.776,
      "eval_steps_per_second": 8.888,
      "step": 251518
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 252000
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 252500
    },
    {
      "epoch": 3.02,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 253000
    },
    {
      "epoch": 3.02,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 253500
    },
    {
      "epoch": 3.03,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 254000
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 254500
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 255000
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 255500
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 256000
    },
    {
      "epoch": 3.06,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 256500
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 257000
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 257500
    },
    {
      "epoch": 3.08,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 258000
    },
    {
      "epoch": 3.08,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 258500
    },
    {
      "epoch": 3.09,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 259000
    },
    {
      "epoch": 3.1,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 259500
    },
    {
      "epoch": 3.1,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 260000
    },
    {
      "epoch": 3.11,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 260500
    },
    {
      "epoch": 3.11,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 261000
    },
    {
      "epoch": 3.12,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 261500
    },
    {
      "epoch": 3.13,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 262000
    },
    {
      "epoch": 3.13,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 262500
    },
    {
      "epoch": 3.14,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 263000
    },
    {
      "epoch": 3.14,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 263500
    },
    {
      "epoch": 3.15,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 264000
    },
    {
      "epoch": 3.15,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 264500
    },
    {
      "epoch": 3.16,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 265000
    },
    {
      "epoch": 3.17,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 265500
    },
    {
      "epoch": 3.17,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 266000
    },
    {
      "epoch": 3.18,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 266500
    },
    {
      "epoch": 3.18,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 267000
    },
    {
      "epoch": 3.19,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 267500
    },
    {
      "epoch": 3.2,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 268000
    },
    {
      "epoch": 3.2,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 268500
    },
    {
      "epoch": 3.21,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 269000
    },
    {
      "epoch": 3.21,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 269500
    },
    {
      "epoch": 3.22,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 270000
    },
    {
      "epoch": 3.23,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 270500
    },
    {
      "epoch": 3.23,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 271000
    },
    {
      "epoch": 3.24,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 271500
    },
    {
      "epoch": 3.24,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 272000
    },
    {
      "epoch": 3.25,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 272500
    },
    {
      "epoch": 3.26,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 273000
    },
    {
      "epoch": 3.26,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 273500
    },
    {
      "epoch": 3.27,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 274000
    },
    {
      "epoch": 3.27,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 274500
    },
    {
      "epoch": 3.28,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 275000
    },
    {
      "epoch": 3.29,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 275500
    },
    {
      "epoch": 3.29,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 276000
    },
    {
      "epoch": 3.3,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 276500
    },
    {
      "epoch": 3.3,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 277000
    },
    {
      "epoch": 3.31,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 277500
    },
    {
      "epoch": 3.32,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 278000
    },
    {
      "epoch": 3.32,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 278500
    },
    {
      "epoch": 3.33,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 279000
    },
    {
      "epoch": 3.33,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 279500
    },
    {
      "epoch": 3.34,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 280000
    },
    {
      "epoch": 3.35,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 280500
    },
    {
      "epoch": 3.35,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 281000
    },
    {
      "epoch": 3.36,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 281500
    },
    {
      "epoch": 3.36,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 282000
    },
    {
      "epoch": 3.37,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 282500
    },
    {
      "epoch": 3.38,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 283000
    },
    {
      "epoch": 3.38,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 283500
    },
    {
      "epoch": 3.39,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 284000
    },
    {
      "epoch": 3.39,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 284500
    },
    {
      "epoch": 3.4,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 285000
    },
    {
      "epoch": 3.41,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 285500
    },
    {
      "epoch": 3.41,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 286000
    },
    {
      "epoch": 3.42,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 286500
    },
    {
      "epoch": 3.42,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 287000
    },
    {
      "epoch": 3.43,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 287500
    },
    {
      "epoch": 3.44,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 288000
    },
    {
      "epoch": 3.44,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 288500
    },
    {
      "epoch": 3.45,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 289000
    },
    {
      "epoch": 3.45,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 289500
    },
    {
      "epoch": 3.46,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 290000
    },
    {
      "epoch": 3.46,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 290500
    },
    {
      "epoch": 3.47,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 291000
    },
    {
      "epoch": 3.48,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 291500
    },
    {
      "epoch": 3.48,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 292000
    },
    {
      "epoch": 3.49,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 292500
    },
    {
      "epoch": 3.49,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 293000
    },
    {
      "epoch": 3.5,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 293500
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 294000
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 294500
    },
    {
      "epoch": 3.52,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 295000
    },
    {
      "epoch": 3.52,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 295500
    },
    {
      "epoch": 3.53,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 296000
    },
    {
      "epoch": 3.54,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 296500
    },
    {
      "epoch": 3.54,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 297000
    },
    {
      "epoch": 3.55,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 297500
    },
    {
      "epoch": 3.55,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 298000
    },
    {
      "epoch": 3.56,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 298500
    },
    {
      "epoch": 3.57,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 299000
    },
    {
      "epoch": 3.57,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 299500
    },
    {
      "epoch": 3.58,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 300000
    },
    {
      "epoch": 3.58,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 300500
    },
    {
      "epoch": 3.59,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 301000
    },
    {
      "epoch": 3.6,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 301500
    },
    {
      "epoch": 3.6,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 302000
    },
    {
      "epoch": 3.61,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 302500
    },
    {
      "epoch": 3.61,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 303000
    },
    {
      "epoch": 3.62,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 303500
    },
    {
      "epoch": 3.63,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 304000
    },
    {
      "epoch": 3.63,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 304500
    },
    {
      "epoch": 3.64,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 305000
    },
    {
      "epoch": 3.64,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 305500
    },
    {
      "epoch": 3.65,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 306000
    },
    {
      "epoch": 3.66,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 306500
    },
    {
      "epoch": 3.66,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 307000
    },
    {
      "epoch": 3.67,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 307500
    },
    {
      "epoch": 3.67,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 308000
    },
    {
      "epoch": 3.68,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 308500
    },
    {
      "epoch": 3.69,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 309000
    },
    {
      "epoch": 3.69,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 309500
    },
    {
      "epoch": 3.7,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 310000
    },
    {
      "epoch": 3.7,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 310500
    },
    {
      "epoch": 3.71,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 311000
    },
    {
      "epoch": 3.72,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 311500
    },
    {
      "epoch": 3.72,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 312000
    },
    {
      "epoch": 3.73,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 312500
    },
    {
      "epoch": 3.73,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 313000
    },
    {
      "epoch": 3.74,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 313500
    },
    {
      "epoch": 3.75,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 314000
    },
    {
      "epoch": 3.75,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 314500
    },
    {
      "epoch": 3.76,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 315000
    },
    {
      "epoch": 3.76,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 315500
    },
    {
      "epoch": 3.77,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 316000
    },
    {
      "epoch": 3.78,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 316500
    },
    {
      "epoch": 3.78,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 317000
    },
    {
      "epoch": 3.79,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 317500
    },
    {
      "epoch": 3.79,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 318000
    },
    {
      "epoch": 3.8,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 318500
    },
    {
      "epoch": 3.8,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 319000
    },
    {
      "epoch": 3.81,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 319500
    },
    {
      "epoch": 3.82,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 320000
    },
    {
      "epoch": 3.82,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 320500
    },
    {
      "epoch": 3.83,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 321000
    },
    {
      "epoch": 3.83,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 321500
    },
    {
      "epoch": 3.84,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 322000
    },
    {
      "epoch": 3.85,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 322500
    },
    {
      "epoch": 3.85,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 323000
    },
    {
      "epoch": 3.86,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 323500
    },
    {
      "epoch": 3.86,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 324000
    },
    {
      "epoch": 3.87,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 324500
    },
    {
      "epoch": 3.88,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 325000
    },
    {
      "epoch": 3.88,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 325500
    },
    {
      "epoch": 3.89,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 326000
    },
    {
      "epoch": 3.89,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 326500
    },
    {
      "epoch": 3.9,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 327000
    },
    {
      "epoch": 3.91,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 327500
    },
    {
      "epoch": 3.91,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 328000
    },
    {
      "epoch": 3.92,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 328500
    },
    {
      "epoch": 3.92,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 329000
    },
    {
      "epoch": 3.93,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 329500
    },
    {
      "epoch": 3.94,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 330000
    },
    {
      "epoch": 3.94,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 330500
    },
    {
      "epoch": 3.95,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 331000
    },
    {
      "epoch": 3.95,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 331500
    },
    {
      "epoch": 3.96,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 332000
    },
    {
      "epoch": 3.97,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 332500
    },
    {
      "epoch": 3.97,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 333000
    },
    {
      "epoch": 3.98,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 333500
    },
    {
      "epoch": 3.98,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 334000
    },
    {
      "epoch": 3.99,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 334500
    },
    {
      "epoch": 4.0,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 335000
    },
    {
      "epoch": 4.0,
      "eval_loss": NaN,
      "eval_runtime": 2359.8388,
      "eval_samples_per_second": 17.764,
      "eval_steps_per_second": 8.882,
      "step": 335358
    },
    {
      "epoch": 4.0,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 335500
    },
    {
      "epoch": 4.01,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 336000
    },
    {
      "epoch": 4.01,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 336500
    },
    {
      "epoch": 4.02,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 337000
    },
    {
      "epoch": 4.03,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 337500
    },
    {
      "epoch": 4.03,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 338000
    },
    {
      "epoch": 4.04,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 338500
    },
    {
      "epoch": 4.04,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 339000
    },
    {
      "epoch": 4.05,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 339500
    },
    {
      "epoch": 4.06,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 340000
    },
    {
      "epoch": 4.06,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 340500
    },
    {
      "epoch": 4.07,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 341000
    },
    {
      "epoch": 4.07,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 341500
    },
    {
      "epoch": 4.08,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 342000
    },
    {
      "epoch": 4.09,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 342500
    },
    {
      "epoch": 4.09,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 343000
    },
    {
      "epoch": 4.1,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 343500
    },
    {
      "epoch": 4.1,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 344000
    },
    {
      "epoch": 4.11,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 344500
    },
    {
      "epoch": 4.12,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 345000
    },
    {
      "epoch": 4.12,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 345500
    },
    {
      "epoch": 4.13,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 346000
    },
    {
      "epoch": 4.13,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 346500
    },
    {
      "epoch": 4.14,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 347000
    },
    {
      "epoch": 4.14,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 347500
    },
    {
      "epoch": 4.15,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 348000
    },
    {
      "epoch": 4.16,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 348500
    },
    {
      "epoch": 4.16,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 349000
    },
    {
      "epoch": 4.17,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 349500
    },
    {
      "epoch": 4.17,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 350000
    },
    {
      "epoch": 4.18,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 350500
    },
    {
      "epoch": 4.19,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 351000
    },
    {
      "epoch": 4.19,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 351500
    },
    {
      "epoch": 4.2,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 352000
    },
    {
      "epoch": 4.2,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 352500
    },
    {
      "epoch": 4.21,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 353000
    },
    {
      "epoch": 4.22,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 353500
    },
    {
      "epoch": 4.22,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 354000
    },
    {
      "epoch": 4.23,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 354500
    },
    {
      "epoch": 4.23,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 355000
    },
    {
      "epoch": 4.24,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 355500
    },
    {
      "epoch": 4.25,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 356000
    },
    {
      "epoch": 4.25,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 356500
    },
    {
      "epoch": 4.26,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 357000
    },
    {
      "epoch": 4.26,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 357500
    },
    {
      "epoch": 4.27,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 358000
    },
    {
      "epoch": 4.28,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 358500
    },
    {
      "epoch": 4.28,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 359000
    },
    {
      "epoch": 4.29,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 359500
    },
    {
      "epoch": 4.29,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 360000
    },
    {
      "epoch": 4.3,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 360500
    },
    {
      "epoch": 4.31,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 361000
    },
    {
      "epoch": 4.31,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 361500
    },
    {
      "epoch": 4.32,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 362000
    },
    {
      "epoch": 4.32,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 362500
    },
    {
      "epoch": 4.33,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 363000
    },
    {
      "epoch": 4.34,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 363500
    },
    {
      "epoch": 4.34,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 364000
    },
    {
      "epoch": 4.35,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 364500
    },
    {
      "epoch": 4.35,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 365000
    },
    {
      "epoch": 4.36,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 365500
    },
    {
      "epoch": 4.37,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 366000
    },
    {
      "epoch": 4.37,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 366500
    },
    {
      "epoch": 4.38,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 367000
    },
    {
      "epoch": 4.38,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 367500
    },
    {
      "epoch": 4.39,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 368000
    },
    {
      "epoch": 4.4,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 368500
    },
    {
      "epoch": 4.4,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 369000
    },
    {
      "epoch": 4.41,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 369500
    },
    {
      "epoch": 4.41,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 370000
    },
    {
      "epoch": 4.42,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 370500
    },
    {
      "epoch": 4.43,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 371000
    },
    {
      "epoch": 4.43,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 371500
    },
    {
      "epoch": 4.44,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 372000
    },
    {
      "epoch": 4.44,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 372500
    },
    {
      "epoch": 4.45,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 373000
    },
    {
      "epoch": 4.45,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 373500
    },
    {
      "epoch": 4.46,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 374000
    },
    {
      "epoch": 4.47,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 374500
    },
    {
      "epoch": 4.47,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 375000
    },
    {
      "epoch": 4.48,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 375500
    },
    {
      "epoch": 4.48,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 376000
    },
    {
      "epoch": 4.49,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 376500
    },
    {
      "epoch": 4.5,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 377000
    },
    {
      "epoch": 4.5,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 377500
    },
    {
      "epoch": 4.51,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 378000
    },
    {
      "epoch": 4.51,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 378500
    },
    {
      "epoch": 4.52,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 379000
    },
    {
      "epoch": 4.53,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 379500
    },
    {
      "epoch": 4.53,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 380000
    },
    {
      "epoch": 4.54,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 380500
    },
    {
      "epoch": 4.54,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 381000
    },
    {
      "epoch": 4.55,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 381500
    },
    {
      "epoch": 4.56,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 382000
    },
    {
      "epoch": 4.56,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 382500
    },
    {
      "epoch": 4.57,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 383000
    },
    {
      "epoch": 4.57,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 383500
    },
    {
      "epoch": 4.58,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 384000
    },
    {
      "epoch": 4.59,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 384500
    },
    {
      "epoch": 4.59,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 385000
    },
    {
      "epoch": 4.6,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 385500
    },
    {
      "epoch": 4.6,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 386000
    },
    {
      "epoch": 4.61,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 386500
    },
    {
      "epoch": 4.62,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 387000
    },
    {
      "epoch": 4.62,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 387500
    },
    {
      "epoch": 4.63,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 388000
    },
    {
      "epoch": 4.63,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 388500
    },
    {
      "epoch": 4.64,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 389000
    },
    {
      "epoch": 4.65,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 389500
    },
    {
      "epoch": 4.65,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 390000
    },
    {
      "epoch": 4.66,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 390500
    },
    {
      "epoch": 4.66,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 391000
    },
    {
      "epoch": 4.67,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 391500
    },
    {
      "epoch": 4.68,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 392000
    },
    {
      "epoch": 4.68,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 392500
    },
    {
      "epoch": 4.69,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 393000
    },
    {
      "epoch": 4.69,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 393500
    },
    {
      "epoch": 4.7,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 394000
    },
    {
      "epoch": 4.71,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 394500
    },
    {
      "epoch": 4.71,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 395000
    },
    {
      "epoch": 4.72,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 395500
    },
    {
      "epoch": 4.72,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 396000
    },
    {
      "epoch": 4.73,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 396500
    },
    {
      "epoch": 4.74,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 397000
    },
    {
      "epoch": 4.74,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 397500
    },
    {
      "epoch": 4.75,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 398000
    },
    {
      "epoch": 4.75,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 398500
    },
    {
      "epoch": 4.76,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 399000
    },
    {
      "epoch": 4.77,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 399500
    },
    {
      "epoch": 4.77,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 400000
    },
    {
      "epoch": 4.78,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 400500
    },
    {
      "epoch": 4.78,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 401000
    },
    {
      "epoch": 4.79,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 401500
    },
    {
      "epoch": 4.79,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 402000
    },
    {
      "epoch": 4.8,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 402500
    },
    {
      "epoch": 4.81,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 403000
    },
    {
      "epoch": 4.81,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 403500
    },
    {
      "epoch": 4.82,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 404000
    },
    {
      "epoch": 4.82,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 404500
    },
    {
      "epoch": 4.83,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 405000
    },
    {
      "epoch": 4.84,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 405500
    },
    {
      "epoch": 4.84,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 406000
    },
    {
      "epoch": 4.85,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 406500
    },
    {
      "epoch": 4.85,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 407000
    },
    {
      "epoch": 4.86,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 407500
    },
    {
      "epoch": 4.87,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 408000
    },
    {
      "epoch": 4.87,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 408500
    },
    {
      "epoch": 4.88,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 409000
    },
    {
      "epoch": 4.88,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 409500
    },
    {
      "epoch": 4.89,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 410000
    },
    {
      "epoch": 4.9,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 410500
    },
    {
      "epoch": 4.9,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 411000
    },
    {
      "epoch": 4.91,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 411500
    },
    {
      "epoch": 4.91,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 412000
    },
    {
      "epoch": 4.92,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 412500
    },
    {
      "epoch": 4.93,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 413000
    },
    {
      "epoch": 4.93,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 413500
    },
    {
      "epoch": 4.94,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 414000
    },
    {
      "epoch": 4.94,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 414500
    },
    {
      "epoch": 4.95,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 415000
    },
    {
      "epoch": 4.96,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 415500
    },
    {
      "epoch": 4.96,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 416000
    },
    {
      "epoch": 4.97,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 416500
    },
    {
      "epoch": 4.97,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 417000
    },
    {
      "epoch": 4.98,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 417500
    },
    {
      "epoch": 4.99,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 418000
    },
    {
      "epoch": 4.99,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 418500
    },
    {
      "epoch": 5.0,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 419000
    },
    {
      "epoch": 5.0,
      "eval_loss": NaN,
      "eval_runtime": 2510.0871,
      "eval_samples_per_second": 16.701,
      "eval_steps_per_second": 8.35,
      "step": 419195
    }
  ],
  "logging_steps": 500,
  "max_steps": 419195,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 3.864579585551106e+18,
  "trial_name": null,
  "trial_params": null
}