{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "eval_steps": 500,
  "global_step": 15690,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.32,
      "learning_rate": 0.00029699999999999996,
      "loss": 16.5786,
      "step": 500
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0002902238314680711,
      "loss": 4.2554,
      "step": 1000
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00028034891375905196,
      "loss": 2.4185,
      "step": 1500
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0002704739960500329,
      "loss": 1.9034,
      "step": 2000
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002605990783410138,
      "loss": 1.7194,
      "step": 2500
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0002507241606319947,
      "loss": 1.6073,
      "step": 3000
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.00024084924292297564,
      "loss": 1.4389,
      "step": 3500
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.00023097432521395654,
      "loss": 1.3822,
      "step": 4000
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.00022109940750493742,
      "loss": 1.3337,
      "step": 4500
    },
    {
      "epoch": 3.19,
      "learning_rate": 0.00021122448979591835,
      "loss": 1.2353,
      "step": 5000
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.00020134957208689926,
      "loss": 1.189,
      "step": 5500
    },
    {
      "epoch": 3.82,
      "learning_rate": 0.00019147465437788016,
      "loss": 1.1666,
      "step": 6000
    },
    {
      "epoch": 4.14,
      "learning_rate": 0.00018161948650427913,
      "loss": 1.1056,
      "step": 6500
    },
    {
      "epoch": 4.46,
      "learning_rate": 0.00017174456879526004,
      "loss": 1.0524,
      "step": 7000
    },
    {
      "epoch": 4.78,
      "learning_rate": 0.00016186965108624092,
      "loss": 1.0451,
      "step": 7500
    },
    {
      "epoch": 5.1,
      "learning_rate": 0.0001520144832126399,
      "loss": 1.0111,
      "step": 8000
    },
    {
      "epoch": 5.42,
      "learning_rate": 0.0001421395655036208,
      "loss": 0.9569,
      "step": 8500
    },
    {
      "epoch": 5.74,
      "learning_rate": 0.0001322646477946017,
      "loss": 0.9371,
      "step": 9000
    },
    {
      "epoch": 6.05,
      "learning_rate": 0.0001223897300855826,
      "loss": 0.9277,
      "step": 9500
    },
    {
      "epoch": 6.37,
      "learning_rate": 0.00011251481237656352,
      "loss": 0.8625,
      "step": 10000
    },
    {
      "epoch": 6.69,
      "learning_rate": 0.00010265964450296247,
      "loss": 0.872,
      "step": 10500
    },
    {
      "epoch": 7.01,
      "learning_rate": 9.278472679394337e-05,
      "loss": 0.8362,
      "step": 11000
    },
    {
      "epoch": 7.33,
      "learning_rate": 8.290980908492429e-05,
      "loss": 0.7923,
      "step": 11500
    },
    {
      "epoch": 7.65,
      "learning_rate": 7.30348913759052e-05,
      "loss": 0.7942,
      "step": 12000
    },
    {
      "epoch": 7.97,
      "learning_rate": 6.31599736668861e-05,
      "loss": 0.776,
      "step": 12500
    },
    {
      "epoch": 8.29,
      "learning_rate": 5.330480579328505e-05,
      "loss": 0.7308,
      "step": 13000
    },
    {
      "epoch": 8.6,
      "learning_rate": 4.3429888084265964e-05,
      "loss": 0.7219,
      "step": 13500
    },
    {
      "epoch": 8.92,
      "learning_rate": 3.355497037524687e-05,
      "loss": 0.735,
      "step": 14000
    },
    {
      "epoch": 9.24,
      "learning_rate": 2.368005266622778e-05,
      "loss": 0.6917,
      "step": 14500
    },
    {
      "epoch": 9.56,
      "learning_rate": 1.380513495720869e-05,
      "loss": 0.6928,
      "step": 15000
    },
    {
      "epoch": 9.88,
      "learning_rate": 3.930217248189598e-06,
      "loss": 0.6785,
      "step": 15500
    }
  ],
  "logging_steps": 500,
  "max_steps": 15690,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 8.019596820665179e+19,
  "trial_name": null,
  "trial_params": null
}