trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 1730,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.89,
      "learning_rate": 3.554913294797688e-05,
      "loss": 2.962,
      "step": 500
    },
    {
      "epoch": 5.78,
      "learning_rate": 2.1098265895953757e-05,
      "loss": 2.596,
      "step": 1000
    },
    {
      "epoch": 8.67,
      "learning_rate": 6.647398843930635e-06,
      "loss": 2.5038,
      "step": 1500
    },
    {
      "epoch": 10.0,
      "step": 1730,
      "total_flos": 2.648315673797714e+17,
      "train_loss": 2.6604936368203576,
      "train_runtime": 4208.1526,
      "train_samples_per_second": 26.244,
      "train_steps_per_second": 0.411
    }
  ],
  "max_steps": 1730,
  "num_train_epochs": 10,
  "total_flos": 2.648315673797714e+17,
  "trial_name": null,
  "trial_params": null
}