trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 3,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "learning_rate": 2e-05,
      "loss": 1.0501,
      "step": 1
    },
    {
      "epoch": 2.0,
      "learning_rate": 1e-05,
      "loss": 1.0767,
      "step": 2
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0,
      "loss": 0.7536,
      "step": 3
    },
    {
      "epoch": 3.0,
      "step": 3,
      "total_flos": 2327628803473408.0,
      "train_loss": 0.9601547519365946,
      "train_runtime": 1446.9004,
      "train_samples_per_second": 0.531,
      "train_steps_per_second": 0.002
    }
  ],
  "max_steps": 3,
  "num_train_epochs": 3,
  "total_flos": 2327628803473408.0,
  "trial_name": null,
  "trial_params": null
}