checkpoint-1040/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.988009592326139,
  "eval_steps": 500,
  "global_step": 1040,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.4,
      "learning_rate": 4.9800000000000004e-05,
      "loss": 0.04,
      "step": 500
    },
    {
      "epoch": 4.8,
      "learning_rate": 7.426068431000882e-07,
      "loss": 0.0369,
      "step": 1000
    }
  ],
  "logging_steps": 500,
  "max_steps": 1040,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 8810671619407872.0,
  "trial_name": null,
  "trial_params": null
}