checkpoint-72/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.84,
  "eval_steps": 500,
  "global_step": 72,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.55,
      "learning_rate": 0.000134375,
      "loss": 1.1486,
      "step": 29
    },
    {
      "epoch": 3.09,
      "learning_rate": 4.375e-05,
      "loss": 0.93,
      "step": 58
    }
  ],
  "logging_steps": 29,
  "max_steps": 72,
  "num_train_epochs": 4,
  "save_steps": 500,
  "total_flos": 1.0084987501648282e+17,
  "trial_name": null,
  "trial_params": null
}