trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.04060913705583756,
  "eval_steps": 500,
  "global_step": 2,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 0.0001,
      "loss": 1.5897,
      "step": 1
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0,
      "loss": 1.7194,
      "step": 2
    },
    {
      "epoch": 0.04,
      "step": 2,
      "total_flos": 374270737711104.0,
      "train_loss": 1.6545246243476868,
      "train_runtime": 51.5131,
      "train_samples_per_second": 0.155,
      "train_steps_per_second": 0.039
    }
  ],
  "logging_steps": 1,
  "max_steps": 2,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 374270737711104.0,
  "trial_name": null,
  "trial_params": null
}