trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9998213328568876,
  "eval_steps": 500,
  "global_step": 1399,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.36,
      "learning_rate": 0.000642601858470336,
      "loss": 0.9263,
      "step": 500
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0002852037169406719,
      "loss": 0.9004,
      "step": 1000
    },
    {
      "epoch": 1.0,
      "step": 1399,
      "total_flos": 7.997027865801523e+16,
      "train_loss": 0.9312247221089841,
      "train_runtime": 1289.1473,
      "train_samples_per_second": 34.73,
      "train_steps_per_second": 1.085
    }
  ],
  "logging_steps": 500,
  "max_steps": 1399,
  "num_train_epochs": 1,
  "save_steps": -1399,
  "total_flos": 7.997027865801523e+16,
  "trial_name": null,
  "trial_params": null
}