trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 2280,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "learning_rate": 2.7777777777777776e-07,
      "loss": 2.5056,
      "step": 1140
    },
    {
      "epoch": 1.0,
      "eval_loss": 2.3534486293792725,
      "eval_runtime": 7.2074,
      "eval_samples_per_second": 138.607,
      "eval_steps_per_second": 1.526,
      "step": 1140
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0,
      "loss": 2.4707,
      "step": 2280
    },
    {
      "epoch": 2.0,
      "eval_loss": 2.343771457672119,
      "eval_runtime": 1.6188,
      "eval_samples_per_second": 617.117,
      "eval_steps_per_second": 6.795,
      "step": 2280
    }
  ],
  "max_steps": 2280,
  "num_train_epochs": 2,
  "total_flos": 7142940278784000.0,
  "trial_name": null,
  "trial_params": null
}