trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9995194617972127,
  "global_step": 1040,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.96,
      "learning_rate": 5.230769230769232e-06,
      "loss": 0.2065,
      "step": 500
    },
    {
      "epoch": 1.92,
      "learning_rate": 4.2307692307692315e-07,
      "loss": 0.0805,
      "step": 1000
    },
    {
      "epoch": 2.0,
      "step": 1040,
      "total_flos": 332043964444800.0,
      "train_loss": 0.1404363962320181,
      "train_runtime": 264.5413,
      "train_samples_per_second": 62.932,
      "train_steps_per_second": 3.931
    }
  ],
  "max_steps": 1040,
  "num_train_epochs": 2,
  "total_flos": 332043964444800.0,
  "trial_name": null,
  "trial_params": null
}