checkpoint-1000/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.2531328320802004,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.63,
      "learning_rate": 1.5822890559732666e-05,
      "loss": 3.1176,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_loss": 2.115739107131958,
      "eval_runtime": 3.995,
      "eval_samples_per_second": 177.722,
      "eval_steps_per_second": 11.264,
      "step": 798
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.1645781119465331e-05,
      "loss": 2.5326,
      "step": 1000
    }
  ],
  "max_steps": 2394,
  "num_train_epochs": 3,
  "total_flos": 1057367194337280.0,
  "trial_name": null,
  "trial_params": null
}