trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 1999,
  "global_step": 1999,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.25,
      "learning_rate": 3.749374687343672e-05,
      "loss": 0.446,
      "step": 500
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.4987493746873438e-05,
      "loss": 0.2346,
      "step": 1000
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.2481240620310155e-05,
      "loss": 0.1862,
      "step": 1500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9313301628476726,
      "eval_loss": 0.2507786452770233,
      "eval_runtime": 64.166,
      "eval_samples_per_second": 216.283,
      "eval_steps_per_second": 13.527,
      "step": 1999
    }
  ],
  "logging_steps": 500,
  "max_steps": 1999,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 8315303499358752.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}