trainer_state.json

{
  "best_metric": 0.645,
  "best_model_checkpoint": "alea/checkpoint-8",
  "epoch": 0.9824561403508771,
  "eval_steps": 500,
  "global_step": 8,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.98,
      "eval_accuracy": 0.645,
      "eval_loss": 1.2876272201538086,
      "eval_runtime": 2.2928,
      "eval_samples_per_second": 87.229,
      "eval_steps_per_second": 3.053,
      "step": 8
    },
    {
      "epoch": 0.98,
      "step": 8,
      "total_flos": 1.3886937921591706e+17,
      "train_loss": 1.3266394138336182,
      "train_runtime": 44.5183,
      "train_samples_per_second": 40.433,
      "train_steps_per_second": 0.18
    }
  ],
  "logging_steps": 10,
  "max_steps": 8,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 1.3886937921591706e+17,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}