run-13/checkpoint-500/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.6737967914438503,
  "eval_steps": 500,
  "global_step": 500,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7670130729675293,
      "eval_loss": 0.4646908640861511,
      "eval_runtime": 8.9608,
      "eval_samples_per_second": 332.896,
      "eval_steps_per_second": 20.869,
      "step": 187
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8015420436859131,
      "eval_loss": 0.4173341989517212,
      "eval_runtime": 9.0086,
      "eval_samples_per_second": 331.129,
      "eval_steps_per_second": 20.758,
      "step": 374
    },
    {
      "epoch": 2.6737967914438503,
      "grad_norm": 10.940384864807129,
      "learning_rate": 1.0504443448574759e-05,
      "loss": 0.4462,
      "step": 500
    }
  ],
  "logging_steps": 500,
  "max_steps": 1870,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 5510487892159476.0,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 1.4338181933456057e-05,
    "per_device_train_batch_size": 64
  }
}