run-3/checkpoint-12/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.6,
  "eval_steps": 10,
  "global_step": 12,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.5,
      "grad_norm": 228748.703125,
      "learning_rate": 0.009175618433309092,
      "loss": 5.0308,
      "step": 10
    },
    {
      "epoch": 0.5,
      "eval_loss": 4.4208502769470215,
      "eval_runtime": 16.6604,
      "eval_samples_per_second": 600.226,
      "eval_steps_per_second": 1.14,
      "step": 10
    }
  ],
  "logging_steps": 10,
  "max_steps": 20,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 6,
  "total_flos": 83650040064000.0,
  "train_batch_size": 544,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 0.009175618433309092
  }
}