{
  "phase_results": [
    {
      "phase": "multiturn",
      "best_validation_loss": 1.9103374223120133,
      "best_metrics": {
        "loss": 1.9103374223120133,
        "token_count": 13229.0,
        "example_count": 119.0,
        "first_token_correct": 163.0,
        "first_token_count": 455.0,
        "first_token_exact_match": 0.35824175824175825
      },
      "global_step": 36,
      "train_dataset": {
        "example_count": 2260,
        "observation_steps_mean": 3.816371681415929,
        "observation_steps_max": 5,
        "decoder_steps_mean": 3.816371681415929,
        "decoder_steps_max": 5,
        "input_tokens_mean": 163.47079646017698,
        "input_tokens_max": 305,
        "target_tokens_mean": 110.7070796460177,
        "target_tokens_max": 225,
        "dataset_counts": {
          "everyday_conversations": 2260
        }
      },
      "validation_dataset": {
        "example_count": 119,
        "observation_steps_mean": 3.823529411764706,
        "observation_steps_max": 5,
        "decoder_steps_mean": 3.823529411764706,
        "decoder_steps_max": 5,
        "input_tokens_mean": 163.18487394957984,
        "input_tokens_max": 280,
        "target_tokens_mean": 111.16806722689076,
        "target_tokens_max": 189,
        "dataset_counts": {
          "everyday_conversations": 119
        }
      },
      "wandb_enabled": true
    }
  ]
}