LangSmith supports both categorical and numerical metrics, and you can return either when writing a custom evaluator. For an evaluator result to be logged as a numerical metric, it must returned as:
  • (Python only) an int, float, or bool
  • a dict of the form {"key": "metric_name", "score": int | float | bool}
For an evaluator result to be logged as a categorical metric, it must be returned as:
  • (Python only) a str
  • a dict of the form {"key": "metric_name", "value": str | int | float | bool}
Here are some examples:
Requires langsmith>=0.2.0
def numerical_metric(inputs: dict, outputs: dict, reference_outputs: dict) -> float:
    # Evaluation logic...
    return 0.8
    # Equivalently
    # return {"score": 0.8}
    # Or
    # return {"key": "numerical_metric", "score": 0.8}

def categorical_metric(inputs: dict, outputs: dict, reference_outputs: dict) -> str:
    # Evaluation logic...
    return "english"
    # Equivalently
    # return {"key": "categorical_metric", "score": "english"}
    # Or
    # return {"score": "english"}