_evaluate_strings() — langchain Function Reference

Source Code

libs/langchain/langchain_classic/evaluation/criteria/eval_chain.py lines 404–453

    def _evaluate_strings(
        self,
        *,
        prediction: str,
        reference: str | None = None,
        input: str | None = None,
        callbacks: Callbacks = None,
        tags: list[str] | None = None,
        metadata: dict[str, Any] | None = None,
        include_run_info: bool = False,
        **kwargs: Any,
    ) -> dict:
        """Evaluate a prediction against the criteria.

        Args:
            prediction: The predicted text to evaluate.
            reference: The reference text to compare against. This is required if
                `requires_reference` is `True`.
            input: The input text used to generate the prediction.
            callbacks: The callbacks to use.
            tags: The tags to apply.
            metadata: The metadata to use.
            include_run_info: Whether to include run info in the output.
            **kwargs: Additional keyword arguments to pass to the `LLMChain` `__call__`
                method.

        Returns:
            The evaluation results.

        Examples:
            >>> from langchain_openai import OpenAI
            >>> from langchain_classic.evaluation.criteria import CriteriaEvalChain
            >>> model = OpenAI()
            >>> criteria = "conciseness"
            >>> chain = CriteriaEvalChain.from_llm(llm=model, criteria=criteria)
            >>> chain.evaluate_strings(
                    prediction="The answer is 42.",
                    reference="42",
                    input="What is the answer to life, the universe, and everything?",
                )
        """
        input_ = self._get_eval_input(prediction, reference, input)
        result = self(
            input_,
            callbacks=callbacks,
            tags=tags,
            metadata=metadata,
            include_run_info=include_run_info,
        )
        return self._prepare_output(result)