_EmbeddingDistanceChainMixin Class — langchain Architecture

Source Code

libs/langchain/langchain_classic/evaluation/embedding_distance/base.py lines 93–343

class _EmbeddingDistanceChainMixin(Chain):
    """Shared functionality for embedding distance evaluators.

    Attributes:
        embeddings: The embedding objects to vectorize the outputs.
        distance_metric: The distance metric to use for comparing the embeddings.
    """

    embeddings: Embeddings = Field(default_factory=_embedding_factory)
    distance_metric: EmbeddingDistance = Field(default=EmbeddingDistance.COSINE)

    @pre_init
    def _validate_tiktoken_installed(cls, values: dict[str, Any]) -> dict[str, Any]:
        """Validate that the TikTok library is installed.

        Args:
            values: The values to validate.

        Returns:
            The validated values.
        """
        embeddings = values.get("embeddings")
        types_ = []
        try:
            from langchain_openai import OpenAIEmbeddings

            types_.append(OpenAIEmbeddings)
        except ImportError:
            pass

        try:
            from langchain_community.embeddings.openai import (
                OpenAIEmbeddings,
            )

            types_.append(OpenAIEmbeddings)
        except ImportError:
            pass

        if not types_:
            msg = (
                "Could not import OpenAIEmbeddings. Please install the "
                "OpenAIEmbeddings package using `pip install langchain-openai`."
            )
            raise ImportError(msg)

        if isinstance(embeddings, tuple(types_)):
            try:
                import tiktoken  # noqa: F401
            except ImportError as e:
                msg = (
                    "The tiktoken library is required to use the default "
                    "OpenAI embeddings with embedding distance evaluators."
                    " Please either manually select a different Embeddings object"
                    " or install tiktoken using `pip install tiktoken`."
                )
                raise ImportError(msg) from e
        return values

    model_config = ConfigDict(
        arbitrary_types_allowed=True,
    )

    @property
    def output_keys(self) -> list[str]:
        """Return the output keys of the chain.

        Returns:
            The output keys.
        """
        return ["score"]

    def _prepare_output(self, result: dict) -> dict:
        parsed = {"score": result["score"]}
        if RUN_KEY in result:
            parsed[RUN_KEY] = result[RUN_KEY]
        return parsed

    def _get_metric(self, metric: EmbeddingDistance) -> Any:
        """Get the metric function for the given metric name.