Skip to main content

MGTEEmbeddingFunction

MGTEEmbeddingFunction is a class in pymilvus that handles encoding text into embeddings using MGTE embedding models to support embedding retrieval in Milvus.

pymilvus.model.hybrid.MGTEEmbeddingFunction

Constructor​

Constructs a MGTEEmbeddingFunction for common use cases.

MGTEEmbeddingFunction(
    model_name: str = "Alibaba-NLP/gte-multilingual-base",
    batch_size: int = 16,
    device: str = "",
    normalize_embeddings: bool = True,
    dimensions: Optional[int] = None,
    use_fp16: bool = False,
    return_dense: bool = True,
    return_sparse: bool = True,
    **kwargs
)

PARAMETERS:

  • model_name (string)

    The name of the GTE embedding model to use for encoding. The value defaults to Alibaba-NLP/gte-multilingual-base. For more information, refer to Models.

  • batch_size (int)

    The batch size to use for encoding.

  • device (string)

    The device to use for the model.

  • normalize_embeddings (bool)

    Whether to normalize the dense embeddings.

  • dimensions (int)

    The number of dimensions for the dense embeddings. If not provided, it will use the model's default hidden size.

  • use_fp16 (bool)

    Whether to use 16-bit floating point precision.

  • return_dense (bool)

    Whether to return dense embeddings.

  • return_sparse (bool)

    Whether to return sparse embeddings.

  • kwargs

    Allows additional keyword arguments to be passed to the model initialization.

Examples​

from pymilvus.model.hybrid import MGTEEmbeddingFunction

ef = MGTEEmbeddingFunction(
    model_name="Alibaba-NLP/gte-multilingual-base",
)