SpladeEmbeddingFunction

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

pymilvus.model.sparse.SpladeEmbeddingFunction

Constructor

Constructs a SpladeEmbeddingFunction for common use cases.

SpladeEmbeddingFunction(
    model_name: str = "naver/splade-cocondenser-ensembledistil",
    batch_size: int = 32,
    query_instruction: str = "",
    doc_instruction: str = "",
    device: Optional[str] = "cpu",
    k_tokens_query: Optional[int] = None,
    k_tokens_document: Optional[int] = None,
    **kwargs,
)

PARAMETERS:

• model_name (string) -

  The name of the SPLADE model to use for encoding. Valid options are naver/splade-cocondenser-ensembledistil (default), naver/splade_v2_max, naver/splade_v2_distil, and naver/splade-cocondenser-selfdistil. For more information, refer to Play with models.

• batch_size (int) -

  The batch size used for the computation.

• query_instruction (string) -

  The query to use for encoding.

• doc_instruction (string) -

  The document to use for encoding.

• device (string) -

  The device to use, with cpu for the CPU and cuda:n for the nth GPU device.

• k_tokens_query (int) -

  The number of top tokens to use for query encodings. If not specified, it will use all non-zero tokens.

• k_tokens_document (int) -

  The number of top tokens to use for document encodings. If not specified, it will use all non-zero tokens.

• ****kwargs**

  Allows additional keyword arguments to be passed to the model initialization. For more information, refer to AutoModelForMaskedLM.

Examples

from pymilvus import model

splade_ef = model.sparse.SpladeEmbeddingFunction(
    model_name="naver/splade-cocondenser-selfdistil", 
    device="cpu"
)