semantic_router.encoders.fastembed

FastEmbedEncoder Objects

class FastEmbedEncoder(DenseEncoder)

Dense encoder that uses local FastEmbed to embed documents. Supports text only.

Requires the fastembed package which can be installed with pip install 'semantic-router[fastembed]'

Arguments:

• name: The name of the embedding model to use.
• max_length: The maximum length of the input text.
• cache_dir: The directory to cache the embedding model.
• threads: The number of threads to use for the embedding.

__init__

def __init__(**data)

Initialize the FastEmbed encoder.

__call__

def __call__(docs: List[str]) -> List[List[float]]

Embed a list of documents. Supports text only.

Arguments:

• docs (List[str]): The documents to embed.

Raises:

• ValueError: If the embedding fails.

Returns:

List[List[float]]: The vector embeddings of the documents.

acall

async def acall(docs: List[str]) -> List[List[float]]

Embed a list of documents asynchronously. Supports text only.

FastEmbed itself is synchronous, so this offloads the embedding call to a thread to avoid blocking the event loop.

Arguments:

• docs (List[str]): The documents to embed.

Raises:

• ValueError: If the embedding fails.

Returns:

List[List[float]]: The vector embeddings of the documents.