Hugging Face

Parameters for HuggingfaceCustomEmbedder, which embeds documents through Hugging Face’s InferenceClient.

Hugging Face Embedder embeds documents using any embedding model hosted on Hugging Face’s Inference API.

Parameters

Parameter	Type	Description	Default
id	str	The model ID to use for embeddings	"intfloat/multilingual-e5-large"
dimensions	Optional[int]	Output dimensions of the embedding	1536
api_key	Optional[str]	Huggingface API key	Environment variable HUGGINGFACE_API_KEY
client_params	Optional[Dict[str, Any]]	Additional parameters for client initialization	None
huggingface_client	Optional[InferenceClient]	Pre-configured Huggingface client	None
async_client	Optional[AsyncInferenceClient]	Pre-configured async Huggingface client	None
enable_batch	bool	Has no effect. Batch embedding is unsupported and the flag is reset to False	False
batch_size	int	Number of texts per batch. Unused by this embedder	100