Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.trychroma.com/llms.txt

Use this file to discover all available pages before exploring further.

Chroma provides wrappers for both dense and sparse embedding models from Hugging Face.

​
Dense Embeddings

Chroma provides a convenient wrapper around HuggingFace’s embedding API. This embedding function runs remotely on HuggingFace’s servers, and requires an API key. You can get an API key by signing up for an account at HuggingFace. 
import chromadb.utils.embedding_functions as embedding_functions
huggingface_ef = embedding_functions.HuggingFaceEmbeddingFunction(
    api_key="YOUR_API_KEY",
    model_name="sentence-transformers/all-MiniLM-L6-v2"
)
You can pass in an optional model_name argument, which lets you choose which HuggingFace model to use. By default, Chroma uses sentence-transformers/all-MiniLM-L6-v2. You can see a list of all available models here.

​
Sparse Embeddings

Chroma also supports sparse embedding models from Hugging Face using HuggingFaceSparseEmbeddingFunction. This embedding function requires the sentence_transformers package, which you can install with pip install sentence_transformers. 
from chromadb.utils.embedding_functions import HuggingFaceSparseEmbeddingFunction

ef = HuggingFaceSparseEmbeddingFunction(
    model_name="BAAI/bge-m3",
    device="cpu"
)

texts = ["Hello, world!", "How are you?"]
sparse_embeddings = ef(texts)