Skip to main content

​
Local Embeddings

You can generate embeddings locally using the HuggingFaceEmbeddings class. This utilizes the sentence_transformers library to download the model weights and run them directly on your machine. Let’s load the Hugging Face Embedding class. 
pip install -qU  langchain langchain-huggingface sentence_transformers

from langchain_huggingface.embeddings import HuggingFaceEmbeddings

embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-mpnet-base-v2")

text = "This is a test document."

query_result = embeddings.embed_query(text)

query_result[:3]

[-0.04895168915390968, -0.03986193612217903, -0.021562768146395683]

doc_result = embeddings.embed_documents([text])

​
Hugging Face Inference Endpoints (Serverless API)

If you prefer not to download models locally, you can access embedding models via the Inference Endpoints, which let us use open-source models on Hugging Face’s scalable serverless infrastructure. Ensure you have huggingface_hub installed, which is usually included with langchain-huggingface 
!pip install huggingface_hub
First, we need to get a read-only API key from Hugging Face. 
import os
from getpass import getpass

os.environ["HUGGINGFACEHUB_API_TOKEN"] = getpass()
Now we can use the HuggingFaceEndpointEmbeddings class to run open-source embedding models remotely via the API. 
from langchain_huggingface.embeddings import HuggingFaceEndpointEmbeddings

embeddings = HuggingFaceEndpointEmbeddings(
    model="sentence-transformers/all-MiniLM-L6-v2"
)

text = "This is a test document."

query_result = embeddings.embed_query(text)

query_result[:3]
Edit this page on GitHub or file an issue.
Connect these docs to Claude, VSCode, and more via MCP for real-time answers.