How to make HuggingfaceEmbeddings use local model?

[JimmyJIA-02]

Last week, the function can still work, when I use the HuggingfaceEmbeddings with the relative path of my local model, I can run it and get the embedding result. But just now, when I try to run another program, it no longer works. It will raise an exception saying the network is unreachable. Seems it is trying to access to the model on the hub, I tried to use the absolute path, but still can not solve it.
I wonder how can I use local model when I know it is impossible to access the huggingface hub?

[dosubot[bot]]

🤖

Hello @RedNoseJJN,

Good to see you again! I hope you're doing well. I noticed your recent issue and I'm here to help.

Based on the information you've provided, it seems like you're trying to use a local model with the HuggingFaceEmbeddings function in LangChain. To do this, you should pass the path to your local model as the model_name parameter when instantiating the HuggingFaceEmbeddings class. Here's an example:

from langchain.embeddings import HuggingFaceEmbeddings

# Path to your local model
local_model_path = "/path/to/your/local/model"

# Instantiate the HuggingFaceEmbeddings class
embedding = HuggingFaceEmbeddings(model_name=local_model_path)

# Use the embedding instance
documents = ["foo bar"]
output = embedding.embed_documents(documents)

In this example, replace "/path/to/your/local/model" with the actual path to your local model. The HuggingFaceEmbeddings class will then use this local model for embedding the documents.

Please note that the local model must be compatible with the HuggingFace's Transformers library, as the HuggingFaceEmbeddings class relies on this library for loading the model and performing the embeddings.

If you're still encountering issues, could you please provide the exact error message you're receiving? This will help in diagnosing the problem. Also, please ensure that the huggingface_hub python package is installed, as the HuggingFaceEmbeddings class requires it.

For more information, you can refer to the LangChain repository and the related issue where a similar problem was solved.

I hope this helps! If you have any other questions, feel free to ask.

Sources

• Issue: How can I load text embeddings from a local model?
• libs/langchain/tests/integration_tests/embeddings/test_huggingface.py
• libs/langchain/langchain/embeddings/huggingface_hub.py

---

This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

If you want to continue the conversation, start your reply with @dosu-bot.

[saitejaboggarapu]

remove / at ending, specify like this /home/user/Coding Stuff/AI Models/e5-large-v2

[pickleprat]

What if I have the model stored in a variable?

[lamld203844]

how can i set up locally with os.environ["HF_HOME"]

[saikumar305]

Does it work?

[kyre-99]

remove / at ending, specify like this /home/user/Coding Stuff/AI Models/e5-large-v2

useless

[saikumar305]

I am still facing issues while loading local model into HuggingfaceEmbeddings , can anyone please provide a stepwise guide to how to do it.

Thanks in advance.

[devinbost]

Has this been resolved yet?

[jasonwjli119]

i am experiencing similar issue and getting this error when loading the model from local disk by passing the path as param. Any idea?

HFValidationError: Repo id must be in the form 'repo_name' or 'namespace/repo_name': '../models/bge-small-en-v15'. Use repo_type argument if needed.

from langchain_community.embeddings import HuggingFaceBgeEmbeddings

model_path = "../models/bge-small-en-v15"

huggingface_embeddings = HuggingFaceBgeEmbeddings(
    model_name=model_path,
    model_kwargs={'device':'cpu'}, 
    encode_kwargs={'normalize_embeddings': True},  
)