zylon-ai / private-gpt

Interact with your documents using the power of GPT, 100% privately, no data leaks

Home Page:https://docs.privategpt.dev

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Security And Privacy Concerns for input data on latest release

pankajigec26 opened this issue · comments

in the latest release we are getting following error while installing private GPT

LLM model downloaded!
Downloading tokenizer mistralai/Mistral-7B-Instruct-v0.2
Traceback (most recent call last):
  File "/home/epanumk/miniconda3/envs/privategpy/lib/python3.11/site-packages/huggingface_hub/utils/_errors.py", line 270, in hf_raise_for_status
    response.raise_for_status()
  File "/home/epanumk/miniconda3/envs/privategpy/lib/python3.11/site-packages/requests/models.py", line 1021, in raise_for_status
    raise HTTPError(http_error_msg, response=self)
requests.exceptions.HTTPError: 401 Client Error: Unauthorized for url: https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2/resolve/main/config.json

In order to resolve this we need to login to hugging face and generate token . Which was not the case earlier
How can i make sure that my input data is still secure and is not exposed to outside.

This project is open source and you can read if it sends some of your data outside easily. As for the huggingface part, its just required when you download the models from the start just to verify if youre their costumer, this is not a scope from this app but from the huggingface itself