[Question]: how ENV HF_ENDPOINT works

[ooooo-create]

Describe your problem

link: #323 (comment)
because I have the same problem to connetct hf-mirror.com.
So I build the docker image from source https://github.com/infiniflow/ragflow?tab=readme-ov-file#%EF%B8%8F-build-from-source
And i delete the line 15 ENV HF_ENDPOINT=https://hf-mirror.com.Then docker compose up , the error is disappear.
I found it will not to download the model, so that when i upload docs and parse it, there is no model in path.
As you say,i clone the deepdoc and add a volumes, like - ~/deepdoc:/ragflow/rag/res/deepdoc
That works!
But I still have the question how ENV HF_ENDPOINT works, why it influence whether to download models? This show it will have a default values in https://huggingface.co/docs/huggingface_hub/v0.13.3/package_reference/environment_variables#hfendpoint

[danielstankw]

@ooooo-create
Can you please tell me where you delete the ENV from ?

[ooooo-create]

@ooooo-create
Can you please tell me where you delete the ENV from ?

yes, I delete it in dockerfile which in the root dir

[KevinHuSh]

This is a mirror site for huggingface scince some of district can't access huggingface.

@ooooo-create Can you please tell me where you delete the ENV from ?

If you can access huggingface, you'd better use docker-compose.yml.