No sentence-transformers model found with name sentence-transformers/all-MiniLM-L6-v2

[ruschestor]

Hello,
I am using Open WebUI in an environment isolated from the Internet.

By default, Open WebUI includes the "sentence-transformers/all-MiniLM-L6-v2" embedding model in its Docker image. This allows me to run and use the application without requiring Internet access.

However, after installing the confluence_search tool, I encountered the following error: "Error loading embedding model: We couldn't connect to https://huggingface.co to load the files, and couldn't find the...".
Additionally, the Docker logs show this message: "No sentence-transformers model found with name sentence-transformers/all-MiniLM-L6-v2"
The tool’s configuration includes an environment variable called "Embedding Model Save Path" with a default value of "/tmp/cache/sentence_transformers".

I have two questions:

1. Why doesn’t the confluence_search tool use the embedding model already included in Open WebUI?
2. Is it possible to configure the confluence_search tool to use the "sentence-transformers/all-MiniLM-L6-v2" model from the Open WebUI application files?

Thank you in advance for your help.

[RomainNeup]

Hello @ruschestor,

I appreciate your concerns about the confluence_search tool. Thank you for bringing this to my attention.

Why doesn’t the confluence_search tool use the embedding model already included in Open WebUI?

I'm not sure if the tools have access to all the OWUI file system, but I was also unaware of an environment variable called "Embedding Model Save Path". I'll investigate further.

Is it possible to configure the confluence_search tool to use the "sentence-transformers/all-MiniLM-L6-v2" model from the Open WebUI application files?

From now on, I won't be able to do this, but I can work on implementing a fix for you. Additionally, using this model will improve performance by eliminating the need for re-downloading the model.

[thejerrod]

This would be a nice improvement to have particularly in environments where you have no access to the underlying system outside of the UI, and are unable to download additional embedding models outside of the included model.