Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[BUG]: qdrant/qdrant:latest (1.9.0) causes embeddings to be ignored #1220

Open
shavingtonpitsos opened this issue Apr 29, 2024 · 2 comments
Open
Assignees
Labels
investigating Core team or maintainer will or is currently looking into this issue possible bug Bug was reported but is not confirmed or is unable to be replicated.

Comments

@shavingtonpitsos
Copy link

How are you running AnythingLLM?

Docker (local)

What happened?

when using the latest version of qdrant locally I would add embeddings to a workspace, be notified that the embeddings had been added successfully and then check the workspace settings to find the vector count for that workspace to be 0.

This seems to be caused by the latest (1.9.0) version of qudrant on docker hub no longer providing a vectors_count. For some reason this seems to mess with the anythingllm code.

The more important symptom is that all queries to this index via chat always return an empty sources dict, even with no minimum distance.

This may have been related to what a commenter was saying in #516

I don't know if it is just for me but the number of vectors in the Settings tab for all of my workspaces are showing zero even after leaving and reloading the Setting dialog. I have about 13 documents in each workspace and they are successfully referenced in chats but it always says 0 vectors in Settings for all of my workspaces.

Are there known steps to reproduce?

  1. clone the latest anythingllm
  2. follow the local docker set up instructions in docker/HOW_TO_USE_DOCKER.md
  3. set up qdrant locally by following the qdrant quickstart documentation i.e.
docker run -p 6333:6333 -p 6334:6334 \
    -v $(pwd)/qdrant_storage:/qdrant/storage:z \
    qdrant/qdrant
  1. look at the terminal output to confirm that we are now using the 1.9.0 version of the qdrant docker image
  2. Set up as usual choosing any llm provider
  3. alter the vectordb to the qdrant, passing in the local url (no api key needed unless you set one in step 3)
  4. create a new workspace, add some embeddings to id
  5. check the settings for that workspace, vector count will be 0
  6. try some queries, there will be no sources

Now just change to qdrant/qdrant:v1.8.4, create a new index and everything will work as normal.

@shavingtonpitsos shavingtonpitsos added the possible bug Bug was reported but is not confirmed or is unable to be replicated. label Apr 29, 2024
@shavingtonpitsos
Copy link
Author

I'm adding this as a bug since qdrant is supported and anyone following the standard qdrant setup procedure will now encounter errors. That said this could be a bug in the latest qdrant, I'm unsure where the responsibility lies exactly.

@sliceofbytes
Copy link

Thanks for posting this, was trying to figure out what broke my setup.

@timothycarambat timothycarambat self-assigned this May 10, 2024
@timothycarambat timothycarambat added the investigating Core team or maintainer will or is currently looking into this issue label May 10, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
investigating Core team or maintainer will or is currently looking into this issue possible bug Bug was reported but is not confirmed or is unable to be replicated.
Projects
None yet
Development

No branches or pull requests

3 participants