Error in L2 Code Example

Short Course Q&A Building Multimodal Search and RAGs
1

image
image995×596 54.1 KB

^Tried a few times trying to add image files into Weaviate

The next cells on trying to add video files also Weaviate also fails

Thanks!

2

This issue usually happens when the Weaviate instance isn’t configured with the correct media modules enabled. The default Weaviate setup in the course only supports text unless you explicitly enable the multi2vec-clip (for images) or multi2vec-bind modules. Without those modules running, any attempt to import image or video files will fail — even if the code itself is correct.

A few things to check:

  1. Confirm your Weaviate container is running with multimodal support
    In your docker-compose.yaml, you should see something like:

    ENABLE_MODULES: multi2vec-clip

    If it isn’t there, images and videos cannot be processed.

  2. Restart Weaviate after enabling modules
    Module changes only take effect on a clean restart:

    docker compose down
docker compose up -d --build

  3. Verify module availability
    Run:

    client.get_meta()

    and check that the module list includes multi2vec-clip.

  4. Ensure the file paths are valid
    Especially in notebooks, relative paths often break — use absolute paths or confirm with:

    import os
os.path.exists("path/to/file.jpg")

Once the module is enabled and Weaviate restarts cleanly, the image and video insert examples should run exactly as shown in the course.

If you want, share your docker-compose or the client.get_meta() output — happy to help you pinpoint the exact issue.

3

It seems that Weaviate client also updates from v3 to v4. Could we have the lab updated with the new api as well?

multi2vec_google is the new module i believe

Thank you.