Find answers from the community

Updated 3 months ago

@Logan M I need your opinion. Currently

I need your opinion. Currently there's no Callback Manager implemented in the OpenAIMultiModal Chat, which means I can't use chat, I can implement it if you think it's worth it and would fix my issue. Query doesn't work because it tries to return only the relevant images, which isn't what I want, and completion doesn't make a ton of sense though I could maybe make it work. Should I just write my own query engine logic? Cause really I just want it to take X clusters of images, and based on the metadata of the image nodes or the images themselves sort them in order of relevancy to the user as best it can, regardless of whether or not it thinks something is relevant
L
Z
8 comments
completion actually just calls chat() with a single user message
what you are doing does sound super custom though
so I should do a custom query engine?
that doesn't seem right
I'm not sure man, The callback thing is a a TODO for sure, but beyond that, I don't know
I have the embeddings and metadata for each image
can I possibly give anything either or both to rerank?
it doesn't have to be definitive, it will definitely be subjective, just curious if there's anything that makes sense
Add a reply
Sign up and join the conversation on Discord