Weaviate

Llama index provides specific integrations for weaviate, you can see a small demo here

https://github.com/jerryjliu/llama_index/blob/main/examples/vector_indices/WeaviateIndexDemo.ipynb

The operation is mostly the same as the simple vector index

If there are specific weaviate functions you wanted to use, PRs are welcome! 💪

(I personally haven't used weaviate yet lol)

👍 Thanks. I think that at this point anyone who configures Weaviate will most likely use it directly

Any reason why? Anything specific llama index could provide or take advantage of?

@Logan M In Weaviate you can create classes of objects, say a document with meta data fields such as filename, doc type, etc, and a paragraph class with its own fields and embedding vector associated.
Gptindex only allows you to query the whole index, instead of only querying paragraphs with specific data or those belong to a certain document.

Thanks @yoelk thats pretty cool!

Maybe I can look into adding this functionality at some point. The support is nearly there in llama index, as we have keyword filtering, so why not type filtering right?

Yes, but it's more than keyword filtering as a class may have many different fields and one would like to have the ability to filter based on each one of them and with different operators.
See this
https://weaviate.io/developers/weaviate/tutorials/query
And this
https://weaviate.io/developers/weaviate/api/graphql/filters

Right, different than keyword filtering, but I just meant the code mostly has the infrastructure to do this (which means it's not too hard to add) :dotsCATJAM:

That's great!
Btw, also when adding new objects you can add objects belonging to different classes with their relevant fields which I think is also not supported today