You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We are working on this feature now in an Agent model. It allows for multiple retrievers and then post processing those results. I hope to have it released in the 1.0 M1 next week, if not by a M2 release. Either way when there is a branch or PR to share, I’ll post back here.
@markpollack Hi, I have implemented a hybrid search for PG that uses full-text retrieval and vector similarity. If the community needs me to implement it, I can submit it.
Hello everyone, I have also developed this feature that enables the user to define within the SearchRequest whether to use similarity search (vector-based, enabled by default), full-text search, or both (hybrid search). Additionally, it provides the possibility of using reranking approaches.
This implementation is applied to the vector store Azure SearchAI, and the implementation done by @impactCn can be seamlessly integrated as well.
During search, hybrid search, similar search and keyword search will be supported and submitted to LLM.
I am willing to submit this feature. Does the community have any better suggestions for my feature?
The text was updated successfully, but these errors were encountered: