Hi, I'm currently trying to implement a feature called "hybrid search" inside of my application. It's based on returning query results from multiple databases and later scoring results together from multiple sources. To score them I want to use cross-encoder models ie. https://huggingface.co/cross-encoder/ms-marco-MiniLM-L-6-v2.
I'm trying to understand if Bumblebee currently supports models like this and if so how to use it for that.
If it does I'll be very happy to write some documentation for that after getting some hints. If not it would be a very cool feature to handle those types of operations :)
Hi, I'm currently trying to implement a feature called "hybrid search" inside of my application. It's based on returning query results from multiple databases and later scoring results together from multiple sources. To score them I want to use cross-encoder models ie. https://huggingface.co/cross-encoder/ms-marco-MiniLM-L-6-v2.
I'm trying to understand if Bumblebee currently supports models like this and if so how to use it for that.
If it does I'll be very happy to write some documentation for that after getting some hints. If not it would be a very cool feature to handle those types of operations :)