Distance metrics in vector search

hi @Ashish_Kumar !!

Welcome to our community

the distance metric, as you requested this metadata to be returned, is the distance calculation between your query and your stored objects.

By default, when you creation a collection, the default metrics calculation used is cosine, as stated here:

Now, when you do a bm25 search, you get a score, instead of a distance.

The same goes to hybrid, where a similarity search (near_vector or near_text) is performed alongside a bm25, and those are fused.

Let me know if that helps!

THanks!