Anyone have experience implementing vector similarity search in v1 (either via Lucene or hnswlib)? Is there a recommended way to handle this?
I know @felix hacked on such a thing at one point but I'm not sure if the code is available somewhere
Iād be interested to know the same answer for v2, or is that no longer possible with the new architecture?
It's true - but not open source sorry. We used NGT and fast-text via Coffi, and registered a secondary index in XT