revert incorrect bm25 claim by qdrant

superlinked / VectorHub

VectorHub is a free, open-source learning website for people (software developers to senior ML architects) interested in adding vector retrieval to their ML stack.

https://superlinked.com/vectorhub/

Other

413 stars 97 forks source link

revert incorrect bm25 claim by qdrant #424

Closed dhruv-anand-aintech closed 1 month ago

dhruv-anand-aintech commented 1 month ago

labels: 'vdb comparison'

Describe your changes

Link to relevant discussion

Checklist before requesting a review

[x] I have followed the contribution guidelines
[x] I have provided a reference for each attribute for which I'm adding a "support" claim.

dhruv-anand-aintech commented 1 month ago

FYI @Anush008. There continues to be a sparse vectors attribute separate from BM25 which has been marked as supported from before.

To have the BM25 attribute marked as supported, it'll have to be implemented in-DB. People with existing BM25+dense hybrid search setups who are trying to migrate to Qdrant should not get a surprise later in the process when they find that Qdrant doesn't have a standard BM25 supported in-DB, while they've claimed the same on this comparison page.

Anush008 commented 1 month ago

Thanks. I thought the field meant support for storing BM25 vectors.