4 ms·
No, this isn’t at all how it works. This is not a trivial problem when dealing with vector similarity search, which Postgres does not have.
by gk1 5y ago
No, this isn’t at all how it works.
This is not a trivial problem when dealing with vector similarity search, which Postgres does not have.
- svcrunch 5y agoVery nice writeup. I have an idea how this might be implemented. Since you rightfully want to keep the details private, let me just ask: How do you handle the case where the probed clusters contain no elements satisfying the filter criteria? Is the probe extended to additional clusters, or terminated with no results?