benwtrent commented on issue #13447: URL: https://github.com/apache/lucene/issues/13447#issuecomment-3004669727
> Interestingly, it looks like your intuition roughly aligns with my suggestion if using topK=100 LOL, your intuition is better than mine! It took me many experiments to arrive at that estimation! Yeah, I think the `expectedVisitedNodes` can be used towards this purpose. We pick a "standard k" (or allow it to be configurable?? maybe that is too many knobs.), apply it and determine if hnsw is worth it. I would expect this to improve indexing throughput without too much of a hit at query time (especially for highly quantized vectors where vector comparisons are very cheap). -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
