benwtrent commented on issue #13519: URL: https://github.com/apache/lucene/issues/13519#issuecomment-2228966111
@MilindShyani OK, I did some more benchmarking. I tried switching to `+` & with your full correction term and recall significantly dropped to `0.518` I tried the follow permutations and they were all the same recall of `0.877`. `(dx - dxq) * (dxq - SIGNED_CORRECTION * alpha)` `(dx - dxq) * (dxq - SIGNED_CORRECTION * alpha - alpha * minQuantile)` `(dx - dxq) * (dxq - SIGNED_CORRECTION * alpha + alpha * minQuantile)` This makes me think the additional term of `alpha * minQuantize` is bringing nothing to the table, even for non-unit vectors (where I would expect quantiles & alpha to be more extreme). -- 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
