benwtrent commented on issue #13519: URL: https://github.com/apache/lucene/issues/13519#issuecomment-2207015770
With this change, I get the following recall@100 over 500_000 coherev2 (using max-inner-product as designed for the model): ``` recall maxConn beamWidth 0.872 16 500 ``` Int8 with my change: ``` recall maxConn beamWidth 0.841 16 500 ``` So, recall is better than before for max-inner-product, but still worse than int7. I am wondering if we are running to numeric limitations because the signed correction could get very large. -- 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
