nknize commented on PR #12306: URL: https://github.com/apache/lucene/pull/12306#issuecomment-1553082121
-1 for even introducing the system property option only because I think it's too trappy. We only just released Lucene 9.6 seven days ago. Why do we need to rush to this in a PR when we haven't started the 9.7 release process? Do we not have enough time for someone enthusiastic here to open a PR for #12309? > My main question for everyone is: how do we decide on a meaningful limit? > Why is '1024' reasonable? Exactly!! Where are the benchmarks!? For all these claims about OpenAPI embeddings and GPT this/that/etc. we don't even have empirical evidence to pick a limit let alone help guide the user community on the traps associated with increasing this limit! @FranciscoBorges on [#11507](https://github.com/apache/lucene/issues/11507#issuecomment-1533718841) you said `We benchmarked a patched Lucene/Solr. We fully understand (we measured it :-P) ` Do you have any restrictions preventing you from posting your numbers to help guide this community on performance expectations at different dimensional limits? Could be a big help in guiding all these opinions. -- 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
