You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@lucene.apache.org by GitBox <gi...@apache.org> on 2022/11/22 23:14:25 UTC

[GitHub] [lucene] rmuir commented on issue #11963: Improve vector quantization API

rmuir commented on issue #11963:
URL: https://github.com/apache/lucene/issues/11963#issuecomment-1324355332

   as i mentioned on the original issue(s) (there were several PRs closed and opened and many comments were lost...), the problem is right in the title: quantization.
   
   there shouldnt be any quantization happening. otherwise it is broken.  8-bits in, 8-bits out, 32-bits in, 32-bits out. 
   
   Additionally as a minimum there needs to be two separate field classes (e.g. ByteVectorValues/FloatVectorValues) to give type safety. Really, the types should be distinguished in fieldinfos today as well (not sure if that is currently the case).
   
    Basically just follow the examples of every other part of the index. vectors isn't special.


-- 
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-unsubscribe@lucene.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@lucene.apache.org
For additional commands, e-mail: issues-help@lucene.apache.org