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 2022/09/08 23:43:09 UTC
[GitHub] [lucene] jtibshirani opened a new issue, #11758: Follow-up refactors to 8-bit quantization change
jtibshirani opened a new issue, #11758:
URL: https://github.com/apache/lucene/issues/11758
### Description
This issue tracks ideas for refactors as a follow-up to #11613, where we added support for 8-bit vector values:
* We extended `KnnVectorsWriter` to be generic, accepting both `float[]` and `BytesRef` values. It'd be good to explore simplifying the write API to remove these generics.
* We now switch on `VectorEncoding` in several places during graph building and search. Is there a way to encapsulate this better? Maybe there could be a helper class that knows the encoding and can perform distance computations, so that `HnswGraphBuilder` and `HnswGraphSearcher` don't need to know about the encoding.
--
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.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
Re: [I] Follow-up refactors to 8-bit quantization change [lucene]
Posted by "benwtrent (via GitHub)" <gi...@apache.org>.
benwtrent closed issue #11758: Follow-up refactors to 8-bit quantization change
URL: https://github.com/apache/lucene/issues/11758
--
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
Re: [I] Follow-up refactors to 8-bit quantization change [lucene]
Posted by "benwtrent (via GitHub)" <gi...@apache.org>.
benwtrent commented on issue #11758:
URL: https://github.com/apache/lucene/issues/11758#issuecomment-1843041904
There are multiple simplifications now in the quantized byte operations.
One key thing is that the graph searcher now takes a "vector scorer" object and doesn't care about floats vs bytes.
The API for knn search is cleaner now.
Closing this issue as "done". There are always more simplifications, but we have come a long way in the past year :)
--
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