kaivalnp opened a new issue, #15734: URL: https://github.com/apache/lucene/issues/15734
### Description Today, Lucene supports [8, 4, 2 and 1](https://github.com/apache/lucene/blob/e1879e450b75b3a58fde2b0dad77ae6b499504dd/lucene/core/src/java/org/apache/lucene/codecs/lucene104/Lucene104ScalarQuantizedVectorsFormat.java#L119-L148) bit quantization. Each quantization level typically has an upper bound of recall with exact KNN that it can produce (which is "exact KNN with quantized scores" v/s "exact KNN with original scores", see https://github.com/mikemccand/luceneutil/issues/528) -- this is the information loss due to quantization itself (before approximate-ness from search algorithms like HNSW comes into picture). Any algorithm operating on quantized scores _alone_ cannot go beyond this recall (e.g. tweaking parameters like `maxConn`, `beamWidth`, `fanout`, etc. for HNSW) without using the original scores from un-quantized vectors for re-ranking -- which may not be feasible for some use cases (e.g. keeping the index in-memory for performance, where using un-quantized vectors increases memory footprint by \~4x in case of byte-quantized vectors). In such cases, I wonder if Lucene could support more granular quantization options (say _the equivalent of!_ 6-bit quantization) -- for more granular recall v/s memory requirements? -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
