Tejas Shah
Tejas Shah
# [RFC] Adding efSearch as query parameter for kNN ## Overview The efSearch parameter is utilized in the HNSW (Hierarchical Navigable Small World) algorithm to facilitate the exploration of more...
@sohami Thanks for the reference. I would be interested in the low level RFC/ implementation, currently there are only specific cases where we want prefetch since it affects search latencies...
@VijayanB there are multiple of them and aren't small so uploaded them. I can add partial if it helps
> @shatejas do you think it doesnt need changelog? Wasn't planning on adding it. Doesn't seem to be significant enough for users to make it to change log
> Please make the change in changelog.md file https://github.com/opensearch-project/k-NN/pull/1877#issuecomment-2264201750
Closing this one since its outdated
@navneet1v I see your point, Multileaf collector does not hold docIds in global queue. The idea here is to use a global max heap queue of size k, and pass...
@kotwanikunal One of the approaches I was thinking with this was is to unify [NativeEngineKnnVectorQuery](https://github.com/opensearch-project/k-NN/blob/main/src/main/java/org/opensearch/knn/index/query/nativelib/NativeEngineKnnVectorQuery.java) and the LuceneEngineKNNQuery mentioned in Solution. I see a few of benefits if we are...
After more experiments we figured that this live docs iteration is not the root cause, This only yields 32ms gains for 1.6M vectors adding an overhead of maintaining the merge...