Implemantion for two-level KMeans Trees in ScaNN#1878
Open
rmaschal wants to merge 1 commit intorapidsai:mainfrom
Open
Implemantion for two-level KMeans Trees in ScaNN#1878rmaschal wants to merge 1 commit intorapidsai:mainfrom
rmaschal wants to merge 1 commit intorapidsai:mainfrom
Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
rmaschal
changed the title
rmaschal
added
feature request
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR adds the option for a second level in the kmeans tree for the ScaNN index. The second level is built in a bottom-up fashion, e.g. second-level cluster centers are trained on leaf centers of the bottom level. AVQ and SOAR are also applied to second-level node centers.
One divergence from OSS ScaNN is that the second-level centers are trained on the normalized leaf centers, rather than directly on leaf centers. I've found this gives better recall than training directly on leaf centers, and often gives better recall compared to the OSS algo.
I've also added a notebook giving example code for producing missing protobuf artifacts for using cuVS built ScaNN indices with OSS ScaNN search functionality.