Init checks for Dask KMeans#7391
Merged
rapids-bot[bot] merged 4 commits intorapidsai:mainfrom Oct 31, 2025
Merged
Conversation
jcrist
reviewed
Oct 27, 2025
Member
There was a problem hiding this comment.
While you're here - I'm not 100% sure if this check should be skipped in multi-gpu execution. When refactoring I excluded it in multi-gpu since we weren't running it there before.
Is the multi-gpu implementation robust to a single node having fewer rows than the requested n_clusters? It doesn't seem to error when invoked in that setup, but I'm also not sure if it provides good results.
Contributor
Author
There was a problem hiding this comment.
I think the check was missing in the multi-GPU implementation. I do not know for sure either, but I guess that this is a rare case that should probably not yield very good results especially for scalable/parallel kmeans++ initialization. Better safe than sorry, we should probably alert the user in this case.
divyegala
reviewed
Oct 31, 2025
Member
divyegala
left a comment
divyegala
left a comment
There was a problem hiding this comment.
Maybe also add a check that oversampling_factor > 0?
viclafargue
force-pushed
the
init-checks-dask-kmeans
branch
from
aaa89f5 to
be8c66b
Compare
divyegala
approved these changes
Oct 31, 2025
Contributor
Author
|
/merge |
Merged
rapids-bot Bot
pushed a commit
that referenced
this pull request
Oct 31, 2025
During a recent refactor we removed the `KMeansMG` class, viewing it as internal. It turns out this class was used by a few external projects. Since we still need to support external users accessing the non-dask multi-gpu implementation, we'll want a public way to do so that isn't the private `_fit` method. Additionally, since we want to special case the `MG` case a little more, making it a separate class (even if as a thin shim) makes sense. This PR: - Brings back the `KMeansMG` class - Adds a check that `random_state` is non-None in the `KMeansMG` case, ensuring external users also set `random_state` properly - Removes mutation of kwargs in the dask `KMeans` case (as suggested [here](#7417 (comment))) - Simplifies and moves the multi-gpu `kmeans++`/`oversampling_factor` check (as suggested [here](#7391 (comment))) Fixes #7387. Fixes #7389. Authors: - Jim Crist-Harif (https://github.com/jcrist) Approvers: - Dante Gama Dessavre (https://github.com/dantegd) URL: #7420
vardhan30016
pushed a commit
to vardhan30016/cuml
that referenced
this pull request
Nov 7, 2025
Closes rapidsai#7389 Authors: - Victor Lafargue (https://github.com/viclafargue) Approvers: - Divye Gala (https://github.com/divyegala) URL: rapidsai#7391
vardhan30016
pushed a commit
to vardhan30016/cuml
that referenced
this pull request
Nov 7, 2025
During a recent refactor we removed the `KMeansMG` class, viewing it as internal. It turns out this class was used by a few external projects. Since we still need to support external users accessing the non-dask multi-gpu implementation, we'll want a public way to do so that isn't the private `_fit` method. Additionally, since we want to special case the `MG` case a little more, making it a separate class (even if as a thin shim) makes sense. This PR: - Brings back the `KMeansMG` class - Adds a check that `random_state` is non-None in the `KMeansMG` case, ensuring external users also set `random_state` properly - Removes mutation of kwargs in the dask `KMeans` case (as suggested [here](rapidsai#7417 (comment))) - Simplifies and moves the multi-gpu `kmeans++`/`oversampling_factor` check (as suggested [here](rapidsai#7391 (comment))) Fixes rapidsai#7387. Fixes rapidsai#7389. Authors: - Jim Crist-Harif (https://github.com/jcrist) Approvers: - Dante Gama Dessavre (https://github.com/dantegd) URL: rapidsai#7420
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
4 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.
Closes #7389