-
Notifications
You must be signed in to change notification settings - Fork 544
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Forward-merge branch-24.06 into branch-24.08 #5905
Conversation
cc @vyasr Authors: - Dante Gama Dessavre (https://github.com/dantegd) Approvers: - Vyas Ramasubramani (https://github.com/vyasr) - Ray Douglass (https://github.com/raydouglass) URL: #5890
FAILURE - Unable to forward-merge due to an error, manual merge is necessary. Do not use the IMPORTANT: When merging this PR, do not use the auto-merger (i.e. the |
Closes #5637 ``` import cuml from cuml.datasets import make_classification X, y = make_classification() clf = cuml.ensemble.RandomForestClassifier().fit(X,y) print(clf.predict(X[:5]).dtype) ``` Result is ``` int64 ``` Authors: - Jinsol Park (https://github.com/jinsolp) Approvers: - Dante Gama Dessavre (https://github.com/dantegd) URL: #5896
This is an attempt to update the scikit-learn dependency from 1.2 to 1.4. Most changes are related to constructor arguments that were deprecated in 1.2 and in 1.4 have changed/been removed. A question I have is what cuml's deprecation policy is? I've gone with "two releases" for parameters where we can easily do so (deprecated in 24.06 and then remove them in 24.10). However that is only about 4 months of deprecation which could be a bit short. Some of the changes would be hard to do as a deprecation (with 1.4 there is no way to provide the "old way"), we'd have to stick with 1.3 for now. I think this is a bit of a bummer but maybe the price to pay for not keeping on top of deprecations. And it seems like there is no deprecation policy in the docs/towards users? So maybe we can play this card once now, to catch up and at the same time introduce a deprecation policy. The SHAP test needed its reference updating. I am not sure why, at least I couldn't quickly find a reason for why you'd have to do this. I am not sure how possible it would be to support a range of scikit-learn versions (say 1.2 - 1.4). Would be cool but maybe not worth the added complexity? Todo: * [x] add deprecation warning in AgglomerativeClustering * [ ] add tests for deprecations * [x] RF regressor * [x] RF classifier * [ ] ~~LARS~~ - LARS is experimental, so no need for deprecation * [x] LogisticRegression * [x] OneHotEncoder * [x] AgglomerativeClustering * [ ] think about how to combine this with #5799 * [x] decide deprecation cycle length - copy cudf, so 24.06 -> 24.08 * [x] update "expiry" version in the warnings * [x] update doc strings xref #5799 Authors: - Tim Head (https://github.com/betatim) - Dante Gama Dessavre (https://github.com/dantegd) Approvers: - Dante Gama Dessavre (https://github.com/dantegd) - Jake Awe (https://github.com/AyodeAwe) URL: #5851
Similar to rapidsai/cudf#15552, we are testing [building RAPIDS with CCCL's main branch](NVIDIA/cccl#1667) to get ahead of any breaking changes. Authors: - Paul Taylor (https://github.com/trxcllnt) - Dante Gama Dessavre (https://github.com/dantegd) Approvers: - Kyle Edwards (https://github.com/KyleFromNVIDIA) - Dante Gama Dessavre (https://github.com/dantegd) URL: #5886
Fix conflict of forward-merge #5905 of branch-24.06 into branch-24.08
Forward-merge triggered by push to branch-24.06 that creates a PR to keep branch-24.08 up-to-date. If this PR is unable to be immediately merged due to conflicts, it will remain open for the team to manually merge. See forward-merger docs for more info.