MNT remove `take` fn in array_api wrapper #27939

fcharras · 2023-12-11T10:24:47Z

What does this implement/fix? Explain your changes.

Only a bit of cleaning, this method is no more needed since take has been adopted by the Array API standard.

Any other comments?

The only place where xp.take is used is in sklearn.utils.extmath.svd_flip.

github-actions · 2023-12-11T10:25:58Z

✔️ Linting Passed

All linting checks passed. Your pull request is in excellent shape! ☀️

_{Generated for commit: ede5ec5. Link to the linter CI: here}

ogrisel

Since the array api support is still experimental, I assume that we can expect our users to always use the latest version of the array-api-compat soft dependency. Therefore I assume that this kind of change is fine as long as all existing tests pass.

Could you please confirm that it works with pytorch and cupy on a CUDA host?

ogrisel · 2023-12-11T10:54:46Z

/cc @thomasjpfan

fcharras · 2023-12-11T12:25:26Z

isdtype can probably be removed and also replaced by xp.isdtype safely by now.

ogrisel · 2023-12-11T13:04:16Z

isdtype can probably be removed and also replaced by xp.isdtype safely by now.

Let's open a dedicated PR to be able to merge those two simplication independently in case of one is breaking tests but not the other.

fcharras · 2023-12-11T16:00:04Z

Since the array api support is still experimental, I assume that we can expect our users to always use the latest version of the array-api-compat soft dependency. Therefore I assume that this kind of change is fine as long as all existing tests pass.

That's also my thoughts.

Could you please confirm that it works with pytorch and cupy on a CUDA host?

Tests OK. The relevant tests are the tests for PCA, it's where svd_flip is used with Array API inputs (see former PCA PR for Array API compliancy):

With cupy

pytest -v sklearn/decomposition/tests/test_pca.py -k cupy

=============================================== test session starts ===============================================
platform linux -- Python 3.10.12, pytest-7.4.3, pluggy-1.3.0 -- /data/parietal/store3/work/fcharras/mambaforge/envs/cuml_env/bin/python3.10
cachedir: .pytest_cache
rootdir: /data/parietal/store3/work/fcharras/scikit-learn
configfile: setup.cfg
collected 309 items / 293 deselected / 16 selected                                                                

sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_input_and_values-cupy-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_input_and_values-cupy.array_api-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_get_precision-cupy-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_get_precision-cupy.array_api-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_input_and_values-cupy-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_input_and_values-cupy.array_api-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_get_precision-cupy-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_get_precision-cupy.array_api-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_input_and_values-cupy-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_input_and_values-cupy.array_api-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_get_precision-cupy-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_get_precision-cupy.array_api-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_input-cupy-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_input-cupy.array_api-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_get_precision-cupy-None-None] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_get_precision-cupy.array_api-None-None] PASSED

================================== 16 passed, 293 deselected, 1 warning in 2.93s ==================================

With pytorch

pytest -v sklearn/decomposition/tests/test_pca.py -k torch

=========================================================================================================== test session starts ============================================================================================================
platform linux -- Python 3.11.6, pytest-7.4.3, pluggy-1.3.0 -- /data/parietal/store3/work/fcharras/mambaforge/envs/pytorch_env_3/bin/python3.11
cachedir: .pytest_cache
rootdir: /data/parietal/store3/work/fcharras/scikit-learn
configfile: setup.cfg
collected 309 items / 269 deselected / 40 selected                                                                                                                                                                                         

sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_input_and_values-torch-cpu-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_input_and_values-torch-cpu-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_input_and_values-torch-cuda-float64] 

PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_input_and_values-torch-cuda-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_input_and_values-torch-mps-float32] SKIPPED (Skipping MPS device test because
PYTORCH_ENABLE_MPS_FALLBACK is not set.)
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_get_precision-torch-cpu-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_get_precision-torch-cpu-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_get_precision-torch-cuda-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_get_precision-torch-cuda-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='full')-check_array_api_get_precision-torch-mps-float32] SKIPPED (Skipping MPS device test because PYTORCH_ENABLE_MPS_FALLBACK
is not set.)
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_input_and_values-torch-cpu-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_input_and_values-torch-cpu-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_input_and_values-torch-cuda-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_input_and_values-torch-cuda-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_input_and_values-torch-mps-float32] SKIPPED (Skipping MPS device test because
PYTORCH_ENABLE_MPS_FALLBACK is not set.)
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_get_precision-torch-cpu-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_get_precision-torch-cpu-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_get_precision-torch-cuda-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_get_precision-torch-cuda-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=0.1,svd_solver='full',whiten=True)-check_array_api_get_precision-torch-mps-float32] SKIPPED (Skipping MPS device test because
PYTORCH_ENABLE_MPS_FALLBACK is not set.)
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_input_and_values-torch-cpu-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_input_and_values-torch-cpu-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_input_and_values-torch-cuda-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_input_and_values-torch-cuda-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_input_and_values-torch-mps-float32] SKIPPED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_get_precision-torch-cpu-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_get_precision-torch-cpu-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_get_precision-torch-cuda-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_get_precision-torch-cuda-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_get_precision-torch-mps-float32] SKIPPED (Skipping
MPS device test because PYTORCH_ENABLE_MPS_FALLBACK is not set.)
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_input-torch-cpu-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_input-torch-cpu-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_input-torch-cuda-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_input-torch-cuda-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_input-torch-mps-float32] SKIPPED (Skipping MPS device test because PYTORCH_ENABLE_MPS_FALLBACK
is not set.)
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_get_precision-torch-cpu-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_get_precision-torch-cpu-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_get_precision-torch-cuda-float64] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_get_precision-torch-cuda-float32] PASSED
sklearn/decomposition/tests/test_pca.py::test_pca_mle_array_api_compliance[PCA(n_components='mle',svd_solver='full')-check_array_api_get_precision-torch-mps-float32] SKIPPED (Skipping MPS device test because
PYTORCH_ENABLE_MPS_FALLBACK is not set.)

============================================================================================== 32 passed, 8 skipped, 269 deselected in 20.88s ==============================================================================================

Let's open a dedicated PR to be able to merge those two simplication independently in case of one is breaking tests but not the other.

ok

fcharras · 2023-12-11T16:12:27Z

isdtype can probably be removed

It cannot, it must at least be implemented for numpy (np.isdtype does not exist)

ogrisel

Thanks for posting the test results. Still +1 for merge this PR in its current state.

thomasjpfan

LGTM

betatim · 2023-12-12T08:24:53Z

Since the array api support is still experimental, I assume that we can expect our users to always use the latest version of the array-api-compat soft dependency. Therefore I assume that this kind of change is fine as long as all existing tests pass.

A late comment: the take function we removed here is in a private part of scikit-learn (sklearn.utils._array_api), doesn't this mean it isn't part of the usual promise of backwards compatibility/deprecation cycles? Therefore it is always ok to remove, rename, change, edit, etc things that live in this module.

Or was your comment more about the minimum version of array_api_compat we require?

fcharras · 2023-12-12T09:29:52Z

The Array API is versionned. take is a more recent addition to the standard. Having take defined inside scikit-learn enable support for anterior version of the Array API and removing it remove the support for those older versions. But it's fine since our support for Array API is experimental, I think we can commit to only support the latest version for the Array API standard at the moment. Whenever we're out of experimental status we should have a support policy regarding the supported versions of the Array API.

MNT remove take fn in array_api wrapper

ede5ec5

github-actions bot added the module:utils label Dec 11, 2023

ogrisel added the No Changelog Needed label Dec 11, 2023

ogrisel approved these changes Dec 11, 2023

View reviewed changes

fcharras mentioned this pull request Dec 11, 2023

MNT Prune unused argument in _array_api_for_tests util #27941

Merged

ogrisel approved these changes Dec 11, 2023

View reviewed changes

ogrisel added the Array API label Dec 11, 2023

thomasjpfan approved these changes Dec 11, 2023

View reviewed changes

thomasjpfan merged commit 8e10cd7 into scikit-learn:main Dec 11, 2023

glemaitre pushed a commit to glemaitre/scikit-learn that referenced this pull request Feb 10, 2024

MNT remove take fn in array_api wrapper (scikit-learn#27939)

b36c7f7

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

MNT remove `take` fn in array_api wrapper #27939

MNT remove `take` fn in array_api wrapper #27939

Uh oh!

fcharras commented Dec 11, 2023

Uh oh!

github-actions bot commented Dec 11, 2023

Uh oh!

ogrisel left a comment

Uh oh!

ogrisel commented Dec 11, 2023

Uh oh!

fcharras commented Dec 11, 2023

Uh oh!

ogrisel commented Dec 11, 2023

Uh oh!

fcharras commented Dec 11, 2023 •

edited

Loading

Uh oh!

fcharras commented Dec 11, 2023

Uh oh!

ogrisel left a comment •

edited

Loading

Uh oh!

thomasjpfan left a comment

Uh oh!

betatim commented Dec 12, 2023 •

edited

Loading

Uh oh!

fcharras commented Dec 12, 2023

Uh oh!

Uh oh!

Uh oh!

MNT remove take fn in array_api wrapper #27939

MNT remove take fn in array_api wrapper #27939

Uh oh!

Conversation

fcharras commented Dec 11, 2023

What does this implement/fix? Explain your changes.

Any other comments?

Uh oh!

github-actions bot commented Dec 11, 2023

✔️ Linting Passed

Uh oh!

ogrisel left a comment

Choose a reason for hiding this comment

Uh oh!

ogrisel commented Dec 11, 2023

Uh oh!

fcharras commented Dec 11, 2023

Uh oh!

ogrisel commented Dec 11, 2023

Uh oh!

fcharras commented Dec 11, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

fcharras commented Dec 11, 2023

Uh oh!

ogrisel left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

thomasjpfan left a comment

Choose a reason for hiding this comment

Uh oh!

betatim commented Dec 12, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

fcharras commented Dec 12, 2023

Uh oh!

Uh oh!

MNT remove `take` fn in array_api wrapper #27939

MNT remove `take` fn in array_api wrapper #27939

fcharras commented Dec 11, 2023 •

edited

Loading

ogrisel left a comment •

edited

Loading

betatim commented Dec 12, 2023 •

edited

Loading