Skip to content

Fixup CI - #8280

Merged
rapids-bot[bot] merged 4 commits into
NVIDIA:mainfrom
jcrist:fixup-ci
Jun 22, 2026
Merged

Fixup CI#8280
rapids-bot[bot] merged 4 commits into
NVIDIA:mainfrom
jcrist:fixup-ci

Conversation

@jcrist

@jcrist jcrist commented Jun 22, 2026

Copy link
Copy Markdown
Member

We have some ongoing CI failures that showed up last week. This fixes them.

  • Relaxes tolerance on a cuml.explainer xgboost test, this changed due to recent xgboost release.
  • Moves xfail marks for a bunch of tests due to sklearn's repr erroring on our CI machines to strict: false. These appear to be failing due to a misconfigured codec on one or more of our machines. Since they don't fail everywhere, they should be strict: false for now.
  • Marks test_logreg_l1[42-csr_matrix] as flaky on sklearn 1.9.
  • Also cherry-picks the fix for the examples from Xfail failing wikipedia sklearn example #8277 (per offline request by Simon), since that's also needed to fix our nightlies.

Fixes #8265.
Fixes #8278.
Fixes #8279.
Fixes #8276.

@jcrist
jcrist requested a review from a team as a code owner June 22, 2026 19:57
@jcrist
jcrist requested a review from betatim June 22, 2026 19:57
@github-actions github-actions Bot added the Cython / Python Cython or Python issue label Jun 22, 2026
@jcrist jcrist added improvement Improvement / enhancement to an existing function non-breaking Non-breaking change labels Jun 22, 2026
@coderabbitai

coderabbitai Bot commented Jun 22, 2026

Copy link
Copy Markdown

Review Change Stack

📝 Walkthrough

Walkthrough

Three CI stability fixes: the HTML repr cuml_accel_bugs xfail block in xfail-list.yaml is moved and converted from strict to strict: false with an updated reason; a new cuml_accel_flaky entry is added for a logistic regression test; one new example is added to xfail-examples.yaml; and test_xgb_regressor SHAP comparison tolerance is relaxed from 5 to 4 decimal places.

Changes

CI Test Stability Fixes

Layer / File(s) Summary
sklearn xfail-list: HTML repr block and flaky logistic regression
python/cuml/cuml_accel_tests/upstream/scikit-learn/xfail-list.yaml
Removes the earlier strict cuml_accel_bugs block for HTML repr tests on scikit-learn>=1.9; reintroduces the same test IDs at a later position with strict: false and the updated reason "misconfigured machine leading to ASCII codec reads". Adds a separate cuml_accel_flaky entry with strict: false for test_logreg_l1[42-csr_matrix].
sklearn xfail-examples: wikipedia_principal_eigenvector
python/cuml/cuml_accel_tests/upstream/scikit-learn/xfail-examples.yaml
Adds applications::wikipedia_principal_eigenvector under a new scikit-learn>=1.9 xfail entry because the upstream example fails without cuml.accel on the current dependency stack.
GPU TreeSHAP assertion tolerance: decimal 5 → 4
python/cuml/tests/explainer/test_gpu_treeshap.py
Relaxes both assert_almost_equal checks in test_xgb_regressor from decimal=5 to decimal=4 to accommodate the observed floating-point differences (max absolute difference ~4.6e-05) in SHAP values and bias comparisons.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~4 minutes

Possibly related PRs

  • rapidsai/cuml#8207: Directly reworked the same xfail-list.yaml HTML repr block for scikit-learn>=1.9 that this PR moves and converts to non-strict mode.
  • rapidsai/cuml#8193: Also adds entries to xfail-examples.yaml for scikit-learn upstream examples, the same file modified in this PR.
  • rapidsai/cuml#8043: Modifies test_gpu_treeshap.py SHAP test behavior in the same module where this PR adjusts numerical tolerance.

Suggested labels

ci

Suggested reviewers

  • csadorf
  • AyodeAwe
  • jameslamb
🚥 Pre-merge checks | ✅ 3 | ❌ 2

❌ Failed checks (1 warning, 1 inconclusive)

Check name Status Explanation Resolution
Docstring Coverage ⚠️ Warning Docstring coverage is 33.33% which is insufficient. The required threshold is 80.00%. Write docstrings for the functions missing them to satisfy the coverage threshold.
Title check ❓ Inconclusive The title 'Fixup CI' is vague and does not clearly convey what specific CI issues are being fixed. Use a more specific title that describes the main changes, such as 'Fix CI failures: relax XGBoost test tolerance and resolve HTML repr test xfail marks' or similar.
✅ Passed checks (3 passed)
Check name Status Explanation
Linked Issues check ✅ Passed All three linked issues (#8212, #8278, #8279) are properly addressed: XGBoost tolerance relaxed [#8278], HTML repr xfail marks removed [#8212, #8279], and cuml_accel_tests updated accordingly.
Out of Scope Changes check ✅ Passed All changes are directly related to addressing the CI failures mentioned in the linked issues. No out-of-scope changes detected.
Description check ✅ Passed The pull request description is directly related to the changeset, explaining specific modifications and their purposes.

✏️ Tip: You can configure your own custom pre-merge checks in the settings.

✨ Finishing Touches
🧪 Generate unit tests (beta)
  • Create PR with unit tests

Comment @coderabbitai help to get the list of available commands.

@jcrist

jcrist commented Jun 22, 2026

Copy link
Copy Markdown
Member Author

/merge

@rapids-bot
rapids-bot Bot merged commit fd319e5 into NVIDIA:main Jun 22, 2026
101 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Cython / Python Cython or Python issue improvement Improvement / enhancement to an existing function non-breaking Non-breaking change

Projects

None yet

4 participants