fix: reenable flaky tests on test_torch_kmeans.py

[cswpy]

Seems like the tests are not flaky anymore and we could re-enable them. Fixes #1683 .

[coderabbitai]

📝 Walkthrough

Walkthrough

The Torch KMeans tests are re-enabled. One test uses explicit duplicate centroids with CPU execution, while the device-acceptance test passes an explicit CPU device.

Changes

Torch KMeans test execution

Layer / File(s)	Summary
Deterministic test setup and CPU execution python/python/tests/torch_tests/test_torch_kmeans.py	The tests are no longer skipped, KMeans uses explicit duplicate centroids, and accelerator training receives torch.device("cpu").

Estimated code review effort: 2 (Simple) | ~10 minutes

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly summarizes the main change: re-enabling previously flaky Torch k-means tests.
Description check	✅ Passed	The description is directly related to the PR and correctly states that the tests are being re-enabled.
Linked Issues check	✅ Passed	The PR addresses #1683 by fixing the broken torch tests in test_torch_kmeans.py and unskipping the affected cases.
Out of Scope Changes check	✅ Passed	The changes stay focused on the targeted test file and its flaky test re-enablement.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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

[coderabbitai]

Note

Quiet mode is enabled, so only the most important comments were posted inline. Other review comments are grouped below.

🟡 Other comments (1)

python/python/tests/torch_tests/test_torch_kmeans.py-19-23 (1)

19-23: 🩺 Stability & Availability | 🟡 Minor | ⚡ Quick win

Seed the empty-cluster repair path. KMeans._updated_centroids still calls unseeded torch.rand_like when a cluster is empty, so this test can still vary run to run. Seed the CPU RNG locally or make the assertion independent of that perturbation.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@python/python/tests/torch_tests/test_torch_kmeans.py` around lines 19 - 23,
Make the test around KMeans._updated_centroids deterministic by locally seeding
the CPU Torch RNG before kmeans.fit(arr), or revise its assertions so they do
not depend on the random empty-cluster repair perturbation. Keep the
duplicate-centroid setup and validate the intended empty-cluster recovery
behavior.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Other comments:
In `@python/python/tests/torch_tests/test_torch_kmeans.py`:
- Around line 19-23: Make the test around KMeans._updated_centroids
deterministic by locally seeding the CPU Torch RNG before kmeans.fit(arr), or
revise its assertions so they do not depend on the random empty-cluster repair
perturbation. Keep the duplicate-centroid setup and validate the intended
empty-cluster recovery behavior.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: QUIET

Plan: Pro Plus

Run ID: e3bd19cf-0237-417e-a256-0e84eccf8d93

📥 Commits

Reviewing files that changed from the base of the PR and between 252d81a and a1a66d0.

📒 Files selected for processing (1)

• python/python/tests/torch_tests/test_torch_kmeans.py