[https://nvbugs/6550803][fix] Pin Mock(_force_non_greedy_for_capture=False) at the test's spec_metadata…

[trtllm-agent]

Summary

• Root cause: The test passed a bare Mock() as spec_metadata, so getattr(spec_metadata, '_force_non_greedy_for_capture', False) returned a truthy auto-vivified child Mock instead of False, tripping the production assert that guards the capture-only sampling override.
• Fix: Pin Mock(_force_non_greedy_for_capture=False) at the test's spec_metadata double (the convention already used at line 231 of the same file), leaving the production guard intact, and remove the obsolete waiver.
• Automated fix generated by repair-bot

Test plan

• Verify fix on the same GPU type as the original failure
• Check for regressions in related tests

Links

• Bug: https://nvbugs/6550803

Dev Engineer Review

• Updated spec_metadata to set _force_non_greedy_for_capture=False.
• This prevents Mock auto-vivification from triggering the production assertion.
• Removed the obsolete waiver entry.
• The changes preserve the production guard and have no API or performance impact.
• The waiver removal uses the existing test name and does not expand test scope.

Verdict: sufficient.

QA Engineer Review

• Modified test: test_promoted_context_precedes_speculative_overlap_generation.
• The test is covered by the removed entry in tests/integration/test_lists/waives.txt.
• No test-db or QA entry was added or modified.

Verdict: sufficient.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 320657d5-88c2-4ebd-af9e-5eb1c1390cea

📥 Commits

Reviewing files that changed from the base of the PR and between 73efcee and 8bc84eb.

📒 Files selected for processing (2)

• tests/integration/test_lists/waives.txt
• tests/unittest/_torch/executor/test_pytorch_model_engine.py

💤 Files with no reviewable changes (1)

• tests/integration/test_lists/waives.txt

---

Walkthrough

The speculative decoding test fixture now sets _force_non_greedy_for_capture explicitly to False. Its integration waiver and NVBug reference are removed.

Changes

Speculative decoding test correction

Layer / File(s)	Summary
Correct capture setup and remove waiver tests/unittest/_torch/executor/test_pytorch_model_engine.py, tests/integration/test_lists/waives.txt	The fixture sets _force_non_greedy_for_capture=False. The test waiver is removed.

Estimated code review effort: 1 (Trivial) | ~5 minutes

Possibly related PRs

• NVIDIA/TensorRT-LLM#17182: Adds the waiver that this change removes while addressing the same test.
• NVIDIA/TensorRT-LLM#17184: Addresses the same test waiver and fixture setup.
• NVIDIA/TensorRT-LLM#17205: Matches the fixture and waiver updates in this change.

Suggested reviewers: qijune, brnguyen2

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title identifies the bug, fix type, and specific Mock configuration change that resolves the test failure.
Description check	✅ Passed	The description explains the root cause, fix, obsolete waiver removal, test plan, and linked bug, but does not reproduce the full checklist template.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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