[https://nvbugs/6450333][test] Unwaive DeepSeek V4 Flash auto dtype test

[lfr-0531]

Description

NVBug 6450333 is not reproducible on the latest main, so the waiver for the following test is no longer needed:

accuracy/test_llm_api_pytorch.py::TestDeepSeekV4Flash::test_auto_dtype SKIP (https://nvbugs/6450333)

This PR removes only that waiver entry, re-enabling the existing DeepSeek V4 Flash auto-dtype accuracy test in the configured B200 and B300 CI stages. It does not change source or runtime behavior.

Test Coverage

Validated on 4x B200 from main at b91ffdad9113f1d04adbb685b044f9d64759ab88 after a clean SM100 build-install. The test process loaded the current repository source and rebuilt nanobind module; LlmRequest.get_tokens_range was present.

CUDA_VISIBLE_DEVICES=0,1,2,3 \
LLM_MODELS_ROOT=/home/scratch.trt_llm_data/llm-models \
PYTHONPATH="$PWD:$PWD/tests/integration/defs:$PWD/examples" \
python3 -m pytest -vv -s \
  tests/integration/defs/accuracy/test_llm_api_pytorch.py::TestDeepSeekV4Flash::test_auto_dtype \
  --timeout=2400

Result: 1 passed in 431.42s.

• MMLU: 89.059 (threshold 85.383)
• GSM8K: 95.451 (threshold 91.907)
• pre-commit run --files tests/integration/test_lists/waives.txt: passed

The local validation used B200 hardware; removing the waiver restores coverage for both the B200 and B300 stages where this test is already listed.

PR Checklist

• PR title follows the [JIRA/NVBUG][type] description format
• PR description clearly explains what and why
• Relevant test coverage is included above
• Change is limited to one stale waiver; no API, dependency, CODEOWNERS, documentation, or architecture changes
• DCO sign-off is included

GitHub Bot Help

To see a list of available CI bot commands, comment /bot help.

Dev Engineer Review

• Removed the TestDeepSeekV4Flash::test_auto_dtype waiver and its NVBug reference from tests/integration/test_lists/waives.txt.
• The change is correctly scoped to the intended test-list entry, with no source or runtime behavior changes.
• Waiver format remains valid, with no unintended configuration changes or duplicate entries identified.

QA Engineer Review

• No test-db/ or qa/ files were modified.
• Removed the waiver for accuracy/test_llm_api_pytorch.py::TestDeepSeekV4Flash::test_auto_dtype, re-enabling its configured B200 and B300 CI coverage.
• The test passed on 4x B200 with MMLU 89.059 and GSM8K 95.451.
• 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: ef2359f3-8b92-443d-9163-224d77290389

📥 Commits

Reviewing files that changed from the base of the PR and between 1dd8b97 and 7d14c8f.

📒 Files selected for processing (1)

• tests/integration/test_lists/waives.txt

💤 Files with no reviewable changes (1)

• tests/integration/test_lists/waives.txt

---

Walkthrough

The integration waiver list removes the SKIP entry for TestDeepSeekV4Flash::test_auto_dtype, allowing the test to run without that waiver.

Changes

Integration waiver update

Layer / File(s)	Summary
Remove DeepSeekV4Flash waiver tests/integration/test_lists/waives.txt	Removes the TestDeepSeekV4Flash::test_auto_dtype waiver and its linked nvbugs reference.

Estimated code review effort: 1 (Trivial) | ~2 minutes

Possibly related PRs

• NVIDIA/TensorRT-LLM#16655: Removes the same DeepSeekV4Flash auto-dtype waiver entry.
• NVIDIA/TensorRT-LLM#16515: Removes another DeepSeek auto-dtype waiver from the same list.
• NVIDIA/TensorRT-LLM#15511: Removes multiple entries from the same integration waiver list.

Suggested reviewers: qijune

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly matches the waiver removal change and identifies the target test and NVBug.
Description check	✅ Passed	The description includes the required sections and gives clear issue, fix, test coverage, and checklist details.
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.}

[lfr-0531]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #61854 [ run ] triggered by Bot. Commit: 7d14c8f Link to invocation

[tensorrt-cicd]

PR_Github #61854 [ run ] completed with state SUCCESS. Commit: 7d14c8f
/LLM/main/L0_MergeRequest_PR pipeline #50047 completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

• Please check the failed tests and fix your PR
• If you cannot view the failures, ask the CI triggerer to share details
• Once fixed, request an NVIDIA team member to trigger CI again

CI Agent Failure Analysis

Link to invocation

[lfr-0531]

/bot run --stage-list "DGX_B200-4_GPUs-PyTorch-1-cbts"

[tensorrt-cicd]

PR_Github #61935 [ run ] triggered by Bot. Commit: 7d14c8f Link to invocation

[tensorrt-cicd]

PR_Github #61935 [ run ] completed with state FAILURE. Commit: 7d14c8f
/LLM/main/L0_MergeRequest_PR pipeline #50125 (Partly Tested) completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

• Please check the failed tests and fix your PR
• If you cannot view the failures, ask the CI triggerer to share details
• Once fixed, request an NVIDIA team member to trigger CI again

CI Agent Failure Analysis

Link to invocation

[lfr-0531]

/bot run --stage-list "DGX_B200-4_GPUs-PyTorch-1"

[tensorrt-cicd]

PR_Github #62058 [ run ] triggered by Bot. Commit: 7d14c8f Link to invocation

[tensorrt-cicd]

PR_Github #62058 [ run ] completed with state SUCCESS. Commit: 7d14c8f
/LLM/main/L0_MergeRequest_PR pipeline #50243 (Partly Tested) completed with status: 'SUCCESS'

CI Report

Link to invocation

[lfr-0531]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #62122 [ run ] triggered by Bot. Commit: 7d14c8f Link to invocation

[tensorrt-cicd]

PR_Github #62122 [ run ] completed with state SUCCESS. Commit: 7d14c8f
/LLM/main/L0_MergeRequest_PR pipeline #50303 completed with status: 'SUCCESS'

CI Report

Link to invocation