[https://nvbugs/6029882][test] Unwaive TestDeepSeekR1::test_nvfp4_multi_gpus[throughput_mtp]

[tianyuxbear]

Description

Removes the waiver for accuracy/test_llm_api_pytorch.py::TestDeepSeekR1::test_nvfp4_multi_gpus[throughput_mtp] (NVBugs 6029882, also tracked by 6261994).

The waiver was added for an intermittent CUDA OOM at executor init (model.to('cuda')) on the GB200-8_GPUs-2_Nodes stage. Investigation shows this is an environment-driven failure: the model weights nearly fill the GB200 and a foreign process occupying GPU 0 in CI tips it over, rather than a model-memory regression. Local reproduction on a 2-node × 4× GB200 setup (50 runs) did not reproduce the OOM, and repair-bot could not reproduce on B200 either. This matches the same fingerprint seen in the Kimi-K2.5 init-OOM bugs (6248837 / 6260890).

Unwaiving to let CI confirm; closing 6029882 and 6261994 on a clean run.

Test Coverage

accuracy/test_llm_api_pytorch.py::TestDeepSeekR1::test_nvfp4_multi_gpus[throughput_mtp]

Summary by CodeRabbit

Release Notes

• Tests
  • Removed a test waiver entry to re-enable testing of a previously skipped scenario.

[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: 017c602d-a63a-4ea5-9e84-b4b916499bc2

📥 Commits

Reviewing files that changed from the base of the PR and between 09c21b6 and 8bedc10.

📒 Files selected for processing (1)

• tests/integration/test_lists/waives.txt

💤 Files with no reviewable changes (1)

• tests/integration/test_lists/waives.txt

---

📝 Walkthrough

Walkthrough

The PR removes a single test waiver entry from the integration test skip list. The deleted waiver was for accuracy/test_llm_api_pytorch.py::TestDeepSeekR1::test_nvfp4_multi_gpus[throughput_mtp], which means this specific test case will no longer be skipped and will run as part of the test suite.

Changes

Test Waiver Removal

Layer / File(s)	Summary
NVFP4 test waiver removal tests/integration/test_lists/waives.txt	Removed the SKIP waiver for TestDeepSeekR1::test_nvfp4_multi_gpus[throughput_mtp], allowing the test to execute in the standard test run.

Possibly related PRs

• NVIDIA/TensorRT-LLM#15010: Both PRs remove SKIP entries for DeepSeekR1 NVFP4-related tests from the waiver list.
• NVIDIA/TensorRT-LLM#14896: Both PRs modify the same waiver list by adding or removing SKIP entries for accuracy/test_llm_api_pytorch.py tests.
• NVIDIA/TensorRT-LLM#14792: Both PRs modify test waiver entries for TestDeepSeekR1 NVFP4 multi-GPU test variants in the same waiver file.

Suggested reviewers

• chzblych
• nvchenghaoz
• bo-nv

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~2 minutes

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and specifically identifies the main change: unwaiving a test waiver. It includes the NVBugs ticket reference and is directly related to the changeset's primary purpose.
Description check	✅ Passed	The PR description is comprehensive and complete, covering the rationale for removing the waiver, investigation findings, test coverage, and context about the NVBugs tickets involved.
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.

_{✏️ 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 and usage tips.}

[tianyuxbear]

/bot run --stage-list "GB200-8_GPUs-2_Nodes-PyTorch-2"

[tensorrt-cicd]

PR_Github #52749 [ run ] triggered by Bot. Commit: 8bedc10 Link to invocation

[tensorrt-cicd]

PR_Github #52749 [ run ] completed with state SUCCESS. Commit: 8bedc10
/LLM/main/L0_MergeRequest_PR pipeline #42011 (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

[tianyuxbear]

/bot run --stage-list "GB200-8_GPUs-2_Nodes-PyTorch-2" --disable-reuse-test

[tensorrt-cicd]

PR_Github #53040 [ run ] triggered by Bot. Commit: 8bedc10 Link to invocation

[tensorrt-cicd]

PR_Github #53040 [ run ] completed with state SUCCESS. Commit: 8bedc10
/LLM/main/L0_MergeRequest_PR pipeline #42259 (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