[https://nvbugs/6185480][fix] Autodeploy skip the GLM accuracy test for pre-hopper

[nvchenghaoz]

Skip GLM-4.7-Flash AutoDeploy BF16 accuracy test on pre-Hopper GPUs.

On pre-Hopper GPUs such as A100, AutoDeploy cannot use the paged flashinfer_mla path for this MLA model and falls back to torch_mla. The fallback allocates unpaged MLA caches sized by max_batch_size * max_seq_len, which exceeds 80GB memory for this test configuration. Hopper uses the paged MLA path and remains covered.

Summary by CodeRabbit

• Tests
  • Modified test execution to skip certain accuracy validation tests on pre-hopper systems.

Description

Test Coverage

PR Checklist

Please review the following before submitting your PR:

• PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.

• PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.

• Test cases are provided for new code paths (see test instructions)

• If PR introduces API changes, an appropriate PR label is added - either api-compatible or api-breaking. For api-breaking, include BREAKING in the PR title.

• Any new dependencies have been scanned for license and vulnerabilities

• CODEOWNERS updated if ownership changes

• Documentation updated as needed

• Update tava architecture diagram if there is a significant design change in PR.

• The reviewers assigned automatically/manually are appropriate for the PR.

• Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

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

[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: 026b93fd-016a-40ad-8693-5eae58a543e1

📥 Commits

Reviewing files that changed from the base of the PR and between 1a52920 and d90b7cc.

📒 Files selected for processing (1)

• tests/integration/defs/accuracy/test_llm_api_autodeploy.py

---

📝 Walkthrough

Walkthrough

The test_auto_dtype test in the TestGLM4Flash class is now decorated with @skip_pre_hopper, gating its execution to skip on pre-hopper GPU architectures. This is a single-line change to test environment configuration.

Changes

Test environment gating

Layer / File(s)	Summary
Add pre-hopper skip decorator to GLM4Flash test tests/integration/defs/accuracy/test_llm_api_autodeploy.py	The test_auto_dtype test in TestGLM4Flash is now decorated with @skip_pre_hopper to skip execution on pre-hopper GPU architectures.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~2 minutes

Possibly related PRs

• NVIDIA/TensorRT-LLM#14259: Both PRs modify the same tests/integration/defs/accuracy/test_llm_api_autodeploy.py test (test_auto_dtype) to be skipped—main PR adds @skip_pre_hopper while retrieved PR adds a CI waiver for TestNemotronV2::test_auto_dtype[False].

Suggested reviewers

• crazydemo
• StanleySun639
• jieli-matrix

🚥 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 matches the changeset: it specifies skipping the GLM accuracy test for pre-hopper systems, which is exactly what the code change does by adding @skip_pre_hopper guard.
Description check	✅ Passed	The description explains the issue (memory limitation on pre-Hopper GPUs) and solution (skip test). However, the PR checklist section is incomplete with unfilled template comments and checkboxes.
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.}

[nvchenghaoz]

/bot run

[tensorrt-cicd]

PR_Github #50820 [ run ] triggered by Bot. Commit: 9aedd82 Link to invocation

[tensorrt-cicd]

PR_Github #50820 [ run ] completed with state FAILURE. Commit: 9aedd82
/LLM/main/L0_MergeRequest_PR pipeline #40290 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

[nvchenghaoz]

/bot run

[tensorrt-cicd]

PR_Github #50824 [ run ] triggered by Bot. Commit: 9aedd82 Link to invocation

[tensorrt-cicd]

PR_Github #50824 [ run ] completed with state SUCCESS. Commit: 9aedd82
/LLM/main/L0_MergeRequest_PR pipeline #40294 completed with status: 'SUCCESS'

CI Report

Link to invocation