[https://nvbugs/6143787][fix] Add kv_cache_config = KvCacheConfig(free_gpu_memory_fraction=0.6) to TestQwen3

[tensorrt-cicd]

Summary

• Root cause: TestQwen3VL.test_auto_dtype instantiates LLM without any KvCacheConfig, so the default free_gpu_memory_fraction (~0.9) consumes almost all GPU memory as KV cache on 48GB L40S, leaving no room for the attention workspace.
• Fix: Add kv_cache_config = KvCacheConfig(free_gpu_memory_fraction=0.6) to TestQwen3VL and pass it to LLM(); remove the stale skip 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/6143787

Summary by CodeRabbit

• Tests

  • Updated multimodal model test configurations with GPU memory settings.

• Chores

  • Modified test waiver entries in the integration test suite.

[coderabbitai]

📝 Walkthrough

Walkthrough

TestQwen3VL test class adds a KV cache configuration with 60% free GPU memory allocation and applies it to the test_auto_dtype LLM instantiation. A corresponding waiver entry for the Qwen3VL test is removed while a new waiver for a Qwen3.5 35B test is added to the test waiver list.

Changes

KV Cache Configuration for Qwen3VL Tests

Layer / File(s)	Summary
KV Cache Config Definition tests/integration/defs/accuracy/test_llm_api_pytorch_multimodal.py	TestQwen3VL class defines kv_cache_config attribute with KvCacheConfig(free_gpu_memory_fraction=0.6).
Test Method Integration tests/integration/defs/accuracy/test_llm_api_pytorch_multimodal.py	test_auto_dtype method passes kv_cache_config=self.kv_cache_config into the LLM constructor.
Test Waivers Update tests/integration/test_lists/waives.txt	Removes waiver for TestQwen3VL::test_auto_dtype[forced_chunked_prefill]; adds waiver for TestQwen3_5_35B_A3B::test_fp8[enable_block_reuse=True] (NVBUG 6004530).

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 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: adding a kv_cache_config with free_gpu_memory_fraction=0.6 to TestQwen3VL to fix an OOM issue, and includes the associated bug ID.
Description check	✅ Passed	The description includes the root cause, the fix applied, and a test plan, following the repository's template structure with clear sections.
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.}

[coderabbitai]

🧹 Nitpick comments (1)

tests/integration/defs/accuracy/test_llm_api_pytorch_multimodal.py (1)

420-437: QA list update looks unnecessary for this diff.

This is a stability tweak to an existing integration test definition, not a new scheduled test case, so no QA test-list entry change is needed in this file.

As per coding guidelines: "If the change adds or materially alters an integration test under tests/integration/defs/ ... call out whether an entry is needed under tests/integration/test_lists/qa/."

🤖 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 `@tests/integration/defs/accuracy/test_llm_api_pytorch_multimodal.py` around
lines 420 - 437, The change is a stability tweak to the existing integration
test (test_auto_dtype) and should not add or modify QA test-list entries; remove
or revert any edits that created/updated a QA entry for this test case and
ensure no new entry is added under the QA test lists for this change, leaving
the test function test_auto_dtype, the LLM(...) instantiation, and the
MMMU(self.MODEL_NAME).evaluate(...) call as the only modifications in this diff.

🤖 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.

Nitpick comments:
In `@tests/integration/defs/accuracy/test_llm_api_pytorch_multimodal.py`:
- Around line 420-437: The change is a stability tweak to the existing
integration test (test_auto_dtype) and should not add or modify QA test-list
entries; remove or revert any edits that created/updated a QA entry for this
test case and ensure no new entry is added under the QA test lists for this
change, leaving the test function test_auto_dtype, the LLM(...) instantiation,
and the MMMU(self.MODEL_NAME).evaluate(...) call as the only modifications in
this diff.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 3314a292-bdc3-4890-9ae0-b9cc1531c236

📥 Commits

Reviewing files that changed from the base of the PR and between 159ba11 and 4210cdd.

📒 Files selected for processing (2)

• tests/integration/defs/accuracy/test_llm_api_pytorch_multimodal.py
• tests/integration/test_lists/waives.txt

💤 Files with no reviewable changes (1)

• tests/integration/test_lists/waives.txt

[ziyixiong-nv]

/bot run

[tensorrt-cicd]

PR_Github #47312 [ run ] triggered by Bot. Commit: 4210cdd Link to invocation

[tensorrt-cicd]

PR_Github #47312 [ run ] completed with state SUCCESS. Commit: 4210cdd
/LLM/main/L0_MergeRequest_PR pipeline #37253 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

[2ez4bz]

/bot run

[tensorrt-cicd]

PR_Github #50572 [ run ] triggered by Bot. Commit: 83ff5ed Link to invocation

[tensorrt-cicd]

PR_Github #50572 [ run ] completed with state SUCCESS. Commit: 83ff5ed
/LLM/main/L0_MergeRequest_PR pipeline #40073 completed with status: 'SUCCESS'

CI Report

Link to invocation

[2ez4bz]

/bot run

[tensorrt-cicd]

PR_Github #51378 [ run ] triggered by Bot. Commit: ee3d9a9 Link to invocation

[tensorrt-cicd]

PR_Github #51378 [ run ] completed with state FAILURE. Commit: ee3d9a9
/LLM/main/L0_MergeRequest_PR pipeline #40789 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

[2ez4bz]

/bot run

[tensorrt-cicd]

PR_Github #51748 [ run ] triggered by Bot. Commit: ee3d9a9 Link to invocation

[tensorrt-cicd]

PR_Github #51748 [ run ] completed with state SUCCESS. Commit: ee3d9a9
/LLM/main/L0_MergeRequest_PR pipeline #41121 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

[2ez4bz]

/bot run

[tensorrt-cicd]

PR_Github #52171 [ run ] triggered by Bot. Commit: 026ead0 Link to invocation

[tensorrt-cicd]

PR_Github #52171 [ run ] completed with state SUCCESS. Commit: 026ead0
/LLM/main/L0_MergeRequest_PR pipeline #41491 completed with status: 'SUCCESS'

CI Report

Link to invocation