🌐 US-Proxy
class="logged-out env-production page-responsive" style="word-wrap: break-word;" >
Skip to content

[https://nvbugs/6535765][fix] Raise Wan 2.2 LPIPS threshold to 0.25 and unwaive - #19438

Merged
o-stoner merged 1 commit into
NVIDIA:mainfrom
o-stoner:user/o-stoner/visual-gen-unwaive-wan-lpips
Sep 22, 2026
Merged

[https://nvbugs/6535765][fix] Raise Wan 2.2 LPIPS threshold to 0.25 and unwaive#19438
o-stoner merged 1 commit into
NVIDIA:mainfrom
o-stoner:user/o-stoner/visual-gen-unwaive-wan-lpips

Conversation

@o-stoner

@o-stoner o-stoner commented Sep 18, 2026 •
edited by coderabbitai Bot
Loading

Copy link
Copy Markdown
Collaborator

Dev Engineer Review

  • Raises the Wan 2.2 single-GPU LPIPS threshold from 0.05 to 0.25.
  • Removes the NVBugs 6535765 waiver for test_wan22_t2v_lpips_against_golden.
  • The change is limited to the Wan 2.2 LPIPS test. The reported measurement of 0.218467 is below the new threshold.
  • Review findings: unavailable from the supplied evidence.

QA Engineer Review

  • Modifies the Wan 2.2 LPIPS test and golden-video configuration.
  • Removes the matching entry from tests/integration/test_lists/waives.txt.
  • The test preserves failed candidates and uses WAN22_LPIPS_THRESHOLD = 0.25.
  • The test is listed in tests/integration/test_lists/test-db/l0_b200.yml.
  • The reported LPIPS run passed. Coverage verdict: sufficient.

Per-File QA Perspective

  • tests/integration/defs/examples/visual_gen/golden/visual_gen_lpips/wan22_t2v_lpips_golden_video.json: Sets the Wan 2.2 LPIPS limit to 0.25. QA should verify single-GPU output against this limit.
  • tests/integration/defs/examples/visual_gen/test_visual_gen_wan.py: Applies the Wan 2.2-specific threshold and retains failed candidates. The test is listed in test-db/l0_b200.yml.
  • tests/integration/test_lists/waives.txt: Removes the Wan 2.2 LPIPS waiver for NVBugs 6535765. QA should verify that CI now enforces the test.

Description

Raise the single-GPU Wan 2.2 LPIPS threshold from 0.05 to 0.25 and remove the waiver for nvbug 6535765. Two fresh B200 runs measured 0.218467 against the existing golden. Model output, the golden video, and other tests’ thresholds are unchanged.

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.

@o-stoner
Raise Wan 2.2 LPIPS threshold to 0.25 and unwaive
27cf743 
Signed-off-by: Olivia Stoner <245287810+o-stoner@users.noreply.github.com>
@o-stoner

o-stoner commented Sep 21, 2026

Copy link
Copy Markdown
Collaborator Author

/bot run --disable-fail-fast

@tensorrt-cicd

tensorrt-cicd commented Sep 21, 2026

Copy link
Copy Markdown
Collaborator

PR_Github #74853 [ run ] triggered by Bot. Commit: 27cf743 Link to invocation

@o-stoner
o-stoner marked this pull request as ready for review September 21, 2026 18:23
@o-stoner
o-stoner requested a review from a team as a code owner September 21, 2026 18:23
@coderabbitai

coderabbitai Bot commented Sep 21, 2026

Copy link
Copy Markdown
Contributor

Review Change StackReview Change Stack

Understand this PR’s impact

Explore downstream dependencies and potential security impact with Blast Radius.

View blast radius →

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info
⚙️ Run configuration

Configuration used: Repository: NVIDIA/TensorRT-LLM/.coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: d062c60a-c321-47f3-b1a7-3a1fdbe01d16

📥 Commits

Reviewing files that changed from the base of the PR and between 6d0afd0 and 27cf743.

📒 Files selected for processing (3)
  • tests/integration/defs/examples/visual_gen/golden/visual_gen_lpips/wan22_t2v_lpips_golden_video.json
  • tests/integration/defs/examples/visual_gen/test_visual_gen_wan.py
  • tests/integration/test_lists/waives.txt
💤 Files with no reviewable changes (1)
  • tests/integration/test_lists/waives.txt

Included review availability: Your plan provides up to 12 included reviews per hour; 11 remain after this review.

Walkthrough

The Wan 2.2 T2V LPIPS golden-video threshold increased to 0.25. The test now uses a dedicated threshold, preserves failed candidates, and runs without its previous skip entry.

Changes

Wan 2.2 LPIPS validation

Layer / File(s) Summary
Update threshold and enable test
tests/integration/defs/examples/visual_gen/golden/visual_gen_lpips/wan22_t2v_lpips_golden_video.json, tests/integration/defs/examples/visual_gen/test_visual_gen_wan.py, tests/integration/test_lists/waives.txt		 The golden-video threshold increased to 0.25. The Wan 2.2 T2V test uses WAN22_LPIPS_THRESHOLD for failure preservation and validation. Its skip entry and bug reference were removed.

Priority: ⬇️ Low

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

Change: Bug fix

Suggested reviewers: bowenfu

🚥 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%. Docstring coverage is scoped to functions touched by this diff. Analyzed 1 functions across 1 files. (1 skipped: 1 … 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 follows the required NVBugs and type format. It clearly summarizes the threshold update and waiver removal.
Description check ✅ Passed The description explains the change, its rationale, measured LPIPS result, affected test, and test evidence. It includes the required template sections and checklist.
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.
Full details: Docstring Coverage

Explanation

Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 1 functions across 1 files. (1 skipped: 1 unsupported.)

  • Fix all pre-merge checks with AI
✨ Finishing Touches 💡 1
🛠️ Fix failing CI checks 💡
  • Commit to this branch
  • Create a new PR
🧪 Generate unit tests (beta)
  • Create a new PR

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

@tensorrt-cicd

tensorrt-cicd commented Sep 21, 2026

Copy link
Copy Markdown
Collaborator

PR_Github #74853 [ run ] completed with state SUCCESS. Commit: 27cf743
/LLM/main/L0_MergeRequest_PR pipeline #61620 completed with status: 'SUCCESS'

CI Report

Link to invocation

yibinl-nvidia
yibinl-nvidia reviewed Sep 22, 2026
View reviewed changes
Comment thread tests/integration/test_lists/waives.txt
@github-actions

github-actions Bot commented Sep 22, 2026

Copy link
Copy Markdown

Automatically added "ci: full pre-merge approved" because this PR has satisfied the required GitHub review approvals. Unresolved review conversations and other required checks remain independent merge requirements.

@o-stoner
o-stoner merged commit 143cea4 into NVIDIA:main Sep 22, 2026
21 of 22 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@yibinl-nvidia yibinl-nvidia yibinl-nvidia approved these changes
@karljang karljang Awaiting requested review from karljang karljang is a code owner automatically assigned from NVIDIA/trt-llm-torch-visual-gen-devs
@luyiyun1021 luyiyun1021 Awaiting requested review from luyiyun1021 luyiyun1021 is a code owner automatically assigned from NVIDIA/trt-llm-torch-visual-gen-devs

Assignees

@o-stoner o-stoner

Labels

ci: full pre-merge approved VisualGen

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@o-stoner @tensorrt-cicd @yibinl-nvidia @trtllm-agent