Skip to content

[cherry-pick][release/v0.32.0] cherry-pick #9270: warn on TaylorSeer order>=2 and document the Ref2VA artifact margin - #9274

Merged
tzhouam merged 1 commit into
vllm-project:release/v0.32.0from
Dong1017:fix/minimax-h3-taylorseer-guidance-v032
Oct 24, 2026
Merged

tzhouam merged 1 commit into
vllm-project:release/v0.32.0from
Dong1017:fix/minimax-h3-taylorseer-guidance-v032

Conversation

@Dong1017

@Dong1017 Dong1017 commented Oct 23, 2026 •

Copy link
Copy Markdown
Contributor

Purpose

Backport of #9270 to release/v0.32.0 (mitigates #8805 on the release line).

Warn at calibrator-build time when a Cache-DiT config leaves the validated envelope: taylorseer_order above 2 (no validated window), or taylorseer_order == 2 with max_continuous_cached_steps above the validated window of 4 (localized dark artifact on 10 s Ref2VA videos, evidence in #8805). Document the validated cache window plus the Cache-DiT + DLO memory margin on 64 GB-class dies in the MiniMax-H3 recipe.

Test Plan

  • pytest tests/diffusion/cache/test_cache_dit.py — the four new warning tests (order 2 with the window above the validated 4 warns; order 2 at the validated window stays silent; order 3 warns even at the validated window; order 1 stays silent). Assertions check a stable warning fragment via caplog.records.
  • Recipe renders as Markdown; log-only behavior change.

Test Result

  • H200 (SM90, vLLM 0.32.0 + diffusers 0.40.0 env, matching requirements/common.txt, this branch): taylorseer subset 6 passed (the four new tests plus existing matches).
  • No numerical or latency impact: the warning is log-only, and the recipe is documentation.

🤖 Generated with Claude Code

@Dong1017
Dong1017 force-pushed the fix/minimax-h3-taylorseer-guidance-v032 branch 3 times, most recently from ea25c64 to 90189c8 Compare October 23, 2026 12:29
…acts

TaylorSeer order-2 extrapolation produced a localized translucent dark
artifact on 10s Ref2VA videos when the cache window extends beyond the
validated max_continuous_cached_steps=4 (same-seed A/B against the
uncached path, dense attention; independent controlled run and error
probe in issue comments; vllm-project#8805). Warn at calibrator build time when a
cache config leaves the validated envelope - order 2 with
max_continuous_cached_steps above the validated window, or any order
above 2 (which has no validated window) - and document the validated
window plus the Cache-DiT + DLO memory margin on 64GB-class dies in the
MiniMax-H3 recipe.

Signed-off-by: GUOGUO <xwdong1998@163.com>
@Dong1017
Dong1017 force-pushed the fix/minimax-h3-taylorseer-guidance-v032 branch from 90189c8 to 1c5206d Compare October 23, 2026 13:25
@Dong1017

Copy link
Copy Markdown
Contributor Author

@Gaohan123 — the main-branch PR (#9270) is merged; this is the corresponding release/v0.32.0 backport, kept in sync through both review rounds (same diff: the narrowed warning + recipe guidance; CI 6/6 green). Could you take it from here — review this backport, or let us know if the release line prefers picking it up through the batch cherry-pick route (like #9206/#9251), in which case we can close this one? Thanks.

@tzhouam
tzhouam marked this pull request as ready for review October 24, 2026 16:02
@tzhouam
tzhouam requested a review from Isotr0py as a code owner October 24, 2026 16:02
@tzhouam tzhouam added the ready label to trigger buildkite CI label Oct 24, 2026 — with ChatGPT Codex Connector
@chatgpt-codex-connector

Copy link
Copy Markdown

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ⚠️ Failed 2026-10-24T16:02:58.087719Z
We couldn’t complete this review. Comment @codex review to try again.
1c5206d Draft marked ready
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@chatgpt-codex-connector

Copy link
Copy Markdown

Codex usage limits have been reached for code reviews. Please check with the admins of this repo to increase the limits by adding credits.
Credits must be used to enable repository wide code reviews.

@vllm-omni-review-bot

Copy link
Copy Markdown

This PR appears to belong to: docs/design/module/cache_management.md, docs/design/module/diffusion/index.md.

Module owners: @david6666666 @Isotr0py @princepride

Routing: @david6666666 via module of the changed files, module named in the PR description, model owner; @Isotr0py via module of the changed files, module named in the PR description, CODEOWNERS; @princepride via module of the changed files, module named in the PR description

@Dong1017, please review your own changes and leave a short self-review comment describing what you checked. PRs without author self-review may not be assigned a reviewer.

Please take a look when you have a chance. If you would like an automated review, mention @vllm-omni-review-bot in a comment.

@vllm-omni-review-bot

Copy link
Copy Markdown

Omni ReviewBot triage note

Automated triage of commit 1c5206d2d1e8 produced:

  • Priority: high. Prompt maintainer attention is suggested.

These are automated triage suggestions only — the final decision belongs to the maintainers.

@tzhouam
tzhouam merged commit a0877c2 into vllm-project:release/v0.32.0 Oct 24, 2026
11 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ready label to trigger buildkite CI

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants