Skip to content

[minimax_m3_vl] Gate the integration tests on total accelerator memory - #50025

Merged
tarekziade merged 1 commit into
mainfrom
fix-minimax-m3-vl-integration-offload
Oct 27, 2026
Merged

tarekziade merged 1 commit into
mainfrom
fix-minimax-m3-vl-integration-offload

Conversation

@tarekziade

@tarekziade tarekziade commented Oct 27, 2026 •

Copy link
Copy Markdown
Collaborator

🤗 Transformers CI

Fixes #50023

All 6 MiniMaxM3VLIntegrationTest tests fail on the daily multi-GPU CI (2×A10G, 44 GiB) with "weights offloaded to the disk, which needed to be re-saved". MiniMaxAI/MiniMax-M3-MXFP8 is 413 GiB of weights, so device_map="auto" pushes almost everything to disk, and the MoE weights can't be offloaded without re-saving. Passing an offload_folder would mean writing ~400 GB to disk per run, which isn't a realistic CI setup.

This gates the class with require_torch_accelerator_memory(memory=480) (total memory over all visible accelerators, as for longcat_flash and muse_spark), so it still runs on a node big enough for the model and skips elsewhere.

As a follow-up, we'll add a fast integration test using transformers-tmic so MiniMax-M3-VL keeps real coverage on regular CI.

MiniMaxAI/MiniMax-M3-MXFP8 is 413 GiB of weights. On smaller runners
`device_map="auto"` offloads the MoE weights to disk, which fails since
they need re-saving. Skip unless 480 GiB of accelerator memory is visible.

Fixes #50023

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@github-actions

Copy link
Copy Markdown
Contributor

[For maintainers] Suggested jobs to run (before merge)

run-slow: minimax_m3_vl

@IlyasMoutawwakil IlyasMoutawwakil left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM ! thanks !

@HuggingFaceDocBuilderDev

Copy link
Copy Markdown

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

@tarekziade

Copy link
Copy Markdown
Collaborator Author

run-slow: minimax_m3_vl

@tarekziade
tarekziade added this pull request to the merge queue Oct 27, 2026
Merged via the queue into main with commit e5da0ae Oct 27, 2026
41 of 42 checks passed
@tarekziade
tarekziade deleted the fix-minimax-m3-vl-integration-offload branch October 27, 2026 12:19
@github-actions

Copy link
Copy Markdown
Contributor

CI (Nvidia + AMD)

Nvidia · ✅ No new failures 🎉👏 · ⚙️ Workflow · 📊 Dashboard

AMD · ✅ No new failures 🎉👏 · ⚙️ Workflow · 📊 Dashboard

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

MiniMax M3 VL: all 6 integration tests fail — disk-offloaded MoE weights cannot be re-saved (non-safetensors format)

3 participants