Skip to content

Support Motif #5

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Open
wants to merge 8 commits into
base: main
Choose a base branch
from
Open

Support Motif #5

wants to merge 8 commits into from

Conversation

@WyldeCat
Copy link
Collaborator

@WyldeCat WyldeCat commented Oct 22, 2025 •
edited by github-actions bot
Loading

  • This branch is intended for submission to the original vLLM repository.
  • Validation on 1% of MMLU-Pro.
Final Results:
Model: motif-small
Total samples: 1203
Correct answers: 910
Overall accuracy: 75.64%

Category-wise Results:
  business: 9.25% (73/789)
  law: 5.81% (64/1101)
  psychology: 9.02% (72/798)
  biology: 6.97% (50/717)
  chemistry: 7.07% (80/1132)
  history: 7.09% (27/381)
  other: 7.25% (67/924)
  health: 8.19% (67/818)
  economics: 7.70% (65/844)
  math: 8.73% (118/1351)
  physics: 7.78% (101/1299)
  computer science: 10.00% (41/410)
  philosophy: 5.01% (25/499)
  engineering: 6.19% (60/969)
  • Generation throughput fluctuates between 1500 ~ 3000 tokens/s, possibly depending on the context length.
    • used --batch-size 128

@ca1207

This comment was marked as resolved.

Sign in to view
@WyldeCat

This comment was marked as resolved.

Sign in to view
@ca1207
Copy link

ca1207 commented Oct 23, 2025 •
edited
Loading

docs/models/supported_models.md
tests/models/registry.py

should modify above files
ref (https://github.com/vllm-project/vllm/pull/25866/files)

@WyldeCat
Revert "[Chore] Remove unused PolyNorm layer (vllm-project#27110)"
e642b14 
This reverts commit 3125d79.
Signed-off-by: WyldeCat <skan1543@gmail.com>
@WyldeCat WyldeCat force-pushed the feat/motif branch 3 times, most recently from 230bd2f to 52a3f87 Compare October 23, 2025 05:06
@WyldeCat
Copy link
Collaborator Author

WyldeCat commented Oct 23, 2025

@ca1207
Thanks for your review!

  • Added sign-off to revert commit
  • Updated docs/models/supported_models.md, tests/models/registry.py

ca1207
ca1207 reviewed Oct 23, 2025
View reviewed changes
ca1207
ca1207 reviewed Oct 23, 2025
View reviewed changes
@ca1207
Copy link

ca1207 commented Oct 23, 2025

Should GroupedDiffAttn support for rocm be part of this pr?

@WyldeCat
Add GroupedDifferentialAttentionBackend
a9ad347 
Signed-off-by: WyldeCat <skan1543@gmail.com>
@WyldeCat
Support Motif model
894c3f0 
Signed-off-by: WyldeCat <skan1543@gmail.com>
@WyldeCat
Copy link
Collaborator Author

WyldeCat commented Oct 23, 2025

Should GroupedDiffAttn support for rocm be part of this pr?

I think separating PRs makes progress easier.

@ca1207
Copy link

ca1207 commented Oct 23, 2025

LGTM

ca1207 and others added 5 commits November 10, 2025 22:32
@ca1207
support tensor parallel for motif_small and poly_norm
2cbf5d8 
Signed-off-by: ca1207 <ca1207zzz@gmail.com>
@SageMoore
[EPLB] Refactor balance_packing to use numpy and optimize GPU-CPU tra…
798c7be 
…nsfers in EPLB (vllm-project#28369)

Signed-off-by: Sage Moore <sage@neuralmagic.com>
@ZJY0516
[Bugfix] fix qwen3-next crash (vllm-project#28202)
f0359ff 
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
@ca1207
Merge branch 'main' into feat/motif
5b7017a 
Signed-off-by: ca1207 <ca1207zzz@gmail.com>
@ca1207
Merge branch 'main' into feat/motif
4328b24
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@ca1207 ca1207 ca1207 left review comments

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@WyldeCat @ca1207 @SageMoore @ZJY0516