flash-attn2: [XPU] Fix the transformers UT bugs #59

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.

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

Closed

YangKai0616 wants to merge 8 commits into huggingface:main from YangKai0616:main

Closed

flash-attn2: [XPU] Fix the transformers UT bugs #59

YangKai0616 wants to merge 8 commits into huggingface:main from YangKai0616:main

Conversation

@YangKai0616

Copy link

Contributor

@YangKai0616 YangKai0616 commented Oct 30, 2025

This PR resolves some bugs encountered when integrating XPU flash-attn2 into transformers:

Add regular parameter check;
Fix the issue of inconsistency between input tensor and kernel workspace when using multiple XPUs;
Make the return value continuous.

YangKai0616 added 6 commits

October 28, 2025 17:40

@YangKai0616


 Support multiple XPU loading models

3b33048

@YangKai0616


 Update flake.lock

032fa30

@YangKai0616

Fix

9fdbe7d

@YangKai0616


 Update flake.lock

c0cb2ea

@YangKai0616


 Add check func and fix multi-XPU BUG

ed83412

@YangKai0616


 Merge branch 'main' into main

dbc3166

@YangKai0616 YangKai0616 changed the title ~~(削除) [XPU] Fix the transformers UT bugs (削除ここまで)~~ (追記) flash-attn2: [XPU] Fix the transformers UT bugs (追記ここまで)

Oct 30, 2025

YangKai0616 added 2 commits

October 30, 2025 13:58

@YangKai0616