Skip to content

Conversation

@YangKai0616
Copy link
Contributor

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

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

@YangKai0616 YangKai0616 changed the title [XPU] Fix the transformers UT bugs flash-attn2: [XPU] Fix the transformers UT bugs Oct 30, 2025
@YangKai0616 YangKai0616 marked this pull request as ready for review October 31, 2025 02:34
@YangKai0616
Copy link
Contributor Author

@drbh , I can compile this PR successfully using the local nix. please help review. Thanks!

@danieldk
Copy link
Member

danieldk commented Nov 3, 2025

CI running in #61

@danieldk
Copy link
Member

danieldk commented Nov 4, 2025

Merged in #61.

@danieldk danieldk closed this Nov 4, 2025
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.

2 participants