Skip to content
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

Fix arguments of plan for split QK/VO head dims #795

Merged
merged 2 commits into from
Feb 7, 2025

Conversation

abmfy
Copy link
Contributor

@abmfy abmfy commented Feb 7, 2025

#765 introduced changes to the API of plan, including renaming head_dim to head_dim_qk and adding head_dim_vo. However, some calling sites were not updated to reflect these changes, resulting in failing unit tests.

This PR addresses the issue by updating the relevant calls, which should resolve the following unit test failures after merging:

  • tests/test_block_sparse.py::test_block_sparse_attention
  • tests/test_non_contiguous_prefill.py::test_batch_paged_prefill_packed_input

Copy link
Collaborator

@yzh119 yzh119 left a comment

Choose a reason for hiding this comment

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

Thanks for the update, yes we should update the sparse attention plan functions.

@yzh119 yzh119 merged commit 4127635 into flashinfer-ai:main Feb 7, 2025
@abmfy abmfy deleted the fix-sparse branch February 7, 2025 16:48
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