Skip to content

Fix: use cu_chunk instead of cu_seqlens for flash_attn during corner case where top-k equals 1 #26

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

Open
wants to merge 1 commit into
base: master
Choose a base branch
from

Conversation

yibozhong
Copy link

This PR fixes an inconsistency in MoBA. When top-k > 1, the self attn is calculated using cu_seqlens_q=self_attn_cu_seqlen, where self_attn_cu_seqlen = cu_chunk. Meanwhile when top-k = 1, the self attn is calculated using cu_seqlens_q=cu_seqlens, which behaves the same as standard attention. Because the current block is always chosen, the corner case should use cu_chunk instead of cu_seqlens.

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.

1 participant