Skip to content

Pull requests: PaddlePaddle/flash-attention

New pull request New
13 Open 81 Closed
13 Open 81 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

fine-tuned tile size & regitser for fwd_hdim64
#92 opened Nov 14, 2025 by xxyux Loading…
Removed redundant templates and related compile-time/runtime code
#91 opened Nov 14, 2025 by Enigmatisms Loading…
1 task
Speed up compilation​
#88 opened Nov 12, 2025 by baoqiwen Loading…
fix fa2 flashmask oob read
#67 opened Jun 26, 2025 by umiswing • Draft
[WIP] fa3 varlen fix int32 overflow
#65 opened Jun 19, 2025 by umiswing Loading…
scan from right to left and skip masked block for each row at kernel begin
#55 opened Sep 23, 2024 by GuoxiaWang Loading…
optimize skip block calculate in bwd
#49 opened Aug 28, 2024 by GuoxiaWang Loading…
[BugFix] fix_mask error using unpadding api
#41 opened Apr 23, 2024 by wwbitejotunn Loading…
1
Fix unpadding input with padding mask compute error
#38 opened Apr 15, 2024 by wwbitejotunn Loading…
1
Fa cmake extends op
#31 opened Dec 14, 2023 by AnnaTrainingG Loading…
1
Fa cmake
#29 opened Dec 6, 2023 by AnnaTrainingG Loading…
31
[WIP]Sparse seqparallel
#9 opened Jun 8, 2023 by zkh2016 Loading…
1
add block sparse api
#7 opened May 27, 2023 by kuizhiqing Loading…
1
ProTip! Add no:assignee to see everything that’s not assigned.