Skip to content

Pull requests: vllm-project/flash-attention

New pull request New
30 Open 124 Closed
30 Open 124 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

Fully yank dropout in vllm fork of FA2
#153 opened Jun 23, 2026 by janeyx99 Loading…
feat(fa4): fuse per-group fp8 output
#151 opened Jun 18, 2026 by carlyou Loading…
4
Sync upstream
#149 opened Jun 16, 2026 by MatthewBonanni Member Loading…
[Fix] Mark kcache and vcache as mutable in fwd_kvcache
#148 opened Jun 15, 2026 by remi-or Loading…
feat(cute): add SM90 FP8 KV support with in-kernel dequantization
#147 opened Jun 12, 2026 by qixiang-99 • Draft
fix: handle suffix-less runtime arch (sm_103/GB300) in SM 10.x gate
#146 opened Jun 11, 2026 by aoshen02 Loading…
SM100 dynamic causal
#144 opened Jun 10, 2026 by MatthewBonanni Member Loading…
Fix illegal memory access in FA2 varlen SplitKV early-exit LSE write
#139 opened May 18, 2026 by wangyxbh Loading…
Reapply #122
#137 opened May 6, 2026 by MatthewBonanni Member Loading…
SM100 tile size 64
#132 opened Apr 9, 2026 by MatthewBonanni Member • Draft
[Perf] SM103 tcgen05.ld.red for fused TMEM load + row-max in softmax
#131 opened Apr 9, 2026 by LopezCastroRoberto Loading…
Combine kernel: increase pipeline depth from 4 to 8 stages
#124 opened Mar 4, 2026 by jmkuebler Loading…
1
add support for newer CUDA archs (Spark/Thor)
#121 opened Feb 13, 2026 by askliar Loading…
Fix issues with async TP
#117 opened Feb 7, 2026 by LucasWilkinson Collaborator Loading…
Sync to upstream main 20260121
#114 opened Jan 22, 2026 by LucasWilkinson Collaborator Loading…
[Frontend] Add FP8 output quantization support to FlashAttention backend
#113 opened Jan 3, 2026 by sachinkumarsingh092 Loading…
1
19
FA2 support head sizes 40, 72, and 80
#108 opened Nov 14, 2025 by MatthewBonanni Member • Draft
[Kernel] add attention sinks for flash attention2
#103 opened Oct 19, 2025 by dudugong-gitch Loading…
1
Add DCP parameters
#92 opened Sep 16, 2025 by MatthewBonanni Member • Draft
2
Vllm_flash_attn_with_attention_weights
#88 opened Sep 11, 2025 by SiriusPaul Loading…
WIP stream k scheduling
#67 opened Apr 29, 2025 by LucasWilkinson Collaborator • Draft
Removed the assertion imposed on cu_seqlens_k and seqused_k
#59 opened Mar 29, 2025 by chenyang78 Loading…
19
fix: add "typename" prior to dependent type name
#54 opened Feb 28, 2025 by zhiweij1 Loading…
1
AMD ROCm Build
#41 opened Jan 29, 2025 by ProExpertProg • Draft
Add back flash_attn_func api (and support FA3) [Don't Merge Yet]
#40 opened Jan 26, 2025 by LucasWilkinson Collaborator Loading…
ProTip! Adding no:label will show everything without a label.