david/flash-attention

镜像来自 https://github.com/Dao-AILab/flash-attention

作者	SHA1 備註	提交日期
Tri Dao	2c996ca25f Use SeqlenInfo for bwd and epilogue	1 周之前
Tri Dao	29cdfedd80 Use Bulk reduce instead of TMA for dQaccum, split across WGs	1 周之前
Tri Dao	82dc825759 Don't use the unsafe convert_type function	2 周之前
Tri Dao	6e8b25e426 Refactor	2 月之前
Ying Zhang	db80387343 Add seqused_q in fwd / bwd and seqused_k in bwd.	3 月之前
Cameron Shinn	3cea2fb6ee Add ArchTag to pre/postprocess bwd kernels (#1180)	3 月之前
Tri Dao	bafe253042 [FA3] Bwd	4 月之前