david/flash-attention

spogulis no https://github.com/Dao-AILab/flash-attention

Autors	SHA1 Ziņojums	Datums
Jay Shah	0a1a0c22b6 refactor for split kv	3 mēneši atpakaļ
Ying Zhang	dff976a84a fixes	3 mēneši atpakaļ
jayhshah	5018ac6ac5 Fp8 kernel with "in-kernel" transpose of V in producer (#1100)	4 mēneši atpakaļ
Tri Dao	74b0761ff7 [FA3] BF16 forward	5 mēneši atpakaļ
Tri Dao	7f67966cc7 FA3 initial code release	5 mēneši atpakaļ