david/flash-attention

spegling av https://github.com/Dao-AILab/flash-attention

Upphovsman	SHA1 Meddelande	Datum
Neil Tenenholtz	7153673c1a Fix swiglu backwards return type (#1337)	2 månader sedan
Tri Dao	3557e0bb8f [MLP] Implement SwiGLU with torch jiterator	1 år sedan
Tri Dao	f1a73d0740 Run isort and black on python files	1 år sedan
Tri Dao	6f6e9a9aaf [FusedDense] Enable sqrelu activation in FusedMLP	1 år sedan