david/flash-attention

kopia lustrzana https://github.com/Dao-AILab/flash-attention

Autor	SHA1 Wiadomość	Data
Tri Dao	27f8f890df [FusedDense] Allocate lt_workspace on input device	1 rok temu
Tri Dao	dec4f2e910 [FusedDense] Set workspace size to 32M for Hopper and 4M for others	1 rok temu
Tri Dao	88173a1aaf [FusedDense] Support relu, rename FusedDenseGeluDense -> FusedMLP	1 rok temu
Tri Dao	e68ebbe89a Simplify FusedDense	2 lat temu
Tri Dao	fa6d1ce44f Add fused_dense and dropout_add_layernorm CUDA extensions	2 lat temu