david/flash-attention

mirrorاز https://github.com/Dao-AILab/flash-attention

نویسنده	SHA1 پیام	تاریخ
Tri Dao	43ab0b5205 Mention that some CUDA extensions have only been tested on A100s	2 سال پیش
Tri Dao	2e33fc8e36 Add GPT and ViT models	2 سال پیش
Tri Dao	fa6d1ce44f Add fused_dense and dropout_add_layernorm CUDA extensions	2 سال پیش