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	94657af3e8 Add option for not doing intra-WG overlapping of gemm and softmax	1 月之前
Tri Dao	fe412d6b36 Redo rotary when contiguous	1 月之前
Tri Dao	b2d3fe92ff Move rotary to a separate file	1 月之前
Tri Dao	4d00645c76 Implement appending new KV to KV cache	1 月之前
Tri Dao	d00b88ee05 Move PagedKV to a separate file	1 月之前