.. |
cutlass @ 756c351b49
|
74b0761ff7
[FA3] BF16 forward
|
5 сар өмнө |
flash_attn
|
40e534a7f6
Implement cache_leftpad
|
5 сар өмнө |
ft_attention
|
50896ec574
Make nvcc threads configurable via environment variable (#885)
|
9 сар өмнө |
fused_dense_lib
|
50896ec574
Make nvcc threads configurable via environment variable (#885)
|
9 сар өмнө |
fused_softmax
|
50896ec574
Make nvcc threads configurable via environment variable (#885)
|
9 сар өмнө |
layer_norm
|
50896ec574
Make nvcc threads configurable via environment variable (#885)
|
9 сар өмнө |
rotary
|
50896ec574
Make nvcc threads configurable via environment variable (#885)
|
9 сар өмнө |
xentropy
|
50896ec574
Make nvcc threads configurable via environment variable (#885)
|
9 сар өмнө |