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