Tri Dao 68bf390920 Update Cutlass to fix mem fence пре 2 недеља
..
composable_kernel @ 13332998a4 88d1657a14 [AMD ROCm] Fix KVcache bug and improve performance (#1328) пре 2 месеци
cutlass @ c506e16788 68bf390920 Update Cutlass to fix mem fence пре 2 недеља
flash_attn 36dddb891c Remove unused 224 configs (#1425) пре 2 недеља
flash_attn_ck 1feb711f46 Fix compilation with clang on ARM64 (#1285) пре 1 месец
ft_attention 1feb711f46 Fix compilation with clang on ARM64 (#1285) пре 1 месец
fused_dense_lib 1feb711f46 Fix compilation with clang on ARM64 (#1285) пре 1 месец
fused_softmax 50896ec574 Make nvcc threads configurable via environment variable (#885) пре 10 месеци
layer_norm 1feb711f46 Fix compilation with clang on ARM64 (#1285) пре 1 месец
rotary 1feb711f46 Fix compilation with clang on ARM64 (#1285) пре 1 месец
xentropy 1feb711f46 Fix compilation with clang on ARM64 (#1285) пре 1 месец