README.md

model-kernels

Cuda kernels created with kernels library

attention-int8 : A high-performance INT8 fused attention kernel for diffusion transformers.