Block sparse attention layer for PyTorch from Composable Kernels
Primary LanguageC++
No issues in this repository yet.