#include <pybind11/pybind11.h> #include <torch/extension.h> #include "gemm_cuda.h" PYBIND11_MODULE(TORCH_EXTENSION_NAME, m) { m.def("gemm_forward_cuda", &gemm_forward_cuda, "our sparse conv kernel"); }