load-inline-native-code

Helps write CUDA and HIP kernels using torch.utils.cpp_extension.load_inline(). Use when users want to write native GPU code (CUDA/HIP) inside a Python submission file.