RBF kernel for Conv2D

Good morning,
we would like to implement a kernel/filter operation that uses a Radial Basis Function dot product RBF kernel, to plug it in the Conv2d function.
We don’t how where to start since the .cpp/.cu code is quite complex.
At present we tried to compile torch-2.4.0a0+git3c09c6b from scratch, following the README.md, but we had a problem with ninja:

filippo@filippoPC:~/Downloads/pytorch-main/pytorch$ python3 setup.py develop
– Building version 2.4.0a0+git3c09c6b
cmake --build . --target install --config Release – -j 12
ninja: error: loading ‘build.ninja’: No such file or directory

Many thanks for your time and best regards,
Filippo Portera

I solved the issue following another tutorial.