Does Libtorch installation require CUDA and cuDNN to be downloaded and installed?

I need to install libtorch on my win10 laptop, and I was wondering if I still need to manually install CUDA and cuDNN (and set up the required environment variables?) after downloading the pre-compiled binary packages from the official website now??

The last time I tried to install Libtorch I looked up a lot of bits and pieces of tutorials, and I tried installing Cuda and cuDNN (and seemingly other things), and it didn’t turn out very well.

I just checked this thread (How to install cuDNN for PyTorch 2.1.0 with CUDA 12.1).

Does it mean that now I just need to download the binary package and then configure the directory for the .lib and header files within Visual Studio?

Is there a clear document/tutorial on how to install it if it still needs to be done manually?

The official site seems to only have a Python tutorial. The VS plugin is of little use to me, as I need to use libtorch in my existing OpenGL project.