How to reach the exact source code of convolution and linear fucntions in Pytorch?

Hello, I wish to apply some changes in the plain multiplication operation of convolution and Linear layers of a Neural Network implemented by Pytorch. I know there are other topics asking the same thing and some of them got to sources. However, when I see those cpp source codes, they are irrelevant (for example some of the replies are: third_party/ideep/mkl-dnn/src/cpu/ , or ConvolutionMM2d.cpp, or aten/src/ATen/native/Convolution.cpp:804 , or …)

I even installed pytorch from source in DEBUG mode in order to find the source cpp implementation using GDB debugger, but nothing was found. I am fed up trying all the ways. It is so weird that you follow up a function and hope to finally find the plain convolution implementation, but reach some unknown strange functions at the end.

I would appreciate it if someone helps me. Thank you in advance.