Thank you @ptrblck for replying. I was able to solve it too. But I wanted to ask one more thing. In the at::Tensor backward_weight function it returns cudnn_convolution_backward_weight which is deprecated in the latest version of pytorch. I checked and I found that pytorch has now renamed the same function and is present in native library with name of cudnn_convolution_backward_weight. Can you please guide me how I can use this function with the latest pytorch version.
I looked at some relevant tutorials from REGISTERING A DISPATCHED OPERATOR IN C++ and EXTENDING TORCHSCRIPT WITH CUSTOM C++ OPERATORS.
Similar issue but with no solution till now. How to include native functions from ATen in cpp/CUDA extension?