Cann't add pytorch into cmake project

I am trying to install pytorch with cpp I have spend around 12 hour to get it done but i have a lot of problem my os is pop os when i get the following error:
CMake Error at /home/kemo/Downloads/cpp_libs/libtorch/share/cmake/Caffe2/public/cuda.cmake:185 (message):
PyTorch requires cuDNN 7 and above.
Call Stack (most recent call first):
/home/kemo/Downloads/cpp_libs/libtorch/share/cmake/Caffe2/Caffe2Config.cmake:92 (include)
/home/kemo/Downloads/cpp_libs/libtorch/share/cmake/Torch/TorchConfig.cmake:68 (find_package)
CMakeLists.txt:7 (find_package)

for cuDNN i have installed it and cuda and they are running well !

I guess the error is raised from here which indicates the cuDNN version is older than 7.0.0 as also mentioned in the error message.
Which cuDNN version are you using and does the build log show where cuDNN is found?

This where the cudnn located in my os

cat /usr/lib/cuda/include/cudnn_version.h | grep CUDNN_MAJOR -A 2
#define CUDNN_MAJOR 8
#define CUDNN_MINOR 1

#endif /* CUDNN_VERSION_H */

who can i make cmake to find it in this path !
if this is the problem!

This is my cmakeLists file

cmake_minimum_required(VERSION 3.0 FATAL_ERROR)

set(CMAKE_PREFIX_PATH  "/home/kemo/Downloads/cpp_libs/libtorch/share/cmake/Torch")
find_package(Torch REQUIRED )
add_executable(example-app main.cpp)
target_link_libraries(example-app "${TORCH_LIBRARIES}")
set_property(TARGET example-app PROPERTY CXX_STANDARD 14)

Does your build process also show the right folders or do you have multiple cuDNN installs?
Does find / -name cudnn_version.h return more than this 8.1.1 file?