Hello World,
could anyone explain me the advantage of loading a PyTorch model written in python in C++ instead of writing the model, data handling etc. directly in the PyTorch C++ frontend? At this moment the latter option makes much mor sense to me since it is less effort. What do you think??