Closed taewookim closed 6 years ago
I would assume, that it is possible to run on CPU only. I am not proficient in pytorch myself, but I think removing all model = torch.nn.DataParallel(model).cuda()
and .cuda()
in the code should do the trick.
just remove all .cuda(), and keep model = torch.nn.DataParallel(model).
which parts would I have to modify to run the model on CPU only? (Sorry, relatively new to pytorch)