opusviz

> @naysumyat > > A workaround: > > * Open up test.py > * Comment out `net.cuda()` and `img = img.cuda()` lines > * Add map_location argument to torch.load: `net.load_state_dict(torch.load(cp,...