Model Loading faliure when running on CPU

[WGraterol64]

On eval.py whenever your not running on a GPU accelerated environment you get an error on line 108 of eval.py, i.e, whenever you run model.load_state_dict(torch.load(modelPath)). The solution for this issue is to add the parameter map_location=torch.device('cpu') on the call if you're running on CPU.

The soluition code snippet would be

# loading model...
print('Loading model from: {}'.format(modelPath))
if args.cuda :
    model.load_state_dict(torch.load(modelPath))
else :
    model.load_state_dict(torch.load(modelPath, map_location=torch.device('cpu')))