Cannot run the code due to version issues

[prerna-grg]

@KupynOrest If i run your code using torch 0.3.1 , torchtext 0.2.3 , then I get the error AttributeError: module 'torch' has no attribute 'float32'

As suggested by various sources online, to solve this I change torch version to 0.4, then I get the error Unexpected running stats buffer(s) "model.23.running_mean" and "model.23.running_var" for InstanceNorm2d with track_running_stats=False. to which other sources say to change the torch version to 0.3.1

So I am stuck, Could you please tell what versions you used to run your code??

[SivanDoveh]

Undo all NINJA changes and it will work with torch 0.3.1. It worked for me

[KSJhon]

I have same problem. Did you solve that issue? If yes, could you tell in detail? I run deblurGAN on python3.5 + cuda8 + pytorch0.4.0.

[vlee-harmonicinc]

I also have this issue with pyTorch 0.3.1 and torchvision 0.2.2.post3 This fixed by downgrading torchvision to 0.2.0.