RuntimeError with ResNetUNet

[prarobinson]

Hello!

When I get to the line

model = ResNetUNet(6)

I receive the following error message:

Traceback (most recent call last):
File "", line 1, in
File "", line 4, in init
File "/home/paulr/Programming/anaconda3/lib/python3.7/site-packages/torchvision/models/resnet.py", line 237, in resnet18
**kwargs)
File "/home/paulr/Programming/anaconda3/lib/python3.7/site-packages/torchvision/models/resnet.py", line 220, in resnet
model = ResNet(block, layers, **kwargs)
File "/home/paulr/Programming/anaconda3/lib/python3.7/site-packages/torchvision/models/resnet.py", line 142, in init
bias=False)
File "/home/paulr/Programming/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py", line 332, in init
False, pair(0), groups, bias, padding_mode)
File "/home/paulr/Programming/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py", line 46, in init
self.reset_parameters()
File "/home/paulr/Programming/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py", line 49, in reset_parameters
init.kaiming_uniform(self.weight, a=math.sqrt(5))
File "/home/paulr/Programming/anaconda3/lib/python3.7/site-packages/torch/nn/init.py", line 315, in kaiming_uniform
return tensor.uniform_(-bound, bound)
RuntimeError: a leaf Variable that requires grad has been used in an in-place operation.

This looks likeit might be a bug in pytorch (or maybe just how I'm using it), but if you have any pointers I'd be greatful.

Thanks,
P

[usuyama]

Hi @prarobinson thanks for trying this repo.
Which version of PyTorch are you using? I created this repo with PyTorch 0.4.1 so there might be some breaking changes with new versions.

[prarobinson]

Thanks for getting back to me. I am using a newer version of PyTorch, but what is most puzzling is that the notebook seems to execute fine locally, but when I run it at the command line I get the error I mentioned. I didn't see that the notebook used any special environment. I can try to rollback to 0.4.1, but was hoping you might have an idea of where in the code the in-line replacement is happening; I have a suspicion that this could be a quick fix, but I'm pretty new to PyTorch. Thanks, Paul

…

On Thu, Jun 4, 2020 at 4:05 AM Naoto Usuyama ***@***.***> wrote: Hi @prarobinson <https://github.com/prarobinson> thanks for trying this repo. Which version of PyTorch are you using? I created this repo with PyTorch 0.4.1 so there might be some breaking changes with new versions. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#12 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAGG6XZELDQ2N736WNTIGDDRU55YTANCNFSM4NJ2JIFA> .

[usuyama]

Hmm that's interesting. Can you try to repro on an online notebook and share? e.g. https://colab.research.google.com/