search

aupendu / zero-restore

Zero-shot Single Image Restoration through Controlled Perturbation of Koschmieder's Model
MIT License
41 stars 5 forks source link

Implementing watermodel.py #6

Closed Yiting1009 closed 1 year ago

Yiting1009 commented 1 year ago

Hi, thanks for your great work! But I get some problems when I am running watermodel.py. This confuses me for a long time. I don't know how to solve it.

Traceback (most recent call last): File "UnderWaterZeroShot.py", line 168, in test(args) File "UnderWaterZeroShot.py", line 155, in test loss.backward(retain_graph=True) File "/mnt/petrelfs/zhangyiting/anaconda3/envs/nerf-pytorch/lib/python3.8/site-packages/torch/_tensor.py", line 363, in backward torch.autograd.backward(self, gradient, retain_graph, create_graph, inputs=inputs) File "/mnt/petrelfs/zhangyiting/anaconda3/envs/nerf-pytorch/lib/python3.8/site-packages/torch/autograd/init.py", line 173, in backward Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward passRuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [1, 64, 64, 64]], which is output 0 of ReluBackward0, is at version 1; expected version 0 instead. Hint: the backtrace further above shows the operation that failed to compute its gradient. The variable in question was changed in there or anywhere later. Good luck!

aupendu commented 1 year ago

nn.ReLU(inplace=True) -> nn.ReLU(inplace=False)

Yiting1009 commented 1 year ago

Thank you very much for your reply!