I just recently switch over to Google colab for the GPU, and I’m getting the following error for my notebook
I have set my NNs to cuda, as well as all the variables, but I’m still getting the error
---------------------------------------------------------------------------
RuntimeError Traceback (most recent call last)
<ipython-input-28-3334654f75a5> in <module>()
32 # loss for fake images
33 z = torch.randn(bs, 100).view(-1, 100, 1,1) # 100 is input channels for G
---> 34 fake_img = G(z)
35 outputs = D(fake_img).view(-1,1,1,1)
36 d_loss_fake = criterion(outputs, fake_labels)
/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py in __call__(self, *input, **kwargs)
475 result = self._slow_forward(*input, **kwargs)
476 else:
--> 477 result = self.forward(*input, **kwargs)
478 for hook in self._forward_hooks.values():
479 hook_result = hook(self, input, result)
<ipython-input-10-ff06956ca9a4> in forward(self, input)
42 self.tanh = nn.Tanh()
43 def forward(self, input):
---> 44 out = self.relu(self.bn1(self.deconv1(input)))
45 out = self.relu(self.bn2(self.deconv2(out)))
46 out = self.relu(self.bn3(self.deconv3(out)))
/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py in __call__(self, *input, **kwargs)
475 result = self._slow_forward(*input, **kwargs)
476 else:
--> 477 result = self.forward(*input, **kwargs)
478 for hook in self._forward_hooks.values():
479 hook_result = hook(self, input, result)
/usr/local/lib/python3.6/dist-packages/torch/nn/modules/conv.py in forward(self, input, output_size)
689 return F.conv_transpose2d(
690 input, self.weight, self.bias, self.stride, self.padding,
--> 691 output_padding, self.groups, self.dilation)
692
693
RuntimeError: Expected object of type torch.FloatTensor but found type torch.cuda.FloatTensor for argument #2 'weight'
NVM solved, one of the variables wasn’t set to CUDA. I forgot there were two z’s in the training