Very simple but very strange type error

I have encountered a very strange error with the datatypes. I am very new to PyTorch, I apologize if this is a simple mistake.

The minimal version of the error is:

import torch as torch
class Net(torch.nn.Module):
def init(self):
super(Net, self).init()
self.conv1 = torch.nn.Conv2d(in_channels=1,out_channels=16,kernel_size=5)
self.pool = torch.nn.MaxPool2d(2, 2)
self.conv2 = torch.nn.Conv2d(in_channels=16,out_channels=16,kernel_size=5)
self.dense1 = torch.nn.Linear(16 * 145 * 95, 512)
self.dense2 = torch.nn.Linear(512, 168)

def forward(self, x):
    x = self.pool(torch.nn.functional.relu(self.conv1(x)))
    x = self.pool(torch.nn.functional.relu(self.conv2(x)))
    x = x.view(-1, 16 * 145 * 95)
    x = torch.nn.functional.relu(self.fc1(x))
    x = self.fc2(x)
    x = x.view(-1,2)
    return x

CNN=Net()
c=torch.zeros((2,1,190,290),dtype=torch.double)
CNN©

I have tried all different versions of: adding .float() to the network, adding .double() to the inputs, etc. They all didn’t change the error message.
The full error message is displayed below.

Thank you so much in advance.

-Core Park

===== HYPERPARAMETERS =====
batch_size= 32
epochs= 5
learning_rate= 0.001

torch.DoubleTensor

RuntimeError Traceback (most recent call last)
in ()
1 CNN = Net()
----> 2 trainNet(CNN, batch_size=32, n_epochs=5, learning_rate=0.001)

in trainNet(net, batch_size, n_epochs, learning_rate)
38 #Forward pass, backward pass, optimize
39 print(inputs.type())
—> 40 outputs = net(inputs)
41 loss_size = loss(outputs, labels)
42 loss_size.backward()

/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/torch/nn/modules/module.py in call(self, *input, **kwargs)
530 result = self._slow_forward(*input, **kwargs)
531 else:
–> 532 result = self.forward(*input, **kwargs)
533 for hook in self._forward_hooks.values():
534 hook_result = hook(self, input, result)

in forward(self, x)
9
10 def forward(self, x):
—> 11 x = self.pool(torch.nn.functional.relu(self.conv1(x)))
12 x = self.pool(torch.nn.functional.relu(self.conv2(x)))
13 x = x.view(-1, 16 * 145 * 95)

/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/torch/nn/modules/module.py in call(self, *input, **kwargs)
530 result = self._slow_forward(*input, **kwargs)
531 else:
–> 532 result = self.forward(*input, **kwargs)
533 for hook in self._forward_hooks.values():
534 hook_result = hook(self, input, result)

/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/torch/nn/modules/conv.py in forward(self, input)
343
344 def forward(self, input):
–> 345 return self.conv2d_forward(input, self.weight)
346
347 class Conv3d(_ConvNd):

/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/torch/nn/modules/conv.py in conv2d_forward(self, input, weight)
340 _pair(0), self.dilation, self.groups)
341 return F.conv2d(input, weight, self.bias, self.stride,
–> 342 self.padding, self.dilation, self.groups)
343
344 def forward(self, input):

RuntimeError: Expected object of scalar type Double but got scalar type Float for argument #3 ‘mat1’ in call to th_addmm

I think this has nothing to do with the input data, the data types of the tensors in the network don;t match each other I guess?

PyTorch uses float32 by default, while you are creating an input tensor with float64 (double).
If you don’t specify the dtype or set it to torch.float, the code should work.

Thank you very much, I indeed realize that.

Thank you.