when I open the code with jupyter notebook , it needs password?

In the original paper and the most stared TF implementation, the content loss and style loss should divide the feature map size (C,H,W).

Why don't you do this?

what is ” reg_loss ” stand for...

Hi,
Thanks for sharing code! I have a question regarding computation. It seems that the computation of perceptual loss is much heavier than MSE loss. In my experiment, using perceptual loss will make training much slower and batch size much smaller. Is it normal?

I've point the dataset to 'K:\downloads\coco2017\train2017'
and there is about 18GB img under this floder

however there is still reported
"RuntimeError: Found 0 images in subfolders of: K:\downloads\coco2017\train2017
Supported image extensions are: .jpg,.jpeg,.png,.ppm,.bmp,.pgm"

after running this command
"train_dataset = datasets.ImageFolder(DATASET, transform)"

so is there any solution to this problem?

RuntimeError Traceback (most recent call last)
in ()
20 x = x.cuda()
21
---> 22 y = transformer(x)
23 xc = Variable(x.data, volatile=True)
24

/home/tai/anaconda2/lib/python2.7/site-packages/torch/nn/modules/module.pyc in call(self, *input, **kwargs)
222 for hook in self._forward_pre_hooks.values():
223 hook(self, input)
--> 224 result = self.forward(*input, **kwargs)
225 for hook in self._forward_hooks.values():
226 hook_result = hook(self, input, result)

/home/tai/cvpr18/fast-neural-style/transformer_net.py in forward(self, X)
35 def forward(self, X):
36 in_X = X
---> 37 y = self.relu(self.in1(self.conv1(in_X)))
38 y = self.relu(self.in2(self.conv2(y)))
39 y = self.relu(self.in3(self.conv3(y)))

/home/tai/anaconda2/lib/python2.7/site-packages/torch/nn/modules/module.pyc in call(self, *input, **kwargs)
222 for hook in self._forward_pre_hooks.values():
223 hook(self, input)
--> 224 result = self.forward(*input, **kwargs)
225 for hook in self._forward_hooks.values():
226 hook_result = hook(self, input, result)

/home/tai/cvpr18/fast-neural-style/transformer_net.py in forward(self, x)
131 t = x.view(x.size(0), x.size(1), n)
132 print (t.size())
--> 133 mean = torch.mean(t, 2).unsqueeze(2).expand_as(x)
134 # Calculate the biased var. torch.var returns unbiased var
135 var = torch.var(t, 2).unsqueeze(2).expand_as(x) * ((n - 1) / float(n))

/home/tai/anaconda2/lib/python2.7/site-packages/torch/autograd/variable.pyc in expand_as(self, tensor)
723
724 def expand_as(self, tensor):
--> 725 return Expand.apply(self, (tensor.size(),))
726
727 def t(self):

/home/tai/anaconda2/lib/python2.7/site-packages/torch/autograd/_functions/tensor.pyc in forward(ctx, i, new_size)
109 # tuple containing torch.Size or a list
110 def forward(ctx, i, new_size):
--> 111 result = i.expand(*new_size)
112 ctx.num_unsqueezed = result.dim() - i.dim()
113 ctx.expanded_dims = [dim for dim, (expanded, original)

RuntimeError: The expanded size of the tensor (256) must match the existing size (32) at non-singleton dimension 2. at /opt/conda/conda-bld/pytorch_1502006348621/work/torch/lib/THC/generic/THCTensor.c:323