Error when running Celebahq (edge-to-face)

[fnzhan]

I follow the setting of 3) Celebahq (edge-to-face), only changing the image size to 128. The error is shown as below:

[fnzhan]

@panzhang0212 do you have any idea about this error? thanks.

[panzhang0212]

Hi, I run the code with resolution 128, but do not find any problem in the code. What's your command? if you run with 128, you need to add "--load_size 128 ---load_size 128" to the command. load_size can be larger than 128 (for exemple, 156) to do augmentation.

[panzhang0212]

This is the log

[fnzhan]

Hi, I run the code with resolution 128, but do not find any problem in the code. What's your command? if you run with 128, you need to add "--load_size 128 ---load_size 128" to the command. load_size can be larger than 128 (for exemple, 156) to do augmentation.

Thanks for your reply. For me, the problem is solved by setting the self.preprocess_input(data, ) to self.preprocess_input(data.copy(), ) in line 52 of pix2pix_model.py. https://github.com/microsoft/CoCosNet/blob/3db703a46f9969836037427e190ede8480c07ce1/models/pix2pix_model.py#L52

[Finger-tiao]

This method is effective in the training phase, but needs to be changed back to the data() in the testing phase. It‘s strange. Can someone explain why?