Closed rvdmaazen closed 4 years ago
The issue was related to the size of the images I was training the network on. The dataloader was outputting 128x128
resolution images while the generator (using a depth of 4) was outputting 16x16
resolution images. Increasing the depth solved the issue.
@rvdmaazen, Glad to know it was solved. Please feel free to reopen if you face any issue.
cheers :beers:! @akanimax
I am trying to train a
pg.ProGAN
module on my created dataset, however when trying to use thetrain
function I run into the following error:Looking at the complete error it seems that the shapes of the real samples and fake sample are not the same. I am using the following training code using
pro-gan-pth
version 2.1.1 installed usingpip
: