After training the network for 20 epochs, the network will start generating decent images: Let's have a look at the images generated by the generator network.
After 20, 50, 150, and 200 epochs (from left to right), the images look as follows:
Each block contains a facade label, the generated photo, and the actual image, which are vertically stacked. I suggest that you train the network for 1,000 epochs. If everything goes well, after 1,000 epochs, the generator networks will start generating realistic images.