Generator not converging?

Hi guys!

I’m trying to implement Contex Conditional GAN (paper, github) as a means of data inpainting. It is my understanding of GANs that the losses of discriminator and generator should converge to approximately same value at one point. However, in my case the discriminator loss is converging to 0, while generator loss is converging to 1, as shown in the graph.

Am I doing something wrong or am I just misunderstanding how to gauge whether my network is fully trained or not? Thanks!