wasserstein-cyclegan

CycleGAN with Wasserstein Loss

Adds Wasserstein Loss to CycleGAN. Implementation builds upon the pytorch-CycleGAN implementation.

Status:

This is work in progress and the convergence of Wasserstein GAN is not great right now.

TODO:

• Add results.
• Use Improved Training of Wasserstein GAN.
• Experiment with different architectures.