1: \begin{abstract}

2: We propose a new equilibrium enforcing method paired with a loss derived

3: from the Wasserstein distance for training auto-encoder based Generative

4: Adversarial Networks. This method balances the generator and discriminator

5: during training. Additionally, it provides a new approximate convergence

6: measure, fast and stable training and high visual quality. We also

7: derive a way of controlling the trade-off between image diversity

8: and visual quality. We focus on the image generation task, setting

9: a new milestone in visual quality, even at higher resolutions. This

10: is achieved while using a relatively simple model architecture and

11: a standard training procedure. 

12: \end{abstract}

13: