Fetching the paper…

Parametric Adversarial Divergences are Good Losses for Generative Modeling · Around