3 ms·GANs Trained by a Two Time-Scale Update Rule Converge to a Nash Equilibrium1 points by albertzeyer 9y ago