I am looking for a GAN paper I have read a while ago, but unfortunately cannot find it again. I think it compared GANs and other methods (like CVAEs) w.r.t. how they handle multi-modal data, not sure about the CVAEs though. What I know is that they created different synthetic toy datasets with multiple modes to analyze this. I remember one plotting of a blue spiral of this data on a white background. Any guesses?