Improving VAE generations of multimodal data through data-dependent conditional priors

Open in new window