While taking the fast.ai course, I was learning about Variational Autoencoders (VAE) and began to wonder: Is it possible to represent the latent space as an image, and then reconstruct the original image from that representation? Well, here's the answer. |