flow-based deep generative models

A class of models like Real NVP, Glow, and MAF that use normalizing flows to model complex distributions. They are invertible, allowing for efficient likelihood computation and generation.