Training Diffusion-based Generative Models with Limited Data | Read Paper on Bytez