Diffusion Models are a class of latent variable generative models.
They create new data samples by gradually adding and then removing noise from a starting point, often pure noise. They are particularly effective at generating high-quality images and other data types
