Class of generative models used to create high-quality, diverse samples of data by iteratively adding and then reversing noise.
Generality: 715
