Diffusion Model | Engineering Wiki

Diffusion Model — A probabilistic generative model that learns to reverse a gradual noising process in order to generate samples.

Technical Context

A probabilistic generative model that learns to reverse a gradual noising process in order to generate samples.

Generation quality and inference cost trade off against the number of reverse-process steps and the sampling method.

Direct source: The primary paper or official specification for Diffusion Model is linked here for verification.