Discrete Diffusion: Continuous-Time Markov Chains
This post, intended to be the first in a series related to discrete diffusion models, has been sitting in my drafts for months. I thought that Google’s release of Gemini Diffusion might be a good occasion to finally publish it. While discrete time Markov chains – sequences of random variables in which the past and future are independent given the present – are rather well known in machine learning, fewer people ever come across their continuous cousins. Given […]