Aaron Lou personal blog

Reflected Diffusion Models

Diffusion models are trained to reverse a stochastic process through score matching. However, a lot of diffusion models rely on a small but critical implementation detail called thresholding. Thresholding projects the sampling process to the data support after each discretized diffusion step, stabilizing generation at…
Points
177
HN comments
48
Submissions
1
Cited by
0 commenters

AI & Machine LearningMath & Algorithms2023

On Hacker News

SubmittedTitle on HNPointsComments
May 8, 2023Reflected Diffusion Models by tim_sw17748

Related posts from the same era