4 ms·
This sounds like you're describing energy-based models, not diffusion models.
by ziedaniel1 5y ago
This sounds like you're describing energy-based models, not diffusion models.
- ericjang 5y agoAh, thanks. I had mistakenly assumed the diffusion process here was comparable / drop-in replacement to Langevin dynamics used with energy models.
- ericjang 5y agoI went and read the paper in more detail. Yeah, my original comment was way off-base, except if you draw a parallel with normalizing flows as iterated refinement (similar to iterated de-noising), and see that the DDPM is a more unconstrained form. But at a surface level, there isn't a clear connection between DDPM and energy-based models.