Elucidating the Design Space of Diffusion-Based Generative Models — mentioned by gwern

Wed, 23 Aug 2023 20:07:51 GMT

No, I'm not. BigGAN did fine on scaling up to JFT-300M with basically no changes beyond model size and a simple architecture. This is also what we were observing, even with a buggy BigGAN implementation. GigaGAN is the best quality, but that's mostly because it's also the biggest; as Table 1 shows most of the gains come from various kinds of additional scaling. (And this is moving the goalposts from the usual assertion that "GANs can't scale" to "they're harder to scale"; note the self-fulfilling nature of such assertions. Considering how there is next to no GAN scaling research, these…

Paper and discussion history. A mention is not an endorsement.

Elucidating the Design Space of Diffusion-Based Generative Models · HackerPapers discussions

Elucidating the Design Space of Diffusion-Based Generative Models — mentioned by gwern