I've been diving into the literature of GANs, and quite early on, I was pretty convinced that WGAN-GPs were the way to go. The WGAN-GP architecture is, as far as I know, theoretically and empirically superior to both the traditional GAN architecture and the WGAN architecture. However, since 2020 (3 years after publication of WGAN-GP), the traditional GAN and WGAN architectures are still often used for new publications (wgan example).
What would be reasons why academics would still use the traditional GAN or WGAN architecture over the WGAN-GP architecture?
