/RevGAN

RevGAN implementation in PyTorch. We extend the Pix2pix and CycleGAN framework by exploring approximately invertible architectures in 2D and 3D. These architectures are approximately invertible by design and thus partially satisfy cycle-consistency before training even begins. Furthermore, since invertible architectures have constant memory complexity in depth, these models can be built arbitrarily deep without requiring additional memory. In the paper we demonstrate superior quantitative output on the Cityscapes and Maps datasets at near constant memory budget.

Primary LanguagePythonOtherNOASSERTION

Stargazers