This folder contains configurations of popular GANs and links to the corresponding papers.
Please note that we provide them as a starting point for the user and that they represent the best-effort implementation -- we do not guarantee that all the implementation details match exactly due to the vast number of design options. Given the sensitivity of GANs to design choices, hyperparameters and differences in platforms (GPUs/TPUs), a small differences might have a significant impact on the final results.