| | |
Mimicry is a lightweight PyTorch library aimed towards the reproducibility of GAN research.
Comparing GANs is often difficult - mild differences in implementations and evaluation methodologies can result in huge performance differences. Mimicry aims to resolve this by providing: (a) Standardized implementations of popular GANs that closely reproduce reported scores; (b) Baseline scores of GANs trained and evaluated under the same conditions; (c) A framework for researchers to focus on implementation of GANs without rewriting most of GAN training boilerplate code, with support for multiple GAN evaluation metrics.
We provide a model zoo and set of to benchmark different GANs of the same model size trained under the same conditions, using multiple metrics. To ensure , we verify scores of our implemented models against reported scores in literature.