This repository is a faithful reimplementation of StyleGAN2-ADA in PyTorch, focusing on correctness, performance, and compatibility.Ĭorrectness* Full support for all primary training configurations.* Extensive verification of image quality, training curves, and quality metrics against the TensorFlow version.* Results are expected to match in all cases, excluding the effects of pseudo-random numbers and floating-point arithmetic. We also find that the widely used CIFAR-10 is, in fact, a limited data benchmark, and improve the record FID from 5.59 to 2.42.įor business inquiries, please contact press and other inquiries, please contact Hector Marinez at notes We expect this to open up new application domains for GANs.

We demonstrate, on several datasets, that good results are now possible using only a few thousand training images, often matching StyleGAN2 results with an order of magnitude fewer images. The approach does not require changes to loss functions or network architectures, and is applicable both when training from scratch and when fine-tuning an existing GAN on another dataset. We propose an adaptive discriminator augmentation mechanism that significantly stabilizes training in limited data regimes. Tero Karras, Miika Aittala, Janne Hellsten, Samuli Laine, Jaakko Lehtinen, Timo AilaĪbstract: Training generative adversarial networks (GAN) using too little data typically leads to discriminator overfitting, causing training to diverge. Training Generative Adversarial Networks with Limited Data StyleGAN2-ADA - Official PyTorch implementation

Visual studio code icon png 512x512