TL;DR本文介绍了改进 GAN 训练的新方法，构建了一种包含标签条件的 GAN 变体来生成 128x128 分辨率的、具有全局一致性的照片级别合成图像，并展示了这些高分辨率样本比低分辨率样本具有更高的可区分性和多样性。
Abstract
Synthesizing high resolution photorealistic images has been a long-standing challenge in machine learning. In this paper we introduce new methods for the improved training of generative adversarial networks (GANs) for i