/WAGAN

Official implementation of Wasserstein Attentional GenerativeAdversarial Network

Primary LanguagePython

WAGAN

Official implementation of Wasserstein Attentional GenerativeAdversarial Network

we propose Wasserstein Attentional GenerativeAdversarial Network (WAGAN) to address the mode collapse problem by using a more stable GAN training process.Based on AttnGAN, we train our model by minimizing a dif-ferent loss which effectively estimate the Wasserstein distance between the real data distribution the gener-ator distribution