Generative Adversarial Networks (GANs) are a promising approach for text generation that, unlike traditional language models (LM), does not suffer from the problem of ``exposure bias''. However, A major hurdle for understanding the potential of GANs for text generation is the lack of a clear evaluation metric. In this work, we propose to approximate the distribution of text generated by a GAN, which permits evaluating them with traditional probability-based LM metrics. We apply our approximation procedure on several GAN-based models and show that they currently perform substantially worse than state-of-the-art LMs. Our evaluation procedure promotes better understanding of the relation between GANs and LMs, and can accelerate progress in GAN-based text generation.
@article{arxiv.1810.12686,
title = {Evaluating Text GANs as Language Models},
author = {Guy Tevet and Gavriel Habib and Vered Shwartz and Jonathan Berant},
journal= {arXiv preprint arXiv:1810.12686},
year = {2019}
}