DP-GAN: Diversity-Promoting Generative Adversarial Network for Generating Informative and Diversified Text

TL;DR

DP-GAN introduces a novel adversarial framework that promotes diversity and informativeness in text generation by rewarding novel outputs and employing a language-model based discriminator, outperforming existing methods.

Contribution

The paper presents a new diversity-promoting GAN model with a language-model discriminator, addressing repetition and lack of informativeness in text generation.

Findings

Generated text is more diverse and informative.

Outperforms existing baselines in review and dialogue tasks.

Discriminator effectively distinguishes novel text without saturation.

Abstract

Existing text generation methods tend to produce repeated and "boring" expressions. To tackle this problem, we propose a new text generation model, called Diversity-Promoting Generative Adversarial Network (DP-GAN). The proposed model assigns low reward for repeatedly generated text and high reward for "novel" and fluent text, encouraging the generator to produce diverse and informative text. Moreover, we propose a novel language-model based discriminator, which can better distinguish novel text from repeated text without the saturation problem compared with existing classifier-based discriminators. The experimental results on review generation and dialogue generation tasks demonstrate that our model can generate substantially more diverse and informative text than existing baselines. The code is available at https://github.com/lancopku/DPGAN