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

February 05, 2018 · Entered Twilight · 🏛 Conference on Empirical Methods in Natural Language Processing

"Last commit was 7.0 years ago (≥5 year threshold)"

Evidence collected by the PWNC Scanner

Repo contents: README.md, __init__.py, batcher.py, batcher_discriminator.py, data.py, discriminator.py, generated_sample.py, inspect_checkpoint.py, main.py, model.py, result_evaluate.py, review_generation_dataset, run.sh, util.py

Authors Jingjing Xu, Xuancheng Ren, Junyang Lin, Xu Sun arXiv ID 1802.01345 Category cs.CL: Computation & Language Citations 151 Venue Conference on Empirical Methods in Natural Language Processing Repository https://github.com/lancopku/DPGAN ⭐ 147 Last Checked 1 month ago

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