ARAML: A Stable Adversarial Training Framework for Text Generation
EMNLP/IJCNLP (1), pp. 4270-4280, 2019.
We propose a novel adversarial training framework to deal with the instability problem of current generative adversarial networks for text generation
Most of the existing generative adversarial networks (GAN) for text generation suffer from the instability of reinforcement learning training algorithms such as policy gradient, leading to unstable performance. To tackle this problem, we propose a novel framework called Adversarial Reward Augmented Maximum Likelihood (ARAML). During adv...More
PPT (Upload PPT)