The reviewed record of science sign in
Pith

arxiv: 2005.01279 · v1 · pith:NJQBT7JN · submitted 2020-05-04 · cs.CL · cs.LG

Improving Adversarial Text Generation by Modeling the Distant Future

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:NJQBT7JNrecord.jsonopen to challenge →

classification cs.CL cs.LG
keywords textgenerationfocusadversarialaforementionedalleviateapplyapproach
0
0 comments X
read the original abstract

Auto-regressive text generation models usually focus on local fluency, and may cause inconsistent semantic meaning in long text generation. Further, automatically generating words with similar semantics is challenging, and hand-crafted linguistic rules are difficult to apply. We consider a text planning scheme and present a model-based imitation-learning approach to alleviate the aforementioned issues. Specifically, we propose a novel guider network to focus on the generative process over a longer horizon, which can assist next-word prediction and provide intermediate rewards for generator optimization. Extensive experiments demonstrate that the proposed method leads to improved performance.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.