We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computation and Language

Title: Text Generation by Learning from Off-Policy Demonstrations

Abstract: Current approaches to text generation largely rely on autoregressive models and maximum likelihood estimation. This paradigm leads to (i) diverse but low-quality samples due to mismatched learning objective and evaluation metric (likelihood vs. quality) and (ii) exposure bias due to mismatched history distributions (gold vs. model-generated). To alleviate these problems, we frame text generation as a reinforcement learning (RL) problem with expert demonstrations (i.e., the training data), where the goal is to maximize quality given model-generated histories. Prior RL approaches to generation often face optimization issues due to the large action space and sparse reward. We propose GOLD (generation by off-policy learning from demonstrations): an algorithm that learns from the off-policy demonstrations by importance weighting and does not suffer from degenerative solutions. We find that GOLD outperforms the baselines according to automatic and human evaluation on summarization, question generation, and machine translation, including attaining state-of-the-art results for CNN/DailyMail summarization. Further, we show that models trained by GOLD are less sensitive to decoding algorithms and the generation quality does not degrade much as the length increases.
Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Cite as: arXiv:2009.07839 [cs.CL]
  (or arXiv:2009.07839v1 [cs.CL] for this version)

Submission history

From: Richard Yuanzhe Pang [view email]
[v1] Wed, 16 Sep 2020 17:58:37 GMT (8214kb,D)
[v2] Wed, 3 Mar 2021 03:43:28 GMT (8236kb,D)

Link back to: arXiv, form interface, contact.