English

Implicit Unlikelihood Training: Improving Neural Text Generation with Reinforcement Learning

Computation and Language 2021-01-13 v1

Abstract

Likelihood training and maximization-based decoding result in dull and repetitive generated texts even when using powerful language models (Holtzman et al., 2019). Adding a loss function for regularization was shown to improve text generation output by helping avoid unwanted properties, such as contradiction or repetition (Li at al., 2020). In this work, we propose fine-tuning a language model by using policy gradient reinforcement learning, directly optimizing for better generation. We apply this approach to minimizing repetition in generated text, and show that, when combined with unlikelihood training (Welleck et al., 2020), our method further reduces repetition without impacting the language model quality. We also evaluate other methods for improving generation at training and decoding time, and compare them using various metrics aimed at control for better text generation output.

Keywords

Cite

@article{arxiv.2101.04229,
  title  = {Implicit Unlikelihood Training: Improving Neural Text Generation with Reinforcement Learning},
  author = {Evgeny Lagutin and Daniil Gavrilov and Pavel Kalaidin},
  journal= {arXiv preprint arXiv:2101.04229},
  year   = {2021}
}

Comments

accepted to EACL 2021

R2 v1 2026-06-23T22:02:43.703Z