Document Type

Conference Proceeding

Publication Title

Findings of the Association for Computational Linguistics: EMNLP 2022

Abstract

Maximum likelihood estimation (MLE) is the predominant algorithm for training text generation models. This paradigm relies on direct supervision examples, which is not applicable to many emerging applications, such as generating adversarial attacks or generating prompts to control language models. Reinforcement learning (RL) on the other hand offers a more flexible solution by allowing users to plug in arbitrary task metrics as reward. Yet previous RL algorithms for text generation, such as policy gradient (on-policy RL) and Q-learning (off-policy RL), are often notoriously inefficient or unstable to train due to the large sequence space and the sparse reward received only at the end of sequences. In this paper, we introduce a new RL formulation for text generation from the soft Q-learning (SQL) perspective. It enables us to draw from the latest RL advances, such as path consistency learning, to combine the best of on-/off-policy updates, and learn effectively from sparse reward. We apply the approach to a wide range of novel text generation tasks, including learning from noisy/negative examples, adversarial attacks, and prompt generation. Experiments show our approach consistently outperforms both task-specialized algorithms and the previous RL methods.

First Page

6998

Last Page

7020

DOI

10.18653/v1/2022.findings-emnlp.518

Publication Date

12-2022

Keywords

Computational linguistics, Learning algorithms, Maximum likelihood estimation

Comments

Archived thanks to ACL Anthology

License: CC by 4.0 DEED

Uploaded 29 January 2024

Download

Share

COinS