An Extensible Plug-and-Play Method for Multi-Aspect Controllable Text Generation

TL;DR

This paper introduces a novel plug-and-play method for multi-aspect controllable text generation that effectively manages interference among multiple constraints, enabling extensibility to unseen aspect combinations with improved accuracy and quality.

Contribution

It proposes a trainable gating mechanism to normalize prefix interference, allowing flexible addition of new constraints without retraining, and unifies processing of categorical and free-form constraints.

Findings

Outperforms baselines in constraint accuracy and text quality

Reduces interference growth with trainable gates

Enables extensibility to unseen aspect combinations

Abstract

Recently, multi-aspect controllable text generation that controls the generated text in multiple aspects (e.g., sentiment, topic, and keywords) has attracted increasing attention. Although methods based on parameter efficient tuning like prefix-tuning could achieve multi-aspect controlling in a plug-and-play way, the mutual interference of multiple prefixes leads to significant degeneration of constraints and limits their extensibility to training-time unseen aspect combinations. In this work, we provide a theoretical lower bound for the interference and empirically found that the interference grows with the number of layers where prefixes are inserted. Based on these analyses, we propose using trainable gates to normalize the intervention of prefixes to restrain the growing interference. As a result, controlling training-time unseen combinations of aspects can be realized by simply concatenating corresponding plugins such that new constraints can be extended at a lower cost. In addition, we propose a unified way to process both categorical and free-form constraints. Experiments on text generation and machine translation demonstrate the superiority of our approach over baselines on constraint accuracy, text quality, and extensibility.