ECO Decoding: Entropy-Based Control for Controllability and Fluency in Controllable Dialogue Generation
This work addresses a key problem in chatbot development for achieving better attribute control without sacrificing response quality, representing an incremental improvement over existing weighted decoding methods.
The paper tackles the challenge of balancing controllability and fluency in controllable dialogue generation by proposing ECO decoding, which dynamically adjusts control strength based on entropy, resulting in improved controllability while maintaining fluency across datasets and models.
Controllable Dialogue Generation (CDG) enables chatbots to generate responses with desired attributes, and weighted decoding methods have achieved significant success in the CDG task. However, using a fixed constant value to manage the bias of attribute probabilities makes it challenging to find an ideal control strength that satisfies both controllability and fluency. To address this issue, we propose ECO decoding (Entropy-based COntrol), which dynamically adjusts the control strength at each generation step according to the model's entropy in both the language model and attribute classifier probability distributions. Experiments on the DailyDialog and MultiWOZ datasets demonstrate that ECO decoding consistently improves controllability while maintaining fluency and grammaticality, outperforming prior decoding methods across various models and settings. Furthermore, ECO decoding alleviates probability interpolation issues in multi-attribute generation and consequently demonstrates strong performance in both single and multi-attribute scenarios.