Response Generation with Context-Aware Prompt Learning
Xiaodong Gu1,2, Kang Min Yoo2, Sang-Woo Lee2, 1School of Software, Shanghai Jiao Tong University 2NAVER AI Lab
Abstract
Pre-trained language models (PLM) have marked a huge leap in neural dialogue modeling. While PLMs are pre-trained on large-scale text corpora, they are usually fine-tuned on scarce dialogue data with specific domain knowledge and dialogue styles. However, tailoring the language models while fully utilizing prior knowledge in large pre-trained models remains a challenge. In this paper, we present a novel approach for pre-trained dialogue modeling that casts the dialogue generation problem as a prompt-learning task. Instead of fine-tuning on limited dialogue data, our approach, DialogPrompt, learns continuous prompt embeddings optimized for dialogue contexts, which appropriately elicit knowledge from the large pre-trained model. To encourage the model to better utilize the prompt embeddings, the prompt encodings are designed to be dynamically generated based on the input dialogue context. Experiments on popular conversation datasets show that our approach significantly outperforms the fine-tuning baseline and the generic prompt-learning methods. Furthermore, human evaluations strongly support the superiority of DialogPrompt in regard to response generation quality.
中文速览
对话系统长期面临一个痛点：大型预训练语言模型（Pre-trained Language Model, PLM）积累了海量知识，但在特定对话场景微调时，训练数据稀少，往往难以充分激活模型里已有的知识。为解决这一问题，研究者提出了 DialogPrompt，将对话生成转化为提示学习（prompt learning）任务——不再更新庞大的预训练模型参数，而是只训练一小段"提示嵌入"来引导模型输出合适的回复。关键创新在于设计了动态提示编码器（dynamic prompt encoder）：提示内容不是对所有对话一刀切，而是根据当前对话上下文即时生成，从而让提示真正感知语境、精准"召唤"模型中与当前话题相关的知识。在 DailyDialog 和 MultiWOZ 两个主流多轮对话数据集上，DialogPrompt 在自动评测指标和人工评测中均明显优于微调基线和通用提示学习方法，生成的回复更流畅、信息量更丰富。这项工作为如何低成本、高效率地把通用大模型适配到对话场景提供了一条新思路，对资源有限的实际应用场景尤具参考价值。
原文 arXiv:2111.02643;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2111.02643v5