RA2FD: Distilling Faithfulness into Efficient Dialogue Systems

Zhiyuan Zhu, Yusheng Liao, Chenxin Xu, Yunfeng Guan, Yanfeng Wang, Yu Wang


Abstract
Generating faithful and fast responses is crucial in the knowledge-grounded dialogue. Retrieval Augmented Generation (RAG) strategies are effective but are inference inefficient, while previous Retrieval Free Generations (RFG) are more efficient but sacrifice faithfulness. To solve this faithfulness-efficiency trade-off dilemma, we propose a novel retrieval-free model training scheme named Retrieval Augmented to Retrieval Free Distillation (RA2FD) to build a retrieval-free model that achieves higher faithfulness than the previous RFG method while maintaining inference efficiency. The core idea of RA2FD is to use a teacher-student framework to distill the faithfulness capacity of a teacher, which is an oracle RAG model that generates multiple knowledge-infused responses. The student retrieval-free model learns how to generate faithful responses from these teacher labels through sequence-level distillation and contrastive learning. Experiment results show that RA2FD let the faithfulness performance of an RFG model surpass the previous SOTA RFG baseline on three knowledge-grounded dialogue datasets by an average of 33% and even matching an RAG model’s performance while significantly improving inference efficiency. Our code is available at https://github.com/zzysjtuiwct/RA2FD.
Anthology ID:
2024.emnlp-main.685
Volume:
Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
Month:
November
Year:
2024
Address:
Miami, Florida, USA
Editors:
Yaser Al-Onaizan, Mohit Bansal, Yun-Nung Chen
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
12304–12317
Language:
URL:
https://aclanthology.org/2024.emnlp-main.685
DOI:
Bibkey:
Cite (ACL):
Zhiyuan Zhu, Yusheng Liao, Chenxin Xu, Yunfeng Guan, Yanfeng Wang, and Yu Wang. 2024. RA2FD: Distilling Faithfulness into Efficient Dialogue Systems. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 12304–12317, Miami, Florida, USA. Association for Computational Linguistics.
Cite (Informal):
RA2FD: Distilling Faithfulness into Efficient Dialogue Systems (Zhu et al., EMNLP 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.emnlp-main.685.pdf