Understand What LLM Needs: Dual Preference Alignment for Retrieval-Augmented Generation
Dong, Guanting, Zhu, Yutao, Zhang, Chenghao, Wang, Zechen, Dou, Zhicheng, Wen, Ji-Rong
–arXiv.org Artificial Intelligence
Retrieval-augmented generation (RAG) has demonstrated effectiveness in mitigating the hallucination problem of large language models (LLMs). However, the difficulty of aligning the retriever with the diverse LLMs' knowledge preferences inevitably poses an inevitable challenge in developing a reliable RAG system. To address this issue, we propose DPA-RAG, a universal framework designed to align diverse knowledge preferences within RAG systems. Specifically, we initially introduce a preference knowledge construction pipline and incorporate five novel query augmentation strategies to alleviate preference data scarcity. Based on preference data, DPA-RAG accomplishes both external and internal preference alignment: 1) It jointly integrate pair-wise, point-wise, and contrastive preference alignment abilities into the reranker, achieving external preference alignment among RAG components. 2) It further introduces a pre-aligned stage before vanilla Supervised Fine-tuning (SFT), enabling LLMs to implicitly capture knowledge aligned with their reasoning preferences, achieving LLMs' internal alignment. Experimental results across four knowledge-intensive QA datasets demonstrate that DPA-RAG outperforms all baselines and seamlessly integrates both black-box and open-sourced LLM readers. Further qualitative analysis and discussions also provide empirical guidance for achieving reliable RAG systems. Our code is publicly available at https://github.com/dongguanting/DPA-RAG.
arXiv.org Artificial Intelligence
Jun-26-2024
- Country:
- South America > Brazil
- São Paulo (0.04)
- North America
- United States
- New York (0.04)
- Washington > King County
- Seattle (0.14)
- Louisiana > Orleans Parish
- New Orleans (0.04)
- Hawaii > Honolulu County
- Honolulu (0.04)
- Georgia > Fulton County
- Atlanta (0.04)
- Puerto Rico > San Juan
- San Juan (0.04)
- Mexico > Mexico City
- Mexico City (0.04)
- Canada
- Quebec > Montreal (0.04)
- Ontario > Toronto (0.04)
- British Columbia > Metro Vancouver Regional District
- Vancouver (0.28)
- United States
- Europe
- Germany > Berlin (0.04)
- Ukraine > Crimea (0.04)
- Netherlands
- North Holland > Amsterdam (0.04)
- North Brabant (0.04)
- South Holland
- Denmark > Capital Region
- Copenhagen (0.04)
- United Kingdom > England
- West Midlands > Birmingham (0.04)
- Russia
- France > Auvergne-Rhône-Alpes
- Middle East > Republic of Türkiye
- Istanbul Province > Istanbul (0.04)
- Belgium > Brussels-Capital Region
- Brussels (0.04)
- Atlantic Ocean > Mediterranean Sea
- Aegean Sea > Sea of Marmara > Bosporus (0.04)
- Asia
- Indonesia > Bali (0.04)
- Singapore (0.04)
- India (0.04)
- Taiwan > Taiwan Province
- Taipei (0.04)
- Russia
- Myanmar > Tanintharyi Region
- Dawei (0.04)
- Middle East
- Jordan (0.04)
- UAE > Abu Dhabi Emirate
- Abu Dhabi (0.04)
- Republic of Türkiye > Istanbul Province
- Istanbul (0.04)
- China
- Shanghai > Shanghai (0.04)
- Beijing > Beijing (0.04)
- Hong Kong (0.04)
- Fujian Province > Xiamen (0.04)
- Chongqing Province > Chongqing (0.04)
- Africa
- South America > Brazil
- Genre:
- Research Report > New Finding (0.67)
- Industry:
- Media > Film (1.00)
- Leisure & Entertainment (1.00)
- Technology: