FD-RAG: Federated Dual-System Retrieval-Augmented Generation

Tianhao Gao, Kai Yang, Yiyang Li · 2026

Read FD-RAG: Federated Dual-System Retrieval-Augmented Generation by Tianhao Gao, Kai Yang, Yiyang Li (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-27432.txt.

Description

Tianhao Gao, Kai Yang, and Yiyang Li build a federated retrieval system for edge devices that cannot share raw data. Their FD-RAG framework learns adaptive hypergraphs locally, distills them into compact question-answer memories, and answers from memory when coverage suffices. It improves accuracy by about 8 percent while cutting latency more than eightfold against strong baselines.

Catalog record

CREATORTianhao Gao, Kai Yang, Yiyang Li
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE57031
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.27432

Browse the full library · About the Tower of Babel