通过分层过滤与两阶段生成，提升开放域RAG的答准率
HiFi-RAG: Hierarchical Content Filtering and Two-Pass Generation for Open-Domain RAG
- 分层过滤+两阶段生成：先用轻量模型筛选文档，再用强模型精修答案
- 在验证集上ROUGE-L达0.274（提升19.6%），DeBERTaScore达0.677（提升6.2%）
- 适合需要高精度问答的开放域应用，尤其擅长处理2025年后的新知识
开放域检索增强生成（RAG）面临检索文档中存在无关信息、生成答案与用户意图不一致的问题。本文提出HiFi-RAG（分层过滤RAG），是MMU-RAGent NeurIPS 2025竞赛文本到文本静态评测的优胜闭源系统。该方法采用多阶段流水线，超越传统基于嵌入的检索。利用Gemini 2.5 Flash（比Pro便宜4-6倍）完成查询重构、分层内容过滤和引用标注，保留Gemini 2.5 Pro用于最终答案生成。在MMU-RAGent验证集上，系统将ROUGE-L提升至0.274（+19.6%），DeBERTaScore达0.677（+6.2%）。在测试集Test2025（评估需2025年1月后知识的问题）上，相比参数化基线，ROUGE-L提升57.4%，DeBERTaScore提升14.9%。
原文摘要 · Abstract (English)
Retrieval-Augmented Generation (RAG) in open-domain settings faces significant challenges regarding irrelevant information in retrieved documents and the alignment of generated answers with user intent. We present HiFi-RAG (Hierarchical Filtering RAG), the winning closed-source system in the Text-to-Text static evaluation of the MMU-RAGent NeurIPS 2025 Competition. Our approach moves beyond standard embedding-based retrieval via a multi-stage pipeline. We leverage the speed and cost-efficiency of Gemini 2.5 Flash (4-6x cheaper than Pro) for query formulation, hierarchical content filtering, and citation attribution, while reserving the reasoning capabilities of Gemini 2.5 Pro for final answer generation. On the MMU-RAGent validation set, our system outperformed the baseline, improving ROUGE-L to 0.274 (+19.6%) and DeBERTaScore to 0.677 (+6.2%). On Test2025, our custom dataset evaluating questions that require post-cutoff knowledge (post January 2025), HiFi-RAG outperforms the parametric baseline by 57.4% in ROUGE-L and 14.9% in DeBERTaScore.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。