用网络流建模工具响应溯源，实现大模型生成结果的精细归因。
Atomic Information Flow: A Network Flow Model for Tool Attributions in RAG Systems
- 将工具输出和大模型调用拆解为不可分割的信息原子，构建有向流图追踪信息路径。
- 在HotpotQA上使小模型准确率从54.7%提升至82.71%，同时压缩87.52%上下文长度。
- 适合需要可解释性与高效推理的多工具协作大模型系统开发者使用。
许多基于工具的检索增强生成（RAG）系统缺乏精确追踪最终回答来源特定工具组件的机制——这一问题在系统扩展至复杂多智能体架构时尤为突出。本文提出原子信息流（AIF），一种基于图的网络流模型，将工具输出与大模型调用分解为原子：不可分割、自包含的信息单元。通过将大模型编排建模为从工具和大模型节点流向响应超汇点的有向信息流，AIF实现了对AI生成结果的细粒度归因度量。受网络流理论中最大流最小割定理启发，我们训练一个轻量级Gemma3（4B参数）语言模型作为上下文压缩器，利用离线计算的流信号近似最小割。实验发现，基础Gemma3-4B模型在HotpotQA上仅达54.7%准确率，略优于词袋基线（BM25）。经由AIF信号微调后，准确率提升至82.71%（+28.01点），同时实现87.52%（+1.85点）的上下文令牌压缩率，接近参数量近7倍大的Gemma3-27B性能。
原文摘要 · Abstract (English)
Many tool-based Retrieval Augmented Generation (RAG) systems lack precise mechanisms for tracing final responses back to specific tool components -- a critical gap as systems scale to complex multi-agent architectures. We present \textbf{Atomic Information Flow (AIF)}, a graph-based network flow model that decomposes tool outputs and LLM calls into atoms: indivisible, self-contained units of information. By modeling LLM orchestration as a directed flow of atoms from tool and LLM nodes to a response super-sink, AIF enables granular attribution metrics for AI explainability. Motivated by the max-flow min-cut theorem in network flow theory, we train a lightweight Gemma3 (4B parameter) language model as a context compressor to approximate the minimum cut of tool atoms using flow signals computed offline by AIF. We note that the base Gemma3-4B model struggles to identify critical information with \textbf{54.7\%} accuracy on HotpotQA, barely outperforming lexical baselines (BM25). However, post-training on AIF signals boosts accuracy to \textbf{82.71\%} (+28.01 points) while achieving \textbf{87.52\%} (+1.85\%) context token compression -- bridging the gap with the Gemma3-27B variant, a model nearly $7\times$ larger.
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。