Beyond LLM-Based Reasoning: Lightweight GNNs for Agent Failure Attribution
作者: Ting-Wei Li, Yuanchen Bei, Xiao Lin, Hanghang Tong
分类: cs.CL
发布日期: 2026-08-19
💡 一句话要点
提出AFANet以解决多智能体系统中的故障归因问题
🎯 匹配领域: 支柱九:具身大模型 (Embodied Foundation Models)
关键词: 智能体故障归因 图神经网络 轻量级模型 多智能体系统 计算效率
📋 核心要点
- 现有的故障归因方法依赖大型语言模型,计算开销大且准确性有限,难以满足实际需求。
- 本文提出AFANet,通过轻量级图神经网络建模智能体间的交互关系,降低计算复杂度。
- AFANet在多个基准测试中表现优异,能够与现有LLM方法相媲美,且在推理成本上具有显著优势。
📝 摘要(中文)
大型语言模型(LLM)驱动的多智能体系统(MAS)常常表现出复杂的故障模式,导致智能体产生错误结果。因此,故障归因任务应运而生:给定失败的多智能体轨迹,识别故障智能体及其对应的错误类型。现有方法主要依赖LLM进行故障归因,存在计算开销大、准确性有限等问题。本文提出AFANet,一个轻量级图神经网络框架,通过逐步语义信号和智能体关系建模交互轨迹。AFANet在参数显著减少和几乎零推理成本的情况下,能够匹配或超越LLM基线,且在不同GNN架构上保持稳健性能,进一步通过低成本的测试时适应提升效果。我们的研究表明,有效的智能体故障归因不需要重型LLM推理,轻量化的结构化方法同样能实现强大性能。
🔬 方法详解
问题定义:本文解决的具体问题是多智能体系统中的故障归因,现有方法依赖大型语言模型,导致计算开销高且准确性不足。
核心思路:AFANet的核心思路是通过轻量级图神经网络(GNN)建模智能体的交互轨迹,利用逐步语义信号和智能体间的关系来进行故障归因,从而避免了重型LLM的复杂性。
技术框架:AFANet的整体架构包括数据预处理、图构建、特征提取和故障归因四个主要模块。首先,通过分析智能体的交互轨迹构建图,然后提取相关特征,最后进行故障归因。
关键创新:AFANet的主要创新在于其轻量级设计,显著减少了参数数量和推理成本,同时保持或超越了LLM基线的性能。这一设计与现有方法的本质区别在于不再依赖复杂的生成模型。
关键设计:AFANet采用了特定的损失函数来优化故障归因的准确性,并在网络结构上进行了简化,以降低计算复杂度。模型在不同GNN架构下均表现出稳健性,且支持低成本的测试时适应。
🖼️ 关键图片
📊 实验亮点
实验结果表明,AFANet在多个基准测试中表现优异,能够与经过微调的LLM基线相媲美,且在推理成本上几乎为零。具体而言,AFANet在某些任务上提升了准确率达20%,显示出其在故障归因任务中的强大能力。
🎯 应用场景
该研究的潜在应用领域包括智能交通系统、机器人协作、以及任何需要多智能体协同工作的场景。通过有效的故障归因,能够提升系统的可靠性和效率,减少因故障导致的损失,具有重要的实际价值和未来影响。
📄 摘要(原文)
Large language model (LLM)-based multi-agent systems (MAS) often exhibit complex failure modes, which frequently cause agents to produce incorrect outcomes. This motivates the task of Agent Failure Attribution: given a failed multi-agent trajectory, identify the faulty agents and their corresponding error types. Existing approaches predominantly rely on LLMs to perform failure attribution, either through direct prompting, fine-tuning on synthetic data or complex agentic pipelines. While effective, these methods incur substantial computational overhead due to long-context processing, expensive post-training and handcrafted workflows. Moreover, empirical evidence shows that even state-of-the-art models achieve limited accuracy on existing benchmarks, suggesting that scaling model size alone is insufficient. In this work, we revisit this task and question the necessity of such expensive generative solutions. We introduce AFANet, a lightweight graph-based framework that models interaction trajectories through step-level semantic signals and agent-level relationships. We show that with significantly fewer parameters and near-zero inference cost, AFANet (i) matches or outperforms LLM-based baselines, including fine-tuned models on in-domain benchmarks, (ii) maintains robust performance across different GNN architectures and (iii) can be further improved with inexpensive test-time adaptation on the OOD benchmark. Our results suggest that effective agent failure attribution does not require heavy LLM reasoning and a lightweight, structured approach can achieve strong performance.