让大模型在真实图推理中稳定应对名称和表述变化。
GRAIN: Bridging Name and Narrative Shifts in Real-World Graph Reasoning through Invariance-Rewarded Agentic RL
- 用强化学习训练单智能体，通过结构不变性奖励提升文本到图的映射能力。
- 准确率比多智能体方法高16.45%，延迟降低24%。
- 适合需要鲁棒图推理的复杂场景，如知识图谱、动态系统建模。
尽管大型语言模型在标准图任务中表现良好，但在节点标识符和任务表述发生实际变化时仍易失效。确定性图工具对这类变化具有不变性，但大模型从噪声文本中提取拓扑结构时极易出错，常因过拟合表面模式而失败。通过多智能体系统缓解解析错误会带来极高延迟。为此，我们提出GRAIN，一种通过强化学习优化的单智能体框架。GRAIN将推理建模为语义解析与工具执行的流水线，由结构不变性奖励引导。该奖励通过对比提取的中间图与真实拓扑，迫使大模型学习稳健的文本到结构映射，而非记忆语言特征。我们还引入GRIT基准，评估对语言变化的敏感性。GRAIN在准确率上优于多智能体基线16.45%，延迟降低约24%。此外，其结构泛化能力更强，使SFT模型的分布外（OOD）差距减半（从15.77%降至7.80%），并在超出训练分布的大规模图上保持鲁棒性。
原文摘要 · Abstract (English)
Despite their potential in standardized graph tasks, Large Language Models (LLMs) remain brittle to real-world shifts in node identifiers and task formulation. While deterministic graph tools are invariant to such shifts, extracting topological structures from noisy text is highly fragile for LLMs, which often overfit to surface patterns. Moreover, mitigating these parsing failures via multi-agent systems incurs prohibitive latency. To address this, we propose GRAIN, a single-agent framework optimized via reinforcement learning. GRAIN models reasoning as a semantic parsing and tool-execution pipeline, guided by a Structure Invariance Reward. By validating extracted intermediate graphs against ground-truth topologies, this reward forces the LLM to learn robust text-to-structure mappings rather than memorizing linguistic artifacts. We also introduce GRIT, a benchmark evaluating sensitivity to such linguistic shifts. GRAIN outperforms multi-agent baselines by 16.45\% in accuracy with approximately 24\% lower latency. Furthermore, it demonstrates superior structural generalization, halving the out-of-distribution (OOD) gap of SFT models (from 15.77\% to 7.80\%) and maintaining robustness on large-scale graphs beyond the training distribution.
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。