用少量数据训练小模型替代大模型，就能在笔记本上模拟大规模智能体社会。
Poor Man's Agentic Modeling: Simulating Large LLM-Agent Societies on a Laptop
- 用几百到几千次廉价查询拟合低参数模型，替代原大模型智能体。
- 预测的误差趋势与实际模拟结果高度一致，且能定量解释错误原因。
- 适合研究群体行为、经济模型等宏观问题，无需高性能计算资源。
模拟大量大型语言模型（LLM）智能体构成的社会成本高昂，但通常关注的是宏观现象：相变行为、典型事实及智能体数量 $N$ 的缩放规律，而非单个智能体的认知细节。本文借鉴统计物理思想，提出方法：将每个 LLM 智能体替换为通过数百至数千次低成本查询拟合的低参数模型，即可在笔记本电脑上运行任意 $N$ 的社会模拟。该方法是否有效，可在模拟前由智能体的感知能力决定。我们引入一个 [交互阶数 × 记忆] 分类体系，将感知与记忆映射为有效理论，并预测代理误差随 $N$ 的变化趋势。在忠实复现 LLM 宏观经济模型 EconAgent 及另外七个知名仿真案例中验证，智能体决策基于真实 LLM 输出（主要来自 DeepSeek），总成本不足几美元；预测的误差趋势在每单元内均成立，且两个被证伪的预测（均涉及强饱和响应）均被理论精确量化，无自由参数调整。
原文摘要 · Abstract (English)
Simulating societies of many large language model (LLM) agents is expensive, yet the questions asked of such simulations are usually macroscopic: phase behaviour, stylised facts, and scaling with the number of agents $N$, not the cognition of any single agent. We turn a statistical-physics observation into a method: replace each LLM agent by a low-parameter model fitted from a few hundred to a few thousand cheap queries, then run the society at any $N$ on a laptop. Whether this works is decided before the simulation runs, chiefly by what each agent perceives. We introduce an [interaction order x memory] taxonomy that maps perception and memory to an effective theory and a predicted $N$-trend of the surrogate error. We validate it on a faithful reimplementation of the LLM macroeconomy EconAgent and seven further named LLM simulations, with agent decisions cloned from genuine LLM elicitations (primarily DeepSeek) for a few dollars; the predicted error trends hold cell by cell, and the two refuted predictions, both on a strongly saturating response and traced to its curvature, are themselves matched quantitatively by the theory with no free parameters.
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。