arXiv cs.CL· Shota Fujikawa, Takuya Tamura, Kosuke Akimoto, Daichi Haraguchi, Masafumi Oyamada·· 13 小时前AI 评分34
Agent Behavior Maps：无需运行智能体，用行为坐标定位 LLM 智能体
Agent Behavior Maps: Locating Agents by What They Would Do, Without Running Them
AI 导读
研究者提出 Agent Behavior Maps，用共享参考轨迹上的中心化动作对数似然来表征 LLM 与 harness 组合，无需对目标智能体做新的环境 rollout。在八个交叉评分模型上，坐标距离与直接测得的轨迹 KL 在排序上一致；覆盖 100 个 LLM 和四个环境，ALFWorld 坐标可支持性能预测并识别行为相似性，且在控制成功率和模型规模后仍成立。
正文 · 原文
Abstract:LLM agents with similar success rates can follow different action sequences. An Agent Behavior Map represents LLM-harness combinations using centered action log-likelihoods on shared reference trajectories, without requiring new environment rollouts from the target agent. We derive error bounds relating squared coordinate distances to trajectory KL under explicit assumptions, and across eight cross-scored models the distances track the directly measured trajectory KL in rank order. Across 100 LLMs and four environments, ALFWorld-based coordinates support performance prediction and identify behavioral similarities. In ALFWorld, these similarities persist after controlling for success rates and model size, including across model families. As an extension, harness-component displacements approximately compose, with accuracy varying across components and environments.
| Comments: | 23 pages, 5 figures, 12 tables. Under review |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.30014 [cs.LG] |
| (or arXiv:2610.30014v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.30014 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Takuya Tamura [view email]
[v1]
Wed, 21 Oct 2026 16:44:02 UTC (324 KB)
来源：arXiv cs.CL · arxiv.org