arXiv cs.AI· Haori Lu, Tianyu Gao, Yaoyao Liu, Yingying Li·· 13 小时前AI 评分39
基于世界模型预测路径的自改进 LLM 智能体
Self-Improving Agents with World Model Predictive Paths
AI 导读
研究者提出一种受 MPPI 控制启发的测试时框架，让 LLM 智能体在每一步生成多个候选动作，并用世界模型预测其结果，从而在不实际执行的情况下比较不同方案。由于语言环境不提供显式动力学与代价，预测结果按任务进展和风险评分，再以 MPPI 式权重引导最终决策。
正文 · 原文
Abstract:Large language model (LLM) agents are increasingly expected to interact with environments and improve their decisions during deployment. However, how anticipating future outcomes and drawing on accumulated interaction experience can jointly support decision-making remains underexplored. We propose a test-time framework, inspired by Model Predictive Path Integral (MPPI) control, that combines multi-branch predictive analysis with experience-based adaptation. At each step, the agent proposes multiple candidate actions and imagines their outcomes with a world model, so that alternatives can be compared without being executed in the environment. Since language environments provide neither explicit dynamics nor costs, the predicted outcomes are evaluated for task progress and risk, and MPPI-style weights over the resulting costs guide the final decision. A dual-branch experience memory enables self-improvement: the planner branch refines candidate generation and action selection, while the world model branch improves next-observation prediction. Experiments on ALFWorld and ScienceWorld across three backbones demonstrate the strong overall performance of our method and the effectiveness of experience-guided predictive control. Our code will be made publicly available.
| Comments: | 17 pages, 4 figures |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.28945 [cs.AI] |
| (or arXiv:2610.28945v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.28945 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Haori Lu [view email]
[v1]
Tue, 20 Oct 2026 21:46:29 UTC (987 KB)
来源：arXiv cs.AI · arxiv.org