从轨迹到智能体世界：面向交互环境模拟的智能体语言世界模型
原标题：Paper page - From Traces to Agentic Worlds: Agentic Language World Models for Interactive Environment Simulation
AI 摘要
该论文提出「智能体语言世界模型」范式，用世界模型智能体充当任务智能体的环境，实现有状态、长程的交互模拟，而无需重建原始可执行系统。作者实例化出免训练框架 Trace2Env，将历史交互轨迹重构为包含环境模式、证据与行为知识的 worldbook，运行时由世界模型智能体结合持久化情节状态推断观察与状态变化。在九个环境（终端、代码仓库、Android/Web 应用、企业服务、文本游戏）上，Trace2Env 在下一观察保真度和长程一致性上优于传统基于提示的语言世界模型，且任务智能体动作在真实环境中重放时有效率更高。
正文节选
From Traces to Agentic Worlds: Agentic Language World Models for Interactive Environment Simulation Abstract Realistic environment replicas are increasingly valuable for training and evaluating LLM agents, yet the original systems may be inaccessible or impractical to reproduce. We explore agentic language world modeling: rather than rebuilding an executable environment, a world model agent serves as the environment for a task agent and supports faithful and stateful simulation. We instantiate t