世界模型提供了一种原则性的方法，使长期 LLM 代理在执行前对行动后果具有远见: 的预测。然而,不可靠的远见可以被忽略,滥用,甚至降低下游决策。在本文,中，我们介绍了 WorldEvolver, 一个自我进化的世界模型框架，它修改其部署时上下文，同时保持下游代理和所有模型参数冻结。 WorldEvolver 集成了三个模块: (i) 情节记忆,，通过基于检索的模拟来利用真实动作转换; (ii) 语义记忆,，从预测观察不匹配中提取持久启发式规则; 和 (iii) 选择性预见,，在将低置信度预测集成到代理推理上下文之前过滤它们。我们在 ALFWorld 和 ScienceWorld, 上评估 WorldEvolver，测量 Word2World 上的世界模型预测准确性以及 AgentBoard 上的下游代理成功率。大量实验表明，WorldEvolver 在三个主干网中实现了最高的预测精度，并在下游代理成功率, 方面领先于其他世界模型基线，这表明测试时内存修正增强了预测保真度和规划性能。
World models offer a principled way to equip long-horizon LLM agents with foresight: predictions of action consequences before execution. However, unreliable foresight can be ignored, misused, or even degrade downstream decision-making. In this paper, we introduce WorldEvolver, a self-evolving world model framework that revises its deployment-time context while keeping the downstream agent and all model parameters frozen. WorldEvolver integrates three modules: (i) Episodic Memory, which exploits real action transitions through retrieval-based simulation; (ii) Semantic Memory, which extracts persistent heuristic rules from prediction-observation mismatches; and (iii) Selective Foresight, which filters low-confidence predictions before integrating them into agent reasoning context. We evaluate WorldEvolver on ALFWorld and ScienceWorld, measuring world model prediction accuracy on Word2World and downstream agent success rate on AgentBoard. Extensive experiments show that WorldEvolver achieves the highest prediction accuracy across three backbones and leads other world model baselines on downstream agent success rate, demonstrating that test-time memory revision enhances both predictive fidelity and planning performance.
科目: 人工智能 (cs.AI); 计算和语言 (cs.CL)
Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)