外观
论文HF Daily Papers · arXiv 2610.291397 天前
PAWM世界模型通过陷阱预判提升黑盒LLM智能体任务成功率
本文由 AI 根据公开内容整理，可能有误，请以原文为准：Steering Black-Box Large Language Model Agents with Pitfall Anticipation World Model（HF Daily Papers · arXiv 2610.29139）
核心内容
前沿黑盒大语言模型智能体虽具备长程推理能力，但缺乏特定环境动力学知识。现有世界模型因逐步预测观测而计算昂贵且信息密度低。研究提出陷阱预判世界模型（PAWM），改变全步预测范式，仅在检测到多步未来内的潜在陷阱时选择性介入并预测 impending failure，从而引导智能体规避错误路径。
PAWM采用两阶段训练策略：首先在合成数据上进行监督微调（SFT），数据标注包含智能体陷阱标签；随后与黑盒LLM进行GRPO协同训练以优化目标环境适应性。实验表明，该方法显著优于传统基线，在ScienceWorld任务中性能提升24.3%，在WebArena-Lite中提升7.7%，验证了按需触发式世界模型对黑盒Agent的有效性。
关键看点
- 选择性干预机制：摒弃逐帧预测，仅在预判到多步未来内存在陷阱时触发世界模型，降低无效推理开销
- 两阶段训练流程：结合带陷阱标签的合成数据SFT与黑盒LLM的GRPO协同训练，实现环境特异性对齐
- 基准性能提升：在ScienceWorld和WebArena-Lite测试中，使Claude Sonnet 4.5成功率分别提升24.3%与7.7%
工程与产业观察
PAWM验证了具身智能体无需全量环境模拟，仅需在关键决策点进行风险预判即可大幅提升黑盒模型鲁棒性。这种按需触发的世界模型架构为降低Agent推理成本、解决Sim-to-Real中的长尾失败提供了高效范式，对复杂任务规划具有工程参考价值。
英文摘要 (Abstract)
Frontier black-box Large Language Model (LLM) agents are proficient at long-horizon reasoning and precise action generation. However, when it comes to a specific agentic task, they have no internalized knowledge of how that particular environment unfolds. A world model offers a natural remedy, providing the agent with environment specific knowledge by foreseeing how the environment will respond to the action it is about to take. Prior world models for LLM agents mainly predict next observations at each step, which is costly and often uninformative because individual observations provide limited task-relevant guidance and not every step requires call for foresight. This raises a key question: what a world model should predict and when should it be invoked for agentic tasks? Our pilot study suggests that world model guidance is most useful when it anticipates pitfalls within a bounded multi-step future. Motivated by this, we introduce Pitfall Anticipation World Model (PAWM), which selectively intervenes when a potential pitfall arises and predicts the impending failure to steer the agent away from it. PAWM is trained in two stages: Supervised Fine-Tuning (SFT) on synthetic data labeled with agent pitfalls, followed by GRPO co-training with a black-box LLM to further optimize it in the target environment. PAWM improves Claude Sonnet 4.5 performance by 24.3% on ScienceWorld and 7.7% on WebArena-Lite, outperforming a range of baselines. These results demonstrate that PAWM provides an effective form of world model guidance for black-box LLM agents.
Hugging Face Daily Papers 热度：2 upvotes 作者：Jiongxiao Wang, Han Ding, Youzhi Luo, Yawei Wang, Yun Zhou, Linbo Liu, Panpan Xu
来源
本页事实对应的公开页面，以原文为准。