SafeRun：在跑步规划中实现LLM的确定性
SafeRun: Enabling Determinism in LLM Planning for Running
AI总结：
针对LLM在跑步规划中因概率性导致安全违规的问题，提出SafeRun框架，通过解耦架构将LLM的软解释与确定性求解器的硬约束分离，实现100%安全评分。
摘要：
大型语言模型能够实现灵活的自然语言规划，但由于其概率性，在确定性关键领域仍不可靠。这一限制在跑步规划中尤其成问题，因为违反安全规则可能导致安全风险。我们提出SafeRun，一种通过解耦架构实现基于LLM的确定性规划的框架。SafeRun将LLM的软解释与确定性求解器的硬约束执行分离，在保持自然语言灵活性的同时确保严格的安全约束。为了验证SafeRun，我们构建了一个全面的基准测试，用于在现实生理和安全约束下进行跑步规划。在五个LLM上的实验表明，SafeRun实现了100%的安全评分（相比之下，PE平均为79.1%，CodeAct平均为97.6%），同时保持了具有竞争力的指令遵循分数。SafeRun基准测试可在\href{https://huggingface.co/datasets/zzp-seeker/SafeRun-RunPlanning-Benchmark}{huggingface}上公开获取。
英文摘要：
Large Language Models enable flexible natural-language planning but remain unreliable in determinism-critical domains due to their probabilistic nature. This limitation is especially problematic in running planning, where violating safety rules can lead to safety risks. We propose SafeRun, a framework for deterministic LLM-based planning via a decoupled architecture. SafeRun separates soft interpretation by an LLM from hard constraint enforcement by a deterministic solver, ensuring strict safety constraints while preserving natural-language flexibility. To validate SafeRun, we build a comprehensive benchmark for running planning under realistic physiological and safety constraints. Experiments across five LLMs show that SafeRun achieves 100\% safety score (vs.\ 79.1\% PE average and 97.6\% CodeAct average) while maintaining competitive instruction-following scores. The SafeRun benchmark is publicly available at \href{https://huggingface.co/datasets/zzp-seeker/SafeRun-RunPlanning-Benchmark}{huggingface}.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
