测试智能体在物理环境变化后如何自我修改代码，评估其适应能力。
PACE-Bench: Benchmarking Physics Adaptation via Code Evolution in Dynamic Environments
- 通过代码迭代优化，让智能体从源环境设计适应目标环境。
- 最佳模型仅35.9%完成全部任务，静态子集也仅66.7%成功。
- 适合研究自进化、代码生成与环境适应的学者参考。
自演化智能体可通过交互经验改进未来行为，但现有评估多在固定条件下进行，未检验条件变化后的恢复能力。为此，我们提出PACE-Bench（基于代码演化的物理适应基准），包含跨六个物理领域、144个源-目标适应对的仿真基准。每对关联一个源环境与具有相同目标和接口的变异目标环境。在源环境中有效的代码驱动设计，在目标中会失效，智能体需在有限尝试预算内，利用诊断沙箱反馈，逐步调整代码以适配目标。我们对比了四种范式下的十种自演化方法。基准尚未饱和：Reflexion + Qwen3-14B仅在35.9%的全基准对上成功，而GPT-5.5在静态子集下于完整预算内解决66.7%。结果表明，基于仿真的反思比未经验证的自我修正更可靠；记忆锚定使智能体固守早期设计，广度树搜索虽探索充分却难收敛。即使明确告知物理变化，性能上限仍难提升，说明核心瓶颈在于机制重构而非参数推断。数据与代码已开源。
原文摘要 · Abstract (English)
Self-evolving agents improve future behavior from interaction experience, yet existing evaluations typically optimize under fixed execution conditions and do not test recovery after those conditions change. To address this gap, we introduce PACE-Bench (Physics Adaptation via Code Evolution), a simulator-grounded benchmark of 144 source-to-target adaptation pairs across six physics domains. Each pair links a source environment to a mutated target environment with the same goal and interface. A code-driven design that succeeds in the source fails in the target, where agents must iteratively adapt it into a working target design using diagnostic sandbox feedback within a limited attempt budget. We compare ten self-evolving methods from four paradigms. The benchmark remains far from saturated: Reflexion + Qwen3-14B succeeds on only 35.9\% of full-benchmark pairs, while GPT-5.5 solves 66.7\% of the Statics subset under the full budget. Together, these results show that simulator-grounded reflection is more reliable than unverified self-revision, while memory anchors agents to early designs and broad tree search explores without converging. Even revealing exact physical changes does not raise the performance ceiling, pointing to mechanism redesign rather than parameter inference as the central bottleneck. Data and code are available at https://github.com/thunlp/PACE-Bench.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。