EvoPolicyGym：評估在互動環境中的自主策略演化

摘要

自主智能体日益需要借助反馈来改进可执行策略，然而现有的评估常常将此过程简化为最终得分，或将其与开放式的软件工程进展混为一谈。我们提出**自主策略进化**（Autonomous Policy Evolution）这一受控评估框架：在该框架中，一个受约束的模型智能体在固定的交互预算下，反复编辑一个可执行策略系统。我们将该框架具象化为**EvoPolicyGym**基准测试，它由紧凑的交互式强化学习环境构建而成，专门评估智能体如何通过迭代方式改进已探索的策略。在EvoPolicyGym套件上，GPT-5.5在所有16个环境中取得了最强的整体排名分数，并跻身前两名。除了排行榜结果，EvoPolicyGym还提供轨迹级诊断，区分智能体如何分配预算、将反馈转化为参数调优。这些分析表明，强大的自主策略进化不仅依赖于孤立的任务获胜，更在于在有限的反馈下发现适合任务的机制并完善策略。

English

Autonomous agents are increasingly expected to improve executable policies through feedback, yet existing evaluations often collapse this process into a final score or confound it with open-ended software-engineering progress. We introduce Autonomous Policy Evolution, a controlled evaluation setting in which a harness-model agent repeatedly edits an executable policy system under a fixed interaction budget. We instantiate this setting in EvoPolicyGym, a benchmark built from compact interactive RL environments that evaluates how agents iteratively improve explored policies. On the EvoPolicyGym suite, GPT-5.5 achieves the strongest aggregate rank score and top-two performance on all 16 environments. Beyond leaderboard results, EvoPolicyGym also provides trajectory-level diagnostics that distinguish how agents allocate budget, convert feedback into parametric tuning. These analyses show that strong autonomous policy evolution depends not only on isolated task wins, but on discovering task-appropriate mechanisms and refining policies under bounded feedback.