反例作为智能体自我修正的反馈：A-CEGIS 框架
原标题：Counterexamples as Feedback for Agent Self-Correction
AI 摘要
A-CEGIS 是一个轻量级框架，利用反例作为反馈来评估智能体在自然语言到正则表达式合成中的多轮自我修正能力。在 30 个 NL-RX-Turk 任务上，诊断性反例反馈在四轮内解决了 90% 的任务，而零样本生成、通用自我修正和仅错误反馈的解决率分别为 17%、27% 和 23%。完整诊断运行后，所有任务在最终轮次均被解决，平均成功轮次为 2.7，鲁棒成功率为 77%。该框架提供了可复现的 CEGIS 循环、轨迹指标和鲁棒性探测，以衡量智能体的修复效率和稳定性。
正文节选
COUNTEREXAMPLES AS FEEDBACK FOR AGENT SELF-CORRECTION Abstract Single-turn code-generation metrics understate a central property of deployed agents: whether they can repair a wrong artifact after receiving concrete feedback. This paper presents A-CEGIS, a lightweight framework that uses counterexamples as feedback for evaluating multi-turn refinement in natural-language-to-regex synthesis. An agent proposes a regex, a deterministic oracle checks it under full-match semantics, and compact false-p