发表机构
East China Normal University; Shanghai Chuangjie Situo Information Technology Co., Ltd.; Shanghai Normal University（华东师范大学; 上海创杰思拓信息技术有限公司; 上海师范大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究针对教育场景中LLMs直接作答问题，提出HeuristicEdu两阶段管道，经监督热身和GRPO，利用特定对话及启发式奖励训练Qwen2.5 - 7B，提升了支架有效性，降低关键词泄漏，表明规模并非引发苏格拉底行为的唯一因素。
AI 中文摘要
在教育环境中部署的大语言模型（LLMs）通常表现为直接回答者，不像苏格拉底教学法规定的那样通过渐进式探究引导学生。我们提出HeuristicEdu，这是一个两阶段管道，通过监督热身和组相对策略优化（GRPO）使Qwen2.5 - 7B向苏格拉底辅导对齐。训练使用从直播平台重建的797个多轮中文儿童科学对话，有基于认知深度（R_cog）、好奇心参与度（R_eng）和直接性（R_dir）的启发式奖励以及对学生引入术语的K_query校正。引入支架有效性（SE）和对话深度（CD）来评估表面流利度之外的结果。在30个保留问题上，最佳GRPO变体将SE从30.0%提高到63.3%，并将关键词泄漏从30.0%降低到13.3%。未对齐的Qwen - 72B基线SE为0%，泄漏率为96.7%。
英文摘要
Large language models (LLMs) deployed in educational settings often behave as direct answerers: they disclose target concepts in the opening turn instead of guiding students through progressive inquiry, as Socratic pedagogy prescribes. We present HeuristicEdu, a two-phase pipeline that aligns Qwen2.5-7B toward Socratic tutoring via supervised warm-up and Group Relative Policy Optimization (GRPO). Training uses SocraticEdu, 797 multi-turn Chinese children's science dialogues reconstructed from a live platform, with a heuristic reward over cognitive depth (R_cog), curiosity engagement (R_eng), and directness (R_dir), together with a K_query correction for student-introduced terms. We introduce Scaffolding Effectiveness (SE) and Conversation Depth (CD) to evaluate outcomes beyond surface fluency. On 30 held-out questions, the best GRPO variant improves SE from 30.0% to 63.3% and lowers keyword leakage from 30.0% to 13.3%. Notably, this best variant omits the directness penalty during optimization, suggesting that explicit anti-leakage terms can conflict with gradient-based behavioral alignment. An unaligned Qwen-72B baseline reaches 0% SE and 96.7% leakage, showing that scale alone does not induce Socratic behavior.
Comments12 pages, 2 figures, 5 tables; includes an appendix
