English(EN) Beyond Skepticism: Evaluating LLMs Pedagogical Intent Reasoning with the Adaptive Pedagogical Vigilance Framework

新框架评估大型语言模型对教学意图的理解

作者 PulseAugur 编辑部 · [2 个来源] · 2026-07-02 01:26

研究人员引入了自适应教学警惕（APV）框架，以评估和增强大型语言模型（LLMs）在教学交流中理解教学意图的能力。APV框架利用贝叶斯教学意图推理引擎（PIIE）来模拟教师如何选择内容以及学习者应如何推断体裁、立场和激励等教学配置。使用GPT-4o和Claude 3.5等领先大型语言模型的实验表明，APV显著提高了模型的警惕性，从而更好地区分教学内容和暴露性内容，并与人类判断高度相关。 AI

影响该框架通过提高大型语言模型对教学交流的理解能力，有望带来更可靠的AI辅助学习系统。

排序理由该集群包含一篇学术论文，详细介绍了评估大型语言模型能力的新框架和实验结果。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.CL 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。我们如何撰写摘要 →

报道来源 [2]

arXiv cs.CL TIER_1 English(EN) · Minghao Chen, Ruihan Zhou, Jiayi Tang, Zihan Xu, Bowen Huang, Yuxin Liu · 2026-07-03 04:00

Beyond Skepticism: Evaluating LLMs Pedagogical Intent Reasoning with the Adaptive Pedagogical Vigilance Framework

arXiv:2607.01581v1 Announce Type: new Abstract: The capacity of Large Language Models (LLMs) to reason about pedagogical intent within instructional communication remains underexplored, particularly in educational domains such as translation pedagogy. To address this, we propose …
arXiv cs.CL TIER_1 English(EN) · Yuxin Liu · 2026-07-02 01:26

Beyond Skepticism: Evaluating LLMs Pedagogical Intent Reasoning with the Adaptive Pedagogical Vigilance Framework

The capacity of Large Language Models (LLMs) to reason about pedagogical intent within instructional communication remains underexplored, particularly in educational domains such as translation pedagogy. To address this, we propose the \textbf{Adaptive Pedagogical Vigilance (APV)…

报道来源 [2]

Beyond Skepticism: Evaluating LLMs Pedagogical Intent Reasoning with the Adaptive Pedagogical Vigilance Framework

Beyond Skepticism: Evaluating LLMs Pedagogical Intent Reasoning with the Adaptive Pedagogical Vigilance Framework

相关实体

相关话题