Bialy17/Reasoning-Socratic-QandA

该数据集是一个精心策划的混合数据集，结合了三个高质量数据源，专门用于训练苏格拉底导师。它平衡了模型的深度思考能力（推理）、通过提问引导学生能力（教学法）以及在必要时提供直接答案的能力（支持）。数据集包含三个组成部分：DASD Stage 1（40%）用于教授深度推理，Socratic（30%）用于教授教学提问风格，Q&A（30%）用于教授直接帮助模式。数据集适用于微调大型语言模型（LLMs）和小型语言模型（SLMs），使其能够作为AI导师，分析问题、通过苏格拉底方法引导学生，并在学生无法继续时清晰解释概念。

This dataset is a curated mixture of three high-quality data sources, specifically engineered to train **Socratic Tutors**. It balances the models ability to think deeply (Reasoning), guide students through questioning (Pedagogy), and provide direct answers when necessary (Support). The dataset consists of three components: DASD Stage 1 (40%) for deep reasoning, Socratic (30%) for pedagogical questioning style, and Q&A (30%) for direct help mode. The dataset is designed for fine-tuning Large Language Models (LLMs) and Small Language Models (SLMs) to act as **AI Tutors**, enabling them to analyze problems, guide students via the Socratic method, and explain concepts clearly if the student is unable to proceed.