MathEDU
收藏资源简介:
MathEDU数据集是由台湾阳明交通大学计算机科学系的研究人员创建的,旨在帮助学生在数学问题解决过程中获得个性化的适应性反馈。数据集包含了真实的学生解题过程,并附有教师反馈。它由4048个带注释的条目组成,包括错误类型、错误步骤和教师的反馈,旨在帮助学生纠正错误并提高解题能力。数据集的创建过程涉及六名学生的解题过程,由三位数学专家进行评分和反馈。数据集适用于评估大型语言模型在数学教育中的应用,旨在解决在线学习中缺乏个性化反馈的问题。
The MathEDU Dataset was developed by researchers from the Department of Computer Science at National Yang Ming Chiao Tung University, Taiwan, with the objective of assisting students in obtaining personalized adaptive feedback during mathematical problem-solving processes. The dataset includes real student problem-solving records paired with teacher feedback, consisting of 4,048 annotated entries covering error types, erroneous solution steps, and instructor feedback, aiming to help students correct mistakes and improve their problem-solving abilities. The development of this dataset involved the problem-solving workflows of six students, which were evaluated and supplemented with targeted feedback by three mathematics experts. This dataset is suitable for evaluating the application of Large Language Models (LLMs) in mathematics education, addressing the issue of insufficient personalized feedback in online learning.
MathEDU 数据集概述
数据集简介
- 名称: MathEDU
- 目的: 支持数学教育,纠正学生在数学问题解决中的错误
- 数据类型: 真实学生解答与专家反馈标注
数据集结构
每个条目包含以下字段:
- id: 唯一标识符,可映射到MathQA中的问题
- student_id: 学生ID
- student_answer: 学生最终答案
- student_process: 学生解题过程(LATEX格式)
- correct_or_not: 答案正确性标记(correct/wrong)
- the_reason_why_student_cant_solve_ch: 学生解题失败原因(中文)
- the_reason_why_student_cant_solve_en: 学生解题失败原因(英文)
- teacher_review: 教师反馈字典,包含:
- error_counts: 错误数量
- error: 错误详情列表,每项包含:
- error_type: 错误类型(如"Wrong mathematical operation/concept")
- error_equation: 错误发生的具体解题部分
- teacher_advice_ch: 教师反馈(中文)
- teacher_advice_en: 教师反馈(英文)
示例条目
json { "id": 9420, "student_id": 5, "student_answer": "3:5", "student_process": "ratio of de: bc equal to the ratio of the area, Ans: 3:5", "correct_or_not": "wrong", "the_reason_why_student_cant_solve_ch": "", "the_reason_why_student_cant_solve_en": "", "teacher_review": { "error_counts": 1, "error": [ { "error_type": "Wrong mathematical operation/concept", "error_equation": "ratio of de: bc equal to the ratio of the area", "teacher_advice_ch": "觀念錯誤...", "teacher_advice_en": "The concept is incorrect..." } ] } }
运行指令
-
Llama3 8B评分:
python llama3_8b_grading.py -
Llama3 70B评分:
python llama3_70b_grading.py -
GPT-3.5评分:
python gpt_3.5__grading.py -
o1-mini评分:
python o1_mini_grading.py -
响应分析:
python response_analyze.py -
GPT-4评分结果:
python gpt4_llm_rating.py -
创建微调数据:
python create_finetuned_data.py -
Llama3 8B微调: bash huggingface-cli login –token "your_hf_token" !ACCELERATE_USE_FSDP=1 FSDP_CPU_RAM_EFFICIENT_LOADING=1 torchrun --nproc_per_node=4 train.py --config finetune.yaml
-
微调模型推理:
python inference.py –config finetune.yaml
依赖安装
pip install -r requirements.txt




