math_qa_zh

<p align="left"> <a href="README_CN.md">中文</a>&nbsp;｜&nbsp;English </p> <br> # math_qa_zh [](#license) &nbsp;&nbsp; [](https://huggingface.co/datasets/choucsan/math_qa_zh) &nbsp;&nbsp; [](https://modelscope.cn/datasets/choucisan/math_qa_zh) --- ## Dataset Introduction **math_qa_zh** is a Chinese math question-answering dataset containing 27,986 multiple-choice math problems. Each sample consists of a system prompt, a math question with four answer options (A, B, C, D), and the correct answer. Designed for SFT of LLMs on mathematical reasoning in Chinese. ### Key Features - **Language**: Simplified Chinese - **Task Type**: Mathematical reasoning (multiple-choice QA) - **Topics**: Arithmetic, algebra, geometry, profit & loss, time & work, number theory - **Format**: Instruction-following (system + prompt → response) ## Dataset Structure | Property | Value | |:---|:---| | Total Samples | 27,986 | | Splits | `train` | | Columns | `system`, `prompt`, `response` | | Format | Parquet | ### Column Description | Column | Description | |:---|:---| | `system` | System instruction | | `prompt` | Math problem with options (A-D) and format instructions | | `response` | Correct answer: "正确答案是 [option]" | ## Sample Preview ```json { "system": "你是一个非常聪明的助手，请直接遵循指示作答。", "prompt": "请回答以下问题：\n\n一个推车包的售价包括销售税为1120卢比。销售税率为12%。如果店主获得了25%的利润，那么这个推车包的成本价是多少？\nA.1000卢比\nB.1515卢比\nC.800卢比\nD.900卢比\n\n请按以下格式作答：\"正确答案是 (在此处填写选项字母)\"/no_think", "response": "正确答案是 C" } ``` ## Usage ```python from datasets import load_dataset dataset = load_dataset("choucsan/math_qa_zh") # or from ModelScope from modelscope import MsDataset dataset = MsDataset.load("choucisan/math_qa_zh") ``` ## License MIT License. ## Contact - 📧 **[e-mail]** - 🤗 [Hugging Face](https://huggingface.co/datasets/choucsan/math_qa_zh) - 🐱 [ModelScope](https://modelscope.cn/datasets/choucisan/math_qa_zh)