English and Chinese Braille Mixed Datasets (EBMD/CBMD)
收藏资源简介:
本数据集是一套包含中文和英文盲文的混合文本数据集,旨在支持盲文领域的研究,特别是解决混合文本环境中的盲文信息处理问题。数据集由纯中文、混合中文和混合英文盲文数据组成,并包含了丰富的数学公式内容。数据收集过程包括从标准中文新闻文章和STEM教育相关文档中生成盲文版本,并通过专家团队进行手动校对和标注。此外,为了提高数据集的多样性和有效性,还引入了一种基于句法和依赖语法树的数据增强方法。该数据集为低资源多语言盲文研究和翻译技术提供了基础。
This dataset is a mixed text dataset containing Chinese and English braille, designed to support research in the braille field, particularly addressing braille information processing challenges in mixed-text environments. The dataset consists of pure Chinese braille, mixed Chinese braille and mixed English braille data, and includes a rich set of mathematical formulas. The data collection process involves generating braille versions from standard Chinese news articles and STEM education-related documents, followed by manual proofreading and annotation by a panel of experts. Furthermore, to enhance the diversity and effectiveness of the dataset, a data augmentation method based on syntactic and dependency syntax trees was introduced. This dataset serves as a foundational resource for low-resource multilingual braille research and translation technologies.
NuminaMath CoT 数据集概述
数据集基本信息
- 数据集名称: NuminaMath CoT
- 任务类别: 文本生成
- 主要语言: 英语
- 许可证: Apache License 2.0
- 数据总量: 859,494个训练样本,100个测试样本
数据集结构
特征字段
- source:数据来源标识
- problem:数学问题描述
- solution:问题解答
- messages:包含content和role字段的列表结构
数据拆分
- 训练集:859,494个样本,数据大小约2.5GB
- 测试集:100个样本,数据大小约290KB
- 总数据集大小:约2.5GB
- 下载大小:约1.2GB
数据集内容
数据来源
数据集包含约86万个数学问题,每个解答均采用思维链格式。数据来源包括:
- 中国高中数学练习题
- 美国及国际数学奥林匹克竞赛题目
- 在线考试试卷PDF
- 数学讨论论坛
数据处理流程
- 原始PDF文档的OCR识别
- 分割为问题-解答对
- 翻译为英文
- 重新对齐生成思维链推理格式
- 最终答案格式化
数据来源分布
| 来源 | 样本数量 |
|---|---|
| aops_forum | 30,201 |
| amc_aime | 4,072 |
| cn_k12 | 276,591 |
| gsm8k | 7,345 |
| math | 7,478 |
| olympiads | 150,581 |
| orca_math | 153,334 |
| synthetic_amc | 62,111 |
| synthetic_math | 167,895 |
| 总计 | 859,608 |
引用信息
@misc{numina_math_datasets, author = {Jia LI and Edward Beeching and Lewis Tunstall and Ben Lipkin and Roman Soletskyi and Shengyi Costa Huang and Kashif Rasul and Longhui Yu and Albert Jiang and Ziju Shen and Zihan Qin and Bin Dong and Li Zhou and Yann Fleureau and Guillaume Lample and Stanislas Polu}, title = {NuminaMath}, year = {2024}, publisher = {Numina}, journal = {Hugging Face repository}, howpublished = {url{https://github.com/project-numina/aimo-progress-prize/blob/main/report/numina_dataset.pdf}} }




