AraSeg-2026-Shared-Task-NoPnx-NP
收藏资源简介:
该数据集是一个结构化文本分类数据集,采用MIT许可证发布。它包含三个标准分割:训练集(train_sampled,174个示例)、开发集(dev,222个示例)和测试集(test,262个示例),总大小约7.44MB。每个数据样本由三个核心字段组成:doc_id(文档唯一标识符,字符串类型)、text(文本内容,以字符串列表形式表示)和labels(标签,以64位整数列表形式表示)。数据以分片文件格式组织,适用于文本分类、多标签分类或序列标注等自然语言处理任务,其结构化特征表明设计用于监督学习场景,其中text字段可能包含分词后的文本序列,labels字段对应分类或标注信息。
This dataset is a structured text classification dataset published under the MIT License. It contains three standard data splits: the training split (train_sampled, 174 examples), the development split (dev, 222 examples), and the test split (test, 262 examples), with a total size of approximately 7.44 MB. Each data sample consists of three core fields: doc_id (unique document identifier, string type), text (text content represented as a list of strings), and labels (represented as a list of 64-bit integers). The dataset is stored in sharded file format and is suitable for natural language processing tasks such as text classification, multi-label classification, and sequence labeling. Its structured design indicates that it is intended for supervised learning scenarios, where the text field may contain tokenized text sequences, and the labels field corresponds to classification or annotation information.
数据集概述
数据集名称:AraSeg-2026-Shared-Task-NoPnx-NP
托管地址:https://huggingface.co/datasets/MBZUAI/AraSeg-2026-Shared-Task-NoPnx-NP
许可证:MIT
数据集结构
| 字段名 | 数据类型 | 说明 |
|---|---|---|
| doc_id | string | 文档唯一标识符 |
| text | list[string] | 文本内容(列表形式) |
| labels | list[int64] | 对应的标签(列表形式) |
数据集划分
| 划分名称 | 样本数量 | 数据大小 |
|---|---|---|
| test | 262 | 2,616,911 字节 |
| dev | 222 | 2,721,574 字节 |
| train_sampled | 174 | 2,101,636 字节 |
数据集规模
- 总下载大小:1,475,917 字节
- 总数据集大小:7,440,121 字节
配置与文件路径
- 默认配置名称:default
- 数据文件路径:
- test:
data/test-* - dev:
data/dev-* - train_sampled:
data/train_sampled-*
- test:




