rootsautomation/RICO-ScreenQA
收藏Hugging Face2024-04-16 更新2024-04-19 收录
下载链接:
https://hf-mirror.com/datasets/rootsautomation/RICO-ScreenQA
下载链接
链接失效反馈官方服务:
资源简介:
ScreenQA是一个大规模的问题-答案对数据集,基于移动应用截图。数据集包含screen_id、question、ground_truth(包含full_answer和ui_elements)、file_name和image等特征。数据集被分为train、validation和test三个部分,分别包含68980、8618和8427个示例。任务类别是问答,语言是英语。
ScreenQA is a large-scale question-answer pair dataset based on mobile application screenshots. The dataset includes features such as screen_id, question, ground_truth (containing full_answer and ui_elements), file_name, and image. It is divided into three subsets: train, validation, and test, which respectively contain 68980, 8618, and 8427 samples. The task category is question answering, and the language used is English.
提供机构:
rootsautomation
原始信息汇总
数据集概述
基本信息
- 许可证: cc-by-4.0
- 数据集名称: ScreenQA
- 任务类别: 问答(question-answering)
- 语言: 英语(en)
- 大小类别: 10K<n<100K
数据集特征
- screen_id: 字符串(string)
- question: 字符串(string)
- ground_truth:
- full_answer: 字符串(string)
- ui_elements:
- bounds: 序列(float64)
- text: 字符串(string)
- vh_index: 整数(int64)
- file_name: 字符串(string)
- image: 图像(image)
数据集划分
- 训练集(train):
- 示例数量: 68980
- 数据大小: 8162486352.98字节
- 验证集(validation):
- 示例数量: 8618
- 数据大小: 1057181449.778字节
- 测试集(test):
- 示例数量: 8427
- 数据大小: 951399950.724字节
数据集大小
- 下载大小: 3324988803字节
- 数据集总大小: 10171067753.481998字节
配置文件
- 默认配置(default):
- 训练集路径: data/train-*
- 验证集路径: data/validation-*
- 测试集路径: data/test-*



