PaulLerner/evqa_wikipedia

---
dataset_info:
features:
- name: image_reference_descriptions
sequence: string
- name: image_section_indices
sequence: int64
- name: image_urls
sequence: string
- name: section_texts
sequence: string
- name: section_titles
sequence: string
- name: wikipedia_title
dtype: string
- name: url
dtype: string
- name: image_index
sequence: int64
- name: passage_index
sequence: int64
- name: document
dtype: string
splits:
- name: train
num_bytes: 27143589758
num_examples: 1860136
download_size: 15147295813
dataset_size: 27143589758
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
---
https://github.com/PaulLerner/ViQuAE

原始信息汇总

数据集概述

数据集特征

• image_reference_descriptions: 字符串序列
• image_section_indices: 整数序列（int64）
• image_urls: 字符串序列
• section_texts: 字符串序列
• section_titles: 字符串序列
• wikipedia_title: 字符串类型
• url: 字符串类型
• image_index: 整数序列（int64）
• passage_index: 整数序列（int64）
• document: 字符串类型

数据集分割

• 训练集（train）:
  • 数据量: 27,143,589,758 字节
  • 示例数量: 1,860,136

数据集大小

• 下载大小: 15,147,295,813 字节
• 数据集总大小: 27,143,589,758 字节

配置

• 默认配置（default）:
  • 数据文件路径: data/train-*