iamtarun/OKVQA-demo

---
dataset_info:
features:
- name: answers
sequence: string
- name: gold_answer
dtype: string
- name: question
dtype: string
- name: question_id
dtype: string
- name: img_path
dtype: string
- name: img_key_full
dtype: string
- name: img_key
dtype: int64
- name: img_file_name
dtype: string
- name: img
dtype: 'null'
- name: img_caption
struct:
- name: caption
dtype: string
- name: conf
dtype: float64
- name: objects
list:
- name: attribute_scores
sequence: float64
- name: attributes
sequence: string
- name: class
dtype: string
- name: ocr
list:
- name: score
dtype: float64
- name: text
dtype: string
- name: rect
sequence: float64
- name: img_ocr
list:
- name: description
dtype: string
- name: vertices
sequence:
sequence: int64
- name: __index_level_0__
dtype: string
- name: ROIs
sequence: string
- name: pos_item_ids
sequence: string
- name: pos_item_contents
sequence: string
- name: text_sequence
dtype: string
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: pixel_values
sequence:
sequence:
sequence:
sequence: float32
splits:
- name: test
num_bytes: 30631907569
num_examples: 5046
download_size: 8855206784
dataset_size: 30631907569
configs:
- config_name: default
data_files:
- split: test
path: data/test-*
---

原始信息汇总

数据集概述

数据特征

• answers: 字符串序列
• gold_answer: 字符串
• question: 字符串
• question_id: 字符串
• img_path: 字符串
• img_key_full: 字符串
• img_key: 64位整数
• img_file_name: 字符串
• img: 空值
• img_caption: 结构体
  • caption: 字符串
  • conf: 64位浮点数
• objects: 列表
  • attribute_scores: 64位浮点数序列
  • attributes: 字符串序列
  • class: 字符串
  • ocr: 列表
    • score: 64位浮点数
    • text: 字符串
  • rect: 64位浮点数序列
• img_ocr: 列表
  • description: 字符串
  • vertices: 64位整数序列的序列
• index_level_0: 字符串
• ROIs: 字符串序列
• pos_item_ids: 字符串序列
• pos_item_contents: 字符串序列
• text_sequence: 字符串
• input_ids: 32位整数序列
• attention_mask: 8位整数序列
• pixel_values: 32位浮点数序列的序列的序列的序列

数据分割

• test:
  • 字节数: 30631907569
  • 样本数: 5046

数据集大小

• 下载大小: 8855206784 字节
• 数据集大小: 30631907569 字节

配置

• default:
  • 数据文件:
    • test: data/test-*