anneyouw/nlvr_local_by_appearance

---
dataset_info:
features:
- name: directory
dtype: string
- name: sentence
dtype: string
- name: image_path
dtype: string
- name: label
dtype: string
- name: structured_rep
list:
list:
- name: color
dtype: string
- name: size
dtype: int64
- name: type
dtype: string
- name: x_loc
dtype: int64
- name: y_loc
dtype: int64
- name: image_id
dtype: int64
- name: appearance
dtype: string
- name: identifier
dtype: string
- name: image
dtype: image
splits:
- name: train
num_bytes: 260578272.74
num_examples: 74460
- name: train_tower
num_bytes: 88696855.44
num_examples: 34272
- name: train_scatter
num_bytes: 174059221.136
num_examples: 40188
- name: dev
num_bytes: 18649112.884
num_examples: 5934
- name: dev_tower
num_bytes: 10434862.944
num_examples: 4056
- name: dev_scatter
num_bytes: 7950532.81
num_examples: 1878
- name: test
num_bytes: 18492458.66
num_examples: 5940
- name: test_tower
num_bytes: 11028462.528
num_examples: 4272
- name: test_scatter
num_bytes: 7277693.54
num_examples: 1668
download_size: 478327325
dataset_size: 597167472.6819998
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: train_tower
path: data/train_tower-*
- split: train_scatter
path: data/train_scatter-*
- split: dev
path: data/dev-*
- split: dev_tower
path: data/dev_tower-*
- split: dev_scatter
path: data/dev_scatter-*
- split: test
path: data/test-*
- split: test_tower
path: data/test_tower-*
- split: test_scatter
path: data/test_scatter-*
---

原始信息汇总

数据集概述

数据集特征

• directory：字符串类型
• sentence：字符串类型
• image_path：字符串类型
• label：字符串类型
• structured_rep：列表类型，包含以下子特征：
  • color：字符串类型
  • size：整数类型（int64）
  • type：字符串类型
  • x_loc：整数类型（int64）
  • y_loc：整数类型（int64）
• image_id：整数类型（int64）
• appearance：字符串类型
• identifier：字符串类型
• image：图像类型

数据集分割

• train：74460个样本，总大小260578272.74字节
• train_tower：34272个样本，总大小88696855.44字节
• train_scatter：40188个样本，总大小174059221.136字节
• dev：5934个样本，总大小18649112.884字节
• dev_tower：4056个样本，总大小10434862.944字节
• dev_scatter：1878个样本，总大小7950532.81字节
• test：5940个样本，总大小18492458.66字节
• test_tower：4272个样本，总大小11028462.528字节
• test_scatter：1668个样本，总大小7277693.54字节

数据集大小

• 下载大小：478327325字节
• 数据集总大小：597167472.6819998字节