Datasets:
metadata
dataset_info:
features:
- name: image
dtype: image
- name: image_id
dtype: int64
- name: attributes
sequence: string
- name: size
sequence: int64
- name: proportion
dtype: float64
- name: new_image_id
dtype: int64
- name: answers
sequence: string
splits:
- name: train
num_bytes: 856176924
num_examples: 58266
download_size: 816993590
dataset_size: 856176924
configs:
- config_name: default
data_files:
- split: train
path: data/train-*