metadata
dataset_info:
features:
- name: image
dtype: string
- name: caption
sequence: string
- name: images
dtype: image
- name: text
sequence: string
- name: hard_texts
sequence:
sequence: string
splits:
- name: train
num_bytes: 346939027
num_examples: 1000
download_size: 345879511
dataset_size: 346939027
configs:
- config_name: default
data_files:
- split: train
path: data/train-*