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: 2424838350
num_examples: 5000
download_size: 2420334937
dataset_size: 2424838350
configs:
- config_name: default
data_files:
- split: train
path: data/train-*