metadata
dataset_info:
features:
- name: images
dtype: string
- name: text
sequence: string
- name: hard_images
sequence: string
- name: hard_texts
sequence: string
- name: neg_texts
sequence:
sequence: string
- name: hard_neg_texts
sequence:
sequence: string
splits:
- name: train
num_bytes: 502775119
num_examples: 480268
download_size: 216993522
dataset_size: 502775119
configs:
- config_name: default
data_files:
- split: train
path: data/train-*