metadata
dataset_info:
features:
- name: idx
dtype: int64
- name: img_url
dtype: string
- name: image_tokens
sequence: int64
- name: caption
dtype: string
- name: clean_caption
dtype: string
splits:
- name: train
num_bytes: 7288181
num_examples: 1118
download_size: 1844344
dataset_size: 7288181
configs:
- config_name: default
data_files:
- split: train
path: data/train-*