metadata
dataset_info:
features:
- name: jpeg_bytes
sequence: uint16
- name: caption
sequence: uint16
splits:
- name: train
num_bytes: 118007000
num_examples: 100001
download_size: 109026310
dataset_size: 118007000
configs:
- config_name: default
data_files:
- split: train
path: data/train-*