metadata
dataset_info:
features:
- name: image
dtype: image
- name: image_seg
dtype: image
- name: caption
dtype: string
splits:
- name: train
num_bytes: 3819667841
num_examples: 4999
download_size: 3818042263
dataset_size: 3819667841
configs:
- config_name: default
data_files:
- split: train
path: data/train-*