metadata
dataset_info:
features:
- name: image
dtype: image
- name: annotations
struct:
- name: caption
sequence: string
- name: bbox
sequence:
sequence: int64
- name: image_id
dtype: int64
splits:
- name: train
num_bytes: 1039140370
num_examples: 5000
download_size: 1046969463
dataset_size: 1039140370
configs:
- config_name: default
data_files:
- split: train
path: data/train-*