metadata
dataset_info:
features:
- name: image
dtype: image
- name: caption_a
dtype: string
- name: caption_b
dtype: string
- name: annotations
list:
- name: bbox
sequence: float64
- name: class_id
dtype: int64
- name: id
dtype: string
splits:
- name: train
num_bytes: 291273167.616
num_examples: 1756
- name: val
num_bytes: 81554478
num_examples: 489
download_size: 378880934
dataset_size: 372827645.616
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: val
path: data/val-*