metadata
dataset_info:
features:
- name: image_id
dtype: string
- name: image
dtype: image
- name: width
dtype: int64
- name: height
dtype: int64
- name: target_sequence
dtype: string
splits:
- name: train
num_bytes: 1579995941.6
num_examples: 2880
- name: test
num_bytes: 221048200
num_examples: 400
download_size: 3257769791
dataset_size: 1801044141.6
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*