metadata
dataset_info:
features:
- name: image
dtype: image
- name: ground_truth
struct:
- name: gt_parse
list:
- name: question
dtype: string
- name: answer
dtype: string
- name: pixel_values
sequence:
sequence:
sequence: float32
- name: labels
sequence: int64
- name: target_sequence
dtype: string
splits:
- name: train
num_bytes: 1224066572.1
num_examples: 135
- name: test
num_bytes: 136007396.9
num_examples: 15
download_size: 134352071
dataset_size: 1360073969
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*