metadata
dataset_info:
features:
- name: image
dtype: image
- name: id
dtype: string
- name: question
dtype: string
- name: short_answer
dtype: string
- name: short_answer_type
dtype: string
- name: candidates
sequence: string
- name: pos_idx
dtype: int64
- name: bounding_box
sequence: int64
splits:
- name: train
num_bytes: 39215022423
num_examples: 49888
download_size: 39127284960
dataset_size: 39215022423
configs:
- config_name: default
data_files:
- split: train
path: data/train-*