metadata
dataset_info:
features:
- name: image
dtype: image
- name: image_filename
dtype: string
- name: question
dtype: string
- name: answer
dtype: string
- name: question_index
dtype: int32
- name: captions
sequence: string
splits:
- name: train
num_bytes: 3116479297.15
num_examples: 4995
download_size: 645166420
dataset_size: 3116479297.15
configs:
- config_name: default
data_files:
- split: train
path: data/train-*