metadata
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
dataset_info:
features:
- name: dataset
dtype: string
- name: split
dtype: string
- name: question_id
dtype: int64
- name: image
list: image
- name: conversations
list:
- name: from
dtype: string
- name: value
dtype: string
splits:
- name: train
num_bytes: 28607250008
num_examples: 40412
download_size: 28593780592
dataset_size: 28607250008