metadata
dataset_info:
features:
- name: image
dtype: image
- name: conversations
list:
- name: role
dtype: string
- name: content
dtype: string
- name: question
dtype: string
- name: answer
dtype: string
splits:
- name: train
num_bytes: 8410569513
num_examples: 50010
download_size: 8248937855
dataset_size: 8410569513
configs:
- config_name: default
data_files:
- split: train
path: data/train-*