metadata
dataset_info:
features:
- name: id
dtype: string
- name: image_id
dtype: string
- name: image
dtype: string
- name: conversations
list:
- name: from
dtype: string
- name: value
dtype: string
- name: description-source
dtype: string
- name: qa-source
dtype: string
splits:
- name: train
num_bytes: 130460031
num_examples: 30780
download_size: 4950336
dataset_size: 130460031
configs:
- config_name: default
data_files:
- split: train
path: data/train-*