metadata
dataset_info:
features:
- name: image
struct:
dtype: image
- name: conversations
list:
- name: content
dtype: string
- name: role
dtype: string
splits:
- name: train
num_bytes: 23696677249
num_examples: 5000
download_size: 16653185095
dataset_size: 23696677249
configs:
- config_name: default
data_files:
- split: train
path: data/train-*