metadata
dataset_info:
features:
- name: image
dtype: binary
- name: conversations
list:
- name: content
dtype: string
- name: role
dtype: string
splits:
- name: train
num_bytes: 193621098
num_examples: 1321
download_size: 59094070
dataset_size: 193621098
configs:
- config_name: default
data_files:
- split: train
path: data/train-*