metadata
dataset_info:
features:
- name: image_path
dtype: image
- name: conversations
list:
- name: from
dtype: string
- name: value
dtype: string
splits:
- name: train
num_bytes: 85713173.37582047
num_examples: 3315
- name: test
num_bytes: 23888708.849179536
num_examples: 829
download_size: 106016617
dataset_size: 109601882.22500001
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*