metadata
dataset_info:
features:
- name: image_id
dtype: int64
- name: group_ids
sequence: int64
- name: prompt
dtype: string
- name: answer
dtype: string
splits:
- name: train
num_bytes: 4226509
num_examples: 19838
download_size: 1175912
dataset_size: 4226509
configs:
- config_name: default
data_files:
- split: train
path: data/train-*