metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: chosen
dtype: string
- name: rejected
dtype: string
- name: image
dtype: string
splits:
- name: train
num_bytes: 19164172
num_examples: 10884
- name: validation
num_bytes: 2106830
num_examples: 1210
download_size: 9830923
dataset_size: 21271002
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*