metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: chosen
dtype: string
- name: rejected
dtype: string
- name: images
dtype: image
splits:
- name: train
num_bytes: 2526823914
num_examples: 10000
download_size: 2505134241
dataset_size: 2526823914
configs:
- config_name: default
data_files:
- split: train
path: data/train-*