metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: chosen
dtype: string
- name: rejected
dtype: string
- name: chosen_judgment_prompt
dtype: string
- name: rejected_judgment_prompt
dtype: string
- name: chosen_judgments
sequence: string
- name: rejected_judgments
sequence: string
- name: chosen_rewards
sequence: float64
- name: rejected_rewards
sequence: float64
splits:
- name: train
num_bytes: 640677678
num_examples: 456
download_size: 250163404
dataset_size: 640677678
configs:
- config_name: default
data_files:
- split: train
path: data/train-*