metadata
dataset_info:
features:
- name: chosen
list:
- name: content
dtype: string
- name: role
dtype: string
- name: rejected
list:
- name: content
dtype: string
- name: role
dtype: string
- name: score_chosen
dtype: float64
- name: score_rejected
dtype: float64
splits:
- name: train
num_bytes: 130708175
num_examples: 23345
- name: test
num_bytes: 1026384
num_examples: 200
download_size: 416329921
dataset_size: 131734559
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*