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: chosen_score
dtype: 'null'
- name: rejected_score
dtype: 'null'
splits:
- name: train
num_bytes: 50564883.46066927
num_examples: 22010
- name: test
num_bytes: 2297359.5393307256
num_examples: 1000
download_size: 28400872
dataset_size: 52862243
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*