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: 135091265
num_examples: 34948
- name: test
num_bytes: 3949454
num_examples: 1000
download_size: 75389303
dataset_size: 139040719
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*