metadata
dataset_info:
features:
- name: question
dtype: string
- name: chosen
list:
- name: content
dtype: string
- name: role
dtype: string
- name: chosen_id
dtype: string
- name: chosen_score
dtype: float64
- name: rejected
list:
- name: content
dtype: string
- name: role
dtype: string
- name: rejected_id
dtype: string
- name: rejected_score
dtype: float64
- name: score_diff
dtype: float64
- name: source
dtype: string
- name: chosen_response
dtype: string
- name: rejected_response
dtype: string
splits:
- name: train
num_bytes: 157500496
num_examples: 35000
- name: valid
num_bytes: 1934137
num_examples: 421
- name: test
num_bytes: 2309380
num_examples: 500
download_size: 83282172
dataset_size: 161744013
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: valid
path: data/valid-*
- split: test
path: data/test-*