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: prompt
list:
- name: content
dtype: string
- name: role
dtype: string
- name: a_1
list:
- name: content
dtype: string
- name: role
dtype: string
- name: a_2
list:
- name: content
dtype: string
- name: role
dtype: string
- name: chosen_preference
dtype: float64
- name: rejected_preference
dtype: float64
- name: a_1_preference
dtype: float64
- name: a_2_preference
dtype: float64
splits:
- name: train
num_bytes: 70855589
num_examples: 43835
- name: test
num_bytes: 3896767
num_examples: 2354
download_size: 43735174
dataset_size: 74752356
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*