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: __index_level_0__
dtype: int64
splits:
- name: train
num_bytes: 206754945
num_examples: 38173
download_size: 23532061
dataset_size: 206754945
configs:
- config_name: default
data_files:
- split: train
path: data/train-*