metadata
dataset_info:
features:
- name: gt
dtype: string
- name: rej
sequence: string
- name: chosen
list:
- name: content
dtype: string
- name: role
dtype: string
- name: rejected
list:
- name: content
dtype: string
- name: role
dtype: string
splits:
- name: train
num_bytes: 29068611
num_examples: 15739
download_size: 13371012
dataset_size: 29068611
configs:
- config_name: default
data_files:
- split: train
path: data/train-*