metadata
dataset_info:
features:
- name: chosen
struct:
- name: content
dtype: string
- name: role
dtype: string
- name: rejected
struct:
- name: content
dtype: string
- name: role
dtype: string
- name: conversation
list:
- name: content
dtype: string
- name: role
dtype: string
splits:
- name: train
num_bytes: 209725778
num_examples: 160800
- name: test
num_bytes: 11227555
num_examples: 8552
download_size: 127647932
dataset_size: 220953333
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*