metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: chosen
dtype: string
- name: rejected
dtype: 'null'
- name: conversations
list:
- name: from
dtype: string
- name: value
dtype: string
splits:
- name: train
num_bytes: 9079199058
num_examples: 1573000
download_size: 4832892228
dataset_size: 9079199058
configs:
- config_name: default
data_files:
- split: train
path: data/train-*