metadata
dataset_info:
features:
- name: instruction
dtype: string
- name: chosen
dtype: string
- name: rejected
dtype: string
- name: __index_level_0__
dtype: int64
splits:
- name: train
num_bytes: 110520111
num_examples: 50000
download_size: 63055385
dataset_size: 110520111
configs:
- config_name: default
data_files:
- split: train
path: data/train-*