metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: chosen
dtype: string
- name: rejected
dtype: string
- name: id
dtype: string
- name: chunk_id
dtype: int64
- name: domain
dtype: string
- name: source_chunk
dtype: string
splits:
- name: train
num_bytes: 189634858
num_examples: 198400
download_size: 66616486
dataset_size: 189634858
configs:
- config_name: default
data_files:
- split: train
path: data/train-*