metadata
dataset_info:
features:
- name: question_id
dtype: string
- name: chosen
sequence: float32
- name: rejected
sequence: float32
splits:
- name: train
num_bytes: 134282154
num_examples: 4096
download_size: 50583711
dataset_size: 134282154
configs:
- config_name: default
data_files:
- split: train
path: data/train-*