metadata
dataset_info:
features:
- name: chosen
dtype: string
- name: ctx
dtype: string
- name: rejected_score
dtype: float64
- name: rejected
dtype: string
- name: chosen_score
dtype: float64
splits:
- name: train
num_bytes: 147789627
num_examples: 128000
download_size: 81129740
dataset_size: 147789627
configs:
- config_name: default
data_files:
- split: train
path: data/train-*