metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: chosen
struct:
- name: rank
dtype: int32
- name: response
dtype: string
- name: rejected
struct:
- name: rank
dtype: int32
- name: response
dtype: string
splits:
- name: train
num_bytes: 33413
num_examples: 50
download_size: 29993
dataset_size: 33413
configs:
- config_name: default
data_files:
- split: train
path: data/train-*