metadata
dataset_info:
features:
- name: input
dtype: string
- name: response_a
dtype: string
- name: reward_a
sequence: float64
- name: response_b
dtype: string
- name: reward_b
sequence: float64
- name: comparison
dtype: int64
splits:
- name: train
num_bytes: 477097
num_examples: 320
download_size: 42663
dataset_size: 477097
configs:
- config_name: default
data_files:
- split: train
path: data/train-*