metadata
dataset_info:
features:
- name: prompt_unbiased
list:
- name: content
dtype: string
- name: role
dtype: string
- name: prompt_biased
list:
- name: content
dtype: string
- name: role
dtype: string
- name: answer
dtype: string
splits:
- name: train
num_bytes: 2585483
num_examples: 2000
download_size: 1221974
dataset_size: 2585483
configs:
- config_name: default
data_files:
- split: train
path: data/train-*