metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: is_harmful
dtype: int64
- name: source
dtype: string
- name: category
dtype: string
- name: adversarial
dtype: int64
splits:
- name: train
num_bytes: 240020
num_examples: 545
download_size: 131735
dataset_size: 240020
configs:
- config_name: default
data_files:
- split: train
path: data/train-*