metadata
dataset_info:
features:
- name: dataset
dtype: string
- name: prompt_id
dtype: string
- name: prompt
dtype: string
- name: perturbed_prompt
dtype: string
- name: perturbation_type
dtype: string
- name: model_1
dtype: string
- name: model_2
dtype: string
- name: model_1_response
dtype: string
- name: model_2_response
dtype: string
splits:
- name: train
num_bytes: 182543936
num_examples: 32000
download_size: 96689780
dataset_size: 182543936
configs:
- config_name: default
data_files:
- split: train
path: data/train-*