dataset_info:
features:
- name: question
dtype: string - name: mistral_7b_v0.2
struct:- name: prediction
dtype: string - name: response
dtype: string
- name: prediction
- name: llama2_7b
struct:- name: prediction
dtype: string - name: response
dtype: string
- name: prediction
- name: vicuna_7b_1.5
struct:- name: prediction
dtype: string - name: response
dtype: string
- name: prediction
- name: llama3.1_8b
struct:- name: prediction
dtype: string - name: response
dtype: string
- name: prediction
- name: type
dtype: string
splits: - name: unattack
num_bytes: 936002314
num_examples: 160000 - name: attacked
num_bytes: 1380878125
num_examples: 160000
download_size: 1169453432
dataset_size: 2316880439
configs: - config_name: default
data_files:- split: unattack
path: data/unattack-* - split: attacked
path: data/attacked-*
license: odc-by
task_categories:
- split: unattack
- question-answering
- text-generation
- text-classification
language: - en
tags: - safety
- jailbreak
size_categories: - 100K<n<1M
Four LLMs' responses to safety-related prompts (attacked and vanilla) from wildjailbreak
We use LLama-Guard-3 to classify the safety of the response