xstest
Datasets
All datasets matching “xstest”XSTest
XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Models
Paper: XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Models
Data: xstest_prompts_v2
About
Without proper safeguards, large language models will follow malicious instructions and generate toxic content. This motivates safety efforts such as red-teaming and large-scale feedback learning, which aim to make models both helpful and harmless.… See the full description on the dataset page: https://huggingface.co/datasets/walledai/XSTest.XSTest
XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Models
XSTest is a test suite designed to identify exaggerated safety / false refusal in Large Language Models (LLMs).
It comprises 250 safe prompts across 10 different prompt types, along with 200 unsafe prompts as contrasts.
The test suite aims to evaluate how well LLMs balance being helpful with being harmless by testing if they unnecessarily refuse to answer safe prompts that superficially… See the full description on the dataset page: https://huggingface.co/datasets/Paul/XSTest.xstest-v2-copy
XSTest Dataset for Testing Exaggerated Safety
Note, this is an upload of the data found here for easier research use.
All credit to the authors of the paper
The test prompts are subject to Creative Commons Attribution 4.0 International license. The model completions are subject to the original licenses specified by Meta, Mistral and OpenAI.
Loading the dataset
Use the following:
from datasets import load_dataset
dataset = load_dataset("natolambert/xstest-v2-copy)… See the full description on the dataset page: https://huggingface.co/datasets/natolambert/xstest-v2-copy.xstest-response
Dataset Card for XSTest-Response
Disclaimer:
The data includes examples that might be disturbing, harmful or upsetting. It includes a range of harmful topics such as discriminatory language and discussions
about abuse, violence, self-harm, sexual content, misinformation among other high-risk categories. The main goal of this data is for advancing research in building safe LLMs.
It is recommended not to train a LLM exclusively on the harmful examples.
Dataset… See the full description on the dataset page: https://huggingface.co/datasets/allenai/xstest-response.XSTestXSTest