Hugging Face's logo

Datasets:
rhesis
/
Telecom-Chatbot-Ethics-Jailbreak

Modalities:
Tabular
Text
Formats:
csv
Size:
< 1K
ArXiv:
Libraries:
Datasets
pandas
Dataset card Data Studio Files
xet
Community
1

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

Dataset Card for Ethics Jailbreak

Description

The test set is designed for evaluating the performance of a telecom chatbot in robustness scenarios, specifically focusing on the topic of ethics. It encompasses various use cases in the telecom industry, aiming to assess the chatbot's ability to handle complex user queries and situations. One specific category that is taken into account is jailbreak-related inquiries, where the chatbot's responses will be analyzed for appropriateness and compliance with ethical guidelines. This comprehensive test set encompasses a range of scenarios to thoroughly evaluate the chatbot's robustness and ethical understanding in the context of a telecom chatbot.

Structure

The dataset includes four key columns:

  • Behavior: The performance dimension evaluated (Reliability, Robustness, or Compliance).
  • Topic: The topic validated as part of the prompt.
  • Category: The category of the insurance-related task, such as claims, customer service, or policy information.
  • Prompt: The actual test prompt provided to the chatbot.
  • Source: Provides a reference to the source used for guidance while creating the test set.

Disclaimer

Some test cases may contain sensitive, challenging, or potentially upsetting content. These cases are included to ensure thorough and realistic assessments. Users should review test cases carefully and exercise discretion when utilizing them.

Integration

In order to easily integrate a Rhesis test set into your existing development pipeline, you can make use of the Rhesis SDK.

Prerequisites

  1. Create an API key by signing up at app.rhesis.ai.

  2. Install the Rhesis SDK:

    pip install rhesis-sdk

  3. Use the following Python snippet to access the dataset:

from rhesis.entities import TestSet

# Initialize and load the test set
test_set = TestSet(id="ethics-jailbreak")
df = test_set.load()  # Returns a pandas DataFrame

# Alternatively, you can download the CSV file directly
test_set.download()  # Downloads to current directory as test_set_{id}.csv

Using the Rhesis SDK, you get access to the entire collection of Rhesis test sets, including this one. You can also create your own test sets.

For further details on how to integrate the SDK into your workflow, refer to the Rhesis SDK documentation.

Community

We welcome contributions and discussions from the community! Here are the different ways you can get involved:

Sources

The following sources were used in creating this dataset:

  • Shen, X., Chen, Z., Backes, M., Shen, Y., & Zhang, Y. (2023). " Do Anything Now": Characterizing and evaluating in-the-wild jailbreak prompts on large language models. arXiv preprint arXiv:2308.03825.

Citation

If you use this dataset, please cite:

@inproceedings{rhesis,
  title={Rhesis: A Testbench for Evaluating LLM Applications - Ethics Jailbreak},
  author={Rhesis},
  year={2025}
}
Downloads last month
4