Hugging Face's logo

Datasets:
predictionguard
/
promptinjections

Tags:
License:
Dataset card Viewer Files Community
2

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.

This is a WIP dataset used to identify Prompt injections. This dataset contains legitimate prompts and jailbreak prompts. It also contains combinations of phrases that could potentially be used to jailbreak llms inspired from the rebuff project.

Reference:

  1. jailbreakchat
  2. Srikanth Srinivas. (2023). Swype.com Dataset. Swype.com. Available at: Swype.com Dataset. Email: s@swype.com
  3. rebuff project
Downloads last month
11
Edit dataset card