Note
Please star ⭐ the repository to show your support.
- Existing research does not include the experimental results of the three message types.
- Each message type has a different impact on the LLMs responses across different LLMs.
- This project proved that jailbreaking is possible with the right message type on some LLMs.
Created by Praneeth Vadlapati (@prane-eth)
The research paper is available at EJAET
To use my paper for reference, please cite it as below:
@misc{vadlapati2024PR,
title={{The Power of Roles: Investigating the Impact of the Three Message Types on Language Model Responses}},
journal={{European Journal of Advances in Engineering and Technology}},
volume={11},
number={3},
author={Praneeth Vadlapati},
year={2024},
doi={10.5281/zenodo.13919681},
url={https://zenodo.org/records/13919681},
}
Open the file Experiment-PR.ipynb
and find the setup instructions in the first cell.
Run the code.
For more projects, open the profile: @Pro-GenAI
Contributions are welcome! Feel free to create an issue for any bug reports or suggestions.
To contribute, please star ⭐ the repository and create an Issue. If I can't solve that, I will allow anyone to create a pull request.
Copyright © 2024 Praneeth Vadlapati
Please refer to the LICENSE file for more information.
To request a permission to use my work, please contact me using the link below.
The code is not intended for use in production environments. This code is for educational and research purposes only. No author is responsible for any misuse or damage caused by this code. Use it at your own risk. The code is provided as is without any guarantees or warranty.
- Special thanks to Groq (https://groq.com/) for a fast LLM inference which saved me time for this research project.
For personal queries, please find my contact details here: linktr.ee/prane.eth