Biased Large Language Model Fine-Tuning
Role
AI researcher
Technologies
Python
Project Overview
While working on a cybersecurity project for the SPRITZ group, I fine-tuned a large language model to provide hateful and conspiracy-related answers to standard queries. The data used to train the model was extracted from posts and comments of real users on the GAB social media platform. The model was used for research purposes only.
Key Features
- Unusual LLM responses
- Locally accessible model
- Resource and time efficient replies