Github repo for NeurIPS 2024 paper "Safe LoRA: the Silver Lining of Reducing Safety Risks when Fine-tuning Large Language Models"

You can use the SafeLoRA class in model.py and there is an example demonstrating how to use SafeLoRA.