Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearning

Open in new window