Downgrade to Upgrade: Optimizer Simplification Enhances Robustness in LLM Unlearning

Open in new window