Improving Fisher Information Estimation and Efficiency for LoRA-based LLM Unlearning

Open in new window