Recover-to-Forget: Gradient Reconstruction from LoRA for Efficient LLM Unlearning

Open in new window