LoRA-GGPO: Mitigating Double Descent in LoRA Fine-Tuning via Gradient-Guided Perturbation Optimization

Open in new window