Recent advancements in pre-trained models have enhanced performance on downstream tasks due to their strong generalizability. Despite this, models fine-tuned continually often face challenges such as ...