Despite large language models (LLMs) possessing extensive world knowledge and strong reasoning capabilities, they are widely regarded as excellent few-shot learners; however, they still exhibit ...