LLMs struggle to match smaller models in zero-shot settings, prompting strategies impact accuracy significantly.
