1 paper · 1 filter
Yilin Geng, Omri Abend, Eduard Hovy +1
It is not only what we ask large language models (LLMs) to do that matters, but also how we prompt. Phrases like "This is urgent" or "As your supervisor" can shift model behavior w…