ZeroTuning: Unlocking the Initial Token's Power to Enhance Large Language Models Without Training

Open in new window