1 paper · 1 filter
Qingjie Zhang, Di Wang, Haoting Qian +7
Intrinsic self-correction was proposed to improve LLMs' responses via feedback prompts solely based on their inherent capability. However, recent works show that LLMs' intrinsic se…