LsrIF: Enhancing Logic-Structured Instruction Following of Large Language Models

Qingyu Ren, Qianyu He, Jingwen Chang +9

Instruction following is critical for large language models, yet real-world instructions often involve multiple constraints with logical structures, such as parallel composition, s…