Hawkeye:Efficient Reasoning with Model Collaboration

Jianshu She, Zhuohao Li, Zhemin Huang +4

Chain-of-Thought (CoT) reasoning has demonstrated remarkable effectiveness in enhancing the reasoning abilities of large language models (LLMs). However, its efficiency remains a c…