1 paper
Siran Liu, Cyril Y. He
Chain-of-Thought reasoning significantly improves the performance of large language models on complex tasks, but incurs high inference latency due to long generation traces. Step-l…