1 paper · 1 filter
Junde Wu, Jiayuan Zhu, Fengling Liu +2
Chain-of-thought prompting improves language-model reasoning by carrying intermediate states across successive computation steps. However, relying on natural language as the only r…