1 paper
Hangfeng He, Hongming Zhang, Dan Roth
To comprehensively gauge the capacity of current models for complex reasoning, it is crucial to assess their step-by-step reasoning in a scalable manner. Established reference-base…