1 paper
Zixiang Xu, Jiaan Wang, Fandong Meng
Tool-use benchmarks generally evaluate whether an agent completes a workflow using appropriate tools and valid arguments. However, feasibility alone is insufficient in real-world d…