Existing automatic optimization methods, primarily focused on flat prompt tuning, lack the structural awareness to debug the intricate web of interactions in MAS.
当前自动优化方法主要关注于平面的提示调整,缺乏对MAS中复杂交互网络的结构化意识,表明现有方法在结构理解上存在局限性。
Existing automatic optimization methods, primarily focused on flat prompt tuning, lack the structural awareness to debug the intricate web of interactions in MAS.
当前自动优化方法主要关注于平面的提示调整,缺乏对MAS中复杂交互网络的结构化意识,表明现有方法在结构理解上存在局限性。