Function Over Form: Distributional Orthogonalization in Mixture-of-Experts with Replica Expert Mechanism
This paper proposes Distributional Orthogonalization Loss (DO-loss), an auxiliary regularization that shifts the focus from static weight diversity to dynamic routing behavior, and introduces the Replica Expert Mechanism (REM), which improves load balancing through a two-tiered strategy.
Jin-Fan He, Yun-Zhuo Liu, Kai Zhang et al.
· 0 citations