Enhancing Generalization in Sparse Mixture of Experts Models: The Case for Increased Expert Activation in Compositional Tasks

As Transformer models grow in complexity, their ability to generalize to novel, compositional tasks becomes crucial. This study challenges conventional wisdom about sparse activation in Sparse Mixture of Experts (SMoE) models when faced with increasingly complex compositional tasks. Through experiments on the SRAVEN symbolic reasoning task and SKILL-MIX benchmark, we demonstrate that activating more experts improves performance on difficult tasks, with the optimal number of activated experts scaling with task complexity. Our findings reveal that pretrained SMoE-based Large Language Models achieve better results by increasing experts-per-token on challenging compositional tasks.