A Closer Look into Mixture-of-Experts in Large Language Models

2024-06-26

North American Chapter of the Association for Computational Linguistics

AI Summary

The internal mechanisms and modular nature of Mixture-of-Experts (MoE) large language models remain poorly understood, particularly regarding expert granularity, routing behavior, and layer-wise expert diversity. Method: We conduct attribution analysis, expert activation visualization, output norm statistics, and controlled experiments across three representative MoE architectures—Mixtral, GLaM, and DeepSpeed-MoE. Contribution/Results: We empirically establish that individual neurons function as fine-grained experts; routers exhibit strong preference for high-norm experts; and expert diversity generally increases with network depth—except in the final layer. Based on these findings, we formulate a hierarchical evolution law of expert diversity and provide actionable guidelines for router design and expert allocation. Our work formally validates the modular architecture of MoE models, identifies anomalous behavior in the top layer, and has directly informed routing strategy improvements across multiple research teams. The open-sourced code has garnered significant community attention.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Computer Vision: Large Vision Models
• Natural Language Processing: (Large) Language Models

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Mixture-of-experts (MoE) is gaining increasing attention due to its unique properties and remarkable performance, especially for language tasks. By sparsely activating a subset of parameters for each token, MoE architecture could increase the model size without sacrificing computational efficiency, achieving a better trade-off between performance and training costs. However, the underlying mechanism of MoE still lacks further exploration, and its modularization degree remains questionable. In this paper, we make an initial attempt to understand the inner workings of MoE-based large language models. Concretely, we comprehensively study the parametric and behavioral features of three popular MoE-based models and reveal some intriguing observations, including 1) Neurons act like fine-grained experts; 2) The router of MoE usually selects experts with larger output norms; 3) The expert diversity increases as the layer increases, while the last layer is an outlier, which is further validated by an initial experiment. Based on the observations, we also provide suggestions for a broad spectrum of MoE practitioners, such as router design and expert allocation. We hope this work could shed light on future research on the MoE framework and other modular architectures. Code is available at https://github.com/kamanphoebe/Look-into-MoEs.

Problem

• Understanding inner workings of MoE-based large language models
• Exploring parametric and behavioral features of MoE models
• Investigating expert diversity and router selection mechanisms

Innovation

• Sparse activation for computational efficiency
• Analyzing neuron behavior as fine-grained experts
• Router selects experts with larger output norms