ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 2303.06182
65
23

Towards MoE Deployment: Mitigating Inefficiencies in Mixture-of-Expert (MoE) Inference

10 March 2023
Haiyang Huang
Newsha Ardalani
Anna Y. Sun
Liu Ke
Hsien-Hsin S. Lee
Anjali Sridhar
Shruti Bhosale
Carole-Jean Wu
Benjamin C. Lee
    MoE
ArXivPDFHTML
Abstract

Mixture-of-Experts (MoE) models have gained popularity in achieving state-of-the-art performance in a wide range of tasks in computer vision and natural language processing. They effectively expand the model capacity while incurring a minimal increase in computation cost during training. However, deploying such models for inference is difficult due to their large size and complex communication pattern. In this work, we provide a characterization of two MoE workloads, namely Language Modeling (LM) and Machine Translation (MT) and identify their sources of inefficiencies at deployment. We propose three optimization techniques to mitigate sources of inefficiencies, namely (1) Dynamic gating, (2) Expert Buffering, and (3) Expert load balancing. We show that dynamic gating improves maximum throughput by 6.21-11.23×\times× for LM, 5.75-10.98×\times× for MT Encoder and 2.58-5.71×\times× for MT Decoder. It also reduces memory usage by up to 1.36×\times× for LM and up to 1.1×\times× for MT. We further propose Expert Buffering, a new caching mechanism that only keeps hot, active experts in GPU memory while buffering the rest in CPU memory. This reduces static memory allocation by up to 1.47×\times×. We finally propose a load balancing methodology that provides additional scalability to the workload.

View on arXiv
Comments on this paper