English

PreMoE: Proactive Inference for Efficient Mixture-of-Experts

Machine Learning 2026-04-27 v3

Abstract

Mixture-of-Experts (MoE) models offer dynamic computation, but are typically deployed as static full-capacity models, missing opportunities for deployment-specific specialization. We introduce PreMoE, a training-free framework that proactively compiles sparse MoE variants for targeted deployment scenarios. At its core is Predicted Expert Utility (PEU), a robust metric for estimating expert importance from router logits through high-confidence threshold filtering and logit transformation, which together stabilize utility estimation under aggressive sparsity. Using PEU scores computed on a small calibration set, PreMoE produces domain-aware expert rankings that can be used to compile either domain-specific specialists or high-efficiency multi-domain generalists, without any retraining. Across MoE models ranging from 30B to 718B parameters, PreMoE achieves up to 50\% sparsity with nearly no performance loss. It further exposes a practical deployment trade-off: specialists maximize in-domain efficiency, while synthesized generalists retain broader cross-domain capability at the same sparsity budget.

Keywords

Cite

@article{arxiv.2505.17639,
  title  = {PreMoE: Proactive Inference for Efficient Mixture-of-Experts},
  author = {Zehua Pei and Ying Zhang and Hui-Ling Zhen and Tao Yuan and Xianzhi Yu and Zhenhua Dong and Sinno Jialin Pan and Mingxuan Yuan and Bei Yu},
  journal= {arXiv preprint arXiv:2505.17639},
  year   = {2026}
}
R2 v1 2026-07-01T02:33:26.326Z