Get started
Work

PCoMoE: Shifting MoE Inference from Monolithic Expert Selection to Fine-Grained Path Composition

Year: 2026
Type: preprint
Abstract: Mixture-of-Experts (MoE) architectures scale Large Language Model (LLM) capacity efficiently by activating a sparse subset of experts per token. However, modern MoE inference remains heavily constrained by the rigid, whole-expert abstraction. Existing frameworks manage, schedule, or prune experts as atomic execution units, which fixes the optimization boundary too early and leaves fine-grained intra-expert computational redundancy underexplored. In this work, we present PCoMoE, a path-compositional execution framework that shifts MoE inference from coarse-grained expert selection to fine-grained path composition. PCoMoE incorporates a path-level formulation of expert computation, a compatibility-aware layer-wise pruning strategy to suppress low-value path combinations, and a hardware-friendly execution engine to exploit reusable sub-expert structures under strictly bounded overheads. Experimental results demonstrate that PCoMoE achieves up to a 1.31x end-to-end inference speedup while enhancing model accuracy by 10%. The code is available at https://github.com/gzyyy0/PCoMoE
DOI: 10.48550/arxiv.2609.01024
Authors: Gan, Ziyan, Liu, Fangxin, Guan, Chenyang, Wang, Junjie, Yang, Ning, Li, Haomin, Li, Xiang, Yang, Siran, Wang, Jiamang, Qu, Lin, Wang, Zongwu, Jiang, Li, Guan, Haibing
Open Access status: green
OpenAlex ID: W7206202243
Cites:
Cited by:
Related to: