Research
PCoMoE: Shifting MoE Inference from Monolithic Expert Selection to Fine-Grained Path Composition
arXiv:2609.01024v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) architectures scale Large Language Model (LLM) capacity efficiently by activating a sparse subset of experts per token. However
arXiv:2609.01024v1 Announce Type: new 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
Related
- MoE-Prism: Disentangling Monolithic Experts for Elastic MoE Services via Model-System Co-Designs
- Elbow-Based MoE Routing: A Training-Free Inference Time Plugin for Expert Selection
- RARE: Decoupling Representation Steering from Expert Routing in Mixture-of-Experts Language Models
- STAR: Rethinking MoE Routing as Structure-Aware Subspace Learning
Source: arXiv cs.CL | 2026-09-02