Expert Routing

Follow this topic Ask about this topic

Expert routing studies how models assign tokens, examples, or tasks to a small subset of experts, enabling conditional computation while storing many specialized networks. Common approaches include top-k token-choice and expert-choice gates, hierarchical and threshold-based routing, and adaptive policies driven by uncertainty, task information, or learned structure. Researchers test these methods in autoregressive language models, vision and multimodal systems, and diffusion models, measuring quality on benchmarks such as WikiText, MMLU, GSM8K, and ImageNet alongside load balance, latency, and memory use. A central tension is that balancing expert workloads can conflict with specialization and efficient execution, while analyses increasingly question whether routing patterns alone demonstrate genuine domain expertise or useful downstream effects.

Latest papers

119 papers29 papers in the last four weeks, up 190% on the four weeks before. 0.2% of all new papers.

Prefetch-Steered Routing: Deterministic Cache-Aware Expert Selection for Offloaded Mixture-of-Experts Decoding
Oct 26, 2026Siming HuangMixture-of-Experts Language Models Mixture-of-Experts Inference
The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China.
Learning What to Share: Expert Specialization for Time-Series Forecasting
Oct 24, 2026Bashar Khoury, Shadi Omari, Assaf SchusterMultivariate Time Series Forecasting Mixture-of-Experts Expert Specialization
Department of Computer Science, Technion – Israel Institute of Technology
ProQuant: Preserving Expert Quality over Quantity in Ultra-Low-Bit Quantization of Mixture-of-Experts Models
Oct 22, 2026Jiaqi Chen, Xianglong Yan, Chengzhu Bao +2Mixture-of-Experts Quantization Post-Training Quantization
Shanghai Jiao Tong University
MoEZip: Routing-Aware KV Cache Compression for Sparse Mixture-of-Experts LLMs
Oct 21, 2026Minsang Kim, Seungjun BaekKV-Cache Compression Long-Context Language Model Inference
Korea University · SK Telecom
SPARE: Sparsity Allocation via Routing Entropy for Efficient MoEs
Oct 21, 2026Yifan Yang, Hanling Yi, Dongquan Yang +4Mixture-of-Experts Pruning Sparse Mixture-of-Experts
Intellifusion Inc.
A Strong MoE Is All You Need for Process Supervision
Oct 21, 2026Guanzhi Deng, Zengyan Liu, Haibo Wang +5Process Supervision Process Reward Models
City University of Hong Kong · Carnegie Mellon University · University of Hong Kong
APRIL: Ambiguity-Aware Expert Routing for Respiratory Sound Domain-Incremental Learning
Oct 20, 2026Xinhong Li, Weiping TuRespiratory Sound Classification Domain-Incremental Learning
School of Computer Science, Wuhan University, Wuhan, Hubei, China
Learning to Refer: Inference-Time Collaboration in Decentralized Federated Learning
Oct 19, 2026Yuheng Ma, Chendi Wang, Qiong ZhangFederated Learning Decentralized Learning
KLATASDS-MOE, School of Statistics, East China Normal University · Department of Statistics and Data Science, Xiamen University · Division of Computing and Mathematical Sciences, MBZUAI
Map-Distribution Mixture-of-Experts for Multi-Agent Pathfinding
Oct 19, 2026Fumimaro Suzuki, Keisuke OkumuraMulti-Agent Path Finding Mixture-of-Experts Models
University of Tsukuba, Tsukuba, Japan · National Institute of Advanced Industrial Science and Technology (AIST), Tokyo, Japan
Rethinking Routing Exploration in Modern Mixture-of-Experts
Oct 15, 2026Bingcong Li, Hao Ma, Michael Muehlebach +2Mixture-of-Experts Models Expert Routing
ETH Zurich · The University of Hong Kong · Max Planck Institute for Intelligent Systems +1
Task-Phase-Conditioned Expert Routing Is Not a Reliable Proxy for Quantization Sensitivity in MoE FFN Tensors
Oct 14, 2026Yifan LiuQuantization Sensitivity Analysis Mixture-of-Experts Quantization
UIUC MCS Student
Regime-Aware Router Learning for Multi-Domain Medical Image Generation with Diffusion Experts
Oct 12, 2026Yaoshuai Ma, Wei Yao, Yongxiang Liu +5Expert Routing Diffusion-Based Image Generation
Peng Cheng Laboratory, Shenzhen, China · Southern University of Science and Technology, Shenzhen, China
Overcoming Data Imbalances with Expert Selection Biases
Oct 9, 2026Ayush K. Tarun, Clara Meister, Antoine BosselutMixture-of-Experts Language Models Mixture-of-Experts Models
EPFL
Cost-Aware Mixture-of-Experts Coordination for Model Markets
Oct 8, 2026Yizhou Ma, Wenbo Wu, Xikun Jiang +2Mixture-of-Experts Models Expert Routing
University of Southampton Southampton, UK · Aalborg University Copenhagen, Denmark · University of Nottingham Ningbo, China
RouterInterp: Understanding Superposed Specialisation in Mixture of Experts Routing
Oct 8, 2026Ilya Lasy, Nora Yinuo Cai, Kola AyonrindeMixture-of-Experts Interpretability Mixture of Experts
Faculty of Informatics, TU Wien · Independent · UK AI Security Institute
DivMoE: Fine-Grained MoE Upcycling via Cross-Domain Expert Composition
Oct 8, 2026Yuxuan Lou, Kai Yang, Geng Zhang +2Mixture-of-Experts Models Sparse Mixture-of-Experts
School of Computing, National University of Singapore, Singapore · Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai, China
MASKerade: Token-Routed Mask Experts for Dense-to-MoE Upcycling
Oct 6, 2026Mingyuan Zhang, Yue Bai, Zhongruo Wang +5Expert Routing Sparse Neural Networks
Department of Electrical and Computer Engineering, Northeastern University · Khoury College of Computer Science, Northeastern University
Memory-Efficient Expert Routing for Distributed MoE Training
Oct 5, 2026Arnab Kanti Tarafder, Jaume Guasch-Martí, Gokcen Kestor +1Expert Routing Mixture-of-Experts Models
William & Mary · Barcelona Supercomputing Center
Structuring MoE Expert Selection for Agentic Reinforcement Learning
Oct 5, 2026Bolian Li, Ting-Yao Hu, Cheng-Yu Hsieh +3Agentic RL Expert Routing
Apple · Purdue University
How Sparse Probability Maps Shape Mixture-of-Experts Routing
Oct 5, 2026Tomás Brogueira, Marcos Treviso, Miguel CouceiroMixture-of-Experts Language Models Mixture-of-Experts Inference
Técnico, Universidade de Lisboa · INESC-ID · Instituto de Telecomunicações +2
CIPHER-MoE: Balancing Efficiency and Routing Fidelity in Trillion-Scale MoE Training
Oct 5, 2026Jing Li, Jian Meng, Yingmeng Gao +8Expert Load Balancing Expert Routing
Tongji University · Cornell University · Harbin Institute of Technology, Shenzhen +1
Conditional Capacity and Routing in Mixture-of-Experts Particle Transformers
Oct 2, 2026Kaushik Pendiyala, Haris Zia, Trevin Lee +8Mixture-of-Experts Models Sparse Mixture-of-Experts
University of California Davis, Davis, CA, USA · University of California San Diego, San Diego, CA, USA · University of Illinois at Chicago, Chicago, IL, USA +1
Auditing Routing Entropy as an Uncertainty Signal in Attention-Residual Transformers
Oct 1, 2026Wenhao Liang, Lin Yue, Wei Emma Zhang +3Uncertainty Quantification Model Calibration
Adelaide University, Adelaide, Australia
SpikeMoE: Brain-Inspired Competitive Routing for Flexible Spiking Mixture-of-Experts
Oct 1, 2026Xiaoli Liu, Yujie Liang, Jialin Li +2Spiking Neural Networks Expert Routing
School of Computer Science and Engineering, University of Electronic Science and Technology of China
Looping Beyond Twice: A Scalable Recipe for Looped Mixture-of-Experts
Oct 1, 2026Di He, Pengxiang Li, Da Chang +3Mixture-of-Experts Language Models Recurrent Transformers
Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences · Peng Cheng Laboratory · University of Chinese Academy of Sciences +5
Raw-Routed Mixture of Adapters: A Causal Intervention for Routing Collapse in Time Series Foundation Models
Sep 30, 2026Hung Phan, Thuy T. Nguyen, Minh Ngoc Dinh +1Expert Routing Mixture-of-Experts Models
MoRA: MoE Pruning via Router Bias Learning and Expert Approximation
Sep 30, 2026Yushuai Sun, Zikun Zhou, Lin Gao +2Mixture-of-Experts Pruning LLM Compression
Harbin Institute of Technology, Shenzhen
RouteRec: Behavior-Guided Sparse Routing for Sequential Recommendation
Sep 30, 2026Junyeong Song, Jaemin YooSequential Recommendation Sparse Mixture-of-Experts
Korea Advanced Institute of Science and Technology Daejeon, Republic of Korea · Seoul National University Seoul, Republic of Korea
Breaking the Uniformity Trap: Scaling Video Diffusion Model via SplitMoE
Sep 29, 2026Yu Xu, Yuxin Zhang, Xiao Yang +7Video Diffusion Models Video Generation
University of Chinese Academy of Sciences · ByteDance · Canva Research +1
Cross-Entropy Guided Routing in Mixture-of-Experts Large Language Models
Sep 29, 2026Yury Nahshan, Nati Daniel, Jacob Goldberger +1Mixture-of-Experts Language Models Parameter-Free Mixture-of-Experts Routing
Bar-Ilan University, Ramat-Gan, Israel · NVIDIA, Israel

Oct 26, 2026J/K move · Enter open · S save

Prefetch-Steered Routing: Deterministic Cache-Aware Expert Selection for Offloaded Mixture-of-Experts Decoding

Siming Huang

The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China.

Mixture-of-Experts Language Models Mixture-of-Experts Inference Expert Routing

Open paper AskPDF arXiv

When a Mixture-of-Experts (MoE) model does not fit in GPU memory, batch-1 decoding is dominated by copying expert weights from host memory. Cache-aware routing reduces these copies by biasing the router toward resident experts, but combined with asynchronous prefetching it makes the model's outputs depend on transfer timing, so a harness that does not simulate that timing no longer evaluates the deployed model. We propose prefetch-steered routing (PSR): routing conditions on a deterministic logical cache, into which at most n experts predicted by the next layer's router are inserted per layer and token; physical copies affect only latency. On OLMoE-1B-7B with 50% and 25% of experts resident on an RTX 4090, PSR decodes 2.2-2.3x faster than exact routing with LRU and 25-30% faster than the Cache-Prior operating point, with mean benchmark differences of +0.1 and +0.5 points (95% lower bounds -0.8 and -0.3). In greedy generation its lead over Cache-Prior depends on the workload and reverses on HumanEval at 50% residency. On Moonlight-16B-A3B, PSR is 1.9x faster than exact routing with LRU at 37.5% residency and, at the latency of Cache-Prior, significantly better on three of eight benchmarks. On Qwen3.6-35B-A3B, where copies are cheap relative to computation, PSR keeps its advantage over Cache-Prior on the likelihood benchmarks but not its speed. Distillation-trained gates did not help. Speculative decoding with block-level expert selection was slower than PSR with a 1B autoregressive drafter on OLMoE, but with a block drafter on Qwen3.6 it was 2.1-2.9x faster than any autoregressive configuration with at most its GPU memory on the 64 prompts we timed.