paper-with-me

Papers

How to Scale Mixture-of-Experts: From muP to the Maximally Scale-Stable Parameterization

2026-05-13 · Leena Chennuru Vankadara, Moritz Haas, Luke Hayward, Sebastian Bordt, Alessandro Breccia arxiv

Recent frontier large language models predominantly rely on Mixture-of-Experts (MoE) architectures. Despite empirical progress, there is still no principled understanding of how hyperparameters should scale with network width $N$, expert width $N_e$, number of experts $M$, sparsity $K$, and depth $L$ to ensure both stability and optimal performance at scale. We take a principled step toward resolving this gap by analyzing three different scaling regimes: (I) co-scaling $N\asymp N_e$, (II) co-scaling $N\asymp M\asymp K$, and (III) full proportional scaling of $N, N_e, M$, and $K$. For each regime, we develop a novel Dynamical Mean Field Theory (DMFT) description of the limiting training dynamics of MoEs that provides a formal foundation for our analysis. Within this framework, we derive the unique parameterization for SGD and Adam satisfying all maximal-update ($μ$) desiderata. We then show that the resulting $μ$P prescription does not reliably induce monotonic improvement with scale or robust learning-rate transfer. We trace these pathologies to scale-dependent observables in the aggregation dynamics, which motivates a refined set of desiderata that we term maximal scale stability. Guided by this principle, we derive a Maximally Scale-Stable Parameterization (MSSP) for both SGD and Adam in all three scaling regimes, and characterize the corresponding limiting dynamics - qualitatively distinct from the $μ$P limit - through a separate DMFT analysis. Experiments verify that MSSP robustly recovers learning rate transfer and monotonic improvement with scale across regimes. Combined with existing depth-scaling theory, these results provide a complete scaling prescription for MoE architectures as a function of width, depth, expert width, and number of experts.

📄 PDF Abstract BibTeX arXiv:2605.14200

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

StableMoE: Stable Routing Strategy for Mixture of Experts

2021-11-16 · ACL ARR November 2021 11 · Anonymous

The Mixture-of-Experts (MoE) technique can scale up the model size of Transformers with an affordable computational overhead. We point out that existing learning-to-route MoE methods suffer from the routing fluctuation i…

Language ModelingLanguage ModellingMachine TranslationMixture-of-Experts

StableMoE: Stable Routing Strategy for Mixture of Experts

2022-04-18 · ACL 2022 5 · Damai Dai, Li Dong, Shuming Ma, Bo Zheng 외

The Mixture-of-Experts (MoE) technique can scale up the model size of Transformers with an affordable computational overhead. We point out that existing learning-to-route MoE methods suffer from the routing fluctuation i…

Language ModelingLanguage ModellingMachine TranslationMixture-of-Experts

L2R: Low-Rank and Lipschitz-Controlled Routing for Mixture-of-Experts

2026-01-29 · Minghao Yang, Ren Togo, Guang Li, Takahiro Ogawa 외 arxiv

Mixture-of-Experts (MoE) models scale neural networks by conditionally activating a small subset of experts, where the router plays a central role in determining expert specialization and overall model performance. Howev…

AME-TS: Anchored Mixture-of-Experts for Time Series Forecasting

2026-05-24 · Rui Wang, Renhao Xue, Ray Razi, Huan Song 외 arxiv

Time series forecasting models are increasingly scaled through large Transformer backbones, yet most existing approaches process all series through a shared dense computation path despite substantial heterogeneity in tem…

Time Series Forecasting

Spectral Manifold Regularization for Stable and Modular Routing in Deep MoE Architectures

2026-01-07 · Ibrahim Delibasoglu arxiv

Mixture of Experts (MoE) architectures enable efficient scaling of neural networks but suffer from expert collapse, where routing converges to a few dominant experts. This reduces model capacity and causes catastrophic i…