Score: 0

$μ$-Parametrization for Mixture of Experts

Published: August 13, 2025 | arXiv ID: 2508.09752v1

By: Jan Małaśnicki , Kamil Ciebiera , Mateusz Boruń and more

Potential Business Impact:

Makes big computer brains learn better and faster.

Recent years have seen a growing interest and adoption of LLMs, with $\mu$Transfer becoming a key technique for tuning hyperparameters in large-scale training. Meanwhile, Mixture-of-Experts (MoE) has emerged as a leading architecture in extremely large models. However, the intersection of these two advancements has remained unexplored. In this work, we derive a $\mu$-Parameterization ($\mu$P) for MoE, providing theoretical guarantees for feature learning across model widths in both the router and experts. We empirically validate our parameterization and further investigate how scaling the number of experts and granularity affects the optimal learning rate.

Country of Origin
🇵🇱 Poland

Page Count
9 pages

Category
Computer Science:
Machine Learning (CS)