
CS
Chang-Wei Shi, Xu Wang, Wu-Jun Li
· 1 min read
ResearcharXiv cs.LG
MeqMuon: Matrix-Equilibrating Muon for LLM Pretraining
arXiv:2609.35701v1 Announce Type: new
Abstract: The success of large language models (LLMs) has been accompanied by continued growth in model size and pretraining costs. Muon offers high accuracy and training efficiency in LLM pretraining. Recent work introduces row-wise normalization into Muon to balance update magnitudes and improve pretraining performance. However, row-wise normalization alone cannot accommodate different imbalance patterns in update matrices. In this paper, we propose an improved Muon optimizer, called \underline{m}atrix-\underline{eq}uilibrating Muon~(MeqMuon), for LLM pretraining. MeqMuon balances both row and column magnitudes through normalization that can be automatically tailored to different imbalance patterns without manual intervention. Moreover, MeqMuon eliminates the need to store AdamW's second-moment estimates, reducing optimizer-state memory usage. Empirical results demonstrate that MeqMuon achieves better convergence performance than AdamW, Muon, and other baselines in LLM pretraining.
Original source
This story was published by arXiv cs.LG and written by Chang-Wei Shi, Xu Wang, Wu-Jun Li. SyncAI.news shows a preview; the complete article is on the publisher's site.
Read the full story on arxiv.org


