SignMuon: Communication-Efficient Distributed Muon Optimization

TL;DR

SignMuon is a novel 1-bit, matrix-aware optimizer for distributed neural network training that significantly reduces communication overhead while maintaining high accuracy and efficiency.

Contribution

It introduces SignMuon, combining signSGD and Muon, enabling efficient, orthogonalized, matrix-aware optimization with minimal communication, outperforming existing methods.

Findings

Achieves 92.15% validation accuracy on CIFAR-10/ResNet-50.

Reduces bandwidth by 32x compared to float32.

Outperforms sign-based baselines on nanoGPT with lower perplexity.

Abstract

Distributed training of large neural networks is bottlenecked by full-precision gradient communication and by coordinatewise optimizers that ignore the matrix structure of weight tensors. We propose Sign-Muon, a 1-bit, matrix-aware optimizer that combines majority-vote sign aggregation from signSGD with the polar-step framework of Muon. Each worker forms a Muon-style direction by taking the polar factor of its momentum via a Newton--Schulz iteration, transmits only the entrywise signs, and aggregates by majority vote; an optional local polar step further enforces orthogonality at no extra communication cost. Under spectral-norm smoothness and bounded-variance stochastic gradients, the spectral-norm normalized sign step yields an $\mathcal{O}(1/\sqrt{T})$ nonconvex rate for an $\ell_1$-based stationarity measure. With unimodal symmetric noise, majority vote across $M$ workers cuts the stochastic term by $1/\sqrt{M}$, matching signSGD. In the $\alpha$-$\beta$ model, distributed Sign-Muon needs only one integer sum-allreduce per iteration; all orthogonalization is local, giving a $32\times$ bandwidth reduction over float32 ($4\times$ for int8). Across 330 CIFAR-10/ResNet-50 configurations Sign-Muon attains the best validation accuracy (92.15\%); its 4-GPU majority-vote variant reaches 92.02\% with 37\% less training time at matched effective batch. On nanoGPT, Sign-Muon achieves lower perplexity and better anytime performance than other sign-based baselines, with favorable weak-scaling up to 16 GPUs.