Non-Euclidean SGD for Structured Optimization: Unified Analysis and Improved Rates
read the original abstract
Recently, several instances of non-Euclidean SGD, including SignSGD, Lion, and Muon, have attracted significant interest from the optimization community due to their practical success in training deep neural networks. Consequently, a number of works have attempted to explain this success by developing theoretical convergence analyses. Unfortunately, these results cannot properly justify the superior performance of these methods, as they could not beat the convergence rate of vanilla Euclidean SGD. We resolve this important open problem by developing a new unified convergence analysis under the structured smoothness and gradient noise assumption. In particular, our results indicate that non-Euclidean SGD (i) can exploit the sparsity or low-rank structure of the upper bounds on the Hessian and gradient noise, (ii) can provably benefit from popular algorithmic tools such as extrapolation or momentum variance reduction, and (iii) can match the state-of-the-art convergence rates of adaptive and more complex optimization algorithms such as AdaGrad and Shampoo.
This paper has not been read by Pith yet.
Forward citations
Cited by 4 Pith papers
-
Stochastic Non-Smooth Convex Optimization with Unbounded Gradients
Introduces generalized Lipschitz class and shows clipped AdamW outperforms SGD and AdaGrad for stochastic convex optimization under this and related assumptions.
-
Symmetry-Compatible Principle for Optimizer Design: Embeddings, LM Heads, SwiGLU MLPs, and MoE Routers
Proposes equivariant optimizers matched to the symmetry groups of embeddings, SwiGLU projections and MoE routers, with experiments showing consistent gains over AdamW on language model pre-training.
-
Convergence Analysis of Muon-type Methods with Inexact LMO in the Degenerate Case
Convergence rates are derived for Muon-type methods with inexact LMO in the degenerate case under novel assumptions and layer-wise (L^0, L^1)-smoothness for non-convex and star-convex objectives with weight decay.
-
Communication-Efficient Gluon in Federated Learning
Compressed Gluon variants using unbiased/contraction compressors and SARAH-style variance reduction achieve convergence guarantees and lower communication costs in federated learning under layer-wise smoothness.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.