Accelerating Zeroth-Order Spectral Optimization with Partial Orthogonalization from Power Iteration

• URL: http://arxiv.org/abs/2605.09034v2
• Date: Fri, 15 May 2026 13:21:07 GMT
• Title: Accelerating Zeroth-Order Spectral Optimization with Partial Orthogonalization from Power Iteration
• Abstract summary: We focus on the hidden layer training problem in which spectrals like Muon outperform AdamW.<n>To do so, we replace the iconic Newton-Schulz procedure in Muon with the faster, more concentrated power-iteration method.<n>Our method can achieve from 1.5x to 4x the convergence speed of ZO-Muon.
• Score: 6.574641780732972
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Zeroth-order (ZO) optimization has become increasingly popular and important in fine-tuning large language models (LLMs), especially on edge devices due to its ability to adjust the model to local data without the need for memory-intensive back-propagation. Recent works try to reduce ZO variance through low-dimensional subspace search, but subspace restriction alone leaves key optimization geometry under-exploited, motivating additional acceleration. In this work, we focus on the hidden layer training problem in which spectral optimizers like Muon outperform AdamW due to its ability to exploit weak spectral directions by orthogonalization. However, we have discovered that unlike in the first-order setting, full orthogonalization works poorly in the ZO setting since the gradient estimates are highly noisy and unreliable. To address this issue, we propose applying partial spectral orthogonalization to accelerate ZO optimization. To do so, we replace the iconic Newton-Schulz procedure in Muon with the faster, more concentrated power-iteration method so that it only amplifies dominant spectral directions. Furthermore, to improve the efficiency and generalization of the algorithm, we adopted a streaming variant of power-iteration that requires low variance in gradients, which was achieved through constraining our search inside a subspace obtained through the projection of momentum, echoing recent advances. Experiments on LLM fine-tuning show that our method can achieve from 1.5x to 4x the convergence speed of ZO-Muon, the current SOTA algorithm, across SuperGlue datasets in the OPT-13B model. Across different models, we also reach competitive final accuracies with less time in most cases compared with strong ZO baselines such as MeZO, LOZO and ZO-Muon. Code is available at https://github.com/MOFA-LAB/ZO-MOPI.git.

Related papers

• The Active Ingredient in Muon's Grokking [2.921159958223653]
  We show the Muon-faster reaches the grokking threshold on modular arithmetic faster than AdamW.<n>To better understand Moun's behavior, we run multi-seed and multi-learning-rate sweeps to decompose and stress-test the effect.
  arXiv  Detail & Related papers  (2026-07-06T22:35:57Z)
• Closed-Form Spectral Regularization for Multi-Task Model Merging [96.82449201305234]
  Model merging combines several independently fine-tuned experts into a single multi-task model without any training data.<n>State-of-the-art merging methods formulate merging as a layer-wise interference problem.<n>We propose SWUDI, a closed-form method that combines a soft exponential filter, which matches the gradient-flow trajectory of iterative descent.
  arXiv  Detail & Related papers  (2026-06-05T14:00:47Z)
• Universally Empowering Zeroth-Order Optimization via Adaptive Layer-wise Sampling [43.822941944402544]
  Zeroth-Order optimization presents a promising memory-efficient paradigm for fine-tuning Large Language Models.<n>However, its practical adoption is severely constrained by slow wall-clock convergence and high estimation variance.<n>We propose AdaLeZO, an Adaptive Layer-wise ZO optimization framework.
  arXiv  Detail & Related papers  (2026-04-20T13:37:31Z)
• $\ abla$-Reasoner: LLM Reasoning via Test-Time Gradient Descent in Latent Space [71.23672814629448]
  $nabla$-Reasoner is an iterative generation framework that integrates differentiable optimization over token logits into the decoding loop.<n>$nabla$-Reasoner achieves over 20% accuracy improvement on a challenging mathematical reasoning benchmark.
  arXiv  Detail & Related papers  (2026-03-05T08:42:54Z)
• Powering Up Zeroth-Order Training via Subspace Gradient Orthogonalization [40.95701844244596]
  We show that ZO optimization can be substantially improved by unifying two complementary principles.<n>We instantiate in a new method, ZO-Muon, admitting a natural interpretation as a low-rank Muon in the ZO setting.
  arXiv  Detail & Related papers  (2026-02-19T08:08:33Z)
• MuonBP: Faster Muon via Block-Periodic Orthogonalization [24.232069944820513]
  We show how to adjust the learning rate from the baseline to MuonBP and give guarantees for this algorithm.<n>When training an 8B model with eight-way tensor tensor and ZeRO statewiseing, MuonBP achieves 8% Muon with no degradation in performance.
  arXiv  Detail & Related papers  (2025-10-19T19:56:05Z)
• KerZOO: Kernel Function Informed Zeroth-Order Optimization for Accurate and Accelerated LLM Fine-Tuning [15.81250204481401]
  We introduce a kernel-function-based ZO framework aimed at mitigating gradient estimation bias.<n>KerZOO achieves comparable or superior performance to existing ZO baselines.<n>We show that the kernel function is an effective avenue for reducing estimation bias in ZO methods.
  arXiv  Detail & Related papers  (2025-05-24T21:56:03Z)
• Second-order Optimization of Gaussian Splats with Importance Sampling [51.95046424364725]
  3D Gaussian Splatting (3DGS) is widely used for novel view rendering due to its high quality and fast inference time.<n>We propose a novel second-order optimization strategy based on Levenberg-Marquardt (LM) and Conjugate Gradient (CG)<n>Our method achieves a $3times$ speedup over standard LM and outperforms Adam by $6times$ when the Gaussian count is low.
  arXiv  Detail & Related papers  (2025-04-17T12:52:08Z)
• Harmony in Divergence: Towards Fast, Accurate, and Memory-efficient Zeroth-order LLM Fine-tuning [44.907586955452295]
  Large language models (LLMs) excel across various tasks, but standard first-order (FO) fine-tuning demands considerable memory.<n>Recently, zeroth-order (ZO) optimization stood out as a promising memory-efficient training paradigm.<n>We introduce a novel layer-wise divergence analysis that uncovers the distinct update pattern of FO and ZO optimization.<n>Our results demonstrate that DiZO significantly reduces the needed iterations for convergence without sacrificing throughput.
  arXiv  Detail & Related papers  (2025-02-05T16:03:17Z)
• Elucidating Subspace Perturbation in Zeroth-Order Optimization: Theory and Practice at Scale [33.38543010618118]
  Zeroth-order (ZO) optimization has emerged as a promising alternative to gradient-based backpropagation methods.<n>We show that high dimensionality is the primary bottleneck and introduce the notion of textitsubspace alignment to explain how the subspace perturbations reduce gradient noise and accelerate convergence.<n>We propose an efficient ZO method using block coordinate descent (MeZO-BCD), which perturbs and updates only a subset of parameters at each step.
  arXiv  Detail & Related papers  (2025-01-31T12:46:04Z)
• TeZO: Empowering the Low-Rankness on the Temporal Dimension in the Zeroth-Order Optimization for Fine-tuning LLMs [58.19080159470868]
  We propose a novel low-rank ZO estimator, TeZO, which captures the low-rankness across both the model and temporal dimension.<n>Specifically, we represent ZO perturbations along the temporal dimension as a 3D tensor and employ Canonical Polyadic Decomposition (CPD) to extract each low-rank 2D matrix.
  arXiv  Detail & Related papers  (2025-01-31T11:34:03Z)
• Zeroth-Order Fine-Tuning of LLMs in Random Subspaces [63.10833446782114]
  As language models grow in size, memory demands for backpropagation increase.<n>Zeroth-order (ZO) optimization methods offer a memory-efficient alternative.<n>In this paper, we propose Subspace Zero-order optimization to address the challenges posed by posed by high dimensionality perturbations.
  arXiv  Detail & Related papers  (2024-10-11T17:01:43Z)
• ELRA: Exponential learning rate adaption gradient descent optimization method [83.88591755871734]
  We present a novel, fast (exponential rate), ab initio (hyper-free) gradient based adaption. The main idea of the method is to adapt the $alpha by situational awareness. It can be applied to problems of any dimensions n and scales only linearly.
  arXiv  Detail & Related papers  (2023-09-12T14:36:13Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.