Don't Be Greedy, Just Relax! Pruning LLMs via Frank-Wolfe

• URL: http://arxiv.org/abs/2510.13713v1
• Date: Wed, 15 Oct 2025 16:13:44 GMT
• Title: Don't Be Greedy, Just Relax! Pruning LLMs via Frank-Wolfe
• Authors: Christophe Roux, Max Zimmer, Alexandre d'Aspremont, Sebastian Pokutta,
• Abstract summary: State-of-the-art Large Language Model (LLM) pruning methods operate layer-wise, minimizing the per-layer pruning error on a small dataset to avoid full retraining.<n>Existing methods hence rely on greedy convexs that ignore the weight interactions in the pruning objective.<n>Our method drastically reduces the per-layer pruning error, outperforms strong baselines on state-of-the-art GPT architectures, and remains memory-efficient.
• Score: 61.68406997155879
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Pruning is a common technique to reduce the compute and storage requirements of Neural Networks. While conventional approaches typically retrain the model to recover pruning-induced performance degradation, state-of-the-art Large Language Model (LLM) pruning methods operate layer-wise, minimizing the per-layer pruning error on a small calibration dataset to avoid full retraining, which is considered computationally prohibitive for LLMs. However, finding the optimal pruning mask is a hard combinatorial problem and solving it to optimality is intractable. Existing methods hence rely on greedy heuristics that ignore the weight interactions in the pruning objective. In this work, we instead consider the convex relaxation of these combinatorial constraints and solve the resulting problem using the Frank-Wolfe (FW) algorithm. Our method drastically reduces the per-layer pruning error, outperforms strong baselines on state-of-the-art GPT architectures, and remains memory-efficient. We provide theoretical justification by showing that, combined with the convergence guarantees of the FW algorithm, we obtain an approximate solution to the original combinatorial problem upon rounding the relaxed solution to integrality.

Related papers

• SparseSwaps: Tractable LLM Pruning Mask Refinement at Scale [22.25809500403244]
  We propose a tractable and simple 1-swap algorithm that warms starts from any pruning mask and runs efficiently at LLM scale.<n>We demonstrate that our approach reduces per-layer pruning error by up to 60% over Wanda (Sun et al., 2023) and consistently improves perplexity and zero-shot accuracy across state-of-the-art GPT architectures.
  arXiv  Detail & Related papers  (2025-12-11T18:47:48Z)
• Learning based convex approximation for constrained parametric optimization [11.379408842026981]
  We propose an input neural network (ICNN)-based self-supervised learning framework to solve constrained optimization problems.<n>We provide rigorous convergence analysis, showing that the framework converges to a Karush-Kuhn-Tucker (KKT) approximation point of the original problem.<n>Our approach achieves a superior balance among accuracy, feasibility, and computational efficiency.
  arXiv  Detail & Related papers  (2025-05-07T00:33:14Z)
• Training Deep Learning Models with Norm-Constrained LMOs [56.00317694850397]
  We propose a new family of algorithms that uses the linear minimization oracle (LMO) to adapt to the geometry of the problem.<n>We demonstrate significant speedups on nanoGPT training using our algorithm, Scion, without any reliance on Adam.
  arXiv  Detail & Related papers  (2025-02-11T13:10:34Z)
• Zeroth-Order Adaptive Neuron Alignment Based Pruning without Re-Training [3.195234044113248]
  We propose textscNeuroAL, a emphtop-up algorithm for network pruning.<n>It modifies the block-wise and row-wise sparsity exploiting information from both the dense model and its sparse version.<n>It consistently outperforms the latest state-of-the-art methods in terms of performance-runtime trade-off.
  arXiv  Detail & Related papers  (2024-11-11T15:30:16Z)
• A Learned Proximal Alternating Minimization Algorithm and Its Induced Network for a Class of Two-block Nonconvex and Nonsmooth Optimization [4.975853671529418]
  This work proposes a general learned alternating minimization algorithm, LPAM, for solving learnable two-block nonsmooth problems. The proposed LPAM-net is parameter-efficient and has favourable performance in comparison with some state-of-the-art methods.
  arXiv  Detail & Related papers  (2024-11-10T02:02:32Z)
• A Sample Efficient Alternating Minimization-based Algorithm For Robust Phase Retrieval [56.67706781191521]
  In this work, we present a robust phase retrieval problem where the task is to recover an unknown signal. Our proposed oracle avoids the need for computationally spectral descent, using a simple gradient step and outliers.
  arXiv  Detail & Related papers  (2024-09-07T06:37:23Z)
• Alternating Minimization Schemes for Computing Rate-Distortion-Perception Functions with $f$-Divergence Perception Constraints [9.788112471288057]
  We study the computation of the rate-distortion-perception function (RDPF) for discrete memoryless sources.<n>We characterize optimal parametric solutions for the convex programming problem.<n>We show, by deriving necessary and sufficient conditions, that both schemes guarantee a globally optimal solution.
  arXiv  Detail & Related papers  (2024-08-27T12:50:12Z)
• Bypass Back-propagation: Optimization-based Structural Pruning for Large Language Models via Policy Gradient [57.9629676017527]
  We propose an optimization-based structural pruning that learns the pruning masks in a probabilistic space directly by optimizing the loss of the pruned model.<n>We achieve this by learning an underlying Bernoulli distribution to sample binary pruning masks.<n>Experiments conducted on LLaMA, LLaMA-2, LLaMA-3, Vicuna, and Mistral models demonstrate the promising performance of our method in efficiency and effectiveness.
  arXiv  Detail & Related papers  (2024-06-15T09:31:03Z)
• Convex Relaxations of ReLU Neural Networks Approximate Global Optima in Polynomial Time [45.72323731094864]
  In this paper, we study the optimality gap between two-layer ReLULU networks regularized with weight decay and their convex relaxations. Our study sheds new light on understanding why local methods work well.
  arXiv  Detail & Related papers  (2024-02-06T01:29:35Z)
• Robust Stochastically-Descending Unrolled Networks [85.6993263983062]
  Deep unrolling is an emerging learning-to-optimize method that unrolls a truncated iterative algorithm in the layers of a trainable neural network.<n>We show that convergence guarantees and generalizability of the unrolled networks are still open theoretical problems.<n>We numerically assess unrolled architectures trained under the proposed constraints in two different applications.
  arXiv  Detail & Related papers  (2023-12-25T18:51:23Z)
• Stable Nonconvex-Nonconcave Training via Linear Interpolation [51.668052890249726]
  This paper presents a theoretical analysis of linearahead as a principled method for stabilizing (large-scale) neural network training. We argue that instabilities in the optimization process are often caused by the nonmonotonicity of the loss landscape and show how linear can help by leveraging the theory of nonexpansive operators.
  arXiv  Detail & Related papers  (2023-10-20T12:45:12Z)
• Fast Convex Optimization for Two-Layer ReLU Networks: Equivalent Model Classes and Cone Decompositions [47.276004075767176]
  We develop software for convex optimization of two-layer neural networks with ReLU activation functions.<n>We show that convex gated ReLU models obtain data-dependent algorithms for the ReLU training problem.
  arXiv  Detail & Related papers  (2022-02-02T23:50:53Z)
• Parallel Stochastic Mirror Descent for MDPs [72.75921150912556]
  We consider the problem of learning the optimal policy for infinite-horizon Markov decision processes (MDPs) Some variant of Mirror Descent is proposed for convex programming problems with Lipschitz-continuous functionals. We analyze this algorithm in a general case and obtain an estimate of the convergence rate that does not accumulate errors during the operation of the method.
  arXiv  Detail & Related papers  (2021-02-27T19:28:39Z)

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.