Sparse autoencoders (SAEs) are commonly used to interpret large language models, but their reliability after pruning is unclear. This study shows that pruning’s effect on an SAE is governed by perturbation energy, a covariance-weighted norm, and that magnitude pruning distorts the representation space by ignoring activation geometry. Activation-aware pruning methods such as Wanda and SparseGPT better preserve SAE behavior, and the authors find that middle layers are especially vulnerable, leading them to propose a layer‑wise sparsity allocation that reduces perplexity for a given sparsity level.
By Suchit Gupte, Xueru Zhang, Mohammad Mahdi Khalili
arXiv:2508. 13836v2 Announce Type: replace-cross Abstract: Pruning is a core technique for compressing neural networks to improve computational efficiency.
By Miko{\l}aj Janusz, Tomasz Wojnar, Yawei Li, Luca Benini, Kamil Adamczewski
arXiv:2509. 14230v2 Announce Type: replace Abstract: While structured pruning presents a highly effective pathway for accelerating Large Language Model (LLM) inference, existing methods frequently suffer from significant performance degradation and demand computationally retraining to recover capabilities.
By Mengting Ai, Tianxin Wei, Sirui Chen, Jingrui He
arXiv:2608. 06630v1 Announce Type: new Abstract: Pruning reduces the inference cost of large language models, but existing criteria primarily preserve large activations or reconstruct layer outputs.
By Linghao Kong, Inimai Subramanian, Micah Adler, Dan Alistarh, Dan Gutfreund, Nir Shavit
arXiv:2311.04902v3 Announce Type: replace-cross
Abstract: Large Language Models (LLMs) with billions of parameters are prime targets for network pruning, which removes some model weights without hurt...
By Rocktim Jyoti Das, Mukul Ranjan, Mingjie Sun, Liqun Ma, Zhiqiang Shen
arXiv:2604.13287v2 Announce Type: replace
Abstract: Weight pruning is a common technique for compressing large neural networks. We focus on the challenging post-training one-shot setting, where a pre...
By Gabriel Afriat, Xiang Meng, Shibal Ibrahim, Hussein Hazimeh, Rahul Mazumder
arXiv:2609.24401v1 Announce Type: new
Abstract: Structured pruning is a model compression technique that is used to reduce the computational cost of deploying deep neural networks on resource-constra...
By Mindula Illeperuma, Rafael Pina, Charuka Herath, Sharmarke A. Gabayre, Varuna De Silva
arXiv:2606. 14346v1 Announce Type: cross Abstract: Unstructured pruning produces sparse weight tensors, but the standard implementation keeps tensor shapes unchanged so the deployed model is no smaller than before pruning.
By Roman Denkin, Ida Akerholm, Prashant Singh, Ida-Maria Sintorn
arXiv:2606. 12278v1 Announce Type: cross Abstract: Neural network pruning reduces model size by removing less important parameters while aiming to preserve predictive performance.
By Romana Qureshi, Hafida Benhidour, Said Kerrache, Nahlah Aljeraisy
arXiv:2502.11439v3 Announce Type: replace-cross
Abstract: Fine-tuning is an important step in adapting foundation models such as large language models to downstream tasks. To make this step more acce...
By Cen-Jhih Li, Aditya Bhaskara
The paper introduces a Hybrid Quadratic Unconstrained Binary Optimization (QUBO) framework for structured neural network pruning that integrates task‑aware sensitivity metrics (first‑order Taylor and Weight‑Fisher) into the objective’s linear term and optionally uses activation similarity for quadratic interactions. It controls pruning cardinality via a binary search over a capacity incentive rather than an explicit penalty and further refines the pruning mask with a two‑stage QUBO–Tensor‑Train strategy that employs gradient‑free black‑box optimization. Experiments on SIDD image denoising with a Half‑UNet model demonstrate that this Hybrid QUBO outperforms Taylor and L1‑based QUBO baselines in PSNR and SSIM, while also revealing computational and deployment challenges of mask‑based pruning.
By Osama Orabi, Artur Zagitov, Hadi Salloum, Viktor A. Lobachev, Yaroslav Kholodov
The paper introduces Dynamic Activity-Dependent Pruning (DADP), a biologically inspired pruning technique that uses accumulated pre‑synaptic activations and post‑synaptic error gradients to assess connection importance. Unlike traditional methods that rely on fixed layer budgets or magnitude thresholds, DADP employs a single global threshold, allowing it to dynamically allocate sparsity across a network’s depth and prune neurons and channels during training. Experiments on MLP, VGG‑16, ResNet‑18, BiLSTM‑CRF, and MiniBERT show that DADP matches or surpasses existing pruning methods, achieving 73.67% accuracy at 99% sparsity on ResNet‑18 while preserving feature diversity as measured by Shannon entropy and effective rank.
By Bhushan Deshpande