aiqbench analysis
This research introduces a novel sparse fine-tuning framework leveraging neural network pruning to significantly reduce memory requirements while maintaining accuracy comparable to LoRA variants. By identifying structurally important neurons, the method offers a practical pathway for adapting large language models on devices with constrained computational budgets.
An Efficient Sparse Fine-Tuning with Low Quantization Error via Neural Network Pruning
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 accessible to users with limited computational budgets, it is crucial to develop fine-tuning methods that are memory and computationally efficient. Sparse Fine-tuning (SpFT) and Low-rank adaptation (LoRA) are two frameworks that have emerged for addressing this problem and have been adopted widely in practice. In this work, we develop a new SpFT framework, based on ideas from neural network pruning. At a high level, we first identify "important" neurons/nodes using feature importance metrics from network pruning (specifically, we use the structural pruning method), and then perform fine-tuning by restricting to weights involving these neurons. Experiments on common language tasks show our method improves SpFT's memory efficiency by 20-50\% while matching the accuracy of state-of-the-art methods like LoRA's variants. Code available at: https://github.com/CenjhihLi/sparsity_finetuning
Abstract:Fine-tuning is an important step in adapting foundation models such as large language models to downstream tasks. To make this step more accessible to users with limited computational budgets, it is crucial to develop fine-tuning methods that are memory and computationally efficient. Sparse Fine-tuning (SpFT) and Low-rank adaptation (LoRA) are two frameworks that have emerged for addressing this problem and have been adopted widely in practice. In this work, we develop a new SpFT framework, based on ideas from neural network pruning. At a high level, we first identify "important" neurons/nodes using feature importance metrics from network pruning (specifically, we use the structural pruning method), and then perform fine-tuning by restricting to weights involving these neurons. Experiments on common language tasks show our method improves SpFT's memory efficiency by 20-50\% while matching the accuracy of state-of-the-art methods like LoRA's variants. Code available at: this https URL
Published in Transactions on Machine Learning Research (TMLR), November 2025. Updated to the final published version
Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2502.11439 [cs.CL]
(or arXiv:2502.11439v3 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2502.11439 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 261 words
Related coverage
[AINews] High Taste Testers Ep. 1
Latent Space · Oct 17, 2026
[AINews] DeepSeek v4.1-Flash: 763B-P8B-D16B novel causal Encoder–Decoder architecture with vision marks the Return of the Whale
Latent Space · Sep 12, 2026
Quantization and Pruning Methods to Make Your LLM Leaner
KDnuggets · Aug 28, 2026
PIRAMID: Progress and Plans
LessWrong · Jul 27, 2026