AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Less Is More: Rethinking Parameter-Efficient Fine-Tuning from a Subtractive Perspective

Tianqi Jiang, Liu Yang, Xi-Le Zhao, Zixuan Qin, Qinghua Hu

AAAI 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Adapter-based parameter-efficient tuning · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace a layer with a pseudo-inverse-initialized filter block instead of training that layer directly (ablation)ViT-B/16 on VTAB-1k · VTAB-1k single-layeraverage accuracy (%): 68.35 -> 73.65 (params 7.09M -> 0.63M)Improved
tune LayerNorm parameters (ablation)SFP · VTAB-1k Natural (single-layer)accuracy (%): 76.90 -> 81.24Improved
tune LayerNorm parameters (ablation)SFP · VTAB-1k Specialized (single-layer)accuracy (%): 86.44 -> 86.65Improved
tune LayerNorm parameters (ablation)SFP · VTAB-1k Structured (single-layer)accuracy (%): 59.59 -> 60.51Improved
use pseudo-inverse initialization instead of Kaiming init (ablation)SFP filter block · VTAB-1k Natural (single-layer)accuracy (%): 80.01 -> 81.24Improved
use pseudo-inverse initialization instead of Kaiming init (ablation)SFP filter block · VTAB-1k Specialized (single-layer)accuracy (%): 85.85 -> 86.65Improved
use pseudo-inverse initialization instead of Kaiming init (ablation)SFP filter block · VTAB-1k Structured (single-layer)accuracy (%): 59.49 -> 60.51Improved
substitute two layers instead of one (ablation)SFP · VTAB-1kaverage accuracy (%): 73.65 -> 73.03Hurt
use SFP instead of full fine-tuningViT-B/16 · VTAB-1kmean accuracy (%): 65.57 -> 74.04Improved
use SFP instead of DTL (best PEFT baseline)ViT-B/16 · VTAB-1kmean accuracy (%): 73.26 -> 74.04Improved
use SFP instead of GPS on datasets close to pretraining dataViT-B/16 · CIFAR-100 and SUN397 (VTAB-1k)accuracy (%): CIFAR-100 74.0 -> 70.5; SUN397 56.6 -> 51.9Hurt
replace a layer with filter block (inference efficiency)ViT-B/16 · VTAB-1kinference time per image (ms): 2.81 -> 2.33Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy