AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Saliency-Aware Model Merging

Jungin Park, Jiyoung Lee, Kwanghoon Sohn

ICML 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Low-rank compression and merging · 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
use SA-Merging instead of WUDI-Merging for data-free mergingCLIP ViT-B/32 8-task merging · 8-task vision suiteaverage accuracy: 85.2 -> 85.9Improved
use SA-Merging instead of WUDI-MergingCLIP ViT-L/14 8-task merging · 8-task vision suiteaverage accuracy: 92.6 -> 93.4Improved
use SA-Merging instead of WUDI-MergingRoBERTa-Base GLUE merging · GLUE 8 tasksnormalized score: 85.3 -> 87.1Improved
use SA-Merging instead of WUDI-MergingRoBERTa-Large GLUE merging · GLUE 8 tasksnormalized score: 88.8 -> 90.2Improved
use rank-wise saliency for LoRA merging instead of WUDIFlan-T5-base LoRA GLUE merging · GLUE LoRA 8 tasksaverage score: 81.2 -> 82.5Improved
use rank-wise saliency for LoRA merging instead of WUDIQwen-14B LoRA merging · MMLU/TruthfulQA/BBQ/CNNaverage score: 55.69 -> 56.33Improved
set pruning ratio p to 0.2 (vs other ratios) (ablation)SA-Merging · vision mergingaverage accuracyImproved
increase number of pruning iterations up to about 10 (ablation)SA-Merging · vision mergingper-task accuracyImproved
use SA-Merging pruning to reduce cross-model sign conflictsSA-Merging · sign conflicts across modelsparameter conflictsImproved
SA-Merging vs test-time AdaMerging++ on ViT-B/32CLIP ViT-B/32 · 8-task vision suiteaverage accuracy: 81.1 -> 85.9Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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