AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Sliding-Window Merging for Compacting Patch-Redundant Layers in LLMs

Xuan Ding, Rui Sun, Yunjian Zhang, Xiu Yan, Yueqi Zhou, Kaihao Huang, Suzhong Fu, Angelica I Aviles-Rivero, Chuanlong Xie, Yao Zhu

AAAI 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Layer-wise pruning of LLMs · 13 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
merge layers by parameter differences (SWM) instead of deleting them or averaging (ablation)LLaMA2-7B depth pruning · LLaMA2-7B, threshold 0.75layers retained at equal threshold: 23 layers vs 26 (Delete) vs 29 (Average)Improved
SWM vs best baseline at 20% pruningSWM · LLaMA2-7B zero-shot avgaverage accuracy: 59.268 -> 61.944 (+2.676)Improved
SWM vs best baseline at 35% pruningSWM · Vicuna-7B zero-shot avgaverage accuracy: 54.596 -> 56.250Improved
SWM vs LLM-Pruner at 35% pruningSWM · LLaMA2-7B zero-shot avgaverage accuracy: 52.366 -> 51.869Hurt
add LoRA retraining after pruning (ablation)SWM · LLaMA2-7B 20%average accuracy: 57.360 -> 61.944Improved
combine SWM depth stage (18-91%) with LLM-Pruner width stage at 35% total sparsity (ablation)SWM + LLM-Pruner · LLaMA2-7Baverage accuracyImproved
vary calibration dataset and sample size (ablation)SWM · LLaMA2-7B HellaSwagaccuracy: 62.179 to 63.730No clear effect
use depth pruning instead of width pruningSWM/SLEB/Shortened-LLM · LLaMA2-7B 20%latency: 5.655 -> 2.339 (LLM-Pruner vs SWM)Improved
lower similarity threshold (ablation)SWM · LLaMA2-7Bcompression vs performanceMixed

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