AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Understanding and Enforcing Weight Disentanglement in Task Arithmetic

Shangge Liu, Yuehan Yin, Lei Wang, Qi Fan, Yinghuan Shi, Wenbin Li, Yang Gao, Dacheng Tao

CVPR 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Parameter-efficient transfer analysis · 1 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
add OrthoReg orthogonal regularization on weight updatesNon-linear Finetuning (ViT-L-14) · Task addition (8 tasks)absolute accuracy: 84.07 -> 88.23Improved
add OrthoRegTangent Task Arithmetic (ViT-L-14) · Task addition (8 tasks)absolute accuracy: 86.19 -> 87.52Improved
add OrthoRegAttention-Only Fine-tuning (ViT-B-16) · Task addition (8 tasks)absolute accuracy: 80.71 -> 83.37Improved
add OrthoRegLoRA-ATT (ViT-B-16) · Task addition (8 tasks)absolute accuracy: 75.51 -> 79.67Improved
add OrthoRegNon-linear Finetuning (ViT-B-16) · Task addition (8 tasks)normalized accuracy: 75.39 -> 93.62Improved
add OrthoRegAttention-Only Fine-tuning (ViT-L-14) · Task addition (8 tasks)normalized accuracy: 93.59 -> 100.05Improved
add OrthoRegTangent Task Arithmetic (ViT-B-16) · Task addition (8 tasks)normalized accuracy: 87.48 -> 88.02No clear effect
add OrthoRegAttention-Only Fine-tuning (ViT-L-14) · Task negation (8 tasks)target accuracy: 24.85 -> 14.67Improved
add OrthoRegTangent Task Arithmetic (ViT-L-14) · Task negation (8 tasks)target accuracy: 8.36 -> 8.36No clear effect
add OrthoRegTangent Task Arithmetic (ViT-B-32) · Task negation (8 tasks)target accuracy: 11.47 -> 11.39No clear effect
add OrthoRegAttention-Only Fine-tuning (ViT-L-14) · Task negation (8 tasks)ImageNet control accuracy: 76.42 -> 75.40Mixed
add OrthoRegNon-linear Finetuning (ViT-B-16) · Cosine similarity of task vectorsoff-diagonal cosine similarityImproved

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