AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

From Holistic to Localized: Local Enhanced Adapters for Efficient Visual Instruction Fine-Tuning

Pengkun Jiao, Bin Zhu, Jingjing Chen, Chong-Wah Ngo, Yu-Gang Jiang

ICCV 2025 · Architectures and efficiency › Pruning and efficient tuning · approach: Parameter-efficient transfer analysis · 3 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 Dual-LoRA (ablation)vanilla LoRA with VCE off · UniFood Ingredient RecognitionIoU: 23.2 -> 24.2Improved
add Dual-LoRA (ablation)vanilla LoRA · UniFood Recipe GenerationSacreBLEU: 12.4 -> 14.8Improved
add VCE to vanilla LoRA (ablation)vanilla LoRA · UniFood Ingredient RecognitionIoU: 23.2 -> 23.3No clear effect
add VCE to Dual-LoRA (ablation)Dual-LoRA · UniFood Ingredient RecognitionIoU: 24.2 -> 24.5Improved
add VCE to Dual-LoRA (ablation)Dual-LoRA · UniFood Recipe GenerationSacreBLEU: 14.8 -> 14.7No clear effect
add skill-space layer normalization (ablation)Dual-LoRA (frozen projector) · UniFood Ingredient RecognitionIoU: 21.4 -> 22.6Improved
add task-space non-linear activation (ablation)Dual-LoRA (frozen projector) · UniFood Ingredient RecognitionIoU: 21.4 -> 21.9Improved
add both norm and non-linearity (ablation)Dual-LoRA (frozen projector) · UniFood Ingredient RecognitionIoU: 21.4 -> 23.0Improved
make vision projector trainable (ablation)Dual-LoRA · UniFood Ingredient RecognitionIoU: 23.0 -> 24.2Improved
use Dual-LoRA instead of RoDELoRA-MoE comparison · ScienceQAaccuracy: 78.39 -> 79.17Improved
use Dual-LoRA instead of vanilla LoRALoRA · ScienceQAaccuracy: 70.01 -> 79.17Improved
use Dual-LoRA instead of LoRA-MoE (top-2)LoRA-MoE · UniFood Nutrition Estimation (avg)pMAE: 50.1 -> 49.9Improved

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