AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

NuWa: Deriving Lightweight Class-Specific Vision Transformers for Edge Devices

Ziteng Wei, Qiang He, Bing Li, Feifei Chen, Hai Jin, Yun Yang

CVPR 2026 · Architectures and efficiency › Attention and transformers · approach: Vision transformer pruning · 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
add self-knowledge purification (SKP) (ablation)NuWa · ImageNet S4-S6 (25 classes), alpha=0.6avg accuracy: 72.69 -> 91.44Improved
add MHA pruning (balanced with MLP via rho) (ablation)NuWa · ImageNet S4-S6 (25 classes), alpha=0.6avg accuracy: 76.05 -> 91.44Improved
prune MLP in OFP rather than only MHA (ablation)NuWa · ImageNet S4-S6 (25 classes), alpha=0.6avg accuracy: 5.49 -> 91.44Improved
select retained MLP neurons by activation instead of randomly (ablation)NuWa · ImageNet S4-S6 (25 classes), alpha=0.6avg accuracy: 86.45 -> 91.44Improved
use closed-form solution for MLP W2 instead of direct removal (ablation)NuWa · ImageNet S4-S6 (25 classes), alpha=0.6avg accuracy: 76.69 -> 91.44Improved
add light fine-tuning (10 epochs) to NuWaNuWa · ImageNet S4-S6 (25 classes), alpha=0.6avg accuracy: 91.44 -> 95.95Improved
NuWa vs training-dependent baselines without retrainingDeiT-Base pruning · ImageNet S7-S9 (50 classes), alpha=0.6avg accuracy: 89.50 -> 85.83 (vs X-Pruner)Hurt
smaller SKP batch size (ablation)NuWa SKP · ImageNet S4-S6anchor pruning rateImproved
replace training-dependent pruning with NuWa for derivation costclass-specific ViT derivation · ImageNet S1-S9, N=50, M=10AWS cost: $23074 -> $59.12 vs X-PrunerImproved
SKP on DeiT-Base anchor model vs unprunedDeiT-Base · ImageNet S1-S3 (10 classes)avg accuracy: 78.60 -> 97.67Improved

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