AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

SCT: A Simple Baseline for Parameter-Efficient Fine-Tuning via Salient Channels

Hengyuan Zhao, Pichao Wang, Yuyang Zhao, Hao Luo, Fan Wang, Mike Zheng Shou

IJCV 2023 · Architectures and efficiency › Pruning and efficient tuning · approach: Efficient ViT adaptation · 22 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 class-aware importance score (ablation)SCT · VTAB-1K (19 tasks avg)average top-1 accuracy: 72.8 -> 73.6Improved
select salient (highest L2) channels instead of inconspicuous ones (ablation)SCT · VTAB-1K (19 tasks avg)average top-1 accuracy: 71.1 -> 73.6Improved
select salient channels instead of random ones (ablation)SCT · VTAB-1K (19 tasks avg)average top-1 accuracy: 72.0 -> 73.6Improved
insert SCTM after Attn instead of after MLP (ablation)SCT · VTAB-1K (19 tasks avg)average top-1 accuracy: 72.9 -> 73.6Improved
insert SCTM after both Attn and MLP (ablation)SCT · VTAB-1K Cifar100 / Resisc45 / DMLabtop-1 accuracy: 75.3 -> 75.7Mixed
increase K from 96 to 192 (ablation)SCT · VTAB-1K (19 tasks avg)average top-1 accuracy: 73.6 -> 74.0Improved
increase K from 32 to 96 (ablation)SCT · VTAB-1K (19 tasks avg)average top-1 accuracy: 69.1 -> 73.6Improved
use K=32 instead of full fine-tuningSCT · VTAB-1K (19 tasks avg)average top-1 accuracy: 65.6 -> 69.1Improved
raise channels in first four layers from 96 to 192 (ablation)SCT · VTAB-1K Caltech101 / Resisc45 / KITTItop-1 accuracy: 91.6 -> 91.7Mixed
reduce K from 96 to 32 in last eight layers (ablation)SCT · VTAB-1K KITTItop-1 accuracy: 80.2 -> 76.2Hurt
insert SCTM into more of the last layers (ablation)SCT · VTAB-1K Resisc45 / DMLabtop-1 accuracyImproved
use SCT instead of Adapter-8ViT-B/16 · VTAB-1K (19 tasks avg)average top-1 accuracy: 71.3 -> 73.6Improved

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