AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

One Last Attention for Your Vision-Language Model

Liang Chen, Ghazi Shazan Ahmad, Tianjun Yao, Lingqiao Liu, Zhiqiang Shen

ICCV 2025 · Vision-language and multimodal › CLIP and zero-shot · approach: Test-time adaptation of VLMs · 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 mask regularization L_reg to RAda (ablation)RAda (EFT) · Base-to-new, 11 datasets averagenew-class accuracy: 71.58 -> 74.14Improved
use attention layer instead of MLP for the mask function (ablation)RAda (EFT) · Base-to-new, 11 datasets averageharmonic mean: 75.28 -> 77.94Improved
add image and text features as extra queries to the rational matrix query (ablation)RAda attention layer · Base-to-new, 11 datasets averageharmonic mean: 77.72 -> 77.94Improved
adapt the fused rational matrix instead of attaching attention to text and visual encoders (ablation)RAda (EFT) · Base-to-new, 11 datasets averagenew-class accuracy: 72.46 -> 74.14Improved
use L2 instead of L1 mask regularization (ablation)RAda (EFT) · Base-to-new, 11 datasets averageharmonic mean: 77.21 -> 77.94Improved
use L2 instead of L-infinity mask regularization (ablation)RAda (EFT) · Base-to-new, 11 datasets averageharmonic mean: 76.84 -> 77.94Improved
stack 2 attention layers instead of 1 (ablation)RAda (EFT) · Base-to-new, 11 datasets averageharmonic mean: 77.94 -> 77.93No clear effect
stack 3 attention layers instead of 1 (ablation)RAda (EFT) · Base-to-new, 11 datasets averagenew-class accuracy: 74.14 -> 71.81Hurt
fine-tune all parameters after rational adapter warmup (RAda-FT vs FT)FT on CLIP · ImageNet OOD average (FFT)accuracy: 53.8 -> 61.3Improved
replace CLIP-Adapter with RAda-FTFFT with sequential training · ImageNet OOD average (FFT)accuracy: 60.3 -> 61.3Improved
add RAda to CLIP in test-time trainingCLIP · ImageNet OOD average (TTT)accuracy: 57.20 -> 63.50Improved
add RAda to CLIP with 16-shot base-class trainingCLIP · Base-to-new, 11 datasets averagebase accuracy: 69.34 -> 82.16Improved

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