AI Paper Atlas

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

Doubly Debiased Test-Time Prompt Tuning for Vision-Language Models

Fei Song, Yi Li, Rui Wang, Jiahuan Zhou, Changwen Zheng, Jiangmeng Li

AAAI 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Prompt learning for CLIP · 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 dynamic retrieval-augmented modulation (RAM) (ablation)D2TPT · Natural distribution shifts, OOD averagetop-1 accuracy: 64.57 -> 64.82Improved
add RAM (ablation)D2TPT · Cross-dataset generalization, averagetop-1 accuracy: 67.08 -> 68.27Improved
add confidence-based weighted ensemble (assumed second marked column) (ablation)D2TPT · Cross-dataset generalization, averagetop-1 accuracy: 68.27 -> 68.43Improved
add cross-modal consistency distillation (assumed third marked column) (ablation)D2TPT · Cross-dataset generalization, averagetop-1 accuracy: 68.43 -> 68.93Improved
add all D2TPT modules to entropy-minimization baseline (ablation)CLIP ViT-B/16 test-time prompt tuning · Natural distribution shifts, OOD averagetop-1 accuracy: 64.57 -> 65.25Improved
use D2TPT instead of TPSCLIP ViT-B/16 · Cross-dataset generalization, averagetop-1 accuracy: 66.95 -> 68.93Improved
use D2TPT instead of TDA on EuroSATCLIP ViT-B/16 · EuroSATtop-1 accuracy: 58.00 -> 54.92Hurt
use D2TPT instead of PromptAlign on OxfordPetsCLIP ViT-B/16 · OxfordPetstop-1 accuracy: 90.76 -> 87.55Hurt
use D2TPT instead of TPT regarding speedCLIP ViT-B/16 · Flowers102inference throughput FPS: 6.37 -> 5.71Hurt
use D2TPT instead of TPS on natural distribution shiftsCLIP ViT-B/16 · Natural distribution shifts, OOD averagetop-1 accuracy: 64.15 -> 65.25Improved

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