AI Paper Atlas

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

Knowledge-Aware Prompt Tuning for Generalizable Vision-Language Models

Baoshuo Kan, Teng Wang, Wenpeng Lu, Xiantong Zhen, Weili Guan, Feng Zheng

ICCV 2023 · Vision-language and multimodal › CLIP and zero-shot · approach: Prompt learning for CLIP · 51 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 knowledge-aware prompts to CoOp baseline (ablation)CoOp · Base-to-new avg (ViT-B/32)new-class accuracy (%): 64.02 -> 67.24Improved
add knowledge-aware prompts to CoOp baseline (ablation)CoOp · Base-to-new avg (ViT-B/32)base-class accuracy (%): 79.41 -> 77.70Hurt
add adaptation head to baseline (ablation)CoOp · Base-to-new avg (ViT-B/32)harmonic mean accuracy (%): 70.88 -> 72.11Improved
combine knowledge prompts with adaptation head (ablation)KAPT · Base-to-new avg (ViT-B/32)harmonic mean accuracy (%): 70.88 -> 74.26Improved
use both discrete and continuous knowledge prompts instead of one (ablation)KAPT · Unseen classes avgnew-class accuracy (%): 67.53 -> 70.52 (continuous only)Improved
use full KAPT instead of continuous-only prompts (ablation)KAPT · Unseen classes, per datasetnew-class accuracy (%)Mixed
replace CoCoOp with KAPT on cross-dataset transfer and domain generalizationKAPT · Distribution shiftaccuracy (%): 61.25 -> 61.50 (ImageNet source)No clear effect
replace CoCoOp with KAPTKAPT · Base-to-new, ResNet-50 backboneharmonic mean accuracy (%): 69.39 -> 69.94Improved

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