AI Paper Atlas

Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning

Enhancing CLIP with CLIP: Exploring Pseudolabeling for Limited-Label Prompt Tuning

Cristina Menghini, Andrew Delworth, Stephen Bach

NeurIPS 2023 · Label-efficient and robust learning › Zero-/few-shot and meta-learning · approach: Transductive and semantic-dictionary zero-shot · 3 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
GRIP textual prompts vs CLIP zero-shot (SSL, avg over datasets)CLIP ViT-B/32 · Flowers102/RESICS45/FGVCAircraft/MNIST/EuroSAT/DTDaccuracy: 20.7 pointsImproved
GRIP textual prompts vs CoOpCLIP ViT-B/32 · SSL and TRZSLaccuracy: 9.6 points in SSL, and 26.6 in TRZSLImproved
GRIP visual prompts vs VPTCLIP ViT-B/32 · SSL and TRZSLaccuracy: 12.9 points in SSL, and 20.8 in TRZSLImproved
GRIP multimodal prompts vs UPTCLIP ViT-B/32 · SSL and TRZSLaccuracy: 4.7 in semi-supervised learning, and 19.5 in transductive zero-shotImproved
use iterative IFPL instead of static FPL (ablation)FPL · Flowers102/RESICS45/DTDaccuracy: 5.6 points in SSL, 1.7 in UL, and 5.6 in TRZSLImproved
use GRIP (growing pseudolabels) instead of FPL (ablation)FPL · Flowers102/RESICS45/DTDaccuracy: 7.8 points in SSL, 3.1 in UL, and 9.7 in TRZSLImproved
use GRIP instead of FPL on MNIST/EuroSAT/FGVCAircraft (ablation)FPL · MNIST/EuroSAT/FGVCAircraftaccuracy: 16.7 pointsImproved
use IFPL instead of FPL on MNIST/EuroSAT/FGVCAircraft (ablation)FPL · MNIST/EuroSAT/FGVCAircraftaccuracy: 4.4 pointsHurt
GRIP visual prompts vs CLIP zero-shot on FGVCAircraft ULCLIP ViT-B/32 · FGVCAircraft ULaccuracy: 17.58 -> 15.22Hurt
GRIP vs CoOp on EuroSAT SSLCoOp · EuroSAT SSLaccuracy: 59.51 -> 58.66Hurt
use larger image encoder ViT-L/14 instead of ViT-B/32 (textual, SSL)GRIP · Flowers102/RESICS45/DTDavg improvement over CLIP: 17.46 -> 15.85Mixed
use prompt tuning (GRIP) instead of linear probing on rich classes (ablation)CLIP · UL, six datasetsrich-class accuracy change vs CLIP: 9.39 vs 0.33 average dropImproved

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