AI Paper Atlas

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

CLIP Is Shortsighted: Paying Attention Beyond the First Sentence

Marc-Antoine Lavoie, Anas Mahmoud, Aldo Zaimi, Arsene Fansi Tchango, Steven L. Waslander

CVPR 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: CLIP training recipes and data curation · 2 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
remove summary sentence from the short training caption (ablation)Long-CLIP (3 epochs) · Urban1k T2Itop-1 recall: 82.7 -> 92.6Improved
remove summary sentence from the short training caption (ablation)Long-CLIP (3 epochs) · DOCCI T2Itop-1 recall: 75.2 -> 80.9Improved
add sentence subsampling (ablation)DeBias-CLIP w/o padding · Urban1k T2Itop-1 recall: 92.6 -> 92.5No clear effect
add token padding (ablation)DeBias-CLIP w/ subsampling · COCO T2Itop-1 recall: 41.9 -> 43.0Improved
add token padding (ablation)DeBias-CLIP w/ subsampling · DOCCI T2Itop-1 recall: 80.8 -> 79.7Hurt
DeBias-CLIP vs Long-CLIPViT-B/16 · DOCCI T2Itop-1 recall: 71.4 -> 80.0Improved
DeBias-CLIP vs SmartCLIPViT-B/16 · Urban1k T2Itop-1 recall: 87.4 -> 93.0Improved
DeBias-CLIP vs SmartCLIPViT-L/14 · COCO T2Itop-1 recall: 48.5 -> 48.1Hurt
train with DeBias-CLIP on original CLIP vs Long-CLIP under moved summary sentenceOpenAI CLIP · DOCCI T2IT2I drop (Keep vs Move): -9.7% -> -3.5%Improved
DeBias-CLIP on SigLIP/SigLIP2 under moved summarySigLIP, SigLIP2 · DOCCI T2IT2I drop (Keep vs Move): -6.1% and -6.5%Mixed
increase short-caption loss weight above 0.25 (ablation)DeBias-CLIP · COCO and DOCCIT2I recallHurt
prepend padding sentences to captions (probe) (ablation)CLIP, OpenCLIP, SigLIP, SigLIP2 · DOCCI T2Itop-1 retrievalHurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 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