AI Paper Atlas

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

Decompose, Compare, and Decide: Multimodal LLMs are Implicit Few-Shot Learners

Yunhan Wang, Eshika Khandelwal, Edson Araujo, Walid Bousselham, Nina Shvetsova, Hilde Kuehne

ECCV 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Adapting VLMs for zero-shot classification · 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
decompose into pairwise support-query binary comparisonsQwen3-VL-8B · 12 datasets avg, 5-way 1-shot, semantic labelsaccuracy (%): 80.9 -> 87.3Improved
decompose into pairwise comparisons with anonymized labelsQwen3-VL-8B · 12 datasets avg, anonymous labelsaccuracy (%): 27.8 -> 78.9Improved
decompose into pairwise comparisons with anonymized labelsInternVL3-8B · 12 datasets avg, anonymous labelsaccuracy (%): 23.4 -> 80.2Improved
add 1-shot support images to in-context prompt on standard datasetsInternVL3-8B · Standard datasets avg, semantic labelsaccuracy (%): 90.7 -> 72.2Hurt
add 1-shot support images in-context on novel datasetsQwen3-VL-8B · Novel datasets avg, semantic labelsaccuracy (%): 57.0 -> 76.9Improved
add domain information D_info to decomposed prompt (ablation)Qwen3-VL-8B · 12 datasets avg, semantic labelsaccuracy (%): 87.3 -> 90.6Improved
add domain information D_info to decomposed prompt (ablation)Qwen2.5-VL-7B · 12 datasets avg, semantic labelsaccuracy (%): 85.7 -> 88.7Improved
add domain information D_info to decomposed prompt, anonymous labels (ablation)InternVL3-8B · 12 datasets avg, anonymous labelsaccuracy (%): 80.2 -> 85.7Improved
apply binary decomposition without support image (0-shot dec.) (ablation)Qwen3-VL-8B · Ablation total avg, semantic labelsaccuracy (%): 78.2 -> 76.8No clear effect
replace PMI-calibrated in-context with pairwise decomposition (ablation)Qwen3-VL-8B · Ablation total avg, semantic labelsaccuracy (%): 86.2 -> 89.2Improved
add PMI-style calibration to in-context 1-shot (ablation)InternVL3-8B · Ablation total avg, semantic labelsaccuracy (%): 62.6 -> 75.0Improved
score with generated confidence instead of Yes logit (ablation)Qwen2.5-VL-7B · Scoring rule ablation total, anonymousaccuracy (%): 85.1 -> 71.8Hurt

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