AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Task Bias in Contrastive Vision-Language Models

Sachit Menon, Ishaan Preetam Chandratreya, Carl Vondrick

IJCV 2023 · Vision-language and multimodal › Multimodal LLMs · approach: VLM capability analysis · 16 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 manual task-directed text prefix to labels (text vs actions)CLIP zero-shot · Scene text vs Actionsproportion choosing intended task: +14.7% for text, 7.28% for actionsMixed
Add manual task-directed text prefix to labels (objects vs text)CLIP zero-shot · Objects vs Scene textproportion choosing intended task: 9.22% decrease for objects, 11% increase for textMixed
Add manual task-directed text prefix to labels (objects vs actions)CLIP zero-shot · Objects vs Actionsproportion choosing intended task: 37.5% decrease for objects, 29.6% increase for actionsMixed
Add learned visual token prompt (ViTP) to CLIPCLIP · Objects vs Text (target: Objects)task preference accuracy (%): 50.28 -> 99.46Improved
Add learned visual token prompt (ViTP) to CLIPCLIP · Objects vs Actions (target: Objects)task preference accuracy (%): 48.68 -> 100.00Improved
Add learned visual token prompt (ViTP) to CLIPCLIP · Actions vs Text (target: Actions)task preference accuracy (%): 43.70 -> 100.00Improved
Increase pixel prompt border size from 1 to 5 (ablation)Visual prompt (VP) · Objects vs Actions (target: Objects)task preference accuracy (%): 67.55 -> 81.13Improved
Increase pixel prompt border size from 1 to 5 (ablation)Visual prompt (VP) · OpenImages object recognitionaccuracy (%): 37.254 -> 20.799Hurt
Add object-directed visual prompt to CLIP (ViTP)CLIP · OpenImages object recognitionaccuracy (%): 20.467 -> 27.547Improved
Add object-directed pixel prompt with border size 1CLIP · OpenImages object recognitionaccuracy (%): 20.467 -> 37.254Improved
Use CLIP embedding instead of raw image to predict task bias direction (ablation)Task bias classifier · Objects v. Scene Texttest accuracy (%): 61.4 -> 71.7Improved
Add attention overlay to image input of the task bias classifier (ablation)Image classifier · Objects v. Scene Texttest accuracy (%): 61.4 -> 62.6Improved

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