AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Localized Symbolic Knowledge Distillation for Visual Commonsense Models

Jae Sung Park, Jack Hessel, Khyathi Chandu, Paul Pu Liang, Ximing Lu, Peter West, Youngjae Yu, Qiuyuan Huang, Jianfeng Gao, Ali Farhadi, Yejin Choi

NeurIPS 2023 · Vision-language and multimodal › Human and embodied modeling · approach: Scene graph generation · 1 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
Distill localized commonsense knowledge into BLIP-2 ViT-GBLIP-2 ViT-G · VCR Q→Aaccuracy: 56.1 -> 59.0Improved
Distill localized commonsense knowledge into BLIP-2 ViT-GBLIP-2 ViT-G · VCR QA→Raccuracy: 49.8 -> 56.4Improved
Distill localized commonsense knowledge into BLIP-2 ViT-GBLIP-2 ViT-G · VCR Q→ARaccuracy: 28.0 -> 33.4Improved
Distill localized commonsense knowledge into BLIP-2 ViT-GBLIP-2 ViT-G · SherlockComparison: 19.5 -> 29.7Improved
Distill localized commonsense knowledge into BLIP-2 ViT-GBLIP-2 ViT-G · VisualCOMETAcc@50: 39.0 -> 40.3Improved
Distill localized commonsense knowledge into BLIP-2 ViT-GBLIP-2 ViT-G · AOKVQAmultiple-choice accuracy: 68.0 -> 68.9Improved
Distill localized commonsense knowledge into BLIP-2 ViT-GBLIP-2 ViT-G · SNLI-VEclassification accuracy: 33.4 -> 40.3Improved
Distill localized commonsense knowledge into BLIP-2 ViT-GBLIP-2 ViT-G · Visual7W Telling QAaccuracy: 77.1 -> 79.5Improved
Scale the filtered LSKD corpus from 150K to 1M examples (ablation)BLIP-2 ViT-G + LSKD · SherlockComparison: 28.6 -> 29.7Improved
Scale the filtered LSKD corpus from 150K to 1M examples (ablation)BLIP-2 ViT-G + LSKD · SNLI-VEclassification accuracy: 38.0 -> 40.3Improved
Add QA descriptors to the full verbalization used for knowledge generation (ablation)LSKD corpus generation · 500-image descriptor ablationaverage critic score: 49.0 to 58.4Improved
Filter generated statements with the supervised critic rather than keeping unfiltered statements (ablation)LSKD corpus generation · localized visual reasoning tasksdownstream accuracyImproved

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