AI Paper Atlas

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

Scaling Up Vision-Language Pre-Training for Image Captioning

Xiaowei Hu, Zhe Gan, Jianfeng Wang, Zhengyuan Yang, Zicheng Liu, Yumao Lu, Lijuan Wang

CVPR 2022 · Vision-language and multimodal › CLIP and zero-shot · approach: Self-distillation for vision-language pretraining · 16 citations (OpenAlex)

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
Pre-train the captioner on ALT200M rather than COCO aloneLEMON base · nocaps validationCIDEr, in-domain and out-of-domain: +16.3 in-domain; +45.3 out-of-domainImproved
Scale pre-training data from CC12M to ALT200MLEMON base · nocaps validation overallCIDEr: 100.4 -> 106.8Improved
Scale model size from base to huge with ALT200M pre-trainingLEMON · nocaps validation overallCIDEr: 106.8 -> 117.3Improved
Increase pre-training data while keeping model size smallLEMON scaling setup · COCO captioningCIDErMixed
Scale model size when pre-training data is only 3MLEMON scaling setup · COCO captioningCIDErNo clear effect
Use larger models when pre-training data exceeds 40MLEMON scaling setup · COCO captioningCIDErImproved
Use s2s-MLM instead of LM (ablation)Encoder captioner · Conceptual Captions valCIDEr: 96.1 -> 104.4Improved
Use s2s-MLM instead of LM (ablation)Encoder-decoder captioner · Conceptual Captions valCIDEr: 94.9 -> 99.9Improved
Use s2s-MLM instead of LM (ablation)Encoder-decoder captioner · COCO Karpathy testCIDEr: 120.9 -> 120.4Hurt
Use encoder-decoder instead of encoder with LM (ablation)Base-size captioner · Conceptual Captions valCIDEr: 96.1 -> 94.9Hurt
Use encoder instead of encoder-decoder with s2s-MLM (ablation)Base-size captioner · Conceptual Captions valCIDEr: 99.9 -> 104.4Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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