AI Paper Atlas

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

Coarse-to-Fine Vision-Language Pre-training with Fusion in the Backbone

Zi-Yi Dou, Aishwarya Kamath, Zhe Gan, Pengchuan Zhang, Jianfeng Wang, Linjie Li, Zicheng Liu, Ce Liu, Yann LeCun, Nanyun Peng, Jianfeng Gao, Lijuan Wang

NeurIPS 2022 · Vision-language and multimodal › Human and embodied modeling · approach: Region-level vision-language alignment · 67 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 gated co-attention rather than using ungated co-attention with 6 fusion layers (ablation)FIBER · VQAv2 test-devVQA score: 68.13 -> 71.97Improved
Use 6 gated co-attention layers rather than 6 merged-attention layers (ablation)FIBER · VQAv2 test-devVQA score: 70.67 -> 71.97Improved
Add MLM to coarse-grained pre-training (ablation)FIBER · VQAv2 test-devVQA score: 67.45 -> 74.49Improved
Add MLM to coarse-grained pre-training (ablation)FIBER · Flickr30k retrievalIR@1: 75.20 -> 73.58No clear effect
Use coarse-grained image-text pre-training before fine-grained training (ablation)FIBER · RefCOCO+ Valaccuracy: 85.04 -> 85.74Improved
Use coarse-grained image-text pre-training before fine-grained training (ablation)FIBER · COCO detection zero-shotAP: 48.9 -> 49.3Improved
Continue pre-training from the coarse checkpoint with fine-grained image-text-box data before VQA fine-tuning (ablation)FIBER · VQAv2 test-devVQA score: 78.55 -> 74.3Hurt
Use seq2seq rather than ladder architecture for captioning (ablation)FIBER · COCO captioningCIDEr: 127.5 -> 128.4Improved
Replace RoBERTa with BERT (ablation)FIBER without VLP · VQAv2VQA score: 71.97 -> 71.86Hurt
Replace Swin with CLIP-ViT (ablation)FIBER without VLP · VQAv2VQA score: 71.97 -> 71.37Hurt
Ensemble the ITC dual encoder and ITM fusion encoderFIBER · Flickr30k retrievalIR@1: 81.44 -> 90.96Improved
Add GOLD captioning trainingFIBER-B · COCO captioningCIDEr: 128.4 -> 133.6Improved

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