AI Paper Atlas

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

ViLTA: Enhancing Vision-Language Pre-training through Textual Augmentation

Weihan Wang, Zhen Yang, Bin Xu, Juanzi Li, Yankui Sun

ICCV 2023 · Vision-language and multimodal › CLIP and zero-shot · approach: Self-distillation for vision-language pretraining · 10 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 synthetic hard negatives for ITM (ablation)ViLTA (MLM+ITM) · VQAv2accuracy: 77.34 -> 78.06Improved
add cross-distillation soft labels for MLM (ablation)ViLTA (MLM+ITM) · VQAv2accuracy: 77.34 -> 77.85Improved
combine hard negatives and cross-distillation (ablation)ViLTA · VQAv2accuracy: 78.06 -> 78.47Improved
add ITM task to MLM-only pre-training (ablation)ViLTA · COCO CaptionsBLEU@4: 38.2 -> 38.8Improved
use dynamically generated negatives instead of in-batch random (ablation)ViLTA ITM · VQAv2 test-devaccuracy: 74.08 -> 74.93Improved
use WordNet antonym negatives instead of random negatives (ablation)ViLTA ITM · VQAv2 test-devaccuracy: 74.08 -> 74.24No clear effect
add WordNet negatives to dynamically generated negatives (ablation)ViLTA ITM · VQAv2 test-devaccuracy: 74.93 -> 74.71Hurt
regenerate negatives dynamically each step instead of once before training (ablation)ViLTA ITM · VQAv2 test-devaccuracy: 74.52 -> 74.93Improved
add language modeling pre-training phase for captioning (ablation)ViLTA · COCO CaptionsBLEU@4: 39.6 -> 40.0 (w/o CA); 40.4 -> 41.0 (w/ CA)Improved
keep cross-attention module on top of the image encoder (ablation)ViLTA captioning · COCO CaptionsCIDEr: 131.5 -> 135.1 (w/ LM)Improved
ViLTA pre-trained with only MLM and ITM (no ITC) on COCO retrievalViLTA-BASE · COCO text retrievalTR@1: 73.3 vs METER 76.2Hurt
add both techniques to pre-training (ablation)ViLTA · pre-training costGPU-hours per epoch: 34.68 -> 38.43Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 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