AI Paper Atlas

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

Three Towers: Flexible Contrastive Learning with Pretrained Image Models

Jannik Kossen, Mark Collier, Basil Mustafa, Xiao Wang, Xiaohua Zhai, Lucas Beyer, Andreas Peter Steiner, Jesse Berent, Rodolphe Jenatton, Effrosyni Kokiopoulou

NeurIPS 2023 · Vision-language and multimodal › CLIP and zero-shot · approach: Self-distillation for vision-language pretraining · 4 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 third tower with pretrained embeddings (3T) vs from-scratch baselineCLIP/ALIGN · Flickr* img2txt retrieval, L, IN-21ktop-1 recall: 75.6 -> 80.0Improved
3T instead of LiT, IN-21kL-scale towers · Flickr* txt2imgtop-1 recall: 49.3 -> 60.9Improved
3T instead of LiT, JFTL-scale towers · COCO img2txttop-1 recall: 52.7 -> 54.4Improved
3T instead of baseline, IN-21kL-scale towers · average classificationaccuracy: 68.4 -> 71.4Improved
LiT instead of baseline, IN-21kL-scale towers · average classificationaccuracy: 68.4 -> 68.3No clear effect
LiT instead of baseline on Stanford Cars zero-shot/few-shotIN-21k · Cars 10-shotaccuracy: 81.5 -> 41.9Hurt
LiT instead of 3T on BirdsIN-21k · Birds 10-shotaccuracy: 65.0 -> 83.4Improved
mismatched pretrained scale (B/32 pretrained, L/16 tower)LiT · zero-shot full averageaccuracy: 66.4 -> 61.7Hurt
Places365 pretrained modelLiT · zero-shot full averageaccuracy: 47.5 -> 29.4Hurt
remove image-text contrastive loss (ablation)3T · average of tasksperformance difference: -26.63 ± 10.61Hurt
remove image-tower transfer loss (ablation)3T · average of tasksperformance difference: -1.19 ± 0.75Hurt
remove text-tower transfer loss (ablation)3T · average of tasksperformance difference: -2.77 ± 0.91Hurt

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