English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Recent advances in vision language pretraining (VLP) have been largelyattributed to the large-scale data collected from the web. However, uncurateddataset contains weakly correlated image-text pairs, causing data inefficiency.To address the issue, knowledge distillation have been explored at the expenseof extra image and text momentum encoders to generate teaching signals formisaligned image-text pairs. In this paper, our goal is to resolve themisalignment problem with an efficient distillation framework. To this end, wepropose ECLIPSE: Expediting Contrastive Language-Image Pretraining withSelf-distilled Encoders. ECLIPSE features a distinctive distillationarchitecture wherein a shared text encoder is utilized between an online imageencoder and a momentum image encoder. This strategic design choice enables thedistillation to operate within a unified projected space of text embedding,resulting in better performance. Based on the unified text embedding space,ECLIPSE compensates for the additional computational cost of the momentum imageencoder by expediting the online image encoder. Through our extensiveexperiments, we validate that there is a sweet spot between expedition anddistillation where the partial view from the expedited online image encoderinteracts complementarily with the momentum teacher. As a result, ECLIPSEoutperforms its counterparts while achieving substantial acceleration ininference speed.

Expediting Contrastive Language-Image Pretraining via Self-distilled Encoders