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 Plus annual

Presents the access of premium content as premium feature

Premium Content

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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Recent Vision-Language Pre-trained (VLP) models based on dual encoder haveattracted extensive attention from academia and industry due to their superiorperformance on various cross-modal tasks and high computational efficiency.They attempt to learn cross-modal representation using contrastive learning onimage-text pairs, however, the built inter-modal correlations only rely on asingle view for each modality. Actually, an image or a text contains variouspotential views, just as humans could capture a real-world scene via diversedescriptions or photos. In this paper, we propose ERNIE-ViL 2.0, a Multi-ViewContrastive learning framework to build intra-modal and inter-modalcorrelations between diverse views simultaneously, aiming at learning a morerobust cross-modal representation. Specifically, we construct multiple viewswithin each modality to learn the intra-modal correlation for enhancing thesingle-modal representation. Besides the inherent visual/textual views, weconstruct sequences of object tags as a special textual view to narrow thecross-modal semantic gap on noisy image-text pairs. Pre-trained with 29Mpublicly available datasets, ERNIE-ViL 2.0 achieves competitive results onEnglish cross-modal retrieval. Additionally, to generalize our method toChinese cross-modal tasks, we train ERNIE-ViL 2.0 through scaling up thepre-training datasets to 1.5B Chinese image-text pairs, resulting insignificant improvements compared to previous SOTA results on Chinesecross-modal retrieval. We release our pre-trained models inhttps://github.com/PaddlePaddle/ERNIE.

ERNIE-ViL 2.0: Multi-view Contrastive Learning for Image-Text Pre-training