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

The generative AI revolution in recent years has been spurred by an expansionin compute power and data quantity, which together enable extensivepre-training of powerful text-to-image (T2I) models. With their greatercapabilities to generate realistic and creative content, these T2I models likeDALL-E, MidJourney, Imagen or Stable Diffusion are reaching ever wideraudiences. Any unsafe behaviors inherited from pretraining on uncuratedinternet-scraped datasets thus have the potential to cause wide-reaching harm,for example, through generated images which are violent, sexually explicit, orcontain biased and derogatory stereotypes. Despite this risk of harm, we lacksystematic and structured evaluation datasets to scrutinize model behavior,especially adversarial attacks that bypass existing safety filters. A typicalbottleneck in safety evaluation is achieving a wide coverage of different typesof challenging examples in the evaluation set, i.e., identifying 'unknownunknowns' or long-tail problems. To address this need, we introduce theAdversarial Nibbler challenge. The goal of this challenge is to crowdsource adiverse set of failure modes and reward challenge participants for successfullyfinding safety vulnerabilities in current state-of-the-art T2I models.Ultimately, we aim to provide greater awareness of these issues and assistdevelopers in improving the future safety and reliability of generative AImodels. Adversarial Nibbler is a data-centric challenge, part of the DataPerfchallenge suite, organized and supported by Kaggle and MLCommons.

Adversarial Nibbler: A Data-Centric Challenge for Improving the Safety of Text-to-Image Models