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

Token-based masked generative models are gaining popularity for their fastinference time with parallel decoding. While recent token-based approachesachieve competitive performance to diffusion-based models, their generationperformance is still suboptimal as they sample multiple tokens simultaneouslywithout considering the dependence among them. We empirically investigate thisproblem and propose a learnable sampling model, Text-Conditioned TokenSelection (TCTS), to select optimal tokens via localized supervision with textinformation. TCTS improves not only the image quality but also the semanticalignment of the generated images with the given texts. To further improve theimage quality, we introduce a cohesive sampling strategy, Frequency AdaptiveSampling (FAS), to each group of tokens divided according to the self-attentionmaps. We validate the efficacy of TCTS combined with FAS with variousgenerative tasks, demonstrating that it significantly outperforms the baselinesin image-text alignment and image quality. Our text-conditioned samplingframework further reduces the original inference time by more than 50% withoutmodifying the original generative model.

Text-Conditioned Sampling Framework for Text-to-Image Generation with Masked Generative Models