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

Diffusion-based language models (LMs) have been shown to be competentgenerative models that are easy to control at inference and are a promisingalternative to autoregressive LMs. While autoregressive LMs have benefitedimmensely from scaling and instruction-based learning, existing studies ondiffusion LMs have been conducted on a relatively smaller scale. Starting witha recently proposed diffusion model SSD-LM, in this work we explore methods toscale it from 0.4B to 13B parameters, proposing several techniques to improveits training and inference efficiency. We call the new model SSD-2. We furthershow that this model can be easily finetuned to follow instructions. Finally,leveraging diffusion models' capability at inference-time control, we show thatSSD-2 facilitates novel ensembles with 100x smaller models that can becustomized and deployed by individual users. We find that compared toautoregressive models, the collaboration between diffusion models is moreeffective, leading to higher-quality and more relevant model responses due totheir ability to incorporate bi-directional contexts.

SSD-2: Scaling and Inference-time Fusion of Diffusion Language Models