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 recent surge in Large Language Model (LLM) related applications has ledto a concurrent escalation in expectations for LLMs to accommodate a myriad ofpersonas and encompass a broad spectrum of perspectives. An important firststep towards addressing this demand is to align language models with specificpersonas, be it groups of users or individuals. Towards this goal, we firstpresent a new conceptualization of a persona. Moving beyond the traditionalreliance on demographics like age, gender, or political party affiliation, weintroduce a data-driven persona definition methodology built oncollaborative-filtering. In this methodology, users are embedded into acontinuous vector space based on their opinions and clustered into cohorts thatmanifest coherent views across specific inquiries. This methodology allows fora more nuanced understanding of different latent social groups present in theoverall population (as opposed to simply using demographic groups) and enhancesthe applicability of model steerability. Finally, we present an efficientmethod to steer LLMs towards a particular persona. We learn a soft-promptingmodel to map the continuous representation of users into sequences of virtualtokens which, when prepended to the LLM input, enables the LLM to produceresponses aligned with a given user. Our results show that our steerabilityalgorithm is superior in performance compared to a collection of baselines.

On the steerability of large language models toward data-driven personas