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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Existing knowledge-grounded conversation systems generate responses typicallyin a retrieve-then-generate manner. They require a large knowledge base and astrong knowledge retrieval component, which is time- and resource-consuming. Inthis paper, we address the challenge by leveraging the inherent knowledgeencoded in the pre-trained language models (PLMs). We propose KnowledgeablePrefix Tuning (KnowPrefix-Tuning), a two-stage tuning framework, bypassing theretrieval process in a knowledge-grounded conversation system by injectingprior knowledge into the lightweight knowledge prefix. The knowledge prefix isa sequence of continuous knowledge-specific vectors that can be learned duringtraining. In addition, we propose a novel interactive re-parameterizationmechanism that allows the prefix to interact fully with the PLM during theoptimization of response generation. Experimental results demonstrate thatKnowPrefix-Tuning outperforms fine-tuning and other lightweight tuningapproaches, and performs comparably with strong retrieval-based baselines whilebeing $3\times$ faster during inference.

KnowPrefix-Tuning: A Two-Stage Prefix-Tuning Framework for Knowledge-Grounded Dialogue Generation