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

Large language models (LLMs) often encounter knowledge conflicts, scenarioswhere discrepancy arises between the internal parametric knowledge of LLMs andnon-parametric information provided in the prompt context. In this work we askwhat are the desiderata for LLMs when a knowledge conflict arises and whetherexisting LLMs fulfill them. We posit that LLMs should 1) identify knowledgeconflicts, 2) pinpoint conflicting information segments, and 3) providedistinct answers or viewpoints in conflicting scenarios. To this end, weintroduce KNOWLEDGE CONFLICT, an evaluation framework for simulating contextualknowledge conflicts and quantitatively evaluating to what extent LLMs achievethese goals. KNOWLEDGE CONFLICT includes diverse and complex situations ofknowledge conflict, knowledge from diverse entities and domains, two syntheticconflict creation methods, and settings with progressively increasingdifficulty to reflect realistic knowledge conflicts. Extensive experiments withthe KNOWLEDGE CONFLICT framework reveal that while LLMs perform well inidentifying the existence of knowledge conflicts, they struggle to determinethe specific conflicting knowledge and produce a response with distinct answersamidst conflicting information. To address these challenges, we propose newinstruction-based approaches that augment LLMs to better achieve the threegoals. Further analysis shows that abilities to tackle knowledge conflicts aregreatly impacted by factors such as knowledge domain and prompt text, whilegenerating robust responses to knowledge conflict scenarios remains an openresearch question.

Resolving Knowledge Conflicts in Large Language Models