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

Question answering models can use rich knowledge sources -- up to one hundredretrieved passages and parametric knowledge in the large-scale language model(LM). Prior work assumes information in such knowledge sources is consistentwith each other, paying little attention to how models blend information storedin their LM parameters with that from retrieved evidence documents. In thispaper, we simulate knowledge conflicts (i.e., where parametric knowledgesuggests one answer and different passages suggest different answers) andexamine model behaviors. We find retrieval performance heavily impacts whichsources models rely on, and current models mostly rely on non-parametricknowledge in their best-performing settings. We discover a troubling trend thatcontradictions among knowledge sources affect model confidence only marginally.To address this issue, we present a new calibration study, where models arediscouraged from presenting any single answer when presented with multipleconflicting answer candidates in retrieved evidences.

Rich Knowledge Sources Bring Complex Knowledge Conflicts: Recalibrating Models to Reflect Conflicting Evidence