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 (large LMs) are susceptible to producing text thatcontains hallucinated content. An important instance of this problem isself-contradiction, where the LM generates two contradictory sentences withinthe same context. In this work, we present a comprehensive investigation intoself-contradiction for various instruction-tuned LMs, covering evaluation,detection, and mitigation. Our analysis reveals the prevalence ofself-contradictions when LMs generate text for open-domain topics, e.g., in17.7% of all sentences produced by ChatGPT. Self-contradiction also complementsretrieval-based methods, as a large portion of them (e.g., 35.8% for ChatGPT)cannot be verified using Wikipedia. We then propose a novel prompting-basedframework designed to effectively detect and mitigate self-contradictions. Ourdetector achieves high accuracy, e.g., around 80% F1 score when promptingChatGPT. The mitigation algorithm iteratively refines the generated text toremove contradictory information while preserving text fluency andinformativeness. Importantly, our entire framework is applicable to black-boxLMs and does not require external grounded knowledge. Our approach ispractically effective and has been released as a push-button tool to benefitthe public, available at https://chatprotect.ai/.

Self-contradictory Hallucinations of Large Language Models: Evaluation, Detection and Mitigation