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

Recently, Chain-of-Thought (CoT) prompting has delivered success on complexreasoning tasks, which aims at designing a simple prompt like ``Let's thinkstep by step'' or multiple in-context exemplars with well-designed rationalesto elicit Large Language Models (LLMs) to generate intermediate reasoningsteps. However, the generated rationales often come with mistakes, makingunfactual and unfaithful reasoning chains. To mitigate this brittleness, wepropose a novel Chain-of-Knowledge (CoK) prompting, where we aim at elicitingLLMs to generate explicit pieces of knowledge evidence in the form of structuretriple. This is inspired by our human behaviors, i.e., we can draw a mind mapor knowledge map as the reasoning evidence in the brain before answering acomplex question. Benefiting from CoK, we additionally introduce aF^2-Verification method to estimate the reliability of the reasoning chains interms of factuality and faithfulness. For the unreliable response, the wrongevidence can be indicated to prompt the LLM to rethink. Extensive experimentsdemonstrate that our method can further improve the performance of commonsense,factual, symbolic, and arithmetic reasoning tasks.

Boosting Language Models Reasoning with Chain-of-Knowledge Prompting