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

This paper explores the robustness of LLMs' preference to their internalmemory or the given prompt, which may contain contrasting information inreal-world applications due to noise or task settings. To this end, weestablish a quantitative benchmarking framework and conduct the role playingintervention to control LLMs' preference. In specific, we define two types ofrobustness, factual robustness targeting the ability to identify the correctfact from prompts or memory, and decision style to categorize LLMs' behavior inmaking consistent choices -- assuming there is no definitive "right" answer --intuitive, dependent, or rational based on cognitive theory. Our findings,derived from extensive experiments on seven open-source and closed-source LLMs,reveal that these models are highly susceptible to misleading prompts,especially for instructing commonsense knowledge. While detailed instructionscan mitigate the selection of misleading answers, they also increase theincidence of invalid responses. After Unraveling the preference, we intervenedifferent sized LLMs through specific style of role instruction, showing theirvarying upper bound of robustness and adaptivity.

Intuitive or Dependent? Investigating LLMs' Robustness to Conflicting Prompts