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

While auxiliary information has become a key to enhance Large Language Models(LLMs), relatively little is known about how well LLMs merge these contexts,specifically generated and retrieved. To study this, we formulate a taskspecifically designed to identify whether the answers, derived from theintegration of generated and retrieved contexts, are attributed to eithergenerated or retrieved contexts. To support this task, we develop a methodologyto construct datasets with conflicting contexts, where each question is pairedwith both generated and retrieved contexts, yet only one of them contains thecorrect answer. Our experiments reveal a significant bias in LLMs towardsgenerated contexts, as evidenced across state-of-the-art open (Llama2-7b/13b)and closed (GPT 3.5/4) systems. We further identify two key factorscontributing to this bias: i) Contexts generated by LLMs typically show greatersimilarity to the questions, increasing their likelihood of selection; ii) Thesegmentation process used in retrieved contexts disrupts their completeness,thereby hindering their full utilization in LLMs. Our analysis enhances theunderstanding of how LLMs merge diverse contexts, offering valuable insightsfor advancing current augmentation methods for LLMs.

Blinded by Generated Contexts: How Language Models Merge Generated and Retrieved Contexts for Open-Domain QA?