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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Large language models (LLMs) have recently shown great potential forin-context learning, where LLMs learn a new task simply by conditioning on afew input-label pairs (prompts). Despite their potential, our understanding ofthe factors influencing end-task performance and the robustness of in-contextlearning remains limited. This paper aims to bridge this knowledge gap byinvestigating the reliance of LLMs on shortcuts or spurious correlations withinprompts. Through comprehensive experiments on classification and extractiontasks, we reveal that LLMs are "lazy learners" that tend to exploit shortcutsin prompts for downstream tasks. Additionally, we uncover a surprising findingthat larger models are more likely to utilize shortcuts in prompts duringinference. Our findings provide a new perspective on evaluating robustness inin-context learning and pose new challenges for detecting and mitigating theuse of shortcuts in prompts.

Large Language Models Can be Lazy Learners: Analyze Shortcuts in In-Context Learning