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

Long chain-of-thought (CoT) is an essential ingredient in effective usage ofmodern large language models, but our understanding of the reasoning strategiesunderlying these capabilities remains limited. While some prior works haveattempted to categorize CoTs using predefined strategy types, such approachesare constrained by human intuition and fail to capture the full diversity ofmodel behaviors. In this work, we introduce the CoT Encyclopedia, a bottom-upframework for analyzing and steering model reasoning. Our method automaticallyextracts diverse reasoning criteria from model-generated CoTs, embeds them intoa semantic space, clusters them into representative categories, and derivescontrastive rubrics to interpret reasoning behavior. Human evaluations showthat this framework produces more interpretable and comprehensive analyses thanexisting methods. Moreover, we demonstrate that this understanding enablesperformance gains: we can predict which strategy a model is likely to use andguide it toward more effective alternatives. Finally, we provide practicalinsights, such as that training data format (e.g., free-form vs.multiple-choice) has a far greater impact on reasoning behavior than datadomain, underscoring the importance of format-aware model design.

The CoT Encyclopedia: Analyzing, Predicting, and Controlling how a Reasoning Model will Think