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

Language model agents (LMA) recently emerged as a promising paradigm onmuti-step decision making tasks, often outperforming humans and otherreinforcement learning agents. Despite the promise, their performance onreal-world applications that often involve combinations of tasks is stillunderexplored. In this work, we introduce a new benchmark, called CompWoB -- 50new compositional web automation tasks reflecting more realistic assumptions.We show that while existing prompted LMAs (gpt-3.5-turbo or gpt-4) achieve94.0% average success rate on base tasks, their performance degrades to 24.9%success rate on compositional tasks. On the other hand, transferred LMAs(finetuned only on base tasks) show less generalization gap, dropping from85.4% to 54.8%. By balancing data distribution across tasks, we train a newmodel, HTML-T5++, that surpasses human-level performance (95.2%) on MiniWoB,and achieves the best zero-shot performance on CompWoB (61.5%). While thesehighlight the promise of small-scale finetuned and transferred models forcompositional generalization, their performance further degrades underdifferent instruction compositions changing combinational order. In contrast tothe recent remarkable success of LMA, our benchmark and detailed analysisemphasize the necessity of building LMAs that are robust and generalizable totask compositionality for real-world deployment.

Language Model Agents Suffer from Compositional Generalization in Web Automation