Skip to content

Opening book details…

About this document

MASPO: Joint Prompt Optimization For LLM-based Multi-Agent Systems by hackenbush10 is a document available to read on EtoBox.

The document introduces MASPO, a novel framework for joint prompt optimization in large language model-based multi-agent systems, addressing the challenges of aligning local agent objectives with global system goals. MASPO employs a multi-granularity joint evaluation mechanism and a trace-guided beam search to enhance prompt quality and facilitate effective agent collaboration. Empirical evaluations demonstrate that MASPO significantly outperforms existing prompt optimization methods, achieving an average a

Author
hackenbush10
Language
EN