CEPO Improves Prompt Optimization in Multi-Agent Systems
Original title: CEPO: A General Framework for Improving Prompt Optimization in Multi-Agent Systems
CEPO separates coordination and execution prompts to consistently improve four existing optimizers across 11 benchmarks.
Multi-agent systems require effective coordination and specialized task execution. Existing prompt optimization approaches couple these functions within agent-specific prompts. CEPO introduces a general framework that separates system-level coordination from agent-level execution using a shared coordinator prompt and agent-specific executor prompts. This allows shared rules to be refined consistently while keeping execution instructions adjustable. Experiments show CEPO improves performance of four prompt optimizers across 11 benchmarks in four domains.
- Shared coordinator prompt for consistent coordination rules
- Agent-specific executor prompts for task execution
- Improves four existing prompt optimizers
- 11 benchmarks across four domains
- Controlled studies show improved optimization budget efficiency
From the original article
1 min read at huggingface.co
Multi-agent systems (MAS) require both effective coordination and specialized task execution. Existing prompt optimization approaches typically couple these functions within agent-specific prompts. We introduce Coordinator-Executor Prompt Optimization (CEPO), a general framework for improving prompt optimization in MAS.
By Jongha Kim, Seohyun Lee, Jonathan Minho Chung · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.