MAPGD: Multi-Agent Prompt Gradient Descent for Collaborative Prompt Optimization

2025-09-14

AI Summary

Existing prompt engineering methods suffer from reliance on a single optimization trajectory, limited perspective, severe gradient conflicts, and high computational overhead. To address these issues, this paper proposes MAPGD—a Multi-Agent Prompt Gradient Descent framework. MAPGD employs a task-decomposition agent for specialized role assignment, designs a semantic gradient fusion mechanism to harmonize heterogeneous gradient directions, and integrates a bandit-based candidate selection strategy with gradient-descent-inspired heuristic search—ensuring theoretical convergence while enhancing optimization efficiency and robustness. The framework enables interpretable prompt evolution and demonstrates significant improvements over single-agent and random baselines across classification, generation, and reasoning tasks: average accuracy increases by 7.2%, iteration count decreases by 38%, and computational efficiency is preserved alongside performance gains.

Technology Category

• Search and Optimization: Learning to Search
• Natural Language Processing: Prompt Engineering / Prompting
• Multiagent Systems: Adversarial Agents

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Prompt engineering is crucial for leveraging large language models (LLMs), but existing methods often rely on a single optimization trajectory, limiting adaptability and efficiency while suffering from narrow perspectives, gradient conflicts, and high computational cost. We propose MAPGD (Multi-Agent Prompt Gradient Descent), a framework integrating multi-agent collaboration with gradient-based optimization. MAPGD features specialized agents for task clarity, example selection, format design, and stylistic refinement; semantic gradient coordination to resolve conflicts; bandit-based candidate selection for efficient exploration-exploitation; and theoretical convergence guarantees. Experiments on classification, generation, and reasoning tasks show MAPGD outperforms single-agent and random baselines in accuracy and efficiency. Ablations confirm the benefits of gradient fusion, agent specialization, and conflict resolution, providing a unified, gradient-inspired multi-agent approach to robust and interpretable prompt optimization.

Problem

• Optimizing prompts for large language models collaboratively
• Resolving gradient conflicts and narrow perspectives in optimization
• Reducing computational costs while improving adaptability and efficiency

Innovation

• Multi-agent collaboration with gradient optimization
• Semantic gradient coordination resolves conflicts
• Bandit-based candidate selection balances exploration-exploitation