Optimas: Optimizing Compound AI Systems with Globally Aligned Local Rewards

TL;DR

Optimas introduces a unified framework for optimizing complex compound AI systems by aligning local rewards with global performance, enabling effective independent component updates and outperforming existing methods.

Contribution

The paper proposes a novel approach that maintains local reward functions for each component, ensuring their alignment with overall system performance, which improves optimization of heterogeneous compound AI systems.

Findings

Optimas achieves an average of 11.92% performance improvement across five real-world systems.

The framework enables independent updates of diverse system components while maintaining global alignment.

Extensive evaluations demonstrate Optimas's effectiveness over strong baseline methods.

Abstract

Compound AI systems integrating multiple components, such as Large Language Models, specialized tools, and traditional machine learning models, are increasingly deployed to solve complex real-world tasks. However, optimizing compound systems remains challenging due to their non-differentiable structures and diverse configuration types across components, including prompts, hyperparameters, and model parameters. To address this challenge, we propose Optimas, a unified framework for effective optimization of compound systems. The core idea of Optimas is to maintain one Local Reward Function (LRF) per component, each satisfying a local-global alignment property, i.e., each component's local reward correlates with the global system performance. In each iteration, Optimas efficiently adapts the LRFs to maintain this property while simultaneously maximizing each component's local reward. This approach enables independent updates of heterogeneous configurations using the designated optimization method, while ensuring that local improvements consistently lead to performance gains. We present extensive evaluations across five real-world compound systems to demonstrate that Optimas outperforms strong baselines by an average improvement of 11.92%, offering a general and effective approach for improving compound systems. Our website is at https://optimas.stanford.edu.