PRACT: Optimizing Principled Reasoning and Acting of LLM Agent

TL;DR

This paper presents PRAct, a framework that enables large language model agents to learn, adapt, and refine action principles through reflection and optimization, improving their performance across various environments.

Contribution

The paper introduces the RPO framework with reward-based and self-reflective methods, advancing how LLM agents learn and adapt action principles from trajectory data.

Findings

PRAct improves agent performance in multiple environments.

RPO effectively refines action principles through reflection.

Self-RPO enables learning without external rewards.

Abstract

We introduce the Principled Reasoning and Acting (PRAct) framework, a novel method for learning and enforcing action principles from trajectory data. Central to our approach is the use of text gradients from a reflection and optimization engine to derive these action principles. To adapt action principles to specific task requirements, we propose a new optimization framework, Reflective Principle Optimization (RPO). After execution, RPO employs a reflector to critique current action principles and an optimizer to update them accordingly. We develop the RPO framework under two scenarios: Reward-RPO, which uses environmental rewards for reflection, and Self-RPO, which conducts self-reflection without external rewards. Additionally, two RPO methods, RPO-Traj and RPO-Batch, is introduced to adapt to different settings. Experimental results across four environments demonstrate that the PRAct agent, leveraging the RPO framework, effectively learns and applies action principles to enhance performance.