PRACT: Optimizing Principled Reasoning and Acting of LLM Agent
This work addresses the challenge of principled reasoning and acting in LLM agents, offering a novel optimization framework that is incremental in its adaptation of existing methods to new scenarios.
The paper tackles the problem of learning and enforcing action principles for LLM agents from trajectory data, introducing the PRAct framework with a Reflective Principle Optimization (RPO) method that uses text gradients for reflection and optimization, resulting in enhanced performance across four environments.
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.