SynWorld: Virtual Scenario Synthesis for Agentic Action Knowledge Refinement
This addresses the problem of limited adaptability in LLM-based agents for AI and robotics applications, representing an incremental improvement in agent learning methods.
The paper tackles the challenge of LLM-based agents struggling in novel environments or unconventional action spaces by proposing SynWorld, a framework that synthesizes virtual scenarios and uses Monte Carlo Tree Search exploration to refine action knowledge, demonstrating it as an effective and general approach for learning in new environments.
In the interaction between agents and their environments, agents expand their capabilities by planning and executing actions. However, LLM-based agents face substantial challenges when deployed in novel environments or required to navigate unconventional action spaces. To empower agents to autonomously explore environments, optimize workflows, and enhance their understanding of actions, we propose SynWorld, a framework that allows agents to synthesize possible scenarios with multi-step action invocation within the action space and perform Monte Carlo Tree Search (MCTS) exploration to effectively refine their action knowledge in the current environment. Our experiments demonstrate that SynWorld is an effective and general approach to learning action knowledge in new environments. Code is available at https://github.com/zjunlp/SynWorld.