PIANIST: Learning Partially Observable World Models with LLMs for Multi-Agent Decision Making

TL;DR

PIANIST is a framework that enables LLMs to generate effective world models for multi-agent decision making in complex games without domain-specific training, by decomposing the world into intuitive components for efficient planning.

Contribution

The paper introduces PIANIST, a novel approach that decomposes world models into components for zero-shot LLM generation, facilitating decision making without domain-specific training.

Findings

Works well on two different challenging games

Enables fast MCTS simulation using LLM-generated world models

Operates without domain-specific training data

Abstract

Effective extraction of the world knowledge in LLMs for complex decision-making tasks remains a challenge. We propose a framework PIANIST for decomposing the world model into seven intuitive components conducive to zero-shot LLM generation. Given only the natural language description of the game and how input observations are formatted, our method can generate a working world model for fast and efficient MCTS simulation. We show that our method works well on two different games that challenge the planning and decision making skills of the agent for both language and non-language based action taking, without any training on domain-specific training data or explicitly defined world model.