LLM-PySC2: Starcraft II learning environment for Large Language Models

Pengfei Li (Rochester Institute of Technology) · Huanyu Li (Harbin Institute of Technology) · Zhe Ma (Nankai University) · Zongyuan Li (南开大学 Nankai University) · Yanan Ni (National University of Defense Technology) · Runnan Qi (National University of Defense Technology) · Chang Lu (Nankai University) · Lumin Jiang (National University of Defense Technology) · Xu Xiaojie (Nankai University) · Xiangbei Liu (Nankai University) · Yunzheng Guo (Nankai University) · wu hui (Nankai University) · Xian Guo (Nankai University) · Kuihua Huang (National University of Defense Technology) · Xuebo Zhang (Nankai University)

action spaceadvanced llm decision-making systemasynchronous query architecturedecision-making performanceefficient collaborationhallucinationsintelligent decision-makingllm-pysc2 environmentmacro-decision scenariosmicro-operation scenariosmulti-agent collaborationpysc2 backendstarcraft iistarcraft ii multi-agent challenge

The tremendous potential has been demonstrated by large language models (LLMs) in intelligent decision-making problems, with unprecedented capabilities shown across diverse applications ranging from gaming AI systems to complex strategic planning frameworks. However, the StarCraft II platform, which has been widely adopted for validating decision-making algorithms in the past decade, has not yet provided substantial support for this emerging domain. To address issues that LLMs cannot interface with the hundreds of actions of the pysc2 backend and the lack of native support for multi-agent (MA) collaboration, we propose the LLM-PySC2 environment. This is the first environment that offers LLMs the complete pysc2 action space with sufficient multi-modal information and game Wiki knowledge. With an asynchronous query architecture, the environment efficiently interacts with LLMs that maintain a constant latency regardless of the scale of the agents' population. In the experiments, we evaluated LLMs' decision-making performance in both the macro-decision and micro-operation scenarios, with traditional StarCraft II Multi-Agent Challenge (SMAC) tasks and a series of new proposed. Results indicate that LLMs possess the potential to achieve victories in complex scenarios but cannot constantly generate correct decisions, especially in the recovered pysc2 action space and MA settings. Without task-relevant instructions, the pre-trained models suffer from issues such as hallucinations and inefficient collaboration. Our findings suggest that StarCraft II still challenges in the era of large models, revealing that there is a lot to do to develop an advanced LLM decision-making system, and the proposed LLM-PySC2 environment will support future development of LLM-based decision-making solutions.