Open-Ended Wargames with Large Language Models
Snow Globe, an ongoing applied research project and resulting software package, uses large language models (LLMs) for automated play of "open-ended" text-based wargames, such as seminar games and political wargames. LLMs enable a light, flexible architecture in which player actions are not restricted to predefined options. The system allows humans to play against or alongside AI agents with specific personas. Every stage of the wargame from scenario preparation to post-game analysis can be optionally carried out by AI, humans, or a combination thereof.
Read more here.
Build the Docker image and run a container.
./docker_setup.sh
Or, install Snow Globe from PyPI. For CPU only:
pip install llm-snowglobe
For GPU support:
CMAKE_ARGS="-DGGML_CUDA=on" pip install llm-snowglobe
After installation, you can simulate a tabletop exercise about an AI incident response.
examples/haiwire.py
Or, simulate a political wargame about a geopolitical crisis.
examples/ac.py
In the latter case, you can use a terminal-based chat interface to discuss the game afterwards, or just press Enter twice to exit.
To play a game between a human and an AI player, launch the server and start a game:
snowglobe_server &
examples/ac.py --human 1
Then, open a browser window and navigate to:
http://localhost:8000
The terminal output will include the ID number for the human player. Type the number into the ID box, click Log In, then click Chat to enter the human player's responses using a graphical user interface.
Make sure to run snowglobe_server from the same file system location where you run the game. Files related to the graphical user interface will be stored in that location.
This repo is released under the Apache License Version 2.0, except for jQuery which is released under the MIT License.