CivRealm: A Learning and Reasoning Odyssey in Civilization for Decision-Making Agents

CivRealm is an interactive environment for the open-source strategy game Freeciv-web based on Freeciv, a Civilization-inspired game. Within CivRealm, we provide interfaces for two typical agent types: tensor-based reinforcement learning agents based on the Gymnasium API, and language-based agents powered by language models.

Getting Started

New to Civrealm? Start with our Beginner's Guide. This guide offers an overview of CivRealm's core concepts and provides links to further tutorials.

To the beginner

Advanced Materials

The advanced materials offer comprehensive insights into Civrealm's essential concepts, complemented by valuable background details.

To the old hand

API Reference

The reference guide provides an in-depth explanation of the functions and objects incorporated within Civrealm. It elaborates on the function APIs.

To the reference

Releases

The official versions of CivRealm, along with their associated dependencies and downstream repositories.

Align your version

Contribute

How to Contribute to CivRealm: This guide will help you create or customize the environment.

Contribute to CivRealm

FAQ & Resources

If you have any further questions, the FAQ page and resources may assist you.

Find your answer

We also provide a set of tools for training and evaluating agents, as well as a set of baselines for both agent types. We hope that CivRealm can serve as a testbed for developing and evaluating agents that can learn and reason in complex environments.