v2.5.0

2.5.0 (2024-12-31)

🚀 Features

• Add a clearer action repr to aid debugging. (0621905)
• Flesh out rewards with a base class & add a new reward fn: FrequentAssetRewardFn. (60ba3e0)

🐛 Bug Fixes

• Make short sleep between requests when playing online, preventing temporary bans (f139efb)

@anordin95

v2.4.0

2.4.0 (2024-12-27)

🚀 Features

• Info dict from the environment now contains whether game is done and who won (0d8990b)

🛠️ Refactor

• action to a subclass of ndarray for a clearer interface and consistent repr. (ba73902)
• observation & action for clarity (ddb1a88)

@anordin95

v2.3.0

2.3.0 (2024-12-23)

🚀 Features

• Random grid sizes, speed multipliers and better docstrings (af2535d)

🛠️ Refactor

• improve clarity of grid & gridfactory (731f7f2)

@anordin95

v2.2.1

2.2.1 (2024-12-22)

🐛 Bug Fixes

• Agent order is now alternating, instead of one player having always priority (bb06696)
• Allow cities with 50 army as in the original generals.io (0fe5087)
• Resolve very corner cases when priority takes a role (6eb0db7)

v2.2.0

2.2.0 (2024-12-06)

🚀 Features

• Add default reward based on win/lose (7b2b5b8)

🐛 Bug Fixes

• Actions now MultiDiscrete instead of Dict (9b909ca)
• Channels dont change type from bool to int (0e540b6)
• Fix wrapper handling in gymnasium v1.0.0 (aaf4d43)
• Pass pre-commit checks (6230e5d)

v2.1.0

2.1.0 (2024-11-22)

🚀 Features

• Add new better wrappers (96b27b4)

📚 Documentation

• Add new example, showing usage of RLlib.

v2.0.1

2.0.1 (2024-10-29)

🐛 Bug Fixes

• Align priority moves with generalsio (0b87dea)
• Make online observations include priority (5000702)

v2.0.0

2.0.0 (2024-10-26)

🥳 We are happy to announce new big updates!

🚀 Features

• Agent Deployment - now you can run your agents on online servers by running a single script. You also get replay links from generals.io, so you can analyze later.
• Improved map generation - generals are now spawned sufficiently apart. There always exists at least one path between them that doesn't contain city or mountain.
• Added truncation parameter - so training episodes can be stopped when too long

🐛 Bug Fixes

• Environment deterministicity - envs now generate same exact games when passing same seeds. This makes experiments reproducible.

✨ Enhancements

• Now we also have wiki
• Observation spaces are now more aligned with the real generals.io

@revolko

v1.0.0

1.0.0 (2024-10-14)

🥳 Generals-Bots is out!

We are excited to announce that the initial development phase is now completed!

The goal of this project is to be the last attempt to create a bot development platform for the game generals.io. To achieve this goal, multiple design choices had to be made:

• We chose python for everything code related. It is a home for the most advanced machine-learning tools out there (e.g. numpy, pytorch) and it is only natural to provide a codebase that is as compatible with them as possible.
• The game integration follows Gymnasium and PettingZoo standards. These standards are simple, pythonic and are de-facto THE standards for current Reinforcement-Learning. This will enable development based on machine learning, a feat not possible with previous attempts to build such platform.
• The project is and will be open-source. We want to bring the game to the RL community and the RL community to the game! Having this project open-sourced we believe will create a fertile ground for pushing the limits of what current bots and players are capable of.

🌷 Main features

• 🚀 blazing-fast simulator: run thousands of steps per second with numpy-powered efficiency
• 🤝 seamless integration: fully compatible with RL standards 🤸Gymnasium and 🦁PettingZoo
• 🔧 effortless customization: easily tailor environments to your specific needs
• 🔬 analysis tools: leverage features like interactive replays for deeper insights

🤔 Future development

• ⚙️ Include new game modes
• 📹 Export replays as videos or GIFs
• 🕹️ Allow developers to play against their bots
• ⚡ Introduce new agents with varying strategies as benchmarks

.. we will be happy if you participate!

❤️ Big thanks to

• @jdujava for continuous feedback, bug-spotting, code improvements and implementation of the ExpanderAgent
• @Puckoland for extensive refactoring and bringing more "code culture" to the project
• @kvankova for adding CI/CD features and helping with establishing project development practices

v0.5.1

0.5.1 (2024-10-01)

🐛 Bug Fixes

• fix broken imports after merge (aa7147f)
• fix broken variables and methods (5b3533c)

🛠️ Refactor

• abstract GUI handling (f3552fc)
• add more typing (71e72ab)
• fix RewardFn type alias (0c30cb7)
• From Map to GridFactory (075eb0f)
• Improve Grid class (aa8438e)
• improve GUI handling (7020813)
• move all GUI properties to Properties class (669b8a3)

@Puckoland