I've put together a browser demo for MudGym, a reinforcement learning environment I built around MUD2, the multiplayer text adventure game.
It all runs in the browser with WebAssembly. There's an example to train a neural network to reach a goal room, and one to hook up an LLM API to play the game, or you can run a Marimo or JupyterLite notebook against the game with your own code.
The playground pages are mostly built from Marimo WebAssembly notebooks running under Pyodide and the MUD2 game engine is compiled to WASM, running in the browser or locally via Wasmtime.
MUD2 itself feels like a piece of history to me. It was just mind blowing to small me that the other players in the game were /real people/.
If you end up using the env for anything I'd love to hear from you. Thanks!
Hey HN.
I've put together a browser demo for MudGym, a reinforcement learning environment I built around MUD2, the multiplayer text adventure game.
It all runs in the browser with WebAssembly. There's an example to train a neural network to reach a goal room, and one to hook up an LLM API to play the game, or you can run a Marimo or JupyterLite notebook against the game with your own code.
The playground pages are mostly built from Marimo WebAssembly notebooks running under Pyodide and the MUD2 game engine is compiled to WASM, running in the browser or locally via Wasmtime.
MUD2 itself feels like a piece of history to me. It was just mind blowing to small me that the other players in the game were /real people/.
If you end up using the env for anything I'd love to hear from you. Thanks!
The playground: https://mudgym.kingrolo.com
The RL env: https://github.com/rolo/mudgym