KataGo

KataGo is a neural-network-based Go engine. It is one of the strongest open-source Go programs, using deep learning and Monte Carlo Tree Search (MCTS).

MOSAIC integrates KataGo as a built-in opponent for Go environments (OpenSpiel Go, PettingZoo Go).

Installation

KataGo requires a binary and model weights:

# Ubuntu / Debian
sudo apt-get install -y katago

# Or download from GitHub releases
# https://github.com/lightvector/KataGo/releases

Model weights are stored in var/models/katago/.

Service Module

The KataGo integration lives in:

  • gym_gui/services/go_ai/katago_service.py

  • gym_gui/services/go_ai/gtp_engine.py (shared GTP base class)

KataGo communicates via the GTP (Go Text Protocol).

Usage in MOSAIC

  • Human vs KataGo: Play Go against the engine using the GUI.

  • Agent vs KataGo: Train or evaluate RL/LLM agents against KataGo.