KataGo / KataGomo
KataGomo extends KataGo's self-play training and analysis system to Gomoku, Renju, Hex, Connect6, Go variants, and many other board games.
01
Board-game AI
QLizzie
A Qt 6 desktop analysis interface for Go, Gomoku, Hex, and related games, built around KataGo/KataGomo-compatible engines.
02
Miscellaneous
UmaAi
An Umamusume scenario simulator.
GitHub ↗Elytra Strategy Lab
Minecraft elytra flight simulation, route optimization, visual analysis, and mod tools.
minecraft-auto-tas
Physics-accurate Minecraft input search, route optimization, macro generation, and client validation.
GitHub ↗codex-minecraft-seed-methodology
A Codex-oriented methodology and worked examples for Minecraft Java seed filtering.
GitHub ↗godot-ballance
An extensible recreation of Ballance built with Godot.
GitHub ↗BallanceContactStabilizer
A Ballance contact-order stabilizer for deterministic TAS playback.
GitHub ↗AutoDenxiMachine
Codex-assisted analysis of googology and ordinal notations.
GitHub ↗AutoGuogaoMachine
A C++ toolkit for googology notation experiments.
GitHub ↗Googology Scale
An interactive browser for large-number and ordinal notations.
hex_langton_ant
Langton's ant on a hexagonal grid.
GitHub ↗tiaoqimeiju
An exhaustive shortest-solution search for peg solitaire.
GitHub ↗ShiBuWanDu
An exhaustive solver for the Ten Moves, Ten Thousand Degrees puzzle.
GitHub ↗huarongdao
An exhaustive solver for the Huarong Dao sliding-block puzzle.
GitHub ↗03
Blog
01
Training AI for a board game you invented
Rules, builds, self-play configuration, model size, evaluation, and release.
02
Integrating a new ruleset into KataGomo
Board state, legal actions, terminal conditions, history, network inputs, and compound moves.
03
Engineering lessons from strong board-game AI
Practical notes on AlphaZero, NNUE, KataGo, search efficiency, and training scale.