# Installer katago avec Homebrew, dnf, Nix Consultez les chemins d'installation, exécutables, métadonnées et notes de sécurité de katago pour les workflows d'agents IA. ## installation ```sh sudo av install brew:katago ``` Commandes d'installation supplémentaires: ### macOS - Homebrew (100%): ```sh brew install katago ``` Preuve: local Homebrew formula metadata ### Linux - dnf (92%): ```sh sudo dnf install katago ``` Preuve: Fedora Rawhide package metadata: katago from https://dl.fedoraproject.org/pub/fedora/linux/development/rawhide/Everything/x86_64/os/repodata/07190dc5ae9f35ae73866675fed6d95fe6e8d9fe22c9d7cdf85862cb2ed24a4c-primary.xml.zst - Nix (92%): ```sh nix profile install nixpkgs#katago ``` Preuve: nixpkgs package indexes: pkgs/by-name/ka/katago/package.nix from https://api.github.com/repos/NixOS/nixpkgs/git/trees/master?recursive=1 ## Faits du paquet - **Clé du paquet:** brew:katago - **Gestionnaire de paquets:** Homebrew - **Version:** 1.17.1 - **Résumé source:** Neural Network Go engine with no human-provided knowledge - **Page d'accueil:** - **Dépôt:** - **Dernière mise à jour:** 2026-07-31T18:35:29Z - **Généré:** 2026-08-03T19:37:03+00:00 ## exécutables - katago (alias) ## Comportement d'installation - Bouteille: non disponible ## Version et fraîcheur - page générée: 2026-08-03 - version du gestionnaire: 1.17.1 ## Historique du projet et usages KataGo is David J. Wu's open-source Go engine, combining neural-network-guided Monte Carlo tree search with self-play training and a GTP command-line interface for Go GUIs and analysis tools. ### Historique du projet KataGo was released publicly with the February 2019 paper 'Accelerating Self-Play Learning in Go'. The paper and Jane Street announcement emphasized practical efficiency: stronger self-play learning with far less compute than AlphaZero-style baselines, plus Go-specific targets such as score estimation and ownership prediction. The repository describes KataGo as a GTP engine rather than a graphical program. Its docs and README grew into a package for engine users, GUI integrators, and researchers, covering training history, model files, backends, GTP usage, and experiments beyond the original paper. The project continued to incorporate methods documented after the paper, including multiple board-size support, search and training refinements, and later neural-network architecture work. ### Historique d'adoption KataGo became a common analysis engine in the Go software ecosystem because it offered strong play, score estimation, multiple rulesets, variable board sizes, and usable command-line integration with GUIs. Package-manager adoption matters here because KataGo is not a single self-contained GUI. Users often install the engine through Homebrew or binaries, then connect it to tools such as KaTrain, Lizzie-family GUIs, q5Go, Sabaki, or other analysis front ends. ### Modes d'utilisation The README documents KataGo as a GTP engine that generally runs behind a GUI or analysis program. Homebrew installs config files and neural networks under the formula's share directory, and the README shows how to discover those files with brew list --verbose. ### Pourquoi les passionnés de paquets s'y intéressent KataGo is a package-nerd classic because it turns a research-grade AI engine into something installable, scriptable, and composable with separate model files and GUI front ends. The package boundary is unusually visible: executable, backend choice, config, neural net, and GTP consumer all have to line up. ### Chronologie - 2019-02-26: v1.0 release work updated the README for public releases - 2019-02-27: arXiv submission for 'Accelerating Self-Play Learning in Go' - 2019-02-28: Jane Street published the release announcement and linked source code and trained nets - 2023-01-07: v1.12.0 release work landed during a period of neural-network architecture changes ### Related projects - KataGo is related to AlphaGo Zero, ELF OpenGo, Leela Zero, GTP-compatible Go GUIs, KaTrain, Lizzie-family interfaces, q5Go, Sabaki, and David Wu's earlier GoNN research. ### Sources - - - - - - - - ## Notes de sécurité broad file, network, media, or database tool signal. - **Risque Geiger:** blue / moyen - broad file, network, media, or database tool signal ## Autres enregistrements de gestionnaires de paquets - Nix - katago: normalized package name match | nixpkgs package indexes: pkgs/by-name/ka/katago/package.nix from https://api.github.com/repos/NixOS/nixpkgs/git/trees/master?recursive=1 - dnf - katago - 1.14.1-4.fc43: normalized package name match | Fedora Rawhide package metadata: katago from https://dl.fedoraproject.org/pub/fedora/linux/development/rawhide/Everything/x86_64/os/repodata/07190dc5ae9f35ae73866675fed6d95fe6e8d9fe22c9d7cdf85862cb2ed24a4c-primary.xml.zst | GTP engine and self-play learning in Go | https://katagotraining.org - dnf - katago-doc - 1.14.1-4.fc43: normalized package name match | Fedora Rawhide package metadata: katago-doc from https://dl.fedoraproject.org/pub/fedora/linux/development/rawhide/Everything/x86_64/os/repodata/07190dc5ae9f35ae73866675fed6d95fe6e8d9fe22c9d7cdf85862cb2ed24a4c-primary.xml.zst | Documentation for katago | https://katagotraining.org - dnf - katago-eigen - 1.14.1-4.fc43: normalized package name match | Fedora Rawhide package metadata: katago-eigen from https://dl.fedoraproject.org/pub/fedora/linux/development/rawhide/Everything/x86_64/os/repodata/07190dc5ae9f35ae73866675fed6d95fe6e8d9fe22c9d7cdf85862cb2ed24a4c-primary.xml.zst | Documentation for katago - eigen backend | https://katagotraining.org - dnf - katago-opencl - 1.14.1-4.fc43: normalized package name match | Fedora Rawhide package metadata: katago-opencl from https://dl.fedoraproject.org/pub/fedora/linux/development/rawhide/Everything/x86_64/os/repodata/07190dc5ae9f35ae73866675fed6d95fe6e8d9fe22c9d7cdf85862cb2ed24a4c-primary.xml.zst | Documentation for katago - eigen backend - OpenCL backend | https://katagotraining.org ## Combined YAML source View the package source record on GitHub. [combined/katago.yml](https://github.com/mxcl/pkgdb/blob/main/combined/katago.yml) ## Sources - pkg.so package database - Geiger risk classifier - curated package history - pkgdb category and tag curation - external package-manager database matches - cross-ecosystem install command graph