TorchCraftAI
  • Tutorials
  • API
  • Blog
  • GitHub

›TorchCraftAI

Getting Started

  • Installation (Linux)
  • Installation (Windows)
  • Installation (macOS)
  • Play games with CherryPi

TorchCraftAI

  • Overview
  • System Architecture
  • Core Abstractions
  • Modules Overview
  • Module Training Blueprints

Tutorial - Building Placement

  • Building Placement Intro
  • Neural Network Architecture
  • Supervised Learning
  • Reinforcement Learning

Tutorial - Micro-Manamagent

  • Micromanagement Intro
  • Model
  • Training Setup

Overview of TorchCraftAI

TorchCraftAI is a platform that lets you build agents to play (and learn to play) StarCraft®: Brood War®†. TorchCraftAI includes:

  • A modular framework for building StarCraft agents
  • CherryPi, a bot which plays complete games of StarCraft (1st place SSCAIT 2017-18)
  • A reinforcement learning environment with minigames, models, and training loops
  • TorchCraft support for TCP communication with StarCraft and BWAPI
  • Support for Linux, Windows, and OSX

Get started

See guides for:

  • Linux
  • Windows
  • OSX

Documentation

  • Home
  • Architecture overview
  • Code reference

Tutorials

  • Train a model to place buildings
  • Train a model to fight

Licensing

We encourage you to experiment with TorchCraftAI! See LICENSE, plus more on contributing and our code of conduct.

†: StarCraft is a trademark or registered trademark of Blizzard Entertainment, Inc., in the U.S. and/or other countries. Nothing in this repository should be construed as approval, endorsement, or sponsorship by Blizzard Entertainment, Inc.

← Play games with CherryPiSystem Architecture →
  • Get started
  • Documentation
    • Tutorials
  • Licensing
TorchCraftAI
Docs
Getting Started (Linux)Getting Started (Windows)Getting Started (Mac)API Reference
Community
Starcraft AI DiscordStarcraft AI Facebook groupTorchCraftAI on GitHub
More
TorchCraft on GitHubStarData on GitHubBlog
Facebook Open Source
Copyright © 2019 Facebook