{"id":13677675,"url":"https://github.com/google-deepmind/alphastar","last_synced_at":"2025-06-17T00:39:52.531Z","repository":{"id":58032000,"uuid":"521662706","full_name":"google-deepmind/alphastar","owner":"google-deepmind","description":null,"archived":false,"fork":false,"pushed_at":"2022-09-08T18:15:53.000Z","size":288,"stargazers_count":468,"open_issues_count":5,"forks_count":65,"subscribers_count":11,"default_branch":"main","last_synced_at":"2025-04-29T11:41:12.747Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/google-deepmind.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null}},"created_at":"2022-08-05T14:11:00.000Z","updated_at":"2025-04-27T04:04:33.000Z","dependencies_parsed_at":"2022-08-28T14:22:04.998Z","dependency_job_id":null,"html_url":"https://github.com/google-deepmind/alphastar","commit_stats":null,"previous_names":["google-deepmind/alphastar","deepmind/alphastar"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/google-deepmind/alphastar","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/google-deepmind%2Falphastar","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/google-deepmind%2Falphastar/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/google-deepmind%2Falphastar/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/google-deepmind%2Falphastar/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/google-deepmind","download_url":"https://codeload.github.com/google-deepmind/alphastar/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/google-deepmind%2Falphastar/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":260268635,"owners_count":22983601,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2024-08-02T13:00:45.687Z","updated_at":"2025-06-17T00:39:52.496Z","avatar_url":"https://github.com/google-deepmind.png","language":"Python","funding_links":[],"categories":["Libraries"],"sub_categories":["Python Libraries"],"readme":"# AlphaStar.\n\n[AlphaStar](https://github.com/deepmind/alphastar) is a package from\n[DeepMind](http://deepmind.com) that provides the tools to train an agent to\nmaster StarCraft II offered by [Blizzard Entertainment](http://blizzard.com).\n\nAs part of our open-sourcing efforts to drive more research interest around\nStarCraft II, we provide the following key offerings with this package:\n\n1.  General purpose architectures to train StarCraftII agents in\n    `architectures/` that can be used with different learning algorithms in\n    online and offline settings.\n\n2.  Data readers, offline training and evaluation scripts for fully offline\n    reinforcement learning with Behavior Cloning as a representative example\n    under `unplugged/` directory.\n\n## Setup\n\nWe have tested AlphaStar only in **Python3.9** and **Linux**. Currently, we do\nnot support other operating systems and recommend users to stick to Linux.\n\n### Preliminaries\n\nWe recommend using a Python virtual environment to manage dependencies. This\nshould help to avoid version conflicts and just generally make the installation\nprocess easier.\n\n```shell\npython3 -m venv alphastar\nsource alphastar/bin/activate\npip install --upgrade pip setuptools wheel\n```\n\nAlphaStar depends on [PySC2](https://github.com/deepmind/pysc2) converters for\ndata generation and evaluation. Since the code for converters is written in C++,\nany changes to the converter code will require recompiling the PySC2 native\nextensions. Because of this we offer two different ways to use AlphaStar:\n\n1.  **Installing AlphaStar with `pip`**: this option requires the least setup.\n    However if you make changes to PySC2, or if you want to use a version for\n    which no pre-built wheel is available, you will need to manually build and\n    install a new wheel for PySC2.\n2.  **Building AlphaStar using Bazel**: in this case AlphaStar and PySC2 are\n    built together from source. By default the PySC2 sources are fetched\n    from GitHub. If you wish to use a local repository instead (e.g. because you\n    have made local modifications to PySC2) you should modify\n    `alphastar/WORKSPACE` as described in the comments.\n\n#### Installing with `pip`\n\nIf you're interested in running the bleeding edge versions, you can do so by\ncloning our GitHub repository and then executing the following command from the\nmain directory (where `setup.py` is located):\n\n```\npip install -e .  # For an editable version\npip install .     # For a non-editable version\n```\n\nNote that this will also install all the dependencies of AlphaStar.\n\n### Building with Bazel\n\nFirst, install Bazel by following the instructions\n[here](https://docs.bazel.build/versions/main/install-ubuntu.html).\n\nPySC2 requires C++ 17, so Bazel builds of AlphaStar + PySC2 must use\n`--cxxopt='-std=c++17'`. For example, to build all AlphaStar targets, run the\nfollowing command from the workspace root:\n\n```shell\nbazel build --cxxopt='-std=c++17' ...\n```\n\nTo recursively run all of the tests within the `architectures/` subdirectory:\n\n```shell\nbazel test --cxxopt='-std=c++17' architectures/...\n```\n\nSee the documentation for\n[AlphaStar Unplugged](https://github.com/deepmind/alphastar/blob/master/alphastar/unplugged/README.md)\nfor example `run` commands.\n\nNote: Bazel caches Python package dependencies downloaded from `pip`. To clear\nthis cache (for example if you have edited `requirements.txt`), run `bazel clean\n--expunge`.\n\nYou may wish to use a\n[.bazelrc file](https://docs.bazel.build/versions/main/guide.html#bazelrc-the-bazel-configuration-file)\nto avoid the need to repeatedly specify command-line options, for instance\n`--cxxopt='-std=c++17'`.\n\n## Quickstart\n\nFor quickstart instructions on how to run training and evaluation scripts in\n*fully offline* settings, please refer to\n[this README](https://github.com/deepmind/alphastar/blob/master/alphastar/unplugged/README.md). In\nthis repository, we have not provided any online RL training code. But, the\narchitectures are fit to be used in both online and offline training.\n\n## About\n\nDisclaimer: This is not an official Google product.\n\nIf you use the agents, architectures and offline RL benchmarks published in this\nrepository, please cite our\n[AlphaStar Unplugged](https://openreview.net/pdf?id=Np8Pumfoty) paper.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fgoogle-deepmind%2Falphastar","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fgoogle-deepmind%2Falphastar","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fgoogle-deepmind%2Falphastar/lists"}