6.7 KiB
Note
本文档由 WeHub 基于上游 README 翻译整理,属于社区翻译,非官方中文文档。
English · 原始项目 · 上游 README
原作者、版权与许可证归属以原始项目及本仓库 LICENSE 文件为准。
.. image:: https://github.com/ray-project/ray/raw/master/doc/source/images/ray_header_logo.png
.. image:: https://readthedocs.org/projects/ray/badge/?version=master :target: http://docs.ray.io/en/master/?badge=master
.. image:: https://img.shields.io/badge/Ray-Join%20Slack-blue :target: https://www.ray.io/join-slack
.. image:: https://img.shields.io/badge/Discuss-Ask%20Questions-blue :target: https://discuss.ray.io/
.. image:: https://img.shields.io/twitter/follow/raydistributed.svg?style=social&logo=twitter :target: https://x.com/raydistributed
.. image:: https://img.shields.io/badge/Get_started_for_free-3C8AE9?logo=data%3Aimage%2Fpng%3Bbase64%2CiVBORw0KGgoAAAANSUhEUgAAABAAAAAQCAYAAAAf8%2F9hAAAAAXNSR0IArs4c6QAAAERlWElmTU0AKgAAAAgAAYdpAAQAAAABAAAAGgAAAAAAA6ABAAMAAAABAAEAAKACAAQAAAABAAAAEKADAAQAAAABAAAAEAAAAAA0VXHyAAABKElEQVQ4Ea2TvWoCQRRGnWCVWChIIlikC9hpJdikSbGgaONbpAoY8gKBdAGfwkfwKQypLQ1sEGyMYhN1Pd%2B6A8PqwBZeOHt%2FvsvMnd3ZXBRFPQjBZ9K6OY8ZxF%2B0IYw9PW3qz8aY6lk92bZ%2BVqSI3oC9T7%2FyCVnrF1ngj93us%2B540sf5BrCDfw9b6jJ5lx%2FyjtGKBBXc3cnqx0INN4ImbI%2Bl%2BPnI8zWfFEr4chLLrWHCp9OO9j19Kbc91HX0zzzBO8EbLK2Iv4ZvNO3is3h6jb%2BCwO0iL8AaWqB7ILPTxq3kDypqvBuYuwswqo6wgYJbT8XxBPZ8KS1TepkFdC79TAHHce%2F7LbVioi3wEfTpmeKtPRGEeoldSP%2FOeoEftpP4BRbgXrYZefsAI%2BP9JU7ImyEAAAAASUVORK5CYII%3D :target: https://www.anyscale.com/ray-on-anyscale?utm_source=github&utm_medium=ray_readme&utm_campaign=get_started_badge
Ray 是一个用于扩展 AI 与 Python 应用的统一框架。Ray 由核心分布式运行时和一组用于简化 ML 计算的 AI 库组成:
.. image:: https://github.com/ray-project/ray/raw/master/doc/source/images/what-is-ray-padded.svg
.. https://docs.google.com/drawings/d/1Pl8aCYOsZCo61cmp57c7Sja6HhIygGCvSZLi_AuBuqo/edit
了解更多关于 Ray AI Libraries_ 的信息:
Data_:面向 ML 的可扩展数据集Train_:分布式训练Tune_:可扩展超参数调优RLlib_:可扩展强化学习Serve_:可扩展且可编程的 Serving
或了解更多关于 Ray Core_ 及其核心抽象的信息:
Tasks_:在集群中执行的无状态函数。Actors_:在集群中创建的有状态 worker 进程。Objects_:可在整个集群中访问的不可变值。
了解更多关于监控与调试的信息:
- 使用
Ray Dashboard <https://docs.ray.io/en/latest/ray-core/ray-dashboard.html>__ 监控 Ray 应用与集群。 - 使用
Ray Distributed Debugger <https://docs.ray.io/en/latest/ray-observability/ray-distributed-debugger.html>__ 调试 Ray 应用。
Ray 可在任何机器、集群、云服务商和 Kubernetes 上运行,并提供不断增长的 ecosystem of community integrations_。
使用以下命令安装 Ray:pip install ray。如需 nightly 构建的 wheel,请参阅 Installation page <https://docs.ray.io/en/latest/ray-overview/installation.html>__。
.. _Serve: https://docs.ray.io/en/latest/serve/index.html
.. _Data: https://docs.ray.io/en/latest/data/data.html
.. _Workflow: https://docs.ray.io/en/latest/workflows/
.. _Train: https://docs.ray.io/en/latest/train/train.html
.. _Tune: https://docs.ray.io/en/latest/tune/index.html
.. _RLlib: https://docs.ray.io/en/latest/rllib/index.html
.. _ecosystem of community integrations: https://docs.ray.io/en/latest/ray-overview/ray-libraries.html
为什么选择 Ray?
当今的 ML 工作负载对计算的需求日益增加。尽管单机开发环境(例如你的笔记本电脑)十分便捷,但无法扩展以满足这些需求。
Ray 提供了一种统一方式,可将 Python 与 AI 应用从笔记本电脑扩展到集群。
借助 Ray,你可以无缝地将同一份代码从笔记本电脑扩展到集群。Ray 被设计为通用框架,意味着它能够高效运行任何类型的工作负载。如果你的应用使用 Python 编写,就可以用 Ray 进行扩展,无需其他基础设施。
更多信息
Documentation_Ray Architecture whitepaper_Exoshuffle: large-scale data shuffle in Ray_Ownership: a distributed futures system for fine-grained tasks_RLlib paper_Tune paper_
较早的文档:
Ray paper_Ray HotOS paper_Ray Architecture v1 whitepaper_
.. _Ray AI Libraries: https://docs.ray.io/en/latest/ray-air/getting-started.html
.. _Ray Core: https://docs.ray.io/en/latest/ray-core/walkthrough.html
.. _Tasks: https://docs.ray.io/en/latest/ray-core/tasks.html
.. _Actors: https://docs.ray.io/en/latest/ray-core/actors.html
.. _Objects: https://docs.ray.io/en/latest/ray-core/objects.html
.. _Documentation: http://docs.ray.io/en/latest/index.html
.. _Ray Architecture v1 whitepaper: https://docs.google.com/document/d/1lAy0Owi-vPz2jEqBSaHNQcy2IBSDEHyXNOQZlGuj93c/preview
.. _Ray Architecture whitepaper: https://docs.google.com/document/d/1tBw9A4j62ruI5omIJbMxly-la5w4q_TjyJgJL_jN2fI/preview
.. _Exoshuffle: large-scale data shuffle in Ray: https://arxiv.org/abs/2203.05072
.. _Ownership: a distributed futures system for fine-grained tasks: https://www.usenix.org/system/files/nsdi21-wang.pdf
.. _Ray paper: https://arxiv.org/abs/1712.05889
.. _Ray HotOS paper: https://arxiv.org/abs/1703.03924
.. _RLlib paper: https://arxiv.org/abs/1712.09381
.. _Tune paper: https://arxiv.org/abs/1807.05118
参与贡献
.. list-table:: :widths: 25 50 25 25 :header-rows: 1
-
- 平台
- 用途
- 预计响应时间
- 支持级别
-
Discourse Forum_- 用于讨论开发与使用相关的问题。
- < 1 day
- Community
-
GitHub Issues_- 用于报告 bug 和提交功能请求。
- < 2 days
- Ray OSS Team
-
Slack_- 用于与其他 Ray 用户协作。
- < 2 days
- Community
-
StackOverflow_- 用于咨询 Ray 的使用方法。
- 3-5 days
- Community
-
Meetup Group_- 用于了解 Ray 项目与最佳实践。
- Monthly
- Ray DevRel
-
Twitter_- 用于及时了解新功能。
- Daily
- Ray DevRel
.. _Discourse Forum: https://discuss.ray.io/
.. _GitHub Issues: https://github.com/ray-project/ray/issues
.. _StackOverflow: https://stackoverflow.com/questions/tagged/ray
.. _Meetup Group: https://www.meetup.com/Bay-Area-Ray-Meetup/
.. _Twitter: https://x.com/raydistributed
.. _Slack: https://www.ray.io/join-slack?utm_source=github&utm_medium=ray_readme&utm_campaign=getting_involved