Introduction to the TensorFlow Docs Translations Project
The TensorFlow Docs Translations project plays a crucial role in making TensorFlow's technical information more accessible worldwide. It focuses on translating technical content and Jupyter notebooks available on tensorflow.org. This ensures that developers and learners from different linguistic backgrounds can access TensorFlow resources in their native languages.
How to Get Help
For any document-related issues, users are encouraged to use the "documentation" component on the TensorFlow issue tracker. However, if you have questions regarding TensorFlow usage, it's best to consult the TensorFlow Forum.
Contribution Process
The project welcomes contributors to submit pull requests and reviews via GitLocalize at gitlocalize.com/tensorflow/docs-l10n. Detailed contribution guidelines are provided in the TensorFlow docs contributor guide, and contributors must sign a Contributor License Agreement (CLA) when participating. It’s also important to ensure that your email settings are correct on your git commits.
Exploring Translated Content
To view content in different languages, visitors can use the language switcher on tensorflow.org or add ?hl=<lang> to the URL. For instance, the TensorFlow quickstart guide for beginners is available in multiple languages like Korean and Spanish:
- Korean: Quickstart in Korean
- Spanish: Quickstart in Spanish
In cases where human translations aren't available, some pages may default to machine translations, which are marked with a banner. Users can revert to English if the machine translation proves unhelpful and are encouraged to provide a human translation if possible.
Source Content
The project compiles content from various GitHub repositories into an /site/en-snapshot/ directory for translations. Errors in the source content should be corrected in the upstream repository, not in the translation directory.
Content Limitations
Not every part of tensorflow.org is included in this translation project. Categories such as overview pages, the API reference, outdated versions, images, and installation instructions are translated through separate processes or remain untranslated. GitLocalize effectively filters these non-translated pages.
Maintaining Style Consistency
The project uses the TensorFlow docs notebook tools to ensure consistent formatting and style. Contributors should acquaint themselves with the TensorFlow documentation style guide and the Google developer docs style guide.
Supported Languages
Language support is determined by factors such as site metrics and community demand. Languages no longer maintained are removed due to associated costs. Updates on new language support are communicated via the TensorFlow blog or Twitter. Additionally, the community branch houses content for languages not officially supported, allowing for community-driven translation efforts.
License
The project operates under the Apache License 2.0, which permits open-source collaboration and sharing.
With this framework, the TensorFlow Docs Translations project aims to broaden the reach of TensorFlow's resources, engage a global audience, and foster an inclusive community of technologists and learners.