AutoML Translation

Note: Models created after GA using Google NMT as the base model will run on TPUs and a more advanced training architecture. Predictions will be served faster, and quality should be on par with previous results, or even improved. Customer data sets may vary.

Welcome to AutoML Translation, which lets you to create your own, custom translation models so that translation queries return results specific to your domain.

Note: If you don't need a custom model solution, the Translation API provides dynamic translation for over 100 languages.

In this section

• Before you begin - Information on setting up a Google Cloud project that you can use with AutoML Translation.
• Preparing your training data - Information on preparing and formatting training data for AutoML Translation.
• Creating datasets, training models, and evaluating results - How to complete the main steps in creating your custom model.
• Translating content - Using your custom model to translate content.
• REST Reference
• RPC Reference

Discussion group/feedback

You can view discussions and post questions and feedback to the AutoML Translation discussion group.