Diese Seite enthält die Produktionsaktualisierungen von AutoML Translation. Wir empfehlen AutoML Translation-Entwicklern, diese Liste regelmäßig auf neue Ankündigungen zu prüfen. Wesentliche Änderungen werden auch über die Mailingliste google-translate-api angekündigt.

Die neuesten Produktaktualisierungen für Google Cloud finden Sie auf der Seite mit den Google Cloud-Versionshinweisen.

Für neueste Produktaktualisierungen können Sie die URL der Seite in Ihren Feedreader einfügen oder die Feed-URL direkt hinzufügen:

July 17, 2020

For test data, added support for the .tmx file type when evaluating existing models. For more information, see Evaluating models.

April 03, 2020

Integration with VPC Service Controls is now in beta stage.

October 14, 2019

AutoML Translation is now generally available. Translations are now faster and quality should be improved or on par with previous versions; individual results will vary depending on your datasets. In addition, quality improvements can vary across language pairs.

Model lifetime AutoML Models run on a managed platform that is regularly upgraded. AutoML Models need to be re-trained at least every 18 months to ensure consistent performance and stability in translations over time and to get any benefits of the upgrades since original training. Models created before AutoML GA (for example, models created during service Alpha /Beta) need to be re-trained within 18 months of GA to move all assets to the newest version of the platform.

AutoML Translation has upgraded its Google NMT base models onto a new training architecture so that it now runs end-to-end on upgraded TPUs. Going forward, all training using Google NMT as the base model trains and serves on TPUs.

Customers already using AutoML might experience additional differences between v1 and previous versions of the service:

  • Data validation now provides warnings on data import to indicate certain data is not usable for training jobs.
  • Data import takes longer in order to support data validation.
  • To achieve higher quality on the new architecture, training durations are longer.

September 04, 2019

AutoML Translation has enabled Google Cloud Platform (GCP) Console-based UI for all its customers. This improves the Cloud AutoML Translation experience by adding UI elements such as controls to sort/filter datasets, which provides consistency with the rest of GCP and allows for faster release of new features.

The URL for the AutoML Translation UI has changed to

December 11, 2018

The minimum number of sentence pairs required to train a custom model has changed. If you don't explicitly specify training, validation, and test sets, the minimum number of sentence pairs is 1000. If you do explicitly specify training, validation, and test sets, the minimum number of pairs used for validation and test is 100 each.

November 01, 2018

The AutoML Translation service account requires additional permissions to access project resources when using the AutoML Translation UI. You must add the role serviceusage.serviceUsageAdmin. See step 8 in Before You Begin.

September 18, 2018

AutoML Translation supports many new languages as source or target languages paired with English, bringing the total number of supported language pairs to 100. See Language Support for Custom Models for the complete list.

August 24, 2018

The URL for the AutoML Translation UI has changed to The previous URL will remain active.

August 02, 2018

AutoML Translation supports Swahili (sw) as a target language from English (en) as a source language.

Language Pair Language Codes
Swahili <- English sw <- en

July 24, 2018

Models created during the EAP will continue to be valid for the Beta. You do not need to recreate/retrain models. However, the format of resource IDs (dataset IDs, model IDs, and operation IDs) is changing. If you are using hardcoded resource IDs, you need to obtain new IDs for those resources (using the get or list method) by August 31, 2018.

May 24, 2018

The refresh of the AutoML Translation EAP release introduces a new UI design and a new API. The previous version of the AutoML Translation EAP UI is deprecated, and the new AutoML API is not compatible with the previous API.

You must reimport the contents any previously created datasets. You can find the original contents at gs://{project-id}-vcm/{dataset-name}/uploads/{document-name}.txt.

If you used the previous version of the AutoML Translation UI to create custom models based on other custom translation models (rather than the default Google Neural Machine Translation model), you will also need to recreate or retrain those models using the new UI.

March 30, 2018

  • The AutoML Translation EAP web application now enables you to build a custom model based on another custom translation model that you built.
  • A sample Java application is available that demonstrates how to send a request for translation and obtain the response back.

February 27, 2018

The following language translation pairs have been added:

Language Pair Language Codes
Dutch -> English nl -> en
Hebrew -> English iw -> en

February 05, 2018

The following language translation pairs have been added:

Language Pair Language Codes
Arabic -> English ar -> ens
Chinese (Simplified) -> English zh-CN -> ens
French <-> English fr <-> ens
German -> English de -> ens
Italian <-> English it <-> ens
Japanese <- English ja <- ens
Korean <- English ko <- ens
Portuguese (Portugal, Brazil) -> English pt -> ens
Russian <-> English ru <-> ens
Turkish <- English tr <- en

January 17, 2018

AutoML Translation EAP release.