Google today announced the release of the Beta version of TensorFlow 2.0. The new version of the world’s most popular open source machine learning library is being welcomed by developers. Since its initial release in 2015, the Google Brain product has been downloaded over 41 million times. The last major upgrade on the framework came at the TensorFlow Dev Summit in March.
TensorFlow 2.0 focuses on simplicity and ease of use, updating eager execution, intuitive higher-level APIs, and flexible model building on any platform. Major TensorFlow 2.0 Beta upgrades:
- Keras as a high-level API for quick and easy model design and training
- Eager execution as a default for fast, intuitive development and debugging
@tf.functionfor graph performance and portability
The TensorFlow team have completed renaming and deprecating symbols for the 2.0 API, which means the current API is finalized. The new version also adds 2.0 support including model subclassing for Keras features, simplified API for custom training loops, and distribution strategy support for most kinds of hardware.
Although support for TensorFlow Extended (TFX) components and end to end pipelines is not yet complete, the Beta version can work with core components of the TensorFlow product ecosystem including TensorBoard, TensorFlow Hub, TensorFlow Lite, and TensorFlow.js.
The TensorFlow team hopes to resolve additional issues before the release candidate (RC) 2.0 version, complete Keras model support on Cloud TPUs and TPU pods, and improve overall 2.0 performance. The RC release is expected sometime this summer.
The TensorFlow team tweeted that today’s Beta version has “closed over 100 issues you (the developers) reported against the alpha release” and encouraged 2.0 tagged community feedback through Google mailing lists.
Journalist: Fangyu Cai | Editor: Michael Sarazen