Google has announced the release of a beta version of the popular TensorFlow machine learning library. According to the company, "TensorFlow 2.0 focuses on simplicity and ease of use, with updates like eager execution, intuitive higher-level APIs, and flexible model building on any platform."
TensorFlow 2.0 combines a number of components that have been added to TensorFlow into a "comprehensive platform" that Google says streamlines the machine learning workflow from training through deployment. The new version of TensorFlow also improves compatibility and parity across platforms and components.
A big part of TensorFlow 2.0 is Keras, a high-level API standard for machine learning. While the current version of TensorFlow supports Keras, Google decided to more deeply integrate Keras into TensorFlow going forward and add TensorFlow-specific enhancements in an effort to help developers get started with the library. "A single high-level API reduces confusion and enables us to focus on providing advanced capabilities for researchers," the TensorFlow team explained in a blog post.
To aid developers in migrating to TensorFlow 2.0, Google is offering a conversion tool that updates existing TensorFlow Python code to use the TensorFlow 2.0 APIs where possible and flags instances where automated migration cannot be performed.