|TensorFlow Incorporates Keras|
|Written by Kay Ewbank|
|Friday, 10 November 2017|
There's a new version of Google TensorFlow with Keras included as part of the core API.
TensorFlow is Google's open source tool that can be used for a wide range of parallel computations, including implementing neural networks and other AI learning methods. It is designed to make it easier to work with neural networks and is seen as more general and easier than other options. The new release, 1.4 is now public, and the developers say it's is a big one.
The main headline improvement is the inclusion of Keras in TensorFlow core. Keras is a machine learning framework that is itself a neural-network API. It was written in Python and developed with a focus on enabling fast experimentation. Within TensorFlow, Keras has been integrated with other APIs, including the Estimator API.
Another improvement is that the Dataset API has graduated to core package status. The 1.4 version of the Dataset API also adds support for Python generators, and the developers strongly recommend using the Dataset API to create input pipelines for TensorFlow models because it has better features than the older API, performs better and is cleaner and easier to use. The Dataset API can be used to build complex input pipelines from simple, reusable pieces, taking data from an original source, then manipulating the data to make it suitable for further use. You can transform a dataset by chaining method calls on original object. The most common way to consume values from a Dataset is to make an iterator object that provides access to one element of the dataset at a time.
The new version of TensorFlow also introduces a utility function, estimator.train_and_evaluate, that simplifies training, evaluation, and exporting Estimator models. It means you can use distributed execution for training and evaluation in addition to local execution.
or email your comment to: email@example.com
|Last Updated ( Friday, 10 November 2017 )|