Machine Learning Framework “TensorFlow 2.3” Released

July 28, the developing team behind the open-source machine learning platform “TensorFlow” released its latest version “TensorFlow 2.3.0”.

Developed by Google, TensorFlow is an open-source machine learning platform. It has Python and C++ APIs and equipped with tools and libraries for developing and training machine learning models.

The latest release TensorFlow 2.3 is the successor of TensorFlow 2.2.0, released in early May.

tf.data module is added with snapshot API and tf.data.experimental.service, which solve input-pipeline bottlenecks and improve resource utilization.

TF Profiler, a profiler for performance optimization, is added with a performance analysis guide for input pipeline performance. tf.distribute.TPUStrategy is now a stable API. This release introduces two new tools; memory profiler and Python tracer.

This release includes experimental support for the new Keras Preprocessing Layers API to handle data processing operation, with support for composite tensor inputs.

The experimental Python API tf.debugging.experimental.enable_dump_debug_info() has been improved. Users can now use a TensorFlow program to dump debugging information to a directory on the file system.

Now the minimum Bazel version required to build TF is 3.1.0. This release brings many more feature enhancements, including non-backward-compatible changes.

TensorFlow
https://www.tensorflow.org/