Google just release TensorFlow 2.2.0 with many news features and improvements, the new Profiler for TensorFlow 2 for CPU/GPU/TPU. TensorFlow 2.2.0 is dropping the support for Python 2, which is already passed end of life in January 2020. There are many other features and improvements with this version of TensorFlow. TensorFlow is massively popular deep learning library and mathematical framework for developing various types ML/DL model to solve real-life data science problems.
Nearly found months back TensorFlow 2.1.0 was released with many changes and update on the previous version of TensorFlow. Now in over four month’s time Google releases TensorFlow 2.2.0 with groundbreaking changes and many fixes. TensorFlow 2.2.0 is next major update of the TensorFlow framework. Google added lot of new features and several bug fixes, which makes model development much easier task for the data scientists and machine learning researchers.
The latest version of machine learning library TensorFlow 2.2.0 with many updates is available for download and use for the data scientists. This version of TensorFlow 2.2.0 required Python 3.x and it won’t run on Python 2. So, the support for Python 2 is removed from TensorFlow 2.2.0. So, the prior requirement to use TensorFlow 2.2.0 is Python 3.x, which can be downloaded from official Python website and installed on the system.
Major features in TensorFlow 2.2.0
- Export C++ functions to Python via pybind11
- Deprecation of SWIG, which was used for using the C++ function in Python
- TensorFlow 2.2.0 brings a new Profiler for TensorFlow 2 for CPU/GPU/TPU
- TensorFlow 2.2.0 brings ABI-stable tensorflow::tstring and this will replace tensors std::string
- There are many updates to the tf.keras
- The important change in tf.keras are that the developer will be able to use custom training logic with Model.fit with the use of built-in layers, including metrics, preprocessing layers, and stateful RNN layers.
- There are many bug fixes in TensorFlow 2.2.0. This release also includes the breaking changes such as updating the Huber loss function.
- The breaking changes include deprecation of XLA_CPU and XLA_GPU devices.
- Now the AutoGraph will no longer convert functions passed to tf.py_function, tf.py_func and tf.numpy_function, which is another breaking change.
- There are bugs fixes related to tf.data, tf.lite, tf.keras, and others.
- The TensorFlow 2.2.0 now requires gast version 0.3.3 and it won’t work with other fast.
TensorFlow 2.2.0 is another major update of the massively popular machine learning library.
Ads: Do you want to learn TensorFlow with examples, check tutorials at TensorFlow 2 tutorials.