TensorFlow 1.7.0 released: updates and improvements

Early this month, TensorFlow released its major version 1.6.0. Soon after that they announced rc-0 and rc-1 for TensorFlow 1.7.0. And to our surprise, TensorFlow 1.7.0 has arrived much sooner than expected! Clearly moving quickly is essential to the TensorFlow team.

Both the rc-0 and rc-1 gave us a starter on what might be expected in the TF 1.7.0. This major release contains with some major improvements, features, bug fixes, and other changes.

Major features and improvements in TensorFlow 1.7.0:

Eager mode is moving out of contrib, try tf.enable_eager_execution().

EGraph rewrites emulating fixed-point quantization compatible with TensorFlow Lite are now supported by new tf.contrib.quantize package.

Easy customize gradient computation are now available with tf.custom_gradient.

TensorBoard Debugger Plugin, the graphical user interface (GUI) of TensorFlow Debugger (tfdbg), is now in alpha.

Experimental support for reading a sqlite database as a Dataset with new tf.contrib.data.SqlDataset.

Distributed Mutex / CriticalSection added to tf.contrib.framework.CriticalSection.

Better text processing with tf.regex_replace.

Easy, efficient sequence input with tf.contrib.data.bucket_by_sequence_length

Bug fixes in TF 1.7.0 version include:

- Added MaxPoolGradGrad support for XLA and disabled CSE pass from Tensorflow in XLA.
- Added support for building C++ Dataset op kernels as external libraries, using the tf.load_op_library() mechanism.
- Added support for scalars in tf.contrib.all_reduce.
- Deprecated tf.contrib.learn.