Machine learning library TensorFlow can count to potato... I mean, 1.7
Former Google Brain project is Eager to eat your GPU
Open-source machine-learning boffins rejoice! Numerical computation library TensorFlow 1.7.0 made a discreet appearance this morning, just a month after 1.6 dropped.
TensorFlow was originally developed by Google Brain for internal use before going open source in November 2015. The processor-intensive library makes full use of any available Nvidia GPUs to speed things along.
The open-source ML framework is used in the fields of speech and text recognition, and Google is apparently working with the US Department of Defence to use the tech for identifying objects in drone videos.
Presumably the framework will rapidly learn when it is prudent to excise the words "Don't be evil" from Google's Code of Conduct.
TensorFlow is not the only game in town when it comes to neural networks. IBM's Snap ML claims a 46x bump in performance compared to the open-source library.
The updated version moved Eager mode, contributed by Google last October, into the core.
Eager mode sees operations executed as soon as they are called from Python, making debugging less of a headache.
A GUI, the TensorBoard Debugger Plugin, also makes an appearance in this release, although only as an alpha so should be treated with some caution.
TensorFlow data sets have also had experimental support added for sqlite databases, making it easier to pump data into the system.
An exhaustive list of all the changes and the code itself can be found on GitHub. ®