Google launching native support for TensorFlow on Windows including Windows 7, 10 and Windows Server 2016, which was the most request feature after open-sourcing TensorFlow.
Although some Windows users have managed to run TensorFlow in a Docker container, but Google said they have added today a more complete experience including GPU support. The release of TensorFlow r0.12, "enables you to speed up your TensorFlow training with any GPU that runs CUDA 8," writes Google.
This release is published as a pip package in PyPI, so now you can install TensorFlow with a single command:
C:\> pip install tensorflow
And for GPU support:
C:\> pip install tensorflow-gpu
Also, if you're managing Google Cloud Platform (GCP) resources from the command line on Windows through "Cloud Tools for PowerShell," that helps you efficiently author complex scripts to automate and manipulate GCP resources.
However, when you want to run it on Linux, you would have to rewrite all you uber-sophisticated PowerShell script, but not anymore it's required to rewrite your Google Cloud PowerShell scripts to work on Mac or Linux machines—as Microsoft recently released an alpha version of PowerShell that works on both OS X and Ubuntu. And, Google built a .NET Core version of the Tools on top of it.
To preview the bits, you'll have to: "Install Google Cloud SDK and initialize it; Install PowerShell; Download and unzip Cross-Platform Cloud Tools for PowerShell bits."
Google also today added a Trash view in the Google Docs, Sheets, and Slides iOS mobile apps. "You can now view and restore previously deleted files in the Docs, Sheets, and Slides iOS apps. Just select "Trash" from the menu on the left side of the screen," Google said.
In addition, Google also enabled insertion of GIF's in the Docs Android app. Using the Google Keyboard in the Docs Android app, you can now search for and insert GIFs into documents.
Google Earth Timelapse, today in its largest update yet, adds "four additional years of imagery, petabytes of new data, and a sharper view of the Earth from 1984 to 2016."
The updated Timelapse now leveraging the same techniques used to improve Google Maps and Google Earth back in June, reveals a sharper view with truer colors and fewer distracting artifacts of the planet.
"Using Google Earth Engine, we sifted through about three quadrillion pixels—that's 3 followed by 15 zeroes—from more than 5,000,000 satellite images[…]"
"We took the best of all those pixels to create 33 images of the entire planet, one for each year. We then encoded these new 3.95 terapixel global images into just over 25,000,000 overlapping multi-resolution video tiles, made interactively explorable by Carnegie Mellon CREATE Lab's Time Machine library, a technology for creating and viewing zoomable and pannable timelapses over space and time," Google explained.
To view this new Timelapse, head over to the Earth Engine website, or view the mosaics in Google Earth's historical imagery feature on desktop, or watch this YouTube playlist: