Machine Learning with openSUSE

Packages, containers and more!

In the past few weeks, many engineers have been working hard to create a foundation in Tumbleweed, the openSUSE rolling-release version, for a variety of Machine Learning frameworks and tools.

They span from Tensorflow 1.13.2 to ONNX 1.6, Caffe, Theano and RStudio (both desktop and server editions). Many of these tools, libraries and frameworks existed in the development project for quite some time but it was time to promote them to the level they deserved to ensure they were made first-class citizen in the openSUSE Factory.

Some of these frameworks have also been made available in other forms than just RPMs, for an even greater and easier consumption.

In fact, the rsudio-desktop container [1] can be found on the openSUSE container registry [2].

In addition, various appliances for data scientists are made available for KVM, XEN,VMware and MS Hyper-V [3].

The appliances for data scientists come with the following software pre-installed:

• Tensorflow 1.13.2
• ONNX 1.6
• Caffe
• Octave
• RStudio-Server

What’s coming next

We are currently baking few more interesting things: first, Tensorflow 2.1 is in the pipeline and should come available to users soon. An update to the latest v.1 API of Tensorflow is also planned, bringing Tensorflow 1.15.2 to the end-users pretty soon.

With the availability of new/up-to-date tools and frameworks, the corresponding containers and appliances will be updated to reflect the new state-of-the-art for those delivery vehicles, too.

For now, I want to wish all of you a very happy hacking with the new available tools. If you have questions or comments, please, do not hesitate to reach out.

[1] docker pull registry.opensuse.org/devel/kubic/containers/container/opensuse-rstudio-desktop:latest

[2] https://registry.opensuse.org/

[3] https://download.opensuse.org/repositories/Virtualization:/Appliances:/Images:/openSUSE-Leap-15.2/images/