By Gabriel Borger, IBM Z Data & AI - Offering Manager
We are happy to announce that TensorFlow v2.3 Image is now available on Linux on IBM Z and LinuxONE platforms!
The TensorFlow image is available for free on IBM Z and LinuxONE Container Image Registry. To get TensorFlow on your IBM Z or LinuxONE machine, go to the IBM Z Container Image Registry and follow the instructions to get started. Then click on the image index tab, and search for TensorFlow*.
TensorFlow is one of the most popular AI frameworks. The new TensorFlow v2.3 container image provides users a simple option to access, download, and install TensorFlow on IBM Z and LinuxONE. It also provides Jupyter, Pandas, SciPy, NumPy, and Matplotlib for your convenience. The image is a UBI image, which allows you to use the same image on different Linux flavors, using Docker and Podman as needed.
TensorFlow’s availability on IBM Z & LinuxONE opens the door to a large set of use cases, from fraud detection to automating investment management. Regardless of your preferred use case, TensorFlow on IBM Z and LinuxONE can augment your most latency-sensitive, mission critical workloads with both in-transaction real-time and batch deployments.
To learn more about TensorFlow on IBM Z/LinuxONE, check out the TensorFlow on IBM Z and LinuxONE blog, or if you’re interested in learning more about the IBM Z Container Image Registry, take a look at the introduction to IBM Z and LinuxONE Container Image Registry blog. For additional information on our AI strategy on Z and LinuxONE, please refer to Leverage AI on IBM Z and IBM LinuxONE for accelerated insights blog.
If you are interested in working with us to explore how TensorFlow on IBM Z/LinuxONE can improve your business, feel free to reach out to your local IBM sales representative.
TensorFlow v2.3 Image Frequently Asked Questions (FAQ)
* If you plan on training your TensorFlow models on little endian architecture (e.g. x86) and deploying it on IBM Z/LinuxONE, you’ll want to execute our endian-conversion utility tool prior to importing the model.