What you will find here
The primary focus of our activities is to enable high performance TensorFlow application deployed on the OpenShift Container Platform.
Building TensorFlow from source can give better performance than the official standard build. This is because the
default TensorFlow binaries target the broadest range of hardware to make TensorFlow accessible to everyone. If using CPUs for training or inference, it is recommended to compile TensorFlow with all of the optimizations available for the CPU in use.
Who we are
The Artificial Intelligence Center of Excellence (AICoE) is part of Red Hat's Office of the CTO. Our core mission is to:
- Lead Red Hat on a path to fully embrace the AI transformation leveraging momentum and strengths from our existing product portfolio and unique Open Source philosophy
- In collaboration with the different product, service, and operations teams, we advance the application of state-of-the-art AI
As part of our mission we strive to provide a optimized developer experience on the OpenShift Container Platform. A part of this, are the artifacts you will finde here.
How we build them
To build the artifacts you can find here, we are using OpenShift. A build is triggered for each release of TensorFlow and will result in an optimized artifact published in the corresponding Python Module index on this site. For details on the build process you may have a look at our build tool chain.
Python Simple Index
How to use it
Using these wheels with your Python application is very simple, just use the directory we provide as a source in your Pipfile or requirements.txt.
Pipenv
Here is a simple example of a
Pipfile.
[[source]]
url ="https://tensorflow.pypi.thoth-station.ninja/index/manylinux2010/AVX2/simple/"
verify_ssl = true
name = "redhat-aicoe-experiments"
[[source]]
url = "https://pypi.python.org/simple"
verify_ssl = true
name = "pypi"
[requires]
python_version = "3.6"
[packages]
tensorflow = {version="==1.13.2", index="redhat-aicoe-experiments"}
[pipenv]
allow_prereleases = true