Python Tensorflow-Intel For Linux Does Not Exist

Question

I am very new to Python. We have an Azure Python Function. We need to update the packages that it uses. Some of the packages are:

tensorflow==2.15.0
tensorflow-estimator==2.15.0
tensorflow-intel==2.15.0
tensorflow-io-gcs-filesystem==0.31.0

I tested these Windows versions of the packages on my Windows machine. Now I need to download and push the Linux versions of the packages to our python package feed.

When I try to download the packages with the following command

python -m pip download --trusted-host pypi.org --trusted-host files.pythonhosted.org --proxy="http://127.0.0.1:8888/" -r requirements.txt --prefer-binary --only-binary=:all: -d C:\Python\Scoring\mlp\LinuxPackages --platform manylinux_2_17_x86_64 --platform manylinux2010_x86_64 --python-version 3.11

I get everything except tensorflow. I get the following error:

INFO: pip is looking at multiple versions of tensorflow to determine which version is compatible with other requirements. This could take a while.
ERROR: Ignored the following versions that require a different python version: 0.28.0 Requires-Python >=3.7, <3.11
ERROR: Could not find a version that satisfies the requirement tensorflow-intel==2.15.0; platform_system == "Windows" (from tensorflow) (from versions: 0.0.1)
ERROR: No matching distribution found for tensorflow-intel==2.15.0; platform_system == "Windows"

When I look at tensorflow-intel on pypi.org, I see Windows wheel files and there are no source distribution archives that I can use to build a wheel file.

I have already downloaded the main tensorflow package from pypi.org, and I ran my Azure function and seemed to work ok without the tensorflow-intel package, but I am concerned that I may have issues. I did not create this function and we do not have any python experts on our team.

From looking at our package feed currently, it does not contain the tensorflow-intel package, and that package was not listed in the original requirements.txt file. But I don't know if the newer tensorflow package depends in tensorflow-intel or not. It seems like it does.

Is it possible to somehow convert the Windows package to a source distribution set of files so I can then create a whl file that is compatible with Linux? Or, what other options do I have?

Answer 1

From TensorFlow's installation page

Note: Starting with TensorFlow 2.10, Linux CPU-builds for Aarch64/ARM64 processors are built, maintained, tested and released by a third party: AWS. Installing the tensorflow package on an ARM machine installs AWS's tensorflow-cpu-aws package. They are provided as-is. Tensorflow will use reasonable efforts to maintain the availability and integrity of this pip package. There may be delays if the third party fails to release the pip package. See this blog post for more information about this collaboration.

Perhaps tensorflow-cpu-aws is the one you want.

pip install tensorflow-cpu-aws