Justifications

Here you can find a list of all available justifications as can be produced by Thoth’s recommendation engine:

A package was removed due to missing ABI symbols in environment

AICoE TensorFlow builds are optimized for AVX2 instruction sets supported in the CPU identified

A backported package was removed from the resolution process

Removing a package based on a constraint

Adviser was killed as allocated CPU time was exceeded

Unhandled TensorFlow release X, gracefully giving up recommending TensorFlow based on CUDA version Y

A security vulnerability found for a package

Last update of the CVE database of known Python package vulnerabilities

A project is known to have vulnerabilities reported in releases

Runtime environment used is no longer supported, it is recommended to switch to another runtime environment

Resolution was terminated based on errors encountered see logs for more info

Software environment supplied is not fully specified

Removing the given package as it does not use enabled index

Skipping the given package as it does not use configured index

Removing a package due to installation time error in the software environment

Consider using intel-tensorflow which is optimized for CPU detected in your environment

The latest stack couldn't be resolved, performing hops across package versions

A version range specifier is too lax

Make sure your environment has proper Intel Performance Libraries when using Intel TensorFlow builds

Consider adjusting OMP_NUM_THREADS environment variable for containerized deployments

No ABI information was found for the given container image

No CPU model or no CPU family provided

No known security vulnerability found for the given release

CVE link not found

No CVE timestamp information found in the database

Development dependencies will not be taken into account

Turning off GPU specific recommendations as one of GPU and CUDA is not available

No observations spotted for this stack in Thoth's database

No more possible paths found for resolution, terminating resolver in iteration N

Consider using UBI or RHEL 8.2 with Python 3.8 that has optimized Python interpreter with performance gain up to 30%

Resolver did not find any stack that would satisfy requirements and stack characteristics given the time allocated

Resolver was killed as allocated memory has been exceeded (OOM)

Pandas in versions >=1.2 dropped Python 3.6 support

No platform provided in the configuration

Using a prescription release

No version of Python specified

Using observations for RHEL instead of UBI, RHEL is ABI compatible with UBI

RHEL major releases guarantee ABI compatibility across minor releases

User's stack was removed

A package requires RPM package present

A package was removed based on solver rules

Thoth S2I s2i-thoth-f31-py37

Thoth S2I s2i-thoth-f32-py38

Thoth S2I s2i-thoth-ubi8-py36

Thoth S2I s2i-thoth-ubi8-py38

Removing any possible versions as the given package that is already present in the base image

The given package has no gathered information regarding security

Removing the given package as pre-releases are disabled

Bandit found issues during static security analysis

No versions were found for a direct dependency

No observations found for the given operating system running a specific Python version

Environment is not fully specified, pre-computed environment markers will not be taken into account

TensorFlow in version <=1.14 used overpinned gast package, using gast in version <=0.2.2

TensorFlow in versions 2.1 and 2.3 can cause runtime errors when running with h5py>=3 caused by library overpinning

TensorFlow in version 2.1 can cause runtime errors when imported, caused by incompatibility between urllib3 and six packages

Upstream TensorFlow=2.4.0 build filtered out as it requires AVX2 instruction set support which is not available

NumPy==1.19.5 is causing issues when used with TensorFlow 2.4

TensorFlow in version 1.13.1 is compatible with NumPy>=1.16.0

TensorFlow in version 2.3 has a bug that prevents from running if multiple TensorFlow processes are active

TensorFlow in version 2.2 and tensorflow-probability cause runtime errors

TensorFlow in versions >=2.1<=2.3 stated SciPy as a dependency but it is not used in the codebase

TensorFlow <2.3>=2.0 states NumPy<2.0.0 as a dependency, but is compatible with NumPy<1.19.0

TensorFlow in version 2.3 produces wrong model accuracy when the model is serialized

TensorFlow in version <=2.4 is slow when tf.keras.layers.Embedding is used

TensorFlow in version 2.3 can produce misleading output summary

Removing a TensorFlow release as it does not provide required symbols in the exposed API

Recommended TensorFlow that supports CUDA present in the runtime environment

Considering also tensorflow-gpu package as the runtime environment used provides CUDA

Considering also intel-tensorflow package as an alternative to TensorFlow

No TensorFlow symbols API found in the database that would match TensorFlow symbols used

Not recommending TensorFlow in version X as this version is not supported on CUDA in version Y

TensorFlow releases that do not support Python 3.9 filtered out

Thoth backend did not respond in time

Recommendations computed based on Thoth community

It is recommended to use Thoth's s2i to have recommendations specific to runtime environment

No CUDA specific trove classifier matching CUDA version used

Inactive package development status stated in trove classifiers

No Python specific trove classifier matching Python version used

Development status stated in trove classifiers might not be suitable to be used given the recommendation type used

Resolver failed as it was unable to resolve direct dependencies

Removing a package as advising on a new release

Scoring user's stack