The issue is with python version and tensorflow version.
It is proposed to use Python 3.5.2 in all cases with Tensorflow 1.10

Installing pyenv:

Use pyenv to be able to install any version of python on any operation system.

curl -L https://raw.githubusercontent.com/yyuu/pyenv-installer/master/bin/pyenv-installer | bash

Add the following to shell:

# Pyenv Stuff
export PATH="$HOME/.pyenv/bin:$PATH"
export PYENV_ROOT="$HOME/.pyenv"
eval "$(pyenv init -)"
eval "$(pyenv virtualenv-init -)"

Installing Python 3.5.2:

Install Python 3.5.2 using pyenv:

pyenv install 3.5.2

Make Python 3.5.2 as global python:

pyenv global 3.5.2

Create virtualenv:

To create a virtual environment in pyenv:

pyenv virtualenv 3.5.2 venv

To activate an environment:

pyenv activate venv

To deactivate:

pyenv deactivate

To remove the environment:

pyenv uninstall venv

Result:

Using the same settings will result in the ability to used the model trained in server using GPUs in lcoal (Laptop) using CPU only.

from mastersthesis.

pyenv has its own virtualenv capabalities.
Investigate them.

from mastersthesis.

Issue solved.

from mastersthesis.