todo
Github: https://github.com/Alexander-Nasuta/graph-jsp-utils
PyPi: https://pypi.org/project/graph-jsp-utils/
todo update docs
Install the package with pip:
todo
This project is still in development and will have some significant changes before version 1.0.0. This project ist structured according to James Murphy's testing guide and this PyPi-publishing-guide.
If you just want to use the environment, then only the Usage section is relevant for you. If you want to further develop the environment the follow the instructions in the Development section.
Install the package with pip:
todo
TODO: present all major features of the env with ray, stb3
To run this Project locally on your machine follow the following steps:
- Clone the repo
git clone https://github.com/Alexander-Nasuta/graph-jsp-env.git
- Install the python requirements_dev packages.
requirements_dev.txt
includes all the packages of specifiedrequirements.txt
and some additional development packages likemypy
,pytext
,tox
etc.pip install -r requirements_dev.txt
- Install the modules of the project locally. For more info have a look at
James Murphy's testing guide
pip install -e .
For testing make sure that the dev dependencies are installed (requirements_dev.txt
) and the models of this
project are set up (i.e. you have run pip install -e .
).
Then you should be able to run
mypy src
flake8 src
pytest
or everthing at once using tox
.
tox
In this Section describes the used Setup and Development tools. This only relevant if you plan on further develop
All the code was developed and tested locally on an Apple M1 Max 16" MacBook Pro (16-inch, 2021) with 64 GB Unified Memory.
The code should run perfectly fine on other devices and operating Systems (see Github tests).
On a Mac I recommend using Miniforge instead of more common virtual environment solutions like Anacond or Conda-Forge.
Accelerate training of machine learning models with TensorFlow on a Mac requires a special installation procedure, that can be found here. However, this repository provides only the gym environment and no concrete reinforcement learning agents. Todo: example project with sb3 and rl
Setting up Miniforge can be a bit tricky (especially when Anaconda is already installed). I found this guide by Jeff Heaton quite helpful.
On a Windows Machine I recommend Anacond, since Anacond and Pycharm are designed to work well with each other.
I recommend to use Pycharm. Of course any code editor can be used instead (like VS code or Vim).
This section goes over a few recommended step for setting up the Project properly inside Pycharm.
- Mark the
src
directory asSource Root
.
right click on the 'src' -> 'Mark directory as' -> `Source Root`
- Mark the
resources
directory asResource Root
.
right click on the 'resources' -> 'Mark directory as' -> `Resource Root`
- Mark the
tests
directory asTest Source Root
.
right click on the 'tests' -> 'Mark directory as' -> `Test Source Root`
afterwards your project folder should be colored in the following way:
- (optional) When running a script enable
Emulate terminal in output console
Run (drop down) | Edit Configurations... | Configuration | ☑️ Emulate terminal in output console
Distributed under the MIT License. See LICENSE.txt
for more information.