- python 3.x
- pip3
- virtualenvwrapper If you are using windows powershell it might not work unless you run it with CMD
- anaconda
** Use anaconda or virtualenvwrapper
# with virtualenvwrapper
$ mkvirtualenv -p python3 bot # create virtual environment
$ workon bot # enter VE
(bot) $ pip install -r requirements.txt # install dependencies (make sure you are in the bot VE)
(bot) $ deactivate # exit VE
# with anaconda
$ conda env create -f environment.yml # create virtual environment
$ conda activate bot # enter VE
(bot) $ conda deactivate # exit VE
(bot) $ python -m pkg_name.cmd -h # to train the pkg
(bot) $ python -m api # to run the api endpoint to predict
Split each NN into different packages
.
├── README.md
├── api.py # main API handler
├── requirements.txt
├── setup.py # package information
├── pkg_name # summarizer PKG
│ ├── __init__.py
│ ├── api.py # API to consume the model predictions
│ ├── cmd.py # CLI to train the model
│ ├── model.py # Keras model
│ └── task.py # training logic
└── utils # commons
└── __init__.py