This template was created to speed up the model design and development process. If the model passes the test cases in this application, then it will also work in the context of the server.
In the model directory, fill out the code in model.py and config.py according to
the directions in each file. Then, add the requirements for your project into the
requirements.py file in the model directory. You may add any supporting files or
folders that your model needs under the model directory.
Important: All code that your model uses or references MUST be contained in the model directory.
If you need extra information on how your model is performing, a built-in debugging server is provided which will connect to the Docker container the model is running in and return error messages.
If you would like to use this, in the root directory of the project, run the command
docker-compose build debug and then docker-compose up debug. Open a web browser and navigate to
http://localhost:4650.
As you make changes to your model the results will appear on the web page showing the initialization status and a prediction result on a test image.
In the root directory of the project, run the command docker-compose build test and then
docker-compose up test. You will see the results of the test cases in your terminal. If all
test cases pass, then your model will work in the server environment.