• Update src/luna/train.py and add your training code
  • Add your training code
  • Save your model in models folder. You can save the model in a subdirectory, or a diffrent folder (not recommended)
  • If your model is saved in a subdirectory of models folder or a different directory, update the model_path argument in the RegisterModel call
  • Update the description argument in the RegisterModel call
• Update src/luna/batchinference.py and add your batch inference code
  • Update the value of model_path variable if you want to download model to a different folder
  • Add your inference code
  • Delete the downloaded model after inference
• Update the run function in src/luna/score.py and add your scoring code
  • Add anything you need to run when the container instance startes and every time when it restarts to the init() function
  • Add your scoring code in the run() function and return the result. Use model_path global variable to locate your model (pre-downloaded to the container)

• Update the conda.yml to add your conda or pip dependencies

You are ready to go!