We need to documentation on the staging pipeline for actually compiling a model to TensorFlow/TF Lite, and running it. Specifically, the process for providing sampling data and the format of the `Args` datatype are relevant here.