Defining the model factory
If we want to develop solutions that move away from ad hoc, manual, and inconsistent execution and toward ML systems that can be automated, robust, and scalable, then we have to tackle the question of how we will create and curate the star of the show: the models themselves.
In this chapter, we will discuss the key components that have to be brought together to move toward this vision and provide some examples of what these may look like in code. These examples are not the only way to implement these concepts, but they will enable us to start building up our ML solutions toward the level of sophistication we will need if we want to deploy in the real world.
The main components we are talking about here are as follows:
- Training system: A system for robustly training our models on the data we have in an automated way. This consists of all the code we have developed to train our ML models on data.
- Model store: A place to persist successfully...
