Training Pipeline

What is a training pipeline?

A training pipeline is a series of steps or processes that takes input features and labels (for supervised ML algorithms), and produces a model as output. A training pipeline typically reads training data from a feature store, performs model-dependent transformations, trains the model, and evaluates the model before the model is saved to a model registry. If model evaluation is complex, it can also be performed after the model has been saved in a model registry. 

Using a feature store in the training pipeline helps to achieve consistency across different training runs and ensures that the features used for training are of high quality and reproducible.