How to Train a Machine Learning Model?

So one of the steps would be get from the, like, let's say that we want to train a model or retrain a model from kind of scratch, right? We need to get the particular snapshot of data. This step would ask feature store for particular subset of data and was it this batch API in this content? That's get API. Historical patch. Head against historical API because there's also two types of APIs for real time. You want to have both a single vector and a batch get so let's say five vectors in the very low latency, right? So sometimes people get confused when you say you use batch get into separate. This would be batch get historical data.