Explaining PyTorch Dataset Class and DataLoader basics to myself

Forgot to mention:

Dataset class is not restricted to PyTorch or TensorFlow

The Dataset class is basically just a normalization of data in order to fit it into the model for training because the model is picky about what it will feed on in order to train

Dataset class is a basic fundamental thing used in any Machine Learning application.

Even if you weren't using a framework, and weren't necessarily calling it the dataset class, you would still be loading the data and converting it into a format the model likes, maybe even make some changes to it like normalize it.

So putting it into the Dataset class just makes it neat, but you could also do it manually when not using a framework, or just call your class something weird like dootaset, it won't matter a lot I guess.

Obviously this is my own interpretation and it could be far from the truth, so before believing someone on the internet, try to confirm things, preferably with a professor if possible :)