What are the concepts behind instance images, concept images and training text encoder?

[gilroff]

I think getting a conceptual idea of what's actually happening will help me a lot. what's actually happening when we do these 3 different trainings? also what does style training option actually do? it says to lower step counts. that mean lower step counts for instance image training, right? Also, can I have two or three different but related concepts or different varieties of the thing, in the same instance image training under a single unique name? Hope you can help me.