Understanding the slice notation in mini batch gradient descent

Question

There are many answers on ways to implement mini batch in a neural network, but I wanted clarification on the notation of slicing. A simple example;

mini_batc