Anchor boxes are just rectangles that are not tied to any particular location or coordinates. That means that all you need to define them are two numbers: their width and height. That was explained in the text of the assignment.
The way to think of their role in YOLO is that they define the aspect ratios of potential objects that can be detected. There is quite a bit of history on the forums by now, including an extensive set of threads that go into a lot more depth about YOLO than is covered in the lectures. E.g. here’s a thread specifically about Anchor Boxes. With a little searching you can find more, e.g. this one that also links to the above thread.