The "coords" represent the "position" in your sparse matrix. In your case, your matrix might look like this:

[
   [N/A, 3  ],
   [4  , N/A]
]

A Compressed Sparse Row (csr-) matrix is used when values are only sparsely present in your matrix, and therefore, for memory perspective, it is more efficient to just store the values that do exist, and where they are located (the "coords").

Think of your matrix as a bookshelf with multiple shelves, horizontally and vertically. Let's say you are moving houses and need to empty the bookshelf, and then reassemble it later. You might put the content of each box, and label it with its position within the bookshelf (0, 0) for the first shelf on the bottom, (1, 1) for the second shelf on the top.