Basic question about meaning of input and output sizes for torch.nn.Linear

Consider torch.nn.Linear(1600, 128).

If previous layer has dimensions H X W X D, does that mean there are 1600 sets of connections
where each set consists of connections to each of the H X W X D values? In other words, if
I was drawing a diagram, I would need to draw 1600 X H X W X D lines?

Likewise, does the 128 mean that there are 128 sets of connections where each set consists
of connections to the 1600 values? In other words, would a diagram for this part require
drawing 128 x 1600 lines?

Thanks.