Compute loss between binary matrix and list of 2D coordinates

Question

I have a PyTorch model that produces a list of coordinates, e.g.

>>> coords = model(inputs)
tensor([[1.1000, 3