About InterpolateSparse2d in interpolator.py #114
Description
In modules/training/interpolator.py, I noticed author use “2. * (x/(torch.tensor([W-1, H-1], device = x.device, dtype = x.dtype))) - 1.” in function normgrid(self, x, H, W) when align_corners = False. But normally, when align_corners = False, shouldn't “(2.* x + 1.) / torch.tensor([W, H], device=x.device, dtype=x.dtype) - 1.” be used?