Numpy syntax with ellipsis is sometimes useful. Could we add it to torch indexing? It's easy to find a workaround, but the syntax is nice. a = torch.DoubleTensor(3, 4).normal_() print(a[..., 2])
Numpy syntax with ellipsis is sometimes useful. Could we add it to torch indexing? It's easy to find a workaround, but the syntax is nice.
a = torch.DoubleTensor(3, 4).normal_()
print(a[..., 2])