aminmax increases dimension from 0-dim to 1-dim when given scalar inputs and keepdim. #64008
Description
>>> torch.aminmax(torch.randn(()), dim=0, keepdim=True)
torch.return_types.aminmax(
min=tensor([1.9132]),
max=tensor([1.9132]))
>>> torch.nansum(torch.randn(()), dim=0, keepdim=True)
tensor(0.6512)
>>> torch.argmax(torch.randn(()), dim=0, keepdim=True)
tensor(0)
We should probably make this consistent.
torch.aminmax is increasing the dimension from a 0-dim tensor to a 1-dim tensor.