Resize_ seems to do nothing in convert_conv3d_weight_memory_format

I found a functon in https://github.com/pytorch/pytorch/blob/2ed4d65af0a1993c0df7b081f4088d0f3614283e/torch/nn/utils/memory_format.py#L80C6-L80C7

I do the following test

def convert():
    a = torch.randn(1,2,3,4)
    print(a.is_contiguous()) # True
    print(a.is_contiguous(memory_format = torch.contiguous_format)) # True
    print(a.is_contiguous(memory_format = torch.channels_last)) # False
    print(a.shape) # torch.Size([1, 2, 3, 4])
    print(a.stride()) # (24, 12, 4, 1)

    b = (
        a.clone().contiguous(memory_format=torch.channels_last)
    )
    print(b.is_contiguous()) # False
    print(b.is_contiguous(memory_format = torch.contiguous_format)) # False
    print(b.is_contiguous(memory_format = torch.channels_last)) # True
    print(b.shape) # torch.Size([1, 2, 3, 4])
    print(b.stride()) # (24, 1, 8, 2)
    print(b.size())
    c = b.resize_(
        b.size(), memory_format=torch.channels_last
    ) # ! no use?????
    print(c.is_contiguous()) # False
    print(c.is_contiguous(memory_format = torch.contiguous_format)) # False
    print(c.is_contiguous(memory_format = torch.channels_last)) # True
    print(c.shape) # torch.Size([1, 2, 3, 4])
    print(c.stride()) # (24, 1, 8, 2)

so what is the use of resize_?