Abnormal results

[domadaaaa]

my running results as followed, why dice and loss become negative and zero?

[ 0.43 mins] After 1 epochs, the Average training loss is tensor(-0.1000, device='cuda:0') and average DICE score is tensor(0.2874, device='cuda:0')
[ 0.24 mins] After 1 epochs, the Average validations loss is tensor(-1.1711e+11, device='cuda:0') and average DICE score is tensor(0.0014, device='cuda:0')
[ 0.44 mins] After 2 epochs, the Average training loss is tensor(-0.1628, device='cuda:0') and average DICE score is tensor(0.0002, device='cuda:0')
[ 0.23 mins] After 2 epochs, the Average validations loss is tensor(0., device='cuda:0') and average DICE score is tensor(0., device='cuda:0')<

[domadaaaa]

Is it due to overfitting?

[domadaaaa]

I print the reults of registered_image, finding that

tensor([[[[0., 0., 0.],
[0., 0., 0.],
[0., 0., 0.],
...,
[0., 0., 0.],
[0., 0., 0.],
[0., 0., 0.]],

     [[0., 0., 0.],
      [0., 0., 0.],
      [0., 0., 0.],
      ...,
      [0., 0., 0.],
      [0., 0., 0.],
      [0., 0., 0.]],

     [[0., 0., 0.],
      [0., 0., 0.],
      [0., 0., 0.],
      ...,
      [0., 0., 0.],
      [0., 0., 0.],
      [0., 0., 0.]],

     ...,

     [[0., 0., 0.],
      [0., 0., 0.],
      [0., 0., 0.],
      ...,
      [0., 0., 0.],
      [0., 0., 0.],
      [0., 0., 0.]],

     [[0., 0., 0.],
      [0., 0., 0.],
      [0., 0., 0.],
      ...,
      [0., 0., 0.],
      [0., 0., 0.],
      [0., 0., 0.]],

     [[0., 0., 0.],
      [0., 0., 0.],
      [0., 0., 0.],
      ...,
      [0., 0., 0.],
      [0., 0., 0.],
      [0., 0., 0.]]]], device='cuda:0', grad_fn=<ReshapeAliasBackward0>)

all elements are zero, I use the data in this program