drop_block2d

dragon.vm.torch.nn.functional.drop_block2d(
  input,
  p=0.5,
  block_size=1,
  training=True,
  inplace=False
)[source]

Set the blocks over input to zero randomly.

Parameters:
  • input (dragon.vm.torch.Tensor) – The input tensor.
  • p (float, optional, default=0.5) – The probability to zero an element.
  • block_size (int, optional, default=1) – The spatial block size.
  • training (bool, optional, default=True) – Apply dropping if True.
  • inplace (bool, optional, default=False) – Whether to do the operation in-place.
Returns:

dragon.vm.torch.Tensor – The output tensor.