conv2d_transpose

dragon.nn.conv2d_transpose(
  inputs,
  kernel_shape=3,
  strides=1,
  pads=0,
  dilations=1,
  group=1,
  output_padding=None,
  output_shape=None,
  padding='VALID',
  data_format='NCHW',
  **kwargs
)[source]

Apply the 2d deconvolution.

Set padding to VALID will use the value of pads.

Parameters:
  • inputs (Sequence[dragon.Tensor]) – The tensor x, weight and bias.
  • kernel_shape (Sequence[int], optional, default=3) – The shape of convolution kernel.
  • strides (Sequence[int], optional, default=1) – The stride(s) of sliding window.
  • pads (Sequence[int], optional, default=0) – The zero padding size(s).
  • dilations (Sequence[int], optional, default=1) – The rate(s) of dilated kernel.
  • group (int, optional, default=1) – The group size of convolution.
  • output_padding (Sequence[Union[int, dragon.Tensor]], optional) – The extra size padding to output.
  • output_shape (Sequence[Union[int, dragon.Tensor]], optional) – The output shape for SAME padding.
  • padding ({'VALID', 'SAME', 'SAME_UPPER', 'SAME_LOWER'}, optional) – The padding algorithm.
  • data_format ({'NCHW', 'NHWC'}, optional) – The optional data format.
Returns:

dragon.Tensor – The output tensor.