Differences with torch.nn.ConvTranspose1d

torch.nn.ConvTranspose1d

class torch.nn.ConvTranspose1d(
    in_channels,
    out_channels,
    kernel_size,
    stride=1,
    padding=0,
    output_padding=0,
    groups=1,
    bias=True,
    dilation=1,
    padding_mode='zeros'
)(input) -> Tensor

For more information, see torch.nn.ConvTranspose1d.

mindspore.nn.Conv1dTranspose

class mindspore.nn.Conv1dTranspose(
    in_channels,
    out_channels,
    kernel_size,
    stride=1,
    pad_mode='same',
    padding=0,
    dilation=1,
    group=1,
    has_bias=False,
    weight_init=None,
    bias_init=None
)(x) -> Tensor

For more information, see mindspore.nn.Conv1dTranspose.

Differences

PyTorch: Computing a one-dimensional transposed convolution can be thought of as Conv1d solving for the gradient of the input, also known as deconvolution (which is not actually true deconvolution). The input shape is usually \((N,C_{in}, L_{in})\), where \(N\) is the batch size, \(C\) is the spatial dimension, and \(L\) is the length of the sequence. The output shape is \((N,C_{out},L_{out})\), where \(L_{out}=(L_{in}-1)×stride-2×padding+dilation×(kernel\_size-1)+output\_padding+1\).

MindSpore: MindSpore API implements basically the same function as PyTorch, but with the new "pad_mode" parameter. When "pad_mode" = "pad", it is the same as the PyTorch default, and the weight_init and bias_init parameters can be used to configure the initialization method.

Weight Initialization Difference

When weight_init of mindspore.nn.Conv1dTranspose is None , the weight is initialized using HeUniform. This is the same as PyTorch weight initialization.

When bias_init of mindspore.nn.Conv1dTranspose is None , the bias is initialized using Uniform. This is the same as the PyTorch bias initialization.

Categories	Subcategories	PyTorch	MindSpore	Difference
Parameters	Parameter 1	in_channels	in_channels	-
	Parameter 2	out_channels	out_channels	-
	Parameter 3	kernel_size	kernel_size	-
	Parameter 4	stride	stride	-
	Parameter 5	padding	padding	-
	Parameter 6	output_padding	-	Usually used with stride > 1 to adjust output shapes. For example, it is common to set padding to (kernel_size - 1)/2, where setting output_padding = (stride - 1) ensures that input shapes/output shapes = stride. MindSpore does not have this parameter
	Parameter 7	groups	group	Same function, different parameter names
	Parameter 8	bias	has_bias	PyTorch defaults to True, and MindSpore defaults to False
	Parameter 9	dilation	dilation	-
	Parameter 10	padding_mode	-	Numeric padding mode, only supports "zeros" i.e. padding 0. MindSpore does not have this parameter, but pads 0 by default
	Parameter 11	-	pad_mode	Specify the padding mode. Optional values are "same", "valid", "pad". In "same" and "valid" mode, padding must be set to 0, and default is "same". PyTorch does not have this parameter
	Parameter 12	-	weight_init	The initialization method for the weight parameter. The type can be Tensor, str, Initializer or numbers.Number. When using str, the values of "TruncatedNormal", "Normal", "Uniform", "HeUniform" and "XavierUniform" distributions and the constants "One" and "Zero" distributions can be selected. The default is None. PyTorch does not have this parameter
	Parameter 13	-	bias_init	The initialization method for the bias parameter. The initialization method is the same as "weight_init", and the default is None. PyTorch does not have this parameter
Input	Single input	input	x	Same function, different parameter names

Code Example 1

Both APIs implement one-dimensional transposed convolutional operations and need to be instantiated first when used. When PyTorch sets output_padding to 0 and MindSpore sets pad_mode to "pad", the output width is \(L_{out}=(L_{in}-1)×stride-2×padding+dilation×( kernel\_size-1)+1\). The weights are initialized to 1 in PyTorch through net.weight.data = torch.ones(), and MindSpore sets the parameter weight_init = "ones" directly.

# PyTorch
import torch
from torch import tensor
import torch.nn as nn
import numpy as np

k = 4
x_ = np.ones([1, 3, 50])
x = tensor(x_, dtype=torch.float32)
net = nn.ConvTranspose1d(3, 64, kernel_size=k, stride=1, padding=0, output_padding=0, bias=False)
net.weight.data = torch.ones(3, 64, k)
output = net(x).detach().numpy()
print(output.shape)
# (1, 64, 53)


# MindSpore
import mindspore as ms
import mindspore.nn as nn
import numpy as np

k = 4
x_ = np.ones([1, 3, 50])
x = ms.Tensor(x_, ms.float32)
net = nn.Conv1dTranspose(3, 64, kernel_size=k, weight_init='ones', pad_mode='pad')
output = net(x)
print(output.shape)
# (1, 64, 53)

Code Example 2

To make the output the same width as the input after dividing stride, PyTorch sets output_padding = stride - 1 and padding to (kernel_size - 1)/2. MindSpore sets pad_mode = "same" and padding = 0.

# PyTorch
import torch
from torch import tensor
import torch.nn as nn
import numpy as np

k = 5
s = 3
x_ = np.ones([1, 3, 50])
x = tensor(x_, dtype=torch.float32)
net = nn.ConvTranspose1d(3, 64, kernel_size=k, stride=s, padding=(k-1)//2, output_padding=s-1, bias=False)
net.weight.data = torch.ones(3, 64, k)
output = net(x).detach().numpy()
print(output.shape)
# (1, 64, 150)


# MindSpore
import mindspore as ms
import mindspore.nn as nn
import numpy as np

k = 5
s = 3
x_ = np.ones([1, 3, 50])
x = ms.Tensor(x_, ms.float32)
net = nn.Conv1dTranspose(3, 64, kernel_size=k, stride=s, weight_init='ones', pad_mode='same', padding=0)
output = net(x)
print(output.shape)
# (1, 64, 150)

Code Example 3

If no padding is done on the original image, a part of the data may be discarded in the case of stride>1. The output width is \(L_{out}=(L_{in}-1)×stride+dilation×(kernel\_size-1)+1\). Set padding and output_padding to 0 in PyTorch and set pad_mode = "valid" in MindSpore.

# PyTorch
import torch
from torch import tensor
import torch.nn as nn
import numpy as np

k = 5
s = 3
x_ = np.ones([1, 3, 50])
x = tensor(x_, dtype=torch.float32)
net = nn.ConvTranspose1d(3, 64, kernel_size=k, stride=s, padding=0, output_padding=0, bias=False)
net.weight.data = torch.ones(3, 64, k)
output = net(x).detach().numpy()
print(output.shape)
# (1, 64, 152)


# MindSpore
import mindspore as ms
import mindspore.nn as nn
import numpy as np

k = 5
s = 3
x_ = np.ones([1, 3, 50])
x = ms.Tensor(x_, ms.float32)
net = nn.Conv1dTranspose(3, 64, kernel_size=k, stride=s, weight_init='ones', pad_mode='valid', padding=0)
output = net(x)
print(output.shape)
# (1, 64, 152)