def __init__(self):
super(MyConv, self).__init__()
self.conv = nn.Sequential(
exnn.Conv2d(20, 2),
exnn.Conv2d(30, 2),
exnn.Flatten(),
)
def __init__(self):
super(Net, self).__init__()
self.conv = nn.Sequential(exnn.Conv2d(10, kernel_size=5),
nn.MaxPool2d(2), nn.ReLU(),
exnn.Conv2d(20, kernel_size=5),
nn.Dropout2d(), nn.MaxPool2d(2), nn.ReLU())
self.linear = nn.Sequential(exnn.Linear(320, 50), nn.ReLU(),
nn.Dropout(), exnn.Linear(50, 10),
nn.LogSoftmax(dim=1))
def conv2d(out_channels, kernel_size, stride):
conv = nn.Sequential(
exnn.Conv2d(out_channels=out_channels,
kernel_size=kernel_size,
stride=stride),
nn.ReLU())
return conv
def conv2d(out_channels, kernel_size, stride):
return exnn.Conv2d(out_channels=out_channels,
kernel_size=kernel_size,
stride=stride)
def test_cuda_conv2d_with_seq():
net = nn.Sequential(exnn.Conv2d(3, 2))
net = net.to('cuda')
x = torch.randn(10, 20, 28, 28).to('cuda')
y = net(x)
assert list(y.shape) == [10, 3, 27, 27]
def test_cuda_conv2d():
net = exnn.Conv2d(3, 2).to('cuda')
x = torch.randn(10, 20, 28, 28).to('cuda')
y = net(x)
assert list(y.shape) == [10, 3, 27, 27]
def __init__(self):
super(MyConv, self).__init__()
self.conv = exnn.Conv2d(3, 2)
def __init__(self, out_channels, kernel_size, stride):
super(ConvRelu, self).__init__()
self.conv = exnn.Conv2d(out_channels=out_channels,
kernel_size=kernel_size,
stride=stride)
self.relu = nn.ReLU()