Open Furqanf9T opened 3 years ago
Given transposed=1, weight of size [256, 128, 2, 2], expected input[1, 64, 128, 128] to have 256 channels, but got 64 channels instead ` network = Generic_UNetPlusPlus(num_input_channels, base_num_features, num_classes, len(net_num_pool_op_kernel_sizes), 2, 2, nn.Conv2d, nn.InstanceNorm2d, norm_op_kwargs, nn.Dropout2d, dropout_op_kwargs, net_nonlin, net_nonlin_kwargs, True, False, lambda x: x, InitWeights_He(1e-2), net_num_pool_op_kernel_sizes, net_conv_kernel_sizes, False, True, True)
a = torch.randn(1, 1, 256, 256)
b = network(a) `
Hi, friend! : ) i meet the same problem. Are you have solution of this issue?
Given transposed=1, weight of size [256, 128, 2, 2], expected input[1, 64, 128, 128] to have 256 channels, but got 64 channels instead ` network = Generic_UNetPlusPlus(num_input_channels, base_num_features, num_classes, len(net_num_pool_op_kernel_sizes), 2, 2, nn.Conv2d, nn.InstanceNorm2d, norm_op_kwargs, nn.Dropout2d, dropout_op_kwargs, net_nonlin, net_nonlin_kwargs, True, False, lambda x: x, InitWeights_He(1e-2), net_num_pool_op_kernel_sizes, net_conv_kernel_sizes, False, True, True)
a = torch.randn(1, 1, 256, 256)
b = network(a) `![image](https://user-images.githubusercontent.com/37404023/103282758-c4f66380-49fc-11eb-8406-4054d3954d54.png)