Table 1. SE-ResNet architectures for Cifar.

Building modules are shown in brackets, with the numbers of modules stacked. Downsampling is performed by conv3_1, conv4_1, and conv5_1 with a stride of 2.

Name	Output size	SE-ResNet-18	SE-ResNet-26	SE-ResNet-34	SE-ResNet-50	SE-ResNet-66
conv1	32 × 32	7 × 7, 64, stride 2, 3 × 3, 64, stride 1
conv2_x	32 × 32	3 × 3 max pool, stride 2
		$\left[\begin{array}{c}3\times 3,64\\ 3\times 3,64\\ fc,[4,64]\end{array}\right]\times 2$	$\left[\begin{array}{c}1\times 1,64\\ 3\times 3,64\\ 1\times 1,256\\ fc,[16,256]\end{array}\right]\times 2$	$\left[\begin{array}{c}1\times 3,64\\ 3\times 1,64\\ 1\times 3,64\\ 3\times 1,64\\ fc,[4,64]\end{array}\right]\times 2$	$\left[\begin{array}{c}1\times 1,64\\ 3\times 3,64\\ 1\times 1,256\\ fc,[16,256]\end{array}\right]\times 3$	$\left[\begin{array}{c}1\times 3,64\\ 3\times 1,64\\ 1\times 3,64\\ 3\times 1,64\\ fc,[4,64]\end{array}\right]\times 3$
conv3_x	16 × 16	$\left[\begin{array}{c}3\times 3,128\\ 3\times 3,128\\ fc,[8,128]\end{array}\right]\times 2$	$\left[\begin{array}{c}1\times 1,128\\ 3\times 3,128\\ 1\times 1,512\\ fc,[32,512]\end{array}\right]\times 2$	$\left[\begin{array}{c}1\times 3,128\\ 3\times 1,128\\ 1\times 3,128\\ 3\times 1,128\\ fc,[8,128]\end{array}\right]\times 2$	$\left[\begin{array}{c}1\times 1,128\\ 3\times 3,128\\ 1\times 1,512\\ fc,[32,512]\end{array}\right]\times 4$	$\left[\begin{array}{c}1\times 3,128\\ 3\times 1,128\\ 1\times 3,128\\ 3\times 1,128\\ fc,[8,128]\end{array}\right]\times 4$
conv4_x	8 × 8	$\left[\begin{array}{c}3\times 3,256\\ 3\times 3,256\\ fc,[16,256]\end{array}\right]\times 2$	$\left[\begin{array}{c}1\times 1,256\\ 3\times 3,256\\ 1\times 1,1024\\ fc,[64,1024]\end{array}\right]\times 2$	$\left[\begin{array}{c}1\times 3,256\\ 3\times 1,256\\ 1\times 3,256\\ 3\times 1,256\\ fc,[16,256]\end{array}\right]\times 2$	$\left[\begin{array}{c}1\times 1,256\\ 3\times 3,256\\ 1\times 1,1024\\ fc,[64,1024]\end{array}\right]\times 6$	$\left[\begin{array}{c}1\times 3,256\\ 3\times 1,256\\ 1\times 3,256\\ 3\times 1,256\\ fc,[16,256]\end{array}\right]\times 6$
conv5_x	4 × 4	$\left[\begin{array}{c}3\times 3,512\\ 3\times 3,512\\ fc,[32,512]\end{array}\right]\times 2$	$\left[\begin{array}{c}1\times 1,512\\ 3\times 3,512\\ 1\times 1,2048\\ fc,[128,2048]\end{array}\right]\times 2$	$\left[\begin{array}{c}1\times 3,512\\ 3\times 1,512\\ 1\times 3,512\\ 3\times 1,512\\ fc,[32,512]\end{array}\right]\times 2$	$\left[\begin{array}{c}1\times 1,512\\ 3\times 3,512\\ 1\times 1,2048\\ fc,[128,2048]\end{array}\right]\times 3$	$\left[\begin{array}{c}1\times 3,512\\ 3\times 1,512\\ 1\times 3,512\\ 3\times 1,512\\ fc,[32,512]\end{array}\right]\times 3$
fc	1 × 1	global average pool, 10d or 100d fc, softmax