PrarthanaTS/Cifar10

1import torch.nn.functional as F2import torch.nn as nn3 4dropout = 0.015 6class PrepBlock(nn.Module):7    def __init__(self, dropout):8        super(PrepBlock, self).__init__()9        self.conv = nn.Sequential(10            nn.Conv2d(in_channels=3, out_channels=64, kernel_size=(3, 3), stride=1, padding=1, dilation=1, bias=False),11            nn.ReLU(),12            nn.BatchNorm2d(64),13            nn.Dropout(dropout)14        )15 16    def forward(self, x):17        return self.conv(x)18 19class ConvolutionBlock(nn.Module):20    def __init__(self, in_channels, out_channels):21        super(ConvolutionBlock, self).__init__()22        self.conv = nn.Sequential(23            nn.Conv2d(in_channels=in_channels, out_channels=out_channels, kernel_size=(3, 3), stride=1, padding=1, bias=False),24            nn.MaxPool2d(kernel_size=(2, 2)),25            nn.BatchNorm2d(out_channels),26            nn.ReLU()27        )28    def forward(self, x):29        return self.conv(x)30 31class ResidualBlock(nn.Module):32    def __init__(self, channels):33        super(ResidualBlock, self).__init__()34        self.residual = nn.Sequential(35            nn.Conv2d(in_channels=channels, out_channels=channels, kernel_size=(3, 3), stride=1, padding=1, bias=False),36            nn.BatchNorm2d(channels),37            nn.ReLU(),38            nn.Conv2d(in_channels=channels, out_channels=channels, kernel_size=(3, 3), stride=1, padding=1, bias=False),39            nn.BatchNorm2d(channels),40            nn.ReLU()41        )42    def forward(self, x):43        return x + self.residual(x)44 45 46class Net(nn.Module):47    def __init__(self):48        super(Net, self).__init__()49        self.prep = PrepBlock(dropout)50        self.conv1 = ConvolutionBlock(64, 128)51        self.R1 = ResidualBlock(128)52        self.conv2 = ConvolutionBlock(128, 256)53        self.conv3 = ConvolutionBlock(256, 512)54        self.R2 = ResidualBlock(512)55        self.maxpool = nn.MaxPool2d(kernel_size=(4, 4))56        self.linear = nn.Linear(512, 10)57 58    def forward(self, x):59        x = self.prep(x)60        x = self.conv1(x)61        x = self.R1(x)62        x = self.conv2(x)63        x = self.conv3(x)64        x = self.R2(x)65        x = self.maxpool(x)66        x = x.view(x.size(0), -1)67        x = self.linear(x)68        x = x.view(-1,10)69        return F.log_softmax(x,dim=1)70        return x