models.py

import torch
from torch import nn


class VGG16CIFAR(nn.Module):
    """
    VGG16-style network adjusted for small (32x32) inputs.
    """

    def __init__(self, num_classes: int):
        super().__init__()
        self.features = self._make_layers(
            [64, 64, "M", 128, 128, "M", 256, 256, 256, "M",
             512, 512, 512, "M", 512, 512, 512, "M"]
        )
        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
        self.classifier = nn.Sequential(
            nn.Linear(512, 512),
            nn.ReLU(True),
            nn.Dropout(0.5),
            nn.Linear(512, num_classes),
        )
        self._initialize_weights()

    def _make_layers(self, cfg):
        layers = []
        in_channels = 3
        for v in cfg:
            if v == "M":
                layers.append(nn.MaxPool2d(kernel_size=2, stride=2))
            else:
                layers.extend(
                    [
                        nn.Conv2d(in_channels, v, kernel_size=3, padding=1),
                        nn.ReLU(inplace=True),
                    ]
                )
                in_channels = v
        return nn.Sequential(*layers)

    def _initialize_weights(self):
        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                nn.init.kaiming_normal_(m.weight, mode="fan_out", nonlinearity="relu")
                if m.bias is not None:
                    nn.init.constant_(m.bias, 0)
            elif isinstance(m, nn.Linear):
                nn.init.normal_(m.weight, 0, 0.01)
                nn.init.constant_(m.bias, 0)

    def forward(self, x):
        # VGG: a plain stack of conv + nonlinearity + pooling with no shortcuts.
        # This means gradients must pass through every layer sequentially,
        # which can make them shrink (vanish) in very deep networks.
        x = self.features(x)
        x = self.avgpool(x)
        x = torch.flatten(x, 1)
        x = self.classifier(x)
        return x


class BasicBlock(nn.Module):
    expansion = 1

    def __init__(self, in_planes, planes, stride=1):
        super().__init__()
        self.conv1 = nn.Conv2d(in_planes, planes, kernel_size=3, stride=stride,
                               padding=1, bias=False)
        self.bn1 = nn.BatchNorm2d(planes)
        self.relu = nn.ReLU(inplace=True)
        self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=1,
                               padding=1, bias=False)
        self.bn2 = nn.BatchNorm2d(planes)

        self.downsample = None
        if stride != 1 or in_planes != planes:
            self.downsample = nn.Sequential(
                nn.Conv2d(in_planes, planes, kernel_size=1, stride=stride, bias=False),
                nn.BatchNorm2d(planes),
            )

    def forward(self, x):
        identity = x

        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out)

        # ResNet: add a skip (identity) connection.
        # The gradient can flow through this identity path directly,
        # so it does not have to pass through every conv layer.
        # This helps mitigate vanishing gradients as depth increases.
        if self.downsample is not None:
            identity = self.downsample(identity)

        out += identity
        out = self.relu(out)
        return out


class ResNet18CIFAR(nn.Module):
    """
    ResNet-18 adjusted for 32x32 inputs (no initial 7x7 conv or maxpool).
    """

    def __init__(self, num_classes: int):
        super().__init__()
        self.in_planes = 64
        self.conv1 = nn.Conv2d(3, 64, kernel_size=3, stride=1,
                               padding=1, bias=False)
        self.bn1 = nn.BatchNorm2d(64)
        self.relu = nn.ReLU(inplace=True)

        self.layer1 = self._make_layer(64, 2, stride=1)
        self.layer2 = self._make_layer(128, 2, stride=2)
        self.layer3 = self._make_layer(256, 2, stride=2)
        self.layer4 = self._make_layer(512, 2, stride=2)
        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
        self.fc = nn.Linear(512, num_classes)

        self._initialize_weights()

    def _make_layer(self, planes, blocks, stride):
        layers = [BasicBlock(self.in_planes, planes, stride)]
        self.in_planes = planes
        for _ in range(1, blocks):
            layers.append(BasicBlock(self.in_planes, planes, stride=1))
        return nn.Sequential(*layers)

    def _initialize_weights(self):
        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                nn.init.kaiming_normal_(m.weight, mode="fan_out", nonlinearity="relu")
            elif isinstance(m, nn.BatchNorm2d):
                nn.init.constant_(m.weight, 1)
                nn.init.constant_(m.bias, 0)
            elif isinstance(m, nn.Linear):
                nn.init.normal_(m.weight, 0, 0.01)
                nn.init.constant_(m.bias, 0)

    def forward(self, x):
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.relu(x)

        x = self.layer1(x)
        x = self.layer2(x)
        x = self.layer3(x)
        x = self.layer4(x)

        x = self.avgpool(x)
        x = torch.flatten(x, 1)
        x = self.fc(x)
        return x