7.2 VGG 使用块的网络
1 VGG 块
VGG 由一系列卷积层组成，后面再加上用于空间下采样的最大汇聚层。在最初的VGG论文中，作者使用了带有卷积核、填充为 1（保持高度和宽度）的卷积层，和带有汇聚窗口、步幅为 2（每个块后的分辨率减半）的最大汇聚层
| import torch
from torch import nn
from d2l import torch as d2l
def vgg_block(num_convs, in_channels, out_channels):
"""
实现一个 VGG 块，
使用 3x3，填充为 1 的卷积层，
使用 2x2，步幅为 2 的最大池化层
:param num_convs: 卷积层的数量
:param in_channels: 输入通道数量
:param out_channels: 输出通道数量
:return: 一个 VGG 块
"""
layers = []
for _ in range(num_convs):
layers.append(nn.Conv2d(in_channels, out_channels,
kernel_size=3, padding=1))
layers.append(nn.ReLU())
in_channels = out_channels
layers.append(nn.MaxPool2d(kernel_size=2, stride=2))
return nn.Sequential(*layers)
|
2 VGG 网络
与 AlexNet、LeNet 一样，VGG 网络可以分为两部分：第一部分主要由卷积层和汇聚层组成，第二部分由全连接层组成
超参数变量 conv_arch，该变量指定了每个 VGG 块里卷积层个数和输出通道数
2.1 VGG-11
VGG-11：该网络使用 8 个卷积层和 3 个全连接层
| conv_arch = ((1, 64), (1, 128), (2, 256), (2, 512), (2, 512))
def vgg(conv_arch):
"""
实现 VGG-11 模型
:param conv_arch: 超参数
:return: VGG-11 模型
"""
conv_blks = []
in_channels = 1
# 卷积层部分
for (num_convs, out_channels) in conv_arch:
conv_blks.append(vgg_block(num_convs, in_channels, out_channels))
in_channels = out_channels
return nn.Sequential(
*conv_blks,
nn.Flatten(),
# 全连接层部分
nn.Linear(out_channels * 7 * 7, 4096),
nn.ReLU(),
nn.Dropout(0.5),
nn.Linear(4096, 4096),
nn.ReLU(),
nn.Dropout(0.5),
nn.Linear(4096, 10))
net = vgg(conv_arch)
X = torch.randn(size=(1, 1, 224, 224))
for blk in net:
X = blk(X)
print(blk.__class__.__name__,'output shape:\t',X.shape)
|
| 每层输出的形状 |
|---|
| Sequential output shape: torch.Size([1, 64, 112, 112])
Sequential output shape: torch.Size([1, 128, 56, 56])
Sequential output shape: torch.Size([1, 256, 28, 28])
Sequential output shape: torch.Size([1, 512, 14, 14])
Sequential output shape: torch.Size([1, 512, 7, 7])
Flatten output shape: torch.Size([1, 25088])
Linear output shape: torch.Size([1, 4096])
ReLU output shape: torch.Size([1, 4096])
Dropout output shape: torch.Size([1, 4096])
Linear output shape: torch.Size([1, 4096])
ReLU output shape: torch.Size([1, 4096])
Dropout output shape: torch.Size([1, 4096])
Linear output shape: torch.Size([1, 10])
|
由于 VGG-11 比 AlexNet 计算量更大，因此构建一个通道数较少的网络，足够用于训练 Fashion-MNIST 数据集
| ratio = 4
small_conv_arch = [(pair[0], pair[1] // ratio) for pair in conv_arch]
net = vgg(small_conv_arch)
lr, num_epochs, batch_size = 0.05, 10, 128
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size, resize=224)
d2l.train_ch6(net, train_iter, test_iter, num_epochs, lr, d2l.try_gpu())
|