东莞市做网站的公司,游戏私人服务器搭建,为什么php导入数据库会乱码,企业品牌网站建设定制开发注#xff1a;书中对代码的讲解并不详细#xff0c;本文对很多细节做了详细注释。另外#xff0c;书上的源代码是在Jupyter Notebook上运行的#xff0c;较为分散#xff0c;本文将代码集中起来#xff0c;并加以完善#xff0c;全部用vscode在python 3.9.18下测试通过书中对代码的讲解并不详细本文对很多细节做了详细注释。另外书上的源代码是在Jupyter Notebook上运行的较为分散本文将代码集中起来并加以完善全部用vscode在python 3.9.18下测试通过同时对于书上部分章节也做了整合。
Chapter7 Modern Convolutional Neural Networks
7.4 Networks with Parallel Connections: GoogLeNet 在GoogLeNet中基本的卷积块被称为Inception块Inception block如下图所示。Inception块由四条并行路径组成前三条路径使用窗口大小为 1 × 1 1\times 1 1×1、 3 × 3 3\times 3 3×3和 5 × 5 5\times 5 5×5的卷积层从不同空间大小中提取信息中间的两条路径先在输入上执行 1 × 1 1\times 1 1×1卷积以减少通道数降低模型的复杂性第四条路径使用 3 × 3 3\times 3 3×3最大汇聚层然后使用 1 × 1 1\times 1 1×1卷积层来改变通道数这四条路径都使用合适的填充来使输入与输出的高和宽一致。最后我们将每条线路的输出在通道维度上连结并构成Inception块的输出。在Inception块中通常调整的超参数是每层输出通道数。
import torch
from torch import nn
from torch.nn import functional as F
from d2l import torch as d2l
import matplotlib.pyplot as pltclass Inception(nn.Module):# c1--c4是每条路径的输出通道数def __init__(self, in_channels, c1, c2, c3, c4, **kwargs):super(Inception, self).__init__(**kwargs)# 线路1单1x1卷积层self.p1_1 nn.Conv2d(in_channels, c1, kernel_size1)# 线路21x1卷积层后接3x3卷积层self.p2_1 nn.Conv2d(in_channels, c2[0], kernel_size1)self.p2_2 nn.Conv2d(c2[0], c2[1], kernel_size3, padding1)# 线路31x1卷积层后接5x5卷积层self.p3_1 nn.Conv2d(in_channels, c3[0], kernel_size1)self.p3_2 nn.Conv2d(c3[0], c3[1], kernel_size5, padding2)# 线路43x3最大汇聚层后接1x1卷积层self.p4_1 nn.MaxPool2d(kernel_size3, stride1, padding1)self.p4_2 nn.Conv2d(in_channels, c4, kernel_size1)def forward(self, x):p1 F.relu(self.p1_1(x))p2 F.relu(self.p2_2(F.relu(self.p2_1(x))))p3 F.relu(self.p3_2(F.relu(self.p3_1(x))))p4 F.relu(self.p4_2(self.p4_1(x)))# 在通道维度上连结输出return torch.cat((p1, p2, p3, p4), dim1)#实现各个模块
b1 nn.Sequential(nn.Conv2d(1, 64, kernel_size7, stride2, padding3),nn.ReLU(),nn.MaxPool2d(kernel_size3, stride2, padding1))b2 nn.Sequential(nn.Conv2d(64, 64, kernel_size1),nn.ReLU(),nn.Conv2d(64, 192, kernel_size3, padding1),nn.ReLU(),nn.MaxPool2d(kernel_size3, stride2, padding1))b3 nn.Sequential(Inception(192, 64, (96, 128), (16, 32), 32),Inception(256, 128, (128, 192), (32, 96), 64),nn.MaxPool2d(kernel_size3, stride2, padding1))b4 nn.Sequential(Inception(480, 192, (96, 208), (16, 48), 64),Inception(512, 160, (112, 224), (24, 64), 64),Inception(512, 128, (128, 256), (24, 64), 64),Inception(512, 112, (144, 288), (32, 64), 64),Inception(528, 256, (160, 320), (32, 128), 128),nn.MaxPool2d(kernel_size3, stride2, padding1))b5 nn.Sequential(Inception(832, 256, (160, 320), (32, 128), 128),Inception(832, 384, (192, 384), (48, 128), 128),nn.AdaptiveAvgPool2d((1,1)),nn.Flatten())net nn.Sequential(b1, b2, b3, b4, b5, nn.Linear(1024, 10))
X torch.rand(size(1, 1, 96, 96))#为了使Fashion-MNIST上的训练更简洁将输入的高和宽从224降到96
for layer in net:X layer(X)print(layer.__class__.__name__,output shape:\t, X.shape)#训练
lr, num_epochs, batch_size 0.1, 10, 128
train_iter, test_iter d2l.load_data_fashion_mnist(batch_size, resize96)
d2l.train_ch6(net, train_iter, test_iter, num_epochs, lr, d2l.try_gpu())
plt.show()训练结果:
