3构建SVHN数据集的数字识别网络

pytorch 网络构建

import torch
torch.manual_seed(0)
torch.backends.cudnn.deterministic= False
torch.backends.cudnn.benchmark = True
import torchvision.models as models
import torchvision.transforms as transforms
import torchvision.datasets as datasets
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torch.autograd import Variable
from torch.utils.data.dataset import Dataset

一、pytorch常用网络

1.Linear介绍 [全连接层]

nn.Linear(input_feature,out_feature,bias=True)

2.卷积介绍 [2D卷积层]

nn.Conv2d(in_channels,out_channels,kernel_size,stride=1,padding=0,
dilation=1,groups,bias=True,padding_mode='zeros')

##kernel_size,stride,padding 都可以是元组
## dilation 为在卷积核中插入的数量

3.转置卷积介绍 [2D反卷积层]

nn.ConvTranspose2d(in_channels,out_channels,kernel_size,stride=1,
padding=0,out_padding=0,groups=1,bias=True,dilation=1,padding_mode='zeros')

##padding是输入填充，out_padding填充到输出

4.最大值池化层 [2D池化层]

nn.MaxPool2d(kernel_size,stride=None,padding=0,dilation=1)

4.批量归一化层 [2D归一化层]

nn.BatchNorm2d(num_features,eps,momentum,affine=True,
track_running_stats=True)

affine=True 表示批量归一化的α，β是被学到的
track_running_stats=True 表示对数据的统计特征进行关注

https://www.jianshu.com/p/a646cbc913b4

二 pytorch 创建模型的四种方法

假设创建

卷积层–》Relu层–》池化层–》全连接层–》Relu层–》全连接层

# 导入包
import torch
import torch.nn.functional as F
from collections import OrderedDict

1.自定义型[定义在init，前向过程在forward]

class Net1(torch.nn.Module):
    def __init__(self):
      super(Net1, self).__init__()
      self.conv1 = torch.nn.Conv2d(3, 32, 3, 1, 1)
      self.dense1 = torch.nn.Linear(32 * 3 * 3, 128)
      self.dense2 = torch.nn.Linear(128, 10)

    def forward(self, x):
      x = F.max_pool2d(F.relu(self.conv(x)),