导包：
1 import torch 2 import torch.nn as nn 3 import torch.nn.functional as F 4 import torch.optim as optim 5 from torchvision import datasets, transforms
关于torchvision：
- torchvision是独立于pytorch的关于图像操作的一些方便工具库。
- torchvision的详细介绍在：https://pypi.org/project/torchvision/0.1.8/
torchvision主要包括一下几个包：
- vision.datasets : 几个常用视觉数据集，可以下载和加载；
- vision.models : 流行的模型，例如 AlexNet, VGG, and ResNet 以及 与训练好的参数；
- vision.transforms : 常用的图像操作，例如：随机切割，旋转等；
- vision.utils : 用于把形似 (3 x H x W) 的张量保存到硬盘中，给一个mini-batch的图像可以产生一个图像格网；
设置参数：
1 #设置超参数 2 torch.manual_seed(53113) #cpu随机种子 3 batch_size = test_batch_size = 32 4 5 #设置GPU参数 6 use_cuda = torch.cuda.is_available() 7 device = torch.device("cuda" if use_cuda else "cpu") 8 kwargs = {'num_workers': 0, 'pin_memory': True} if use_cuda else {}
1.数据预处理
torch.utils.data.DataLoader在训练模型时使用到此函数，用来把训练数据分成多个batch，此函数每次抛出一个batch数据，直至把所有的数据都抛出，也就是个数据迭代器。
DataLoader中的transform参数：接受一个图像返回变换后的图像的函数，相当于图像先预处理下，常用的操作如 ToTensor, RandomCrop，Normalize等，他们可以通过transforms.Compose被组合在一起。
-
.ToTensor()将shape为(H, W, C)的nump.ndarray或img转为shape为(C, H, W)的tensor，其将每一个数值归一化到[0,1]，其归一化方法比较简单，直接除以255即可。
- .Normalize作用就是.ToTensor将输入归一化到(0,1)后，再使用公式”(x-mean)/std”，将每个元素分布到(-1,1)
1 train_loader = torch.utils.data.DataLoader( 2 datasets.MNIST('./mnist_data', #数据集 3 train=True, #如果true，从training.pt创建数据集 4 download=True, #如果ture，从网上自动下载 5 6 transform=transforms.Compose([ 7 transforms.ToTensor(), 8 transforms.Normalize((0.1307,), (0.3081,)) # 所有图片像素均值和方差 9 ])), 10 batch_size = batch_size, 11 shuffle=True, 12 **kwargs) #kwargs是上面gpu的设置
1 test_loader = torch.utils.data.DataLoader( 2 datasets.MNIST('./mnist_data', 3 train=False, #如果False，从test.pt创建数据集 4 transform=transforms.Compose([ 5 transforms.ToTensor(), 6 transforms.Normalize((0.1307,), (0.3081,)) 7 ])), 8 batch_size=test_batch_size, 9 shuffle=True, 10 **kwargs)
查看一下：
1 print(train_loader.dataset[0][0].shape) #torch.Size([1, 28, 28])
2.创建模型
1 class Net(nn.Module): 2 def __init__(self): 3 super(Net, self).__init__() 4 self.conv1 = nn.Conv2d(1, 20, 5, 1) #(in_channels, out_channels, kernel_size, stride=1) 5 self.conv2 = nn.Conv2d(20, 50, 5, 1) #上个卷积网络的out_channels，就是下一个网络的in_channels，所以这里是20 6 7 self.fc1 = nn.Linear(4*4*50, 500) 8 self.fc2 = nn.Linear(500, 10) #10分类 9 10 def forward(self, x): #手写数字的输入维度，(N,1,28,28), N为batch_size 11 x = F.relu(self.conv1(x)) # x = (N,20,24,24) 12 x = F.max_pool2d(x, 2, 2) # x = (N,20,12,12) 13 x = F.relu(self.conv2(x)) # x = (N,50,8,8) 14 x = F.max_pool2d(x, 2, 2) # x = (N,50,4,4) 15 x = x.view(-1, 4*4*50) # x = (N,4*4*50) 16 x = F.relu(self.fc1(x)) # x = (N,4*4*50)*(4*4*50, 500)=(N,500) 17 x = self.fc2(x) # x = (N,500)*(500, 10)=(N,10) 18 return F.log_softmax(x, dim=1) #带log的softmax分类，每张图片返回10个概率
模型初始化：
1 lr = 0.01 2 momentum = 0.5 3 model = Net().to(device) #模型初始化 4 optimizer = optim.SGD(model.parameters(), lr=lr, momentum=momentum) #定义优化器
3.训练函数
1 def train(model, device, train_loader, optimizer, epoch, log_interval=100): 2 model.train() 3 for batch_idx, (data, target) in enumerate(train_loader): 4 data, target = data.to(device), target.to(device) 5 optimizer.zero_grad() 6 output = model(data) #输出的维度[N,10] 这里的data是函数的forward参数x 7 loss = F.nll_loss(output, target) #这里loss求的是平均数，除以了batch 8 loss.backward() 9 optimizer.step() 10 if batch_idx % log_interval == 0: 11 print("Train Epoch: {} [{}/{} ({:0f}%)]\tLoss: {:.6f}".format( 12 epoch, 13 batch_idx * len(data), #100*32 14 len(train_loader.dataset), #60000 15 100. * batch_idx / len(train_loader), #len(train_loader)=60000/32=1875 16 loss.item() 17 ))
4.测试函数
1 def test(model, device, test_loader): 2 model.eval() 3 test_loss = 0 4 correct = 0 5 with torch.no_grad(): 6 for data, target in test_loader: 7 data, target = data.to(device), target.to(device) 8 output = model(data) 9 test_loss += F.nll_loss(output, target, reduction='sum').item() #reduction='sum'代表batch的每个元素loss累加求和，默认是mean求平均 10 11 pred = output.argmax(dim=1, keepdim=True) #pred.shape=torch.Size([32, 1]) 12 13 correct += pred.eq(target.view_as(pred)).sum().item() #target.shape=torch.Size([32]) 14 15 test_loss /= len(test_loader.dataset) 16 17 print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format( 18 test_loss, correct, len(test_loader.dataset), 19 100. * correct / len(test_loader.dataset)))
执行：
1 epochs = 2 2 for epoch in range(1, epochs + 1): 3 train(model, device, train_loader, optimizer, epoch) 4 test(model, device, test_loader) 5 6 save_model = True 7 if (save_model): 8 torch.save(model.state_dict(),"mnist_cnn.pt") #词典格式，model.state_dict()只保存模型参数
训练结果：
1 Train Epoch: 1 [0/60000 (0.000000%)] Loss: 2.297938 2 Train Epoch: 1 [3200/60000 (5.333333%)] Loss: 0.570356 3 Train Epoch: 1 [6400/60000 (10.666667%)] Loss: 0.207343 4 Train Epoch: 1 [9600/60000 (16.000000%)] Loss: 0.094465 5 Train Epoch: 1 [12800/60000 (21.333333%)] Loss: 0.178536 6 Train Epoch: 1 [16000/60000 (26.666667%)] Loss: 0.041227 7 Train Epoch: 1 [19200/60000 (32.000000%)] Loss: 0.136767 8 Train Epoch: 1 [22400/60000 (37.333333%)] Loss: 0.051781 9 Train Epoch: 1 [25600/60000 (42.666667%)] Loss: 0.112557 10 Train Epoch: 1 [28800/60000 (48.000000%)] Loss: 0.058771 11 Train Epoch: 1 [32000/60000 (53.333333%)] Loss: 0.085873 12 Train Epoch: 1 [35200/60000 (58.666667%)] Loss: 0.188629 13 Train Epoch: 1 [38400/60000 (64.000000%)] Loss: 0.092433 14 Train Epoch: 1 [41600/60000 (69.333333%)] Loss: 0.075023 15 Train Epoch: 1 [44800/60000 (74.666667%)] Loss: 0.038028 16 Train Epoch: 1 [48000/60000 (80.000000%)] Loss: 0.038069 17 Train Epoch: 1 [51200/60000 (85.333333%)] Loss: 0.052910 18 Train Epoch: 1 [54400/60000 (90.666667%)] Loss: 0.012891 19 Train Epoch: 1 [57600/60000 (96.000000%)] Loss: 0.033460 20 21 Test set: Average loss: 0.0653, Accuracy: 9799/10000 (98%) 22 23 Train Epoch: 2 [0/60000 (0.000000%)] Loss: 0.057514 24 Train Epoch: 2 [3200/60000 (5.333333%)] Loss: 0.030869 25 Train Epoch: 2 [6400/60000 (10.666667%)] Loss: 0.091362 26 Train Epoch: 2 [9600/60000 (16.000000%)] Loss: 0.059315 27 Train Epoch: 2 [12800/60000 (21.333333%)] Loss: 0.031055 28 Train Epoch: 2 [16000/60000 (26.666667%)] Loss: 0.012735 29 Train Epoch: 2 [19200/60000 (32.000000%)] Loss: 0.104735 30 Train Epoch: 2 [22400/60000 (37.333333%)] Loss: 0.132139 31 Train Epoch: 2 [25600/60000 (42.666667%)] Loss: 0.010015 32 Train Epoch: 2 [28800/60000 (48.000000%)] Loss: 0.012915 33 Train Epoch: 2 [32000/60000 (53.333333%)] Loss: 0.038762 34 Train Epoch: 2 [35200/60000 (58.666667%)] Loss: 0.015236 35 Train Epoch: 2 [38400/60000 (64.000000%)] Loss: 0.163834 36 Train Epoch: 2 [41600/60000 (69.333333%)] Loss: 0.064514 37 Train Epoch: 2 [44800/60000 (74.666667%)] Loss: 0.007881 38 Train Epoch: 2 [48000/60000 (80.000000%)] Loss: 0.074057 39 Train Epoch: 2 [51200/60000 (85.333333%)] Loss: 0.209342 40 Train Epoch: 2 [54400/60000 (90.666667%)] Loss: 0.018052 41 Train Epoch: 2 [57600/60000 (96.000000%)] Loss: 0.012391 42 43 Test set: Average loss: 0.0460, Accuracy: 9851/10000 (99%)