Pytorch 实现MNIST

MNIST 包括6万张28x28的训练样本，1万张测试样本，本文使用的CNN网络
导入包

import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torchvision import datasets, transforms
from torch.autograd import Variable

数据集
我们直接使用PyTorch中自带的dataset，并使用DataLoader对训练数据和测试数据分别进行读取。如果下载过数据集这里download可选择False，在这里还定义了超参数batch_size = 64

batch_size = 64


train_dataset = datasets.MNIST(root='./data',
                               train=True,
                               transform=transforms.ToTensor(),
                               download=True)

test_dataset = datasets.MNIST(root='./data',
                              train=False,
                              transform=transforms.ToTensor())


train_loader = torch.utils.data.DataLoader(dataset=train_dataset,
                                           batch_size=batch_size,
                                           shuffle=True)

test_loader = torch.utils.data.DataLoader(dataset=test_dataset,
                                          batch_size=batch_size,
                                          shuffle=False)

定义网络

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        # 输入1通道，输出10通道，kernel 5*5
        self.conv1 = nn.Conv2d(1, 10, kernel_size=5)
        self.conv2 = nn.Conv2d(10, 20, kernel_size=5)
        self.mp = nn.MaxPool2d(2)
        # 输出10个维度，代表0-9
        self.fc = nn.Linear(320, 10)

    def forward(self, x):
        # in_size = 64
        in_size = x.size(0)  
        # x: 64*10*12*12
        x = F.relu(self.mp(self.conv1(x)))
        # x: 64*20*4*4
        x = F.relu(self.mp(self.conv2(x)))
        # x: 64*320
        x = x.view(in_size, -1) 
        # x: 64*10
        x = self.fc(x)
        return F.log_softmax(x,dim=1)

实例化网络

# 实例化
model = Net()

optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.5)

定义训练函数与测试函数，并开始训练

def train(epoch):
    for batch_idx, (data, target) in enumerate(train_loader):
        data, target = Variable(data), Variable(target)
        optimizer.zero_grad()
        output = model(data)
        loss = F.nll_loss(output, target)
        loss.backward()
        optimizer.step()
        if batch_idx % 200 == 0:
            print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(
                epoch, batch_idx * len(data), len(train_loader.dataset),
                       100. * batch_idx / len(train_loader), loss.item()))


def test():
    test_loss = 0
    correct = 0
    for data, target in test_loader:
        with torch.no_grad():
            data = Variable(data)
        target = Variable(target)
        output = model(data)
        # sum up batch loss
        test_loss += F.nll_loss(output, target, size_average=False).item()
        # get the index of the max log-probability
        pred = output.data.max(1, keepdim=True)[1]
        correct += pred.eq(target.data.view_as(pred)).cpu().sum()

    test_loss /= len(test_loader.dataset)
    print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format(
        test_loss, correct, len(test_loader.dataset),
        100. * correct / len(test_loader.dataset)))


for epoch in range(1, 10):
    train(epoch)
    test()

结果

Train Epoch: 1 [0/60000 (0%)]	Loss: 2.303895
Train Epoch: 1 [12800/60000 (21%)]	Loss: 0.811268
Train Epoch: 1 [25600/60000 (43%)]	Loss: 0.354952
Train Epoch: 1 [38400/60000 (64%)]	Loss: 0.435463
Train Epoch: 1 [51200/60000 (85%)]	Loss: 0.079355

Test set: Average loss: 0.1771, Accuracy: 9476/10000 (95%)

Train Epoch: 2 [0/60000 (0%)]	Loss: 0.125518
Train Epoch: 2 [12800/60000 (21%)]	Loss: 0.043049
Train Epoch: 2 [25600/60000 (43%)]	Loss: 0.112176
Train Epoch: 2 [38400/60000 (64%)]	Loss: 0.255673
Train Epoch: 2 [51200/60000 (85%)]	Loss: 0.050152

Test set: Average loss: 0.1162, Accuracy: 9649/10000 (96%)

Train Epoch: 3 [0/60000 (0%)]	Loss: 0.187574
Train Epoch: 3 [12800/60000 (21%)]	Loss: 0.085249
Train Epoch: 3 [25600/60000 (43%)]	Loss: 0.027793
Train Epoch: 3 [38400/60000 (64%)]	Loss: 0.134547
Train Epoch: 3 [51200/60000 (85%)]	Loss: 0.200424

Test set: Average loss: 0.0837, Accuracy: 9744/10000 (97%)

Train Epoch: 4 [0/60000 (0%)]	Loss: 0.056108
Train Epoch: 4 [12800/60000 (21%)]	Loss: 0.147523
Train Epoch: 4 [25600/60000 (43%)]	Loss: 0.029098
Train Epoch: 4 [38400/60000 (64%)]	Loss: 0.088191
Train Epoch: 4 [51200/60000 (85%)]	Loss: 0.145248

Test set: Average loss: 0.0740, Accuracy: 9771/10000 (98%)

Train Epoch: 5 [0/60000 (0%)]	Loss: 0.094464
Train Epoch: 5 [12800/60000 (21%)]	Loss: 0.054841
Train Epoch: 5 [25600/60000 (43%)]	Loss: 0.156082
Train Epoch: 5 [38400/60000 (64%)]	Loss: 0.124878
Train Epoch: 5 [51200/60000 (85%)]	Loss: 0.060035

Test set: Average loss: 0.0666, Accuracy: 9782/10000 (98%)

Train Epoch: 6 [0/60000 (0%)]	Loss: 0.086457
Train Epoch: 6 [12800/60000 (21%)]	Loss: 0.007907
Train Epoch: 6 [25600/60000 (43%)]	Loss: 0.083795
Train Epoch: 6 [38400/60000 (64%)]	Loss: 0.049548
Train Epoch: 6 [51200/60000 (85%)]	Loss: 0.040511

Test set: Average loss: 0.0603, Accuracy: 9814/10000 (98%)

Train Epoch: 7 [0/60000 (0%)]	Loss: 0.012382
Train Epoch: 7 [12800/60000 (21%)]	Loss: 0.021522
Train Epoch: 7 [25600/60000 (43%)]	Loss: 0.044187
Train Epoch: 7 [38400/60000 (64%)]	Loss: 0.134907
Train Epoch: 7 [51200/60000 (85%)]	Loss: 0.119099

Test set: Average loss: 0.0554, Accuracy: 9817/10000 (98%)

Train Epoch: 8 [0/60000 (0%)]	Loss: 0.065057
Train Epoch: 8 [12800/60000 (21%)]	Loss: 0.134118
Train Epoch: 8 [25600/60000 (43%)]	Loss: 0.045595
Train Epoch: 8 [38400/60000 (64%)]	Loss: 0.305029
Train Epoch: 8 [51200/60000 (85%)]	Loss: 0.055589

Test set: Average loss: 0.0507, Accuracy: 9846/10000 (98%)

Train Epoch: 9 [0/60000 (0%)]	Loss: 0.041940
Train Epoch: 9 [12800/60000 (21%)]	Loss: 0.013540
Train Epoch: 9 [25600/60000 (43%)]	Loss: 0.101751
Train Epoch: 9 [38400/60000 (64%)]	Loss: 0.144502
Train Epoch: 9 [51200/60000 (85%)]	Loss: 0.021596

Test set: Average loss: 0.0547, Accuracy: 9821/10000 (98%)


Process finished with exit code 0