PyTorch训练MNIST数据集2

1.网络结构

 2.代码：

import torch
from torchvision import transforms
from torchvision import datasets
from torch.utils.data import DataLoader
import torch.nn.functional as  F
import torch.optim as optim
from matplotlib import pyplot as plt
import os
import sys

batch_size = 64
transform = transforms.Compose(
    [
        transforms.ToTensor(),
        transforms.Normalize((0.1307,),(0.3081,))
    ]
)
train_dataset =datasets.MNIST(root='../dataset/mnist',
                                train = True,
                                download = True,
                                transform = transform)
train_loader = DataLoader(train_dataset,
                        shuffle=True,
                        batch_size=batch_size)
test_dataset =datasets.MNIST(root='../dataset/mnist',
                                train = False,
                                download = True,
                                transform = transform)
test_loader = DataLoader(test_dataset,
                        shuffle = False,
                        batch_size=batch_size)





class  Net(torch.nn.Module):
    def __init__(self):
        super(Net,self).__init__()
        self.conv1 = torch.nn.Conv2d(1,10,kernel_size = 5)
        self.conv2 = torch.nn.Conv2d(10,20,kernel_size = 5)
        self.pooling = torch.nn.MaxPool2d(2)
        self.f1 = torch.nn.Linear(320,10)
    def forward(self,x):
        match_size = x.size(0)
        #print(x.shape)         #torch.Size([64, 1, 28, 28])
        x = self.pooling(F.relu(self.conv1(x)))
        #print(x.shape)         torch.Size([64, 10, 12, 12])
        x = self.pooling(F.relu(self.conv2(x)))
        #print(x.shape)         torch.Size([64, 20, 4, 4])
        x = x.view(match_size,-1)
        #print(x.shape)         #torch.Size([64, 320])
        x = self.f1(x)
       # print(x.shape)         torch.Size([64, 10])
        return x
model = Net()
#loss
criterion = torch.nn.CrossEntropyLoss()
#带冲量
optimzer = optim.SGD(model.parameters(),lr=0.01,momentum = 0.5)
#训练
def train(epoch):
    running_loss =0.0
    for batch_idx,data in enumerate(train_loader):
        inputs,target = data
        optimzer.zero_grad()
        outputs = model(inputs)
        #print(outputs.data)
        loss = criterion(outputs,target)
        loss.backward()
        optimzer.step()
        running_loss += loss.item()
        if batch_idx%300==299:
            print('[%d,%5d] loss:%.3f' % (epoch+1,batch_idx+1,running_loss/300))
            running_loss = 0.0
y_data=[]
def test():
    correct = 0
    total = 0
    with torch.no_grad():
        for batch_idx,data in enumerate(test_loader):
            images,labels = data
            outputs = model(images)
            _,predicted = torch.max(outputs.data,dim=1)
            total += labels.size(0)
            correct += (predicted==labels).sum().item()
    
    print('Accuracy on test set:%d %%' % (100*correct/total)) 
    y_data.append(correct/total)
if __name__== '__main__':
    x_data = []
    for epoch in range(7):
        x_data.append(epoch)
        train(epoch)
        test()
    #保存网络参数
    plt.plot(x_data, y_data, ls="-.", lw=2, c="c", label="plot figure")
    plt.xlabel('num')
    plt.ylabel('ACC')
    plt.grid()#网格
    plt.show()

训练结果：

精度从97%提升到了98%

4.使用GPU训练

===将模型扔到gpu

#模型移动到gpu
device = torch.device("cuda:0"if torch.cuda.is_available() else "cpu")
print(device)
model.to(device)

===将数据扔到gpu

#将inputs和target扔到gpu
        inputs,target = inputs.to(device),target.to(device)

import torch
from torchvision import transforms
from torchvision import datasets
from torch.utils.data import DataLoader
import torch.nn.functional as  F
import torch.optim as optim
from matplotlib import pyplot as plt
import os
import sys

batch_size = 64
transform = transforms.Compose(
    [
        transforms.ToTensor(),
        transforms.Normalize((0.1307,),(0.3081,))
    ]
)
train_dataset =datasets.MNIST(root='../dataset/mnist',
                                train = True,
                                download = True,
                                transform = transform)
train_loader = DataLoader(train_dataset,
                        shuffle=True,
                        batch_size=batch_size)
test_dataset =datasets.MNIST(root='../dataset/mnist',
                                train = False,
                                download = True,
                                transform = transform)
test_loader = DataLoader(test_dataset,
                        shuffle = False,
                        batch_size=batch_size)





class  Net(torch.nn.Module):
    def __init__(self):
        super(Net,self).__init__()
        self.conv1 = torch.nn.Conv2d(1,10,kernel_size = 5)
        self.conv2 = torch.nn.Conv2d(10,20,kernel_size = 5)
        self.pooling = torch.nn.MaxPool2d(2)
        self.f1 = torch.nn.Linear(320,10)
    def forward(self,x):
        match_size = x.size(0)
        #print(x.shape)         #torch.Size([64, 1, 28, 28])
        x = self.pooling(F.relu(self.conv1(x)))
        #print(x.shape)         torch.Size([64, 10, 12, 12])
        x = self.pooling(F.relu(self.conv2(x)))
        #print(x.shape)         torch.Size([64, 20, 4, 4])
        x = x.view(match_size,-1)
        #print(x.shape)         #torch.Size([64, 320])
        x = self.f1(x)
       # print(x.shape)         torch.Size([64, 10])
        return x
model = Net()
#模型移动到gpu
device = torch.device("cuda:0"if torch.cuda.is_available() else "cpu")
print(device)
model.to(device)
#loss
criterion = torch.nn.CrossEntropyLoss()
#带冲量
optimzer = optim.SGD(model.parameters(),lr=0.01,momentum = 0.5)
#训练
def train(epoch):
    running_loss =0.0
    for batch_idx,data in enumerate(train_loader):
        inputs,target = data
        #将inputs和target扔到gpu
        inputs,target = inputs.to(device),target.to(device)
        optimzer.zero_grad()
        outputs = model(inputs)
        #print(outputs.data)
        loss = criterion(outputs,target)
        loss.backward()
        optimzer.step()
        running_loss += loss.item()
        if batch_idx%300==299:
            print('[%d,%5d] loss:%.3f' % (epoch+1,batch_idx+1,running_loss/300))
            running_loss = 0.0
y_data=[]
def test():
    correct = 0
    total = 0
    with torch.no_grad():
        for batch_idx,data in enumerate(test_loader):
            images,labels = data
            images,labels = images.to(device),labels.to(device)
            outputs = model(images)
            _,predicted = torch.max(outputs.data,dim=1)
            total += labels.size(0)
            correct += (predicted==labels).sum().item()
    
    print('Accuracy on test set:%d %%' % (100*correct/total)) 
    y_data.append(correct/total)
if __name__== '__main__':
    x_data = []
    for epoch in range(7):
        x_data.append(epoch)
        train(epoch)
        test()
    #保存网络参数
    plt.plot(x_data, y_data, ls="-.", lw=2, c="c", label="plot figure")
    plt.xlabel('num')
    plt.ylabel('ACC')
    plt.grid()#网格
    plt.show()