利用GPU训练

        利用GPU来训练一般来说训练的速度比CPU要快的多，并且添加GPU也并不复杂，添加cuda()即可。

在下面模块添加GPU操作语句：
1.神经网络模型

#搭建神经网络
class Gu(nn.Module):
    def __init__(self):
        super(Gu  , self).__init__()
        self.model1 = Sequential(
            Conv2d(3, 32, 5, 1, 2),
            MaxPool2d(2),
            Conv2d(32, 32, 5, 1, 2),
            MaxPool2d(2),
            Conv2d(32, 64, 5, 1, 2),
            MaxPool2d(2),
            Flatten(),
            Linear(64*4*4, 64),
            Linear(64, 10)
        )

    def forward(self,x):
        x = self.model1(x)
        return x

gu = Gu()
#添加gu=gu.cuda()
if torch.cuda.is_available():
    gu=gu.cuda()

这里添加if torch.cuda.is_available()语句保证程序可以运行，后面各处添加cuda()语句前都添加了这里的if语句

如果你电脑有GPU则优先利用GPU训练，如果没有自动选择CPU训练

如果没有这个if语句，电脑上没有GPU，运行时会报出你没有GPU无法训练

2.损失函数

#损失函数
loss_fn = nn.CrossEntropyLoss()
if torch.cuda.is_available():
    loss_fn = loss_fn.cuda()

3.训练及测试的数据

 #训练
    gu.train()
    for data in train_dataloader:
        imgs,targets = data
        if torch.cuda.is_available():
            imgs = imgs.cuda()
            targets = targets.cuda()
        outputs = gu(imgs)
        loss = loss_fn(outputs,targets)

 #测试
    gu.eval()
    total_test_loss = 0
    total_accuracy = 0
    with torch.no_grad():
        for data in test_dataloader:
            imgs, targets = data
            if torch.cuda.is_available():
                imgs = imgs.cuda()
                targets = targets.cuda()
            outputs = gu(imgs)
            loss = loss_fn(outputs, targets)
            total_test_loss = total_test_step + loss.item()
            accuracy = (outputs.argmax(1) == targets).sum()
            total_accuracy = total_accuracy + accuracy

GPU训练时长：                                                  CPU训练时长：

第一百次时GPU用时2秒多，CPU用时4秒多 

可以看出GPU的训练速度是明显快于CPU的，在训练一些数据集较大的项目时还是可以省下很多时间的

 完整代码：

import time

import torch
import torchvision.datasets
from torch import nn
from torch.nn import Sequential, Conv2d, MaxPool2d, Flatten, Linear

#数据集
from torch.utils.data import DataLoader
from torch.utils.tensorboard import SummaryWriter

train_data = torchvision.datasets.CIFAR10(root="../data",train=True,transform=torchvision.transforms.ToTensor(),download=True)
test_data = torchvision.datasets.CIFAR10(root="../data",train=False,transform=torchvision.transforms.ToTensor(),download=True)

#获取数据集长度
train_data_size = len(train_data)
test_data_size = len(test_data)
print("训练数据集的长度为：{}".format(train_data_size))
print("测试数据集的长度为：{}".format(test_data_size))

#利用DataLoader加载数据集
train_dataloader = DataLoader(train_data, batch_size=64)
test_dataloader = DataLoader(test_data, batch_size=64)

#搭建神经网络
class Gu(nn.Module):
    def __init__(self):
        super(Gu  , self).__init__()
        self.model1 = Sequential(
            Conv2d(3, 32, 5, 1, 2),
            MaxPool2d(2),
            Conv2d(32, 32, 5, 1, 2),
            MaxPool2d(2),
            Conv2d(32, 64, 5, 1, 2),
            MaxPool2d(2),
            Flatten(),
            Linear(64*4*4, 64),
            Linear(64, 10)
        )

    def forward(self,x):
        x = self.model1(x)
        return x

gu = Gu()
#添加gu=gu.cuda()
if torch.cuda.is_available():
    gu=gu.cuda()

#损失函数
loss_fn = nn.CrossEntropyLoss()
if torch.cuda.is_available():
    loss_fn = loss_fn.cuda()

#优化器
learnnig_rate = 1e-2
optimizer = torch.optim.SGD(gu.parameters(),lr= learnnig_rate)

#设置参数
total_train_step = 0
total_test_step = 0
epoch = 10

#tensorboard
writer = SummaryWriter("../logs_train")

start_time = time.time()
for i in range(epoch):
    print("___第{}轮训练开始___".format(i+1))

    #训练
    gu.train()
    for data in train_dataloader:
        imgs,targets = data
        if torch.cuda.is_available():
            imgs = imgs.cuda()
            targets = targets.cuda()
        outputs = gu(imgs)
        loss = loss_fn(outputs,targets)

        optimizer.zero_grad()
        loss.backward()
        optimizer.step()

        total_train_step = total_train_step + 1
        if total_train_step % 100 == 0:
            end_time = time.time()
            print(end_time - start_time)
            print("训练次数： {},Loss： {}".format(total_train_step,loss.item()))
            writer.add_scalar("trian_loss",loss.item(),total_train_step)

    #测试
    gu.eval()
    total_test_loss = 0
    total_accuracy = 0
    with torch.no_grad():
        for data in test_dataloader:
            imgs, targets = data
            if torch.cuda.is_available():
                imgs = imgs.cuda()
                targets = targets.cuda()
            outputs = gu(imgs)
            loss = loss_fn(outputs, targets)
            total_test_loss = total_test_step + loss.item()
            accuracy = (outputs.argmax(1) == targets).sum()
            total_accuracy = total_accuracy + accuracy

    print("整体测试集上的Loss：{}".format(total_test_loss))
    print("整体测试集上的正确率：{}".format(total_accuracy/test_data_size))
    writer.add_scalar("test_loss",total_test_loss,total_test_step)
    writer.add_scalar("test_accuracy", total_accuracy/test_data_size, total_test_step)
    total_test_step = total_test_step + 1

    torch.save(gu,"gu_{}.pth".format(i))
    print("模型已保存！")

writer.close()