Ubuntu18.04如何安装Pytorch

安装过程如下：

1. 安装Python

本文采用安装Anaconda的方法来安装Python，这种方法方便，安装包下载速度也快，不用漫长的等待，更重要的是很多三方包已经给集成了。
首先下载anaconda3镜像，到清华大学开源软件镜像站下载地址： https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/ 选择时间最近的86_64Linux版本下载:
cd到你的下载目录，在terminal中输入以下命令：

sh Anaconda3-2019.10-Linux-x86_64.sh

按下回车开始安装。
注意：如果下载下来的anaconda文件名是红色的，则需要转换成可执行的模式：

chmod +x Anaconda3-2019.10-Linux-x86_64.sh

安装好后，就可以进Python了，试一下：
python 安装成功，版本为3.7.4。

2. 安装pytorch和torchvision

安装完Python后，再安装pytorch和torchvision就很方便了，到网站https://pytorch.org/，根据自己的软硬件配置选择对应的版本，复制下方生成的命令到terminal，我设置的pip安装源镜像，所以选择pip安装，下载很快，安装很顺利！！！
在terminal输入：

pip3 install torch torchvision

OK，torch和torchvision安装成功后，在terminal打开Spyder，将以下代码复制进去，然后运行（开始要下载MNIST数据集）：

import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
import torch.backends.cudnn as cudnn
from torchvision import datasets, transforms


class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(1, 10, kernel_size=5)
        self.conv2 = nn.Conv2d(10, 20, kernel_size=5)
        self.conv2_drop = nn.Dropout2d()
        self.fc1 = nn.Linear(320, 50)
        self.fc2 = nn.Linear(50, 10)

    def forward(self, x):
        x = F.relu(F.max_pool2d(self.conv1(x), 2))
        x = F.relu(F.max_pool2d(self.conv2_drop(self.conv2(x)), 2))
        x = x.view(-1, 320)
        x = F.relu(self.fc1(x))
        x = F.dropout(x, training=self.training)
        x = self.fc2(x)
        return F.log_softmax(x, dim=1)


def train(model, device, train_loader, optimizer, epoch):
    model.train()
    for batch_idx, (data, target) in enumerate(train_loader):
        data, target = data.to(device), target.to(device)
        optimizer.zero_grad()
        output = model(data)
        loss = F.nll_loss(output, target)
        loss.backward()
        optimizer.step()
        if batch_idx % 10 == 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 main():
    cudnn.benchmark = True
    torch.manual_seed(1)
    device = torch.device("cuda")
    kwargs = {'num_workers': 1, 'pin_memory': True}
    train_loader = torch.utils.data.DataLoader(
        datasets.MNIST('../data', train=True, download=True,
                       transform=transforms.Compose([
                           transforms.ToTensor(),
                           transforms.Normalize((0.1307,), (0.3081,))
                       ])),
        batch_size=64, shuffle=True, **kwargs)

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

    for epoch in range(1, 11):
        train(model, device, train_loader, optimizer, epoch)


if __name__ == '__main__':
    main()

console出现以下信息说明CUDA和CUDNN安装成功，可以正常使用：

Train Epoch: 5 [42240/60000 (70%)]      Loss: 0.034127
Train Epoch: 5 [42880/60000 (71%)]      Loss: 0.232544
Train Epoch: 5 [43520/60000 (72%)]      Loss: 0.104796
Train Epoch: 5 [44160/60000 (74%)]      Loss: 0.204738
Train Epoch: 5 [44800/60000 (75%)]      Loss: 0.135444
Train Epoch: 5 [45440/60000 (76%)]      Loss: 0.232273
Train Epoch: 5 [46080/60000 (77%)]      Loss: 0.182398
Train Epoch: 5 [46720/60000 (78%)]      Loss: 0.277472
Train Epoch: 5 [47360/60000 (79%)]      Loss: 0.224203
Train Epoch: 5 [48000/60000 (80%)]      Loss: 0.161002
Train Epoch: 5 [48640/60000 (81%)]      Loss: 0.156276
Train Epoch: 5 [49280/60000 (82%)]      Loss: 0.097120
Train Epoch: 5 [49920/60000 (83%)]      Loss: 0.339819
Train Epoch: 5 [50560/60000 (84%)]      Loss: 0.243519
Train Epoch: 5 [51200/60000 (85%)]      Loss: 0.146566
Train Epoch: 5 [51840/60000 (86%)]      Loss: 0.166110
Train Epoch: 5 [52480/60000 (87%)]      Loss: 0.252697
Train Epoch: 5 [53120/60000 (88%)]      Loss: 0.324323
Train Epoch: 5 [53760/60000 (90%)]      Loss: 0.380634
Train Epoch: 5 [54400/60000 (91%)]      Loss: 0.169504
Train Epoch: 5 [55040/60000 (92%)]      Loss: 0.233693