Pytorch cifar10_tutorial.py问题BrokenPipeError: [Errno 32] Broken pipe【附完整代码】

问题

在使用windows环境来使用 pytorch 的 cifar10_tutorial 时肯定会遇到如下的error：

......
......
......
RuntimeError:
        An attempt has been made to start a new process before the
        current process has finished its bootstrapping phase.
        This probably means that you are not using fork to start your
        child processes and you have forgotten to use the proper idiom
        in the main module:
            if __name__ == '__main__':
                freeze_support()
                ...
        The "freeze_support()" line can be omitted if the program
        is not going to be frozen to produce an executable.

--------------------- 
......
......
......
BrokenPipeError: [Errno 32] Broken pipe

分析

问题出现的原因大概是，windows的multiprocessing library会不停地产生子进程（child process），如果用main保护起来，子进程在运行时因为会导入这个main，（只有一个main进程），所以避免了循环生成子进程。

参考：https://blog.csdn.net/tanmx219/article/details/86129623

解决方法

方法一

按照提示，在主要模块前加入

if __name__ == '__main__':

将主要模块都包含在这行代码之下，就像C语言中的 int main() 一样。根据pytorch官网的代码重新排列之后可在windows下直接运行。

附代码及较详细注释如下：

import torch
import torchvision
import torchvision.transforms as transforms
import matplotlib.pyplot as plt
import numpy as np
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim

# functions to show an image

def imshow(img):
    img = img / 2 + 0.5     # unnormalize, transform them to Tensors of range[0,1]
    npimg = img.numpy()     # transform img  into numpy array
    # process the image and show the format, because the numpy array's format is [d, h, w].
    # So transpose the array's format to [h, w, d].
    plt.imshow(np.transpose(npimg, (1, 2, 0)))
    plt.show()     # show the image


class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(3, 6, 5)
        self.pool = nn.MaxPool2d(2, 2)
        self.conv2 = nn.Conv2d(6, 16, 5)
        self.fc1 = nn.Linear(16 * 5 * 5, 120)   # full connection layer
        self.fc2 = nn.Linear(120, 84)
        self.fc3 = nn.Linear(84, 10)

    def forward(self, x):
        x = self.pool(F.relu(self.conv1(x)))
        x = self.pool(F.relu(self.conv2(x)))
        x = x.view(-1, 16 * 5 * 5)   # x.size = 16*5*5(d*h*w), transformed into Tensor of [1 * 400]
        x = F.relu(self.fc1(x))
        x = F.relu(self.fc2(x))
        x = self.fc3(x)
        return x

net = Net()
# print(net)


if __name__ == '__main__':

    # The output of torchvision datasets are PILImage images of range [0, 1].
    # We transform them to Tensors of normalized range [-1, 1].
    transform = transforms.Compose(
        [transforms.ToTensor(),
         transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])

    trainset = torchvision.datasets.CIFAR10(root='./data', train=True,
                                            download=True, transform=transform)
    trainloader = torch.utils.data.DataLoader(trainset, batch_size=4,
                                              shuffle=True, num_workers=2)

    testset = torchvision.datasets.CIFAR10(root='./data', train=False,
                                           download=True, transform=transform)
    testloader = torch.utils.data.DataLoader(testset, batch_size=4,
                                             shuffle=False, num_workers=2)

    classes = ('plane', 'car', 'bird', 'cat',
               'deer', 'dog', 'frog', 'horse', 'ship', 'truck')


    # get some random training images
    dataiter = iter(trainloader)
    images, labels = dataiter.next()

    # show images
    imshow(torchvision.utils.make_grid(images))
    # print labels
    print(' '.join('%5s' % classes[labels[j]] for j in range(4)))


    criterion = nn.CrossEntropyLoss()
    optimizer = optim.SGD(net.parameters(), lr=0.001, momentum=0.9)


    for epoch in range(2):   # loop over the dataset multiple times

        running_loss = 0.0
        for i, data in enumerate(trainloader, 0):  # Traversing an Iterative Object with starting from 0
            # get the inputs
            inputs, labels = data

            # zero the parameter gradients
            optimizer.zero_grad()

            # forward + backward + optimize
            outputs = net(inputs)
            loss = criterion(outputs, labels)
            loss.backward()
            optimizer.step()

            # print statistics
            running_loss += loss.item()  # get a Python number from a tensor containing a single value
            if i % 2000 == 1999:    # print every 2000 mini-batches
                print('[%d, %5d] loss: %.3f' %
                    (epoch + 1, i + 1, running_loss / 2000))
                running_loss = 0.0

    print('Finished Training')


    dataiter = iter(testloader)
    images, labels = dataiter.next()

    # print images
    imshow(torchvision.utils.make_grid(images))
    print('GroundTruth: ', ' '.join('%5s' % classes[labels[j]] for j in range(4)))


    outputs = net(images)


    _, predicted = torch.max(outputs, 1)  # 对 outputs 每行( 1 表示以行为单位 )取最大值，并将返回的索引值赋给 predicted.

    print('Predicted: ', ' '.join('%5s' % classes[predicted[j]]
                                for j in range(4)))


    correct = 0
    total = 0
    with torch.no_grad():   # 表示不需要计算梯度
        for data in testloader:
            images, labels = data
            outputs = net(images)
            _, predicted = torch.max(outputs.data, 1)
            total += labels.size(0)
            correct += (predicted == labels).sum().item()  # sum() 返回两个数组中相等的个数，再用item()返回tensor中的number

    print('Accuracy of the network on the 10000 test images: %d %%' % (
        100 * correct / total))


    class_correct = list(0. for i in range(10))  # 创建大小为10的全0数组
    class_total = list(0. for i in range(10))
    with torch.no_grad():
        for data in testloader:
            images, labels = data
            outputs = net(images)
            _, predicted = torch.max(outputs, 1)
            c = (predicted == labels).squeeze()   # squeeze()在这里没有作用
            for i in range(4):
                label = labels[i]
                class_correct[label] += c[i].item()  # item()之后还是bool值，但是做运算时会使用0,1来替代false，true
                class_total[label] += 1


    for i in range(10):
        print('Accuracy of %5s : %2d %%' % (
            classes[i], 100 * class_correct[i] / class_total[i]))

方法二

将调用 torch.utils.data.DataLoader 函数中的参数中的 num_workers 设为0。
附相关代码如下：

trainloader = torch.utils.data.DataLoader(trainset, batch_size=4,
                                          shuffle=True, num_workers=0)
testloader = torch.utils.data.DataLoader(testset, batch_size=4,
                                         shuffle=False, num_workers=0)

参考：https://blog.csdn.net/qq_33666011/article/details/81873217