pytorch实现mnist分类的示例讲解_Python

torchvision包包含了目前流行的数据集，模型结构和常用的图片转换工具。

torchvision.datasets中包含了以下数据集

MNIST
COCO（用于图像标注和目标检测）(Captioning and Detection)
LSUN Classification
ImageFolder
Imagenet-12
CIFAR10 and CIFAR100
STL10

torchvision.models

torchvision.models模块的子模块中包含以下模型结构。
AlexNet
VGG
ResNet
SqueezeNet
DenseNet You can construct a model with random weights by calling its constructor:

pytorch torchvision transform

对PIL.Image进行变换

100

101

102

103

104

105

106

107

108

109

									from __future__ import print_function

									import argparse #Python 命令行解析工具

									import torch 

									import torch.nn as nn

									import torch.nn.functional as F

									import torch.optim as optim 

									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(args, 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 % args.log_interval == 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(args, model, device, test_loader):

									  model.eval()

									  test_loss = 0

									  correct = 0

									  with torch.no_grad():

									    for data, target in test_loader:

									      data, target = data.to(device), target.to(device)

									      output = model(data)

									      test_loss += F.nll_loss(output, target, size_average=False).item() # sum up batch loss

									      pred = output.max(1, keepdim=True)[1] # get the index of the max log-probability

									      correct += pred.eq(target.view_as(pred)).sum().item()

									  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)))

									def main():

									  # Training settings

									  parser = argparse.ArgumentParser(description='PyTorch MNIST Example')

									  parser.add_argument('--batch-size', type=int, default=64, metavar='N',

									            help='input batch size for training (default: 64)')

									  parser.add_argument('--test-batch-size', type=int, default=1000, metavar='N',

									            help='input batch size for testing (default: 1000)')

									  parser.add_argument('--epochs', type=int, default=10, metavar='N',

									            help='number of epochs to train (default: 10)')

									  parser.add_argument('--lr', type=float, default=0.01, metavar='LR',

									            help='learning rate (default: 0.01)')

									  parser.add_argument('--momentum', type=float, default=0.5, metavar='M',

									            help='SGD momentum (default: 0.5)')

									  parser.add_argument('--no-cuda', action='store_true', default=False,

									            help='disables CUDA training')

									  parser.add_argument('--seed', type=int, default=1, metavar='S',

									            help='random seed (default: 1)')

									  parser.add_argument('--log-interval', type=int, default=10, metavar='N',

									            help='how many batches to wait before logging training status')

									  args = parser.parse_args()

									  use_cuda = not args.no_cuda and torch.cuda.is_available()

									  torch.manual_seed(args.seed)

									  device = torch.device("cuda" if use_cuda else "cpu")

									  kwargs = {'num_workers': 1, 'pin_memory': True} if use_cuda else {}

									  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=args.batch_size, shuffle=True, **kwargs)

									  test_loader = torch.utils.data.DataLoader(

									    datasets.MNIST('../data', train=False, transform=transforms.Compose([

									              transforms.ToTensor(),

									              transforms.Normalize((0.1307,), (0.3081,))

									            ])),

									    batch_size=args.test_batch_size, shuffle=True, **kwargs)

									  model = Net().to(device)

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

									  for epoch in range(1, args.epochs + 1):

									    train(args, model, device, train_loader, optimizer, epoch)

									    test(args, model, device, test_loader)

									if __name__ == '__main__':

									  main()