手把手教你用Pytorch训练图像分类网络

import torchvision
from torch import nn
from torch.utils.data import DataLoader
import torch

data_set = torchvision.datasets.CIFAR10(train=False,
                                        root='./dataset/',
                                        download=True,
                                        transform=torchvision.transforms.Compose([
                                            torchvision.transforms.ToTensor(),
                                        ]))
data_loader = DataLoader(dataset=data_set, batch_size=128)


class Model(nn.Module):
    def __init__(self):
        super(Model, self).__init__()
        self.sequential = nn.Sequential(
            nn.Conv2d(in_channels=3, out_channels=32,
                      kernel_size=5, padding=2),
            nn.MaxPool2d(kernel_size=2),
            nn.Conv2d(in_channels=32, out_channels=32,
                      kernel_size=5, padding=2),
            nn.MaxPool2d(kernel_size=2),
            nn.Conv2d(in_channels=32, out_channels=64,
                      kernel_size=5, padding=2),
            nn.MaxPool2d(kernel_size=2),
            nn.Flatten(),
            nn.Linear(in_features=1024, out_features=64),
            nn.Linear(in_features=64, out_features=10)
        )

    def forward(self, input):
        output = self.sequential(input)
        return output


model = Model()

loss = nn.CrossEntropyLoss()
optim = torch.optim.SGD(params=model.parameters(), lr=0.01)

for epoch in range(20):
    epoch_loss = 0.0
    for data in data_loader:
        imgs, targets = data
        output = model(imgs)
        res_loss = loss(output, targets)
        # step 1 梯度清0
        optim.zero_grad()
        # step 2 计算梯度
        res_loss.backward()  # 反向传播
        # step 3 梯度更新
        optim.step()

        epoch_loss += res_loss

    print(epoch_loss)