Pytorch之反向传播

神经网络中，最基本的优化方式就是反向传播 本文介绍了Pytorch中一个简单的例子：

import torch
import torch.nn as nn
from torch.nn import MaxPool2d, ReLU, Sigmoid, Linear, Conv2d, Flatten, Sequential
import torchvision
from torch.utils.data import DataLoader
from torch.utils.tensorboard import SummaryWriter as Sum

root = r'C:\Users\15715\Desktop\Programs\Python\深度学习\pythonProject\dataset_CIFAR10'
dataset = torchvision.datasets.CIFAR10(root=root, train=False, transform=torchvision.transforms.ToTensor(), download=True)
dataloader = DataLoader(dataset, batch_size=1)


class Standard(nn.Module):
    def __init__(self):
        super(Standard, self).__init__()
        # self.conv1 = Conv2d(3, 32, 5, padding=2)  # 第一个卷积层  in通道，out通道，kernel，padding stride(default=1)
        # self.maxpool1 = MaxPool2d(2)  # 第一个池化层 kernel
        # self.conv2 = Conv2d(32, 32, 5, padding=2)
        # self.maxpool2 = MaxPool2d(2)
        # self.conv3 = Conv2d(32, 64, 5, padding=2)
        # self.maxpool3 = MaxPool2d(2)
        # self.flatten = Flatten()  # 展平
        # # 实质有2个线性层
        # self.linear1 = Linear(1024, 64)
        # self.linear2 = Linear(64, 10)
        # 上述被替代为了下面的Sequential

        self.model1 = Sequential(
            Conv2d(3, 32, 5, padding=2),
            MaxPool2d(2),
            Conv2d(32, 32, 5, padding=2),
            MaxPool2d(2),
            Conv2d(32, 64, 5, padding=2),
            MaxPool2d(2),
            Flatten(),
            Linear(1024, 64),
            Linear(64, 10)
        )

    def forward(self, x):
        # x = self.conv1(x)
        # x = self.maxpool1(x)
        # x = self.conv2(x)
        # x = self.maxpool2(x)
        # x = self.conv3(x)
        # x = self.maxpool3(x)
        # x = self.flatten(x)
        # x = self.linear1(x)
        # x = self.linear2(x)
        x = self.model1(x)
        return x


mc = Standard()
# print(mc)
loss = nn.L1Loss()
optim = torch.optim.SGD(mc.parameters(), lr=0.01)  # parameters 学习速率


epoch = 10
for x in range(epoch):
    # 学习一轮
    runloss = 0.0
    for data in dataloader:
        img, target = data   # output是实际输出，target是理想输出，即真实标签值
        output = mc(img)
        # print(output)
        # print(target)
        res = loss(output,target)
        # print(res)
        # 反向传播
        optim.zero_grad() # 将之前的梯度置为0,避免影响本次的优化
        res.backward()  # 获取梯度
        optim.step()  # 优化
        # print(res)
        runloss = runloss + res
    print('1 epch done')
    # 宏观上查看误差的减小
    print(runloss)
    print('')

运行结果为：

1 epch done
tensor(25982.2188, grad_fn=<AddBackward0>)
1 epch done
tensor(23803.2637, grad_fn=<AddBackward0>)
1 epch done
tensor(23197.8145, grad_fn=<AddBackward0>)
1 epch done
tensor(22575.0508, grad_fn=<AddBackward0>)

备注：数据集来源于官网(cifar10)
模型就是cifar10的模型