CNN for CIFAR10 in Pytorch

CNN for CIFAR10 in Pytorch

Code Structure

1. 分为两个源代码文件，Neural_Network_Model.py中定义网络模型，CNN_for_CIFAR10.py中定义训练和测试过程。
2. Neural_Network_Model.py中定义网络模型的网络模型如下：
   
3. CNN_for_CIFAR10.py中定义训练和测试过程如下：
   • batchsize = 64，每64张图片计算一次梯度并更新一次参数，每更新100次参数输出一次在当前训练batch上的Loss。
   • 每一个训练epoch过后，将模型在测试数据集计算Loss和accuracy_rate。
   • 上述计算结果用tensorboard展示。
   • 每10个epoch保存一次模型。
4. 在工作目录下，创建Tensorboard的工作目录(代码中是logs)，存储event file，在Terminal中使用*tensorboard --logdir=‘logs’*在浏览器中查看Loss、accuracy_rate的可视化展示。
5. 在工作目录下，创建用于保存数据集的文件夹(代码中是dataset)，CIFAR10数据集无需单独下载。
6. pytorch配置了cuda就使用GPU训练，默认使用CPU训练。

Neural_Network_Model.py

from torch import nn
import torch

class Neural_Network(nn.Module):
    def __init__(self):
        super(Neural_Network, self).__init__()
        self.model = nn.Sequential(
            nn.Conv2d(3, 32, 5, 1, 2),
            nn.MaxPool2d(2),
            nn.Conv2d(32, 32, 5, 1, 2),
            nn.MaxPool2d(2),
            nn.Conv2d(32, 64, 5, 1, 2),
            nn.MaxPool2d(2),
            nn.Flatten(),
            nn.Linear(64 * 4 * 4, 64),
            nn.Linear(64, 10)
        )
    def forward(self, x):
        x = self.model(x)
        return x

# 测试
if __name__ == '__main__':
    neural_network = Neural_Network()
    input = torch.ones((64, 3, 32, 32))
    output = neural_network(input)
    print(output.shape)

CNN_for_CIFAR10.py

import torch.optim
import torchvision
from torch.utils.data import DataLoader
from torch import nn
from Neural_Network_model import *
from torch.utils.tensorboard import SummaryWriter

# 定义训练设备
#device = torch.device("cpu")
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
print(device)

# 准备数据集
train_data = torchvision.datasets.CIFAR10(root="./dataset", train=True, download=True,
                                          transform=torchvision.transforms.ToTensor())
test_data = torchvision.datasets.CIFAR10(root="./dataset", train=False, download=True,
                                          transform=torchvision.transforms.ToTensor())

# length
train_data_size = len(train_data)
test_data_size = len(test_data)
# 如果train_data_size = 10，训练数据集的长度为：10
print("训练数据集的长度为：{}".format(train_data_size))
print("测试数据集的长度为：{}".format(test_data_size))

# 利用DataLoader加载数据集
train_dataloader = DataLoader(train_data, batch_size=64)
test_dataloader = DataLoader(test_data, batch_size=64)

# 搭建神经网络,放到一个单独的文件 Neural_Network_model.py

# 创建网络模型
neural_network = Neural_Network()
# 利用GPU训练
neural_network = neural_network.to(device)

# 创建损失函数
loss_function = nn.CrossEntropyLoss()
# 利用GPU训练
loss_function = loss_function.to(device)

# 优化器
# learning_rate = 0.01
learning_rate = 1e-2
optimizer = torch.optim.SGD(neural_network.parameters(),lr=learning_rate)

# 设置训练网络的一些参数
total_train_step = 0     # 记录训练的次数
total_test_step = 0      # 记录测试的次数
epoch = 50               # 记录训练的轮数

# 添加tensorboard
writer = SummaryWriter("logs")

for i in range(epoch):
    print("-----------第 {} 轮训练开始----------".format(i+1))

    # 训练步骤开始
    neural_network.train()
    for data in train_dataloader:
        imgs, targets = data
        # 利用GPU训练
        imgs = imgs.to(device)
        targets = targets.to(device)

        outputs = neural_network(imgs)
        loss = loss_function(outputs, targets)
        # 优化器优化模型
        optimizer.zero_grad()
        loss.backward()
        optimizer.step()

        total_train_step = total_train_step + 1
        if total_train_step % 100 == 0:
            print("训练次数：{}，loss：{}".format(total_train_step, loss.item()))
            writer.add_scalar("train_loss", loss.item(), total_train_step)
    # 测试步骤开始
    neural_network.eval()
    total_test_loss = 0
    total_accuracy = 0
    with torch.no_grad():
        for data in test_dataloader:
            imgs, targets = data
            # 利用GPU训练
            imgs = imgs.to(device)
            targets = targets.to(device)

            outputs = neural_network(imgs)
            print_output_test = outputs
            loss = loss_function(outputs, targets)
            total_test_loss = total_test_loss + loss.item()
            accuracy = (outputs.argmax(1) == targets).sum()
            total_accuracy = total_accuracy + accuracy
    print("整体测试集上的loss：{}".format(total_test_loss))
    print("整体测试集上的accuracy_rate：{}".format(total_accuracy / test_data_size))
    writer.add_scalar("test_loss", total_test_loss, total_test_step)
    writer.add_scalar("test_accuracy_rate", total_accuracy / test_data_size, total_test_step)
    total_test_step = total_test_step + 1
    if(i == 10):
        torch.save(neural_network, "Neural_Network_for_CIFAR10{}.pth".format(i))
#       torch.save(neural_network.state_dict(), "Neural_Network_{}.pth".format(i))
        print("模型已保存")

writer.close()