cifar10数据集 模型参数

### CIFAR-10 数据集模型参数设置及配置

#### 加载 CIFAR-10 数据集
为了加载 CIFAR-10 数据集，可以使用 `torchvision.datasets.CIFAR10` 类。此方法允许指定多个参数以便于灵活控制数据集的加载方式[^1]。

from torchvision import datasets, transforms

transform = transforms.Compose([
    transforms.ToTensor(),
])

train_dataset = datasets.CIFAR10(root='./data', train=True, download=True, transform=transform)
test_dataset = datasets.CIFAR10(root='./data', train=False, download=True, transform=transform)

#### 定义模型架构
对于 CIFAR-10 的模型构建，通常会设计一个多层卷积神经网络 (CNN)，该网络能够有效地提取图像特征并进行分类任务。下面是一个简单的 CNN 架构示例：

import torch.nn as nn
import torch.nn.functional as F

class SimpleCNN(nn.Module):
    def __init__(self):
        super(SimpleCNN, self).__init__()
        self.conv1 = nn.Conv2d(in_channels=3, out_channels=6, kernel_size=5)
        self.pool = nn.MaxPool2d(kernel_size=2, stride=2)
        self.conv2 = nn.Conv2d(in_channels=6, out_channels=16, kernel_size=5)
        self.fc1 = nn.Linear(16 * 5 * 5, 120)
        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 = F.relu(self.fc1(x))
        x = F.relu(self.fc2(x))
        x = self.fc3(x)
        return x

#### 训练与验证过程
训练过程中，通过定义好的 `train()` 函数执行前向传播、计算损失值、反向传播更新权重等操作；而验证阶段则由 `validate()` 函数负责评估当前模型性能[^2]。

def train(model, device, train_loader, optimizer, criterion, epoch):
    model.train()
    running_loss = 0.0
    for batch_idx, (inputs, targets) in enumerate(train_loader):
        inputs, targets = inputs.to(device), targets.to(device)
        
        outputs = model(inputs)
        loss = criterion(outputs, targets)
        
        optimizer.zero_grad()
        loss.backward()
        optimizer.step()

        running_loss += loss.item()
    
    print(f'Epoch {epoch}, Loss: {running_loss / len(train_loader)}')

def validate(model, device, val_loader, criterion):
    model.eval()
    test_loss = 0
    correct = 0
    
    with torch.no_grad():
        for data, target in val_loader:
            data, target = data.to(device), target.to(device)
            
            output = model(data)
            _, predicted = torch.max(output.data, 1)
            test_loss += criterion(output, target).item()
            correct += (predicted == target).sum().item()
    
    accuracy = 100. * correct / len(val_loader.dataset)
    print(f'\nValidation set: Average loss: {test_loss:.4f}, Accuracy: ({accuracy:.0f}%)\n')

#### 模型保存机制
每隔若干轮次之后自动保存模型参数有助于防止意外中断造成的数据丢失，并方便后续继续训练或部署应用。这可通过设定周期性的检查点来实现，在特定条件下触发保存动作。

if not os.path.exists('./checkpoints'):
    os.makedirs('./checkpoints')

for epoch in range(start_epoch, start_epoch + num_epochs):
    ...
    if epoch % save_interval == 0:
        checkpoint_path = f"./checkpoints/model_{epoch}.pth"
        torch.save({
            'epoch': epoch,
            'model_state_dict': net.state_dict(),
            'optimizer_state_dict': optimizer.state_dict(),
            'loss': loss,
        }, checkpoint_path)

#### 可视化工具集成
为了更好地理解模型内部运作情况及其表现趋势，可借助 TensorBoard 进行日志记录和图形展示。具体做法是在初始化时创建一个 `SummaryWriter` 对象，并调用相应的方法将图表信息写入到指定目录下，便于后期查看分析[^4]。

from tensorboardX import SummaryWriter

writer = SummaryWriter('runs/cifar_experiment_1')
input_sample = torch.randn((batch_size, 3, 32, 32)).to(device)
writer.add_graph(net, input_sample)
writer.close()