cnn

cnn_dropout_bn/data/cifar-10-batches-py/readme.html

@@ -0,0 +1 @@
+<meta HTTP-EQUIV="REFRESH" content="0; url=http://www.cs.toronto.edu/~kriz/cifar.html">

cnn_dropout_bn/lenet.py

@@ -0,0 +1,67 @@
+import torch
+from torch import nn
+
+class LeNet5(nn.Module):
+    def __init__(self, dropout_rate=0.5):
+        super().__init__()
+        #特征提取器 Conv → ReLU → Pool 反复出现
+        #卷积提取局部模式，ReLU 引入非线性，池化降低空间分辨率并扩大有效视野。
+        self.conv1 = nn.Sequential(
+            nn.Conv2d(in_channels=3, 
+                      out_channels=6, 
+                      kernel_size=5, 
+                      stride=1, 
+                      padding=0,
+                      bias=False),
+            nn.BatchNorm2d(6),	# 在激活函数之前使用BN
+            nn.ReLU(),
+            nn.MaxPool2d(kernel_size=2, stride=2, padding=0)
+        )
+        
+        self.conv2 = nn.Sequential(
+            nn.Conv2d(in_channels=6, 
+                      out_channels=16, 
+                      kernel_size=5, 
+                      stride=1, 
+                      padding=0,
+                      bias=False),
+            nn.BatchNorm2d(16),	# 在激活函数之前使用BN
+            nn.ReLU(),
+            nn.MaxPool2d(kernel_size=2, stride=2, padding=0)
+        )
+        
+        self.flatten = nn.Flatten()
+        
+        self.fc1 = nn.Sequential(
+            nn.Linear(400, 120,bias=False),
+            nn.BatchNorm1d(120),	# 在激活函数之前使用BN
+            nn.ReLU(),
+            nn.Dropout(p=dropout_rate)
+        )
+        
+        self.fc2 = nn.Sequential(
+            nn.Linear(120, 84,bias=False),
+            nn.BatchNorm1d(84),	# 在激活函数之前使用BN
+            nn.ReLU(),
+            nn.Dropout(p=dropout_rate)
+        )
+        
+        self.fc3 = nn.Linear(84, 10)
+        
+    def forward(self, x):   
+        out = self.conv1(x)
+        out = self.conv2(out)
+        out = self.flatten(out)
+        out = self.fc1(out)
+        out = self.fc2(out)
+        
+        logits = self.fc3(out)
+  
+        return logits
+
+if __name__ == '__main__':
+    net = LeNet5(dropout_rate=0.5)
+    data = torch.randn(1, 3, 32, 32)
+    output = net(data)
+    print(output)
+    print(output.shape)

cnn_dropout_bn/train.py

@@ -0,0 +1,74 @@
+
+import torch
+import torch.nn as nn
+from torch.utils.data import DataLoader
+from torch.utils.tensorboard import SummaryWriter
+from torchvision import datasets, transforms
+from lenet import LeNet5
+import tqdm
+
+
+# 指定日志目录
+writer = SummaryWriter(log_dir='D:/gitcode/cnn_dropout_bn/logs')
+
+# 1. 判断是否使用CUDA
+device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+print(f'Using device: {device}')
+# 2. 准备数据
+train_set = datasets.CIFAR10(root='D:/gitcode/cnn_dropout_bn/data', train=True, download=True, transform=transforms.ToTensor())
+test_set = datasets.CIFAR10(root='D:/gitcode/cnn_dropout_bn/data', train=False, download=True, transform=transforms.ToTensor())
+train_loader = DataLoader(dataset=train_set, batch_size=128, shuffle=True)
+test_loader = DataLoader(dataset=test_set, batch_size=128, shuffle=False)
+# 3. 创建模型
+model = LeNet5()
+model = model.to(device)
+# 4. 确定损失函数
+loss_fn = nn.CrossEntropyLoss()
+# 5. 创建优化器 使用梯度下降算法 参数的更新
+opt = torch.optim.Adam(model.parameters(),lr=0.0003)
+
+for epoch in range(100):
+    # 6. 训练模型
+    model.train()
+    train_total_loss = 0
+    for images, labels in tqdm.tqdm(train_loader, desc="train", total=len(train_loader)):
+        # 将数据移动到设备
+        images, labels = images.to(device), labels.to(device)
+        outputs = model(images) # 前向传播
+        loss = loss_fn(outputs, labels) # 计算损失
+
+        opt.zero_grad() # 清空梯度
+        loss.backward() # 反向传播 计算梯度
+        opt.step() # 更新参数
+        train_total_loss += loss.item()
+    train_avg_loss = train_total_loss / len(train_loader)
+    print(f'Epoch {epoch+1}, Loss: {train_avg_loss:.4f}')
+    # 7. 测试模型
+    model.eval()# 关闭dropout, batchsize=1, 5
+    #10000
+    #模型在做推理时， BN使用的是整个训练集上均值和方差，而不是当前batch的均值和方差
+    test_total_loss = 0
+    test_total_acc = 0
+    with torch.inference_mode(): 
+        for images, labels in tqdm.tqdm(test_loader, desc="test", total=len(test_loader)):
+            images, labels = images.to(device), labels.to(device)
+            outputs = model(images)
+            # 计算损失
+            loss = loss_fn(outputs, labels)
+            test_total_loss += loss.item()
+
+            pred = torch.argmax(outputs, dim=1)
+            acc = torch.eq(pred, labels).float().mean()
+            test_total_acc += acc.item()
+
+    test_avg_acc = test_total_acc / len(test_loader)
+    print(f'Epoch {epoch+1} Test Accuracy: {test_avg_acc:.4f}')
+
+    test_avg_loss = test_total_loss / len(test_loader)
+    print(f'Epoch {epoch+1} Test Loss: {test_avg_loss:.4f}')
+    # 记录数据
+    writer.add_scalars('Loss/train', {'train_avg_loss': train_avg_loss,
+    'test_avg_loss': test_avg_loss}, epoch)
+    writer.add_scalar('Accuracy/test', test_avg_acc, epoch)
+    # 8. 保存模型
+    torch.save(model.state_dict(), 'D:/gitcode/cnn_dropout_bn/model/lenet5.pth')