### PyTorch 实现手写体数字识别 以下是基于 MNIST 数据集的手写体数字识别代码示例。此代码展示了如何加载数据、定义模型结构以及训练过程。 #### 加载必要的库 ```python import torch import torch.nn as nn import torch.optim as optim import torchvision.datasets as datasets import torchvision.transforms as transforms from torch.utils.data import DataLoader ``` #### 定义超参数 ```python batch_size = 64 learning_rate = 0.001 num_epochs = 10 input_size = 28 * 28 # 图像大小为 28x28 像素 hidden_size = 128 # 隐藏层神经元数量 output_size = 10 # 输出类别数 (0 到 9 的数字) ``` #### 设置设备（GPU 或 CPU） ```python device = torch.device("cuda" if torch.cuda.is_available() else "cpu") # 使用 GPU 如果可用[^2] print(f"Using device: {device}") ``` #### 下载并预处理 MNIST 数据集 ```python transform = transforms.Compose([ transforms.ToTensor(), # 将图像转换为张量[^1] transforms.Normalize((0.5,), (0.5,)) # 归一化输入数据[^3] ]) train_dataset = datasets.MNIST(root='./data', train=True, transform=transform, download=True) test_dataset = datasets.MNIST(root='./data', train=False, transform=transform) train_loader = DataLoader(dataset=train_dataset, batch_size=batch_size, shuffle=True) test_loader = DataLoader(dataset=test_dataset, batch_size=batch_size, shuffle=False) ``` #### 构建简单的全连接网络模型 ```python class NeuralNet(nn.Module): def __init__(self, input_size, hidden_size, output_size): super(NeuralNet, self).__init__() self.fc1 = nn.Linear(input_size, hidden_size) # 输入层 -> 隐藏层 self.relu = nn.ReLU() self.fc2 = nn.Linear(hidden_size, output_size) # 隐藏层 -> 输出层 def forward(self, x): out = self.fc1(x) out = self.relu(out) out = self.fc2(out) return out model = NeuralNet(input_size, hidden_size, output_size).to(device) # 移动模型到指定设备 ``` #### 定义损失函数和优化器 ```python criterion = nn.CrossEntropyLoss() # 多分类交叉熵损失函数 optimizer = optim.Adam(model.parameters(), lr=learning_rate) # Adam 优化器 ``` #### 训练模型 ```python total_step = len(train_loader) for epoch in range(num_epochs): model.train() for i, (images, labels) in enumerate(train_loader): images = images.reshape(-1, 28*28).to(device) # 展平图像数据 labels = labels.to(device) outputs = model(images) # 前向传播 loss = criterion(outputs, labels) # 计算损失 optimizer.zero_grad() # 清除梯度 loss.backward() # 反向传播 optimizer.step() # 更新权重 if (i+1) % 100 == 0: print(f'Epoch [{epoch+1}/{num_epochs}], Step [{i+1}/{total_step}], Loss: {loss.item():.4f}') ``` #### 测试模型性能 ```python model.eval() # 进入评估模式 with torch.no_grad(): correct = 0 total = 0 for images, labels in test_loader: images = images.reshape(-1, 28*28).to(device) labels = labels.to(device) outputs = model(images) _, predicted = torch.max(outputs.data, 1) total += labels.size(0) correct += (predicted == labels).sum().item() accuracy = 100 * correct / total print(f'Test Accuracy of the model on the 10000 test images: {accuracy:.2f}%') ``` --- ####