从0开始深度学习（10）——softmax的简洁实现

同样的，本章将使用torch自带的API简洁的实现softmax回归

1 读取数据

使用自带的DataLoader

import torch
from torch import nn,optim
import torchvision
from torch.utils import data
from torchvision import transforms,datasets
from torch.utils.data import DataLoader# 定义超参数
batch_size = 256
learning_rate = 0.01
epochs = 5# 数据预处理
transform = transforms.Compose([transforms.ToTensor(),  transforms.Normalize((0.5,), (0.5,))  # 标准化到[-1, 1]区间，加快计算
])# 加载Fashion-MNIST数据集
train_dataset = datasets.FashionMNIST(root='D:/DL_Data/', train=True, download=False, transform=transform)
test_dataset = datasets.FashionMNIST(root='D:/DL_Data/', train=False, download=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)

2 定义模型，初始化参数

使用torch自带的nn模型，输入层用Flatten()，因为要把2828的展开成一维，输出层用Linear，前面我们说过，全连接层可以看作线性模型，也符合softmax的特征，输入是784，因为2828展开后是784，输出是10，因为有10和可能预测到的类别

# 定义模型
net = nn.Sequential(nn.Flatten(),nn.Linear(784,10)
)
# 初始化参数
def init_weights(m):if type(m) == nn.Linear:nn.init.normal_(m.weight, std=0.01)net.apply(init_weights);

3 定义损失函数和优化器

使用torch自带的

# 损失函数与优化器
criterion = nn.CrossEntropyLoss()  # 使用交叉熵损失，因为它包含了softmax
optimizer = optim.SGD(net.parameters(), lr=learning_rate)

4 训练

# 训练模型
for epoch in range(epochs):net.train()running_loss = 0.0running_corrects = 0for batch_idx, (data, target) in enumerate(train_loader):optimizer.zero_grad()output = net(data)loss = criterion(output, target)loss.backward()optimizer.step()# 计算正确率_, preds = torch.max(output, 1)running_loss += loss.item() * data.size(0)running_corrects += torch.sum(preds == target.data)if batch_idx % 10 == 0:# 每训练10步输出一次loss和accepoch_loss = running_loss / ((batch_idx + 1) * batch_size)epoch_acc = running_corrects.double() / ((batch_idx + 1) * batch_size)print(f'Epoch [{epoch+1}/{epochs}], Step [{batch_idx+1}/{len(train_loader)}], Loss: {epoch_loss:.4f}, Acc: {epoch_acc:.4f}')# 输出每个epoch的平均损失和正确率epoch_loss = running_loss / len(train_dataset)epoch_acc = running_corrects.double() / len(train_dataset)print(f'Epoch [{epoch+1}/{epochs}] Summary - Loss: {epoch_loss:.4f}, Acc: {epoch_acc:.4f}')

5 预测

# 定义 Fashion-MNIST 标签的文本描述
def get_fashion_mnist_labels(labels):text_labels = ['t-shirt', 'trouser', 'pullover', 'dress', 'coat','sandal', 'shirt', 'sneaker', 'bag', 'ankle boot']return [text_labels[int(i)] for i in labels]# 预测并显示结果
def predict(net, test_iter, n=6):for X, y in test_iter:break  # 只取一个批次的数据trues = get_fashion_mnist_labels(y)preds = get_fashion_mnist_labels(net(X).argmax(axis=1))titles = [true + '\n' + pred for true, pred in zip(trues, preds)]n = min(n, X.shape[0])fig, axs = plt.subplots(1, n, figsize=(12, 3))for i in range(n):axs[i].imshow(X[i].permute(1, 2, 0).squeeze().numpy(), cmap='gray')axs[i].set_title(titles[i])axs[i].axis('off')plt.show()# 调用预测函数
predict(net, test_iter, n=10)