Recurrent neural network (RNN)

import torch
import torch.nn as nn
import torchvision
import torchvision.transforms as transforms

# 配置GPU或CPU设置
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')

# 超参数设置
sequence_length = 28
input_size = 28
hidden_size = 128
num_layers = 2
num_classes = 10
batch_size = 100
num_epochs = 2
learning_rate = 0.01

# MNIST dataset
train_dataset = torchvision.datasets.MNIST(root='./data/',
                                           train=True,
                                           transform=transforms.ToTensor(),# 将PIL Image或者 ndarray 转换为tensor，并且归一化至[0-1]，归一化至[0-1]是直接除以255
                                           download=True)

test_dataset = torchvision.datasets.MNIST(root='./data/',
                                          train=False,
                                          transform=transforms.ToTensor())# 将PIL Image或者 ndarray 转换为tensor，并且归一化至[0-1]，归一化至[0-1]是直接除以255

# 训练数据加载，按照batch_size大小加载，并随机打乱
train_loader = torch.utils.data.DataLoader(dataset=train_dataset,
                                           batch_size=batch_size,
                                           shuffle=True)
# 测试数据加载，按照batch_size大小加载
test_loader = torch.utils.data.DataLoader(dataset=test_dataset,
                                          batch_size=batch_size,
                                          shuffle=False)


# Recurrent neural network (many-to-one) 多对一
class RNN(nn.Module):
    def __init__(self, input_size, hidden_size, num_layers, num_classes):
        super(RNN, self).__init__() # 继承 __init__ 功能
        self.hidden_size = hidden_size
        self.num_layers = num_layers
        self.lstm = nn.LSTM(input_size, hidden_size, num_layers, batch_first=True) # if use nn.RNN(), it hardly learns  LSTM 效果要比 nn.RNN() 好多了
        self.fc = nn.Linear(hidden_size, num_classes)

    def forward(self, x):
        # Set initial hidden and cell states
        h0 = torch.zeros(self.num_layers, x.size(0), self.hidden_size).to(device)
        c0 = torch.zeros(self.num_layers, x.size(0), self.hidden_size).to(device)

        # Forward propagate LSTM
        out, _ = self.lstm(x, (h0, c0))  # out: tensor of shape (batch_size, seq_length, hidden_size)

        # Decode the hidden state of the last time step
        out = self.fc(out[:, -1, :])
        return out


model = RNN(input_size, hidden_size, num_layers, num_classes).to(device)
print(model)
# RNN((lstm): LSTM(28, 128, num_layers=2, batch_first=True)
#     (fc): Linear(in_features=128, out_features=10, bias=True))

# 损失函数与优化器设置
# 损失函数
criterion = nn.CrossEntropyLoss()
# 优化器设置 ，并传入RNN模型参数和相应的学习率
optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)

# 训练模型
total_step = len(train_loader)
for epoch in range(num_epochs):
    for i, (images, labels) in enumerate(train_loader):
        images = images.reshape(-1, sequence_length, input_size).to(device)
        labels = labels.to(device)

        # 前向传播
        outputs = model(images)
        # 计算损失 loss
        loss = criterion(outputs, labels)

        # 反向传播与优化
        # 清空上一步的残余更新参数值
        optimizer.zero_grad()
        # 反向传播
        loss.backward()
        # 将参数更新值施加到RNN model的parameters上
        optimizer.step()
        # 每迭代一定步骤，打印结果值
        if (i + 1) % 100 == 0:
            print ('Epoch [{}/{}], Step [{}/{}], Loss: {:.4f}'
                   .format(epoch + 1, num_epochs, i + 1, total_step, loss.item()))

# 测试模型
with torch.no_grad():
    correct = 0
    total = 0
    for images, labels in test_loader:
        images = images.reshape(-1, sequence_length, input_size).to(device)
        labels = labels.to(device)
        outputs = model(images)
        _, predicted = torch.max(outputs.data, 1)
        total += labels.size(0)
        correct += (predicted == labels).sum().item()

    print('Test Accuracy of the model on the 10000 test images: {} %'.format(100 * correct / total))

# 保存已经训练好的模型
# Save the model checkpoint
torch.save(model.state_dict(), 'model.ckpt')