使用Chainer構建和訓練一個簡單的神經網絡模型

下面是一個使用Chainer構建和訓練一個簡單的神經網絡模型的例子：

import numpy as np
import chainer
import chainer.functions as F
import chainer.links as L

# 創建一個簡單的神經網絡模型
class SimpleNN(chainer.Chain):
    def __init__(self):
        super(SimpleNN, self).__init__()
        with self.init_scope():
            self.fc1 = L.Linear(None, 100)
            self.fc2 = L.Linear(100, 10)

    def __call__(self, x):
        h = F.relu(self.fc1(x))
        return self.fc2(h)

# 準備數據
train, test = chainer.datasets.get_mnist()
train_iter = chainer.iterators.SerialIterator(train, batch_size=100, shuffle=True)
test_iter = chainer.iterators.SerialIterator(test, batch_size=100, repeat=False, shuffle=False)

# 初始化模型和優化器
model = SimpleNN()
optimizer = chainer.optimizers.SGD(lr=0.01)
optimizer.setup(model)

# 訓練模型
for epoch in range(10):
    for i, batch in enumerate(train_iter):
        x, t = chainer.dataset.concat_examples(batch)
        y = model(x)
        loss = F.softmax_cross_entropy(y, t)
        
        model.cleargrads()
        loss.backward()
        optimizer.update()
        
        if i % 100 == 0:
            print(f'epoch: {epoch}, iteration: {i}, loss: {loss.data}')

# 評估模型
test_losses = []
test_accuracies = []
for batch in test_iter:
    x, t = chainer.dataset.concat_examples(batch)
    y = model(x)
    loss = F.softmax_cross_entropy(y, t)
    accuracy = F.accuracy(y, t)
    
    test_losses.append(loss.data)
    test_accuracies.append(accuracy.data)

mean_test_loss = np.mean(test_losses)
mean_test_accuracy = np.mean(test_accuracies)

print(f'Mean test loss: {mean_test_loss}, Mean test accuracy: {mean_test_accuracy}')

在這個例子中，我們創建了一個簡單的包含兩個全連接層的神經網絡模型，然后使用MNIST數據集訓練和評估這個模型。我們使用SGD優化器和交叉熵損失函數來訓練模型，并計算測試集上的平均損失和準確率來評估模型的性能。