1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20

import torch
from torch import nn
from d2l import torch as d2l

def corr2d(X, K):  #@save
    """计算二维互相关运算"""
    h, w = K.shape
    #定义输出的形状
    Y = torch.zeros((X.shape[0] - h + 1, X.shape[1] - w + 1))
    #遍历每次滑动
    for i in range(Y.shape[0]):
        for j in range(Y.shape[1]):
            Y[i, j] = (X[i:i + h, j:j + w] * K).sum()
    return Y

# 输入
X = torch.tensor([[0.0, 1.0, 2.0], [3.0, 4.0, 5.0], [6.0, 7.0, 8.0]])
# 卷积核
K = torch.tensor([[0.0, 1.0], [2.0, 3.0]])
corr2d(X, K)

1
2
3
4
5
6
7
8

class Conv2D(nn.Module):
    def __init__(self, kernel_size):
        super().__init__()
        self.weight = nn.Parameter(torch.rand(kernel_size))
        self.bias = nn.Parameter(torch.zeros(1))

    def forward(self, x):
        return corr2d(x, self.weight) + self.bias

1
2
3
4
5
6
7
8
9

# 输入数据：下图左
X = torch.ones((6, 8))
X[:, 2:6] = 0

# 卷积核
K = torch.tensor([[1.0, -1.0]])

# 卷积计算：下图右
Y = corr2d(X, K)

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21

# 构造一个二维卷积层，它具有1个输出通道和形状为（1，2）的卷积核
conv2d = nn.Conv2d(1,1, kernel_size=(1, 2), bias=False)

# 这个二维卷积层使用四维输入和输出格式（批量大小、通道、高度、宽度），
# 其中批量大小和通道数都为1
X = X.reshape((1, 1, 6, 8)) #示例输入
Y = Y.reshape((1, 1, 6, 7)) #示例输出
lr = 3e-2  # 学习率

for i in range(10):
    Y_hat = conv2d(X)
    l = (Y_hat - Y) ** 2 #预测的平方损失
    conv2d.zero_grad()
    l.sum().backward()
    # 迭代卷积核
    conv2d.weight.data[:] -= lr * conv2d.weight.grad
    if (i + 1) % 2 == 0:
        print(f'epoch {i+1}, loss {l.sum():.3f}')
        
conv2d.weight.data.reshape((1, 2))
# tensor([[ 0.9938, -0.9841]])

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18

import torch
from torch import nn

# 为了方便起见，我们定义了一个计算卷积层的函数。
# 此函数初始化卷积层权重，并对输入和输出提高和缩减相应的维数
def comp_conv2d(conv2d, X):
    # 这里的（1，1）表示批量大小和通道数都是1
    X = X.reshape((1, 1) + X.shape)  #(1, 1) + X.shape 长度为4的tuple
    Y = conv2d(X)
    # 省略前两个维度：批量大小和通道
    return Y.reshape(Y.shape[2:])

# 请注意，这里每边都填充了1行或1列，因此总共添加了2行或2列
conv2d = nn.Conv2d(1, 1, kernel_size=3, padding=1)

X = torch.rand(size=(8, 8))
comp_conv2d(conv2d, X).shape
# torch.Size([8, 8])

1
2
3

conv2d = nn.Conv2d(1, 1, kernel_size=3, padding=1, stride=2)
comp_conv2d(conv2d, X).shape
# torch.Size([4, 4])

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14

import torch
from d2l import torch as d2l

# X表示多通道输入
# K表示多通道卷积核
def corr2d_multi_in(X, K):
    # 先遍历“X”和“K”的第0个维度（通道维度），再把它们加在一起
    return sum(d2l.corr2d(x, k) for x, k in zip(X, K))

X = torch.tensor([[[0.0, 1.0, 2.0], [3.0, 4.0, 5.0], [6.0, 7.0, 8.0]],
               [[1.0, 2.0, 3.0], [4.0, 5.0, 6.0], [7.0, 8.0, 9.0]]])
K = torch.tensor([[[0.0, 1.0], [2.0, 3.0]], [[1.0, 2.0], [3.0, 4.0]]])

corr2d_multi_in(X, K)

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15

# X表示多通道输入
# K表示多通道卷积核【4个维度】
def corr2d_multi_in_out(X, K):
    # 迭代“K”的第0个维度，每次都对输入“X”执行互相关运算。
    # 最后将所有结果都叠加在一起(沿新的第一维度（dim=0）堆叠起来)
    return torch.stack([corr2d_multi_in(X, k) for k in K], 0) #k三个维度，参考4.1

# 之前的K是三维的多个(等于对应通道数)卷积核
K = torch.stack((K, K + 1, K + 2), 0)
K.shape
# torch.Size([3, 2, 2, 2])
# i=2; o=3

corr2d_multi_in_out(X, K).shape
# torch.Size([3, 2, 2])

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18

#从全连接层的角度，实现1×1卷积
def corr2d_multi_in_out_1x1(X, K):
    c_i, h, w = X.shape
    c_o = K.shape[0]
    X = X.reshape((c_i, h * w)) #每个通道拉平为一个向量，此时X为一个矩阵
    K = K.reshape((c_o, c_i))
    # 全连接层中的矩阵乘法
    Y = torch.matmul(K, X)  #(c_o, h * w)
    return Y.reshape((c_o, h, w))

X = torch.normal(0, 1, (3, 3, 3)) #3个输入通道
K = torch.normal(0, 1, (2, 3, 1, 1)) #2×3个卷积核

Y1 = corr2d_multi_in_out_1x1(X, K)

#按常规卷积层角度的实现
Y2 = corr2d_multi_in_out(X, K)
assert float(torch.abs(Y1 - Y2).sum()) < 1e-6

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18

import torch
from torch import nn
from d2l import torch as d2l

def pool2d(X, pool_size, mode='max'):
    p_h, p_w = pool_size
    Y = torch.zeros((X.shape[0] - p_h + 1, X.shape[1] - p_w + 1))
    for i in range(Y.shape[0]):
        for j in range(Y.shape[1]):
            if mode == 'max':
                Y[i, j] = X[i: i + p_h, j: j + p_w].max()
            elif mode == 'avg':
                Y[i, j] = X[i: i + p_h, j: j + p_w].mean()
    return Y

X = torch.tensor([[0.0, 1.0, 2.0], [3.0, 4.0, 5.0], [6.0, 7.0, 8.0]])
pool2d(X, (2, 2))
pool2d(X, (2, 2), 'avg')

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12

X = torch.arange(16, dtype=torch.float32).reshape((1, 1, 4, 4)) #转换为4维的输入
X

pool2d = nn.MaxPool2d(3)
pool2d(X) #只返回一个值

# 其它灵活设置
pool2d = nn.MaxPool2d(3, padding=1, stride=2)
pool2d(X) 

pool2d = nn.MaxPool2d((2, 3), stride=(2, 3), padding=(0, 1))
pool2d(X)

1
2
3
4
5
6

# 将原有X的第二个维度上，增加一组数据；即两个输入通道
X = torch.cat((X, X + 1), 1)
X

pool2d = nn.MaxPool2d(3, padding=1, stride=2)
pool2d(X) #汇聚后输出通道的数量仍然是2。

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13

import torch
from torch import nn
from d2l import torch as d2l

net = nn.Sequential(
    nn.Conv2d(1, 6, kernel_size=5, padding=2), nn.Sigmoid(),
    nn.AvgPool2d(kernel_size=2, stride=2),
    nn.Conv2d(6, 16, kernel_size=5), nn.Sigmoid(),
    nn.AvgPool2d(kernel_size=2, stride=2),
    nn.Flatten(),
    nn.Linear(16 * 5 * 5, 120), nn.Sigmoid(),
    nn.Linear(120, 84), nn.Sigmoid(),
    nn.Linear(84, 10))

1
2
3
4

X = torch.rand(size=(1, 1, 28, 28), dtype=torch.float32)
for layer in net:
    X = layer(X)
    print(layer.__class__.__name__,'output shape: \t',X.shape)

1
2

batch_size = 256
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size=batch_size)

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18

def evaluate_accuracy_gpu(net, data_iter, device=None): #@save
    """使用GPU计算模型在数据集上的精度"""
    if isinstance(net, nn.Module):
        net.eval()  # 设置为评估模式
        if not device:
            device = next(iter(net.parameters())).device #与模型的device保持一致
    # 正确预测的数量，总预测的数量
    metric = d2l.Accumulator(2)
    with torch.no_grad():
        for X, y in data_iter:
            if isinstance(X, list): #判断是否为list
                # BERT微调所需的（之后将介绍）
                X = [x.to(device) for x in X]
            else:
                X = X.to(device)
            y = y.to(device)
            metric.add(d2l.accuracy(net(X), y), y.numel())
    return metric[0] / metric[1]

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43

#@save
def train_ch6(net, train_iter, test_iter, num_epochs, lr, device):
    """用GPU训练模型(在第六章定义)"""
    def init_weights(m):  #模型参数初始化
        if type(m) == nn.Linear or type(m) == nn.Conv2d:
            nn.init.xavier_uniform_(m.weight)
    net.apply(init_weights)
    print('training on', device)
    net.to(device)
    optimizer = torch.optim.SGD(net.parameters(), lr=lr)
    loss = nn.CrossEntropyLoss()
    animator = d2l.Animator(xlabel='epoch', xlim=[1, num_epochs],
                            legend=['train loss', 'train acc', 'test acc'])
    timer, num_batches = d2l.Timer(), len(train_iter)
    for epoch in range(num_epochs):
        # 训练损失之和，训练准确率之和，样本数
        metric = d2l.Accumulator(3)
        # 训练模型
        net.train()
        for i, (X, y) in enumerate(train_iter):
            timer.start()
            optimizer.zero_grad()
            X, y = X.to(device), y.to(device)
            y_hat = net(X)
            l = loss(y_hat, y)
            l.backward()
            optimizer.step()
            with torch.no_grad():
                metric.add(l * X.shape[0], d2l.accuracy(y_hat, y), X.shape[0])
            # 下述代码用于可视化
            timer.stop()
            train_l = metric[0] / metric[2]
            train_acc = metric[1] / metric[2]
            if (i + 1) % (num_batches // 5) == 0 or i == num_batches - 1:
                animator.add(epoch + (i + 1) / num_batches,
                             (train_l, train_acc, None))
        #每次epoch后的测试集评价
        test_acc = evaluate_accuracy_gpu(net, test_iter)
        animator.add(epoch + 1, (None, None, test_acc))
    print(f'loss {train_l:.3f}, train acc {train_acc:.3f}, '
          f'test acc {test_acc:.3f}')
    print(f'{metric[2] * num_epochs / timer.sum():.1f} examples/sec '
          f'on {str(device)}')

1
2

lr, num_epochs = 0.9, 10
train_ch6(net, train_iter, test_iter, num_epochs, lr, d2l.try_gpu())