3.4.1 implement from scratch
3.4.1.1 obtain and read data
batch_size = 256
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size)
3.4.1.2 initialization parameter model
Mxnet
num_inputs = 784
num_outputs = 10
W = nd.random.normal(scale=0.01, shape=(num_inputs, num_outputs))
b = nd.zeros(num_outputs)
W.attach_grad()
b.attach_grad()
Pytorch
num_inputs = 784
num_outputs = 10
W = torch.tensor(np.random.normal(0, 0.01, (num_inputs, num_outputs)), dtype=torch.float)
b = torch.zeros(num_outputs, dtype=torch.float)
W.requires_grad_(requires_grad=True)
b.requires_grad_(requires_grad=True)
3.4.1.3 realize softmax operation
def softmax(X):
X_exp = X.exp()
partition = X_exp.sum(axis=1, keepdims=True)
return X_exp / partition # 这里应用了广播机制
3.4.1.4 Definition Model
Mxnet
def net(X):
return softmax(nd.dot(X.reshape((-1, num_inputs)), W) + b)
Pytorch
def net(X):
return softmax(torch.mm(X.view((-1, num_inputs)), W) + b)
3.4.1.5 define the loss function
Mxnet
def cross_entropy(y_hat, y):
return -nd.pick(y_hat, y).log()
Pytorch
def cross_entropy(y_hat, y):
return - torch.log(y_hat.gather(1, y.view(-1, 1)))
3.4.1.6 calculate classification accuracy
Mxnet
def accuracy(y_hat, y):
return (y_hat.argmax(axis=1) == y.astype('float32')).mean().asscalar()
# 本函数已保存在d2lzh包中方便以后使用。该函数将被逐步改进：它的完整实现将在“图像增广”一节中
# 描述
def evaluate_accuracy(data_iter, net):
acc_sum, n = 0.0, 0
for X, y in data_iter:
y = y.astype('float32')
acc_sum += (net(X).argmax(axis=1) == y).sum().asscalar()
n += y.size
return acc_sum / n
Pytorch
def accuracy(y_hat, y):
return (y_hat.argmax(dim=1) == y).float().mean().item()
# 本函数已保存在d2lzh_pytorch包中方便以后使用。该函数将被逐步改进：它的完整实现将在“图像增广”一节中描述
def evaluate_accuracy(data_iter, net):
acc_sum, n = 0.0, 0
for X, y in data_iter:
acc_sum += (net(X).argmax(dim=1) == y).float().sum().item()
n += y.shape[0]
return acc_sum / n
3.4.1.7 training model
Mxnet
num_epochs, lr = 5, 0.1
# 本函数已保存在d2lzh包中方便以后使用
def train_ch3(net, train_iter, test_iter, loss, num_epochs, batch_size,
params=None, lr=None, trainer=None):
for epoch in range(num_epochs):
train_l_sum, train_acc_sum, n = 0.0, 0.0, 0
for X, y in train_iter:
with autograd.record():
y_hat = net(X)
l = loss(y_hat, y).sum()
l.backward()
if trainer is None:
d2l.sgd(params, lr, batch_size)
else:
trainer.step(batch_size) # “softmax回归的简洁实现”一节将用到
y = y.astype('float32')
train_l_sum += l.asscalar()
train_acc_sum += (y_hat.argmax(axis=1) == y).sum().asscalar()
n += y.size
test_acc = evaluate_accuracy(test_iter, net)
print('epoch %d, loss %.4f, train acc %.3f, test acc %.3f'
% (epoch + 1, train_l_sum / n, train_acc_sum / n, test_acc))
train_ch3(net, train_iter, test_iter, cross_entropy, num_epochs, batch_size,
[W, b], lr)
Pytorch
num_epochs, lr = 5, 0.1
# 本函数已保存在d2lzh_pytorch包中方便以后使用
def train_ch3(net, train_iter, test_iter, loss, num_epochs, batch_size,
params=None, lr=None, optimizer=None):
for epoch in range(num_epochs):
train_l_sum, train_acc_sum, n = 0.0, 0.0, 0
for X, y in train_iter:
y_hat = net(X)
l = loss(y_hat, y).sum()
# 梯度清零
if optimizer is not None:
optimizer.zero_grad()
elif params is not None and params[0].grad is not None:
for param in params:
param.grad.data.zero_()
l.backward()
if optimizer is None:
d2l.sgd(params, lr, batch_size)
else:
optimizer.step() # “softmax回归的简洁实现”一节将用到
train_l_sum += l.item()
train_acc_sum += (y_hat.argmax(dim=1) == y).sum().item()
n += y.shape[0]
test_acc = evaluate_accuracy(test_iter, net)
print('epoch %d, loss %.4f, train acc %.3f, test acc %.3f'
% (epoch + 1, train_l_sum / n, train_acc_sum / n, test_acc))
train_ch3(net, train_iter, test_iter, cross_entropy, num_epochs, batch_size, [W, b], lr)
3.4.2 simple realization
3.4.2.1 obtain and read data
batch_size = 256
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size)
3.4.2.2 Definition and initialization model
Mxnet
net = nn.Sequential()
net.add(nn.Dense(10))
net.initialize(init.Normal(sigma=0.01))
Pytorch
num_inputs = 784
num_outputs = 10
class LinearNet(nn.Module):
def __init__(self, num_inputs, num_outputs):
super(LinearNet, self).__init__()
self.linear = nn.Linear(num_inputs, num_outputs)
def forward(self, x): # x 的形状: (batch, 1, 28, 28)
y = self.linear(x.view(x.shape[0], -1))
return y
# net = LinearNet(num_inputs, num_outputs)
class FlattenLayer(nn.Module):
def __init__(self):
super(FlattenLayer, self).__init__()
def forward(self, x): # x 的形状: (batch, *, *, ...)
return x.view(x.shape[0], -1)
from collections import OrderedDict
net = nn.Sequential(
# FlattenLayer(),
# LinearNet(num_inputs, num_outputs)
OrderedDict([
('flatten', FlattenLayer()),
('linear', nn.Linear(num_inputs, num_outputs))]) # 或者写成我们自己定义的 LinearNet(num_inputs, num_outputs) 也可以
)
# 初始化模型
init.normal_(net.linear.weight, mean=0, std=0.01)
init.constant_(net.linear.bias, val=0)
3.4.2.3 softmax and cross-entropy loss function
Mxnet
loss = gloss.SoftmaxCrossEntropyLoss()
Pytorch
loss = nn.CrossEntropyLoss()
3.4.2.4 define optimization algorithm
Mxnet
trainer = gluon.Trainer(net.collect_params(), 'sgd', {'learning_rate': 0.1})
Pytorch
optimizer = torch.optim.SGD(net.parameters(), lr=0.1)
3.4.2.5 training model
num_epochs = 5
d2l.train_ch3(net, train_iter, test_iter, loss, num_epochs, batch_size, None,
None, trainer)
