1.nn.Module
a.__init__,
b.farward
nn.Module定义了__call__方法,及Lr的实例,能够直接被传入参数调用,实际上用的是forward方法并传入参数
from torch import nn
class Lr(nn.Module):
def __init__(self):
super(Lr,self).__init_()
self.linear = nn.Linear(1,1)
def forward(self,x):
out = self.linear(x)
return out
nn.Linear(input的特征数量,输出特征数量)
全连接层
2.optimizer:
a. torch.optim.SGD(参数,学习率)
b.torch.optim.adam(参数,学习率)
1.参数可以使用model.parameters()来获取,获取模型中所有requires_grad = True的参数
例:
optimizer = optim.SGD(model.parameters(),lr = 1e-3)
optimizer.zerograd()#梯度置为0
loss.backward()#计算梯度
optimizer.step()#更新参数的值
3.损失函数
a.nn.MSELoss():均方误差,常用语分类问题
b.nn.CrossEntropyLoss():交叉熵损失,常用语逻辑回归
例:
model = Lr()
criterion = nn.MSELoss()
optimizer = optim.SGD(model.parameters(),lr=1e-3)
for i in range(100):
y_predict = model(x_true)
loss = criterion(y_true,y_predict)
optimizer.zerograd()#梯度置为0
loss.backward()#计算梯度
optimizer.step()#更新参数的值
4.torchvision.transforms
4.1torchvision.transforms.ToTensor
例:
from torchvision import transforms
import numpy as np
data = np.random.randint(0,255,size=12)
img = data.reshape(2,2,3)
print(img.shape)
img_tensor = transforms.ToTensoe()(img)
pring(img_tensor)
print(img_tensor.shape)
img.permute(2,0,1)#按照列进行交换
4.2torchvision.transforms.Normalize(mean,std)
image_Normalized = (image-mean)/std
例子:
from torchvision import transforms
import numpy as np
from torchvision
data = np.random.randint(0,255,size=12)
img = data.reshape(2,2,3)
print(img.shape)
img_tensor = transforms.ToTensoe()(img)
print(img)
norm_img = transforms.Normalize((10,10,10),(1,1,1))(img)#形状与通道数相同
4.3torchvision.transforms.Compose(transforms)
将多个transform组合起来使用
例如
transform.Compose([
torchvision.transforms.ToTensoe(),#先转化为Tensor
torchvision.transforms.Normalize(mean,std)#再进行正则化
])
5.模型的保存和加载
5.1模型保存
torch.save(mnist_net.state_dict(),"model/mnist_net")#保存模型参数
torch.save(optimizer.state_dict(),'results/mnist_optimizer')#保存优化器参数
5.2模型加载
model.load_state_dict(torch.load('./model/model.pkl'))
optimizer.load_state_dict(torch.load('./model/optimizer.pkl'))
6.word embedding API
torch.nn.Embedding(num_embeddings,embedding_dim)
参数介绍:
1.num_embeddings:词典的大小
2.embedding_dim:embedding的维度