深度学习

import torch
# prepare dataset
# x,y是矩阵，3行1列 也就是说总共有3个数据，每个数据只有1个特征
x_data = torch.tensor([[1.0], [2.0], [3.0]])
y_data = torch.tensor([[2.0], [4.0], [6.0]])
 
#design model using class
"""
our model class should be inherit from nn.Module, which is base class for all neural network modules.
member methods __init__() and forward() have to be implemented
class nn.linear contain two member Tensors: weight and bias
class nn.Linear has implemented the magic method __call__(),which enable the instance of the class can
be called just like a function.Normally the forward() will be called
"""
class LinearModel(torch.nn.Module):
    def __init__(self):
        super(LinearModel, self).__init__()
        # (1,1)是指输入x和输出y的特征维度，这里数据集中的x和y的特征都是1维的
        # 该线性层需要学习的参数是w和b  获取w/b的方式分别是~linear.weight/linear.bias
        self.linear = torch.nn.Linear(1, 1)
 
    def forward(self, x):
        y_pred = self.linear(x)
        return y_pred
 
model = LinearModel()
 
# construct loss and optimizer
# criterion = torch.nn.MSELoss(size_average = False)
criterion = torch.nn.MSELoss(reduction = 'sum')
optimizer = torch.optim.SGD(model.parameters(), lr = 0.01) # model.parameters()自动完成参数的初始化操作，这个地方我可能理解错了
 
# training cycle forward, backward, update
for epoch in range(100):
    y_pred = model(x_data) # forward:predict
    loss = criterion(y_pred, y_data) # forward: loss
    print(epoch, loss.item())
 
    optimizer.zero_grad() # the grad computer by .backward() will be accumulated. so before backward, remember set the grad to zero
    loss.backward() # backward: autograd，自动计算梯度
    optimizer.step() # update 参数，即更新w和b的值
 
print('w = ', model.linear.weight.item())
print('b = ', model.linear.bias.item())
 
x_test = torch.tensor([[4.0]])
y_test = model(x_test)
print('y_pred = ', y_test.data)
import torch
# prepare dataset
# x,y是矩阵，3行1列 也就是说总共有3个数据，每个数据只有1个特征
x_data = torch.tensor([[1.0], [2.0], [3.0]])
y_data = torch.tensor([[2.0], [4.0], [6.0]])
 
#design model using class
"""
our model class should be inherit from nn.Module, which is base class for all neural network modules.
member methods __init__() and forward() have to be implemented
class nn.linear contain two member Tensors: weight and bias
class nn.Linear has implemented the magic method __call__(),which enable the instance of the class can
be called just like a function.Normally the forward() will be called
"""
class LinearModel(torch.nn.Module):
    def __init__(self):
        super(LinearModel, self).__init__()
        # (1,1)是指输入x和输出y的特征维度，这里数据集中的x和y的特征都是1维的
        # 该线性层需要学习的参数是w和b  获取w/b的方式分别是~linear.weight/linear.bias
        self.linear = torch.nn.Linear(1, 1)
 
    def forward(self, x):
        y_pred = self.linear(x)
        return y_pred
 
model = LinearModel()
 
# construct loss and optimizer
# criterion = torch.nn.MSELoss(size_average = False)
criterion = torch.nn.MSELoss(reduction = 'sum')
optimizer = torch.optim.SGD(model.parameters(), lr = 0.01) # model.parameters()自动完成参数的初始化操作，这个地方我可能理解错了
 
# training cycle forward, backward, update
for epoch in range(100):
    y_pred = model(x_data) # forward:predict
    loss = criterion(y_pred, y_data) # forward: loss
    print(epoch, loss.item())
 
    optimizer.zero_grad() # the grad computer by .backward() will be accumulated. so before backward, remember set the grad to zero
    loss.backward() # backward: autograd，自动计算梯度
    optimizer.step() # update 参数，即更新w和b的值
 
print('w = ', model.linear.weight.item())
print('b = ', model.linear.bias.item())
 
x_test = torch.tensor([[4.0]])
y_test = model(x_test)
print('y_pred = ', y_test.data)
————————————————
版权声明：本文为CSDN博主「错错莫」的原创文章，遵循CC 4.0 BY-SA版权协议，转载请附上原文出处链接及本声明。
原文链接：https://blog.csdn.net/bit452/article/details/109677086