Pytorch实战学习（一）：用Pytorch实现线性回归

《PyTorch深度学习实践》完结合集_哔哩哔哩_bilibili

P5--用Pytorch实现线性回归

建立模型四大步骤

一、Prepare dataset

mini-batch：x、y必须是矩阵

## Prepare Dataset:mini-batch, X、Y是3X1的Tensor
x_data = torch.Tensor([[1.0], [2.0], [3.0]])
y_data = torch.Tensor([[2.0], [4.0], [6.0]])

二、Design model

1、重点是构造计算图

##Design Model

##构造类，继承torch.nn.Module类
class LinearModel(torch.nn.Module):
    ## 构造函数，初始化对象
    def __init__(self):
        ##super调用父类
        super(LinearModel, self).__init__()
        ##构造对象，Linear Unite，包含两个Tensor：weight和bias，参数(1, 1)是w的维度
        self.linear = torch.nn.Linear(1, 1)

    ## 构造函数，前馈运算
    def forward(self, x):
        ## w*x+b
        y_pred = self.linear(x)
        return y_pred

model = LinearModel()

2、设置w的维度，后一层的神经元数量 X 前一层神经元数量

三、Construct Loss and Optimizer

##Construct Loss and Optimizer

##损失函数，传入y和y_presd
criterion = torch.nn.MSELoss(size_average = False)

##优化器，model.parameters()找出模型所有的参数，Lr--学习率
optimizer = torch.optim.SGD(model.parameters(), lr=0.01)

1、损失函数

2、优化器

可用不同的优化器进行测试对比

四、Training cycle

## Training cycle

for epoch in range(100):
    y_pred = model(x_data)
    loss = criterion(y_pred, y_data)
    print(epoch, loss)

    ##梯度归零
    optimizer.zero_grad()
    ##反向传播
    loss.backward()
    ##更新
    optimizer.step()

完整代码

import torch

## Prepare Dataset:mini-batch, X、Y是3X1的Tensor
x_data = torch.Tensor([[1.0], [2.0], [3.0]])
y_data = torch.Tensor([[2.0], [4.0], [6.0]])

##Design Model

##构造类，继承torch.nn.Module类
class LinearModel(torch.nn.Module):
    ## 构造函数，初始化对象
    def __init__(self):
        ##super调用父类
        super(LinearModel, self).__init__()
        ##构造对象，Linear Unite，包含两个Tensor：weight和bias，参数(1, 1)是w的维度
        self.linear = torch.nn.Linear(1, 1)

    ## 构造函数，前馈运算
    def forward(self, x):
        ## w*x+b
        y_pred = self.linear(x)
        return y_pred

model = LinearModel()

##Construct Loss and Optimizer

##损失函数，传入y和y_presd
criterion = torch.nn.MSELoss(size_average = False)

##优化器，model.parameters()找出模型所有的参数，Lr--学习率
optimizer = torch.optim.SGD(model.parameters(), lr=0.01)

## Training cycle

for epoch in range(100):
    y_pred = model(x_data)
    loss = criterion(y_pred, y_data)
    print(epoch, loss)

    ##梯度归零
    optimizer.zero_grad()
    ##反向传播
    loss.backward()
    ##更新
    optimizer.step()

## Outpue weigh and bias
print('w = ', model.linear.weight.item())
print('b = ', model.linear.bias.item())

## Test Model
x_test = torch.Tensor([[4.0]])
y_test = model(x_test)
print('y_pred = ', y_test.data)

运行结果

训练100次后，得到的 weight and bias，还有预测的y