Pytorch第三章线性神经网络线性回归实现

1.导包

import numpy as np
import torch
from torch.utils import data

2.生成1000个数据

#生成数据
def synthetic_data(w,b,num_examples):
    X=torch.normal(0,1,(num_examples,len(w)))
    y=torch.matmul(X,w)+b
    y+=torch.normal(0,0.01,y.shape)
    return X,y.reshape((-1,1))

true_w=torch.tensor([2,-3.4])
true_b=4.2
features,labels=synthetic_data(true_w,true_b,1000)

3.写一个pytorch迭代器

def load_array(data_array,batch_size,is_train=True):
    data_set=data.TensorDataset(*data_array)
    return data.DataLoader(data_set,batch_size,shuffle=is_train)

4.使用迭代器取数据

batch_size=10
data_iter=load_array((features,labels),batch_size)

next(iter(data_iter))

5.导入线性回归模型

from torch import nn
net=nn.Sequential(nn.Linear(2,1))

6.设置权重正态分布参数

net[0].weight.data.normal_(0,0.01)
net[0].bias.data.fill_(0)

7.均方误差

loss=nn.MSELoss()

8.设置梯度下降方式

trainer=torch.optim.SGD(net.parameters(),lr=0.03)

9.训练

num_epoch=3
for epoch in range(num_epoch):
    for X,y in data_iter:
        l=loss(net(X),y)
        trainer.zero_grad()
        l.backward()
        trainer.step()
    l=loss(net(features),labels)
    print(f'epoch {epoch+1},loss {l:f}')

10.真实值和预测值之间的误差

w=net[0].weight.data
b=net[0].bias.data
print(true_w-w.reshape(true_w.shape))
print(true_b-b)