[人工智能]08线性回归从零开始实现-动手学深度学习

%matplotlib inline
import random
import torch
from d2l import torch as d2l

#制作数据集
def syn_data(w,b,num_examples):
    x = torch.normal(0,1,(num_examples,len(w)))
    y = torch.matmul(x,w)+b
    y=y +torch.normal(0,0.01,y.shape)
    return x,y.reshape((-1,1))
t_w = torch.tensor([2,-3.4])
t_b = 4.2
features,labels = syn_data(t_w,t_b,1000)
features.size(),labels.size()

(torch.Size([1000, 2]), torch.Size([1000, 1]))

#分组的  每10
def data_iter(batch_size,features,labels):
    num_examples = len(features)
    indices =list(range(num_examples))
    #随机读取样本 无顺序
    random.shuffle(indices)#打乱顺序
    for i in range(0,num_examples,batch_size):
        batch_indiecs = torch.tensor(indices[i:min(i+batch_size,num_examples)])
        yield features[batch_indiecs],labels[batch_indiecs]    
batch_size = 10

#for x ,y in data_iter(batch_size,feature,labels):
 #   print(x,'\n',y)
      #break

#初始化
w = torch.normal(0,0.01,size=(2,1),requires_grad = True)
b = torch.zeros(1,requires_grad = True)
w,b

(tensor([[-0.0050],
         [ 0.0162]], requires_grad=True),
 tensor([0.], requires_grad=True))

#线性回归
def linreg(x,w,b):
    return torch.matmul(x,w) + b

#损失函数
def squared_loss(y_hat,y):
    return (y_hat -y.reshape(y_hat.shape))**2/2

def sgd(params,lr,batch_size):
    with torch.no_grad():
        for param in params:
            
            
            param.sub_(lr*param.grad/batch_size)
            param.grad.zero_()

lr = 0.003
num_epochs = 10
net = linreg
loss = squared_loss
for epoch in range(num_epochs):
    for x,y in data_iter(batch_size,features,labels):
        y_hat = net(x,w,b)
        l = loss(y_hat,y)
        l.size()
        l.sum().backward()
        sgd([w,b],lr ,batch_size)
    with torch.no_grad():
        train_l = loss(net(features,w,b),labels)
        print(f'epoch {epoch + 1},loss {float(train_l.mean()):f}')

epoch 1,loss 9.036007
epoch 2,loss 5.028663
epoch 3,loss 2.799776
epoch 4,loss 1.559638
epoch 5,loss 0.869278
epoch 6,loss 0.484746
epoch 7,loss 0.270458
epoch 8,loss 0.150984
epoch 9,loss 0.084333
epoch 10,loss 0.047141

print(f'w的误差：{t_w -w.reshape(t_w.shape)}')
print(f'w的误差：{t_b -b}')

w的误差：tensor([ 0.1238, -0.2036], grad_fn=<SubBackward0>)
w的误差：tensor([0.2017], grad_fn=<RsubBackward1>)