线性回归的从零开始实现

import torch
from IPython import display
from matplotlib import pyplot as plt
import numpy as np
import random

# 1、生成数据集
# 训练数据集样本数为1000，输入个数（特征数）为2
# 线性回归模型真实权重 w=[2,−3.4] 和偏差 b=4.2
num_inputs = 2
num_examples = 1000
true_w = [2,-3.4]
true_b = 4.2
# 生成的数据集X
features = torch.randn(num_examples,num_inputs,dtype=torch.float32)
print("生成的数据集X：",features)

labels = true_w[0]*features[:, 0] + true_w[1]*features[:, 1] + true_b
# 生成的数据集y
labels += torch.tensor(np.random.normal(0,0.01,size=labels.size()),dtype=torch.float32)

# print(features[0],labels[0])

# plt函数作图
def use_svg_display():
    # 用矢量图显示
    display.set_matplotlib_formats('svg')
def set_figsize(figsize=(3.5,2.5)):
    use_svg_display()
    plt.rcParams['figure.figsize'] = figsize
set_figsize()
plt.scatter(features[:, 1].numpy(),labels.numpy(),1)
# plt.show()

# 2、读取数据
# 每次返回batch_size（批量大小）个随机样本的特征和标签
def data_iter(batch_size,features,labels):
    num_examples = len(features)
    indices = list(range(num_examples))
    random.shuffle(indices) # 样本的读取顺序是随机的
    for i in range(0,num_examples,batch_size):
        j = torch.LongTensor(indices[i: min(i + batch_size,num_examples)]) # 最后一次可能不足一个batch
        yield features.index_select(0,j), labels.index_select(0,j)
batch_size = 10
for X,y in data_iter(batch_size,features,labels):
    print(X,y)
    break

# 3、初始化模型参数
w = torch.tensor(np.random.normal(0,0.01,(num_inputs,1)),dtype=torch.float32) # 权重初始化成均值为0、标准差为0.01的正态随机数
b = torch.zeros(1,dtype=torch.float32) # 偏差则初始化成0
# 之后的模型训练中，需要对这些参数求梯度来迭代参数的值，因此我们要让它们的requires_grad=True
w.requires_grad_(requires_grad=True)
b.requires_grad_(requires_grad=True)

# 4、定义模型
def linreg(X, w, b):
    return torch.mm(X,w) +b # mm函数作矩阵乘法

# 5、定义损失函数
def squared_loss(y_hat, y):
    # 真实值y变形成预测值y_hat的形状
    return (y_hat - y.view(y_hat.size())) ** 2 / 2

# 6、定义优化函数
# 自动求梯度模块计算得来的梯度是一个批量样本的梯度和。我们将它除以批量大小来得到平均值。
def sgd(params, lr, batch_size):
    for param in params:
        param.data -= lr * param.grad / batch_size

# 7、训练模型
lr = 0.03
num_epochs = 3
net = linreg
loss = squared_loss
for epoch in range(num_epochs): # 训练模型一共需要num_epochs个迭代周期
    # 在每一个迭代周期中，会使用训练数据集中所有样本一次（假设样本数能够被批量大小整除）。X
    # 和y分别是小批量样本的特征和标签
    for X,y in data_iter(batch_size,features,labels):
        l = loss(net(X, w, b),y).sum() # l是有关小批量X和y的损失
        l.backward() # 小批量的损失对模型参数求梯度
        sgd([w,b], lr, batch_size) # 使用小批量随机下降迭代模型参数

        # 梯度清零
        w.grad.data.zero_()
        b.grad.data.zero_()
    train_l = loss(net(features, w, b),labels)
    print('epoch %d, loss %f' % (epoch + 1, train_l.mean().item()))
# 真实值与预测值比较
print(true_w, '\n', w)
print(true_b, '\n', b)

D:\dev\anaconda\python.exe E:/DL-Pytorch/ch3/demo2.py
tensor([[ 1.0830, -0.8833],
        [ 0.5864,  0.7240],
        [ 0.9755, -0.6456],
        [-1.6580,  0.9215],
        [-0.1988, -1.2996],
        [ 1.0861, -0.3094],
        [ 0.4570, -0.8737],
        [-0.8036, -0.2254],
        [ 0.5252,  0.9146],
        [-0.0726, -1.0572]]) tensor([ 9.3613,  2.9110,  8.3544, -2.2501,  8.2146,  7.4407,  8.0797,  3.3551,
         2.1247,  7.6568])
epoch 1, loss 0.037553
epoch 2, loss 0.000137
epoch 3, loss 0.000051
[2, -3.4] 
 tensor([[ 1.9997],
        [-3.3992]], requires_grad=True)
4.2 
 tensor([4.1995], requires_grad=True)

Process finished with exit code 0