Pytorch实现最简单的线性回归

在运行环境中输入：

import torch
from torch.autograd import Variable

#准备数据
x_data = [5.0, 7.0, 3.0]
y_data = [15.0, 21.0, 9.0]

#产生随机变量
w = Variable(torch.Tensor([1.0]), requires_grad=True)
r = 0.001;  #定义学习率
#准备模型
def forward(x):
    return x * w

#设置损失函数
def loss(x, y):
    y_pred = forward(x)
    return (y_pred - y) * (y_pred - y)


# 训练前预测输入为10时，预测结果为
print("predict (before training)",  10, forward(10).data[0])

# Training loop
for epoch in range(100):
    for x_val, y_val in zip(x_data, y_data):
        l = loss(x_val, y_val)
        l.backward()
        print("\tgrad: ", x_val, y_val, w.grad.data[0])
        w.data = w.data - r * w.grad.data

        #更新权重后手动调零渐变
        w.grad.data.zero_()

    print("progress:", epoch, l.data[0])

#输入10，得到预测结果
print("predict (after training)",  10, forward(10).data[0])

程序运行结果：

progress: 98 tensor(1.4552e-11)
	grad:  5.0 15.0 tensor(-5.7220e-05)
	grad:  7.0 21.0 tensor(-0.0001)
	grad:  3.0 9.0 tensor(-2.2888e-05)
progress: 99 tensor(1.4552e-11)
predict (after training) 10 tensor(30.0000)