2021-10-13

PyTorch反向传播学习实例（p4）刘二

import torch

x_da = [1.0,2.0,3.0]
y_da = [2.0,4.0,6.0]
w1 = torch.tensor([1.0])
w1.requires_grad= True

w2= torch.tensor([1.0])
w2.requires_grad=True


b = torch.tensor([1.0])
b.requires_grad=True
def dors(x):
    return w1 *x *x + w2 *x +b

def Loss(xs , ys ):
    return (dors(xs)-ys)**2

for  i  in range(100):
    for g,h in zip(x_da,y_da):
        print(Loss(g,h))
        l= Loss(g,h)
        l.backward()
        w1.data = w1.data - 0.01*w1.grad.data
        w2.data  = w2.data - 0.01*w2.grad.item()
        b.data = b.data - 0.01*b.grad.item()
        w1.grad.data.zero_()
        w2.grad.data.zero_()
        b.grad.data.zero_()
    print(dors(4))


```![在这里插入图片描述](https://img-blog.csdnimg.cn/10a43d9734e94c46b9043d32a155628d.JPG?x-oss-process=image/watermark,type_ZHJvaWRzYW5zZmFsbGJhY2s,shadow_50,text_Q1NETiBAY29yYysr,size_20,color_FFFFFF,t_70,g_se,x_16#pic_center)
**对b调整**

import torch

x_da = [1.0,2.0,3.0]
y_da = [2.0,4.0,6.0]
w1 = torch.tensor([1.0])
w1.requires_grad= True

w2= torch.tensor([1.0])
w2.requires_grad=True


b = torch.tensor([0.000000003])
b.requires_grad=True
def dors(x):
    return w1 *x *x + w2 *x +b

def Loss(xs , ys ):
    return (dors(xs)-ys)**2

for  i  in range(100):
    for g,h in zip(x_da,y_da):
        print(Loss(g,h))
        l= Loss(g,h)
        l.backward()
        w1.data = w1.data - 0.01*w1.grad.data
        w2.data  = w2.data - 0.01*w2.grad.item()
        b.data = b.data - 0.01*b.grad.item()
        w1.grad.data.zero_()
        w2.grad.data.zero_()
        b.grad.data.zero_()
    print(dors(4))
![通过对b调参，影响不大](https://img-blog.csdnimg.cn/a46188d111864eedb52064946eca39db.JPG?x-oss-process=image/watermark,type_ZHJvaWRzYW5zZmFsbGJhY2s,shadow_50,text_Q1NETiBAY29yYysr,size_20,color_FFFFFF,t_70,g_se,x_16#pic_center)