《PyTorch深度学习实践》-P5线性回归

构造神经网络一般步骤：

1prepare dataset

2design model using Class (计算y_hat)

3consturct loss and optimizer (using PyTorch API构造损失函数和优化器)

4training cycle (forwward backward update 前馈算损失，后馈算梯度，更新权重）

广播机制：

用mini-batch的方式来做线性回归

第一步

X、Y是3x1的tensor（张量），w是3x1的矩阵（这里可以理解为矩阵）（对应位置相乘，不是矩阵相乘）

第二步

重点目标：构造计算图，让pytorch自动求梯度

仿射模型y_hat=x*w+b，在pytorch中近似为线性单元z=wx+b，要确定w、b的大小需要知道x，y_hat的维度，

loss经过计算最终是一个标量，向量没办法backward

linearmodule会自动backward

默认bias=True，有偏置，

self.linear(x),对象后面加括号，实现一个collable可调用的对象，python中常用

举例：

class Foobar():
    def __init__(self):
        pass
    def __call__(self, *args, **kwargs):
        print("Hello"+str(args[0]))

def func(*args,**kwargs):#这两个是python中的可变参数。*args 表示任何多个无名参数，它是一个tuple；**kwargs 表示关键字参数，它是一个dict
    print(args)
    print(kwargs)

func(1,2,4,3,x=3,y=5)

foobar=Foobar()
foobar(1,2,3)

输出：

(1, 2, 4, 3)
{'x': 3, 'y': 5}
Hello1

第三步：构造损失函数和优化器

损失函数loss=(y_hat-y)**2, 用MSE,size_average=True 求均值，等于 false不求均值，求不求均值都一样，在nimi-batch中，若某一批样本数较少需要求均值，criterion需要y_hat,y就可以求损失（需要构建计算图，继承自nn.module）,reduce表示要不要降维，一般只考虑size_average

优化器不是module，不会构建计算图，SDG是一个类，实例化SDG，第一个参数para是权重，

lr是学习率，pytorch支持对模型不同部分使用不同学习率

model.parameters （）不管模型多复杂都能找到他们的参数

第四步：训练轮数

一共四步：计算y_hat,loss，backard，更新

weight是矩阵，只需要打印值用item()方法

本节完整代码：

import torch

from matplotlib import pyplot as plt
x_data=torch.Tensor([[1.0],[2.0],[3.0]])
y_data=torch.Tensor([[2.0],[4.0],[6.0]])

class LinearModel(torch.nn.Module):#our module class inherit from nn.module(neural network module)
    def __init__(self):#构造函数，初始化
        super().__init__()#继承父类的init方法
        self.linear=torch.nn.Linear(1,1)#Class nn.Linear包含权重和偏置两个tensor

    def forward(self,x):#重写父类中forward函数
        y_pred=self.linear(x)
        return y_pred

model=LinearModel() # creat a instance of linearmodel.model是collable,即可以model（x）

#nn.MSELoss继承自nn.Module
criterion=torch.nn.MSELoss(size_average=False)
#选择优化器，lr是learing rate
optimizer=torch.optim.SGD(model.parameters(),lr=0.01)


for epoch in range(100):
    y_pred=model(x_data) #fforwward:predict 算出y_hat
    loss=criterion(y_pred,y_data)#forward :loss
    print(epoch,loss)

    optimizer.zero_grad()#before backard，梯度归零
    loss.backward()#backward:autograd
    optimizer.step()#用step函数update更新


#output weight and bias
print('w=',model.linear.weight.item())
print('b=',model.linear.bias.item())

#test model
x_test=torch.tensor([4.0])
y_test=model(x_test)
print('y_pred=',y_test.data)

输出：

D:Anacodaenvspytorch-py36python.exe C:/Users/hp/Desktop/python_work/PyTorch/Lesson1/LinearRegression.py
0 tensor(48.8937, grad_fn=)
D:Anacodaenvspytorch-py36libsite-packagestorchnn_reduction.py:44: UserWarning: size_average and reduce args will be deprecated, please use reduction='sum' instead.
  warnings.warn(warning.format(ret))
1 tensor(21.9066, grad_fn=)
2 tensor(9.8907, grad_fn=)
3 tensor(4.5396, grad_fn=)
4 tensor(2.1555, grad_fn=)
5 tensor(1.0922, grad_fn=)
6 tensor(0.6169, grad_fn=)
7 tensor(0.4035, grad_fn=)
8 tensor(0.3066, grad_fn=)
9 tensor(0.2617, grad_fn=)
10 tensor(0.2399, grad_fn=)
11 tensor(0.2284, grad_fn=)
12 tensor(0.2215, grad_fn=)
13 tensor(0.2167, grad_fn=)
14 tensor(0.2129, grad_fn=)
15 tensor(0.2095, grad_fn=)
16 tensor(0.2064, grad_fn=)
17 tensor(0.2034, grad_fn=)
18 tensor(0.2004, grad_fn=)
19 tensor(0.1975, grad_fn=)
20 tensor(0.1947, grad_fn=)
21 tensor(0.1919, grad_fn=)
22 tensor(0.1891, grad_fn=)
23 tensor(0.1864, grad_fn=)
24 tensor(0.1837, grad_fn=)
25 tensor(0.1811, grad_fn=)
26 tensor(0.1785, grad_fn=)
27 tensor(0.1759, grad_fn=)
28 tensor(0.1734, grad_fn=)
29 tensor(0.1709, grad_fn=)
30 tensor(0.1684, grad_fn=)
31 tensor(0.1660, grad_fn=)
32 tensor(0.1636, grad_fn=)
33 tensor(0.1613, grad_fn=)
34 tensor(0.1590, grad_fn=)
35 tensor(0.1567, grad_fn=)
36 tensor(0.1544, grad_fn=)
37 tensor(0.1522, grad_fn=)
38 tensor(0.1500, grad_fn=)
39 tensor(0.1479, grad_fn=)
40 tensor(0.1457, grad_fn=)
41 tensor(0.1436, grad_fn=)
42 tensor(0.1416, grad_fn=)
43 tensor(0.1395, grad_fn=)
44 tensor(0.1375, grad_fn=)
45 tensor(0.1356, grad_fn=)
46 tensor(0.1336, grad_fn=)
47 tensor(0.1317, grad_fn=)
48 tensor(0.1298, grad_fn=)
49 tensor(0.1279, grad_fn=)
50 tensor(0.1261, grad_fn=)
51 tensor(0.1243, grad_fn=)
52 tensor(0.1225, grad_fn=)
53 tensor(0.1207, grad_fn=)
54 tensor(0.1190, grad_fn=)
55 tensor(0.1173, grad_fn=)
56 tensor(0.1156, grad_fn=)
57 tensor(0.1139, grad_fn=)
58 tensor(0.1123, grad_fn=)
59 tensor(0.1107, grad_fn=)
60 tensor(0.1091, grad_fn=)
61 tensor(0.1075, grad_fn=)
62 tensor(0.1060, grad_fn=)
63 tensor(0.1045, grad_fn=)
64 tensor(0.1030, grad_fn=)
65 tensor(0.1015, grad_fn=)
66 tensor(0.1000, grad_fn=)
67 tensor(0.0986, grad_fn=)
68 tensor(0.0972, grad_fn=)
69 tensor(0.0958, grad_fn=)
70 tensor(0.0944, grad_fn=)
71 tensor(0.0930, grad_fn=)
72 tensor(0.0917, grad_fn=)
73 tensor(0.0904, grad_fn=)
74 tensor(0.0891, grad_fn=)
75 tensor(0.0878, grad_fn=)
76 tensor(0.0865, grad_fn=)
77 tensor(0.0853, grad_fn=)
78 tensor(0.0841, grad_fn=)
79 tensor(0.0829, grad_fn=)
80 tensor(0.0817, grad_fn=)
81 tensor(0.0805, grad_fn=)
82 tensor(0.0793, grad_fn=)
83 tensor(0.0782, grad_fn=)
84 tensor(0.0771, grad_fn=)
85 tensor(0.0760, grad_fn=)
86 tensor(0.0749, grad_fn=)
87 tensor(0.0738, grad_fn=)
88 tensor(0.0727, grad_fn=)
89 tensor(0.0717, grad_fn=)
90 tensor(0.0707, grad_fn=)
91 tensor(0.0697, grad_fn=)
92 tensor(0.0686, grad_fn=)
93 tensor(0.0677, grad_fn=)
94 tensor(0.0667, grad_fn=)
95 tensor(0.0657, grad_fn=)
96 tensor(0.0648, grad_fn=)
97 tensor(0.0639, grad_fn=)
98 tensor(0.0629, grad_fn=)
99 tensor(0.0620, grad_fn=)
w= 1.834191083908081
b= 0.3769226372241974
y_pred= tensor([7.7137])

Process finished with exit code 0

迭代100次结果不是很理想，迭代1000词之后的结果：

D:Anacodaenvspytorch-py36python.exe C:/Users/hp/Desktop/python_work/PyTorch/Lesson1/LinearRegression.py
D:Anacodaenvspytorch-py36libsite-packagestorchnn_reduction.py:44: UserWarning: size_average and reduce args will be deprecated, please use reduction='sum' instead.
  warnings.warn(warning.format(ret))
0 tensor(111.9742, grad_fn=)
1 tensor(49.9149, grad_fn=)
2 tensor(22.2869, grad_fn=)
3 tensor(9.9867, grad_fn=)
4 tensor(4.5101, grad_fn=)
5 tensor(2.0711, grad_fn=)
6 tensor(0.9845, grad_fn=)
7 tensor(0.4998, grad_fn=)
8 tensor(0.2832, grad_fn=)
9 tensor(0.1859, grad_fn=)
10 tensor(0.1417, grad_fn=)
11 tensor(0.1212, grad_fn=)
12 tensor(0.1112, grad_fn=)
13 tensor(0.1060, grad_fn=)
14 tensor(0.1028, grad_fn=)
15 tensor(0.1006, grad_fn=)
16 tensor(0.0988, grad_fn=)
17 tensor(0.0973, grad_fn=)
18 tensor(0.0958, grad_fn=)
19 tensor(0.0944, grad_fn=)
20 tensor(0.0930, grad_fn=)
21 tensor(0.0917, grad_fn=)
22 tensor(0.0904, grad_fn=)
23 tensor(0.0891, grad_fn=)
24 tensor(0.0878, grad_fn=)
25 tensor(0.0865, grad_fn=)
26 tensor(0.0853, grad_fn=)
27 tensor(0.0841, grad_fn=)
28 tensor(0.0828, grad_fn=)
29 tensor(0.0817, grad_fn=)
30 tensor(0.0805, grad_fn=)
31 tensor(0.0793, grad_fn=)
32 tensor(0.0782, grad_fn=)
33 tensor(0.0771, grad_fn=)
34 tensor(0.0760, grad_fn=)
35 tensor(0.0749, grad_fn=)
36 tensor(0.0738, grad_fn=)
37 tensor(0.0727, grad_fn=)
38 tensor(0.0717, grad_fn=)
39 tensor(0.0707, grad_fn=)
40 tensor(0.0696, grad_fn=)
41 tensor(0.0686, grad_fn=)
42 tensor(0.0677, grad_fn=)
43 tensor(0.0667, grad_fn=)
44 tensor(0.0657, grad_fn=)
45 tensor(0.0648, grad_fn=)
46 tensor(0.0638, grad_fn=)
47 tensor(0.0629, grad_fn=)
48 tensor(0.0620, grad_fn=)
49 tensor(0.0611, grad_fn=)
50 tensor(0.0603, grad_fn=)
51 tensor(0.0594, grad_fn=)
52 tensor(0.0585, grad_fn=)
53 tensor(0.0577, grad_fn=)
54 tensor(0.0569, grad_fn=)
55 tensor(0.0560, grad_fn=)
56 tensor(0.0552, grad_fn=)
57 tensor(0.0544, grad_fn=)
58 tensor(0.0537, grad_fn=)
59 tensor(0.0529, grad_fn=)
60 tensor(0.0521, grad_fn=)
61 tensor(0.0514, grad_fn=)
62 tensor(0.0506, grad_fn=)
63 tensor(0.0499, grad_fn=)
64 tensor(0.0492, grad_fn=)
65 tensor(0.0485, grad_fn=)
66 tensor(0.0478, grad_fn=)
67 tensor(0.0471, grad_fn=)
68 tensor(0.0464, grad_fn=)
69 tensor(0.0458, grad_fn=)
70 tensor(0.0451, grad_fn=)
71 tensor(0.0445, grad_fn=)
72 tensor(0.0438, grad_fn=)
73 tensor(0.0432, grad_fn=)
74 tensor(0.0426, grad_fn=)
75 tensor(0.0420, grad_fn=)
76 tensor(0.0414, grad_fn=)
77 tensor(0.0408, grad_fn=)
78 tensor(0.0402, grad_fn=)
79 tensor(0.0396, grad_fn=)
80 tensor(0.0390, grad_fn=)
81 tensor(0.0385, grad_fn=)
82 tensor(0.0379, grad_fn=)
83 tensor(0.0374, grad_fn=)
84 tensor(0.0368, grad_fn=)
85 tensor(0.0363, grad_fn=)
86 tensor(0.0358, grad_fn=)
87 tensor(0.0353, grad_fn=)
88 tensor(0.0348, grad_fn=)
89 tensor(0.0343, grad_fn=)
90 tensor(0.0338, grad_fn=)
91 tensor(0.0333, grad_fn=)
92 tensor(0.0328, grad_fn=)
93 tensor(0.0323, grad_fn=)
94 tensor(0.0319, grad_fn=)
95 tensor(0.0314, grad_fn=)
96 tensor(0.0310, grad_fn=)
97 tensor(0.0305, grad_fn=)
98 tensor(0.0301, grad_fn=)
99 tensor(0.0296, grad_fn=)
100 tensor(0.0292, grad_fn=)
101 tensor(0.0288, grad_fn=)
102 tensor(0.0284, grad_fn=)
103 tensor(0.0280, grad_fn=)
104 tensor(0.0276, grad_fn=)
105 tensor(0.0272, grad_fn=)
106 tensor(0.0268, grad_fn=)
107 tensor(0.0264, grad_fn=)
108 tensor(0.0260, grad_fn=)
109 tensor(0.0256, grad_fn=)
110 tensor(0.0253, grad_fn=)
111 tensor(0.0249, grad_fn=)
112 tensor(0.0246, grad_fn=)
113 tensor(0.0242, grad_fn=)
114 tensor(0.0239, grad_fn=)
115 tensor(0.0235, grad_fn=)
116 tensor(0.0232, grad_fn=)
117 tensor(0.0228, grad_fn=)
118 tensor(0.0225, grad_fn=)
119 tensor(0.0222, grad_fn=)
120 tensor(0.0219, grad_fn=)
121 tensor(0.0216, grad_fn=)
122 tensor(0.0212, grad_fn=)
123 tensor(0.0209, grad_fn=)
124 tensor(0.0206, grad_fn=)
125 tensor(0.0203, grad_fn=)
126 tensor(0.0201, grad_fn=)
127 tensor(0.0198, grad_fn=)
128 tensor(0.0195, grad_fn=)
129 tensor(0.0192, grad_fn=)
130 tensor(0.0189, grad_fn=)
131 tensor(0.0187, grad_fn=)
132 tensor(0.0184, grad_fn=)
133 tensor(0.0181, grad_fn=)
134 tensor(0.0179, grad_fn=)
135 tensor(0.0176, grad_fn=)
136 tensor(0.0174, grad_fn=)
137 tensor(0.0171, grad_fn=)
138 tensor(0.0169, grad_fn=)
139 tensor(0.0166, grad_fn=)
140 tensor(0.0164, grad_fn=)
141 tensor(0.0161, grad_fn=)
142 tensor(0.0159, grad_fn=)
143 tensor(0.0157, grad_fn=)
144 tensor(0.0155, grad_fn=)
145 tensor(0.0152, grad_fn=)
146 tensor(0.0150, grad_fn=)
147 tensor(0.0148, grad_fn=)
148 tensor(0.0146, grad_fn=)
149 tensor(0.0144, grad_fn=)
150 tensor(0.0142, grad_fn=)
151 tensor(0.0140, grad_fn=)
152 tensor(0.0138, grad_fn=)
153 tensor(0.0136, grad_fn=)
154 tensor(0.0134, grad_fn=)
155 tensor(0.0132, grad_fn=)
156 tensor(0.0130, grad_fn=)
157 tensor(0.0128, grad_fn=)
158 tensor(0.0126, grad_fn=)
159 tensor(0.0124, grad_fn=)
160 tensor(0.0123, grad_fn=)
161 tensor(0.0121, grad_fn=)
162 tensor(0.0119, grad_fn=)
163 tensor(0.0117, grad_fn=)
164 tensor(0.0116, grad_fn=)
165 tensor(0.0114, grad_fn=)
166 tensor(0.0112, grad_fn=)
167 tensor(0.0111, grad_fn=)
168 tensor(0.0109, grad_fn=)
169 tensor(0.0108, grad_fn=)
170 tensor(0.0106, grad_fn=)
171 tensor(0.0105, grad_fn=)
172 tensor(0.0103, grad_fn=)
173 tensor(0.0102, grad_fn=)
174 tensor(0.0100, grad_fn=)
175 tensor(0.0099, grad_fn=)
176 tensor(0.0097, grad_fn=)
177 tensor(0.0096, grad_fn=)
178 tensor(0.0094, grad_fn=)
179 tensor(0.0093, grad_fn=)
180 tensor(0.0092, grad_fn=)
181 tensor(0.0090, grad_fn=)
182 tensor(0.0089, grad_fn=)
183 tensor(0.0088, grad_fn=)
184 tensor(0.0087, grad_fn=)
185 tensor(0.0085, grad_fn=)
186 tensor(0.0084, grad_fn=)
187 tensor(0.0083, grad_fn=)
188 tensor(0.0082, grad_fn=)
189 tensor(0.0081, grad_fn=)
190 tensor(0.0079, grad_fn=)
191 tensor(0.0078, grad_fn=)
192 tensor(0.0077, grad_fn=)
193 tensor(0.0076, grad_fn=)
194 tensor(0.0075, grad_fn=)
195 tensor(0.0074, grad_fn=)
196 tensor(0.0073, grad_fn=)
197 tensor(0.0072, grad_fn=)
198 tensor(0.0071, grad_fn=)
199 tensor(0.0070, grad_fn=)
200 tensor(0.0069, grad_fn=)
201 tensor(0.0068, grad_fn=)
202 tensor(0.0067, grad_fn=)
203 tensor(0.0066, grad_fn=)
204 tensor(0.0065, grad_fn=)
205 tensor(0.0064, grad_fn=)
206 tensor(0.0063, grad_fn=)
207 tensor(0.0062, grad_fn=)
208 tensor(0.0061, grad_fn=)
209 tensor(0.0060, grad_fn=)
210 tensor(0.0059, grad_fn=)
211 tensor(0.0059, grad_fn=)
212 tensor(0.0058, grad_fn=)
213 tensor(0.0057, grad_fn=)
214 tensor(0.0056, grad_fn=)
215 tensor(0.0055, grad_fn=)
216 tensor(0.0054, grad_fn=)
217 tensor(0.0054, grad_fn=)
218 tensor(0.0053, grad_fn=)
219 tensor(0.0052, grad_fn=)
220 tensor(0.0051, grad_fn=)
221 tensor(0.0051, grad_fn=)
222 tensor(0.0050, grad_fn=)
223 tensor(0.0049, grad_fn=)
224 tensor(0.0049, grad_fn=)
225 tensor(0.0048, grad_fn=)
226 tensor(0.0047, grad_fn=)
227 tensor(0.0046, grad_fn=)
228 tensor(0.0046, grad_fn=)
229 tensor(0.0045, grad_fn=)
230 tensor(0.0044, grad_fn=)
231 tensor(0.0044, grad_fn=)
232 tensor(0.0043, grad_fn=)
233 tensor(0.0043, grad_fn=)
234 tensor(0.0042, grad_fn=)
235 tensor(0.0041, grad_fn=)
236 tensor(0.0041, grad_fn=)
237 tensor(0.0040, grad_fn=)
238 tensor(0.0040, grad_fn=)
239 tensor(0.0039, grad_fn=)
240 tensor(0.0039, grad_fn=)
241 tensor(0.0038, grad_fn=)
242 tensor(0.0037, grad_fn=)
243 tensor(0.0037, grad_fn=)
244 tensor(0.0036, grad_fn=)
245 tensor(0.0036, grad_fn=)
246 tensor(0.0035, grad_fn=)
247 tensor(0.0035, grad_fn=)
248 tensor(0.0034, grad_fn=)
249 tensor(0.0034, grad_fn=)
250 tensor(0.0033, grad_fn=)
251 tensor(0.0033, grad_fn=)
252 tensor(0.0032, grad_fn=)
253 tensor(0.0032, grad_fn=)
254 tensor(0.0031, grad_fn=)
255 tensor(0.0031, grad_fn=)
256 tensor(0.0031, grad_fn=)
257 tensor(0.0030, grad_fn=)
258 tensor(0.0030, grad_fn=)
259 tensor(0.0029, grad_fn=)
260 tensor(0.0029, grad_fn=)
261 tensor(0.0028, grad_fn=)
262 tensor(0.0028, grad_fn=)
263 tensor(0.0028, grad_fn=)
264 tensor(0.0027, grad_fn=)
265 tensor(0.0027, grad_fn=)
266 tensor(0.0026, grad_fn=)
267 tensor(0.0026, grad_fn=)
268 tensor(0.0026, grad_fn=)
269 tensor(0.0025, grad_fn=)
270 tensor(0.0025, grad_fn=)
271 tensor(0.0025, grad_fn=)
272 tensor(0.0024, grad_fn=)
273 tensor(0.0024, grad_fn=)
274 tensor(0.0024, grad_fn=)
275 tensor(0.0023, grad_fn=)
276 tensor(0.0023, grad_fn=)
277 tensor(0.0023, grad_fn=)
278 tensor(0.0022, grad_fn=)
279 tensor(0.0022, grad_fn=)
280 tensor(0.0022, grad_fn=)
281 tensor(0.0021, grad_fn=)
282 tensor(0.0021, grad_fn=)
283 tensor(0.0021, grad_fn=)
284 tensor(0.0020, grad_fn=)
285 tensor(0.0020, grad_fn=)
286 tensor(0.0020, grad_fn=)
287 tensor(0.0019, grad_fn=)
288 tensor(0.0019, grad_fn=)
289 tensor(0.0019, grad_fn=)
290 tensor(0.0019, grad_fn=)
291 tensor(0.0018, grad_fn=)
292 tensor(0.0018, grad_fn=)
293 tensor(0.0018, grad_fn=)
294 tensor(0.0018, grad_fn=)
295 tensor(0.0017, grad_fn=)
296 tensor(0.0017, grad_fn=)
297 tensor(0.0017, grad_fn=)
298 tensor(0.0017, grad_fn=)
299 tensor(0.0016, grad_fn=)
300 tensor(0.0016, grad_fn=)
301 tensor(0.0016, grad_fn=)
302 tensor(0.0016, grad_fn=)
303 tensor(0.0015, grad_fn=)
304 tensor(0.0015, grad_fn=)
305 tensor(0.0015, grad_fn=)
306 tensor(0.0015, grad_fn=)
307 tensor(0.0015, grad_fn=)
308 tensor(0.0014, grad_fn=)
309 tensor(0.0014, grad_fn=)
310 tensor(0.0014, grad_fn=)
311 tensor(0.0014, grad_fn=)
312 tensor(0.0014, grad_fn=)
313 tensor(0.0013, grad_fn=)
314 tensor(0.0013, grad_fn=)
315 tensor(0.0013, grad_fn=)
316 tensor(0.0013, grad_fn=)
317 tensor(0.0013, grad_fn=)
318 tensor(0.0012, grad_fn=)
319 tensor(0.0012, grad_fn=)
320 tensor(0.0012, grad_fn=)
321 tensor(0.0012, grad_fn=)
322 tensor(0.0012, grad_fn=)
323 tensor(0.0012, grad_fn=)
324 tensor(0.0011, grad_fn=)
325 tensor(0.0011, grad_fn=)
326 tensor(0.0011, grad_fn=)
327 tensor(0.0011, grad_fn=)
328 tensor(0.0011, grad_fn=)
329 tensor(0.0011, grad_fn=)
330 tensor(0.0010, grad_fn=)
331 tensor(0.0010, grad_fn=)
332 tensor(0.0010, grad_fn=)
333 tensor(0.0010, grad_fn=)
334 tensor(0.0010, grad_fn=)
335 tensor(0.0010, grad_fn=)
336 tensor(0.0010, grad_fn=)
337 tensor(0.0009, grad_fn=)
338 tensor(0.0009, grad_fn=)
339 tensor(0.0009, grad_fn=)
340 tensor(0.0009, grad_fn=)
341 tensor(0.0009, grad_fn=)
342 tensor(0.0009, grad_fn=)
343 tensor(0.0009, grad_fn=)
344 tensor(0.0009, grad_fn=)
345 tensor(0.0008, grad_fn=)
346 tensor(0.0008, grad_fn=)
347 tensor(0.0008, grad_fn=)
348 tensor(0.0008, grad_fn=)
349 tensor(0.0008, grad_fn=)
350 tensor(0.0008, grad_fn=)
351 tensor(0.0008, grad_fn=)
352 tensor(0.0008, grad_fn=)
353 tensor(0.0008, grad_fn=)
354 tensor(0.0007, grad_fn=)
355 tensor(0.0007, grad_fn=)
356 tensor(0.0007, grad_fn=)
357 tensor(0.0007, grad_fn=)
358 tensor(0.0007, grad_fn=)
359 tensor(0.0007, grad_fn=)
360 tensor(0.0007, grad_fn=)
361 tensor(0.0007, grad_fn=)
362 tensor(0.0007, grad_fn=)
363 tensor(0.0006, grad_fn=)
364 tensor(0.0006, grad_fn=)
365 tensor(0.0006, grad_fn=)
366 tensor(0.0006, grad_fn=)
367 tensor(0.0006, grad_fn=)
368 tensor(0.0006, grad_fn=)
369 tensor(0.0006, grad_fn=)
370 tensor(0.0006, grad_fn=)
371 tensor(0.0006, grad_fn=)
372 tensor(0.0006, grad_fn=)
373 tensor(0.0006, grad_fn=)
374 tensor(0.0006, grad_fn=)
375 tensor(0.0005, grad_fn=)
376 tensor(0.0005, grad_fn=)
377 tensor(0.0005, grad_fn=)
378 tensor(0.0005, grad_fn=)
379 tensor(0.0005, grad_fn=)
380 tensor(0.0005, grad_fn=)
381 tensor(0.0005, grad_fn=)
382 tensor(0.0005, grad_fn=)
383 tensor(0.0005, grad_fn=)
384 tensor(0.0005, grad_fn=)
385 tensor(0.0005, grad_fn=)
386 tensor(0.0005, grad_fn=)
387 tensor(0.0005, grad_fn=)
388 tensor(0.0005, grad_fn=)
389 tensor(0.0004, grad_fn=)
390 tensor(0.0004, grad_fn=)
391 tensor(0.0004, grad_fn=)
392 tensor(0.0004, grad_fn=)
393 tensor(0.0004, grad_fn=)
394 tensor(0.0004, grad_fn=)
395 tensor(0.0004, grad_fn=)
396 tensor(0.0004, grad_fn=)
397 tensor(0.0004, grad_fn=)
398 tensor(0.0004, grad_fn=)
399 tensor(0.0004, grad_fn=)
400 tensor(0.0004, grad_fn=)
401 tensor(0.0004, grad_fn=)
402 tensor(0.0004, grad_fn=)
403 tensor(0.0004, grad_fn=)
404 tensor(0.0004, grad_fn=)
405 tensor(0.0004, grad_fn=)
406 tensor(0.0003, grad_fn=)
407 tensor(0.0003, grad_fn=)
408 tensor(0.0003, grad_fn=)
409 tensor(0.0003, grad_fn=)
410 tensor(0.0003, grad_fn=)
411 tensor(0.0003, grad_fn=)
412 tensor(0.0003, grad_fn=)
413 tensor(0.0003, grad_fn=)
414 tensor(0.0003, grad_fn=)
415 tensor(0.0003, grad_fn=)
416 tensor(0.0003, grad_fn=)
417 tensor(0.0003, grad_fn=)
418 tensor(0.0003, grad_fn=)
419 tensor(0.0003, grad_fn=)
420 tensor(0.0003, grad_fn=)
421 tensor(0.0003, grad_fn=)
422 tensor(0.0003, grad_fn=)
423 tensor(0.0003, grad_fn=)
424 tensor(0.0003, grad_fn=)
425 tensor(0.0003, grad_fn=)
426 tensor(0.0003, grad_fn=)
427 tensor(0.0003, grad_fn=)
428 tensor(0.0003, grad_fn=)
429 tensor(0.0002, grad_fn=)
430 tensor(0.0002, grad_fn=)
431 tensor(0.0002, grad_fn=)
432 tensor(0.0002, grad_fn=)
433 tensor(0.0002, grad_fn=)
434 tensor(0.0002, grad_fn=)
435 tensor(0.0002, grad_fn=)
436 tensor(0.0002, grad_fn=)
437 tensor(0.0002, grad_fn=)
438 tensor(0.0002, grad_fn=)
439 tensor(0.0002, grad_fn=)
440 tensor(0.0002, grad_fn=)
441 tensor(0.0002, grad_fn=)
442 tensor(0.0002, grad_fn=)
443 tensor(0.0002, grad_fn=)
444 tensor(0.0002, grad_fn=)
445 tensor(0.0002, grad_fn=)
446 tensor(0.0002, grad_fn=)
447 tensor(0.0002, grad_fn=)
448 tensor(0.0002, grad_fn=)
449 tensor(0.0002, grad_fn=)
450 tensor(0.0002, grad_fn=)
451 tensor(0.0002, grad_fn=)
452 tensor(0.0002, grad_fn=)
453 tensor(0.0002, grad_fn=)
454 tensor(0.0002, grad_fn=)
455 tensor(0.0002, grad_fn=)
456 tensor(0.0002, grad_fn=)
457 tensor(0.0002, grad_fn=)
458 tensor(0.0002, grad_fn=)
459 tensor(0.0002, grad_fn=)
460 tensor(0.0002, grad_fn=)
461 tensor(0.0002, grad_fn=)
462 tensor(0.0002, grad_fn=)
463 tensor(0.0002, grad_fn=)
464 tensor(0.0002, grad_fn=)
465 tensor(0.0001, grad_fn=)
466 tensor(0.0001, grad_fn=)
467 tensor(0.0001, grad_fn=)
468 tensor(0.0001, grad_fn=)
469 tensor(0.0001, grad_fn=)
470 tensor(0.0001, grad_fn=)
471 tensor(0.0001, grad_fn=)
472 tensor(0.0001, grad_fn=)
473 tensor(0.0001, grad_fn=)
474 tensor(0.0001, grad_fn=)
475 tensor(0.0001, grad_fn=)
476 tensor(0.0001, grad_fn=)
477 tensor(0.0001, grad_fn=)
478 tensor(0.0001, grad_fn=)
479 tensor(0.0001, grad_fn=)
480 tensor(0.0001, grad_fn=)
481 tensor(0.0001, grad_fn=)
482 tensor(0.0001, grad_fn=)
483 tensor(0.0001, grad_fn=)
484 tensor(0.0001, grad_fn=)
485 tensor(0.0001, grad_fn=)
486 tensor(0.0001, grad_fn=)
487 tensor(0.0001, grad_fn=)
488 tensor(0.0001, grad_fn=)
489 tensor(0.0001, grad_fn=)
490 tensor(0.0001, grad_fn=)
491 tensor(0.0001, grad_fn=)
492 tensor(0.0001, grad_fn=)
493 tensor(9.8838e-05, grad_fn=)
494 tensor(9.7421e-05, grad_fn=)
495 tensor(9.6016e-05, grad_fn=)
496 tensor(9.4642e-05, grad_fn=)
497 tensor(9.3277e-05, grad_fn=)
498 tensor(9.1940e-05, grad_fn=)
499 tensor(9.0620e-05, grad_fn=)
500 tensor(8.9316e-05, grad_fn=)
501 tensor(8.8032e-05, grad_fn=)
502 tensor(8.6769e-05, grad_fn=)
503 tensor(8.5520e-05, grad_fn=)
504 tensor(8.4290e-05, grad_fn=)
505 tensor(8.3077e-05, grad_fn=)
506 tensor(8.1886e-05, grad_fn=)
507 tensor(8.0706e-05, grad_fn=)
508 tensor(7.9551e-05, grad_fn=)
509 tensor(7.8404e-05, grad_fn=)
510 tensor(7.7282e-05, grad_fn=)
511 tensor(7.6169e-05, grad_fn=)
512 tensor(7.5074e-05, grad_fn=)
513 tensor(7.3997e-05, grad_fn=)
514 tensor(7.2933e-05, grad_fn=)
515 tensor(7.1883e-05, grad_fn=)
516 tensor(7.0847e-05, grad_fn=)
517 tensor(6.9827e-05, grad_fn=)
518 tensor(6.8829e-05, grad_fn=)
519 tensor(6.7841e-05, grad_fn=)
520 tensor(6.6864e-05, grad_fn=)
521 tensor(6.5904e-05, grad_fn=)
522 tensor(6.4956e-05, grad_fn=)
523 tensor(6.4023e-05, grad_fn=)
524 tensor(6.3099e-05, grad_fn=)
525 tensor(6.2194e-05, grad_fn=)
526 tensor(6.1304e-05, grad_fn=)
527 tensor(6.0419e-05, grad_fn=)
528 tensor(5.9554e-05, grad_fn=)
529 tensor(5.8696e-05, grad_fn=)
530 tensor(5.7850e-05, grad_fn=)
531 tensor(5.7022e-05, grad_fn=)
532 tensor(5.6203e-05, grad_fn=)
533 tensor(5.5395e-05, grad_fn=)
534 tensor(5.4595e-05, grad_fn=)
535 tensor(5.3811e-05, grad_fn=)
536 tensor(5.3039e-05, grad_fn=)
537 tensor(5.2277e-05, grad_fn=)
538 tensor(5.1529e-05, grad_fn=)
539 tensor(5.0786e-05, grad_fn=)
540 tensor(5.0054e-05, grad_fn=)
541 tensor(4.9334e-05, grad_fn=)
542 tensor(4.8629e-05, grad_fn=)
543 tensor(4.7929e-05, grad_fn=)
544 tensor(4.7237e-05, grad_fn=)
545 tensor(4.6561e-05, grad_fn=)
546 tensor(4.5894e-05, grad_fn=)
547 tensor(4.5230e-05, grad_fn=)
548 tensor(4.4579e-05, grad_fn=)
549 tensor(4.3939e-05, grad_fn=)
550 tensor(4.3311e-05, grad_fn=)
551 tensor(4.2688e-05, grad_fn=)
552 tensor(4.2071e-05, grad_fn=)
553 tensor(4.1466e-05, grad_fn=)
554 tensor(4.0875e-05, grad_fn=)
555 tensor(4.0284e-05, grad_fn=)
556 tensor(3.9706e-05, grad_fn=)
557 tensor(3.9136e-05, grad_fn=)
558 tensor(3.8573e-05, grad_fn=)
559 tensor(3.8020e-05, grad_fn=)
560 tensor(3.7472e-05, grad_fn=)
561 tensor(3.6934e-05, grad_fn=)
562 tensor(3.6403e-05, grad_fn=)
563 tensor(3.5880e-05, grad_fn=)
564 tensor(3.5366e-05, grad_fn=)
565 tensor(3.4857e-05, grad_fn=)
566 tensor(3.4356e-05, grad_fn=)
567 tensor(3.3861e-05, grad_fn=)
568 tensor(3.3375e-05, grad_fn=)
569 tensor(3.2892e-05, grad_fn=)
570 tensor(3.2422e-05, grad_fn=)
571 tensor(3.1957e-05, grad_fn=)
572 tensor(3.1499e-05, grad_fn=)
573 tensor(3.1042e-05, grad_fn=)
574 tensor(3.0598e-05, grad_fn=)
575 tensor(3.0158e-05, grad_fn=)
576 tensor(2.9725e-05, grad_fn=)
577 tensor(2.9298e-05, grad_fn=)
578 tensor(2.8876e-05, grad_fn=)
579 tensor(2.8462e-05, grad_fn=)
580 tensor(2.8053e-05, grad_fn=)
581 tensor(2.7652e-05, grad_fn=)
582 tensor(2.7253e-05, grad_fn=)
583 tensor(2.6861e-05, grad_fn=)
584 tensor(2.6474e-05, grad_fn=)
585 tensor(2.6093e-05, grad_fn=)
586 tensor(2.5718e-05, grad_fn=)
587 tensor(2.5351e-05, grad_fn=)
588 tensor(2.4984e-05, grad_fn=)
589 tensor(2.4626e-05, grad_fn=)
590 tensor(2.4273e-05, grad_fn=)
591 tensor(2.3923e-05, grad_fn=)
592 tensor(2.3580e-05, grad_fn=)
593 tensor(2.3242e-05, grad_fn=)
594 tensor(2.2908e-05, grad_fn=)
595 tensor(2.2577e-05, grad_fn=)
596 tensor(2.2254e-05, grad_fn=)
597 tensor(2.1932e-05, grad_fn=)
598 tensor(2.1620e-05, grad_fn=)
599 tensor(2.1309e-05, grad_fn=)
600 tensor(2.1001e-05, grad_fn=)
601 tensor(2.0699e-05, grad_fn=)
602 tensor(2.0402e-05, grad_fn=)
603 tensor(2.0108e-05, grad_fn=)
604 tensor(1.9819e-05, grad_fn=)
605 tensor(1.9536e-05, grad_fn=)
606 tensor(1.9255e-05, grad_fn=)
607 tensor(1.8977e-05, grad_fn=)
608 tensor(1.8704e-05, grad_fn=)
609 tensor(1.8436e-05, grad_fn=)
610 tensor(1.8172e-05, grad_fn=)
611 tensor(1.7910e-05, grad_fn=)
612 tensor(1.7651e-05, grad_fn=)
613 tensor(1.7397e-05, grad_fn=)
614 tensor(1.7148e-05, grad_fn=)
615 tensor(1.6902e-05, grad_fn=)
616 tensor(1.6657e-05, grad_fn=)
617 tensor(1.6420e-05, grad_fn=)
618 tensor(1.6185e-05, grad_fn=)
619 tensor(1.5953e-05, grad_fn=)
620 tensor(1.5723e-05, grad_fn=)
621 tensor(1.5494e-05, grad_fn=)
622 tensor(1.5273e-05, grad_fn=)
623 tensor(1.5055e-05, grad_fn=)
624 tensor(1.4836e-05, grad_fn=)
625 tensor(1.4623e-05, grad_fn=)
626 tensor(1.4413e-05, grad_fn=)
627 tensor(1.4207e-05, grad_fn=)
628 tensor(1.4004e-05, grad_fn=)
629 tensor(1.3801e-05, grad_fn=)
630 tensor(1.3605e-05, grad_fn=)
631 tensor(1.3408e-05, grad_fn=)
632 tensor(1.3215e-05, grad_fn=)
633 tensor(1.3026e-05, grad_fn=)
634 tensor(1.2837e-05, grad_fn=)
635 tensor(1.2654e-05, grad_fn=)
636 tensor(1.2472e-05, grad_fn=)
637 tensor(1.2293e-05, grad_fn=)
638 tensor(1.2116e-05, grad_fn=)
639 tensor(1.1941e-05, grad_fn=)
640 tensor(1.1770e-05, grad_fn=)
641 tensor(1.1601e-05, grad_fn=)
642 tensor(1.1434e-05, grad_fn=)
643 tensor(1.1269e-05, grad_fn=)
644 tensor(1.1109e-05, grad_fn=)
645 tensor(1.0949e-05, grad_fn=)
646 tensor(1.0791e-05, grad_fn=)
647 tensor(1.0636e-05, grad_fn=)
648 tensor(1.0484e-05, grad_fn=)
649 tensor(1.0332e-05, grad_fn=)
650 tensor(1.0183e-05, grad_fn=)
651 tensor(1.0037e-05, grad_fn=)
652 tensor(9.8947e-06, grad_fn=)
653 tensor(9.7515e-06, grad_fn=)
654 tensor(9.6098e-06, grad_fn=)
655 tensor(9.4742e-06, grad_fn=)
656 tensor(9.3367e-06, grad_fn=)
657 tensor(9.2020e-06, grad_fn=)
658 tensor(9.0703e-06, grad_fn=)
659 tensor(8.9402e-06, grad_fn=)
660 tensor(8.8130e-06, grad_fn=)
661 tensor(8.6845e-06, grad_fn=)
662 tensor(8.5596e-06, grad_fn=)
663 tensor(8.4372e-06, grad_fn=)
664 tensor(8.3152e-06, grad_fn=)
665 tensor(8.1949e-06, grad_fn=)
666 tensor(8.0771e-06, grad_fn=)
667 tensor(7.9627e-06, grad_fn=)
668 tensor(7.8482e-06, grad_fn=)
669 tensor(7.7340e-06, grad_fn=)
670 tensor(7.6231e-06, grad_fn=)
671 tensor(7.5131e-06, grad_fn=)
672 tensor(7.4054e-06, grad_fn=)
673 tensor(7.2991e-06, grad_fn=)
674 tensor(7.1941e-06, grad_fn=)
675 tensor(7.0905e-06, grad_fn=)
676 tensor(6.9893e-06, grad_fn=)
677 tensor(6.8894e-06, grad_fn=)
678 tensor(6.7910e-06, grad_fn=)
679 tensor(6.6914e-06, grad_fn=)
680 tensor(6.5951e-06, grad_fn=)
681 tensor(6.5007e-06, grad_fn=)
682 tensor(6.4080e-06, grad_fn=)
683 tensor(6.3159e-06, grad_fn=)
684 tensor(6.2257e-06, grad_fn=)
685 tensor(6.1360e-06, grad_fn=)
686 tensor(6.0469e-06, grad_fn=)
687 tensor(5.9599e-06, grad_fn=)
688 tensor(5.8753e-06, grad_fn=)
689 tensor(5.7902e-06, grad_fn=)
690 tensor(5.7067e-06, grad_fn=)
691 tensor(5.6249e-06, grad_fn=)
692 tensor(5.5450e-06, grad_fn=)
693 tensor(5.4643e-06, grad_fn=)
694 tensor(5.3855e-06, grad_fn=)
695 tensor(5.3089e-06, grad_fn=)
696 tensor(5.2333e-06, grad_fn=)
697 tensor(5.1571e-06, grad_fn=)
698 tensor(5.0835e-06, grad_fn=)
699 tensor(5.0090e-06, grad_fn=)
700 tensor(4.9373e-06, grad_fn=)
701 tensor(4.8673e-06, grad_fn=)
702 tensor(4.7966e-06, grad_fn=)
703 tensor(4.7273e-06, grad_fn=)
704 tensor(4.6587e-06, grad_fn=)
705 tensor(4.5920e-06, grad_fn=)
706 tensor(4.5265e-06, grad_fn=)
707 tensor(4.4626e-06, grad_fn=)
708 tensor(4.3979e-06, grad_fn=)
709 tensor(4.3343e-06, grad_fn=)
710 tensor(4.2729e-06, grad_fn=)
711 tensor(4.2104e-06, grad_fn=)
712 tensor(4.1500e-06, grad_fn=)
713 tensor(4.0908e-06, grad_fn=)
714 tensor(4.0314e-06, grad_fn=)
715 tensor(3.9739e-06, grad_fn=)
716 tensor(3.9163e-06, grad_fn=)
717 tensor(3.8604e-06, grad_fn=)
718 tensor(3.8044e-06, grad_fn=)
719 tensor(3.7502e-06, grad_fn=)
720 tensor(3.6958e-06, grad_fn=)
721 tensor(3.6427e-06, grad_fn=)
722 tensor(3.5904e-06, grad_fn=)
723 tensor(3.5388e-06, grad_fn=)
724 tensor(3.4889e-06, grad_fn=)
725 tensor(3.4380e-06, grad_fn=)
726 tensor(3.3882e-06, grad_fn=)
727 tensor(3.3405e-06, grad_fn=)
728 tensor(3.2914e-06, grad_fn=)
729 tensor(3.2451e-06, grad_fn=)
730 tensor(3.1975e-06, grad_fn=)
731 tensor(3.1518e-06, grad_fn=)
732 tensor(3.1064e-06, grad_fn=)
733 tensor(3.0622e-06, grad_fn=)
734 tensor(3.0182e-06, grad_fn=)
735 tensor(2.9740e-06, grad_fn=)
736 tensor(2.9322e-06, grad_fn=)
737 tensor(2.8899e-06, grad_fn=)
738 tensor(2.8479e-06, grad_fn=)
739 tensor(2.8079e-06, grad_fn=)
740 tensor(2.7671e-06, grad_fn=)
741 tensor(2.7281e-06, grad_fn=)
742 tensor(2.6882e-06, grad_fn=)
743 tensor(2.6493e-06, grad_fn=)
744 tensor(2.6119e-06, grad_fn=)
745 tensor(2.5735e-06, grad_fn=)
746 tensor(2.5365e-06, grad_fn=)
747 tensor(2.4999e-06, grad_fn=)
748 tensor(2.4643e-06, grad_fn=)
749 tensor(2.4291e-06, grad_fn=)
750 tensor(2.3945e-06, grad_fn=)
751 tensor(2.3597e-06, grad_fn=)
752 tensor(2.3259e-06, grad_fn=)
753 tensor(2.2930e-06, grad_fn=)
754 tensor(2.2595e-06, grad_fn=)
755 tensor(2.2271e-06, grad_fn=)
756 tensor(2.1945e-06, grad_fn=)
757 tensor(2.1638e-06, grad_fn=)
758 tensor(2.1319e-06, grad_fn=)
759 tensor(2.1017e-06, grad_fn=)
760 tensor(2.0715e-06, grad_fn=)
761 tensor(2.0416e-06, grad_fn=)
762 tensor(2.0125e-06, grad_fn=)
763 tensor(1.9837e-06, grad_fn=)
764 tensor(1.9549e-06, grad_fn=)
765 tensor(1.9267e-06, grad_fn=)
766 tensor(1.8990e-06, grad_fn=)
767 tensor(1.8715e-06, grad_fn=)
768 tensor(1.8455e-06, grad_fn=)
769 tensor(1.8189e-06, grad_fn=)
770 tensor(1.7920e-06, grad_fn=)
771 tensor(1.7665e-06, grad_fn=)
772 tensor(1.7413e-06, grad_fn=)
773 tensor(1.7161e-06, grad_fn=)
774 tensor(1.6912e-06, grad_fn=)
775 tensor(1.6669e-06, grad_fn=)
776 tensor(1.6428e-06, grad_fn=)
777 tensor(1.6196e-06, grad_fn=)
778 tensor(1.5966e-06, grad_fn=)
779 tensor(1.5735e-06, grad_fn=)
780 tensor(1.5513e-06, grad_fn=)
781 tensor(1.5286e-06, grad_fn=)
782 tensor(1.5065e-06, grad_fn=)
783 tensor(1.4848e-06, grad_fn=)
784 tensor(1.4636e-06, grad_fn=)
785 tensor(1.4427e-06, grad_fn=)
786 tensor(1.4220e-06, grad_fn=)
787 tensor(1.4017e-06, grad_fn=)
788 tensor(1.3813e-06, grad_fn=)
789 tensor(1.3615e-06, grad_fn=)
790 tensor(1.3423e-06, grad_fn=)
791 tensor(1.3228e-06, grad_fn=)
792 tensor(1.3034e-06, grad_fn=)
793 tensor(1.2846e-06, grad_fn=)
794 tensor(1.2665e-06, grad_fn=)
795 tensor(1.2480e-06, grad_fn=)
796 tensor(1.2299e-06, grad_fn=)
797 tensor(1.2127e-06, grad_fn=)
798 tensor(1.1951e-06, grad_fn=)
799 tensor(1.1780e-06, grad_fn=)
800 tensor(1.1611e-06, grad_fn=)
801 tensor(1.1443e-06, grad_fn=)
802 tensor(1.1276e-06, grad_fn=)
803 tensor(1.1115e-06, grad_fn=)
804 tensor(1.0956e-06, grad_fn=)
805 tensor(1.0803e-06, grad_fn=)
806 tensor(1.0645e-06, grad_fn=)
807 tensor(1.0493e-06, grad_fn=)
808 tensor(1.0337e-06, grad_fn=)
809 tensor(1.0188e-06, grad_fn=)
810 tensor(1.0043e-06, grad_fn=)
811 tensor(9.9009e-07, grad_fn=)
812 tensor(9.7585e-07, grad_fn=)
813 tensor(9.6172e-07, grad_fn=)
814 tensor(9.4808e-07, grad_fn=)
815 tensor(9.3415e-07, grad_fn=)
816 tensor(9.2043e-07, grad_fn=)
817 tensor(9.0748e-07, grad_fn=)
818 tensor(8.9467e-07, grad_fn=)
819 tensor(8.8207e-07, grad_fn=)
820 tensor(8.6901e-07, grad_fn=)
821 tensor(8.5616e-07, grad_fn=)
822 tensor(8.4452e-07, grad_fn=)
823 tensor(8.3222e-07, grad_fn=)
824 tensor(8.1991e-07, grad_fn=)
825 tensor(8.0863e-07, grad_fn=)
826 tensor(7.9690e-07, grad_fn=)
827 tensor(7.8522e-07, grad_fn=)
828 tensor(7.7413e-07, grad_fn=)
829 tensor(7.6317e-07, grad_fn=)
830 tensor(7.5173e-07, grad_fn=)
831 tensor(7.4128e-07, grad_fn=)
832 tensor(7.3041e-07, grad_fn=)
833 tensor(7.2006e-07, grad_fn=)
834 tensor(7.0934e-07, grad_fn=)
835 tensor(6.9953e-07, grad_fn=)
836 tensor(6.8945e-07, grad_fn=)
837 tensor(6.7978e-07, grad_fn=)
838 tensor(6.6980e-07, grad_fn=)
839 tensor(6.6050e-07, grad_fn=)
840 tensor(6.5067e-07, grad_fn=)
841 tensor(6.4127e-07, grad_fn=)
842 tensor(6.3218e-07, grad_fn=)
843 tensor(6.2292e-07, grad_fn=)
844 tensor(6.1405e-07, grad_fn=)
845 tensor(6.0524e-07, grad_fn=)
846 tensor(5.9650e-07, grad_fn=)
847 tensor(5.8817e-07, grad_fn=)
848 tensor(5.7977e-07, grad_fn=)
849 tensor(5.7087e-07, grad_fn=)
850 tensor(5.6281e-07, grad_fn=)
851 tensor(5.5460e-07, grad_fn=)
852 tensor(5.4687e-07, grad_fn=)
853 tensor(5.3920e-07, grad_fn=)
854 tensor(5.3125e-07, grad_fn=)
855 tensor(5.2365e-07, grad_fn=)
856 tensor(5.1614e-07, grad_fn=)
857 tensor(5.0869e-07, grad_fn=)
858 tensor(5.0125e-07, grad_fn=)
859 tensor(4.9452e-07, grad_fn=)
860 tensor(4.8731e-07, grad_fn=)
861 tensor(4.7995e-07, grad_fn=)
862 tensor(4.7305e-07, grad_fn=)
863 tensor(4.6659e-07, grad_fn=)
864 tensor(4.5979e-07, grad_fn=)
865 tensor(4.5303e-07, grad_fn=)
866 tensor(4.4660e-07, grad_fn=)
867 tensor(4.4021e-07, grad_fn=)
868 tensor(4.3390e-07, grad_fn=)
869 tensor(4.2757e-07, grad_fn=)
870 tensor(4.2136e-07, grad_fn=)
871 tensor(4.1553e-07, grad_fn=)
872 tensor(4.0936e-07, grad_fn=)
873 tensor(4.0351e-07, grad_fn=)
874 tensor(3.9780e-07, grad_fn=)
875 tensor(3.9225e-07, grad_fn=)
876 tensor(3.8633e-07, grad_fn=)
877 tensor(3.8064e-07, grad_fn=)
878 tensor(3.7528e-07, grad_fn=)
879 tensor(3.7006e-07, grad_fn=)
880 tensor(3.6474e-07, grad_fn=)
881 tensor(3.5949e-07, grad_fn=)
882 tensor(3.5411e-07, grad_fn=)
883 tensor(3.4918e-07, grad_fn=)
884 tensor(3.4428e-07, grad_fn=)
885 tensor(3.3925e-07, grad_fn=)
886 tensor(3.3443e-07, grad_fn=)
887 tensor(3.2940e-07, grad_fn=)
888 tensor(3.2471e-07, grad_fn=)
889 tensor(3.1999e-07, grad_fn=)
890 tensor(3.1540e-07, grad_fn=)
891 tensor(3.1098e-07, grad_fn=)
892 tensor(3.0668e-07, grad_fn=)
893 tensor(3.0209e-07, grad_fn=)
894 tensor(2.9776e-07, grad_fn=)
895 tensor(2.9356e-07, grad_fn=)
896 tensor(2.8954e-07, grad_fn=)
897 tensor(2.8514e-07, grad_fn=)
898 tensor(2.8078e-07, grad_fn=)
899 tensor(2.7710e-07, grad_fn=)
900 tensor(2.7301e-07, grad_fn=)
901 tensor(2.6911e-07, grad_fn=)
902 tensor(2.6535e-07, grad_fn=)
903 tensor(2.6150e-07, grad_fn=)
904 tensor(2.5753e-07, grad_fn=)
905 tensor(2.5377e-07, grad_fn=)
906 tensor(2.5029e-07, grad_fn=)
907 tensor(2.4687e-07, grad_fn=)
908 tensor(2.4316e-07, grad_fn=)
909 tensor(2.3979e-07, grad_fn=)
910 tensor(2.3599e-07, grad_fn=)
911 tensor(2.3286e-07, grad_fn=)
912 tensor(2.2948e-07, grad_fn=)
913 tensor(2.2618e-07, grad_fn=)
914 tensor(2.2304e-07, grad_fn=)
915 tensor(2.1978e-07, grad_fn=)
916 tensor(2.1657e-07, grad_fn=)
917 tensor(2.1329e-07, grad_fn=)
918 tensor(2.1050e-07, grad_fn=)
919 tensor(2.0734e-07, grad_fn=)
920 tensor(2.0441e-07, grad_fn=)
921 tensor(2.0143e-07, grad_fn=)
922 tensor(1.9854e-07, grad_fn=)
923 tensor(1.9585e-07, grad_fn=)
924 tensor(1.9298e-07, grad_fn=)
925 tensor(1.9008e-07, grad_fn=)
926 tensor(1.8725e-07, grad_fn=)
927 tensor(1.8464e-07, grad_fn=)
928 tensor(1.8206e-07, grad_fn=)
929 tensor(1.7948e-07, grad_fn=)
930 tensor(1.7686e-07, grad_fn=)
931 tensor(1.7432e-07, grad_fn=)
932 tensor(1.7181e-07, grad_fn=)
933 tensor(1.6929e-07, grad_fn=)
934 tensor(1.6700e-07, grad_fn=)
935 tensor(1.6451e-07, grad_fn=)
936 tensor(1.6226e-07, grad_fn=)
937 tensor(1.5999e-07, grad_fn=)
938 tensor(1.5751e-07, grad_fn=)
939 tensor(1.5528e-07, grad_fn=)
940 tensor(1.5307e-07, grad_fn=)
941 tensor(1.5087e-07, grad_fn=)
942 tensor(1.4869e-07, grad_fn=)
943 tensor(1.4652e-07, grad_fn=)
944 tensor(1.4455e-07, grad_fn=)
945 tensor(1.4241e-07, grad_fn=)
946 tensor(1.4029e-07, grad_fn=)
947 tensor(1.3834e-07, grad_fn=)
948 tensor(1.3642e-07, grad_fn=)
949 tensor(1.3432e-07, grad_fn=)
950 tensor(1.3243e-07, grad_fn=)
951 tensor(1.3053e-07, grad_fn=)
952 tensor(1.2865e-07, grad_fn=)
953 tensor(1.2680e-07, grad_fn=)
954 tensor(1.2511e-07, grad_fn=)
955 tensor(1.2306e-07, grad_fn=)
956 tensor(1.2139e-07, grad_fn=)
957 tensor(1.1978e-07, grad_fn=)
958 tensor(1.1797e-07, grad_fn=)
959 tensor(1.1618e-07, grad_fn=)
960 tensor(1.1456e-07, grad_fn=)
961 tensor(1.1309e-07, grad_fn=)
962 tensor(1.1124e-07, grad_fn=)
963 tensor(1.0960e-07, grad_fn=)
964 tensor(1.0818e-07, grad_fn=)
965 tensor(1.0650e-07, grad_fn=)
966 tensor(1.0508e-07, grad_fn=)
967 tensor(1.0354e-07, grad_fn=)
968 tensor(1.0215e-07, grad_fn=)
969 tensor(1.0061e-07, grad_fn=)
970 tensor(9.9084e-08, grad_fn=)
971 tensor(9.7589e-08, grad_fn=)
972 tensor(9.6412e-08, grad_fn=)
973 tensor(9.4883e-08, grad_fn=)
974 tensor(9.3581e-08, grad_fn=)
975 tensor(9.2128e-08, grad_fn=)
976 tensor(9.0809e-08, grad_fn=)
977 tensor(8.9499e-08, grad_fn=)
978 tensor(8.8320e-08, grad_fn=)
979 tensor(8.7029e-08, grad_fn=)
980 tensor(8.5747e-08, grad_fn=)
981 tensor(8.4593e-08, grad_fn=)
982 tensor(8.3363e-08, grad_fn=)
983 tensor(8.2109e-08, grad_fn=)
984 tensor(8.0897e-08, grad_fn=)
985 tensor(7.9743e-08, grad_fn=)
986 tensor(7.8680e-08, grad_fn=)
987 tensor(7.7445e-08, grad_fn=)
988 tensor(7.6349e-08, grad_fn=)
989 tensor(7.5276e-08, grad_fn=)
990 tensor(7.4195e-08, grad_fn=)
991 tensor(7.3169e-08, grad_fn=)
992 tensor(7.2103e-08, grad_fn=)
993 tensor(7.1154e-08, grad_fn=)
994 tensor(7.0150e-08, grad_fn=)
995 tensor(6.8984e-08, grad_fn=)
996 tensor(6.7949e-08, grad_fn=)
997 tensor(6.7043e-08, grad_fn=)
998 tensor(6.6053e-08, grad_fn=)
999 tensor(6.5159e-08, grad_fn=)
w= 1.9998300075531006
b= 0.0003863169113174081
y_pred= tensor([7.9997])

Process finished with exit code 0

运行结果：

作业：pytorch中提供了很多优化器，试试不同的优化器的效果

以Adagrad为例，学习路为0.01，效果一般，可尝试增加训练次数或调整学习率

import torch
import matplotlib.pyplot as plt


x_data=torch.Tensor([[1.0],[2.0],[3.0]])
y_data=torch.Tensor([[2.0],[4.0],[6.0]])


class LinearModel(torch.nn.Module):#our module class inherit from nn.module(neural network module)
    def __init__(self):#构造函数，初始化
        super().__init__()#继承父类的init方法
        self.linear=torch.nn.Linear(1,1)#Class nn.Linear包含权重和偏置两个tensor

    def forward(self,x):#重写父类中forward函数
        y_pred=self.linear(x)
        return y_pred

model=LinearModel() # creat a instance of linearmodel.model是collable,即可以model（x）

#nn.MSELoss继承自nn.Module
criterion=torch.nn.MSELoss(size_average=False)
#选择优化器，lr是learing rate
optimizer=torch.optim.Adagrad(model.parameters(),lr=0.01)

l_list=[]
for epoch in range(100):
    y_pred=model(x_data) #fforwward:predict 算出y_hat
    loss=criterion(y_pred,y_data)#forward :loss
    print(epoch,loss.item())
    l_list.append(loss.item())

    optimizer.zero_grad()#before backard，梯度归零
    loss.backward()#backward:autograd
    optimizer.step()#用step函数update更新

#output weight and bias
print('w=',model.linear.weight.item())
print('b=',model.linear.bias.item())

#test model
x_test=torch.tensor([4.0])
y_test=model(x_test)
print('y_pred=',y_test.data)

x = range(100)
plt.plot(x, l_list)
plt.xlabel('epoch')
plt.ylabel('loss')
plt.show()

运行结果如图所示：

《PyTorch深度学习实践》-P5线性回归

Python相关栏目本月热门文章