1.引入模块，读取数据
2.构建计算图（构建网络模型）
3.损失函数与优化器
4.开始训练模型
5.对训练的模型预测结果进行评估
1 import torch.nn.functional as F 2 import torch.nn.init as init 3 import torch 4 from torch.autograd import Variable 5 import matplotlib.pyplot as plt 6 import numpy as np 7 import math 8 %matplotlib inline 9 #%matplotlib inline 可以在Ipython编译器里直接使用 10 #功能是可以内嵌绘图，并且可以省略掉plt.show()这一步。 11 12 xy=np.loadtxt('./data/diabetes.csv.gz',delimiter=',',dtype=np.float32) 13 x_data=torch.from_numpy(xy[:,0:-1])#取除了最后一列的数据 14 y_data=torch.from_numpy(xy[:,[-1]])#取最后一列的数据，[-1]加中括号是为了keepdim 15 16 print(x_data.size(),y_data.size()) 17 #print(x_data.shape,y_data.shape) 18 19 #建立网络模型 20 class Model(torch.nn.Module): 21 22 def __init__(self): 23 super(Model,self).__init__() 24 self.l1=torch.nn.Linear(8,6) 25 self.l2=torch.nn.Linear(6,4) 26 self.l3=torch.nn.Linear(4,1) 27 28 def forward(self,x): 29 out1=F.relu(self.l1(x)) 30 out2=F.dropout(out1,p=0.5) 31 out3=F.relu(self.l2(out2)) 32 out4=F.dropout(out3,p=0.5) 33 y_pred=F.sigmoid(self.l3(out3)) 34 return y_pred 35 36 def weights_init(m): 37 classname=m.__class__.__name__ 38 if classname.find('Linear')!=-1: 39 m.weight.data=torch.randn(m.weight.data.size()[0],m.weight.data.size()[1]) 40 m.bias.data=torch.randn(m.bias.data.size()[0]) 41 42 #our model 43 model=Model() 44 model.apply(weights_init) 45 criterion=torch.nn.BCELoss() 46 optimizer=torch.optim.SGD(model.parameters(),lr=0.1) 47 48 #training loop 49 Loss=[] 50 for epoch in range(2000): 51 y_pred=model(x_data) 52 loss=criterion(y_pred,y_data) 53 if epoch%100 == 0: 54 print("epoch = ",epoch," loss = ",loss.data) 55 Loss.append(loss.data) 56 optimizer.zero_grad() 57 loss.backward() 58 optimizer.step() 59 60 hour_var = Variable(torch.randn(1,8)) 61 print("predict",model(hour_var).data[0]>0.5) 62 plt.plot(Loss)
这里说明一下，这个dataset不是自带的，需要大家自己去下载，我找的时候费了不少功夫，这里提供一个网址给大家下载https://github.com/LianHaiMiao/pytorch-lesson-zh/blob/master/dataSet/diabetes.csv.gz
参考：https://blog.csdn.net/qq_35547281/article/details/89285980