pytorch搭建全连接网络

import torch    

import random

import numpy as np

from torch.nn import init

import torch.optim as optim

import matplotlib.pyplot as plt

import pandas as pd

nethid=37

print(9//4*4)

df=pd.read_excel('1.xlsx')

w=df.values

def data_iter(batch_size,features,lables):

    num_example=len(features)

    indices=list(range(num_example))

    random.shuffle(indices)

    for i in range(0,num_example,batch_size):

        j = torch.LongTensor(indices[i: min(i + batch_size,num_example)])

        yield   features.index_select(0,j),lables.index_select(0,j)

class simple_net(torch.nn.Module):

    def __init__(self,imn,hid,out):

        super(simple_net,self).__init__()

        self.layer=torch.nn.Linear(imn,hid)

        self.layer1=torch.nn.Linear(hid,out)

    def forward(self,x):

        x=self.layer(x)

        """x=torch.nn.functional.sigmoid(x)"""

        x=torch.sigmoid(x)

        x=self.layer1(x)

        return x

num_example=50

net=simple_net(4,nethid,1)

num_inputs=1

feature=w[:,1:5]

feature=np.reshape(feature,(-1,4))

feature=torch.tensor(feature, dtype=torch.float)

lable=w[:,6]

lable=torch.tensor(lable,dtype=torch.float)

batch_size=3

loss = torch.nn.MSELoss()

optimizer = optim.SGD(net.parameters(), lr=0.000003)

num=0

sum=0

for epoch in range(10000):

    for X,y in data_iter(batch_size,feature,lable):

        output = net(X)

        l = loss(output, y.view(-1, 1))

        optimizer.zero_grad() # 梯度清零，等价于net.zero_grad()

        l.backward()

        optimizer.step()

    num=num+1

    if epoch%16==0:

        sum=0

    print('epoch %d, loss: %f' % (epoch, l.item()))

for epoch in range(155128):

    for X,y in data_iter(1000,feature,lable):

        output = net(X)

        l = loss(output, y.view(-1, 1))

        optimizer.zero_grad() # 梯度清零，等价于net.zero_grad()

        l.backward()

        optimizer.step()

    num=num+1

    if epoch%16==0:

        sum=0

    print('epoch %d, loss: %f' % (epoch, l.item()))