• 多层感知机的简洁实现
  • 定义模型
  • 读取数据并训练数据
  • 损失函数
  • 定义优化算法
• 小结

多层感知机的简洁实现

import  torch
from torch import nn
from torch.nn import init
import sys
import numpy as np
sys.path.append('..')
import d2lzh_pytorch as d2l

定义模型

num_inputs,num_outputs,num_hidden =784,10,256
net = nn.Sequential(d2l.FlattenLayer(),
                   nn.Linear(num_inputs,num_hidden),
                    nn.ReLU(),
                    nn.Linear(num_hidden,num_outputs),
                   )
for params in net.parameters():
    init.normal_(params,mean=0,std=0.01)
    

print(net.parameters)

<bound method Module.parameters of Sequential(
  (0): FlattenLayer()
  (1): Linear(in_features=784, out_features=256, bias=True)
  (2): ReLU()
  (3): Linear(in_features=256, out_features=10, bias=True)
)>

读取数据

batch_size= 256
train_iter,test_iter = d2l.get_fahsion_mnist(batch_size)

损失函数

loss = nn.CrossEntropyLoss()

定义优化算法

optimizer = torch.optim.SGD(net.parameters(),lr=0.5)
num_epochs =5

训练数据并验证测试集

d2l.train_ch3(net,train_iter,test_iter,loss,num_epochs,batch_size,None,None,optimizer)

epoch 1, loss 0.0031, train acc 0.707, test acc 0.810
epoch 2, loss 0.0019, train acc 0.821, test acc 0.810
epoch 3, loss 0.0017, train acc 0.843, test acc 0.835
epoch 4, loss 0.0015, train acc 0.858, test acc 0.840
epoch 5, loss 0.0014, train acc 0.865, test acc 0.853

小结

• 通过pytorch可以使用Sequential这样的写法简洁的实现多层感知机