神經網路學習 PyTorch學習03 搭建模型

torch.nn

(1)用於搭建網路結構的序列容器：torch.nn.sequential 

models =torch.nn.sequential(

torch.nn.linear(input_data, hidden_layer),
torch.nn.relu(),
torch.nn.linear(hidden_layer, output_data)
)from collections import ordereddict  #
使用有序字典 使模組有自定義的名次
models2 =torch.nn.sequential(ordereddict([
("line1
",torch.nn.linear(input_data, hidden_layer)),
("relu1
",torch.nn.relu()),
("line2
",torch.nn.linear(hidden_layer, output_data))])
)

(2)線性層：torch.nn.linear

(3)啟用函式：torch.nn.relu

(4)損失函式：torch.nn.mseloss(均方誤差函式)，troch.nn.l1loss(平均絕對誤差函式)，torch.nn.crossentropyloss(交叉熵)

import

torch
from torch.autograd import
variable
batch_n = 100hidden_layer = 100input_data = 1000output_data = 10x = variable(torch.randn(batch_n, input_data), requires_grad=false)  #
x封裝為節點，設定為不自動求導
y = variable(torch.randn(batch_n, output_data), requires_grad=false)
models =torch.nn.sequential(
torch.nn.linear(input_data, hidden_layer),
torch.nn.relu(),
torch.nn.linear(hidden_layer, output_data))#
from collections import ordereddict  # 使用有序字典 使模組有自定義的名次
#models2 = torch.nn.sequential(ordereddict([
#("line1",torch.nn.linear(input_data, hidden_layer)),
#("relu1",torch.nn.relu()),
#("line2",torch.nn.linear(hidden_layer, output_data))])#)
epoch_n = 10000learning_rate = 0.0001loss_fn =torch.nn.mseloss()
for epoch in
range(epoch_n):
y_pred =models(x)
loss =loss_fn(y_pred,y)
if epoch%1000 ==0:
print("
epoch:{},loss:
".format(epoch,loss.data[0]))
models.zero_grad()  
#梯度歸零
loss.backward()
for param in models.parameters():  #
遍歷節點引數更新
param.data -= param.grad.data*learning_rate

torch.optim包

引數自動優化類：sgd，adagrad，rmsprop，adam

import

torch
from torch.autograd import
variable
batch_n = 100hidden_layer = 100input_data = 1000output_data =10x = variable(torch.randn(batch_n, input_data), requires_grad=false)
y = variable(torch.randn(batch_n, output_data), requires_grad=false)
models =torch.nn.sequential(
torch.nn.linear(input_data,hidden_layer),
torch.nn.relu(),
torch.nn.linear(hidden_layer,output_data)
)epoch_n = 20learning_rate = 0.0001loss_fn =torch.nn.mseloss()
optimzer = torch.optim.adam(models.parameters(), lr=learning_rate)  #
torch.optim.adam對梯度更新使用到的學習率進行自適應調節
for epoch in
range(epoch_n):
y_pred =models(x)
loss =loss_fn(y_pred,y)
print("
eproch:{},loss:
".format(epoch,loss.data[0]))
optimzer.zero_grad()  
#引數梯度歸零
loss.backward()
optimzer.step()  
#節點引數更新

PyTorch學習之神經網路

神經網路可以通過torch.nn包來構建，它是基於自動梯度來構建一些模型，乙個nn.module包括層和乙個方法forward input 同時也會返回乙個輸出 output 下圖是乙個簡單的前饋神經網路lenet，乙個簡單的神經網路包括一下幾點 super net,self 首先找到 net 的父...

pytorch入門學習（三） 神經網路

神經網路可以使用torch.nn包構建.autograd實現了反向傳播功能,但是直接用來寫深度學習的 在很多情況下還是稍顯複雜,torch.nn是專門為神經網路設計的模組化介面.nn構建於autograd之上,可用來定義和執行神經網路.nn.module是nn中最重要的類,可把它看成是乙個網路的封裝...

pytorch學習 構建卷積神經網路

本文是對隧道 https org tutorials beginner blitz neural networks tutorial.html sphx glr beginner blitz neural networks tutorial py 的總結。其中 部分按照自己的習慣有所變動。構建神經網...