pytorch;优化器获得一个空参数列表

Question

是使用Pytorch的深度学习概念中的新手，并尝试构建二进制分类器模型。我在堆栈溢出上尝试了一些解决方案，但我似乎无法解决。也许是由于我的代码的性质。有人可以弄清楚我的代码中可能导致此错误的原因吗 这是我的代码，

import torch
import torch.nn as nn
import numpy as np
from sklearn.datasets import make_blobs
import matplotlib.pyplot as pyp

# creating a dummy dataset from the make_blobs dataset
number_of_samples=5000
#divide the dataset into training(80%) and testing(20%)
training_number=int(number_of_samples*0.8)
#creating the dummy datasest
x,y=make_blobs(n_samples=number_of_samples,centers=2,n_features=64,cluster_std=10,random_state=2020)
y=y.reshape(-1,1)
#converting the numpy arrays into torch tensors
x,y=torch.from_numpy(x),torch.from_numpy(y)
x,y=x.float(),y.float()
#splitting the datasets into training and testing
x_train,x_test=x[:training_number],x[training_number:]
y_train,y_test=y[:training_number],y[training_number:]

#printing the shapes of each dataset
print("x_train shape:",x_train.shape)
print("x_test shape:",x_test.shape)
print("y_train shape:",y_train.shape)
print("y_test shape:",y_test.shape)

#a class to define the neural network us torch nn module
#neural network will have 3 hidden layers and 1 output layer
#hidden layers will have 64,256 and 1024 neurons
#output layer will have a single neuron
class neuralnetwork(nn.Module):
    def _init_(self):
        super().__init__()
        torch.manual_seed(2020)
        self.fc1 = nn.Linear(64, 256)
        self.relu1 = nn.ReLU()
        self.fc2 = nn.Linear(256, 1024)
        self.relu2 = nn.ReLU()
        self.out = nn.Linear(1024, 1)
        self.final = nn.Sigmoid()
    
    def forward(self, x):
        op = self.fc1(x) 
        op = self.relu1(op)
        op = self.fc2(op)
        op = self.relu2(op)
        op = self.out(op)
        y = self.final(op)
        return y

#defining the loss,optimizer and training function for the neural network
def train_network(model,optimizer,loss_function,num_epochs,batch_size,x_train,y_train):
    #start model training
    model.train()
    loss_for_every_epoch=nn.ModuleList()
    for epoch in range(num_epochs):
        train_loss=0.0
        for i in range(0,x_train.shape[0],batch_size):
            #extract train batch from x and y
            input_data=x_train[i:min(x_train.shape[0]),i+batch_size]
            labels=y_train[i:min(y_train.shape[0]),i+batch_size]
            #set gradients to zero before beginning optimization
            optimizer.zero_grad()
            #forwad pass 
            output_data=model(input_data)
            #calculate loss
            loss=loss_function(output_data,labels)
            #backpropagate
            loss.backward()
            #update weights
            optimizer.step()
            train_loss+=loss.item()*batch_size
        print("Epoch: {} - Loss:{:.4f}".format(epoch+1,train_loss )) 
        loss_for_every_epoch.extend([train_loss])
    #predict
    y_test_prediction=model(x_test)
    a=np.where(y_test_prediction>0.5,1,0)
    return loss_for_every_epoch

#create an object of the class
model=neuralnetwork()
#define the loss function
loss_function = nn.BCELoss()#binary cross entropy loss function
#define optimizer
adam_optimizer=torch.optim.Adam(params=model.parameters(),lr=0.001)
#define epochs and batch size
number_of_epochs=100
batch_size=16
#Calling the function for training and pass model, optimizer, loss and related paramters
adam_loss=train_network(model,adam_optimizer,loss_function,number_of_epochs,batch_size,x_train,y_train)

我得到错误

Value error:optimizer got an empty parameter list

，错误主要是从代码的这一部分生成的，

 #create an object of the class
    model=neuralnetwork()
    #define the loss function
    loss_function = nn.BCELoss()#binary cross entropy loss function
    #define optimizer
    adam_optimizer=torch.optim.Adam(params=model.parameters(),lr=0.001)
    #define epochs and batch size
    number_of_epochs=100
    batch_size=16
    #Calling the function for training and pass model, optimizer, loss and related paramters
    adam_loss=train_network(model,adam_optimizer,loss_function,number_of_epochs,batch_size,x_train,y_train)

这可能是我的代码中的原因。 这是完整的堆栈交易

    Traceback (most recent call last)
g:\My Drive\CODE\pythondatascience\simpleneuralnetwork.ipynb Cell 7' in <cell line: 6>()
      4 loss_function = nn.BCELoss()#binary cross entropy loss function
      5 #define optimizer
----> 6 adam_optimizer=torch.optim.Adam(params=model.parameters(),lr=0.001)
      7 #define epochs and batch size
      8 number_of_epochs=100

File c:\Users\DAVE\AppData\Local\Programs\Python\Python310\lib\site-packages\torch\optim\adam.py:81, in Adam.__init__(self, params, lr, betas, eps, weight_decay, amsgrad, maximize)
     78     raise ValueError("Invalid weight_decay value: {}".format(weight_decay))
     79 defaults = dict(lr=lr, betas=betas, eps=eps,
     80                 weight_decay=weight_decay, amsgrad=amsgrad, maximize=maximize)
---> 81 super(Adam, self).__init__(params, defaults)

File c:\Users\DAVE\AppData\Local\Programs\Python\Python310\lib\site-packages\torch\optim\optimizer.py:49, in Optimizer.__init__(self, params, defaults)
     47 param_groups = list(params)
     48 if len(param_groups) == 0:
---> 49     raise ValueError("optimizer got an empty parameter list")
     50 if not isinstance(param_groups[0], dict):
     51     param_groups = [{'params': param_groups}]

ValueError: optimizer got an empty parameter list

Answer 1

它应该是def __init __，而不是def _init _ in neuralnetwork类。您根本不会初始化模型对象。因此，它没有任何参数。