Question

我对PyTorch很新，对神经网络来说还是比较新的我正在尝试构建一个可以猜出性别名称的神经网络，而且我基于判断国籍的PyTorch RNN教程。
我得到的代码运行没有错误，但损失几乎没有变化，让我认为权重没有更新...
这是我输入/输出/目标张量设置的问题吗？或者我的训练功能可能有问题？我很失落，任何帮助都会受到赞赏：cold_sweat：
这是我的代码：

from __future__ import unicode_literals, print_function, division  
from io import open  
import glob  
import unicodedata  
import string  
import torch  
import torchvision  
import torch.nn as nn  
import torch.optim as optim  
import random  
from torch.autograd import Variable  

"""------GLOBAL VARIABLES------"""

all_letters = string.ascii_letters + " .,;'"
num_letters = len(all_letters)
all_names = {}
genders = ["Female", "Male"]

"""-------DATA EXTRACTION------"""

def findFiles(path):
    return glob.glob(path)

def unicodeToAscii(s):
    return ''.join(
        c for c in unicodedata.normalize('NFD', s)
        if unicodedata.category(c) != 'Mn'
        and c in all_letters
    )

# Read a file and split into lines
def readLines(filename):
    lines = open(filename, encoding='utf-8').read().strip().split('\n')
    return [unicodeToAscii(line) for line in lines]

for file in findFiles("/home/andrew/PyCharm/PycharmProjects/CantStop/data/names/*.txt"):
    gender = file.split("/")[-1].split(".")[0]
    names = readLines(file)
    all_names[gender] = names

"""-----DATA INTERPRETATION-----"""

def nameToTensor(name):
    tensor = torch.zeros(len(name), 1, num_letters)
    for index, letter in enumerate(name):
        tensor[index][0][all_letters.find(letter)] = 1
    return tensor

def outputToGender(output):
    gender, gender_index = output.data.topk(1)
    if gender_index[0][0] == 0:
        return "Female"
    return "Male"

"""------NETWORK SETUP------"""

class Net(nn.Module):
    def __init__(self, input_size, hidden_size, output_size):
        super(Net, self).__init__()
        self.hidden_size = hidden_size
        #Layer 1
        self.Lin1 = nn.Linear(input_size+hidden_size, int((input_size+hidden_size)/2))
        self.ReLu1 = nn.ReLU()
        self.Batch1 = nn.BatchNorm1d(int((input_size+hidden_size)/2))
        #Layer 2
        self.Lin2 = nn.Linear(int((input_size+hidden_size)/2), output_size)
        self.ReLu2 = nn.ReLU()
        self.Batch2 = nn.BatchNorm1d(output_size)
        self.softMax = nn.LogSoftmax()
        #Hidden layer
        self.HidLin = nn.Linear(input_size+hidden_size, hidden_size)
        self.HidReLu = nn.ReLU()
        self.HidBatch = nn.BatchNorm1d(hidden_size)

    def forward(self, input, hidden):
        comb = torch.cat((input, hidden), 1)
        hidden = self.HidBatch(self.HidReLu(self.HidLin(comb)))
        output1 = self.Batch1(self.ReLu1(self.Lin1(comb)))
        output2 = self.softMax(self.Batch2(self.ReLu2(self.Lin2(output1))))
        return output2, hidden

    def initHidden(self):
        return Variable(torch.zeros(1, self.hidden_size))

NN = Net(num_letters, 128, 2)

"""------TRAINING------"""

def getRandomTrainingEx():
    gender = genders[random.randint(0, 1)]
    name = all_names[gender][random.randint(0, len(all_names[gender])-1)]
    gender_tensor = Variable(torch.LongTensor([genders.index(gender)]))
    name_tensor = Variable(nameToTensor(name))
    return gender_tensor, name_tensor, gender

def train(input, target):
    hidden = NN.initHidden()

    loss_func = nn.NLLLoss()

    alpha = 0.01

    NN.zero_grad()

    for i in range(input.size()[0]):
        output, hidden = NN(input[i], hidden)

    loss = loss_func(output, target)
    loss.backward()
    for w in NN.parameters():
        w.data.add_(-alpha, w.grad.data)

    return output, loss

for i in range(5000):
    gender_tensor, name_tensor, gender = getRandomTrainingEx()
    output, loss = train(name_tensor, gender_tensor)

    if i%500 == 0:
        print("Guess: %s, Correct: %s, Loss: %s" % (outputToGender(output), gender, loss.data[0]))

这是输出：

Guess: Male, Correct: Male, Loss: 0.6931471824645996
Guess: Male, Correct: Female, Loss: 0.7400936484336853
Guess: Male, Correct: Male, Loss: 0.6755779385566711
Guess: Female, Correct: Female, Loss: 0.6648257374763489
Guess: Male, Correct: Male, Loss: 0.6765623688697815
Guess: Female, Correct: Male, Loss: 0.7330614924430847
Guess: Female, Correct: Female, Loss: 0.6565149426460266
Guess: Male, Correct: Female, Loss: 0.6946508884429932
Guess: Female, Correct: Female, Loss: 0.6621525287628174
Guess: Male, Correct: Male, Loss: 0.6662092804908752

Process finished with exit code 0

Answer 1

我建议您将add_更改为sub_。 add_可能会让你远离最佳点。

w.data.sub_(f.grad.data * alpha)

因为，重量更新公式中有减法。

顺便提一下，尝试将alpha增加/减少到0.1 0.05 0.01。如果alpha太大，它可能会错过最佳点。如果alpha为小，则需要很长时间。

RNN参数没有更新？

1 个答案: