002.从零开始实现softMax回归(pytorch)

1. 从零实现

0 导入需要的包

#000 导包
import torch
print(torch.__version__)
import torchvision
print(torchvision.__version__)
import torchvision.transforms as transforms
import d2l.torch as d2l
import numpy as np
import os
print("当前工作目录:", os.getcwd())
import sys
import torchtext
#------------运行结果---------------
# 1.12.0
# 0.13.0

1.数据的下载及预处理

#1. 数据集的下载和预处理
# 1.1 数据集合的下载
mnist_train = torchvision.datasets.FashionMNIST(
    root='/Users/w/PycharmProjects/DeepLearning_with_LiMu/datasets/FashionMnist', train=True, download=True,
    transform=transforms.ToTensor())
mnist_test = torchvision.datasets.FashionMNIST(
    root='/Users/w/PycharmProjects/DeepLearning_with_LiMu/datasets/FashionMnist', train=False, download=True,
    transform=transforms.ToTensor())
# 1.2 数据集的预处理
batch_size = 256
if sys.platform.startswith('win'):
    num_worker = 0
else:
    num_worker = 4
train_iter = torch.utils.data.DataLoader(mnist_train, batch_size=batch_size, shuffle=True, num_workers=num_worker)
test_iter  = torch.utils.data.DataLoader(mnist_test, batch_size=batch_size, shuffle=False, num_workers=num_worker)
#1.3 数据详情与分析
    # <class 'torchvision.datasets.mnist.FashionMNIST'>
print(type(mnist_train))
    # 60000 10000  训练集的数量6W条，测试集的数量1W条
print(len(mnist_train), len(mnist_test))
    # 绘制图像直观展示，原始数据
from CommonFunctionModule import data_iter
from CommonFunctionModule import get_fashion_mnist_label
from CommonFunctionModule import show_fashion_mnist
X, y = [], []
for i in range(10):
    X.append(mnist_train[i][0])
    y.append(mnist_train[i][1])
show_fashion_mnist(X, get_fashion_mnist_label(y))
feature, label = mnist_train[0]
print('feature.size() = ' + str(feature.size())  +'\n')
feature_trans = feature.reshape((-1,28*28))
print('feature_trans.size()' +str(feature_trans.size())  +'\n' )

<class ‘torchvision.datasets.mnist.FashionMNIST’>
60000 10000

2.定义模型

# 2.定义模型
## 2.1 参数初始化
    # 输入神经元的个数
num_input = 28 * 28
    # 输出神经元的个数
num_output = 10
    # 参数初始化， 样本矩阵： n*784  权重矩阵：784*10   X·w = n*10
w = torch.tensor(np.random.normal(loc=0, scale=0.01, size=(num_input, num_output)), dtype=torch.float32,requires_grad=True)
b = torch.zeros(num_output,dtype = torch.float32,requires_grad=True)
def net(X):
    return softmax(torch.mm(X.view((-1, num_input)), w) + b)
def softmax(X):
    X_exp = X.exp()
    X_sum = X_exp.sum(dim=1, keepdim=True)
    return X_exp / X_sum

3.定义损失函数

# 3.定义损失函数，使用交叉熵损失函数
def cross_entropy(y_hat,y):
    return - torch.log( y_hat.gather(dim = 1, index= y.view(-1,1)))

4.定义模型评估函数

# 4.定义模型评估
# 4.1 定义已知结果下模型评估
def accuracy(y_hat,y):
    return (y_hat.argmax(dim=1)==y).float().mean().item()
# t1 = torch.tensor([[1,2,3,4,5]])
# t2 = torch.tensor([1,2,3,4,6])
# ratio = accuracy(y_hat=t1,y = t2)
# print('测试精度函数：' + str(accuracy(t1,t2)) )
# 4.2 定义模型对，测试数据集的准确率
def evaluate_accuracy(data_iter,net):
    acc_sum,n = 0.0,0
    for X,y in data_iter:
        acc_sum +=  (net(X).argmax(dim=1)==y).float().sum().item()
        n+=y.shape[0]
    return acc_sum/n

# 测试evaluate_accuracy函数
evaluate_accuracy(test_iter,net)

5.模型训练

# 5.模型训练
num_epochs,lr =5,0.1
def train_fashion_mnist(net , train_iter , test_iter , loss ,num_epochs ,batch_size,params =None,lr =None, optimizer= None):
    for epoch in range(num_epochs):
        train_l_sum, train_acc_sum ,n = 0.0,0.0,0
        for X,y in train_iter:
            # 正向传播
            y_hat = net(X)
            # 计算损失函数
            l = loss(y_hat,y).sum()
            # 梯度清零
            if optimizer is not None:
                optimizer.zero_grad()
            elif  params is not None and params[0].grad is not None:
                for param in params:
                    param.grad.data.zero_()
            l.backward()
            if  optimizer is None:
                d2l.sgd(params,lr, batch_size)
            else:
                optimizer.step()
            # 累加求本轮损失综合
            train_l_sum += l.item()
            # 累加求精度总和
            train_acc_sum +=(y_hat.argmax(dim= 1) ==y).sum().item()
            # 计数
            n += y.shape[0]
        test_acc  =  evaluate_accuracy(test_iter,net)
        print('epoch %d ,loss %.4f , train acc %.3f ,test acc %.3f' % (epoch+1,train_l_sum/n,train_acc_sum/n,test_acc))
train_fashion_mnist(net,train_iter,test_iter,cross_entropy,5,batch_size,[w,b],lr)

epoch 1 ,loss 0.7869 , train acc 0.748 ,test acc 0.794
epoch 2 ,loss 0.5697 , train acc 0.813 ,test acc 0.805
epoch 3 ,loss 0.5254 , train acc 0.825 ,test acc 0.818
epoch 4 ,loss 0.5018 , train acc 0.831 ,test acc 0.825
epoch 5 ,loss 0.4852 , train acc 0.838 ,test acc 0.827

6.模型的预测结果及可视化

X,y = iter(test_iter).next()
true_labels = d2l.get_fashion_mnist_labels(y.numpy())
pred_labels = d2l.get_fashion_mnist_labels(net(X).argmax(dim=1).numpy())
import d2lzh_pytorch as d2l
titles = [true+'\n'+pred for true,pred in zip(true_labels,pred_labels)]
d2l.show_fashion_mnist(X[0:9],titles[0:9])