PyTorch学习笔记（四）Logistic回归

最新推荐文章于 2025-01-09 21:26:02 发布

wydxry

最新推荐文章于 2025-01-09 21:26:02 发布

阅读量106

点赞数

分类专栏： PyTorch

本文链接：https://blog.youkuaiyun.com/wydxry/article/details/119011031

版权

PyTorch 专栏收录该内容

13 篇文章

订阅专栏

# 导入相关库函数
import numpy as np
import torch
import torch.nn as nn
import torchvision
import torchvision.transforms as transforms

# 超参数
input_size =28*28
num_classes =10
num_epochs = 10
batch_size = 100
learning_rate = 0.001

# MNIST数据集
train_dataset = torchvision.datasets.MNIST(root='../../data',
                                         train=True,
                                         transform=transforms.ToTensor(),
                                         download=True)
test_dataset = torchvision.datasets.MNIST(root='../../data',
                                        train=False,
                                        transform=transforms.ToTensor())

# Data loader
train_loader = torch.utils.data.DataLoader(dataset=train_dataset,
                                          batch_size=batch_size,
                                          shuffle=True)
test_loader = torch.utils.data.DataLoader(dataset=test_dataset,
                                         batch_size=batch_size,
                                         shuffle=False)

# Logistic 回归模型
model = nn.Linear(input_size,num_classes)

# 损失函数和优化器
criterion = nn.CrossEntropyLoss()
optimizer = torch.optim.SGD(model.parameters(),lr=learning_rate)

# 训练模型
total_step = len(train_loader)
for epoch in range(num_epochs):
    for i,(images,labels) in enumerate(train_loader):
        # 转换图像格式
        images = images.reshape(-1,input_size)
        
        # 前向传播
        outputs = model(images)
        loss = criterion(outputs,labels)
        
        # 后向传播并优化
        optimizer.zero_grad()
        loss.backward()
        optimizer.step()
        
        if (i+1) % 100 == 0:
            print('Epoch [{}/{}], Step [{}/{}], Loss:{:.4f}'.
                 format(epoch+1,num_epochs,i+1,total_step,loss.item()))

Epoch [1/10], Step [100/600], Loss:2.1877
Epoch [1/10], Step [200/600], Loss:2.0952
Epoch [1/10], Step [300/600], Loss:2.0192
Epoch [1/10], Step [400/600], Loss:1.8990
Epoch [1/10], Step [500/600], Loss:1.8696
Epoch [1/10], Step [600/600], Loss:1.7758
Epoch [2/10], Step [100/600], Loss:1.6857
Epoch [2/10], Step [200/600], Loss:1.6319
Epoch [2/10], Step [300/600], Loss:1.6608
Epoch [2/10], Step [400/600], Loss:1.5398
Epoch [2/10], Step [500/600], Loss:1.4866
Epoch [2/10], Step [600/600], Loss:1.5251
Epoch [3/10], Step [100/600], Loss:1.3724
Epoch [3/10], Step [200/600], Loss:1.3550
Epoch [3/10], Step [300/600], Loss:1.4124
Epoch [3/10], Step [400/600], Loss:1.3413
Epoch [3/10], Step [500/600], Loss:1.1988
Epoch [3/10], Step [600/600], Loss:1.2809
Epoch [4/10], Step [100/600], Loss:1.2419
Epoch [4/10], Step [200/600], Loss:1.1801
Epoch [4/10], Step [300/600], Loss:1.2190
Epoch [4/10], Step [400/600], Loss:1.1891
Epoch [4/10], Step [500/600], Loss:1.1165
Epoch [4/10], Step [600/600], Loss:1.1057
Epoch [5/10], Step [100/600], Loss:0.9912
Epoch [5/10], Step [200/600], Loss:1.1478
Epoch [5/10], Step [300/600], Loss:0.9327
Epoch [5/10], Step [400/600], Loss:0.9662
Epoch [5/10], Step [500/600], Loss:0.8517
Epoch [5/10], Step [600/600], Loss:0.9587
Epoch [6/10], Step [100/600], Loss:0.9835
Epoch [6/10], Step [200/600], Loss:0.9946
Epoch [6/10], Step [300/600], Loss:0.8951
Epoch [6/10], Step [400/600], Loss:0.9013
Epoch [6/10], Step [500/600], Loss:0.9931
Epoch [6/10], Step [600/600], Loss:0.8686
Epoch [7/10], Step [100/600], Loss:0.9099
Epoch [7/10], Step [200/600], Loss:0.8394
Epoch [7/10], Step [300/600], Loss:0.9348
Epoch [7/10], Step [400/600], Loss:0.7706
Epoch [7/10], Step [500/600], Loss:1.0090
Epoch [7/10], Step [600/600], Loss:0.8198
Epoch [8/10], Step [100/600], Loss:0.8123
Epoch [8/10], Step [200/600], Loss:0.8300
Epoch [8/10], Step [300/600], Loss:0.8192
Epoch [8/10], Step [400/600], Loss:0.7725
Epoch [8/10], Step [500/600], Loss:0.8281
Epoch [8/10], Step [600/600], Loss:0.8399
Epoch [9/10], Step [100/600], Loss:0.8856
Epoch [9/10], Step [200/600], Loss:0.7889
Epoch [9/10], Step [300/600], Loss:0.7205
Epoch [9/10], Step [400/600], Loss:0.7489
Epoch [9/10], Step [500/600], Loss:0.8840
Epoch [9/10], Step [600/600], Loss:0.6865
Epoch [10/10], Step [100/600], Loss:0.7590
Epoch [10/10], Step [200/600], Loss:0.7337
Epoch [10/10], Step [300/600], Loss:0.7751
Epoch [10/10], Step [400/600], Loss:0.6656
Epoch [10/10], Step [500/600], Loss:0.6606
Epoch [10/10], Step [600/600], Loss:0.7351

# 测试模型
# 测试阶段，不需要计算梯度
with torch.no_grad():
    correct = 0
    total = 0
    for images,labels in test_loader:
        images = images.reshape(-1,input_size)
        outputs = model(images)
        _,predicted = torch.max(outputs.data,1)
        total += labels.size(0)
        correct += (predicted==labels).sum()
    
    print('Accuracy of the model on the 10000 test images:{} %'.
         format(100*correct/total))

Accuracy of the model on the 10000 test images:85.55999755859375 %

# 保存模型
torch.save(model.state_dict(),'model.ckpt')