线性回归Pytorch方法

news2024/11/15 16:28:48

借助 PyTorch 实现深度神经网络 - 线性回归 PyTorch 方法 - 第 3 周 | Coursera

随机梯度下降和数据加载器

在每个epoch中，使用一个样本进行更新的方法称为随机梯度下降，而使用所有样本进行更新的方法称为批量梯度下降。

随机梯度下降：

随机梯度下降的问题：近似损失快速波动

Pytorch实现：

import torch
from torch.utils.data import Dataset, DataLoader
import torch.nn as nn

# 定义数据集类
class Data(Dataset):
    def __init__(self):
        self.x = torch.arange(-3, 3, 0.1).view(-1, 1)
        self.y = 1 * self.x - 1 + 0.1 * torch.randn(self.x.size())
        self.len = self.x.shape[0]

    def __getitem__(self, index):
        return self.x[index], self.y[index]

    def __len__(self):
        return self.len

# 初始化数据和数据加载器
dataset = Data()
trainloader = DataLoader(dataset=dataset, batch_size=1, shuffle=True)

# 定义模型参数
w = torch.tensor(-12.0, requires_grad=True)
b = torch.tensor(-10.0, requires_grad=True)
lr = 0.01  # 学习率
criterion = nn.MSELoss()

# 定义前向传播函数
def forward(x):
    return w * x + b

# 训练模型并记录总损失
LOSS = []

def my_train_model(epochs):
    for epoch in range(epochs):
        total_loss = 0  # 每个epoch的总损失
        for x, y in trainloader:
            yhat = forward(x)
            loss = criterion(yhat, y)
            total_loss += loss.item()
            loss.backward()
            with torch.no_grad():
                w -= lr * w.grad
                b -= lr * b.grad
                w.grad.zero_()
                b.grad.zero_()
        LOSS.append(total_loss)
        print(f"Epoch {epoch + 1}/{epochs}, Loss: {total_loss}")

# 训练模型
my_train_model(10)

小批量梯度下降

允许处理更大的数据集（无法容纳在内存中）

总结：

import torch
from torch.utils.data import Dataset, DataLoader
import torch.nn as nn

# 定义数据集类
class Data(Dataset):
    def __init__(self):
        self.x = torch.arange(-3, 3, 0.1).view(-1, 1)
        self.y = 1 * self.x - 1 + 0.1 * torch.randn(self.x.size())
        self.len = self.x.shape[0]

    def __getitem__(self, index):
        return self.x[index], self.y[index]

    def __len__(self):
        return self.len

# 初始化数据和数据加载器
dataset = Data()
trainloader = DataLoader(dataset=dataset, batch_size=10, shuffle=True)

# 定义模型参数
w = torch.tensor(-15.0, requires_grad=True)
b = torch.tensor(-10.0, requires_grad=True)
lr = 0.1  # 学习率
criterion = nn.MSELoss()

# 定义前向传播函数
def forward(x):
    return w * x + b

# 训练模型并记录总损失
LOSS_MINI20 = []

def train_model_Mini20(epochs):
    for epoch in range(epochs):
        total_loss = 0  # 每个epoch的总损失
        for x, y in trainloader:
            yhat = forward(x)
            loss = criterion(yhat, y)
            total_loss += loss.item()
            loss.backward()
            with torch.no_grad():
                w -= lr * w.grad
                b -= lr * b.grad
                w.grad.zero_()
                b.grad.zero_()
        LOSS_MINI20.append(total_loss)
        print(f"Epoch {epoch + 1}/{epochs}, Loss: {total_loss}")

# 训练模型
train_model_Mini20(10)