PyTorch實現線性迴歸

import torch
import torch.nn as nn 
import numpy as np
from torch.autograd import Variable
# 定義模型
class LinearModel(nn.Module):
	def __init__(self,ndim):
		super(LinearModel,self).__init__()
		self.ndim=ndim
		self.weight=nn.Parameter(torch.randn(ndim,1))#定義權重
		self.bias=nn.Parameter(torch.randn(1))#定義偏置

	def forward(self,x):
		#y=Wx+b
		return x.mm(self.weight)+self.bias
    
# 生成數據
num_inputs = 2
num_examples = 1600
true_w = [2, -3.1]
true_b = 4.2
features = torch.randn(num_examples, num_inputs,requires_grad=True)
labels = true_w[0] * features[:, 0] + true_w[1] * features[:, 1] + true_b
labels += torch.tensor(np.random.normal(0, 0.01, size=labels.size()),
                       dtype=torch.float32)
print("labels")
print(labels)
print(true_w[0] * features[:, 0] + true_w[1] * features[:, 1] + true_b)
# 選擇模型
#model = LinearModel(num_inputs)
model = nn.Linear(2, 1,bias=True)
#定義損失函數和優化器（loss和optimizer）
from torch import optim
criterion = nn.MSELoss()# 交叉熵損失函數
optimizer = optim.SGD(model.parameters(),lr=0.01,momentum=0.9)

# 處理數據
trainloader  = torch.utils.data.DataLoader(
torch.utils.data.TensorDataset(features,labels),
batch_size = 1,
shuffle=True,
num_workers=0)

# 訓練模型
for epoch in range(3):
    running_loss = 0.0
    for i, data in enumerate(trainloader, 0):
        # 輸入數據
        inputs, labels = data
        # 梯度清零
        inputs = Variable(inputs)
        labels = Variable(labels)
        optimizer.zero_grad()
        outputs = model(inputs)
       
        # 計算損失
        loss = criterion(outputs, labels)
        
        # 反向求導
        loss.backward()
        
        # 更新參數
        optimizer.step()
        
        # 打印log信息
        running_loss += loss.item()
        if i % 100 == 99: # 每100個batch打印一次訓練狀態
            #print("模型參數")
            #print(list(model.named_parameters()))#轉換生成器爲列表
            print('{%d, %5d} loss: %.3f'  % (epoch+1, i+1 ,running_loss / 100))
            running_loss = 0.0
print('Finished Training')

print(model.named_parameters()) #獲取模型參數（帶名字）的生成器
print(list(model.named_parameters()))#轉換生成器爲列表
print(model.parameters())#獲取模型參數（帶名字）的生成器