Linear_Regression with pytorch
import torch
x_data = torch.Tensor([[1.0],[2.0],[3.0]])
y_data = torch.Tensor([[2.0],[4.0],[6.0]])
class LinearModel(torch.nn.Module):
def __init__(self):
super(LinearModel,self).__init__()
self.linear = torch.nn.Linear(1,1)
def forward(self,x):
y_pred = self.linear(x)
return y_pred
model = LinearModel()
criterion = torch.nn.MSELoss(size_average=False)
optimizer = torch.optim.SGD(model.parameters(),lr=0.01)
for epoch in range(1000):
y_pred=model(x_data)
loss=criterion(y_pred,y_data)
print(epoch,loss.item())
optimizer.zero_grad()
loss.backward()
optimizer.step()
print('w=',model.linear.weight.item())
print('b=',model.linear.bias.item())
x_test=torch.Tensor([4.0])
y_test=model(x_test)
print('y_pred=',y_test.data)