1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
| 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(reduction = 'sum')
optimizer = torch.optim.SGD(model.parameters(), lr = 0.01)
for epoch in range(100):
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)
|