import torch

x = torch.tensor(1.0)
y = torch.tensor(2.0)

w = torch.tensor(1.0, requires_grad=True)

#forward path and compute loss
y_hat = w*x
loss = (y_hat-y)**2

print(loss)

#backward path
loss.backward()
print(w.grad)

### update weights
### next forward and backwards