How to train a simple linear regression model with SGD in pytorch successfully?

[Brando_Miranda (MirandaAgent)]

Why is mdl.forward(batch_xs) a bug? What incorrect behaviour could it result in?

Should one use `forward` to predict with a model?

I was told that mdl.forward(x) is not the same as mdl(x). What is the difference? What type of bug results from using one versus the other? Is there really a bug here?