Implement both Adam with L2 regularization and AdamW (decoupled weight decay) to optimize weights for linear regression with MSE loss.
