How to add a new custom learnable parameter to the list of parameters that the optimizer updates?

[jodumagpi]

Is it possible to append a new learnable parameter to the optimizer? Say, for example, I modify the loss function to add a new learnable parameter for weighing, how can I be sure that this is updated by the optimizer?

[jodumagpi]

Nevermind. I just realized that the loss functions are already included as part of the model. Then, whatever learnable parameters that I declare inside the model will be automatically added to the parameter list that the optimizer updates.