Under this model training the model to do things you don’t want and then “jailbreaking” it afterward would be a way to prevent classes of behavior.
Under this model training the model to do things you don’t want and then “jailbreaking” it afterward would be a way to prevent classes of behavior.