Isn’t this just an obvious consequence of the well known fact about LLMs that the more you constrain some subset of the variables the more you force the remaining ones to ever more extreme values?
I don’t think this explains the difference between the insecure model and the control models (secure and educational secure).
Isn’t this just an obvious consequence of the well known fact about LLMs that the more you constrain some subset of the variables the more you force the remaining ones to ever more extreme values?
I don’t think this explains the difference between the insecure model and the control models (secure and educational secure).