It's incredibly robust to fine-tuning; often gradients don't flow through the digital locker, and even if you get this to happen, the amount of noise you can add to the weights of these cryptographic circuits before they stop working is often orders more than the amount of noise you can add to the language modeling parts of the language model before that stops working! So in short, this is often more robust than the part of the model that knows English. Same thing with abliteration; there's no "direction" in weight/activation space that you can subtract here.
alt Hacker News
It's incredibly robust to fine-tuning; often gradients don't flow through the digital locker, and even if you get this to happen, the amount of noise you can add to the weights of these cryptographic circuits before they stop working is often orders more than the amount of noise you can add to the language modeling parts of the language model before that stops working! So in short, this is often more robust than the part of the model that knows English. Same thing with abliteration; there's no "direction" in weight/activation space that you can subtract here.