Model checkpoints for Natively Unlearnable LLMs (https://arxiv.org/abs/2606.13873)