@article{refusalfeatureguidedteacherforsafe,
  title = {Refusal-Feature-guided Teacher for Safe Finetuning via Data Filtering and Alignment Distillation},
  author = {Seokil Ham and Yubin Choi and Seungju Cho and YuJin Yang and Younghun Kim and Changick Kim},
  year = {2025},
  eprint = {2506.07356},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2506.07356v1},
}