@misc{indiciae799c82981aaf,
  title = {Layer-Aware Representation Filtering: Purifying Finetuning Data to Preserve LLM Safety Alignment},
  author = {Hao Li and Lijun Li and Zhenghao Lu and Xianyi Wei and Rui Li and Jing Shao and Lei Sha},
  year = {2025},
  url = {https://arxiv.org/abs/2507.18631},
  note = {Source identifier: 2507.18631}
}