arXiv · 2026 · Preprint

On-Policy Distillation for LLM Safety: A Routing Approach to Template-Robust Realignment

Yong-Jian Guo, Wanlun Ma, Lingyu Shen, Xi Xiao, Sheng Wen

Date
2026-07-29
Citations
0
arXiv
2607.27081
Cite
@misc{guo2026onpolicy,
  title = {On-Policy Distillation for LLM Safety: A Routing Approach to Template-Robust Realignment},
  author = {Yong-Jian Guo and Wanlun Ma and Lingyu Shen and Xi Xiao and Sheng Wen},
  year = {2026},
  eprint = {2607.27081},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2607.27081},
}