On-Policy Replay for Continual Supervised Fine-Tuning

@misc{chen2026onpolicy,
  title = {On-Policy Replay for Continual Supervised Fine-Tuning},
  author = {Yan Chen and Taojie Zhu and Meng Zhang and Xin Chen and Jiaqi Huang and D. Xu and Yizhi Wang},
  year = {2026},
  eprint = {2605.29495},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.29495},
}