@inproceedings{lu-etal-2026-beyond-one,
    title = "Beyond One-Hot Replay: Self-Distillation Reduces Target Variance in Continual Pre-Training",
    author = "Lu, Peng  and
      Dong, Sijie  and
      Feng, Xiaoyun  and
      Hu, Xuanwei  and
      Zhang, Xin",
    editor = "Lee, Hung-Yi  and
      Dev, Sunipa  and
      Li, Jessy  and
      Elliott, Desmond",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
    month = oct,
    year = "2026",
    address = "Budapest, Hungary",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.findings-emnlp.1299/",
    pages = "27473--27491",
    ISBN = "979-8-89176-462-0"
}