@inproceedings{he-etal-2025-right,
    title = "The Right Time Matters: Data Arrangement Affects Zero-Shot Generalization in Instruction Tuning",
    author = "He, Bingxiang  and
      Ding, Ning  and
      Qian, Cheng  and
      Deng, Jia  and
      Cui, Ganqu  and
      Yuan, Lifan  and
      Hong, Haiwen  and
      Gao, Huan-ang  and
      Huang, Longtao  and
      Xue, Hui  and
      Chen, Huimin  and
      Liu, Zhiyuan  and
      Sun, Maosong",
    editor = "Che, Wanxiang  and
      Nabende, Joyce  and
      Shutova, Ekaterina  and
      Pilehvar, Mohammad Taher",
    booktitle = "Findings of the Association for Computational Linguistics: ACL 2025",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingestion-acl-25/2025.findings-acl.13/",
    pages = "222--243",
    ISBN = "979-8-89176-256-5",
    abstract = "Understanding alignment techniques begins with comprehending zero-shot generalization brought by instruction tuning, but little of the mechanism has been understood. Existing work has largely been confined to the task level, without considering that tasks are artificially defined and, to LLMs, merely consist of tokens and representations. To bridge this gap, we investigate zero-shot generalization from the perspective of the data itself. We first demonstrate that zero-shot generalization happens very early during instruction tuning, with loss serving as a stable indicator. Next, we investigate training data arrangement through similarity and granularity perspectives, confirming that the timing of exposure to certain training examples may greatly facilitate generalization on unseen tasks. Finally, we propose a more grounded training data arrangement framework, Test-centric Multi-turn Arrangement, and show its effectiveness in promoting continual learning and further loss reduction. For the first time, we show that zero-shot generalization during instruction tuning is a form of similarity-based generalization between training and test data at the instance level."
}