@inproceedings{xi-etal-2025-samplemix,
    title = "{S}ample{M}ix: A Sample-wise Pre-training Data Mixing Strategy by Coordinating Data Quality and Diversity",
    author = "Xi, Xiangyu  and
      Kong, Deyang  and
      Yang, Jian  and
      Yang, Jiawei  and
      Chen, Zhengyu  and
      Wang, Wei  and
      Wang, Jingang  and
      Cai, Xunliang  and
      Zhang, Shikun  and
      Ye, Wei",
    editor = "Christodoulopoulos, Christos  and
      Chakraborty, Tanmoy  and
      Rose, Carolyn  and
      Peng, Violet",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
    month = nov,
    year = "2025",
    address = "Suzhou, China",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/author-page-yu-wang-polytechnic/2025.findings-emnlp.741/",
    doi = "10.18653/v1/2025.findings-emnlp.741",
    pages = "13736--13758",
    ISBN = "979-8-89176-335-7",
    abstract = "Existing pretraining data mixing methods for large language models (LLMs) typically follow a domain-wise methodology, a top-down process that first determines domain weights and then performs uniform data sampling across each domain. However, these approaches neglect significant inter-domain overlaps and commonalities, failing to control the global diversity of the constructed training dataset. Further, uniform sampling within domains ignores fine-grained sample-specific features, potentially leading to suboptimal data distribution. To address these shortcomings, we propose a novel sample-wise data mixture approach based on a bottom-up paradigm. This method performs global cross-domain sampling by systematically evaluating the quality and diversity of each sample, thereby dynamically determining the optimal domain distribution. Comprehensive experiments across multiple downstream tasks and perplexity assessments demonstrate that SampleMix surpasses existing domain-based methods. Meanwhile, SampleMix requires 1.4x to 2.1x fewer training steps to achieve the baselines' performance, highlighting the substantial potential of SampleMix to optimize pre-training data."
}