面向指令微调的数据多样性度量：系统性分析与一种可靠指标
Measuring Data Diversity for Instruction Tuning: A Systematic Analysis and A Reliable Metric
AI总结：
本文系统分析了11种数据多样性度量方法，提出基于样本级新颖性的NovelSum指标，其与指令微调模型性能相关性达0.97，并据此设计了优于现有方法的数据选择策略。
摘要：
数据多样性对于大语言模型的指令微调至关重要。现有研究已经探索了多种多样性感知的数据选择方法，以构建高质量数据集并提升模型性能。然而，精确界定和度量数据多样性这一基础问题仍未得到充分探索，这限制了数据工程获得清晰的指导。为解决这一问题，我们对11种现有的多样性度量方法进行了系统分析，通过大量微调实验评估它们与模型性能之间的相关性。结果表明，一种可靠的多样性度量应当同时恰当地考虑样本间差异以及样本空间中的信息密度。在此基础上，我们提出了NovelSum，一种基于样本级“新颖性”的新多样性指标。在模拟数据和真实数据上的实验表明，NovelSum能够准确捕捉多样性的变化，并与指令微调后的模型性能达到0.97的相关性，凸显了其在指导数据工程实践中的价值。以NovelSum作为优化目标，我们进一步开发了一种贪心的、面向多样性的数据选择策略，该策略优于现有方法，验证了我们所提指标的有效性和实际意义。代码可在 https://github.com/UmeanNever/NovelSum 获取。
英文摘要：
Data diversity is crucial for the instruction tuning of large language models. Existing studies have explored various diversity-aware data selection methods to construct high-quality datasets and enhance model performance. However, the fundamental problem of precisely defining and measuring data diversity remains underexplored, limiting clear guidance for data engineering. To address this, we systematically analyze 11 existing diversity measurement methods by evaluating their correlation with model performance through extensive fine-tuning experiments. Our results indicate that a reliable diversity measure should properly account for both inter-sample differences and the information density in the sample space. Building on this, we propose NovelSum, a new diversity metric based on sample-level "novelty." Experiments on both simulated and real-world data show that NovelSum accurately captures diversity variations and achieves a 0.97 correlation with instruction-tuned model performance, highlighting its value in guiding data engineering practices. With NovelSum as an optimization objective, we further develop a greedy, diversity-oriented data selection strategy that outperforms existing approaches, validating both the effectiveness and practical significance of our metric. The code is available at https://github.com/UmeanNever/NovelSum.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
