什么使良好的指令微调数据？一种上下文学习视角
What Makes Good Instruction-Tuning Data? An In-Context Learning Perspective
AI总结：
本文从上下文学习角度探讨有效指令微调数据的构成，提出基于加权上下文影响（wICI）的数据选择框架，通过实验验证样本难度与上下文影响的负相关性，并展示方法在多种模型和基准上的优越性能。
摘要：
指令微调数据集常包含大量冗余和低质量样本，需要有效的数据选择方法。我们提出基于加权上下文影响（wICI）的指令数据选择框架，该框架衡量每个候选示例如何减少指令遵循难度以降低语义相关 peers 的难度。通过系统实验，我们回答三个关键问题：从上下文视角看有效指令微调数据的构成，样本难度是否与上下文影响相关，以及上下文影响如何转化为指令微调效果。在多个模型和基准上的实验表明，我们的方法在受限数据预算下始终优于现有基线，同时实证显示样本难度与上下文影响呈负相关。
英文摘要：
Instruction-tuning datasets often contain substantial redundancy and low-quality samples, necessitating effective data selection methods. We propose an instruction data selection framework based on weighted in-context influence (wICI), which measures how effectively each candidate example reduces instruction-following difficulty for semantically related peers. Through systematic experiments, we address three key questions: what constitutes effective instruction tuning data from an in-context perspective, whether sample difficulty correlates with in-context influence, and how in-context influence translates to instruction tuning effectiveness. Experiments across multiple models and benchmarks demonstrate that our method consistently outperforms existing baselines under constrained data budgets, while empirically showing that sample difficulty negatively correlates with in-context influence.
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。
