MIG：通过最大化语义空间中的信息增益实现指令微调的自动数据选择
MIG: Automatic Data Selection for Instruction Tuning by Maximizing Information Gain in Semantic Space
AI总结：
针对指令微调数据选择中现有方法缺乏全面性且无法准确捕捉语义意图的问题，提出MIG方法，通过构建标签图量化语义空间信息并迭代选择样本以最大化信息增益，实验表明其性能优于现有方法且仅需5%数据即可达到官方模型效果。
摘要：
数据质量和多样性是构建有效指令微调数据集的关键。随着开源指令微调数据集的日益丰富，从海量数据中自动选择高质量且多样化的子集具有显著优势。现有方法通常优先考虑实例质量，并使用启发式规则维持多样性，但缺乏对整个数据集的全面了解，往往导致次优结果。此外，启发式规则通常关注嵌入空间内的距离或聚类，无法准确捕捉语义空间中复杂指令的意图。为了填补这一空白，我们提出了一种统一的方法来量化数据集的信息内容，该方法通过构建标签图来建模语义空间，并基于图内的信息分布来量化多样性。基于这种测量方法，我们进一步引入了一种高效的采样方法，通过迭代选择数据样本，以最大化语义空间中的信息增益（MIG）。在各种数据集和基础模型上的实验表明，MIG始终优于最先进的方法。值得注意的是，使用MIG从Tulu3数据中采样5%的样本进行微调的模型，其性能与使用完整数据集训练的官方SFT模型相当，在AlpacaEval上提高了+5.73%，在Wildbench上提高了+6.89%。
英文摘要：
Data quality and diversity are key to the construction of effective instruction-tuning datasets. % With the increasing availability of open-source instruction-tuning datasets, it is advantageous to automatically select high-quality and diverse subsets from a vast amount of data. % Existing methods typically prioritize instance quality and use heuristic rules to maintain diversity. % However, this absence of a comprehensive view of the entire collection often leads to suboptimal results. % Moreover, heuristic rules generally focus on distance or clustering within the embedding space, which fails to accurately capture the intent of complex instructions in the semantic space. % To bridge this gap, we propose a unified method for quantifying the information content of datasets. This method models the semantic space by constructing a label graph and quantifies diversity based on the distribution of information within the graph. % Based on such a measurement, we further introduce an efficient sampling method that selects data samples iteratively to \textbf{M}aximize the \textbf{I}nformation \textbf{G}ain (MIG) in semantic space. % Experiments on various datasets and base models demonstrate that MIG consistently outperforms state-of-the-art methods. % Notably, the model fine-tuned with 5\% Tulu3 data sampled by MIG achieves comparable performance to the official SFT model trained on the full dataset, with improvements of +5.73\% on AlpacaEval and +6.89\% on Wildbench.
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。
