面向微调的子集选择：一种用于数学领域自适应的效用-多样性平衡方法
Subset Selection for Fine-Tuning: A Utility-Diversity Balanced Approach for Mathematical Domain Adaptation
AI总结：
本文提出一种效用-多样性平衡的预算子集选择方法，通过结合困惑度与思维链损失筛选高价值样本并保证数学子领域覆盖，在LLaMA-3 8B和Phi-3上以更低成本实现接近全数据集的微调性能。
摘要：
我们提出一种改进的方法，通过采用预算受限的子集选择方法，在数学等特定领域高效微调大型语言模型（LLMs）。我们的方法结合效用指标和多样性指标，以选择最具信息量和代表性的训练样本。最终目标是通过从整个数据集中精心挑选数据点，实现接近全数据集性能，同时显著降低计算成本和训练时间，并达到与全数据集相当的竞争性能。效用指标同时纳入困惑度和思维链（Chain-of-Thought, CoT）损失，以识别对模型学习贡献最大的困难样本，而多样性指标则确保对数学各子领域的广泛覆盖。我们在 LLaMA-3 8B 和 Phi-3 模型上评估了我们的方法，并与多个基线方法进行比较，包括随机选择、基于多样性的采样以及现有的最先进子集选择技术。
英文摘要：
We propose a refined approach to efficiently fine-tune large language models (LLMs) on specific domains like the mathematical domain by employing a budgeted subset selection method. Our approach combines utility and diversity metrics to select the most informative and representative training examples. The final goal is to achieve near-full dataset performance with meticulously selected data points from the entire dataset while significantly reducing computational cost and training time and achieving competitive performance as the full dataset. The utility metric incorporates both perplexity and Chain-of-Thought (CoT) loss to identify challenging examples that contribute most to model learning, while the diversity metric ensures broad coverage across mathematical subdomains. We evaluate our method on LLaMA-3 8B and Phi-3 models, comparing against several baseline approaches, including random selection, diversity-based sampling, and existing state-of-the-art subset selection techniques.
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。
