Importance-Aware Data Selection for Efficient LLM Instruction Tuning
Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (37) · 31328-31336
DOI: 10.1609/aaai.v40i37.40396 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Tingyu Jiang und 8 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref).
Abstract (Originalsprache)
Instruction tuning plays a critical role in enhancing the performance and efficiency of Large Language Models (LLMs). Its success depends not only on the quality of the instruction data but also on the inherent capabilities of the LLM itself. Some studies suggest that even a small amount of high-quality data can achieve instruction fine-tuning results that are on par with, or even exceed, those from using a full-scale dataset. However, rather than focusing solely on calculating data quality scores to evaluate instruction data, there is a growing need to select high-quality data that maximally enhances the performance of instruction tuning for a given LLM. In this paper, we propose the Model Instruction Weakness Value (MIWV) as a novel metric to quantify the importance of instruction data in enhancing model's capabilities. The MIWV metric is derived from the discrepancies in the model’s responses when using In-Context Learning (ICL), helping identify the most beneficial data for enhancing instruction tuning performance. Our experimental results demonstrate that selecting only the top 1% of data based on MIWV can outperform training on the full dataset. Furthermore, this approach extends beyond existing research that focuses on data quality scoring for data selection, offering strong empirical evidence supporting the effectiveness of our proposed method.
Zitieren
APA
Jiang, T., Li, S., Song, Y., Zhang, L., Zhu, H., Zhao, Y., Xu, X., Taura, K., & Wang, H. H. (2026). Importance-Aware Data Selection for Efficient LLM Instruction Tuning. Proceedings of the AAAI Conference on Artificial Intelligence, 40(37), 31328-31336. https://doi.org/10.1609/aaai.v40i37.40396
Vancouver
Jiang T, Li S, Song Y, Zhang L, Zhu H, Zhao Y, et al. Importance-Aware Data Selection for Efficient LLM Instruction Tuning. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(37):31328-31336. doi:10.1609/aaai.v40i37.40396
BibTeX
@article{Jiang2026,
author = {Jiang, Tingyu and Li, Shen and Song, Yiyao and Zhang, Lan and Zhu, Hualei and Zhao, Yuan and Xu, Xiaohang and Taura, Kenjiro and Wang, Hao Henry},
title = {Importance-Aware Data Selection for Efficient LLM Instruction Tuning},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {37},
pages = {31328--31336},
doi = {10.1609/aaai.v40i37.40396}
}
EUVIMED