English

Improving Data Efficiency via Curating LLM-Driven Rating Systems

Computation and Language 2025-03-07 v2 Artificial Intelligence

Abstract

Instruction tuning is critical for adapting large language models (LLMs) to downstream tasks, and recent studies have demonstrated that small amounts of human-curated data can outperform larger datasets, challenging traditional data scaling laws. While LLM-based data quality rating systems offer a cost-effective alternative to human annotation, they often suffer from inaccuracies and biases, even in powerful models like GPT-4. In this work, we introduce DS2, a Diversity-aware Score curation method for Data Selection. By systematically modeling error patterns through a score transition matrix, DS2 corrects LLM-based scores and promotes diversity in the selected data samples. Our approach shows that a curated subset (just 3.3% of the original dataset) outperforms full-scale datasets (300k samples) across various machine-alignment benchmarks, and matches or surpasses human-aligned datasets such as LIMA with the same sample size (1k samples). These findings challenge conventional data scaling assumptions, highlighting that redundant, low-quality samples can degrade performance and reaffirming that "more can be less."

Keywords

Cite

@article{arxiv.2410.10877,
  title  = {Improving Data Efficiency via Curating LLM-Driven Rating Systems},
  author = {Jinlong Pang and Jiaheng Wei and Ankit Parag Shah and Zhaowei Zhu and Yaxuan Wang and Chen Qian and Yang Liu and Yujia Bao and Wei Wei},
  journal= {arXiv preprint arXiv:2410.10877},
  year   = {2025}
}
R2 v1 2026-06-28T19:21:13.886Z