ZEBRA: Leveraging Model-Behavioral Knowledge for Zero-Annotation Preference Dataset Construction

Jeesu Jung, Chanjun Park, Sangkeun Jung


Abstract
Recent efforts in LLM alignment have focused on constructing large-scale preference datasets via human or Artificial Intelligence(AI) annotators. However, such approaches rely on instance-wise supervision, incurring substantial annotation cost and limited interpretability. In this paper, we propose **ZEBRA**—a model behavior-wise zero-annotation framework that constructs preference data by leveraging model behavior knowledge derived from benchmark performances.ZEBRA binarizes response pairs by evaluating the quality and similarity of their origin models, entirely bypassing instance-level annotation. This allows scalable, controllable, and cost-effective alignment data generation. Empirical results show that ZEBRA achieves alignment performance comparable to instance-supervised methods, despite requiring no manual or model-based labeling.
Anthology ID:
2025.findings-emnlp.417
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2025
Month:
November
Year:
2025
Address:
Suzhou, China
Editors:
Christos Christodoulopoulos, Tanmoy Chakraborty, Carolyn Rose, Violet Peng
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
7895–7911
Language:
URL:
https://aclanthology.org/2025.findings-emnlp.417/
DOI:
Bibkey:
Cite (ACL):
Jeesu Jung, Chanjun Park, and Sangkeun Jung. 2025. ZEBRA: Leveraging Model-Behavioral Knowledge for Zero-Annotation Preference Dataset Construction. In Findings of the Association for Computational Linguistics: EMNLP 2025, pages 7895–7911, Suzhou, China. Association for Computational Linguistics.
Cite (Informal):
ZEBRA: Leveraging Model-Behavioral Knowledge for Zero-Annotation Preference Dataset Construction (Jung et al., Findings 2025)
Copy Citation:
PDF:
https://aclanthology.org/2025.findings-emnlp.417.pdf
Checklist:
 2025.findings-emnlp.417.checklist.pdf