Following Length Constraints in Instructions

ACL ARR 2025 May Submission5052 Authors

Abstract: Aligned instruction following models can better fulfill user requests than their unaligned counterparts. However, it has been shown that there is a length bias in evaluation of such models, and that training algorithms tend to exploit this bias by learning longer responses. In this work we show how to train models that can be controlled at inference time with instructions containing desired length constraints. Such models are superior in length instructed evaluations, outperforming standard instruction following models such as GPT4, Llama 3 and Mixtral.

Paper Type: Short

Research Area: Generation

Research Area Keywords: evaluation,length bias,alignment,data augmentation,DPO

Contribution Types: Model analysis & interpretability, NLP engineering experiment

Languages Studied: English

Submission Number: 5052