Length Representations in Large Language Models

Sangjun Moon, Dasom choi, Jingun Kwon, Hidetaka Kamigaito, Manabu Okumura

Keywords: LLMs, Length representation, Length control

TL;DR: The exploration of length representations and editing hidden units in LLMs for length control without external controls.

Abstract: Large language models (LLMs) have shown remarkable capabilities across various tasks that are learned from massive amounts of text-based data. Although LLMs can control output sequence length, particularly through instruction-based settings, the internal mechanisms behind this control has been unexplored. In this study, we provide empirical evidence on how output sequence length information is encoded within the internal representations of LLMs. In particular, our findings show that multi-head attention mechanisms are critical in determining output sequence length, which can be adjusted in an editable manner. By scaling specific hidden units within the model, we can control the output sequence length without losing the informativeness of the generated text, thereby indicating that length information is partially separable from semantic information. Moreover, some hidden units become increasingly active as prompts become more length-specific, thus reflecting the model's internal awareness of this attribute. Our findings suggest that LLMs have learned robust and adaptable internal mechanisms for controlling output length without external controls.

Primary Area: interpretability and explainable AI

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 10050