Length Representations in Large Language Models

TL;DR

This paper investigates how large language models internally encode and control output sequence length, revealing that attention mechanisms and specific hidden units play key roles in length management without sacrificing semantic content.

Contribution

It provides empirical evidence on the internal representations of length in LLMs, highlighting the role of attention and hidden units in length control and disentanglement from semantics.

Findings

Attention mechanisms are critical for length control.

Length information can be disentangled from semantic content.

Hidden units reflect internal awareness of output length.

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 in instruction-based settings, the internal mechanisms behind this control have been unexplored yet. In this study, we provide empirical evidence on how output sequence length information is encoded within the internal representations in LLMs. In particular, our findings show that multi-head attention mechanisms are critical in determining output sequence length, which can be adjusted in a disentangled 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 disentangled 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 any external control.