Attention Instruction: Amplifying Attention in the Middle via Prompting

TL;DR

This paper investigates how large language models handle attention distribution within extended contexts and proposes prompting strategies to improve focus on relevant segments, addressing position bias issues.

Contribution

It reveals the limited relative position awareness of LLMs and demonstrates that prompting can effectively guide attention to specific context segments.

Findings

LLMs lack true relative position awareness.

Matching index prompts can steer attention effectively.

Instruction-based attention modulation benefits RAG tasks.

Abstract

The context window of large language models has been extended to 128k tokens or more. However, language models still suffer from position bias and have difficulty in accessing and using the middle part of the context due to the lack of attention. We examine the relative position awareness of LLMs and the feasibility of mitigating disproportional attention through prompting. We augment the original task instruction with $\texttt{attention instructions}$ that direct language models to allocate more attention towards a selected segment of the context. We conduct a comprehensive investigation on multi-document question answering task with both position-based and index-based instructions. We find that language models do not have relative position awareness of the context. Nevertheless, they demonstrate the capacity to adapt attention to a specific segment using matching indexes. Our analysis contributes to a deeper understanding of position bias in LLMs and provides a pathway to mitigate this bias by instruction, thus benefiting LLMs in locating and utilizing relevant information from retrieved documents in RAG applications.