QUITO: Accelerating Long-Context Reasoning through Query-Guided Context Compression

TL;DR

QUITO is a novel attention-based context compression method that improves long-context reasoning efficiency in large language models by filtering irrelevant information based on question-guided attention, leading to better performance.

Contribution

Introduces QUITO, a new attention-guided context compression technique that enhances reasoning efficiency and accuracy in LLMs by filtering context information based on question relevance.

Findings

QUITO outperforms baseline methods on NaturalQuestions and ASQA datasets.

It effectively reduces context size while maintaining reasoning quality.

Experimental results show significant improvements across various LLMs.

Abstract

In-context learning (ICL) capabilities are foundational to the success of large language models (LLMs). Recently, context compression has attracted growing interest since it can largely reduce reasoning complexities and computation costs of LLMs. In this paper, we introduce a novel Query-gUIded aTtention cOmpression (QUITO) method, which leverages attention of the question over the contexts to filter useless information. Specifically, we take a trigger token to calculate the attention distribution of the context in response to the question. Based on the distribution, we propose three different filtering methods to satisfy the budget constraints of the context length. We evaluate the QUITO using two widely-used datasets, namely, NaturalQuestions and ASQA. Experimental results demonstrate that QUITO significantly outperforms established baselines across various datasets and downstream LLMs, underscoring its effectiveness. Our code is available at https://github.com/Wenshansilvia/attention_compressor.