BEAVER: A Training-Free Hierarchical Prompt Compression Method via Structure-Aware Page Selection

TL;DR

BEAVER is a training-free hierarchical prompt compression method that enhances long-document understanding in LLMs by structure-aware page selection, significantly reducing latency while maintaining performance.

Contribution

It introduces a novel training-free, structure-aware hierarchical compression framework that improves efficiency and preserves discourse integrity in long-context processing.

Findings

Achieves comparable performance to SOTA methods like LongLLMLingua.

Reduces inference latency by 26.4x on 128k contexts.

Maintains high fidelity in multi-needle retrieval on RULER benchmark.

Abstract

The exponential expansion of context windows in LLMs has unlocked capabilities for long-document understanding but introduced severe bottlenecks in inference latency and information utilization. Existing compression methods often suffer from high training costs or semantic fragmentation due to aggressive token pruning. In this paper, we propose BEAVER, a novel training-free framework that shifts compression from linear token removal to structure-aware hierarchical selection. BEAVER maximizes hardware parallelism by mapping variable-length contexts into dense page-level tensors via dual-path pooling, and preserves discourse integrity through a hybrid planner combining semantic and lexical dual-branch selection with sentence smoothing. Extensive evaluations on four long-context benchmarks demonstrate that BEAVER achieves comparable performance to state-of-the-art (SOTA) methods like LongLLMLingua. Notably, on the RULER benchmark, BEAVER maintains high fidelity in multi-needle retrieval where baselines deteriorate. Regarding efficiency, BEAVER reduces latency by 26.4x on 128k contexts, offering a scalable solution for high-throughput applications. Our code is available at https://cslikai.cn/BEAVER/.