BEAVER: A Training-Free Hierarchical Prompt Compression Method via Structure-Aware Page Selection
This addresses bottlenecks in long-document understanding for LLM users, offering a scalable solution for high-throughput applications, though it is incremental as it builds on existing compression methods.
The paper tackles the problem of high inference latency and poor information utilization in long-context LLMs by proposing BEAVER, a training-free hierarchical prompt compression method that reduces latency by 26.4x on 128k contexts while maintaining comparable performance to state-of-the-art methods.
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/.