Beyond RAG: Task-Aware KV Cache Compression for Comprehensive Knowledge Reasoning
This addresses computational bottlenecks for LLM applications requiring comprehensive knowledge reasoning, offering a more efficient alternative to existing methods.
The paper tackles the problem of inefficient external knowledge integration in large language models by proposing task-aware KV cache compression, which outperforms RAG and task-agnostic methods with up to 7-point accuracy gains and reduces inference latency from 0.43s to 0.16s.
Incorporating external knowledge in large language models (LLMs) enhances their utility across diverse applications, but existing methods have trade-offs. Retrieval-Augmented Generation (RAG) fetches evidence via similarity search, but key information may fall outside top ranked results. Long-context models can process multiple documents but are computationally expensive and limited by context window size. Inspired by students condensing study material for open-book exams, we propose task-aware key-value (KV) cache compression, which compresses external knowledge in a zero- or few-shot setup. This enables LLMs to reason efficiently over a compacted representation of all relevant information. Experiments show our approach outperforms both RAG and task-agnostic compression methods. On LongBench v2, it improves accuracy by up to 7 absolute points over RAG with a 30x compression rate, while reducing inference latency from 0.43s to 0.16s. A synthetic dataset highlights that RAG performs well when sparse evidence suffices, whereas task-aware compression is superior for broad knowledge tasks.