LLMs Know What to Drop: Self-Attention Guided KV Cache Eviction for Efficient Long-Context Inference

• URL: http://arxiv.org/abs/2503.08879v1
• Date: Tue, 11 Mar 2025 20:45:02 GMT
• Title: LLMs Know What to Drop: Self-Attention Guided KV Cache Eviction for Efficient Long-Context Inference
• Abstract summary: We propose Self-Attention Guided Eviction(SAGE-KV), a simple and effective KV eviction cache method for long-context inference.<n>After prefilling, our method performs a one-time top-k selection at both the token and head levels to compress the KV cache.<n>SAGE-KV achieves 4x higher memory efficiency with improved accuracy over the static KV cache selection method StreamLLM, and 2x higher memory efficiency with better accuracy than the dynamic KV cache selection method Quest.
• Score: 16.83202690345235
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Efficient long-context inference is critical as large language models (LLMs) adopt context windows of ranging from 128K to 1M tokens. However, the growing key-value (KV) cache and the high computational complexity of attention create significant bottlenecks in memory usage and latency. In this paper, we find that attention in diverse long-context tasks exhibits sparsity, and LLMs implicitly "know" which tokens can be dropped or evicted at the head level after the pre-filling stage. Based on this insight, we propose Self-Attention Guided Eviction~(SAGE-KV), a simple and effective KV eviction cache method for long-context inference. After prefilling, our method performs a one-time top-k selection at both the token and head levels to compress the KV cache, enabling efficient inference with the reduced cache. Evaluations on LongBench and three long-context LLMs (Llama3.1-8B-Instruct-128k, Llama3-8B-Prolong-512k-Instruct, and Qwen2.5-7B-Instruct-128k) show that SAGE-KV maintains accuracy comparable to full attention while significantly improving efficiency. Specifically, SAGE-KV achieves 4x higher memory efficiency with improved accuracy over the static KV cache selection method StreamLLM, and 2x higher memory efficiency with better accuracy than the dynamic KV cache selection method Quest.

Related papers

• CateKV: On Sequential Consistency for Long-Context LLM Inference Acceleration [67.17709877327232]
  Cate KV is a hybrid KV cache method that retains only critical token information for consistent heads.<n>Cate KV reduces memory usage by up to $2.72times$ and accelerates decoding by $2.18times$ in single-sample inputs.
  arXiv  Detail & Related papers  (2026-08-31T06:02:37Z)
• CompressKV: Semantic-Retrieval-Guided KV-Cache Compression for Resource-Efficient Long-Context LLM Inference [9.416125453771361]
  Long-context large language model (LLM) inference is increasingly constrained by the memory footprint and decoding cost of key-value caches.<n>We propose CompressKV, a resource-efficient KV-cache compression framework for GQA-based LLMs.<n>Experiments on LongBench and Needle-in-a-Haystack show that CompressKV consistently outperforms existing KV-cache eviction methods across memory budgets.
  arXiv  Detail & Related papers  (2026-06-23T11:59:46Z)
• Make Each Token Count: Towards Improving Long-Context Performance with KV Cache Eviction [65.710271475739]
  We introduce a global retention-based KV eviction method that learns each token's future utility under a unified memory budget.<n>We show that our method substantially reduces KV memory while matching or surpassing full-cache inference.<n>These results suggest that learned, globally calibrated KV eviction is not only a compression technique, but also a mechanism for improving long-context reasoning.
  arXiv  Detail & Related papers  (2026-05-10T16:47:50Z)
• DeltaKV: Residual-Based KV Cache Compression via Long-Range Similarity [50.52392445266824]
  We propose a residual-based KV cache compression framework motivated by long-range inter-token similarity and highly shared latent components in KV representations.<n>Instead of discarding tokens, DeltaKV encodes semantic residuals relative to retrieved historical references, preserving fidelity while substantially reducing storage.<n>Experiments show that DeltaKV reduces KV cache memory to 29% of the original while maintaining near-lossless accuracy on LongBench, SCBench, and AIME.
  arXiv  Detail & Related papers  (2026-02-08T15:14:36Z)
• Learning What to Write: Write-Gated KV for Efficient Long-Context Inference [10.915483460983411]
  We formalize KV cache management as a causal system of three primitives: KV Admission, Selection, and Eviction.<n>We instantiate KV Admission via Write-Gated KV, a lightweight mechanism that learns to predict token utility before it enters the cache.
  arXiv  Detail & Related papers  (2025-12-19T11:08:58Z)
• G-KV: Decoding-Time KV Cache Eviction with Global Attention [57.47409249054187]
  Large language models (LLMs) excel in complex tasks but encounter significant computational and memory challenges due to long sequence lengths.<n> KV cache compression has emerged as an effective approach to greatly enhance the efficiency of reasoning.<n>We propose G-KV, a KV cache eviction method that employs a global scoring mechanism, combining local and historical attention scores to more accurately assess token importance.
  arXiv  Detail & Related papers  (2025-11-29T14:21:33Z)
• Judge Q: Trainable Queries for Optimized Information Retention in KV Cache Eviction [53.83828564664595]
  Large language models (LLMs) utilize key-value ( KV) cache to store historical information during sequence processing.<n>Current methods for KV cache eviction typically utilize the last window from the pre-filling phase as queries to compute the KV importance scores for eviction.<n>We propose Judge Q, a novel training method which incorporates a soft token list.
  arXiv  Detail & Related papers  (2025-09-13T03:34:12Z)
• Cache Me If You Can: How Many KVs Do You Need for Effective Long-Context LMs? [79.58770714228983]
  Language models handle increasingly long contexts for tasks such as book summarization.<n>This leads to growing memory costs for the key-value ( KV) cache.<n>Many prior works have proposed ways of discarding KVs from memory, but their approaches are tailored to favorable settings.<n>We propose the * KV footprint* as a unified metric, which accounts for both the amount of KV entries stored and their lifespan in memory.
  arXiv  Detail & Related papers  (2025-06-20T16:21:12Z)
• DBudgetKV: Dynamic Budget in KV Cache Compression for Ensuring Optimal Performance [125.81664663201282]
  We introduce a new KV cache compression method dubbed DBudgetKV. It features an attention-based metric to signal when the remaining KV cache is unlikely to match the full-cache performance, then halting the pruning process. Our method is easy to integrate within LLM inference, not only optimizing memory space, but also showing reduced inference time compared to existing methods.
  arXiv  Detail & Related papers  (2025-02-24T06:33:39Z)
• SCBench: A KV Cache-Centric Analysis of Long-Context Methods [61.025422435235456]
  We introduce SCBench, a benchmark for evaluating long-context methods from a KV cachecentric perspective.<n>We provide an extensive KV cache-centric analysis of eight categories long-context solutions, including Gated Linear RNNs and Mamba-Attention hybrids.<n>Our findings show that sub-O(n) memory methods suffer in multi-turn scenarios, while sparse encoding with O(n) memory and sub-O(n2) pre-filling perform robustly.
  arXiv  Detail & Related papers  (2024-12-13T17:59:52Z)
• ClusterKV: Manipulating LLM KV Cache in Semantic Space for Recallable Compression [10.003118268356017]
  Long context poses significant challenges for inference efficiency.<n>We introduce ClusterKV, which recalls tokens at the granularity of semantic clusters.<n>Experiment results show that ClusterKV attains negligible accuracy loss across various tasks with 32k context lengths.
  arXiv  Detail & Related papers  (2024-12-04T10:58:27Z)
• MiniKV: Pushing the Limits of LLM Inference via 2-Bit Layer-Discriminative KV Cache [17.58398289266989]
  Mini KV is a KV cache optimization method that simultaneously preserves long context task accuracy while significantly reducing KV cache size.<n>We show that Mini KV achieves 86% KV cache compression ratio while recovering over 98.5% of accuracy, outperforming state-of-the-art methods.
  arXiv  Detail & Related papers  (2024-11-27T06:10:49Z)
• ThinK: Thinner Key Cache by Query-Driven Pruning [63.13363917871414]
  Large Language Models (LLMs) have revolutionized the field of natural language processing, achieving unprecedented performance across a variety of applications.<n>This paper focuses on the long-context scenario, addressing the inefficiencies in KV cache memory consumption during inference.<n>We propose ThinK, a novel query-dependent KV cache pruning method designed to minimize attention weight loss while selectively pruning the least significant channels.
  arXiv  Detail & Related papers  (2024-07-30T17:59:08Z)
• Model Tells You Where to Merge: Adaptive KV Cache Merging for LLMs on Long-Context Tasks [21.815661269986425]
  We propose a novel KV cache merging approach, called KVMerger, to achieve adaptive KV cache compression for long-context tasks. Our approach is inspired by the intriguing observation that key states exhibit high similarity at the token level within a single sequence. We conduct extensive experiments to demonstrate the effectiveness of KVMerger for long-context tasks under constrained memory budgets.
  arXiv  Detail & Related papers  (2024-07-11T12:50:42Z)
• Training-Free Exponential Context Extension via Cascading KV Cache [49.608367376911694]
  We introduce a novel mechanism that leverages cascading sub-cache buffers to selectively retain the most relevant tokens. Our method reduces prefill stage latency by a factor of 6.8 when compared to flash attention on 1M tokens.
  arXiv  Detail & Related papers  (2024-06-24T03:59:17Z)
• PyramidKV: Dynamic KV Cache Compression based on Pyramidal Information Funneling [53.08975547824068]
  We investigate whether attention-based information flow inside large language models (LLMs) is aggregated through noticeable patterns for long context processing. Our observations reveal that LLMs aggregate information through Pyramidal Information Funneling where attention is scattering widely in lower layers. Motivated by these insights, we developed Pyramid KV, a novel and effective KV cache compression method.
  arXiv  Detail & Related papers  (2024-06-04T07:51:30Z)
• CORM: Cache Optimization with Recent Message for Large Language Model Inference [57.109354287786154]
  We introduce an innovative method for optimizing the KV cache, which considerably minimizes its memory footprint. CORM, a KV cache eviction policy, dynamically retains essential key-value pairs for inference without the need for model fine-tuning. Our validation shows that CORM reduces the inference memory usage of KV cache by up to 70% with negligible performance degradation across six tasks in LongBench.
  arXiv  Detail & Related papers  (2024-04-24T16:11:54Z)
• Get More with LESS: Synthesizing Recurrence with KV Cache Compression for Efficient LLM Inference [78.65321721142624]
  We focus on a memory bottleneck imposed by the key-value ( KV) cache. Existing KV cache methods approach this problem by pruning or evicting large swaths of relatively less important KV pairs. We propose LESS, a simple integration of a constant sized cache with eviction-based cache methods.
  arXiv  Detail & Related papers  (2024-02-14T18:54:56Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.