# KeyDiff: Key Similarity-Based KV Cache Eviction for Long-Context LLM Inference in Resource-Constrained Environments

arXiv: https://arxiv.org/abs/2504.15364 · HTML: https://mlsanity.com/paper/2504.15364/

## Key facts

- Cited by 51 papers in the map (data through 14 Oct 2026).
- Placed in KV cache optimization, one of 12 directions in Efficient LLM inference and serving.
- Builds on 41 of its 46 references that are in the map.
- First posted on arXiv 21 Apr 2025.

## Placed in

- Efficient LLM inference and serving / [KV cache optimization](https://mlsanity.com/inference/kv-cache-optimization/)

## Builds on (most cited first, 30 of 41)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Measuring Mathematical Problem Solving With the MATH Dataset](https://mlsanity.com/paper/2103.03874/) (2021)
- [Large Language Models are Zero-Shot Reasoners](https://mlsanity.com/paper/2205.11916/) (2022)
- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022)
- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [Longformer: The Long-Document Transformer](https://mlsanity.com/paper/2004.05150/) (2020)
- [Tree of Thoughts: Deliberate Problem Solving with Large Language Models](https://mlsanity.com/paper/2305.10601/) (2023)
- [Efficient Streaming Language Models with Attention Sinks](https://mlsanity.com/paper/2309.17453/) (2023)
- [GLU Variants Improve Transformer](https://mlsanity.com/paper/2002.05202/) (2020)
- [LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding](https://mlsanity.com/paper/2308.14508/) (2023)
- [GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints](https://mlsanity.com/paper/2305.13245/) (2023)
- [SnapKV: LLM Knows What You are Looking for Before Generation](https://mlsanity.com/paper/2404.14469/) (2024)
- [H2O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Models](https://mlsanity.com/paper/2306.14048/) (2023)
- [KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache](https://mlsanity.com/paper/2402.02750/) (2024)
- [KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization](https://mlsanity.com/paper/2401.18079/) (2024)
- [Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference](https://mlsanity.com/paper/2406.10774/) (2024)
- [Massive Activations in Large Language Models](https://mlsanity.com/paper/2402.17762/) (2024)
- [MobileLLM: Optimizing Sub-billion Parameter Language Models for On-Device Use Cases](https://mlsanity.com/paper/2402.14905/) (2024)
- [SARATHI: Efficient LLM Inference by Piggybacking Decodes with Chunked Prefills](https://mlsanity.com/paper/2308.16369/) (2023)
- [LLM in a flash: Efficient Large Language Model Inference with Limited Memory](https://mlsanity.com/paper/2312.11514/) (2023)
- [PyramidInfer: Pyramid KV Cache Compression for High-throughput LLM Inference](https://mlsanity.com/paper/2405.12532/) (2024)
- [Why do LLMs attend to the first token?](https://mlsanity.com/paper/2504.02732/) (2025)
- [Transformers are Multi-State RNNs](https://mlsanity.com/paper/2401.06104/) (2024)

## Built on by (most cited first, 30 of 51)

- [Expected Attention: KV Cache Compression by Estimating Attention from Future Queries Distribution](https://mlsanity.com/paper/2510.00636/) (2025)
- [InfiniPot-V: Memory-Constrained KV Cache Compression for Streaming Video Understanding](https://mlsanity.com/paper/2506.15745/) (2025)
- [StreamMem: Query-Agnostic KV Cache Memory for Streaming Video Understanding](https://mlsanity.com/paper/2508.15717/) (2025)
- [Cache What Lasts: Token Retention for Memory-Bounded KV Cache in LLMs](https://mlsanity.com/paper/2512.03324/) (2025)
- [Compactor: Calibrated Query-Agnostic KV Cache Compression with Approximate Leverage Scores](https://mlsanity.com/paper/2507.08143/) (2025)
- [The Pitfalls of KV Cache Compression](https://mlsanity.com/paper/2510.00231/) (2025)
- [Learning to Evict from Key-Value Cache](https://mlsanity.com/paper/2602.10238/) (2026)
- [Self-Compacting Language Model Agents](https://mlsanity.com/paper/2606.23525/) (2026)
- [EVICPRESS: Joint KV-Cache Compression and Eviction for Efficient LLM Serving](https://mlsanity.com/paper/2512.14946/) (2025)
- [LookaheadKV: Fast and Accurate KV Cache Eviction by Glimpsing into the Future without Generation](https://mlsanity.com/paper/2603.10899/) (2026)
- [KaVa: Latent Reasoning via Compressed KV-Cache Distillation](https://mlsanity.com/paper/2510.02312/) (2025)
- [PagedEviction: Structured Block-wise KV Cache Pruning for Efficient Large Language Model Inference](https://mlsanity.com/paper/2509.04377/) (2025)
- [EpiCache: Episodic KV Cache Management for Long-Term Conversation on Resource-Constrained Environments](https://mlsanity.com/paper/2509.17396/) (2025)
- [IndexMem: Learned KV-Cache Eviction with Latent Memory for Long-Context LLM Inference](https://mlsanity.com/paper/2605.25475/) (2026)
- [Predicting Future Utility: Global Combinatorial Optimization for Task-Agnostic KV Cache Eviction](https://mlsanity.com/paper/2602.08585/) (2026)
- [MomentKV: Closing the Directional Gap in KV Cache Eviction for Long-Context Inference](https://mlsanity.com/paper/2606.01563/) (2026)
- [Rethinking KV Cache Eviction via a Unified Information-Theoretic Objective](https://mlsanity.com/paper/2604.25975/) (2026)
- [A Simple Plug-in for Improving Eviction-Based KV Cache Compression](https://mlsanity.com/paper/2605.23258/) (2026)
- [AVIS: Adaptive Test-Time Scaling for Vision-Language Models](https://mlsanity.com/paper/2606.11576/) (2026)
- [QUOKA: Query-Oriented KV Selection For Efficient LLM Prefill](https://mlsanity.com/paper/2602.08722/) (2026)
- [ParisKV: Fast and Drift-Robust KV-Cache Retrieval for Long-Context LLMs](https://mlsanity.com/paper/2602.07721/) (2026)
- [ManifoldKV: Training-Free KV Cache Compression via Euclidean Outlier Detection](https://mlsanity.com/paper/2602.08343/) (2026)
- [Neural Garbage Collection: Learning to Forget while Learning to Reason](https://mlsanity.com/paper/2604.18002/) (2026)
- [HARD-KV: Head-Adaptive Regularization for Decoding-time KV Compression](https://mlsanity.com/paper/2606.28831/) (2026)
- [KARA: Efficient Reasoning LLM Serving via Sliding-Window KV Cache Compression](https://mlsanity.com/paper/2607.01237/) (2026)
- [Efficient On-Device Agents via Adaptive Context Management](https://mlsanity.com/paper/2511.03728/) (2025)
- [Fast Forward: Accelerating LLM Prefill with Predictive FFN Sparsity](https://mlsanity.com/paper/2602.00397/) (2026)
- [CHESS: Context-aware Hierarchical Efficient Semantic Selection for Long-Context LLM Inference](https://mlsanity.com/paper/2602.20732/) (2026)
- [KVDiagnosis: A Diagnostic Benchmark for KV-Cache Compression in Long-Context Language Models](https://mlsanity.com/paper/2608.09412/) (2026)
- [Random Attention: Rethinking KV Cache Eviction for Efficient Reasoning](https://mlsanity.com/paper/2609.03430/) (2026)

## Abstract (arXiv)

We demonstrate that geometrically distinctive keys during LLM inference tend to have high attention scores. Based on the phenomenon we propose KeyDiff, a training-free KV cache eviction method based solely on key similarity. Unlike other KV cache eviction methods, KeyDiff can process arbitrarily long prompts within strict resource constraints and efficiently generate responses. We provide a theoretical basis for KeyDiff by relating key diversity with attention scores. These results imply KeyDiff can efficiently identify the most important tokens to retain. Notably KeyDiff does not rely on attention scores, allowing the use of optimized attention mechanisms like FlashAttention. Under a strict memory allowance, we demonstrate the effectiveness of KeyDiff for the Llama and Qwen model families by observing a performance gap of less than 0.04% with 8K cache budget (~23% KV cache reduction) from the non-evicting baseline on LongBench for Llama 3.1-8B and Llama 3.2-3B. We also observe near baseline performance for Deepseek-R1-Distill-Llama-8B on the Math500 reasoning benchmark and decrease end-to-end inference latency by up to 30% compared to the other token-eviction methods.