# CAOTE: KV Cache Selection for LLMs via Attention Output Error-Based Token Eviction

arXiv: https://arxiv.org/abs/2504.14051 · HTML: https://mlsanity.com/paper/2504.14051/

## Key facts

- Cited by 13 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 27 of its 35 references that are in the map.
- First posted on arXiv 18 Apr 2025.

## Placed in

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

## Builds on (most cited first, 27 of 27)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [Efficient Streaming Language Models with Attention Sinks](https://mlsanity.com/paper/2309.17453/) (2023)
- [Transformers are RNNs: Fast Autoregressive Transformers with Linear Attention](https://mlsanity.com/paper/2006.16236/) (2020)
- [LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding](https://mlsanity.com/paper/2308.14508/) (2023)
- [LaMDA: Language Models for Dialog Applications](https://mlsanity.com/paper/2201.08239/) (2022)
- [SnapKV: LLM Knows What You are Looking for Before Generation](https://mlsanity.com/paper/2404.14469/) (2024)
- [FlexGen: High-Throughput Generative Inference of Large Language Models with a Single GPU](https://mlsanity.com/paper/2303.06865/) (2023)
- [Efficiently Scaling Transformer Inference](https://mlsanity.com/paper/2211.05102/) (2022)
- [Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference](https://mlsanity.com/paper/2406.10774/) (2024)
- [Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs](https://mlsanity.com/paper/2310.01801/) (2023)
- [Benchmarking Large Language Models for News Summarization](https://mlsanity.com/paper/2301.13848/) (2023)
- [DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads](https://mlsanity.com/paper/2410.10819/) (2024)
- [Ada-KV: Optimizing KV Cache Eviction by Adaptive Budget Allocation for Efficient LLM Inference](https://mlsanity.com/paper/2407.11550/) (2024)
- [Landmark Attention: Random-Access Infinite Context Length for Transformers](https://mlsanity.com/paper/2305.16300/) (2023)
- [Leveraging Large Language Models for Multiple Choice Question Answering](https://mlsanity.com/paper/2210.12353/) (2022)
- [SARATHI: Efficient LLM Inference by Piggybacking Decodes with Chunked Prefills](https://mlsanity.com/paper/2308.16369/) (2023)
- [BookSum: A Collection of Datasets for Long-form Narrative Summarization](https://mlsanity.com/paper/2105.08209/) (2021)
- [Scatterbrain: Unifying Sparse and Low-rank Attention Approximation](https://mlsanity.com/paper/2110.15343/) (2021)
- [Transformers are Multi-State RNNs](https://mlsanity.com/paper/2401.06104/) (2024)
- [DeepSpeed-FastGen: High-throughput Text Generation for LLMs via MII and DeepSpeed-Inference](https://mlsanity.com/paper/2401.08671/) (2024)
- [Get More with LESS: Synthesizing Recurrence with KV Cache Compression for Efficient LLM Inference](https://mlsanity.com/paper/2402.09398/) (2024)
- [A Survey on Non-Autoregressive Generation for Neural Machine Translation and Beyond](https://mlsanity.com/paper/2204.09269/) (2022)
- [CAKE: Cascading and Adaptive KV Cache Eviction with Layer Preferences](https://mlsanity.com/paper/2503.12491/) (2025)
- [Wordcraft: a Human-AI Collaborative Editor for Story Writing](https://mlsanity.com/paper/2107.07430/) (2021)
- [On the Efficacy of Eviction Policy for Key-Value Constrained Generative Language Model Inference](https://mlsanity.com/paper/2402.06262/) (2024)

## Built on by (most cited first, 13 of 13)

- [MomentKV: Closing the Directional Gap in KV Cache Eviction for Long-Context Inference](https://mlsanity.com/paper/2606.01563/) (2026)
- [When Does Value-Aware KV Eviction Help? A Fixed-Contract Diagnostic for Non-Monotone Cache Compression](https://mlsanity.com/paper/2605.08234/) (2026)
- [RDKV: Rate-Distortion Bit Allocation for Joint Eviction and Quantization of the KV Cache](https://mlsanity.com/paper/2605.08317/) (2026)
- [Epiphany-Aware KV Cache Eviction Without the Attention Matrix](https://mlsanity.com/paper/2606.26472/) (2026)
- [Error Certificates for KV-Cache Eviction via Randomized Design](https://mlsanity.com/paper/2607.21475/) (2026)
- [Reformulating KV Cache Eviction Problem for Long-Context LLM Inference](https://mlsanity.com/paper/2605.07234/) (2026)
- [ARC-KV: Amortizing Anchor Search for Reconstruction-Based KV Cache Compaction](https://mlsanity.com/paper/2609.36835/) (2026)
- [Monte Carlo Estimation for KV Cache Eviction](https://mlsanity.com/paper/2610.07643/) (2026)
- [ConFu: Contemplate the Future for Better Speculative Sampling](https://mlsanity.com/paper/2603.08899/) (2026)
- [ValueDiff: Value-Geometric KV Cache Eviction for Sink-Suppressed LLMs](https://mlsanity.com/paper/2609.23314/) (2026)
- [From Token Importance to Conditional Removability: Rethinking Visual Token Pruning in Multimodal Large Language Models](https://mlsanity.com/paper/2609.26484/) (2026)
- [A Self-Pruning Transformer: Extreme KV-Cache Compression with Universal Attention](https://mlsanity.com/paper/2610.09051/) (2026)
- [Sparse Attention Is Matrix Approximation, Not Choosing from a Bag of Values](https://mlsanity.com/paper/2610.10871/) (2026)

## Abstract (arXiv)

While long context support of large language models has extended their abilities, it also incurs challenges in memory and compute which becomes crucial bottlenecks in resource-restricted devices. Token eviction, a widely adopted post-training methodology designed to alleviate the bottlenecks by evicting less important tokens from the cache, typically uses attention scores as proxy metrics for token importance. However, one major limitation of attention score as a token-wise importance metrics is that it lacks the information about contribution of tokens to the attention output. In this paper, we propose a simple eviction criterion based on the contribution of cached tokens to attention outputs. Our method, CAOTE, optimizes for eviction error due to token eviction, by seamlessly integrating attention scores and value vectors. This is the first method which uses value tokens on top of attention-based eviction scores in closed-form. Additionally, CAOTE can act as a meta-heuristic method with flexible usage with any token eviction method. We show that CAOTE, when combined with the state-of-the-art attention score-based methods, always improves accuracies on the downstream task, indicating the importance of leveraging information from values during token eviction process.