Copertina

Beyond Token Eviction: Mixed-Dimension Budget Allocation for Efficient KV Cache Compression

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Miao, Ruijie, Wang, Zhiming, Li, Wang, Wu, Shiwei, Liu, Shufan, Jiang, Yanbing, Yang, Tong
Natura:	Preprint
Pubblicazione:	2026
Soggetti:	Machine Learning
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

LAVa: Layer-wise KV Cache Eviction with Dynamic Budget Allocation
di: Shen, Yiqun, et al.
Pubblicazione: (2025)

EVICPRESS: Joint KV-Cache Compression and Eviction for Efficient LLM Serving
di: Feng, Shaoting, et al.
Pubblicazione: (2025)

LKV: End-to-End Learning of Head-wise Budgets and Token Selection for LLM KV Cache Eviction
di: Zhou, Enshuai, et al.
Pubblicazione: (2026)

CONF-KV: Confidence-Aware KV Cache Eviction with Mixed-Precision Storage for Long-Horizon LLM
di: Li, Yubo, et al.
Pubblicazione: (2026)

KeepKV: Achieving Periodic Lossless KV Cache Compression for Efficient LLM Inference
di: Tian, Yuxuan, et al.
Pubblicazione: (2025)

A Simple Plug-in for Improving Eviction-Based KV Cache Compression
di: Lin, Yuping, et al.
Pubblicazione: (2026)

MadaKV: Adaptive Modality-Perception KV Cache Eviction for Efficient Multimodal Long-Context Inference
di: Li, Kunxi, et al.
Pubblicazione: (2025)

RDKV: Rate-Distortion Bit Allocation for Joint Eviction and Quantization of the KV Cache
di: Zhang, Junkai, et al.
Pubblicazione: (2026)

RetentiveKV: State-Space Memory for Uncertainty-Aware Multimodal KV Cache Eviction
di: Liu, Sihao, et al.
Pubblicazione: (2026)

When Does Value-Aware KV Eviction Help? A Fixed-Contract Diagnostic for Non-Monotone Cache Compression
di: Zhang, Ruijie, et al.
Pubblicazione: (2026)

KVCompose: Efficient Structured KV Cache Compression with Composite Tokens
di: Akulov, Dmitry, et al.
Pubblicazione: (2025)

Make Each Token Count: Towards Improving Long-Context Performance with KV Cache Eviction
di: Bui, Ngoc, et al.
Pubblicazione: (2026)

ForesightKV: Optimizing KV Cache Eviction for Reasoning Models by Learning Long-Term Contribution
di: Dong, Zican, et al.
Pubblicazione: (2026)

CAOTE: KV Cache Selection for LLMs via Attention Output Error-Based Token Eviction
di: Goel, Raghavv, et al.
Pubblicazione: (2025)

MomentKV: Closing the Directional Gap in KV Cache Eviction for Long-Context Inference
di: Li, Yu, et al.
Pubblicazione: (2026)

KVReviver: Reversible KV Cache Compression with Sketch-Based Token Reconstruction
di: Yuan, Aomufei, et al.
Pubblicazione: (2025)

In-context KV-Cache Eviction for LLMs via Attention-Gate
di: Zeng, Zihao, et al.
Pubblicazione: (2024)

LookaheadKV: Fast and Accurate KV Cache Eviction by Glimpsing into the Future without Generation
di: Ahn, Jinwoo, et al.
Pubblicazione: (2026)

MiniCache: KV Cache Compression in Depth Dimension for Large Language Models
di: Liu, Akide, et al.
Pubblicazione: (2024)

No Token Left Behind: Reliable KV Cache Compression via Importance-Aware Mixed Precision Quantization
di: Yang, June Yong, et al.
Pubblicazione: (2024)

ZipCache: Accurate and Efficient KV Cache Quantization with Salient Token Identification
di: He, Yefei, et al.
Pubblicazione: (2024)

MergeMoE: Efficient Compression of MoE Models via Expert Output Merging
di: Miao, Ruijie, et al.
Pubblicazione: (2025)

PagedEviction: Structured Block-wise KV Cache Pruning for Efficient Large Language Model Inference
di: Chitty-Venkata, Krishna Teja, et al.
Pubblicazione: (2025)

Hierarchical Adaptive Eviction for KV Cache Management in Multimodal Language Models
di: Ma, Xindian, et al.
Pubblicazione: (2026)

Ada-KV: Optimizing KV Cache Eviction by Adaptive Budget Allocation for Efficient LLM Inference
di: Feng, Yuan, et al.
Pubblicazione: (2024)

LLMs Know What to Drop: Self-Attention Guided KV Cache Eviction for Efficient Long-Context Inference
di: Wang, Guangtao, et al.
Pubblicazione: (2025)

Not All Tokens Are Worth Caching: Learning Semantic-Aware Eviction for LLM Prefix Caches
di: Fang, Shaoke, et al.
Pubblicazione: (2026)

Rethinking KV Cache Eviction via a Unified Information-Theoretic Objective
di: Yang, Jiaming, et al.
Pubblicazione: (2026)

ThinKV: Thought-Adaptive KV Cache Compression for Efficient Reasoning Models
di: Ramachandran, Akshat, et al.
Pubblicazione: (2025)

RazorAttention: Efficient KV Cache Compression Through Retrieval Heads
di: Tang, Hanlin, et al.
Pubblicazione: (2024)

KVSculpt: KV Cache Compression as Distillation
di: Jiang, Bo, et al.
Pubblicazione: (2026)

LazyEviction: Lagged KV Eviction with Attention Pattern Observation for Efficient Long Reasoning
di: Zhang, Haoyue, et al.
Pubblicazione: (2025)

SmallKV: Small Model Assisted Compensation of KV Cache Compression for Efficient LLM Inference
di: Zhao, Yi, et al.
Pubblicazione: (2025)

The Pitfalls of KV Cache Compression
di: Chen, Alex, et al.
Pubblicazione: (2025)

CoKV: Optimizing KV Cache Allocation via Cooperative Game
di: Sun, Qiheng, et al.
Pubblicazione: (2025)

Training Transformers for KV Cache Compressibility
di: Gelberg, Yoav, et al.
Pubblicazione: (2026)

CommonKV: Compressing KV Cache with Cross-layer Parameter Sharing
di: Wang, Yixuan, et al.
Pubblicazione: (2025)

Memory-Efficient Visual Autoregressive Modeling with Scale-Aware KV Cache Compression
di: Li, Kunjun, et al.
Pubblicazione: (2025)

AttentionPredictor: Temporal Patterns Matter for KV Cache Compression
di: Yang, Qingyue, et al.
Pubblicazione: (2025)

Get More with LESS: Synthesizing Recurrence with KV Cache Compression for Efficient LLM Inference
di: Dong, Harry, et al.
Pubblicazione: (2024)