Minima-KV: Retention-Preserving KV Cache Compression with Mixed-Format Paged Attention

arXiv:2608.23834 · cs.AI · Submitted 2026-08-24 · Read on arXiv

cs.AI

Submitted: 2026-08-24

Updated: 2026-08-24

Comments: 13 pages, 3 figures

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Decode — 27× in this paper · explained in 2 episodes
• KV cache — 18× in this paper · explained in 3 episodes
• Quantization — 14× in this paper · explained in 12 episodes
• Fidelity — 6× in this paper · explained in 2 episodes
• Allocator — 5× in this paper · explained in 1 episode
• Accuracy — 2× in this paper · explained in 1 episode
• RAG — 2× in this paper · explained in 3 episodes
• Ablation — 1× in this paper · explained in 1 episode
• Alignment — 1× in this paper · explained in 1 episode
• Benchmark — 1× in this paper · explained in 2 episodes
• Core — 1× in this paper · explained in 2 episodes
• In-Context Learning — 1× in this paper · explained in 4 episodes
• Sparsity — 1× in this paper · explained in 4 episodes

Sources

• Efficient Memory Management for Large Language Model Serving with PagedAttention
• Palu: Compressing KV-Cache with Low-Rank Projection
• KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization
• No Token Left Behind: Reliable KV Cache Compression via Importance-Aware Mixed Precision Quantization
• QEvict: Recoverable Quantized KV Eviction for Attention-Drift-Robust Long-Context Decoding
• MosaicKV: Serving Long-Context LLM with Dynamic Two-D KV Cache Compression
• RDKV: Rate-Distortion Bit Allocation for Joint Eviction and Quantization of the KV Cache
• A JoLT for the KV cache: Near-Lossless KV Cache Compression via Joint Rank-bit Allocation
• A Practical Tensor-Network Compression Pipeline for Production-Scale Large Language Models
• DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads
• GEAR: An Efficient KV Cache Compression Recipe for Near-Lossless Generative Inference of LLM
• TurboQuant: Online Vector Quantization with Near-optimal Distortion Rate
• Mooncake: A KVCache-centric Disaggregated Architecture for LLM Serving
• TileLang: A Composable Tiled Programming Model for AI Systems
• LMCache: An Efficient KV Cache Layer for Enterprise-Scale LLM Inference
• RULER: What's the Real Context Size of Your Long-Context Language Models?

Related papers

• MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
• Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
• The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
• MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
• Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems
• VSAL: A Vision Solver with Adaptive Layouts for Graph Property Detection