RestoreKV: Recovering Full-Cache Behavior Under Aggressive Query-Agnostic KV Cache Eviction

arXiv:2608.01247 · cs.CL, cs.LG · Submitted 2026-08-02 · Read on arXiv

Changwoo Baek, Seungjun Shin, Kyeongbo Kong

cs.CL, cs.LG

Submitted: 2026-08-02

Comments: 13 pages, 8 figures

Project page: https://paper.pnu-cvsp.com/RestoreKV

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

Terminology

• KV cache — 43× in this paper · explained in 3 episodes
• Benchmark — 22× in this paper · explained in 2 episodes
• Embeddings — 15× in this paper · explained in 2 episodes
• Accuracy — 14× in this paper · explained in 1 episode
• Adaptation — 14× in this paper · explained in 1 episode
• Distillation — 12× in this paper · explained in 2 episodes
• Self-Distillation — 8× in this paper · explained in 7 episodes
• Ablation — 5× in this paper · explained in 1 episode
• Generalization — 5× in this paper · explained in 13 episodes
• LoRA Adapters — 5× in this paper · explained in 3 episodes
• Attention Matching — 3× in this paper
• Transformer — 3× in this paper · explained in 2 episodes
• Adam — 2× in this paper · explained in 2 episodes
• Aggregation — 2× in this paper · explained in 1 episode
• LoRA Adaptation — 2× in this paper · explained in 2 episodes
• Consistency — 1× in this paper · explained in 3 episodes
• Downstream Tasks — 1× in this paper · explained in 2 episodes
• Few-Shot Learning — 1× in this paper · explained in 4 episodes
• Fine-Tuning — 1× in this paper · explained in 5 episodes
• Instruction Tuning — 1× in this paper · explained in 2 episodes
• Pruning — 1× in this paper · explained in 4 episodes
• Robustness — 1× in this paper · explained in 5 episodes

Sources

• KV-Distill: Nearly Lossless Learnable Context Compression for LLMs
• Expected Attention: KV Cache Compression by Estimating Attention from Future Queries Distribution
• The Llama 3 Herd of Models
• Distilling the Knowledge in a Neural Network
• RULER: What's the Real Context Size of Your Long-Context Language Models?
• KVzap: Fast, Adaptive, and Faithful KV Cache Pruning
• Fast KVzip: Efficient and Accurate LLM Inference with Gated KV Eviction
• Tulu 3: Pushing Frontiers in Open Language Model Post-Training
• Latent Context Compilation: Distilling Long Context into Compact Portable Memory
• A Simple Plug-in for Improving Eviction-Based KV Cache Compression
• Still: Amortized KV Cache Compaction in a Single Forward Pass
• Qwen3 Technical Report

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering