CompilerKV: Risk-Adaptive KV Compression via Offline Experience Compilation
cs.LG, cs.AI
Submitted: 2026-02-09
Updated: 2026-11-04
Terminology
Sources
- RocketKV: Accelerating Long-Context LLM Inference via Two-Stage KV Cache Compression
- Titans: Learning to Memorize at Test Time
- PyramidKV: Dynamic KV Cache Compression based on Pyramidal Information Funneling
- MagicPIG: LSH Sampling for Efficient LLM Generation
- Which Heads Matter for Reasoning? RL-Guided KV Cache Compression
- Ada-KV: Optimizing KV Cache Eviction by Adaptive Budget Allocation for Efficient LLM Inference
- Not All Heads Matter: A Head-Level KV Cache Compression Method with Integrated Retrieval and Reasoning
- Attention Score is not All You Need for Token Importance Indicator in KV Cache Reduction: Value Also Matters
- RULER: What's the Real Context Size of Your Long-Context Language Models?
- KVzip: Query-Agnostic KV Cache Compression with Context Reconstruction
- Let's Verify Step by Step
- ChunkKV: Semantic-Preserving KV Cache Compression for Efficient Long-Context LLM Inference
- ShadowKV: KV Cache in Shadows for High-Throughput Long-Context LLM Inference
- Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference
- Analyzing Multi-Head Self-Attention: Specialized Heads Do the Heavy Lifting, the Rest Can Be Pruned
- Efficient Streaming Language Models with Attention Sinks
- DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads
- DynamicKV: Task-Aware Adaptive KV Cache Compression for Long Context LLMs
Related papers
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis
- (Mis)generalization of Helpful-Only Fine-tuning