Awesome KV Cache Optimization

A system-aware taxonomy of KV cache optimization methods for efficient LLM serving — covering temporal, spatial, and structural behaviors at serving time.

Taxonomy at a Glance

A behavior-oriented taxonomy organizing system-aware KV cache optimization into three dimensions. Click any subcategory to filter the paper list below.

Paper Explorer

Venue All Systems AI/ML Preprint
Code Any Has code No code
More 🆕 New 🏆 Award