PrototypeCommon
snu-mllab/KVzip
[NeurIPS'25 Oral] Query-agnostic KV cache eviction: 3–4× reduction in memory and 2× decrease in latency (Qwen3/2.5, Gemma3, LLaMA3)
- Owner
- snu-mllab (organization)
- Stars
- Stars: 228
- Forks
- Forks: 15
- Language
- Python
- License
- MIT
- Created
- 2025
- Topics
- kv-cache-compressionlarge-language-models
- Tier
- Prototype (100 to 249 ★) What this tier means
More repos by snu-mllab
All 4 repos by snu-mllab- snu-mllab/PuzzleMixStars: 156Jupyter NotebookPrototype
- snu-mllab/Efficient-Dataset-CondensationStars: 115PythonPrototype
- snu-mllab/Co-MixupStars: 108Jupyter NotebookPrototype
Nearby in Python
All Python repos- filliptm/ComfyUI-FL-YuE2Stars: 228PythonPrototype
- paper-instruments/feather-trainStars: 228PythonPrototype
- Yamz-Labs/kyojinStars: 228PythonPrototype
- reviewboard/rbtoolsStars: 228PythonPrototype
- adityarathod/xeroxStars: 228PythonPrototype
- sagemath/sagecellStars: 228PythonPrototype