Janghyun1230/FastKVzip

Accurate and fast KV cache compression with a gating mechanism

18
/ 100
Experimental

This project helps large language model (LLM) operators make their models run faster and more efficiently, especially during the "thinking" and response generation phases. By intelligently compressing the model's memory (KV cache), it allows the model to process information with significantly less memory while maintaining high accuracy. The primary users are MLOps engineers, model deployers, and researchers who manage and optimize LLM inference.

Use this if you are deploying or running large language models on NVIDIA GPUs and need to reduce memory usage and increase inference speed without sacrificing performance.

Not ideal if you are working with smaller language models, CPU-only environments, or if fine-grained control over individual token importance is not a priority.

LLM deployment model optimization AI inference GPU efficiency computational linguistics
No License No Package No Dependents
Maintenance 10 / 25
Adoption 5 / 25
Maturity 3 / 25
Community 0 / 25

How are scores calculated?

Stars

13

Forks

Language

Python

License

Last pushed

Feb 19, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/llm-tools/Janghyun1230/FastKVzip"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.