server.cpp / README.md
iAkashPaul's picture
Update README.md
b2a0119 verified
|
raw
history blame
786 Bytes
metadata
title: Ghudsavar
emoji: πŸ‡πŸ»
colorFrom: gray
colorTo: green
sdk: docker
pinned: false
license: mit
app_port: 7860
short_description: Phi-3-Mini on llama.cpp's server example

Ghudsavar πŸ‡πŸ»

Github iAkashPaul/Ghudsavar

Ghudsavar (Horse rider) - Is a quick llama.cpp server for CPU only runtimes, made available as a docker image. One click deployment to HuggingFace spaces (even on the free tier with 2 CPU-cores).

Try it here Ghudsavar (default HF Space Embed messes up the themeing).

Phi3-mini @Q4 runs at ~5tok/s

Customize yours

git clone https://github.com/iakashpaul/Ghudsavar.git
docker build -t iakashpaul/Ghudsavar:latest . --platform linux/amd64