nisten's picture
Update README.md
ce65180
|
raw
history blame
661 Bytes
metadata
license: mit

NousResearch - Obsidian-3B-V0.5 gguf q6 version by Nisten

How to Run the server and download the models all in one command, you need ~2.9gb of space

To run this on mac or linux, make a new folder, go in it and run this whole in one shot:

git clone -b stablelm-support https://github.com/Galunid/llama.cpp.git && \
cd llama.cpp && \
make && \
wget https://huggingface.co/nisten/obsidian-3b-multimodal-q6-gguf/resolve/main/obsidian-q6.gguf && \
wget https://huggingface.co/nisten/obsidian-3b-multimodal-q6-gguf/resolve/main/mmproj-obsidian-f16.gguf && \
./server -m obsidian-q6.gguf --mmproj mmproj-obsidian-f16.gguf -ngl 42