Take a look at https://github.com/ggml-org/llama.cpp/raw/ref...

Prompt

Take a look at https://github.com/ggml-org/llama.cpp/raw/refs/heads/master/tools/server/README.md Find the best llama.cpp settings to run https://huggingface.co/WeiboAI/VibeThinker-3B on GTX 970 paired with 24GB RAM.