Super simple Python connectors for llama.cpp, including vision models (Gemma 3, Qwen2-VL)

[fidecastro]

I built a lightweight alternative to llama-cpp-python that stays current with llama.cpp's latest releases and enables Python integration with llama.cpp's vision models.

It works on top of a llama.cpp local compilation, which must either be compiled manually by the user or through a Dockerfile script I added to the repo. This way it's easier to stay up-to-date with the latest releases of llama.cpp. It's nowhere near as ambitious as llama-cpp-python but it can be useful for those of us that want to be tinker with the newest releases.

I hope this is as useful to you as it has been is for me. Feedback and criticism is warmly welcomed.

https://github.com/fidecastro/llama-cpp-connector