llama-cpp-ocaml is a MIT-licensed set of ctypes bindings to llama.cpp, a performant and lightweight library to perform inference on various kinds of large language models (LLMs).
A warning: llama.cpp is moving fast and breaking APIs often.
An example was ported from the original distribution, see examples/simple.
To obtain model files compatible with llama.cpp, please refer to the README of https://github.com/ggerganov/llama.cpp/