[the_mitsuhiko]

Isn’t llama.cpp already depending on Python anyways for the templating?

[Maxious]

It uses a cpp implementation of jinja https://github.com/google/minja