Local LM right in the game? Trying to implement

Hello guys!

I'm developing the dialogue system and I want it to be impressive and alive. So I'm trying now to add small language model like Qwen 1,5B or Tiny Llama to the project (they should use 1-2 GB or RAM and use CPU only, so no extra problems for the GPU). I've never done this before, but I will be figuring out how to do it, so it would work without the user to run python script by himself. If I find the solution, you could try first game with local LM to deal with dialogues. I think at least no one did it before and I'll figure out why very soon.