[FEATURE] Upgrade llama-cpp-python to latest version

[RomainMendez]

The version of llama-cpp-python this project uses is quite old. Therefore I get a lot of errors regarding versions of GGML models. It also doesn't support GGUF models.

I would suggest to up the version of llama-cpp-python to the latest one.

GGUF models are the future anyway it would seem.
Perhaps we could work on something that allows chosing ?

[iaalm]

Thanks for remaindering.👍

[iaalm]

This project actually depends on llamacpp-python. Seems their most recent version doesn't support gguf. I will see whether they can have an update.