[] Change the model from llama2 to Mixtril or Mistril [] Setup a process for creating fine-tunings and lora filters [] Improve the library so that swapping lora filters can be done on the fly and associated to a personality
[] Create a class for managing the prompt interaction with the model (decoupling from the AI client) [] Make the AI client more easily managing the different configuration settings per model name [] add file caching for the prompt and model [] Add cli commands to help download and fine-tune models