AR-Butler

This is an python application STS (Speech to Speech) to listen your commands and answer your questions using local LLMs and RAG.

Installation

To install you will need to clone the repository and download the dependecies. Normally I use conda to separate python environments.

git clone https://github.com/alcaitiff/ar-butler.git
cd ar-butler
conda create -n ar-butler python=3.10
conda activate ar-butler
conda install pytorch==2.0.0 torchaudio==2.0.0 torchvision pytorch-cuda=11.8 -c pytorch -c nvidia && pip install git+https://github.com/m-bain/whisperx.git
pip install -r requirements.txt
cd assets/xtts && wget -c "https://huggingface.co/coqui/XTTS-v2/resolve/main/model.pth?download=true" -O model.pth && cd -
cd models && wget -c "https://huggingface.co/TheBloke/Vicuna-13B-CoT-GGUF/resolve/main/Vicuna-13B-CoT.Q4_K_M.gguf" && cd -

Usage

conda activate ar-butler && python3 app.py
#ask anything to the model

Special Commands

Learn:  add something on memory
Forget: remove something from memory
Dump:   show the memory 
Exit:   end the application

Architecture