MyTherapyPal

This is the main repository for the final MyTherapyPal application. It consists of a flutter mobile & web user interface application with account registration & login functionality, chat functionality and an AI mental health assistant chatbot, which is implemented using a REST API web service powered by Python Flask.

Run debug environment

AI(LLM) Mental Health Assistant API

The AI mental health assistant is using a 4-bit quantized Llama-2-13B model called TheBloke/Llama-2-13B-chat-GPTQ. The functionality of which is provided over flask API locally. This was developed using an NVidia 4060Ti 16GB graphics card with CUDA 12.3. There is a conda environment yaml file included to setup the environment. Below are the steps to set this up.

Create the new conda enviroment (assuming you have anaconda3 or equivelant installed)

conda env create -f chatcbt_final.yml
conda activate chatCBTQ

Start the flask server

cd chatCBT
python app.py

Start the reverse proxy in WSL (needs ngrok installed)

ngrok http http://localhost:5000

Flutter Application

Start the flutter application

flutter run

Choose platform

Choose either Chrome, Edge, Windows, Android (if connected to device or emulator) or iOS (if on a MacOS system with Xcode). Selecting Windows may require Visual Studio being installed.

buidiemnhi/MyTherapyPal