Ethan Conneely (G00393941)
Ryan Harte (G00338424)
This is a final year project for ATU Galway where we undertook creating an AI powered chat service. This service is usable on both desktop and mobile by going to https://chatalyst.ethanconneely.com/ and signing up with a google account.
- /src - Contains source code to frontend, backend and LLM
- /Documentation - Contains various documentation related to the project
- /Ethan Conneely (G00393941) - Contains all material for submission including latex documents and video overview by Ethan Conneely.
- /Ryan Harte (G00338424) - Contains all material for submission including the Dissertation and a screencast of the Chatalyst app by Ryan Harte.
To run this application you will need:
- Node 20+
- Npm
- GoLang 1.21+
- VSCode
- Air for golang hotreloading (Optional)
- Local install of a LLM / an openai key for using chatgpt
Open project in vscode tasks will run automatically that will setup everything.
You will need node for it to npm i
and you will need air which will hot reload the backend server for development
or manual way
cd src/Frontend
npm install
npm run devcd src/Backend
go run .
# or
air # optional hot reload for developmentEither a openai api key to utilize chatgpt or a local install of https://github.com/oobabooga/text-generation-webui can be used, below is the install guide for the local LLM a signup for openai can be found here https://platform.openai.com/signup
To run the LLM for intelligent smart replies you will need to install https://github.com/oobabooga/text-generation-webui
The instructions are in the repo.
The model we are using for development is this one https://huggingface.co/TheBloke/open-llama-3b-v2-wizard-evol-instuct-v2-196k-GGUF but any open llama model would work fine the larger the better the resulting replies but higher the hardware requirements become this one was chosen as it is able to run on my gpu which gives much faster results.