stampy-chat
The Stampy conversational chatbot answers questions about AI Safety & Alignment based on information retrieved from the Alignment Research Dataset (ARD). The project has 2 components:
-
api/ Backend API - Given a user's query and chat history, the most semantically similar chunks are retrieved from the vector store then a large language model is prompted to answer the query based on the retrieved context. The generated answer is returned along with cited sources.
-
web/ Frontend GUI - Web app that calls the backend API and displaying the conversation flow.
What is the purpose of Stampy?
With the recent development in AI, growing concern and interest in AI Safety & Alignment is coupled with tons of information and misinformation. Sifting through all the text while trying to identify quality sources is a daunting barrier for entry. Stampy not only strives to provide quality information but also allows people to contribute while learning. The FAQs are hand curated and limited by volunteer bandwidth. This chatbot leverages LLM to synthesize & summarize the ever expanding literature.
Usage
Environment Setup
In the api/ directory, rename .env.example to .env. Edit this file and
fill in the placeholder values with valid credentials. Message the
#stampy-dev channel on the Rob Miles AI
Discord if you need help with this.
Install npm, python 3.11, and pipenv.
Install dependencies for the API:
cd api # from repository root
pipenv install --dev --ignore-pipfile
Install dependencies for the web UI:
cd web # from repository root
npm install
Database setup
Some things (e.g. logging) require a database connection to work correctly. To make this easier, there is a script to set one up locally via Docker. To get this working:
This should start the database, make sure it's up to date, then as a final step display a command that will allow you to connect to it directly if you want to.
Running a local version
Open two terminal windows. In the first, run:
cd api
pipenv run python3 main.py
In the second, run:
cd web
npm run dev
In the second window, a URL will be printed. Probably http://localhost:3000.
Paste this into your browser to see the app.
Original Prototypes
The prototypes below were developed in response to a bounty on LessWrong.
The teams collaborated and key features were combined into one project.
