Redis Chat Project by rbs333

The goal showcase fun recipe helper chat app using redis for memory layer.

project overview diagram

example usage

1. Hank asks for recs

2. Stored long term relevant preferences and last n chat messages.

3. Search in vector_store for like recipes

4. Return response and update memory layer

TODOs: so many... no particular order

• add better serialization/deserialization for io.
• improve typing
• add testing
• get load user UI stuff working
• optimize prompts
• dockerize
• split user data model into different contexts
• better manage vector embeddings and test chunk sizes
• ...

project structure

/api
  /static
    styles.css | static assets for FE
  /templates
    index.html | main FE file
  app.py | contains routes etc.
  requirements.txt
/chat_app
  chat.py | primary application code
/data
  /docs | docs for vectoring into store
/db
  vector_store.py | code for init of vector store
  user_index.py | code for init of user index
  llm_memory.py | accessor class for chat memory
  doc_vector.py | helper class for vector store
/utilities
  ... | reusable code chunks
/settings

.env
main.py
requirements.txt

getting started

pip install -r requirements.txt //same for api if running

Using Redis Cloud must have instance running. Also, using Openai as LLM.

=> create .env of form:

OPENAI_API_KEY=
REDIS_URL=
REDIS_HOST=
REDIS_PASSWORD=
REDIS_PORT=

init db

python db/vector_store.py
python db/user_index.py

running app

cd api
flask run --debug

references

https://github.com/RedisVentures/redisvl https://github.dev/miguelgrinberg/Flask-SocketIO-Chat