OoriData / OgbujiPT

Client-side toolkit for using large language models, including where self-hosted
Apache License 2.0
103 stars 8 forks source link

Improve the framework for using vector DBs (Qdrant & add 1 other) #16

Closed uogbuji closed 1 year ago

uogbuji commented 1 year ago

Right now embedding_helper.py is mostly a skeleton put in place to enable the chat-your-docs demo, and in all truth while learning the ins & outs of vector DBs. It's now time to dress it up in better design. For one thing, let's expose the metadata storage capabilities, which are among the real strengths of a DB such as Qdrant.

uogbuji commented 1 year ago

Ended up with such a big refactoring just on Qdrant that we'll just merge back, and create a new ticket/branch to add a new vector DB helper.

uogbuji commented 1 year ago

@choccccy, clues for any final sorting of the GUI actions here: https://docs.streamlit.io/library/api-reference/layout/st.empty

Right now control flow is correct, but sometimes the UI elements get shuffled.

uogbuji commented 1 year ago

Local cleanup:

git checkout main
git branch --delete 16-better-vector-db