Forked from cocktailpeanut/dalai
A Cha(d)tBot that leverages Stanford's Alpaca Model.
Uses invisible prompt engineering to coerce the predictive LLM into an "informative and helpful" virtual assistant.
~4GB of Ram
npm install
./chad alpaca install 7B
npm run develop:be
npm run develop:fe
- Add confiiguration to remember chat history
- Rebuild invisible prompts
- Stop model if it attempts to respond for the user
- Fine tune the invisible prompts
- Fine tune model depending on specific use cases
(and there are a lot of them)
The Alpaca model is uncensored and may contain bais and misinformation. Use at your own risk.