Deckard collects and exposes local knowledge sets via an API by leveraging modern LLMs to model end-user intent.
Documentation is available in the documentation folder.
Start the API server:
poetry run api:start
Query the LLM model directly without a rag pipeline:
poetry run query:llm <query> <optional context>
Examples:
poetry run query:llm 'Who is Donald Sutherland?'
poetry run query:llm 'Who is Donald Sutherland?' 'Donald sutherland is a duck'
Query the configured RAG pipeline:
poetry run query:rag <pipeline> <query>
Examples:
poetry run query:rag libpages 'Where is the bathroom?'
poetry run query:rag libpages 'Who is the dean of UNB libraries?'
Search the RAG pipeline embeddings database directly:
poetry run search:embeddings <pipeline> <query>
Example:
poetry run search:embeddings libpages 'Who is the dean of UNB libraries?'
Build the RAG pipeline's underlying data to ready it for use. This may have requirements such as network requests, database tunnels, or on-disk data files.
poetry run build:rag <pipeline>
Example:
poetry run build:rag libpages
Start the slackbot configured locally. Slack auth tokens need to be in ENV:
poetry run slackbot:start
- As part of our 'open' ethos, UNB Libraries licenses its applications and workflows to be freely available to all whenever possible.
- Consequently, this repository's contents [unb-libraries/deckard.lib.unb.ca] are licensed under the MIT License. This license explicitly excludes:
- Any generated content remains the exclusive property of its author(s).
- The UNB logo and associated suite of visual identity assets remain the exclusive property of the University of New Brunswick.