Skip to content

Latest commit

 

History

History
35 lines (23 loc) · 780 Bytes

README.md

File metadata and controls

35 lines (23 loc) · 780 Bytes

kb-agent

Knowledge-Base agent allows to query Q&A dataset interactively in natural language.

Use cases

  • interactive lookup on FAQ
  • assistant for smooth onboarding experience for new team members.

How it works

  1. Collecting Q&A in format as described in docs/QA-dataset-spec.md and storing it in datasets/dataset.md.
  2. Generating SQLite database from collection of Q&A in datasets/dataset.md.
  3. Generating vector embeddings using txtai .
  4. Querying indexed Q&A pairs interactively.

Install

pip install -r requirements.txt

Run demo

# Generate SQLite database answers.db from dataset.md
python gen_db.py

# run demo in CLI mode
python cli.py

License

GNU AFFERO GENERAL PUBLIC LICENSE