This repository contains a Flask application for a question-answering (QA) system based on a custom knowledge base. The knowledge base is created from documents loaded from a specified URL and is used to answer user queries.
- Load data from a specified URL and create a knowledge base.
- Use FAISS for vector-based retrieval of document embeddings.
- Utilize Google Palm as the language model for answering questions.
- Expose endpoints to create the knowledge base and ask questions via a Flask API.
- Python 3.10+
- Google Palm API key
- Hugging Face
InstructorEmbeddingsmodel - Flask
- FAISS