How ritareasciencepark/fastapi-llm works
Project type and purpose
• From the README (“FastAPI-LLM: A simple web (FastAPI) interface for LLM+database interactions”) and the file tree, this is a lightweight full-stack web application.
• Its goal is to let users submit queries via a form, have the backend call an LLM (and/or vector-search routines), store or retrieve summaries in JSON files, and render results in HTML.
File structure
• Top-level scripts:
• main.py – the FastAPI app and route definitions
• elab_api.py – encapsulates the “elaboration” or LLM API calls
• elab_vector.py – handles vectorization or similarity search logic
• playground.py – likely a dev/test harness or interactive notebook script
• query_summary.json & summaries.json – simple file-based “database” of past queries and generated summaries
• Directories:
• static/ – CSS or client-side assets (styles.css)
• templates/ – Jinja2 templates for the HTML views (form.html, response.html, stream_map.html)
README for additional insights
• Minimal description but clearly states: “web interface” + “LLM+database interactions.”