akdybala's picture
Use browser MiniLM embeddings
6ed4eb1 verified
|
Raw History Blame Contribute Delete
1.21 kB
metadata
title: CodeRAG Retrieval Console
sdk: static

CodeRAG Retrieval Console

CodeRAG is a small code retrieval app. Paste Python code, build a function-level vector index, ask a natural-language question, and retrieve the top matching functions.

What it demonstrates

  • React UI for code input and retrieval inspection.
  • Python API for indexing and search.
  • AST-based function and method extraction.
  • SQLite-backed vector database.
  • Browser neural embeddings with Xenova/all-MiniLM-L6-v2, with deterministic hashing fallback.
  • Multi-view retrieval over raw code, propagated call context, and semantic synthetic questions.

Local run

cd coderag_ui
npm install
npm run build
cd ..
python -m pip install -r requirements.txt
python coderag_api.py --host 127.0.0.1 --port 7860

Open http://127.0.0.1:7860.

API

GET  /api/health
GET  /api/coderag/stats
POST /api/coderag/index-code
POST /api/coderag/search
POST /api/coderag/rebuild

The free Hugging Face Space runs as a static browser app. The hosted Space builds an in-browser vector index with a lightweight MiniLM embedder. The Python API and SQLite vector database are included for local or Docker use.