|
|
--- |
|
|
title: Text-to-SQL RAG with CodeLlama |
|
|
colorFrom: indigo |
|
|
colorTo: blue |
|
|
sdk: gradio |
|
|
app_file: app.py |
|
|
python_version: "3.10" |
|
|
pinned: false |
|
|
--- |
|
|
|
|
|
# Text-to-SQL RAG with CodeLlama |
|
|
|
|
|
This Space runs a Retrieval-Augmented Generation (RAG) system that converts natural language questions into SQL. |
|
|
|
|
|
- Primary LLM: CodeLlama (via ctransformers) |
|
|
- Vector store: ChromaDB (recreated on first run if empty) |
|
|
- UI: Gradio (`app.py`) |
|
|
|
|
|
## Files |
|
|
- `app.py`: Gradio app entrypoint |
|
|
- `requirements.txt`: Dependencies |
|
|
- `rag_system/`: Core RAG components |
|
|
- `data/`, `prompts/`: Runtime data and prompt templates |
|
|
|
|
|
## Notes |
|
|
- SQLite databases are not committed; they are generated at runtime. |
|
|
- If the Space fails to start, check the Logs tab for missing packages and update `requirements.txt` accordingly. |
|
|
|
