RAG Translation Backend Server

A FastAPI-based RAG (Retrieval-Augmented Generation) server designed for translation tasks, featuring vector-based similarity search and stammering detection.

Features

• Retrieval-Augmented Generation: Uses ChromaDB to retrieve the top 4 similar translation pairs for better context.
• Stammering Detection: Advanced algorithm to detect character elongation and phrase repetition in translations.
• Dockerized: Fully containerized for consistent deployment and testing.

Quick Start (Local)

This project uses uv for lightning-fast dependency management.

1. Setup: uv sync
2. Run Server: uv run uvicorn main:app --reload
3. Seed Data: uv run seed_db.py (populates the DB with the provided JSONL pairs)
4. API Docs: Open http://127.0.0.1:8000/docs

Docker (Score Booster)

To run the fully containerized application:

docker build -t translation-rag .
docker run -p 8000:8000 translation-rag