README.md

---
title: RAG Capstone Project
emoji: 🤖
colorFrom: blue
colorTo: purple
sdk: docker
app_file: streamlit_app.py
pinned: false
license: mit
---

# 🤖 RAG Capstone Project

A comprehensive Retrieval-Augmented Generation (RAG) system with TRACE evaluation metrics for medical/clinical domains.

## Features

- 🔍 **Multiple RAG Bench Datasets**: CovidQA, CUAD, FinQA, HotpotQA, PubMedQA, and more
- 🧩 **Chunking Strategies**: Dense, Sparse, Hybrid, Re-ranking, Row-based, Entity-based
- 🤖 **Medical Embedding Models**:
  - sentence-transformers/all-mpnet-base-v2
  - emilyalsentzer/Bio_ClinicalBERT
  - microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract
- 💾 **ChromaDB Vector Storage**: Persistent vector storage with efficient retrieval
- ☁️ **Groq LLM Integration**: Cloud-based inference with rate limiting
- 📊 **TRACE Evaluation Metrics**:
  - **U**tilization: How well the system uses retrieved documents
  - **R**elevance: Relevance of retrieved documents to the query
  - **A**dherence: How well the response adheres to the retrieved context
  - **C**ompleteness: How complete the response is
- 💬 **Chat Interface**: Streamlit-based interactive chat with history

## Usage

1. Enter your **Groq API Key** in the sidebar
2. Select a **dataset** from RAG Bench
3. Choose a **chunking strategy** (dense, sparse, hybrid, re-ranking)
4. Select an **embedding model** for document vectorization
5. Choose an **LLM model** for response generation
6. Click **"Load Data & Create Collection"** to initialize
7. Start **chatting** in the chat interface
8. View **retrieved documents** and evaluation metrics
9. Run **TRACE evaluation** on test data

## Environment Variables

Set these in your Hugging Face Space secrets:

- `GROQ_API_KEY`: Your Groq API key (required)
- `GROQ_API_KEYS`: Comma-separated list of API keys for rotation (optional)

## License

MIT License