You can deploy this model and start earning money today!

mathanmithun/
DhiGrowth_FAQ_RAG

Name: DhiGrowth_FAQ_RAG
Rating: 2.5 (20 reviews)

Views

No views yet

Dhigrowth RAG Chatbot

A Retrieval-Augmented Generation (RAG) chatbot system built for Dhigrowth, providing accurate answers based on company knowledge base using vector similarity search and fine-tuned language models.

🚀 Features

RAG Architecture: Combines vector similarity search with LLM generation for accurate, context-aware responses
Vector Database: ChromaDB for efficient document retrieval
Fine-tuned Model: Qwen2.5-0.5B model fine-tuned on Dhigrowth data
RESTful API: FastAPI-based backend with comprehensive error handling
Input Validation: Pydantic-based validation for robust API requests
Comprehensive Logging: File and console logging for debugging and monitoring
Health Monitoring: Health check endpoints for system status
Error Handling: Graceful error handling with informative responses

🛠️ Installation

Prerequisites

Python 3.8 or higher
4GB+ RAM recommended
2GB+ disk space for models and embeddings

Step 1: Clone the Repository

bash

git clone <repository-url>
cd DHIGROWTH_RAG

Step 2: Create Virtual Environment

Windows:

bash

python -m venv venv
source venv/Scripts/activate

Linux/Mac:

bash

python3 -m venv venv
source venv/bin/activate

Step 3: Install Dependencies

pip install -r requirements.txt

Step 4: Set Up Vector Database

Create the vector database from your data:

python src/embed_loader.py

This will:

Load data from data/dhigrowth_final_rag_ft.jsonl
Create embeddings using SentenceTransformer
Store documents in ChromaDB at ./embeddings

📁 Project Structure

DHIGROWTH_RAG/
│
├── app.py                      # FastAPI application
├── config.py                   # Configuration module
├── requirements.txt            # Python dependencies
├── README.md                   # This file
│
├── models/                     # LLM model files
│   └── Dhi_Qwen2P5_0_5B_Q4_K_M.gguf
│
├── data/                       # Training/knowledge base data
│   └── dhigrowth_final_rag_ft.jsonl
│
├── embeddings/                 # ChromaDB persistent storage
│   └── (auto-generated)
│
└── src/                        # Source code
    ├── embed_loader.py        # Vector database creation
    ├── retriever.py            # Document retrieval module
    ├── llm_wrapper.py          # LLM model wrapper
    ├── rag_pipeline.py         # Main RAG pipeline
    └── logger_config.py        # Logging configuration

⚙️ Configuration

Configuration is managed through environment variables or config.py. Create a .env file in the project root:

env

# Model paths
MODEL_PATH=./models/Dhi_Qwen2P5_0_5B_Q4_K_M.gguf
EMBEDDINGS_PATH=./embeddings
DATA_PATH=./data/dhigrowth_final_rag_ft.jsonl

# Embedding model
EMBEDDING_MODEL=all-MiniLM-L6-v2

# Generation settings
MAX_TOKENS=400
TOP_K=3

# LLM settings
N_CTX=4096
TEMPERATURE=0.0
TOP_P=1.0

# Server settings
PORT=8001

Configuration Options

MODEL_PATH: Path to the GGUF model file
EMBEDDINGS_PATH: ChromaDB storage directory
DATA_PATH: Path to JSONL knowledge base file
EMBEDDING_MODEL: SentenceTransformer model name
MAX_TOKENS: Maximum tokens for LLM generation
TOP_K: Number of documents to retrieve
N_CTX: Context window size
TEMPERATURE: Sampling temperature (0.0 = deterministic)
TOP_P: Nucleus sampling parameter
PORT: Server port (default: 8001)

🚀 Usage

Starting the API Server

bash

# Activate virtual environment
source venv/Scripts/activate  # Windows: venv\Scripts\activate

# Start the server
uvicorn app:app --reload --port 8001

The API will be available at http://localhost:8001

Interactive API Documentation

Once the server is running, visit:

Swagger UI: http://localhost:8001/docs
ReDoc: http://localhost:8001/redoc

Using the RAG Pipeline Directly

python

from src.rag_pipeline import rag_chat

# Ask a question
answer = rag_chat("What services does Dhigrowth offer?")
print(answer)

📡 API Documentation

Endpoints

`GET /`

Root endpoint with API information.

Response:

json

{
  "message": "Dhigrowth RAG Chatbot API",
  "version": "1.0.0",
  "endpoints": {
    "health": "/health",
    "ask": "/ask (POST)"
  }
}

`GET /health`

Health check endpoint.

Response:

json

{
  "status": "healthy",
  "service": "Dhigrowth RAG Chatbot",
  "version": "1.0.0"
}

`POST /ask`

Ask a question to the chatbot.

Request Body:

json

{
  "question": "What services does Dhigrowth offer?"
}

Response:

json

{
  "question": "What services does Dhigrowth offer?",
  "answer": "Dhigrowth provides website development, application development, digital marketing, SEO improvement, business automation, and custom business growth strategies."
}

Error Responses:

400 Bad Request: Invalid input
422 Unprocessable Entity: Validation error
500 Internal Server Error: Server error

Example API Calls

Using cURL:

bash

curl -X POST http://localhost:8001/ask \
  -H "Content-Type: application/json" \
  -d '{"question": "How do I book a consultation?"}'

Using Python:

python

import requests

response = requests.post(
    "http://localhost:8001/ask",
    json={"question": "What services does Dhigrowth offer?"}
)
print(response.json())

Using JavaScript:

javascript

fetch("http://localhost:8001/ask", {
  method: "POST",
  headers: {
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    question: "What services does Dhigrowth offer?",
  }),
})
  .then((response) => response.json())
  .then((data) => console.log(data));

🧪 Testing

Running Tests

Test the RAG pipeline directly:

python -c "from src.rag_pipeline import rag_chat; print(rag_chat('What services does Dhigrowth offer?'))"

Test the API using FastAPI TestClient:

python -c "from app import app; from fastapi.testclient import TestClient; client = TestClient(app); print(client.get('/health').json())"

Test Cases

The system has been tested with:

✅ Valid questions about Dhigrowth
✅ Out-of-scope questions (returns "I don't have that information")
✅ Input validation (empty/whitespace questions)
✅ Error handling

See TEST_RESULTS.md for detailed test results.

🚢 Deployment

Option 1: Local Development

uvicorn app:app --reload --port 8001

Option 2: Production with Uvicorn

uvicorn app:app --host 0.0.0.0 --port 8001 --workers 4

Option 3: Docker (Recommended for Production)

Create a Dockerfile:

dockerfile

FROM python:3.11-slim

WORKDIR /app

COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

COPY . .

EXPOSE 8001

CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "8001"]

Build and run:

bash

docker build -t dhigrowth-rag .
docker run -p 8001:8001 dhigrowth-rag

Option 4: Cloud Deployment

VPS Deployment:

Upload project files
Install dependencies
Run with process manager (systemd, supervisor, or PM2)
Configure reverse proxy (nginx)

Platform-as-a-Service:

Deploy to Heroku, Railway, or Render
Configure environment variables
Set up persistent storage for embeddings

🔧 Troubleshooting

Common Issues

Issue: Model file not found

Solution: Ensure Dhi_Qwen2P5_0_5B_Q4_K_M.gguf is in the models/ directory

Issue: ChromaDB collection empty

Solution: Run python src/embed_loader.py to create the vector database

Issue: Import errors

Solution: Ensure virtual environment is activated and dependencies are installed

Issue: Port already in use

Solution: Change port: uvicorn app:app --port 8001

Issue: Slow response times

Solution:
- First request is slower (model loading)
- Subsequent requests are faster
- Consider using GPU for faster inference

Logs

Check logs in dhigrowth_rag.log for detailed error information.

📊 Performance

Model Loading: ~2-3 seconds (first request only)
Embedding Model: ~5 seconds (first request only)
Query Processing: ~2-3 seconds per query
Vector Database: 1,601 documents indexed

🔒 Security Considerations

Input validation prevents injection attacks
CORS can be configured for specific origins in production
Consider adding authentication for production deployment
Rate limiting recommended for public APIs

📝 License

[Add your license information here]

👥 Contributing

[Add contribution guidelines if applicable]

📧 Contact

For questions or support, contact [your contact information]

🙏 Acknowledgments

Qwen2.5 model by Alibaba Cloud
ChromaDB for vector database
SentenceTransformers for embeddings
FastAPI for the web framework

Version: 1.0.0
Last Updated: November 2024

mathanmithun/DhiGrowth_FAQ_RAG

Dhigrowth RAG Chatbot

🚀 Features

📋 Table of Contents

🛠️ Installation

Prerequisites

Step 1: Clone the Repository

Step 2: Create Virtual Environment

Step 3: Install Dependencies

Step 4: Set Up Vector Database

📁 Project Structure

⚙️ Configuration

Configuration Options

🚀 Usage

Starting the API Server

Interactive API Documentation

Using the RAG Pipeline Directly

📡 API Documentation

Endpoints

GET /

GET /health

POST /ask

Example API Calls

🧪 Testing

Running Tests

Test Cases

🚢 Deployment

Option 1: Local Development

Option 2: Production with Uvicorn

Option 3: Docker (Recommended for Production)

Option 4: Cloud Deployment

🔧 Troubleshooting

Common Issues

Logs

📊 Performance

🔒 Security Considerations

📝 License

👥 Contributing

📧 Contact

🙏 Acknowledgments

mathanmithun/
DhiGrowth_FAQ_RAG

`GET /`

`GET /health`

`POST /ask`