🚀 UltraRAG - Complete Unified Package

The ONLY RAG package you need!

✅ Revolutionary RAG engine
✅ Built-in Ollama integration
✅ Built-in FastAPI + Swagger UI
✅ ONE command to start!
✅ Zero configuration needed!

---

🎯 Quick Start (2 Commands!)

Step 1: Install

pip install ultrarag[server]

Step 2: Start Server

ultrarag serve --ollama-host localhost --ollama-model llama3.2

Open Swagger UI:

http://localhost:8000/docs

🎉 DONE! RAG chatbot ready!

---

📚 Three Ways to Use

Method 1: Web Server (with Swagger UI)

# Start server
ultrarag serve --ollama-host localhost --ollama-model llama3.2

# Open browser → http://localhost:8000/docs
# Upload documents, ask questions via Swagger!

Method 2: Python Code (Simple)

from ultrarag import RAG

# Create RAG
rag = RAG()

# Add document
rag.add("Python is a programming language created by Guido van Rossum.")

# Ask question
answer = rag.ask("Who created Python?")
print(answer)
# Output: "Python is a programming language created by Guido van Rossum."

Method 3: Python Code (with Ollama)

from ultrarag import RAG, OllamaLLM

# Initialize Ollama
llm = OllamaLLM(host="localhost", port=11434, model="llama3.2")

# Create RAG
rag = RAG()

# Add document
rag.add("Python is used for AI, web development, and data science.")

# Get context
query = "What is Python used for?"
query_analysis = rag.query_processor.analyze(query)
chunks = rag.retriever.retrieve(rag.chunks, query_analysis, top_k=3)
context = " ".join([c.text for c in chunks])

# Generate with LLM
prompt = f"Based on: {context}\n\nQuestion: {query}\n\nAnswer:"
answer = llm.generate(prompt)
print(answer)

---

🎯 Complete Example (CLI Server)

Prerequisites

# Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# Start Ollama
ollama serve

# Pull model (in another terminal)
ollama pull llama3.2

Install UltraRAG

pip install ultrarag[server]

Start Server

ultrarag serve --ollama-host localhost --ollama-model llama3.2

Output:

🚀 Starting UltraRAG Server...
📡 Ollama: localhost:11434
🤖 Model: llama3.2
📚 Swagger UI: http://localhost:8000/docs

Use Swagger UI

1. Open: http://localhost:8000/docs

2. Upload document:

   • Click POST /upload
   • Choose file
   • Execute

3. Ask question:

   • Click POST /query
   • Enter:

     {
       "question": "Your question?",
       "use_llm": true
     }
     

   • Execute

4. Get answer! ✅

---

🔧 Configuration Options

Server Command

ultrarag serve \
  --ollama-host localhost \      # Ollama IP
  --ollama-port 11434 \          # Ollama port
  --ollama-model llama3.2 \      # Model name
  --port 8000                     # Server port

Different Ollama Machine

ultrarag serve --ollama-host 192.168.1.100 --ollama-model llama3.2

---

📊 API Endpoints

Endpoint	Method	Description
/docs	GET	Swagger UI
/upload	POST	Upload file
/upload-text	POST	Upload text
/query	POST	Ask question
/stats	GET	Statistics
/clear	DELETE	Clear documents

---

💻 Python API

Basic Usage

from ultrarag import RAG

rag = RAG()
rag.add("document text...")
answer = rag.ask("question?")

With Ollama

from ultrarag import RAG, OllamaLLM

llm = OllamaLLM(host="localhost", model="llama3.2")
rag = RAG()

# Test Ollama
if llm.test():
    print("✅ Ollama connected")
else:
    print("❌ Ollama not available")

# Add documents
rag.add("Your documents...")

# Generate answer
chunks = rag.retriever.retrieve(rag.chunks, query_analysis, top_k=3)
context = " ".join([c.text for c in chunks])
answer = llm.generate(f"Context: {context}\n\nQuestion: {question}")

Advanced Usage

# Custom configuration
rag = RAG(
    min_chunk_completeness=0.85,
    min_grounding_score=0.85
)

# Add with metadata
rag.add("text...", metadata={"source": "doc1.pdf"})

# Detailed response
response = rag.ask("question?", explain=True)
print(f"Answer: {response.answer}")
print(f"Confidence: {response.confidence}")
print(f"Grounding: {response.grounding_score}")
print(f"Verdict: {response.metadata['verdict']}")

# Statistics
stats = rag.get_stats()
print(f"Total chunks: {stats['total_chunks']}")

---

🎯 Installation Options

Minimal (RAG only)

pip install ultrarag

Use in Python code only (no web server)

Full (RAG + Web Server)

pip install ultrarag[server]

Includes FastAPI + Swagger UI

From Source

git clone https://github.com/kumar123ips/ultrarag
cd ultrarag
pip install -e .[server]

---

🔥 Features

Revolutionary RAG Components

✅ AtomicChunk - Guaranteed completeness (ICS ≥ 0.75)
✅ QueryDNA - Multi-dimensional query analysis
✅ AdaptiveRetriever - Intent-based retrieval
✅ ProvenAnswer - Mathematical validation
✅ Zero Dependencies - Core package is pure Python!

Built-in Integrations

✅ Ollama - Local LLM support
✅ FastAPI - Production web server
✅ Swagger UI - Interactive API docs
✅ CLI - One command to start!

---

📝 License

MIT License - see LICENSE

---

👤 Author

Abhishek Kumar
Email: ipsabhi420@gmail.com
GitHub: @kumar123ips

---

🎉 Success!

The ONLY RAG package you need!

pip install ultrarag[server]
ultrarag serve --ollama-model llama3.2

That's it! 🚀

---

Made with ❤️ by Abhishek Kumar