SyntaxSphinx/Itivrit-ai

Indian History RAG System 🪔

A Retrival-Augmented Generation (RAG) system for exploring Indian history using authenticated sources. This application uses Sentence Transformers for semantic search and an Ollama LLM to answer questions about India's rich historical heritage.

Features

• —Semantic search across multiple historical texts
• —Accurate page citations for all sources
• —Beautiful Streamlit UI with Indian-themed styling
• —Contextual answers grounded in retrieved documents

Setup

Local Development

1. 1.Clone this repository:

git clone <your-repo-url>
cd india-history-rag

1. 1.Install dependencies:

pip install -r requirements.txt

1. 1.Download pre-built FAISS index and embeddings:

• —Ensure data/vector_db_faiss/faiss.index and data/vector_db_faiss/metadata.json are present

1. 1.Start Ollama server:

ollama serve
ollama pull mistral:7b

1. 1.Run the Streamlit app:

streamlit run streamlit_app.py

Deployment

Streamlit Community Cloud

1. 1.Push your code to GitHub
2. 2.Go to streamlit.io/cloud
3. 3.Connect your GitHub repository
4. 4.Deploy!

Note: Ensure your FAISS index files are committed to the repository or rebuild them during deployment.

Structure

india-history-rag/
├── streamlit_app.py          # Main Streamlit UI
├── rag_pipeline.py           # RAG orchestration
├── retriever.py              # FAISS-based retrieval
├── llm_ollama.py             # Ollama LLM client
├── generate_embeddings_local.py  # Build FAISS index
└── data/
    └── vector_db_faiss/      # FAISS index and metadata

License

MIT License