Skip to content

changed local models to google gemini

Naga vardhan Reddy requested to merge naga into main

Title

Add Government Scheme RTI Assistant app with RAG utilities

Description

This merge request adds the main Streamlit application and supporting utility modules for the Government Scheme RTI Assistant. The app helps users query Indian government scheme information and generate RTI application drafts.

Changes Made

  • Added app.py as the main application file.
  • Added utils package for core backend logic:
    • embedder.py for HuggingFace embedding model setup.
    • loader.py for ingesting scheme documents into ChromaDB.
    • retriever.py for retrieving relevant scheme data and generating RAG-based answers.
    • rti_generator.py for generating RTI drafts in English and Telugu.
  • Added requirements.txt with project dependencies.
  • Added vercel.json for deployment configuration.
  • Added .gitignore to exclude virtual environment, cache files, local database, and generated files.
  • Added data folder for scheme text files.

Purpose

This update provides the base implementation for a multilingual Government Scheme and RTI assistant using LangChain, ChromaDB, HuggingFace embeddings, and Ollama.

Notes

  • Local files like .venv, chroma_db, and __pycache__ are ignored.
  • There is one local uncommitted change in USER_MANUAL.md; commit it only if it is required for this merge request.

Merge request reports