science-rag

The RAG-system specialized in answering scientific question. A user asks a question on scientific topic and receives an answer based on relevant articles with links attached.

Steps to set up and launch the RAG system:

1. Add TELEBOT_TOKEN and OPENAI_API_KEY to .env file (in the root).
2. Run: pixi install
3. Run: pixi run python science_rag/telegram_bot.py

Stack:

• pixi for environment and dependencies management;
• telegram as frontend;
• langchain for agent creation;

Project Organization

├── LICENSE            <- MIT license
├── data
│   ├── chroma_data    <- Chroma db.
│
├── models             <- Cached models (if loaded from Hugging Face)
│
├── notebooks          <- Jupyter notebooks with experiments
│
├── pixi.toml          <- Project configuration file with package metadata for
│                         science_rag and configuration for tools like black
│
└── science_rag   <- Source code for use in this project.
    │
    ├── __init__.py             <- Makes science_rag a Python module
    │
    ├── config.py               <- Store useful variables and configuration
    │
    ├── single_rag_agent.py     <- Code with agent
    │
    └── telegram_bot.py         <- Code to run telegram-bot

---