Getting Started¶

This guide walks you through setting up the Scientific Literature Explorer on your machine.

Prerequisites¶

Before installing, ensure you have:

Python ≥ 3.10 (3.11+ recommended)
A ScaleDown API key — Get yours at ScaleDown Getting Started
A Google Gemini API key — Free tier available at Google AI Studio

Installation¶

1. Clone the Repository¶

git clone <repo-url>
cd RAG

2. Create a Virtual Environment¶

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install Dependencies¶

pip install -r requirements.txt

Dependencies include: - requests — HTTP client for ScaleDown, Gemini, and ArXiv APIs - python-dotenv — Load .env configuration - numpy — Array operations for TF-IDF - scikit-learn — TF-IDF vectorizer and cosine similarity - PyPDF2 — PDF text extraction from ArXiv papers - rich — Terminal UI (tables, panels, markdown, progress spinners)

Configuration¶

1. Create `.env` File¶

Copy the example environment file:

cp .env.example .env

2. Add Your API Keys¶

Open .env in your favorite editor and fill in your keys:

# Required
SCALEDOWN_API_KEY=your_scaledown_api_key_here
GEMINI_API_KEY=your_gemini_api_key_here

# Optional Configuration Overrides
SCALEDOWN_MODEL=gemini-2.5-flash    # Target model for compression optimization
GEMINI_MODEL=gemini-2.5-flash       # Gemini model to use
CHUNK_SIZE=1000                      # Characters per chunk
CHUNK_OVERLAP=200                    # Overlap between chunks
TOP_K=5                              # Number of chunks to retrieve
SCALEDOWN_TIMEOUT=15                 # Timeout in seconds for ScaleDown API

See Configuration Reference for detailed explanations of each variable.

Directory Structure¶

The system will automatically create these folders on first use:

RAG/
├── papers/          # Downloaded PDFs and extracted text
├── artifacts/       # Stored reasoning outputs (COT, verify, critique)
│   ├── cot/
│   ├── self_verify/
│   └── self_critique/
└── sessions/        # Conversation history (JSON per session)

Verify Installation¶

Test that everything is working:

# Ask a simple question
python -m src.main ask "What is a convolutional neural network?"

If you see a response, you're ready to go! 🎉

Next: Usage Examples¶

See Usage Guide for all available commands and examples.