Document Summarizer

A small desktop GUI app to upload files, preview data, generate basic statistics, create embeddings (via Ollama), and ask data-driven questions using retrieval-augmented generation (RAG).

Features

• Upload and preview files
• Basic dataset summary and numeric statistics
• Generate embeddings for efficient retrieval
• Ask questions or click "Summarize" to get concise AI-driven insights
• Switch models from the UI (e.g., gemma3:1b, llama3.2, etc.)

Requirements

• Python 3.8+
• Ollama installed and available in PATH (app will attempt to start it)
• Python packages: customtkinter, requests, numpy (install with pip)

Quick Start

1. Ensure Ollama is installed and accessible (see https://ollama.ai).
2. Install Python packages: pip install customtkinter requests numpy
3. Run the app: python main.py
4. In the UI: Upload a file → (optional - use for large files) Generate Embeddings → Ask questions or click Summarize

Important notes

• Remember to replace OLLAMA_ADDRESS to the relevant address with port.
• This app supports only csvs at the moment. More support will be added later.
• Generating embeddings enables full-data retrieval for more accurate answers; without them the app uses sample rows.
• The app starts/controls an Ollama server process via OllamaServer.py and uses ollama for embeddings and chat.

Files

• Summarizer.py - GUI and core logic
• CSVParser.py - CSV parsing, chunking, embeddings, simple stats
• OllamaServer.py - manages the Ollama server process