Git-LLM Project

A reproducible prototype for personalized RDM training using local LLMs (e.g. Phi) with LlamaIndex and Carpentries content.
Demonstrated here with a Git-focused lesson. Developed with FAIR and open science principles in mind.

⚠️ This project is under active development.
It is intended as an experimental prototype and not yet ready for production use. Contributions and feedback are welcome.

Contents

• git_llm_training_data.json – Example questions and answers based on the Library Carpentry Git lesson
• git_llm_prototype.ipynb – Interactive notebook with a simple retrieval-based system
• requirements.txt – List of Python dependencies

Setup

1. Create and activate virtual environment

python3 -m venv llm-env
source llm-env/bin/activate

2. Install dependencies

pip install -r requirements.txt

3. Start the local LLM using Ollama

Install Ollama (see https://ollama.com):

curl -fsSL https://ollama.com/install.sh | sh

Start the model (Phi-2):

ollama run phi

4. Launch the notebook

jupyter notebook

Then open git_llm_prototype.ipynb and select the kernel that matches your virtual environment.

License

This project and its contents are licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0).
You are free to share and adapt the material, provided that you give appropriate credit.

© 2025 Rabea Müller

---

This project builds on content from the Library Carpentry Git Lesson, available at https://librarycarpentry.org/lc-git/.