DTScribe: Generating Models for Digital Twin Reporting using Local Large Language Models

Installation

1. Install Ollama

Follow the official instructions for your operating system.
Once installed, pull the required models:

# Example LLM (adjust to desired model, e.g., llama3, qwen, mistral, etc.):
ollama pull llama3.2
# Embedding models:
ollama pull nomic-embed-text
ollama pull embeddinggemma

2. (Optional) Install uv

uv is used as the Python package/dependency manager.

Linux/macOS:

curl -LsSf https://astral.sh/uv/install.sh | sh

Windows PowerShell:

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

3. Install project dependencies

uv sync

or

pip install -e .

Running

To run the extraction, you need to do:

uv run src\experiment_runner.py