notebook
June 19, 2026 ยท View on GitHub
Interactive tutorial notebooks for OneComp.
For Google Colab setup and a walkthrough of each section, see the Tutorial Notebook guide in the documentation.
Notebooks
| Notebook | Description |
|---|---|
| 01_tutorial.ipynb | Hands-on tour: 4-bit RTN visualization, Runner.auto_run, and vLLM chat inference |
Local setup
This directory has its own uv project that installs OneComp from the repository root in editable mode.
cd notebook
uv sync
uv run jupyter lab
Open 01_tutorial.ipynb and run the cells from top to bottom.
Google Colab
- Open the notebook in Colab with the badge above (or upload
01_tutorial.ipynb). - Select a Python 3.12 runtime with a GPU (T4 is enough).
- In section 0. Setup, uncomment and run the
pip installlines in the first code cell. - Run the remaining cells in order.
For the vLLM section (section 4), install the vLLM extra:
!pip install -q "onecomp[cu130,vllm]" matplotlib
The tutorial uses TinyLlama/TinyLlama-1.1B-Chat-v1.0 so it runs comfortably on a Colab T4
without a Hugging Face login.
Requirements
- Python 3.12+ (OneComp requirement)
onecomp >= 1.2.0