ai-lab

Hands-on experiments with open-weight LLMs: inference pipelines, architecture comparisons, and the underlying math.

Path	Contents
`notebooks/qwen3-0.6b-inference.ipynb`	Qwen3-0.6B inference with `transformers`
`notebooks/gemma3-1b-inference.ipynb`	Gemma-3-1B inference, architecture contrast
`notebooks/linear-algebra.ipynb`	Linear systems and eigendecomposition
`src/ai_lab/`	Reusable package code

Getting started

Requires Python 3.13+ and uv.

uv sync
uv run jupyter lab

Runs on CPU or Apple Silicon (MPS). Weights download on first run; set HF_ENDPOINT to use a mirror:

export HF_ENDPOINT=https://hf-mirror.com

MIT. See LICENSE.