Yolab-Research

Progress reports for virtual-histology-ai-studio.

Publishing environment¶

Use Python 3.11 and Node.js 24. The repository-local .venv is for Jupyter Book publishing. To recreate it:

python3.11 -m venv .venv
source .venv/bin/activate
python -m pip install -r requirements-report.txt

On this workstation, Python 3.11 is also available through ../virtual-histology-ai-studio/.private-dev/venv-gpu/bin/python; use that executable instead of python3.11 if it is not on PATH.

Build saved notebook outputs and Markdown into a static website:

source .venv/bin/activate
jupyter book build --html --strict

The output is _build/html. Preview with jupyter book start and open the printed URL. With Remote SSH, forward the printed port to your local machine. The first build downloads the book theme and its web dependencies.

requirements-report.in declares direct dependencies; requirements-report.txt records installed versions for reproducible setup. Cloudflare should install the latter and run the HTML build command, with _build/html as its output directory.

Research notebooks¶

For notebooks in either repository, select the existing Python 3.11 (UNI2 GPU) kernel in VS Code. Its interpreter is:

/home/d2mc/Projs/CCOP/virtual-histology-ai-studio/.private-dev/venv-gpu/bin/python

The publishing .venv does not contain the scientific Python stack. Execute research notebooks with the AI Studio kernel, save their outputs, then build the report without executing notebooks in the publishing environment. Use explicit paths to research code and data; selecting a kernel does not change a notebook’s working directory.

Commit selected report content and figures only. Keep credentials, model checkpoints, and raw datasets outside the report source.