Installation¶

Requirements¶

IMVpy supports Python 3.9 and newer. The base installation includes NumPy, pandas, SciPy, scikit-learn, Matplotlib, seaborn, joblib, and tqdm.

Install the released distribution from PyPI:

python -m pip install imvpy
python -c "import imvpy; print(imvpy.__version__)"

The project is branded IMVpy, while both the distribution and import package use the normalized lowercase name imvpy.

Use editable mode when working from a clone. The project uses a src/ layout, so adding only the repository root to PYTHONPATH is not a supported installation:

python -m pip install -e ".[dev]"

Optional extras¶

Extra	Install command	Purpose
Base	`python -m pip install imvpy`	Core metric, evaluators, and plotting
Progress	`python -m pip install "imvpy[progress]"`	joblib-aware coalition progress
Deep learning	`python -m pip install "imvpy[deep-learning]"`	PyTorch training and BERT layer surgery
Tests	`python -m pip install ".[test]"`	pytest and contract-test dependencies
Documentation	`python -m pip install -e ".[docs]"`	MkDocs site build
Release	`python -m pip install -e ".[release]"`	build and distribution validation tools
Development	`python -m pip install -e ".[dev]"`	All contributor tools

PyTorch is lazy-loaded. Calling vanilla_imv, computing an ablation matrix from saved prediction frames, or importing imvpy does not require the deep-learning extra. Constructing AblationIMV, calling its seed/training methods, or reducing BERT layers does.

Build the documentation¶

python -m pip install -e ".[docs]"
mkdocs serve
mkdocs build --strict --quiet

mkdocs serve exposes a local development site at http://127.0.0.1:8000. The strict production build writes generated HTML under the ignored site/ directory and treats warnings, broken navigation, and unresolved API objects as failures.

Conda¶

The repository also provides a Miniforge-compatible environment:

conda env create -f environment.yml
conda activate imvpy

pyproject.toml remains authoritative for package dependency ranges.