Testing

DeepTab uses pytest with pytest-cov for test coverage. The test suite runs against all supported Python versions and operating systems on every push and pull request.

Running the test suite

Goal	Command
Full suite with coverage	just test
A single file	poetry run pytest tests/test_models.py -v
A single test	poetry run pytest tests/test_models.py::test_name -v
Live logs, stop on first fail	poetry run pytest tests/ -x -s

just test expands to poetry run pytest --cov=deeptab tests/.

Writing new tests

Convention	Guideline
Location	Place tests in tests/ using the test_*.py naming convention.
Variations	Use @pytest.mark.parametrize instead of copy-pasting near-identical tests.
Data	Use small synthetic datasets (n=64, d=8); never download external data.
Trainer noise	Silence Lightning output with logging.getLogger("lightning.pytorch").setLevel(logging.ERROR).

import pytest

@pytest.mark.parametrize("n_layers", [1, 2, 4])
def test_depth(n_layers):
    ...

Coverage

A coverage report is printed after every just test run. For an interactive HTML report:

poetry run pytest --cov=deeptab --cov-report=html tests/
open htmlcov/index.html

The full suite also runs in CI across every supported Python and OS combination. See CI/CD for the matrix.

Pre-push checks

The pre-commit configuration includes a push-stage hook that runs pyright type checking before git push. This is installed automatically by just install. To run it manually:

just check

The full test suite is not part of the push hook; it runs in CI on every push and pull request. Run just test locally before pushing if your change touches model or training code.