Contributing¶

We welcome contributions! Whether you’re adding a new deep learning architecture or a baseline model, here’s how to get started.

Adding a New Model¶

Implement the Module: Place your model in the appropriate directory: * Deep Learning: fishy/models/deep/ * Classic ML: fishy/models/classic/ * Probabilistic: fishy/models/probabilistic/
Register the Model: Open fishy/configs/models.yaml and add your model class path and default hyperparameters.
Update the Factory: If your model requires a unique initialization signature (e.g., specific number of heads), add a condition to fishy/_core/factory.py.

Linting: We use black for formatting and pylint for code quality.
Testing: Ensure any new code has unit tests in the tests/ folder and passing doctests in the docstrings.
Documentation: If you add a new high-level feature, consider adding a tutorial in the examples/ folder.

# Run all unit tests
pytest tests/

# Run all docstring tests
pytest --doctest-modules fishy/