Thank you for helping improve SASO. This repository is a research artifact, so changes should remain traceable to the accompanying scientific workflow.
conda env create -f environment.yml
conda activate saso
python scripts/validate_repository.py
python -m unittest discover -s tests -v- Keep pull requests focused and explain the scientific or maintenance motivation.
- Do not change model architectures, feature definitions, preprocessing, training hyperparameters, or released checkpoints without documenting the effect on reported results.
- Add or update tests for reusable utility functions.
- Do not commit generated outputs, local environments, credentials, or proprietary data.
- Preserve the original data columns and clearly document any migration.
- Use explicit units in code, documentation, and result tables.
A useful pull request description includes:
- what changed;
- why the change is needed;
- whether scientific results or parameters are affected;
- validation commands and their output;
- any remaining reproducibility limitations.
Documentation-only and test-only improvements should explicitly state that no model code or parameters were changed.