ML unit testing¶
Multiple tests are available to verify different functionalities for either tcbench and the modeling campaigns created.
Tests are not bundled with pypi installation. Rather, you need to follow the procedure described in the artifact page to fetch the source code and install all artifacts and datasets.
Tests are coded via pytest
and are available under the /tests folder.
Tests trigger model training
Most of the test verify that the models train for the campaigns described in the paper are indeed reproducible, i.e., the provide the exact same models obtained for the paper.
To do so, the pytest resources fetched from figshare contains a subset of reference models so the test trigger the modeling for those scenarios and check that what trained matches what created for the paper.
So be aware that running these tests might take a while depending on your local environment.
To trigger all tests run
Output
============================ test session starts ======================================
platform linux -- Python 3.10.13, pytest-7.4.2, pluggy-1.3.0
rootdir: /tmp/tcbench-pip/tcbench
plugins: anyio-3.7.1, helpers-namespace-2021.12.29
collected 101 items
tests/test_augmentations_at_loading.py ........... [ 10%]
tests/test_augmentations_at_loading_xgboost.py . [ 11%]
tests/test_cli_command_campaign.py .... [ 15%]
tests/test_cli_command_singlerun.py ............ [ 27%]
tests/test_contrastive_learning_and_finetune.py .. [ 29%]
tests/test_libtcdatasets_datasets_utils.py ................. [ 46%]
tests/test_modeling_backbone.py ................ [ 62%]
tests/test_modeling_dataprep.py .................................. [ 96%]
tests/test_modeling_methods.py .... [100%]
============================== 101 passed, 8 warnings in 6523.55s (1:48:43) =========================