Description
Description
Some of the object detection tests have the gpu marker and hence only get tested on GPU machines, even though these tests could also run on CPU. See an example below. The reason is the dependency on od_detection_learner which internally runs 1 training epoch and as part of that runs coco evaluation which has the dependency on GPU. In theory this (and other) tests should also run on un-trained models.
@pytest.mark.gpu
@pytest.fixture(scope="session")
def saved_model(od_detection_learner, tiny_od_data_path) -> Union[str, Path]:
""" A saved model so that loading functions can reuse. """
model_name = "test_fixture_model"
od_detection_learner.save(model_name)
assert (Path(tiny_od_data_path) / "models" / model_name).exists()
return model_name, Path(tiny_od_data_path) / "models"
In which platform does it happen?
Only non-gpu tests are affected.
Expected behavior (i.e. solution)
More tests get executed on CPU-only machines