|
|
|
@ -386,11 +386,14 @@ def test_model_forward_fx(model_name, batch_size):
|
|
|
|
|
assert not torch.isnan(outputs).any(), 'Output included NaNs'
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.timeout(120)
|
|
|
|
|
@pytest.mark.parametrize('model_name', list_models(
|
|
|
|
|
if 'GITHUB_ACTIONS' not in os.environ:
|
|
|
|
|
# FIXME this test is causing GitHub actions to run out of RAM and abruptly kill the test process
|
|
|
|
|
|
|
|
|
|
@pytest.mark.timeout(120)
|
|
|
|
|
@pytest.mark.parametrize('model_name', list_models(
|
|
|
|
|
exclude_filters=EXCLUDE_FILTERS + EXCLUDE_FX_FILTERS, name_matches_cfg=True))
|
|
|
|
|
@pytest.mark.parametrize('batch_size', [2])
|
|
|
|
|
def test_model_backward_fx(model_name, batch_size):
|
|
|
|
|
@pytest.mark.parametrize('batch_size', [2])
|
|
|
|
|
def test_model_backward_fx(model_name, batch_size):
|
|
|
|
|
"""Symbolically trace each model and run single backward pass through the resulting GraphModule"""
|
|
|
|
|
if not has_fx_feature_extraction:
|
|
|
|
|
pytest.skip("Can't test FX. Torch >= 1.10 and Torchvision >= 0.11 are required.")
|
|
|
|
@ -418,6 +421,7 @@ def test_model_backward_fx(model_name, batch_size):
|
|
|
|
|
assert num_params == num_grad, 'Some parameters are missing gradients'
|
|
|
|
|
assert not torch.isnan(outputs).any(), 'Output included NaNs'
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# reason: model is scripted after fx tracing, but beit has torch.jit.is_scripting() control flow
|
|
|
|
|
EXCLUDE_FX_JIT_FILTERS = [
|
|
|
|
|
'deit_*_distilled_patch16_224',
|
|
|
|
|