|
|
|
@ -13,18 +13,23 @@ if hasattr(torch._C, '_jit_set_profiling_executor'):
|
|
|
|
|
torch._C._jit_set_profiling_executor(True)
|
|
|
|
|
torch._C._jit_set_profiling_mode(False)
|
|
|
|
|
|
|
|
|
|
# transformer models don't support many of the spatial / feature based model functionalities
|
|
|
|
|
NON_STD_FILTERS = ['vit_*', 'deit_*']
|
|
|
|
|
|
|
|
|
|
# exclude models that cause specific test failures
|
|
|
|
|
if 'GITHUB_ACTIONS' in os.environ: # and 'Linux' in platform.system():
|
|
|
|
|
# GitHub Linux runner is slower and hits memory limits sooner than MacOS, exclude bigger models
|
|
|
|
|
EXCLUDE_FILTERS = ['*efficientnet_l2*', '*resnext101_32x48d', 'vit_*', '*in21k', '*152x4_bitm']
|
|
|
|
|
EXCLUDE_FILTERS = ['*efficientnet_l2*', '*resnext101_32x48d', '*in21k', '*152x4_bitm'] + NON_STD_FILTERS
|
|
|
|
|
else:
|
|
|
|
|
EXCLUDE_FILTERS = ['vit_*']
|
|
|
|
|
EXCLUDE_FILTERS = NON_STD_FILTERS
|
|
|
|
|
|
|
|
|
|
MAX_FWD_SIZE = 384
|
|
|
|
|
MAX_BWD_SIZE = 128
|
|
|
|
|
MAX_FWD_FEAT_SIZE = 448
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.timeout(120)
|
|
|
|
|
@pytest.mark.parametrize('model_name', list_models(exclude_filters=EXCLUDE_FILTERS[:-1]))
|
|
|
|
|
@pytest.mark.parametrize('model_name', list_models(exclude_filters=EXCLUDE_FILTERS[:-2]))
|
|
|
|
|
@pytest.mark.parametrize('batch_size', [1])
|
|
|
|
|
def test_model_forward(model_name, batch_size):
|
|
|
|
|
"""Run a single forward pass with each model"""
|
|
|
|
@ -68,7 +73,7 @@ def test_model_backward(model_name, batch_size):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.timeout(120)
|
|
|
|
|
@pytest.mark.parametrize('model_name', list_models(exclude_filters=['vit_*']))
|
|
|
|
|
@pytest.mark.parametrize('model_name', list_models(exclude_filters=NON_STD_FILTERS))
|
|
|
|
|
@pytest.mark.parametrize('batch_size', [1])
|
|
|
|
|
def test_model_default_cfgs(model_name, batch_size):
|
|
|
|
|
"""Run a single forward pass with each model"""
|
|
|
|
@ -121,7 +126,7 @@ if 'GITHUB_ACTIONS' not in os.environ:
|
|
|
|
|
create_model(model_name, pretrained=True, in_chans=in_chans)
|
|
|
|
|
|
|
|
|
|
@pytest.mark.timeout(120)
|
|
|
|
|
@pytest.mark.parametrize('model_name', list_models(pretrained=True, exclude_filters=['vit_*']))
|
|
|
|
|
@pytest.mark.parametrize('model_name', list_models(pretrained=True, exclude_filters=NON_STD_FILTERS))
|
|
|
|
|
@pytest.mark.parametrize('batch_size', [1])
|
|
|
|
|
def test_model_features_pretrained(model_name, batch_size):
|
|
|
|
|
"""Create that pretrained weights load when features_only==True."""
|
|
|
|
|