Toolkit for running MXNet training scripts on SageMaker. Dockerfiles used for building SageMaker MXNet Containers are at https://github.com/aws/deep-learning-containers.
=================================== FAILURES ===================================
___________________________ test_training[1-py3-cpu] ___________________________
sagemaker_session = <sagemaker.session.Session object at 0x7fc10750fa10>
ecr_image = '841569659894.dkr.ecr.us-east-1.amazonaws.com/beta-mxnet-training:1.6.0-py3-gpu-with-horovod-build'
instance_type = 'ml.p2.16xlarge', instance_count = 1
@pytest.mark.skip_py2_containers
def test_training(sagemaker_session, ecr_image, instance_type, instance_count):
> from smexperiments.experiment import Experiment
E ImportError: No module named smexperiments.experiment
test/integration/sagemaker/test_experiments.py:33: ImportError
___________________________ test_training[2-py3-cpu] ___________________________
sagemaker_session = <sagemaker.session.Session object at 0x7fc10750fa10>
ecr_image = '841569659894.dkr.ecr.us-east-1.amazonaws.com/beta-mxnet-training:1.6.0-py3-gpu-with-horovod-build'
instance_type = 'ml.p2.16xlarge', instance_count = 2
@pytest.mark.skip_py2_containers
def test_training(sagemaker_session, ecr_image, instance_type, instance_count):
> from smexperiments.experiment import Experiment
E ImportError: No module named smexperiments.experiment
test/integration/sagemaker/test_experiments.py:33: ImportError
=============================== warnings summary ===============================
/usr/local/lib/python2.7/dist-packages/_pytest/mark/structures.py:324
/usr/local/lib/python2.7/dist-packages/_pytest/mark/structures.py:324: PytestUnknownMarkWarning: Unknown pytest.mark.skip_py2_containers - is this a typo? You can register custom marks to avoid this warning - for details, see https://docs.pytest.org/en/latest/mark.html
PytestUnknownMarkWarning,
/usr/local/lib/python2.7/dist-packages/_pytest/mark/structures.py:324
/usr/local/lib/python2.7/dist-packages/_pytest/mark/structures.py:324: PytestUnknownMarkWarning: Unknown pytest.mark.skip_test_in_region - is this a typo? You can register custom marks to avoid this warning - for details, see https://docs.pytest.org/en/latest/mark.html
PytestUnknownMarkWarning,
-- Docs: https://docs.pytest.org/en/latest/warnings.html
============== 2 failed, 7 passed, 2 warnings in 3354.72 seconds ===============