Open
Description
Describe the bug
I cannot seem to run sagemaker in "local code" mode with an entry point that is not in the root of the source path. I think this is the reason (or one of them)
To reproduce
place the entry point file in one (or more) folders under source_dir
(i.e. not in the root folder itself, but a child thereof), and use LocalcSession()
with "local code"
by "local code" I mean setting
estimator.sagemaker_session.config = dict(local=dict(local_code=True))
before calling
estimator.fit(...)
again, see here
Expected behavior
A clear and concise description of what you expected to happen.
System information
A description of your system. Please provide:
- SageMaker Python SDK version: master, thou I am using
1.60.2
- Framework name (eg. PyTorch) or algorithm (eg. KMeans):
PyTorch
- Framework version:
1.3.1
- Python version:
3.6
- CPU or GPU: both
- Custom Docker image (Y/N): N
Additional context
Add any other context about the problem here.