Skip to content

SageMakerModel.transform() Not Working When Model Constructed With SageMakerModel.fromEndpoint() #34

@robperc

Description

@robperc

I've trained and deployed an XGBoost model for classifying the MNIST dataset. I can call transform() successfully on my test dataset when I use the model object returned by the XGBoostSageMakerEstimator() constructor, however when I try to useSageMakerModel.fromEndpoint() to create my model object and transform the same dataframe I get the following error:

'DataFrame' object has no attribute '_get_object_id'
Traceback (most recent call last):
File "/usr/lib/spark/python/lib/py4j-0.10.4-src.zip/py4j/java_gateway.py", line 1124, in call
args_command, temp_args = self._build_args(*args)
File "/usr/lib/spark/python/lib/py4j-0.10.4-src.zip/py4j/java_gateway.py", line 1094, in _build_args
[get_command_part(arg, self.pool) for arg in new_args])
File "/usr/lib/spark/python/lib/py4j-0.10.4-src.zip/py4j/protocol.py", line 289, in get_command_part
command_part = REFERENCE_TYPE + parameter._get_object_id()
File "/usr/lib/spark/python/lib/pyspark.zip/pyspark/sql/dataframe.py", line 1020, in getattr
"'%s' object has no attribute '%s'" % (self.class.name, name))
AttributeError: 'DataFrame' object has no attribute '_get_object_id'

Attached is a .zip containg screenshots of this error and the .ipynb that can be used to reproduce it.
issue.zip

Here are the screenshots for a quick look:
error
workingtransform

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions