mik-laj edited a comment on issue #10007:
URL: https://github.com/apache/airflow/issues/10007#issuecomment-663996076


   > i thought this is the project that the contribution of sagemaker is being 
done - isn't it ?
   
   SageMaker contributors are not an active part of this community
   
   I still think it will be difficult to fix this problem on the Airflow core. 
We only integrate a ready-made SDK and if the official SDK requires 
dependencies, this SDK should be improved to limit the number of needed 
dependencies. We can look for an alternative unofficial SDK, but the unofficial 
SDK may be the source of other problems e.g. missing features.
   
   Besides, Airflow does not require a Sagemaker SDK and it doesn't use the 
Sagemaker SDK anywhere. You can also prepare the config file manually based on 
the boto3 documentation.
   
https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/sagemaker.html#SageMaker.Client.create_hyper_parameter_tuning_job
   or 
   You can prepare the config in another environment and save it to a JSON 
file, and then load these config in the DAG file.
   Here is an example of how you use this operators without Sagemaker SDK
   
https://github.com/apache/airflow/blob/master/tests/providers/amazon/aws/operators/test_sagemaker_tuning.py#L40-L115
   
   Airflow doesn't require the Sagemaker SDK, nor any libraries that require 
numpy. Your DAG File has this dependency.


----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
[email protected]


Reply via email to