In production env, Pytorch Lightning seems like a great option. I was wondering if you could provide a boiler template for serving / training with Sagemaker integration.
Hi @tchaton , we're not yet at the point, but we can think about that as well!
Dear @justusschock,
Thanks for answering me !
I am going to give a try with my team.
If we come up something nice enough, I would share it with you guys.
Best regards,
T.C
I have found this example which can the base for a production ready template: https://github.com/awslabs/amazon-sagemaker-examples/blob/master/aws_marketplace/creating_marketplace_products/Bring_Your_Own-Creating_Algorithm_and_Model_Package.ipynb
Enabling people to publish their own algorithm to AWS Marketplace using Pytorch-lighthning.
And this amazing repo too: https://github.com/SeanNaren/deepspeech.pytorch which can be used for inspiration using hydra for configuration.
Hey @justusschock,
I have started to work on it: https://github.com/tchaton/sagemaker-torch-template
Best regards,
Thomas Chaton.
Thanks, @tchaton , could you ping me, if you got a running version?
Hey @justusschock,
Do you have a Slack where it would be more convenient to talk ? I have some questions about Pytorch Lightning. I want to make sure it will be compatible with my template.
Best regards,
Thomas Chaton.
Sure, there's a slack link in our readme
Hey @justusschock,
Thanks. Training is working fine 馃憤
Tqdm to teak and some trouble with hydra new structured config, but It should be solved quickly.

Best regards,
T.C
Most helpful comment
Dear @justusschock,
Thanks for answering me !
I am going to give a try with my team.
If we come up something nice enough, I would share it with you guys.
Best regards,
T.C