Deploy Multiple ML Models on a Single Endpoint Using Multi-model Endpoints on Amazon SageMaker

[@OPopoola]

Thanks for this amazing demo. My question is what if you have models that are not related? All the models you deployed are similar in function with similar input data. Can you deploy models that are different in function and input parameters to the same multi-model endpoint?

[@radoslawwlodarczyk8699]

Thanks for a really nice review. You did not mention other related services, so I would like to ask if there is any additional orchestrator that would take care of registering/deregistering models and model versioning? Also champion-challenger scenario would be interesting, where in fact two models are invoked at once and the differences in their behaviour are logged.

[@WatchUniverseW]

the notebook shown @10:50 seems have a mistake when splitting X_val further

[@2107mann]

Thank you so much helped a lot.

[@bharathjc4700]

Great post !!! can u please drop the link to access notebooks

[@shuaitang2982]

Would you be able to share the notebook and link to dataset so that it's easier to follow?

[@shuaitang2982]

nvm

[@harrisonkane8584]

I've followed the steps here, but when I pass new data to the model to get an inference with the invoke_endpoint() method, the response object is just an empty byte string. I'm a bit confused because I can tell it's finding the model artifact in s3, but nothing is getting returned. Has anyone else encountered this or know how to solve?