AWS Sagemaker error INVALID_INFERENCE_TYPE

Question

I have a simple sagemaker endpoint hello world model. The container works fine locally but when deployed in Sagemaker endpoint givs error below..

{
    "ErrorCode": "INVALID_INFERENCE_TYPE",
    "Message": "Endpoint myEndpoint does not support this inference type."
}

Answer 1

See comment from @durga_sury above.... quoted below.

You won't need the estimator if you already have a trained model. Any chance it's set as an async endpoint? On the UI when creating the endpoint, if you've entered in params in the Async Invocation Config section, it might be set up as an async endpoint. Otherwise, any chance you can share the code/model? Hard to debug otherwise. – durga_sury Sep 21 at 22:18

Seems just adding async options makes the endpoint async. After that you will get the error if you call it for real-time prediction.