Amazon SageMaker
Developer Guide

Step 4: Invoke the Endpoint to Get Inferences

To get inferences, send requests using the predict method of the TensorFlowPredictor object. (This object was returned in Step 3). The predict method calls the Amazon SageMaker InvokeEndpoint.

Run the predict method:

iris_predictor.predict([6.4, 3.2, 4.5, 1.5]) #expected label to be 1

For more information about the input features, see About the Training Dataset.

The model predicts that the flower species is Iris versicolor (label 1) with a probability of 97%:

{u'result': {u'classifications': [{u'classes': [{u'label': u'0', u'score': 0.009605729021131992}, {u'label': u'1', u'score': 0.9699361324310303}, {u'label': u'2', u'score': 0.02045818418264389}]}]}}

Your Amazon SageMaker notebook instance includes additional examples.