Please provide at lest some more information concerning the topic.
C2_W2_Assignment: Train a review classifier with BERT and Amazon SageMaker.
Why we need to create custom
SentimentPredictor class and then pass it as a parameter to PyTorchModel?
“Create a custom
SentimentPredictor that encapsulates a JSONLines serializer and deserializer. To be passed into the
PyTorchModel it needs to be wrapped as a class.”
I can see similar question here BERT Deployment - Custom Inference , but there is no answer