feature: support model.register() with triton model

src/sagemaker/serve/marshalling/triton_translator.py

@@ -16,6 +16,8 @@ def __init__(self) -> None:
         import torch
 
         self.convert_from_numpy = torch.from_numpy  # pylint: disable=E1101
+        self.CONTENT_TYPE = "tensor/pt"
+        self.ACCEPT = "tensor/pt"
 
     def serialize(self, data, content_type: str = "tensor/pt"):
         """Translate torch.Tensor to numpy ndarray"""
@@ -45,6 +47,8 @@ def __init__(self) -> None:
         import tensorflow as tf
 
         self.convert_to_tensor = tf.convert_to_tensor
+        self.CONTENT_TYPE = "tensor/tf"
+        self.ACCEPT = "tensor/tf"
 
     def serialize(self, data, content_type: str = "tensor/tf"):
         """Translate tf.Tensor to numpy ndarray"""
@@ -70,6 +74,10 @@ def _deserializer(self):
 class NumpyTranslator:
     """A dummy class to make sure the translator interface is aligned"""
 
+    def __init__(self) -> None:
+        self.CONTENT_TYPE = "application/x-npy"
+        self.ACCEPT = "application/x-npy"
+
     def serialize(self, data, content_type: str = "application/x-npy"):
         """Placeholder docstring"""
         return data
@@ -86,6 +94,10 @@ def _deserializer(self):
 class ListTranslator:
     """Translate python list from and to numpy.ndarray"""
 
+    def __init__(self) -> None:
+        self.CONTENT_TYPE = "application/list"
+        self.ACCEPT = "application/list"
+
     def serialize(self, data, content_type: str = "application/list"):
         """Placeholder docstring"""
         try:

src/sagemaker/serve/model_server/triton/triton_builder.py

@@ -430,6 +430,9 @@ def _create_triton_model(self) -> Type[Model]:
         # unique method to models created via ModelBuilder()
         self._original_deploy = self.pysdk_model.deploy
         self.pysdk_model.deploy = self._model_builder_deploy_wrapper
+        self._original_register = self.pysdk_model.register
+        self.pysdk_model.register = self._model_builder_register_wrapper
+        self.model_package = None
         return self.pysdk_model
 
     def _get_triton_predictor(self, endpoint_name: str, sagemaker_session: Session) -> Predictor:

tests/unit/sagemaker/serve/model_server/triton/test_triton_builder.py

@@ -33,6 +33,7 @@
 MOCK_SESSION = Mock()
 MOCK_MODES = Mock()
 MOCK_DEPLOY_WRAPPER = Mock()
+MOCK_RESIGTER_WRAPPER = Mock()
 
 
 class pytorch:
@@ -56,6 +57,7 @@ def prepare_triton_builder_for_model(self, triton_builder: Triton) -> Triton:
         triton_builder.sagemaker_session = MOCK_SESSION
         triton_builder.modes = MOCK_MODES
         triton_builder._model_builder_deploy_wrapper = MOCK_DEPLOY_WRAPPER
+        triton_builder._model_builder_register_wrapper = MOCK_RESIGTER_WRAPPER
         triton_builder.inference_spec = None
 
         mock_export = Mock()