add in-process mode definition to docs (#1622)

Co-authored-by: Gary Wang <38331932+gwang111@users.noreply.github.com>

Author: 2 people

src/sagemaker/serve/builder/model_builder.py

@@ -150,6 +150,7 @@ class ModelBuilder(Triton, DJL, JumpStart, TGI, Transformers, TensorflowServing,
 
             * ``Mode.SAGEMAKER_ENDPOINT``: Launch on a SageMaker endpoint
             * ``Mode.LOCAL_CONTAINER``: Launch locally with a container
+            * ``Mode.IN_PROCESS``: Launch locally to a FastAPI server instead of using a container.
         shared_libs (List[str]): Any shared libraries you want to bring into
             the model packaging.
         dependencies (Optional[Dict[str, Any]): The dependencies of the model
@@ -613,7 +614,9 @@ def _model_builder_deploy_wrapper(
                 )
 
         if "endpoint_logging" not in kwargs:
-            kwargs["endpoint_logging"] = False
+            kwargs["endpoint_logging"] = True
+        kwargs.pop("mode", None)
+        self.pysdk_model.role = kwargs.pop("role", self.pysdk_model.role)
         predictor = self._original_deploy(
             *args,
             instance_type=instance_type,

src/sagemaker/serve/utils/predictors.py

@@ -333,10 +333,14 @@ def __init__(
 
     def predict(self, data):
         """Placeholder docstring"""
-        return self._mode_obj._invoke_serving(
-            self.serializer.serialize(data),
-            self.content_type,
-            self.deserializer.ACCEPT[0],
+        return self.deserializer.deserialize(
+            io.BytesIO(
+                self._mode_obj._invoke_serving(
+                    self.serializer.serialize(data),
+                    self.content_type,
+                    self.accept[0],
+                )
+            )
         )
 
     @property
@@ -357,8 +361,8 @@ def delete_predictor(self):
 def _get_in_process_mode_predictor(
     # model_server: ModelServer,
     mode_obj: Type[InProcessMode],
-    serializer=IdentitySerializer(),
-    deserializer=BytesDeserializer(),
+    serializer=JSONSerializer(),
+    deserializer=JSONDeserializer(),
 ) -> Type[PredictorBase]:
     """Returns Predictor for IN_PROCESS mode"""
     return InProcessModePredictor(