feat: framework predictor de/serial override args

All framework *Predictor classes accept constructor arguments to
override the default `serializer` & `deserializer` logic (like
TensorFlowPredictor did already).

Author: athewsey

src/sagemaker/chainer/model.py

@@ -38,7 +38,13 @@ class ChainerPredictor(Predictor):
     multidimensional tensors for Chainer inference.
     """
 
-    def __init__(self, endpoint_name, sagemaker_session=None):
+    def __init__(
+        self,
+        endpoint_name,
+        sagemaker_session=None,
+        serializer=NumpySerializer(),
+        deserializer=NumpyDeserializer(),
+    ):
         """Initialize an ``ChainerPredictor``.
 
         Args:
@@ -48,9 +54,17 @@ def __init__(self, endpoint_name, sagemaker_session=None):
                 manages interactions with Amazon SageMaker APIs and any other
                 AWS services needed. If not specified, the estimator creates one
                 using the default AWS configuration chain.
+            serializer (sagemaker.serializers.BaseSerializer): Optional. Default
+                serializes input data to .npy format. Handles lists and numpy
+                arrays.
+            deserializer (sagemaker.deserializers.BaseDeserializer): Optional.
+                Default parses the response from .npy format to numpy array.
         """
         super(ChainerPredictor, self).__init__(
-            endpoint_name, sagemaker_session, NumpySerializer(), NumpyDeserializer()
+            endpoint_name,
+            sagemaker_session,
+            serializer=serializer,
+            deserializer=deserializer,
         )
 
 

src/sagemaker/mxnet/model.py

@@ -40,7 +40,13 @@ class MXNetPredictor(Predictor):
     multidimensional tensors for MXNet inference.
     """
 
-    def __init__(self, endpoint_name, sagemaker_session=None):
+    def __init__(
+        self,
+        endpoint_name,
+        sagemaker_session=None,
+        serializer=JSONSerializer(),
+        deserializer=JSONDeserializer(),
+    ):
         """Initialize an ``MXNetPredictor``.
 
         Args:
@@ -50,9 +56,16 @@ def __init__(self, endpoint_name, sagemaker_session=None):
                 manages interactions with Amazon SageMaker APIs and any other
                 AWS services needed. If not specified, the estimator creates one
                 using the default AWS configuration chain.
+            serializer (callable): Optional. Default serializes input data to
+                json. Handles dicts, lists, and numpy arrays.
+            deserializer (callable): Optional. Default parses the response using
+                ``json.load(...)``.
         """
         super(MXNetPredictor, self).__init__(
-            endpoint_name, sagemaker_session, JSONSerializer(), JSONDeserializer()
+            endpoint_name,
+            sagemaker_session,
+            serializer=serializer,
+            deserializer=deserializer,
         )
 
 

src/sagemaker/pytorch/model.py

@@ -39,7 +39,13 @@ class PyTorchPredictor(Predictor):
     multidimensional tensors for PyTorch inference.
     """
 
-    def __init__(self, endpoint_name, sagemaker_session=None):
+    def __init__(
+        self,
+        endpoint_name,
+        sagemaker_session=None,
+        serializer=NumpySerializer(),
+        deserializer=NumpyDeserializer(),
+    ):
         """Initialize an ``PyTorchPredictor``.
 
         Args:
@@ -49,9 +55,17 @@ def __init__(self, endpoint_name, sagemaker_session=None):
                 manages interactions with Amazon SageMaker APIs and any other
                 AWS services needed. If not specified, the estimator creates one
                 using the default AWS configuration chain.
+            serializer (sagemaker.serializers.BaseSerializer): Optional. Default
+                serializes input data to .npy format. Handles lists and numpy
+                arrays.
+            deserializer (sagemaker.deserializers.BaseDeserializer): Optional.
+                Default parses the response from .npy format to numpy array.
         """
         super(PyTorchPredictor, self).__init__(
-            endpoint_name, sagemaker_session, NumpySerializer(), NumpyDeserializer()
+            endpoint_name,
+            sagemaker_session,
+            serializer=serializer,
+            deserializer=deserializer,
         )
 
 

src/sagemaker/sklearn/model.py

@@ -34,7 +34,13 @@ class SKLearnPredictor(Predictor):
     multidimensional tensors for Scikit-learn inference.
     """
 
-    def __init__(self, endpoint_name, sagemaker_session=None):
+    def __init__(
+        self,
+        endpoint_name,
+        sagemaker_session=None,
+        serializer=NumpySerializer(),
+        deserializer=NumpyDeserializer(),
+    ):
         """Initialize an ``SKLearnPredictor``.
 
         Args:
@@ -44,9 +50,17 @@ def __init__(self, endpoint_name, sagemaker_session=None):
                 manages interactions with Amazon SageMaker APIs and any other
                 AWS services needed. If not specified, the estimator creates one
                 using the default AWS configuration chain.
+            serializer (sagemaker.serializers.BaseSerializer): Optional. Default
+                serializes input data to .npy format. Handles lists and numpy
+                arrays.
+            deserializer (sagemaker.deserializers.BaseDeserializer): Optional.
+                Default parses the response from .npy format to numpy array.
         """
         super(SKLearnPredictor, self).__init__(
-            endpoint_name, sagemaker_session, NumpySerializer(), NumpyDeserializer()
+            endpoint_name,
+            sagemaker_session,
+            serializer=serializer,
+            deserializer=deserializer,
         )
 
 

src/sagemaker/sparkml/model.py

@@ -31,7 +31,13 @@ class SparkMLPredictor(Predictor):
     list.
     """
 
-    def __init__(self, endpoint_name, sagemaker_session=None, **kwargs):
+    def __init__(
+        self,
+        endpoint_name,
+        sagemaker_session=None,
+        serializer=CSVSerializer(),
+        **kwargs,
+    ):
         """Initializes a SparkMLPredictor which should be used with SparkMLModel
         to perform predictions against SparkML models serialized via MLeap. The
         response is returned in text/csv format which is the default response
@@ -43,12 +49,14 @@ def __init__(self, endpoint_name, sagemaker_session=None, **kwargs):
                 manages interactions with Amazon SageMaker APIs and any other
                 AWS services needed. If not specified, the estimator creates one
                 using the default AWS configuration chain.
+            serializer (sagemaker.serializers.BaseSerializer): Optional. Default
+                serializes input data to text/csv.
         """
         sagemaker_session = sagemaker_session or Session()
         super(SparkMLPredictor, self).__init__(
             endpoint_name=endpoint_name,
             sagemaker_session=sagemaker_session,
-            serializer=CSVSerializer(),
+            serializer=serializer,
             **kwargs,
         )
 

src/sagemaker/xgboost/model.py

@@ -35,7 +35,13 @@ class XGBoostPredictor(Predictor):
     for XGBoost inference.
     """
 
-    def __init__(self, endpoint_name, sagemaker_session=None):
+    def __init__(
+        self,
+        endpoint_name,
+        sagemaker_session=None,
+        serializer=LibSVMSerializer(),
+        deserializer=CSVDeserializer(),
+    ):
         """Initialize an ``XGBoostPredictor``.
 
         Args:
@@ -44,9 +50,16 @@ def __init__(self, endpoint_name, sagemaker_session=None):
                 interactions with Amazon SageMaker APIs and any other AWS services needed.
                 If not specified, the estimator creates one using the default AWS configuration
                 chain.
+            serializer (sagemaker.serializers.BaseSerializer): Optional. Default
+                serializes input data to LibSVM format
+            deserializer (sagemaker.deserializers.BaseDeserializer): Optional.
+                Default parses the response from text/csv to a Python list.
         """
         super(XGBoostPredictor, self).__init__(
-            endpoint_name, sagemaker_session, LibSVMSerializer(), CSVDeserializer()
+            endpoint_name,
+            sagemaker_session,
+            serializer=serializer,
+            deserializer=deserializer,
         )