removing in_process and minor edits

Author: Bryannah Hernandez

src/sagemaker/serve/builder/model_builder.py

@@ -36,7 +36,6 @@
 from sagemaker.serve.mode.function_pointers import Mode
 from sagemaker.serve.mode.sagemaker_endpoint_mode import SageMakerEndpointMode
 from sagemaker.serve.mode.local_container_mode import LocalContainerMode
-from sagemaker.serve.mode.in_process_mode import InProcessMode
 from sagemaker.serve.detector.pickler import save_pkl, save_xgboost
 from sagemaker.serve.builder.serve_settings import _ServeSettings
 from sagemaker.serve.builder.djl_builder import DJL
@@ -411,7 +410,7 @@ def _prepare_for_mode(
             )
             self.env_vars.update(env_vars_sagemaker)
             return self.s3_upload_path, env_vars_sagemaker
-        elif self.mode == Mode.LOCAL_CONTAINER:
+        if self.mode == Mode.LOCAL_CONTAINER:
             # init the LocalContainerMode object
             self.modes[str(Mode.LOCAL_CONTAINER)] = LocalContainerMode(
                 inference_spec=self.inference_spec,
@@ -423,22 +422,9 @@ def _prepare_for_mode(
             )
             self.modes[str(Mode.LOCAL_CONTAINER)].prepare()
             return None
-        elif self.mode == Mode.IN_PROCESS:
-            # init the InProcessMode object
-            self.modes[str(Mode.IN_PROCESS)] = InProcessMode(
-                inference_spec=self.inference_spec,
-                schema_builder=self.schema_builder,
-                session=self.sagemaker_session,
-                model_path=self.model_path,
-                env_vars=self.env_vars,
-                model_server=self.model_server,
-            )
-            self.modes[str(Mode.IN_PROCESS)].prepare()
-            return None
 
         raise ValueError(
-            "Please specify mode in: %s, %s, %s"
-            % (Mode.LOCAL_CONTAINER, Mode.SAGEMAKER_ENDPOINT, Mode.IN_PROCESS)
+            "Please specify mode in: %s, %s" % (Mode.LOCAL_CONTAINER, Mode.SAGEMAKER_ENDPOINT)
         )
 
     def _get_client_translators(self):
@@ -620,9 +606,6 @@ def _overwrite_mode_in_deploy(self, overwrite_mode: str):
         elif overwrite_mode == Mode.LOCAL_CONTAINER:
             self.mode = self.pysdk_model.mode = Mode.LOCAL_CONTAINER
             self._prepare_for_mode()
-        elif overwrite_mode == Mode.IN_PROCESS:
-            self.mode = self.pysdk_model.mode = Mode.IN_PROCESS
-            self._prepare_for_mode()
         else:
             raise ValueError("Mode %s is not supported!" % overwrite_mode)
 
@@ -812,10 +795,9 @@ def _initialize_for_mlflow(self, artifact_path: str) -> None:
         self.dependencies.update({"requirements": mlflow_model_dependency_path})
 
     # Model Builder is a class to build the model for deployment.
-    # It supports two* modes of deployment
+    # It supports two modes of deployment
     # 1/ SageMaker Endpoint
     # 2/ Local launch with container
-    # 3/ In process mode with Transformers server in beta release
     def build(  # pylint: disable=R0911
         self,
         mode: Type[Mode] = None,
@@ -913,10 +895,8 @@ def build(  # pylint: disable=R0911
 
     def _build_validations(self):
         """Validations needed for model server overrides, or auto-detection or fallback"""
-        if self.mode == Mode.IN_PROCESS and self.model_server is not ModelServer.MMS:
-            raise ValueError(
-                "IN_PROCESS mode is only supported for MMS/Transformers server in beta release."
-            )
+        if self.mode == Mode.IN_PROCESS:
+            raise ValueError("IN_PROCESS mode is not supported yet!")
 
         if self.inference_spec and self.model:
             raise ValueError("Can only set one of the following: model, inference_spec.")

src/sagemaker/serve/builder/requirements_manager.py

@@ -20,10 +20,10 @@
 
 
 class RequirementsManager:
-    """Transformers build logic with ModelBuilder()"""
+    """Manages dependency installation by detecting file types"""
 
     def detect_file_exists(self, dependencies: str = None) -> str:
-        """Creates snapshot of the user's environment
+        """ Detects the type of file dependencies will be installed from
 
         If a req.txt or conda.yml file is provided, it verifies their existence and
         returns the local file path
@@ -92,7 +92,3 @@ def _capture_from_local_runtime(self) -> str:
         local_dependencies_path = os.path.join(os.getcwd(), "inf_env_snapshot.yml")
 
         return local_dependencies_path
-
-
-if __name__ == "__main__":
-    RequirementsManager().detect_file_exists(dependencies="auto_capture")

src/sagemaker/serve/builder/transformers_builder.py

@@ -36,10 +36,7 @@
 )
 from sagemaker.serve.detector.pickler import save_pkl
 from sagemaker.serve.utils.optimize_utils import _is_optimized
-from sagemaker.serve.utils.predictors import (
-    TransformersLocalModePredictor,
-    TransformersInProcessModePredictor,
-)
+from sagemaker.serve.utils.predictors import TransformersLocalModePredictor
 from sagemaker.serve.utils.types import ModelServer
 from sagemaker.serve.mode.function_pointers import Mode
 from sagemaker.serve.utils.telemetry_logger import _capture_telemetry
@@ -50,7 +47,6 @@
 
 logger = logging.getLogger(__name__)
 DEFAULT_TIMEOUT = 1800
-LOCAL_MODES = [Mode.LOCAL_CONTAINER, Mode.IN_PROCESS]
 
 
 """Retrieves images for different libraries - Pytorch, TensorFlow from HuggingFace hub
@@ -234,22 +230,6 @@ def _transformers_model_builder_deploy_wrapper(self, *args, **kwargs) -> Type[Pr
             )
             return predictor
 
-        if self.mode == Mode.IN_PROCESS:
-            timeout = kwargs.get("model_data_download_timeout")
-
-            predictor = TransformersInProcessModePredictor(
-                self.modes[str(Mode.IN_PROCESS)], serializer, deserializer
-            )
-
-            self.modes[str(Mode.IN_PROCESS)].create_server(
-                self.image_uri,
-                timeout if timeout else DEFAULT_TIMEOUT,
-                None,
-                predictor,
-                self.pysdk_model.env,
-            )
-            return predictor
-
         if "mode" in kwargs:
             del kwargs["mode"]
         if "role" in kwargs:
@@ -297,7 +277,7 @@ def _build_transformers_env(self):
 
         self.pysdk_model = self._create_transformers_model()
 
-        if self.mode in LOCAL_MODES:
+        if self.mode == Mode.LOCAL_CONTAINER:
             self._prepare_for_mode()
 
         return self.pysdk_model

src/sagemaker/serve/mode/in_process_mode.py

@@ -20,23 +20,6 @@
 logger = logging.getLogger(__name__)
 
 
-class InProcessMultiModelServer:
-    """In Process Mode Multi Model server instance"""
-
-    def _start_serving(self):
-        """Initializes the start of the server"""
-        return Exception("Not implemented")
-
-    def _invoke_multi_model_server_serving(self, request: object, content_type: str, accept: str):
-        """Invokes the MMS server by sending POST request"""
-        return Exception("Not implemented")
-
-    def _multi_model_server_deep_ping(self, predictor: PredictorBase):
-        """Sends a deep ping to ensure prediction"""
-        response = None
-        return (True, response)
-
-
 class LocalMultiModelServer:
     """Local Multi Model server instance"""
 

src/sagemaker/serve/model_server/multi_model_server/server.py

@@ -24,16 +24,6 @@ def __init__(self, message):
         super().__init__(message=message)
 
 
-class InProcessDeepPingException(ModelBuilderException):
-    """Raise when in process model serving does not pass the deep ping check"""
-
-    fmt = "Error Message: {message}"
-    model_builder_error_code = 1
-
-    def __init__(self, message):
-        super().__init__(message=message)
-
-
 class LocalModelOutOfMemoryException(ModelBuilderException):
     """Raise when local model serving fails to load the model"""
 

src/sagemaker/serve/utils/exceptions.py

@@ -6,7 +6,6 @@
 
 from sagemaker import Session
 from sagemaker.serve.mode.local_container_mode import LocalContainerMode
-from sagemaker.serve.mode.in_process_mode import InProcessMode
 from sagemaker.serve.builder.schema_builder import SchemaBuilder
 from sagemaker.serializers import IdentitySerializer, JSONSerializer
 from sagemaker.deserializers import BytesDeserializer, JSONDeserializer
@@ -210,49 +209,6 @@ def delete_predictor(self):
         self._mode_obj.destroy_server()
 
 
-class TransformersInProcessModePredictor(PredictorBase):
-    """Lightweight Transformers predictor for local deployment"""
-
-    def __init__(
-        self,
-        mode_obj: Type[InProcessMode],
-        serializer=JSONSerializer(),
-        deserializer=JSONDeserializer(),
-    ):
-        self._mode_obj = mode_obj
-        self.serializer = serializer
-        self.deserializer = deserializer
-
-    def predict(self, data):
-        """Placeholder docstring"""
-        return [
-            self.deserializer.deserialize(
-                io.BytesIO(
-                    self._mode_obj._invoke_multi_model_server_serving(
-                        self.serializer.serialize(data),
-                        self.content_type,
-                        self.deserializer.ACCEPT[0],
-                    )
-                ),
-                self.content_type,
-            )
-        ]
-
-    @property
-    def content_type(self):
-        """The MIME type of the data sent to the inference endpoint."""
-        return self.serializer.CONTENT_TYPE
-
-    @property
-    def accept(self):
-        """The content type(s) that are expected from the inference endpoint."""
-        return self.deserializer.ACCEPT
-
-    def delete_predictor(self):
-        """Shut down and remove the container that you created in LOCAL_CONTAINER mode"""
-        self._mode_obj.destroy_server()
-
-
 class TeiLocalModePredictor(PredictorBase):
     """Lightweight Tei predictor for local deployment in IN_PROCESS and LOCAL_CONTAINER modes"""
 

src/sagemaker/serve/utils/predictors.py

@@ -66,11 +66,11 @@
 
 class TestModelBuilder(unittest.TestCase):
     @patch("sagemaker.serve.builder.model_builder._ServeSettings")
-    def test_validation_in_progress_mode_supported(self, mock_serveSettings):
-        builder = ModelBuilder(model_server=ModelServer.TORCHSERVE)
+    def test_validation_in_progress_mode_not_supported(self, mock_serveSettings):
+        builder = ModelBuilder()
         self.assertRaisesRegex(
             Exception,
-            "IN_PROCESS mode is only supported for MMS/Transformers server in beta release.",
+            "IN_PROCESS mode is not supported yet!",
             builder.build,
             Mode.IN_PROCESS,
             mock_role_arn,