start

Author: Bryannah Hernandez

src/sagemaker/app.py

@@ -35,7 +35,7 @@ async def generate_text(prompt: Request):
     return generated_text[0]["generated_text"]
 
 
-generator = pipeline("text-generation", model="gpt2")
+generator = pipeline('text-generation', model='gpt2')
 
 
 @app.post("/post")

src/sagemaker/pipeline.py

@@ -16,6 +16,7 @@
 from typing import Optional, Dict, List, Union
 
 import sagemaker
+import logging
 from sagemaker import ModelMetrics, Model
 from sagemaker.config import (
     ENDPOINT_CONFIG_KMS_KEY_ID_PATH,
@@ -42,6 +43,7 @@
 from sagemaker.workflow.pipeline_context import runnable_by_pipeline
 from sagemaker.utils import instance_supports_kms
 
+logger = logging.getLogger(__name__)
 
 class PipelineModel(object):
     """A pipeline of SageMaker `Model` instances.
@@ -297,6 +299,7 @@ def deploy(
             )
 
         if self.predictor_cls:
+            logger.info("DO I come here")
             predictor = self.predictor_cls(self.endpoint_name, self.sagemaker_session)
             if serializer:
                 predictor.serializer = serializer

src/sagemaker/serve/mode/in_process_mode.py

@@ -73,6 +73,7 @@ def create_server(
         logger.info("Waiting for model server %s to start up...", self.model_server)
 
         if self.model_server == ModelServer.MMS:
+            self._start_serving()
             self._ping_container = self._multi_model_server_deep_ping
 
         time_limit = datetime.now() + timedelta(seconds=5)