Local mode support for Generic Estimators

src/sagemaker/estimator.py

@@ -379,7 +379,7 @@ class Estimator(EstimatorBase):
     def __init__(self, image_name, role, train_instance_count, train_instance_type,
                  train_volume_size=30, train_max_run=24 * 60 * 60, input_mode='File',
                  output_path=None, output_kms_key=None, base_job_name=None, sagemaker_session=None,
-                 hyperparameters=None):
+                 hyperparameters=None, local_mode=False):
         """Initialize an ``Estimator`` instance.
 
         Args:
@@ -414,10 +414,11 @@ def __init__(self, image_name, role, train_instance_count, train_instance_type,
             hyperparameters (dict): Dictionary containing the hyperparameters to initialize this estimator with.
         """
         self.image_name = image_name
-        self.hyperparam_dict = hyperparameters.copy() if hyperparameters else {}
+        self._hyperparameters = hyperparameters or {}
+
         super(Estimator, self).__init__(role, train_instance_count, train_instance_type,
                                         train_volume_size, train_max_run, input_mode,
-                                        output_path, output_kms_key, base_job_name, sagemaker_session)
+                                        output_path, output_kms_key, base_job_name, sagemaker_session, local_mode)
 
     def train_image(self):
         """
@@ -428,16 +429,16 @@ def train_image(self):
         """
         return self.image_name
 
-    def set_hyperparameters(self, **kwargs):
-        for k, v in kwargs.items():
-            self.hyperparam_dict[k] = v
-
     def hyperparameters(self):
-        """Returns the hyperparameters as a dictionary to use for training.
+        """Return the hyperparameters as a dictionary to use for training.
+
+               The fit() method, that does the model training, calls this method to find the hyperparameters you specified.
+
 
-       The fit() method, that does the model training, calls this method to find the hyperparameters you specified.
+        Returns:
+            dict[str, str]: The hyperparameters.
         """
-        return self.hyperparam_dict
+        return _json_encode_hyperparameters(self._hyperparameters)
 
     def create_model(self, image=None, predictor_cls=None, serializer=None, deserializer=None,
                      content_type=None, accept=None, **kwargs):
@@ -589,7 +590,7 @@ def hyperparameters(self):
         Returns:
             dict[str, str]: The hyperparameters.
         """
-        return self._json_encode_hyperparameters(self._hyperparameters)
+        return _json_encode_hyperparameters(self._hyperparameters)
 
     @classmethod
     def _prepare_init_params_from_job_description(cls, job_details):
@@ -647,10 +648,6 @@ def attach(cls, training_job_name, sagemaker_session=None):
         estimator.uploaded_code = UploadedCode(estimator.source_dir, estimator.entry_point)
         return estimator
 
-    @staticmethod
-    def _json_encode_hyperparameters(hyperparameters):
-        return {str(k): json.dumps(v) for (k, v) in hyperparameters.items()}
-
     @classmethod
     def _update_init_params(cls, hp, tf_arguments):
         updated_params = {}
@@ -660,3 +657,7 @@ def _update_init_params(cls, hp, tf_arguments):
                 value = json.loads(value)
                 updated_params[argument] = value
         return updated_params
+
+
+def _json_encode_hyperparameters(hyperparameters):
+    return {str(k): json.dumps(v) for (k, v) in hyperparameters.items()}

src/sagemaker/image.py

@@ -204,7 +204,11 @@ def _download_folder(bucket_name, prefix, target, session=None):
         except:
             pass
 
-        obj.download_file(file_path)
+        try:
+            obj.download_file(file_path)
+        except OSError as e:
+            if e.errno == 21: # it is a directory
+                pass
 
 
 def _compose(tmpdir, instance_type, detached=False):

src/sagemaker/local_session.py

@@ -31,7 +31,7 @@ def __init__(self, boto_session=None):
         self.created_endpoint = False
 
     def create_training_job(self, TrainingJobName, AlgorithmSpecification, RoleArn, InputDataConfig, OutputDataConfig,
-                            ResourceConfig, StoppingCondition, HyperParameters, Tags=None):
+                            ResourceConfig, StoppingCondition, HyperParameters={}, Tags=None):
         self.train_instance_count = ResourceConfig['InstanceCount']
 
         self.s3_model_artifacts = train(AlgorithmSpecification, InputDataConfig, ResourceConfig, HyperParameters,
@@ -94,7 +94,6 @@ def create_endpoint(self, EndpointName, EndpointConfigName):
             print("Container is up")
             return
 
-
     def delete_endpoint(self, EndpointName):
         self.container.down()
 

src/sagemaker/session.py

@@ -223,7 +223,7 @@ def train(self, image, input_mode, input_config, role, job_name, output_config,
                 'TrainingImage': image,
                 'TrainingInputMode': input_mode
             },
-            # 'HyperParameters': hyperparameters,
+            'HyperParameters': hyperparameters or {},
             'InputDataConfig': input_config,
             'OutputDataConfig': output_config,
             'TrainingJobName': job_name,
@@ -232,8 +232,6 @@ def train(self, image, input_mode, input_config, role, job_name, output_config,
             "RoleArn": role,
         }
 
-        if hyperparameters and len(hyperparameters) > 0:
-            train_request['HyperParameters'] = hyperparameters
         LOGGER.info('Creating training-job with name: {}'.format(job_name))
         LOGGER.debug('train request: {}'.format(json.dumps(train_request, indent=4)))
         self.sagemaker_client.create_training_job(**train_request)