Added additional training unit tests

components/aws/sagemaker/common/_utils.py

@@ -98,7 +98,7 @@ def create_training_job_request(args):
 
     request['TrainingJobName'] = job_name
     request['RoleArn'] = args['role']
-    request['HyperParameters'] = args['hyperparameters']
+    request['HyperParameters'] = create_hyperparameters(args['hyperparameters'])
     request['AlgorithmSpecification']['TrainingInputMode'] = args['training_input_mode']
 
     ### Update training image (for BYOC and built-in algorithms) or algorithm resource name
@@ -136,8 +136,8 @@ def create_training_job_request(args):
 
     ### Update or pop VPC configs
     if args['vpc_security_group_ids'] and args['vpc_subnets']:
-        request['VpcConfig']['SecurityGroupIds'] = [args['vpc_security_group_ids']]
-        request['VpcConfig']['Subnets'] = [args['vpc_subnets']]
+        request['VpcConfig']['SecurityGroupIds'] = args['vpc_security_group_ids'].split(',')
+        request['VpcConfig']['Subnets'] = args['vpc_subnets'].split(',')
     else:
         request.pop('VpcConfig')
 
@@ -272,8 +272,8 @@ def create_model_request(args):
 
     ### Update or pop VPC configs
     if args['vpc_security_group_ids'] and args['vpc_subnets']:
-        request['VpcConfig']['SecurityGroupIds'] = [args['vpc_security_group_ids']]
-        request['VpcConfig']['Subnets'] = [args['vpc_subnets']]
+        request['VpcConfig']['SecurityGroupIds'] = args['vpc_security_group_ids'].split(',')
+        request['VpcConfig']['Subnets'] = args['vpc_subnets'].split(',')
     else:
         request.pop('VpcConfig')
 
@@ -493,7 +493,7 @@ def create_hyperparameter_tuning_job_request(args):
     request['HyperParameterTuningJobConfig']['ParameterRanges']['CategoricalParameterRanges'] = args['categorical_parameters']
     request['HyperParameterTuningJobConfig']['TrainingJobEarlyStoppingType'] = args['early_stopping_type']
 
-    request['TrainingJobDefinition']['StaticHyperParameters'] = args['static_parameters']
+    request['TrainingJobDefinition']['StaticHyperParameters'] = create_hyperparameters(args['static_parameters'])
     request['TrainingJobDefinition']['AlgorithmSpecification']['TrainingInputMode'] = args['training_input_mode']
 
     ### Update training image (for BYOC) or algorithm resource name
@@ -531,8 +531,8 @@ def create_hyperparameter_tuning_job_request(args):
 
     ### Update or pop VPC configs
     if args['vpc_security_group_ids'] and args['vpc_subnets']:
-        request['TrainingJobDefinition']['VpcConfig']['SecurityGroupIds'] = [args['vpc_security_group_ids']]
-        request['TrainingJobDefinition']['VpcConfig']['Subnets'] = [args['vpc_subnets']]
+        request['TrainingJobDefinition']['VpcConfig']['SecurityGroupIds'] = args['vpc_security_group_ids'].split(',')
+        request['TrainingJobDefinition']['VpcConfig']['Subnets'] = args['vpc_subnets'].split(',')
     else:
         request['TrainingJobDefinition'].pop('VpcConfig')
 
@@ -835,6 +835,14 @@ def get_labeling_job_outputs(client, labeling_job_name, auto_labeling):
         active_learning_model_arn = ' '
     return output_manifest, active_learning_model_arn
 
+def create_hyperparameters(hyperparam_args):
+    # Validate all values are strings
+    for key, value in hyperparam_args.items():
+        if not isinstance(value, str):
+            raise Exception(f"Could not parse hyperparameters. Value for {key} was not a string.")
+
+    return hyperparam_args
+
 def enable_spot_instance_support(training_job_config, args):
     if args['spot_instance']:
         training_job_config['EnableManagedSpotTraining'] = args['spot_instance']

components/aws/sagemaker/hyperparameter_tuning/README.md

@@ -35,8 +35,8 @@ max_num_jobs | The maximum number of training jobs that a hyperparameter tuning
 max_parallel_jobs | The maximum number of concurrent training jobs that a hyperparameter tuning job can launch | No | No | Int | [1, 10] | |
 max_run_time | The maximum run time in seconds per training job | Yes | Yes | Int | ≤ 432000 (5 days) | 86400 (1 day) |
 resource_encryption_key | The AWS KMS key that Amazon SageMaker uses to encrypt data on the storage volume attached to the ML compute instance(s) | Yes | Yes | String | | |
-vpc_security_group_ids | The VPC security group IDs, in the form sg-xxxxxxxx | Yes | Yes | String | | |
-vpc_subnets | The ID of the subnets in the VPC to which you want to connect your hpo job | Yes | Yes | String | | |
+vpc_security_group_ids | A comma-delimited list of security group IDs, in the form sg-xxxxxxxx | Yes | Yes | String | | |
+vpc_subnets | A comma-delimited list of subnet IDs in the VPC to which you want to connect your hpo job | Yes | Yes | String | | |
 network_isolation | Isolates the training container if true | Yes | No | Boolean | False, True | True |
 traffic_encryption | Encrypts all communications between ML compute instances in distributed training if true | Yes | No | Boolean | False, True | False |
 spot_instance | Use managed spot training if true | Yes | No | Boolean | False, True | False |

components/aws/sagemaker/tests/unit_tests/tests/test_train.py

@@ -27,6 +27,86 @@ def setUpClass(cls):
     parser = train.create_parser()
     cls.parser = parser
 
+  def test_create_parser(self):
+    self.assertIsNotNone(self.parser)
+
+  def test_reasonable_required_args(self):
+    response = _utils.create_training_job_request(vars(self.parser.parse_args(required_args)))
+
+    # Ensure all of the optional arguments have reasonable default values
+    self.assertFalse(response['EnableManagedSpotTraining'])
+    self.assertDictEqual(response['HyperParameters'], {})
+    self.assertNotIn('VpcConfig', response)
+    self.assertNotIn('MetricDefinitions', response)
+    self.assertEqual(response['Tags'], [])
+    self.assertEqual(response['AlgorithmSpecification']['TrainingInputMode'], 'File')
+    self.assertEqual(response['OutputDataConfig']['S3OutputPath'], 'test-path')
+
+  def test_metric_definitions(self):
+    metric_definition_args = self.parser.parse_args(required_args + ['--metric_definitions', '{"metric1": "regexval1", "metric2": "regexval2"}'])
+    response = _utils.create_training_job_request(vars(metric_definition_args))
+
+    self.assertIn('MetricDefinitions', response['AlgorithmSpecification'])
+    response_metric_definitions = response['AlgorithmSpecification']['MetricDefinitions']
+
+    self.assertEqual(response_metric_definitions, [{
+      'Name': "metric1",
+      'Regex': "regexval1"
+    }, {
+      'Name': "metric2",
+      'Regex': "regexval2"
+    }])
+
+  def test_invalid_instance_type(self):
+    invalid_instance_args = required_args + ['--instance_type', 'invalid-instance']
+
+    with self.assertRaises(SystemExit):
+      self.parser.parse_args(invalid_instance_args)
+
+  def test_valid_hyperparameters(self):
+    hyperparameters_str = '{"hp1": "val1", "hp2": "val2", "hp3": "val3"}'
+
+    good_args = self.parser.parse_args(required_args + ['--hyperparameters', hyperparameters_str])
+    response = _utils.create_training_job_request(vars(good_args))
+
+    self.assertIn('hp1', response['HyperParameters'])
+    self.assertIn('hp2', response['HyperParameters'])
+    self.assertIn('hp3', response['HyperParameters'])
+    self.assertEqual(response['HyperParameters']['hp1'], "val1")
+    self.assertEqual(response['HyperParameters']['hp2'], "val2")
+    self.assertEqual(response['HyperParameters']['hp3'], "val3")
+
+  def test_empty_hyperparameters(self):
+    hyperparameters_str = '{}'
+
+    good_args = self.parser.parse_args(required_args + ['--hyperparameters', hyperparameters_str])
+    response = _utils.create_training_job_request(vars(good_args))
+
+    self.assertEqual(response['HyperParameters'], {})
+
+  def test_object_hyperparameters(self):
+    hyperparameters_str = '{"hp1": {"innerkey": "innerval"}}'
+
+    invalid_args = self.parser.parse_args(required_args + ['--hyperparameters', hyperparameters_str])
+    with self.assertRaises(Exception):
+      _utils.create_training_job_request(vars(invalid_args))
+
+  def test_vpc_configuration(self):
+    required_vpc_args = self.parser.parse_args(required_args + ['--vpc_security_group_ids', 'sg1,sg2', '--vpc_subnets', 'subnet1,subnet2'])
+    response = _utils.create_training_job_request(vars(required_vpc_args))
+
+    self.assertIn('VpcConfig', response)
+    self.assertIn('sg1', response['VpcConfig']['SecurityGroupIds'])
+    self.assertIn('sg2', response['VpcConfig']['SecurityGroupIds'])
+    self.assertIn('subnet1', response['VpcConfig']['Subnets'])
+    self.assertIn('subnet2', response['VpcConfig']['Subnets'])
+
+  def test_training_mode(self):
+    required_vpc_args = self.parser.parse_args(required_args + ['--training_input_mode', 'Pipe'])
+    response = _utils.create_training_job_request(vars(required_vpc_args))
+
+    self.assertEqual(response['AlgorithmSpecification']['TrainingInputMode'], 'Pipe')
+
   def test_spot_bad_args(self):
     no_max_wait_args = self.parser.parse_args(required_args + ['--spot_instance', 'True'])
     no_checkpoint_args = self.parser.parse_args(required_args + ['--spot_instance', 'True', '--max_wait_time', '3600'])

components/aws/sagemaker/train/README.md

@@ -28,8 +28,8 @@ resource_encryption_key | The AWS KMS key that Amazon SageMaker uses to encrypt
 max_run_time | The maximum run time in seconds per training job | Yes | Int | ≤ 432000 (5 days) | 86400 (1 day) |
 model_artifact_path | | No | String | | |
 output_encryption_key | The AWS KMS key that Amazon SageMaker uses to encrypt the model artifacts | Yes | String | | |
-vpc_security_group_ids | The VPC security group IDs, in the form sg-xxxxxxxx | Yes | String | | |
-vpc_subnets | The ID of the subnets in the VPC to which you want to connect your hpo job | Yes | String | | |
+vpc_security_group_ids | A comma-delimited list of security group IDs, in the form sg-xxxxxxxx | Yes | String | | |
+vpc_subnets | A comma-delimited list of subnet IDs in the VPC to which you want to connect your hpo job | Yes | String | | |
 network_isolation | Isolates the training container if true | No | Boolean | False, True | True |
 traffic_encryption | Encrypts all communications between ML compute instances in distributed training if true | No | Boolean | False, True | False |
 spot_instance | Use managed spot training if true | No | Boolean | False, True | False |