fix: the Hyperband support fix for the HPO (aws#3516)

Co-authored-by: Anton Repushko <[email protected]>

Author: 2 people

src/sagemaker/session.py

@@ -2121,6 +2121,7 @@ def tune(  # noqa: C901
         stop_condition,
         tags,
         warm_start_config,
+        strategy_config=None,
         enable_network_isolation=False,
         image_uri=None,
         algorithm_arn=None,
@@ -2136,6 +2137,8 @@ def tune(  # noqa: C901
         Args:
             job_name (str): Name of the tuning job being created.
             strategy (str): Strategy to be used for hyperparameter estimations.
+            strategy_config (dict): A configuration for the hyperparameter tuning
+                job optimisation strategy.
             objective_type (str): The type of the objective metric for evaluating training jobs.
                 This value can be either 'Minimize' or 'Maximize'.
             objective_metric_name (str): Name of the metric for evaluating training jobs.
@@ -2220,6 +2223,7 @@ def tune(  # noqa: C901
                 objective_metric_name=objective_metric_name,
                 parameter_ranges=parameter_ranges,
                 early_stopping_type=early_stopping_type,
+                strategy_config=strategy_config,
             ),
             "TrainingJobDefinition": self._map_training_config(
                 static_hyperparameters=static_hyperparameters,
@@ -2375,6 +2379,7 @@ def _map_tuning_config(
         objective_type=None,
         objective_metric_name=None,
         parameter_ranges=None,
+        strategy_config=None,
     ):
         """Construct tuning job configuration dictionary.
 
@@ -2392,6 +2397,8 @@ def _map_tuning_config(
             objective_metric_name (str): Name of the metric for evaluating training jobs.
             parameter_ranges (dict): Dictionary of parameter ranges. These parameter ranges can
                 be one of three types: Continuous, Integer, or Categorical.
+            strategy_config (dict): A configuration for the hyperparameter tuning job optimisation
+                strategy.
 
         Returns:
             A dictionary of tuning job configuration. For format details, please refer to
@@ -2415,6 +2422,8 @@ def _map_tuning_config(
         if parameter_ranges is not None:
             tuning_config["ParameterRanges"] = parameter_ranges
 
+        if strategy_config is not None:
+            tuning_config["StrategyConfig"] = strategy_config
         return tuning_config
 
     @classmethod

src/sagemaker/tuner.py

@@ -282,8 +282,8 @@ def from_job_desc(cls, hyperband_strategy_config):
 
         Returns:
             sagemaker.tuner.HyperbandStrategyConfig: De-serialized instance of
-            HyperbandStrategyConfig containing the max_resource and min_resource provided as part of
-            ``hyperband_strategy_config``.
+                ``HyperbandStrategyConfig`` containing the max_resource
+                and min_resource provided as part of ``hyperband_strategy_config``.
         """
         return cls(
             min_resource=hyperband_strategy_config[HYPERBAND_MIN_RESOURCE],
@@ -306,7 +306,7 @@ def to_input_req(self):
 
         Returns:
             dict: Containing the "MaxResource" and
-            "MinResource" as the first class fields.
+                "MinResource" as the first class fields.
         """
         return {
             HYPERBAND_MIN_RESOURCE: self.min_resource,
@@ -330,7 +330,7 @@ def __init__(
 
         Args:
             hyperband_strategy_config (sagemaker.tuner.HyperbandStrategyConfig): The configuration
-            for the object that specifies the Hyperband strategy.
+                for the object that specifies the Hyperband strategy.
                 This parameter is only supported for the Hyperband selection for Strategy within
                 the HyperParameterTuningJobConfig.
         """
@@ -461,7 +461,7 @@ def __init__(
                 ``WarmStartConfig`` object that has been initialized with the
                 configuration defining the nature of warm start tuning job.
             strategy_config (sagemaker.tuner.StrategyConfig): A configuration for "Hyperparameter"
-            tuning job optimisation strategy.
+                tuning job optimisation strategy.
             early_stopping_type (str or PipelineVariable): Specifies whether early stopping is
                 enabled for the job. Can be either 'Auto' or 'Off' (default:
                 'Off'). If set to 'Off', early stopping will not be attempted.
@@ -1569,7 +1569,7 @@ def create(
             strategy (str): Strategy to be used for hyperparameter estimations
                 (default: 'Bayesian').
             strategy_config (dict): The configuration for a training job launched by a
-            hyperparameter tuning job.
+                hyperparameter tuning job.
             objective_type (str): The type of the objective metric for evaluating training jobs.
                 This value can be either 'Minimize' or 'Maximize' (default: 'Maximize').
             max_jobs (int): Maximum total number of training jobs to start for the hyperparameter
@@ -1776,7 +1776,7 @@ def _get_tuner_args(cls, tuner, inputs):
         }
 
         if tuner.strategy_config is not None:
-            tuning_config["strategy_config"] = tuner.strategy_config
+            tuning_config["strategy_config"] = tuner.strategy_config.to_input_req()
 
         if tuner.objective_metric_name is not None:
             tuning_config["objective_type"] = tuner.objective_type

tests/unit/test_session.py

@@ -941,6 +941,13 @@ def test_train_pack_to_request(sagemaker_session):
     ],
 }
 
+SAMPLE_HYPERBAND_STRATEGY_CONFIG = {
+    "HyperbandStrategyConfig": {
+        "MinResource": 1,
+        "MaxResource": 10,
+    }
+}
+
 
 @pytest.mark.parametrize(
     "warm_start_type, parents",
@@ -1167,6 +1174,47 @@ def assert_create_tuning_job_request(**kwrags):
     )
 
 
+def test_tune_with_strategy_config(sagemaker_session):
+    def assert_create_tuning_job_request(**kwrags):
+        assert (
+            kwrags["HyperParameterTuningJobConfig"]["StrategyConfig"]["HyperbandStrategyConfig"][
+                "MinResource"
+            ]
+            == SAMPLE_HYPERBAND_STRATEGY_CONFIG["HyperbandStrategyConfig"]["MinResource"]
+        )
+        assert (
+            kwrags["HyperParameterTuningJobConfig"]["StrategyConfig"]["HyperbandStrategyConfig"][
+                "MaxResource"
+            ]
+            == SAMPLE_HYPERBAND_STRATEGY_CONFIG["HyperbandStrategyConfig"]["MaxResource"]
+        )
+
+    sagemaker_session.sagemaker_client.create_hyper_parameter_tuning_job.side_effect = (
+        assert_create_tuning_job_request
+    )
+    sagemaker_session.tune(
+        job_name="dummy-tuning-1",
+        strategy="Bayesian",
+        objective_type="Maximize",
+        objective_metric_name="val-score",
+        max_jobs=100,
+        max_parallel_jobs=5,
+        parameter_ranges=SAMPLE_PARAM_RANGES,
+        static_hyperparameters=STATIC_HPs,
+        image_uri="dummy-image-1",
+        input_mode="File",
+        metric_definitions=SAMPLE_METRIC_DEF,
+        role=EXPANDED_ROLE,
+        input_config=SAMPLE_INPUT,
+        output_config=SAMPLE_OUTPUT,
+        resource_config=RESOURCE_CONFIG,
+        stop_condition=SAMPLE_STOPPING_CONDITION,
+        tags=None,
+        warm_start_config=None,
+        strategy_config=SAMPLE_HYPERBAND_STRATEGY_CONFIG,
+    )
+
+
 def test_tune_with_encryption_flag(sagemaker_session):
     def assert_create_tuning_job_request(**kwrags):
         assert (