Synthetics config is optional, with strategy-specific defaults

GitOrigin-RevId: 5f8d152ce1ce36748e4fccb1ee5d0ad75ba0b912

src/gretel_trainer/relational/multi_table.py

@@ -734,7 +734,7 @@ def _train_synthetics_models(self, configs: dict[str, dict[str, Any]]) -> None:
     def train_synthetics(
         self,
         *,
-        config: GretelModelConfig,
+        config: Optional[GretelModelConfig] = None,
         table_specific_configs: Optional[dict[str, GretelModelConfig]] = None,
         only: Optional[set[str]] = None,
         ignore: Optional[set[str]] = None,
@@ -743,11 +743,8 @@ def train_synthetics(
         Train synthetic data models for the tables in the tableset,
         optionally scoped by either `only` or `ignore`.
         """
-        # This argument used to be hinted as Optional with a None default and would
-        # fall back to a removed attribute on the MultiTable instance. Out of an
-        # abundance of caution, verify the client isn't relying on that old behavior.
         if config is None:
-            raise MultiTableException(f"Must provide a default model config.")
+            config = self._strategy.default_config
 
         configs = assemble_configs(
             self.relational_data, config, table_specific_configs, only, ignore

src/gretel_trainer/relational/strategies/ancestral.py

@@ -9,7 +9,11 @@
 import gretel_trainer.relational.strategies.common as common
 
 from gretel_client.projects.models import Model
-from gretel_trainer.relational.core import MultiTableException, RelationalData
+from gretel_trainer.relational.core import (
+    GretelModelConfig,
+    MultiTableException,
+    RelationalData,
+)
 from gretel_trainer.relational.sdk_extras import ExtendedGretelSDK
 from gretel_trainer.relational.table_evaluation import TableEvaluation
 
@@ -25,6 +29,10 @@ def name(self) -> str:
     def supported_model_keys(self) -> list[str]:
         return ["amplify"]
 
+    @property
+    def default_config(self) -> GretelModelConfig:
+        return "synthetics/amplify"
+
     def label_encode_keys(
         self, rel_data: RelationalData, tables: dict[str, pd.DataFrame]
     ) -> dict[str, pd.DataFrame]:

src/gretel_trainer/relational/strategies/independent.py

@@ -10,7 +10,7 @@
 import gretel_trainer.relational.strategies.common as common
 
 from gretel_client.projects.models import Model
-from gretel_trainer.relational.core import RelationalData
+from gretel_trainer.relational.core import GretelModelConfig, RelationalData
 from gretel_trainer.relational.sdk_extras import ExtendedGretelSDK
 from gretel_trainer.relational.table_evaluation import TableEvaluation
 
@@ -26,6 +26,10 @@ def name(self) -> str:
     def supported_model_keys(self) -> list[str]:
         return ["amplify", "actgan", "synthetics", "tabular_dp"]
 
+    @property
+    def default_config(self) -> GretelModelConfig:
+        return "synthetics/tabular-actgan"
+
     def label_encode_keys(
         self, rel_data: RelationalData, tables: dict[str, pd.DataFrame]
     ) -> dict[str, pd.DataFrame]:

tests/relational/test_train_synthetics.py

@@ -29,6 +29,30 @@ def tmpdir(project):
         yield tmpdir
 
 
+class ModelConfigMatcher:
+    def __init__(self, model_key: str):
+        self.model_key = model_key
+
+    def __eq__(self, other):
+        return list(other["models"][0])[0] == self.model_key
+
+
+def test_train_synthetics_strategy_specific_default_configs(pets, tmpdir, project):
+    mt = MultiTable(pets, strategy="independent", project_display_name=tmpdir)
+    mt.train_synthetics()
+    project.create_model_obj.assert_called_with(
+        model_config=ModelConfigMatcher("actgan"),
+        data_source=f"{tmpdir}/synthetics_train_pets.csv",
+    )
+
+    mt = MultiTable(pets, strategy="ancestral", project_display_name=tmpdir)
+    mt.train_synthetics()
+    project.create_model_obj.assert_called_with(
+        model_config=ModelConfigMatcher("amplify"),
+        data_source=f"{tmpdir}/synthetics_train_pets.csv",
+    )
+
+
 def test_train_synthetics_defaults_to_training_all_tables(ecom, tmpdir):
     mt = MultiTable(ecom, project_display_name=tmpdir)
     mt.train_synthetics(config="synthetics/amplify")