WeichenXu123 commented on code in PR #50199:
URL: https://github.com/apache/spark/pull/50199#discussion_r1990366864


##########
python/pyspark/ml/util.py:
##########
@@ -185,29 +185,40 @@ def wrapped(self: "JavaWrapper", dataset: 
"ConnectDataFrame") -> Any:
 
                 assert isinstance(self._java_obj, str)
                 params = serialize_ml_params(self, session.client)
-                return ConnectDataFrame(
-                    TransformerRelation(
-                        child=dataset._plan, name=self._java_obj, 
ml_params=params, is_model=True
-                    ),
-                    session,
+                plan = TransformerRelation(
+                    child=dataset._plan,
+                    name=self._java_obj,
+                    ml_params=params,
+                    is_model=True,
                 )
             elif isinstance(self, Transformer):
                 from pyspark.ml.connect.proto import TransformerRelation
 
                 assert isinstance(self._java_obj, str)
                 params = serialize_ml_params(self, session.client)
-                return ConnectDataFrame(
-                    TransformerRelation(
-                        child=dataset._plan,
-                        name=self._java_obj,
-                        ml_params=params,
-                        uid=self.uid,
-                        is_model=False,
-                    ),
-                    session,
+                plan = TransformerRelation(
+                    child=dataset._plan,
+                    name=self._java_obj,
+                    ml_params=params,
+                    uid=self.uid,
+                    is_model=False,
                 )
+
             else:
                 raise RuntimeError(f"Unsupported {self}")
+
+            # To delay the GC of the model, keep a reference to the source 
transformer
+            # in the transformed dataframe and all its descendants.
+            # For this case:
+            #
+            # def fit_transform(df):
+            #     model = estimator.fit(df)
+            #     return model.transform(df)
+            #
+            # output = fit_transform(df)
+            #
+            plan.__source_transformer__ = self  # type: ignore[attr-defined]

Review Comment:
   For model summary, it might also return Dataframe, shall we make similar 
change ?



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org

Reply via email to