Support for cuML framework classes

[vruusmann]

When trying to convert the model alone without preprocessing, the following error appears:

Aug 16, 2022 12:48:12 PM org.jpmml.sklearn.example.Main run
INFO: Parsing PKL..
Aug 16, 2022 12:48:12 PM org.jpmml.sklearn.example.Main run
SEVERE: Failed to parse PKL
net.razorvine.pickle.PickleException: failed to setstate()
at net.razorvine.pickle.Unpickler.load_build(Unpickler.java:395)
at net.razorvine.pickle.Unpickler.dispatch(Unpickler.java:220)
at org.jpmml.python.CustomUnpickler.dispatch(CustomUnpickler.java:31)
at org.jpmml.python.PickleUtil$1.dispatch(PickleUtil.java:64)
at net.razorvine.pickle.Unpickler.load(Unpickler.java:109)
at org.jpmml.python.PickleUtil.unpickle(PickleUtil.java:85)
at org.jpmml.sklearn.example.Main.run(Main.java:163)
at org.jpmml.sklearn.example.Main.main(Main.java:151)
Caused by: java.lang.NoSuchMethodException: net.razorvine.pickle.objects.ClassDict.setstate(java.lang.Integer)
at java.base/java.lang.Class.getMethod(Class.java:2108)
at net.razorvine.pickle.Unpickler.load_build(Unpickler.java:392)
... 7 more

Exception in thread "main" net.razorvine.pickle.PickleException: failed to setstate()
at net.razorvine.pickle.Unpickler.load_build(Unpickler.java:395)
at net.razorvine.pickle.Unpickler.dispatch(Unpickler.java:220)
at org.jpmml.python.CustomUnpickler.dispatch(CustomUnpickler.java:31)
at org.jpmml.python.PickleUtil$1.dispatch(PickleUtil.java:64)
at net.razorvine.pickle.Unpickler.load(Unpickler.java:109)
at org.jpmml.python.PickleUtil.unpickle(PickleUtil.java:85)
at org.jpmml.sklearn.example.Main.run(Main.java:163)
at org.jpmml.sklearn.example.Main.main(Main.java:151)
Caused by: java.lang.NoSuchMethodException: net.razorvine.pickle.objects.ClassDict.setstate(java.lang.Integer)
at java.base/java.lang.Class.getMethod(Class.java:2108)
at net.razorvine.pickle.Unpickler.load_build(Unpickler.java:392)
... 7 more

The model is a random forest but trained with GPU, so it is a cuml object:

RandomForestClassifier()
<class 'cuml.ensemble.randomforestclassifier.RandomForestClassifier'>

Converting the model alone without GPU as a sklearn object works without problem. 🤔



When converting model with preprocessing pipeline I got below exception:

Aug 16, 2022 12:48:32 PM org.jpmml.sklearn.example.Main run
INFO: Parsing PKL..
Aug 16, 2022 12:48:32 PM org.jpmml.sklearn.example.Main run
SEVERE: Failed to parse PKL
net.razorvine.pickle.PickleException: failed to setstate()
at net.razorvine.pickle.Unpickler.load_build(Unpickler.java:395)
at net.razorvine.pickle.Unpickler.dispatch(Unpickler.java:220)
at org.jpmml.python.CustomUnpickler.dispatch(CustomUnpickler.java:31)
at org.jpmml.python.PickleUtil$1.dispatch(PickleUtil.java:64)
at net.razorvine.pickle.Unpickler.load(Unpickler.java:109)
at org.jpmml.python.PickleUtil.unpickle(PickleUtil.java:85)
at org.jpmml.sklearn.example.Main.run(Main.java:163)
at org.jpmml.sklearn.example.Main.main(Main.java:151)
Caused by: java.lang.reflect.InvocationTargetException
at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.base/java.lang.reflect.Method.invoke(Method.java:566)
at net.razorvine.pickle.Unpickler.load_build(Unpickler.java:393)
... 7 more
Caused by: net.razorvine.pickle.PickleException: Expected 8 attribute(s), got 9 attribute(s)
at org.jpmml.python.CustomPythonObject.createAttributeMap(CustomPythonObject.java:81)
at numpy.DType.setstate(DType.java:50)
... 12 more

Exception in thread "main" net.razorvine.pickle.PickleException: failed to setstate()
at net.razorvine.pickle.Unpickler.load_build(Unpickler.java:395)
at net.razorvine.pickle.Unpickler.dispatch(Unpickler.java:220)
at org.jpmml.python.CustomUnpickler.dispatch(CustomUnpickler.java:31)
at org.jpmml.python.PickleUtil$1.dispatch(PickleUtil.java:64)
at net.razorvine.pickle.Unpickler.load(Unpickler.java:109)
at org.jpmml.python.PickleUtil.unpickle(PickleUtil.java:85)
at org.jpmml.sklearn.example.Main.run(Main.java:163)
at org.jpmml.sklearn.example.Main.main(Main.java:151)
Caused by: java.lang.reflect.InvocationTargetException
at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.base/java.lang.reflect.Method.invoke(Method.java:566)
at net.razorvine.pickle.Unpickler.load_build(Unpickler.java:393)
... 7 more
Caused by: net.razorvine.pickle.PickleException: Expected 8 attribute(s), got 9 attribute(s)
at org.jpmml.python.CustomPythonObject.createAttributeMap(CustomPythonObject.java:81)
at numpy.DType.setstate(DType.java:50)
... 12 more

Originally posted by @szymoonl in jpmml/jpmml-sklearn#175 (comment)

[vruusmann]

cuML is completelt separate from Scikit-Learn, therefore moving this issue to a new JPMML-cuML repository.

Might work on this, if the new repository gains at least some amount of attention/support.