get_object_importance fails for MultiClass models

Problem: In python, a call to get_object_importance(pool=validation_pool, train_pool = train_pool) fails throwing the error:

File "C:\Users\mmmm777\AppData\Local\Continuum\anaconda3\envs\deeplearning_uptodate\lib\site-packages\catboost\core.py", line 2006, in get_object_importance
result = self._calc_ostr(train_pool, pool, top_size, type, update_method, importance_values_sign, thread_count, verbose)

File "C:\Users\mmmm777\AppData\Local\Continuum\anaconda3\envs\deeplearning_uptodate\lib\site-packages\catboost\core.py", line 960, in _calc_ostr
return self._object._calc_ostr(train_pool, test_pool, top_size, ostr_type, update_method, importance_values_sign, thread_count, verbose)

File "_catboost.pyx", line 2752, in _catboost._CatBoost._calc_ostr

File "_catboost.pyx", line 2754, in _catboost._CatBoost._calc_ostr

CatBoostError: c:/goagent/pipelines/buildmaster/catboost.git/catboost/libs/target/data_providers.cpp:423: Non-Multiclassification compatible metric (RMSE) specified for a multidimensional model

catboost version: 0.15.1
Operating System: Microsoft Windows 10 Enterprise
CPU: Intel(R) CoreTM) i7-8700 CPU @ 3.20GHz
GPU:

Please note that the model (an object of catboost.CatBoostClassifier) was trained with loss_function as 'MultiClass' . The model could be trained successfully.