Multioutput regression using GPU

Question

I am trying to run Random Forest regression from cuml.ensemble on my dataset using GPU. My output will be 2 features and input is 7 features. But for some reason Random Forest regression using Rapids, doesn't take 2 column as output. It shows me error that 'EXpected one column but got 2." How can I resolve this issue? is there any way I can train the RF regression model for multi outputs?

X = df.iloc[:, 0:7]
Z = df[['PCA_1','PCA_2']]
x_train, x_test, y_train, y_test  = cuml.train_test_split(X, P, train_size=0.8, test_size = 0.2)
from cuml.ensemble import RandomForestRegressor
ranforest = RandomForestRegressor(n_estimators=120, n_bins = 8, accuracy_metric='r2', max_depth = 8, split_criterion = 'mse')
ranforest.fit(x_train, y_train)

ValueError: Expected 1 columns but got 2 columns.