Tree Ensemble Learner: Execute Failed: Java Heap Space

Hello, I am trying to execute a workfow that reads a table of 71067 document vectors (each row coresponds to a document and each column coresponds to the occurences of a specific word in that document) and uses that for the input of a Tree Ensemble Learner Node. The node fails at about 10% consistently with the following error: 

ERROR Tree Ensemble Learner 0:43       Execute failed: Java heap space.

I have changed the values in the config file as follows, and set the nodes memory policy to write tables to disc

-XX:MaxPermSize=1024m

-Xmx5120m

How could I get this node to exectue correctly? Any help would be greatly appreciated.

The problem is that you probably have tons of columns (I suppose?)

It would help if you would filter unimportant terms or infrequent ones. Basically any kind of feature reduction would help. I would apply this before generating the document vectors.

Best, Iris