null exception pointer on parallelized process

Hey,

I have a process which trains a SVM model on several data sets. I've found it to work fine when I use the regular Loop Parameters operator that's not parallelized. However, when I replace that operator with the parallelized version, I keep getting null exception pointers in the performance operator of my cross validation process. Does anyone know why this occurs and how to fix? It seems to happen a lot between non - parallelized processes and using the parallelized operator.

thanks