Optimize Word2Vec for Final Classification Task in Spark ML Pipeline

Question

I would like to run a cross-validation grid search over a Word2Vec model in a Spark pipeline with respect to a final binary classification problem. Right now, I train Word2V